Structure of acid-soluble and acid-insoluble glycogen

and their responses to changes in glycogen levels

in skeletal muscle

by

Phillip David Barnes

This thesis is presented for the degree of

Doctor of Philosophy of the University of Western Australia

Faculty of Life and Physical Sciences

School of Sport Science, Exercise and Health (2010)

Supervisor: Professor Paul A. Fournier

“The reasonable man adapts himself to the world; the unreasonable one

persists in trying to adapt the world to himself. Therefore all progress depends

on the unreasonable man.”

George Bernard Shaw

i

Declaration

The work involved in designing and conducting the studies described in this

thesis has been carried out primarily by Phillip D. Barnes (the candidate). The

thesis outline and experimental design of the studies was developed and

planned by the candidate in consultation with Professor Paul A. Fournier (the

candidate’s supervisor). All participant recruitment and management was

carried out entirely by the candidate, along with the actual organisation,

implementation and performance of the experiments. In addition, the candidate

was responsible for all data analysis and original drafting of the thesis and peer-

reviewed publications. Professor Paul A. Fournier has provided feedback for

further drafts and completion of the thesis and manuscripts.

Signed:

Phillip D. Barnes Paul A. Fournier

(Candidate) (Supervisor)

ii

Acknowledgements

Many people have helped me throughout the course of my PhD studies and I

would like to thank you all.

Firstly to my participants, both humans and rodents, without whose help I would

never have been able to conduct my research. I would like to extend my

deepest thanks. To the guys that so enthusiastically volunteered for my first

study, you are all invaluable, your willingness to participate was one of the few

things that prevented me from having a stress-induced meltdown. Thank you

also to Dr Anish Singh for volunteering your valuable time and expertise.

I would also like to extend a big thank you to Associate Professor Peta Clode at

the Centre for Microscopy, Characterisation and Analysis for so generously

donating your time to assist me with my electron microscopy work. Without your

invaluable assistance I would never have been able to complete my research.

To my family, Mum, Dad and Jeffrey for giving me the space I needed when I

needed it and the support I required when I required it. I am truly blessed to

have such a loving and understanding family, thank you.

Thank you to my friends, who remained my friends despite me regularly

disappearing into the laboratory for months on end only to surface for a week or

two before vanishing again. The rare times I was able to have a beer with you

really helped me through the final stages of my work.

Finally, I would like to thank Professor Paul Fournier for your supervision and

guidance, for putting up with my incessant need to argue the point and

entertaining my enthusiastic hypotheses. I cannot imagine that I would have

iii

been able to complete this work under the supervision of another, thank you. I

will never forget the influence you have had on my life.

iv

Publications

The work appearing in Chapter 2 of this thesis has been published in the

following peer reviewed journal:

Barnes PD, Singh A & Fournier PA. (2009). Homogenization-dependent

responses of acid-soluble and acid-insoluble glycogen to exercise and

refeeding in human muscles. Metabolism, Clinical and Experimental 58, 1832-

1839.

v

Abbreviations

% percentage

AIG acid insoluble glycogen

AMPK adenosine monophosphate-activated protein kinase

ASG acid soluble glycogen

BE branching enzyme

°C degrees centigrade

CHAPS 3[(3-cholamidopropyl)dimethylammonio]-1-

propanesulfonate

cm centimetre

CO2 carbon dioxide

CV coefficient of variance

Da Dalton

d.w. dry weight

g gravity

G1P glucose 1-phosphate

G6P glucose 6-phosphate

GL liver glycogen-binding regulatory subunit of protein

phosphatase 1

GM muscle glycogen-binding regulatory subunit of

protein phosphatase 1

GNIP glycogenin interacting protein

vi

GNIP2 glycogenin interacting protein 2

GP glycogen phosphorylase

GS glycogen synthase

h hour

H+ hydrogen ion

HCl hydrochloric acid

HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

kDa kiloDalton

kV kilovolts

L litre

LG lysosomal glycogen

M molar

ml millilitre

mg milligram

MG macroglycogen

Mg2+ magnesium ion

min minutes

mm millimetre

mM millimolar

mmol millimoles

Mn2+ manganese ion

µl microliters

vii

NaOH sodium hydroxide

nm nanometres

O2 oxygen

P phosphate

PCA perchloric acid

PG proglycogen

PhK phosphotylase kinase

PP1 protein phosphatase 1

PPP1R6 R6 regulatory subunit of protein phosphatase 1

PTG protein targeting to glycogen subunit of protein

phosphatase 1

SDS sodium dodecyl sulphate

s.r. sarcoplasmic reticulum

TCA trichloroacetic acid

TEM transmission electron microscopy

U units

UDP uridine diphosphate

UDPG uridine diphosphoglucose

UTP uridine triphosphate

V�O2 rate of oxygen consumption per minute

v/v volume to volume

w/v weight to volume

viii

Table of Contents

Declaration ........................................................................................................... i

Acknowledgements ............................................................................................. ii

Publications........................................................................................................ iv

Abbreviations ...................................................................................................... v

List of figures .....................................................................................................xiii

List of tables ..................................................................................................... xvi

Abstract..... .......................................................................................................xvii

Chapter 1 Literature Review ............................................................................... 1

1.1 Introduction ........................................................................................ 2

1.2 The discovery of glycogen ................................................................. 2

1.3 Current views on glycogen structure ................................................ 10

1.4 Brief overview of the discovery of the enzymes involved in the

breakdown of glycogen .................................................................... 15

1.5 Discovery of the enzymes involved in the synthesis of glycogen ..... 19

1.5.1 Glycogenin and the initiation of glycogen synthesis de novo......... 22

1.6 Glycosome ....................................................................................... 29

1.7 Cellular distribution of glycogen ....................................................... 34

ix

1.8 Acid-soluble and acid-insoluble glycogen ........................................ 38

1.9 Homogenisation-free extraction of acid-soluble and acid-insoluble

glycogen: artefact of tissue extraction? ............................................ 46

1.10 Statement of the problem ................................................................. 49

Chapter 2 Homogenisation-dependent responses of acid-soluble and acid-

insoluble glycogen to exercise and re-feeding in human muscles.... 51

2.1 Introduction ...................................................................................... 52

2.2 Materials and methods ..................................................................... 56

2.2.1 Materials ........................................................................................ 56

2.2.2 Participants.................................................................................... 56

2.2.3 Exercise and re-feeding protocol ................................................... 56

2.2.4 Anthropometric data and �O2 peak measurement ........................... 59

2.2.5 Muscle biopsies ............................................................................. 59

2.2.6 Acid extraction of muscle glycogen ............................................... 60

2.2.7 Glycogen determination ................................................................ 61

2.2.8 Expression of results and treatment and analysis of data ............. 62

2.3 Results ............................................................................................. 63

2.3.1 Glycogen yield of homogenisation-dependent and independent

protocols ........................................................................................ 63

x

2.3.2 Effect of exercise and re-feeding on ASG and AIG levels in

human muscles ............................................................................. 65

2.4 Discussion ........................................................................................ 70

Chapter 3 Molecular size distribution of acid-soluble and acid-insoluble

glycogen and the effect of extraction protocol .................................. 78

3.1 Introduction ...................................................................................... 79

3.2 Experimental procedures ................................................................. 83

3.2.1 Materials ........................................................................................ 83

3.2.2 Animals ......................................................................................... 83

3.2.3 Tissue sampling ............................................................................ 83

3.2.4 Acid extraction of muscle glycogen ............................................... 84

3.2.5 Molecular size distribution analysis using transmission electron

microscopy .................................................................................... 85

3.2.6 Glycogen determination ................................................................ 85

3.2.7 Expression of results and treatment and analysis of data ............. 86

3.3 Results ............................................................................................. 87

3.3.1 Optimisation of glycogen extraction: effect of repeated

homogenisation of glycogen on its molecular size determination

by gel filtration chromatography and transmission electron

microscopy .................................................................................... 87

xi

3.3.2 Optimisation of AIG extraction ....................................................... 91

3.3.3 Acid solubility of extracted AIG ...................................................... 94

3.3.4 Effect of pronase treatment on molecular size distribution of

glycogen ........................................................................................ 95

3.3.5 Molecular size distribution of ASG and AIG extracted with

homogenisation-free and homogenisation-dependent protocols ... 98

3.4 Discussion ...................................................................................... 105

Chapter 4 Effect of exercise and re-feeding on the molecular size

distribution of acid-soluble and acid-insoluble glycogen in

skeletal muscle............................................................................... 112

4.1 Introduction .................................................................................... 113

4.2 Experimental procedures ............................................................... 116

4.2.1 Materials ...................................................................................... 116

4.2.2 Animals ....................................................................................... 116

4.2.3 Exercise protocol ......................................................................... 116

4.2.4 Tissue sampling .......................................................................... 119

4.2.5 Acid extraction of muscle glycogen ............................................. 119

4.2.6 Extraction of AIG ......................................................................... 119

4.2.7 Molecular size distribution analysis using transmission electron

microscopy .................................................................................. 120

xii

4.2.8 Glycogen determination .............................................................. 121

4.2.9 Expression of results and treatment and analysis of data ........... 122

4.3 Results ........................................................................................... 123

4.3.1 Effect of exercise and re-feeding on the levels of glycogen,

AIG and ASG............................................................................... 123

4.3.2 Effect of exercise and re-feeding on the molecular size

distribution of AIG and ASG. ....................................................... 125

4.4 Discussion ...................................................................................... 127

Chapter 5 General Discussion ........................................................................ 131

5.1 General discussion ......................................................................... 132

Chapter 6 References ..................................................................................... 143

6.1 References ..................................................................................... 144

xiii

List of figures

Figure 1.1 Methylation of glycogen for chain length determination ................. 4

Figure 1.2 Structure of glycogen as originally proposed by Staudinger and

Husemann (1937) .......................................................................... 5

Figure 1.3 Structure of glycogen as originally proposed by Haworth and

colleagues (1937) .......................................................................... 5

Figure 1.4 Structure of glycogen as originally proposed by Meyer and

Bernfeld (1940). ............................................................................. 8

Figure 1.5 The revised Meyer model as proposed by Whelan and

colleagues (1970) .......................................................................... 9

Figure 1.6 The revised version of the Meyer and Bernfeld model

commonly referred to as the Whelan model of glycogens

structure ....................................................................................... 12

Figure 1.7 Disruptive phosphorylation of the glycogen molecule by

phosphorylase.............................................................................. 16

Figure 1.8 The cooperation of GP and debranching enzyme as required

for complete digestion of glycogen............................................... 17

Figure 1.9 Initiation of glycogen synthesis by a protein primer as proposed

by Krisman and Barengo (1975). ................................................. 24

xiv

Figure 1.10 The complex formed between glycogenin and GS during the

initiation of glycogen synthesis as proposed by Smythe and

colleagues .................................................................................... 27

Figure 1.11 Diagrammatic representation of the muscle glycosome with its

associated proteins ...................................................................... 31

Figure 2.1 Experimental design of the study. ................................................ 58

Figure 2.2 A comparison of total glycogen in human muscle determined

using a homogenisation-free protocol and a homogenisation-

dependent protocol ...................................................................... 64

Figure 2.3 Pattern of response of total muscle glycogen to exercise and

recovery ....................................................................................... 66

Figure 2.4 Effect of exercise and recovery on (A) the pattern of response

of ASG and AIG using a homogenisation-free protocol and (B)

changes in concentrations of ASG and AIG ................................. 68

Figure 2.5 Effect of exercise and recovery on (A) the pattern of response

of ASG and AIG using our homogenisation-dependent protocol

and (B) changes in concentrations of ASG and AIG. ................... 69

Figure 3.1 Effect of extensive homogenisation on glycogen molecular size

distribution using gel filtration chromatography ............................ 90

Figure 3.2 Effect of extensive homogenisation on glycogen molecular size

distribution using transmission electron microscopy .................... 90

Figure 3.3 Incubation of AIG pellet with various extraction buffers. .............. 93

xv

Figure 3.4 Acid solubility of pronase-extracted AIG ...................................... 94

Figure 3.5 The effect of pronase digestion on the glycogen molecular size

distribution A) without the inclusion of acarbose and B) in the

presence of acarbose .................................................................. 97

Figure 3.6 Extraction of ASG and AIG for TEM size distribution analysis ..... 99

Figure 3.7 Electron microscopy of purified A) AIG and B) ASG extracted

using the homogenisation-dependent protocol .......................... 101

Figure 3.8 Glycogen molecular size distributions of ASG and AIG

extracted using the homogenisation-dependent protocol

expressed by size frequency...................................................... 102

Figure 3.9 Electron microscopy of purified A) AIG and B) ASG extracted

using the homogenisation-free protocol ..................................... 103

Figure 3.10 Glycogen molecular size distributions of ASG and AIG

extracted using the homogenisation-free protocol expressed

by size frequency ....................................................................... 104

Figure 4.1 Exercise and muscle sampling protocol ..................................... 118

Figure 4.2 Changes in ASG and AIG in response to a 3-minute bout of

intense exercise and recovery ................................................... 124

Figure 4.3 Molecular size distributions of A) AIG and B) ASG pre-exercise,

post-exercise and after 24 hours of recovery ............................. 126

xvi

List of tables

Table 1.1 Structural parameters of a mature β-glycogen ............................. 13

xvii

Abstract

Muscle glycogen extracted in the presence of acid results in an acid soluble

(ASG) and acid insoluble (AIG) fraction, with AIG levels reported in most recent

studies to be the most responsive to changes in muscle glycogen levels. The

different acid-solubilities of these two glycogen fractions have been explained

on the grounds that AIG corresponds to a lower molecular weight glycogen

species referred to as proglycogen (PG), whereas ASG corresponds to

macroglycogen (MG). Given that the extraction protocol adopted in those recent

studies did not include a homogenisation step, the first objective of this thesis

was to determine whether the inclusion of such a step can affect ASG and AIG

responses to changes in muscle glycogen levels. We found that the patterns of

change in AIG and ASG levels to exercise and re-feeding in humans is highly

sensitive to the protocol of extraction, with ASG being the most responsive

fraction when a homogenisation step is included, but AIG when glycogen is

extracted without a homogenisation step. Given the currently held view that the

acid-solubility of glycogen is determined by its size, with AIG corresponding to a

glycogen population of low molecular weight, our next objective was to compare

the molecular sizes of AIG and ASG from rat muscles extracted using a

homogenisation-free and homogenisation-dependent protocol. Against

expectation, both AIG and ASG were found to have a similar average molecular

size and pattern of molecular size distribution irrespective of the extraction

protocol. The different solubility between AIG and ASG is more likely the result

of the binding of different complements of proteins to AIG compared to ASG as

suggested by AIG becoming acid-soluble after treatment with proteases. Given

that the responses of the molecular sizes of AIG and ASG to changes in muscle

xviii

glycogen levels have never been examined before, our third objective was to

perform such an analysis examining for the first time how the molecular size

distributions of AIG and ASG from homogenised muscle extracts respond in

rats subjected to exercise and recovery. To this end, groups of fasted rats were

sampled before and immediately after an intense three-minute bout of

swimming as well as 24 hours post-exercise. At rest, the molecular size

distributions of both AIG and ASG were again similar. However, immediately

after exercise, the molecular size distribution of ASG shifted markedly towards

glycogen particles of smaller sizes, whereas that of AIG changed little. After 24

hours of recovery, the molecular size distributions of AIG and ASG were similar,

with their average molecular sizes being comparable to those found prior to

exercise. In agreement with these findings, all changes in total glycogen levels

were accounted for by ASG. Such different responses of AIG and ASG to

exercise suggest that these glycogen species correspond to physiologically

distinct glycogen populations, with the mechanism underlying their different acid

solubilities remaining to be elucidated.

1

Chapter 1

Literature Review

2

1.1 Introduction

Since the discovery of glycogen by Claude Bernard in the 1850’s and the

subsequent realisation of its important role in whole body metabolism, there

have been countless studies into the structure, regulation and metabolism of

this molecule. Glycogen, the body’s store of carbohydrates, acts as a rapidly

available but limited source of fuel. In particular, muscle glycogen, which

accounts for 50-80% of the body’s total glycogen stores (Shearer & Graham,

2002), is the site for intramuscular glucose storage and provides a major fuel for

muscular work. Whilst the total energy derived from glycogen is limited

compared to lipid, glycogen metabolism is known to influence whole body fuel

homeostasis, exercise performance and the onset of fatigue, and is implicated

in metabolic diseases such as diabetes mellitus where insulin-stimulated

glycogen storage is impaired. It is not surprising, therefore, that the metabolism

and regulation of glycogen has been the object of an impressive volume of

research.

1.2 The discovery of glycogen

In 1843, Claude Bernard reported that cane sugar administered intravenously to

an animal was completely excreted in the urine, but when treated with digestive

enzymes before injection, the sugar was assimilated into the animal’s body

(Bernard, 1843). This important finding led Bernard to investigate the metabolic

fate of the sugar entering the blood stream from the digestive tract. Ultimately,

these experiments led Bernard to discover that the liver was able to produce

sugar from a sugar-forming substance found within the liver itself (Bernard,

1855), and then later described the extraction and isolation of this substance, or

3

“la matiere glycogene” (Bernard, 1857). Shortly after the discovery of liver

glycogen, Sanson (1857) reported that skeletal muscle contained an almost

identical substance to that found by Bernard.

Bernard also demonstrated that when glycogen is completely hydrolysed only

glucose remains, but when digested in the presence amylases, maltose is

formed (Bernard, 1857). Since maltose is a disaccharide containing an α-1,4-D-

glucosidic bond, this suggested that glycogen consists of glucose residues

linked in a chain via α-1,4-D-glucosidic bonds (Young, 1957). However, the

number of glucose residues involved remained unclear.

In order to determine the length of the glucose chains, Haworth and Percival

(1932) examined the hydrolysis of fully methylated rabbit liver glycogen.

Methylation of a polysaccharide binds all hydroxyl groups not involved in

bonding to a methyl residue (Figure 1.1). The glucose residue at the end of a

chain is thus expected to have one more methyl residue attached then those

located along the chain. By measuring the amount of tetra-methyl glucose

liberated in relation to tri-methyl glucose, they concluded that glycogen has a

linear chain length of at least 12 glucose residues (Haworth & Percival, 1932).

Further studies using methylation and periodate oxidation analysis led to the

finding that the length of glucose chains in native glycogen can range from 10 to

18 residues; with the majority found between 10 and 14 (Halsall et al., 1947). It

was also speculated that the whole glycogen molecule is much larger than just

12 residues and that multiple chains are joined by some form of linkage (Young,

1957).

4

Tetramethyl glucose Trimethyl glucose and MeOH

Figure 1.1 Methylation of glycogen for chain length determination. Figure

modified from Haworth and Percival (1932).

The discovery of some di-methyl glucose as a hydrolysis product of methylated

glycogen provided evidence of such branching, as glucose linked at three

carbons would have only two exposed hydroxyl groups and appear as di-methyl

glucose (Bell, 1937; Haworth et al., 1937). This led Haworth and colleagues

(1937) to propose that, like starch, glycogen consists of glucose units joined at

positions 1 and 4 in linear chains and that these chains are linked together by a

bond connecting the reducing end of one chain with one of the hydroxyl groups

of an adjoining chain (Haworth et al., 1937).

Staudinger and Husemann (1937) proposed an alternative structure for the

glycogen molecule. Using viscosity measurements, they provided evidence that

the glycogen particle is almost spherical and of a molecular weight much

greater then described by other assays. In an effort to explain this data and the

previous methylation results, they proposed a “comb-like” structure of glycogen

with a central chain of 1,4-linked glucose up to 100 residues long, from which

5

chains of 12 residues would attach at carbons 2, 3 and 6 of each residue

(Figure 1.2; Staudinger & Husemann, 1937).

R = chain of 12 to 18 glucose residues

Figure 1.2 Structure of glycogen as originally proposed by Staudinger

and Husemann (1937). Figure modified from Bell (1937).

Figure 1.3 Structure of glycogen as originally proposed by Haworth and

colleagues (1937). Figure modified from Bell (1937).

6

One major difficulty with Staudinger and Husemann’s proposal is that upon

methylation and hydrolysis, this macromolecular structure would yield tri and

tetra-methyl glucose from the side chains, and, hydrolysis of the central chain

would yield un-substituted D-glucose with no di-methyl glucose fraction formed.

This structure thus, failed to explain the di-methyl glucose product found

previously and no evidence has been presented since then to support the

formation of free glucose after hydrolysis (Bell, 1937).

While the aforementioned research was being performed, some evidence that

glycogens had a large molecular weight was provided by the measurement of

glycogen’s osmotic properties. By analysing both rabbit liver and rabbit muscle

glycogen, Oakley and Young (1936) found mean particle weights of up to 2 x

106, similar to that found with viscosity measurements (Staudinger &

Husemann, 1937). This was later confirmed with osmotic pressure analysis of

glycogen from many different species, with mean molecular weights ranging

between 2 x 105 and 2.5 x 106 (Carter & Record, 1939).

On the basis of the large molecular weight of glycogen together with the

methylation results, Haworth and colleagues proposed a laminated structure for

glycogen (Haworth et al., 1937). This structure consisted of chains of glucosyl

residues linked by a hydroxyl group other than at carbon 1 and 4 of a non-

terminal glucose residue (Figure 1.3). If a glucose residue was involved in three

linkages, only two of the five hydroxyl groups would be available for

methylation, and would yield dimethyl glucose upon hydrolysis (Haworth et al.,

1937). Having identified small amounts of 2,3-dimethyl glucose among the

hydrolysis products of fully methylated glycogen, the authors speculated the

presence of a 1,6 bond between chains; however the nature of the bond was

7

not elucidated (Haworth et al., 1937). Haworth and colleagues further

postulated that this structure could accommodate between 3000 and 5000

glucose residues with a molecular weight of 500 000 to 800 000 (Haworth et al.,

1939).

Barker and colleagues (1941) reported that 2,3-di-methyl glucose made up 3%

of the hydrolysis products from methylated rice starch, providing additional

evidence for the involvement of carbon 6 in linking chains within

polysaccharides (Barker et al., 1941). The isolation of isomaltose, a

disaccharide consisting of two glucose residues linked via α-1,6-D-glucosidic

bond, from the hydrolysis products of acetylated glycogen supported the

proposed branching pattern of the glycogen molecule (Wolfrom & O'Neill, 1949).

A few years later, Meyer and colleagues (1940; 1941) proposed an irregular,

highly branched tree-like structure based on data from the enzymatic digestion

of glycogen (Figure 1.4). Commercial mussel glycogen, with a chain length of

11 residues, was digested with wheat β-amylase, reducing the average chain

length to 5.5 residues. Under these conditions, the β-amylase digestion is

limited to those glucose residues exterior to the branching point (Meyer &

Bernfeld, 1940; Meyer & Fuld, 1941). The Meyer model of glycogen structure

also assumes that all chains had equal growth and all non-reducing ends were

found at the surface of the molecule (Meyer & Bernfeld, 1940; Meyer & Fuld,

1941). Interestingly, their findings could not be used to distinguish between

single and multiple branching of the glycogen molecule. Thus, their proposed

multiple branched structure, now known to be correct, was not fully

experimentally grounded at the time (Manners, 1991).

8

Figure 1.4 Structure of glycogen as originally proposed by Meyer and

Bernfeld (1940).

By the 1940’s, the proposed structure of glycogen as linear chains of glucose

residues branching from each other at specific branch points was generally

accepted. However, the first unambiguous evidence for glycogen’s multiple

branching was presented by Larner and co-workers in 1952. Using step-wise

enzymatic degradation of glycogen with muscle phosphorylase and amylo-1,6-

glucosidase, they demonstrated conclusively that multiple branches emanate

from a single glucose chain (Larner et al., 1952).

Later, Peat, Whelan and their colleagues (1952, 1956) provided further

evidence of multiple branching within polysaccharides and also introduced the

concept of A, B and C chains (Peat et al., 1952, 1956). In type A chains, only

carbon 1 of the reducing end is involved in linking the chain to the

polysaccharide, whereas only carbon 6 of the glucose residues is engaged in

type C chains. The attachment of a type B chain uses carbon 1 of the reducing

end as well as carbon 6 of one or more residues (Peat et al., 1952). The degree

of branching in the molecule can then be expressed as the ratio of “A” to “B”

chains (Manners, 1991).

9

Combining step-wise enzymatic degradation with enzymatic determination of

chain length, Whelan and colleagues presented a revised version of the Meyer

model for glycogen (Figure 1.5; Gunja-Smith et al., 1970). They showed that

both A and B chains were of a similar length, averaging 14 residues, with non-

reducing chain ends within the molecule that are not accessible to β-amylase

and phosphorylase. To explain these results, the revised model proposed half

the B chains carrying and average of two A chains and the other half carrying

an average of two B chains (Gunja-Smith et al., 1970). This model, referred to

as the Whelan model, is now the generally accepted model of glycogen

structure (Manners, 1991).

Figure 1.5 The revised Meyer model as proposed by Whelan and

colleagues (1970). Diagrammatic representation of a glycogen

with A:B ratio of ~1:1 and degree of branching of 2.

10

1.3 Current views on glycogen structure

It is important to remember that the proposed models for glycogen structure are

not to be taken literally. Gunja-Smith and colleagues (1970) stated that their

model “is intended only to express certain concepts and is not to be regarded

as precisely defining glycogen structure”. It is also highly unlikely that, due to

the degree of branching and size variations, any two glycogens from animal

tissue have identical structures (Stetten & Stetten, 1960). Although not fully

understood, many of the details of this complex structure have been elucidated.

From examination of glycogen from multiple sources we know that its average

chain length is approximately 12 units; however, individual chains can range

from 6 to more than 50 glucose residues (Gunja-Smith et al., 1970; Akai et al.,

1971; Craig et al., 1988). In addition A and B chains are found in similar

numbers with an A:B-chain ratio of 0.7 to 1.0.

Glycogen structure is also described in terms of exterior and interior chains.

Exterior chains are the portion of the glucose residues from the final branch

point to the non-reducing end of a chain, and internal chains are the portion of

residues between two branch points. Mathematically, there is an equal number

of exterior and interior chains in a glycogen molecule (Manners, 1991). The A

chain, being un-branched, consists only of one exterior chain and is found

almost exclusively in the outermost portion of the glycogen particle, although it

is possible for “buried” A chains to exist within glycogen itself (French, 1964).

The B chain, in contrast, consists of one external chain with one or more

internal chains, and most B chains are found within the glycogen molecule. In

mammalian glycogen, B chains have, on average, two branch points, with

11

interior chain lengths averaging four glucose residues in length. Also, glycogen

is taken as 90% α-1,4-D-glucosidic bonds, and α-1,6-D-glucosidic bonds

represent less than 10% of all the bonds in a glycogen particle (Manners, 1991).

As mentioned above, the current model for glycogen structure in mammalian

tissue is based upon that proposed by Whelan and co-workers (1970). This

model presents glycogen as a roughly spherical structure arranged in

concentric tiers about a core (Figure 1.6; Gunja-Smith et al., 1970). Each B

chain has two branch points and as such the number of chains in any tier is

double that of the previous tier (Melendez et al., 1998). This organisation into

concentric tiers represents layers of branched chains enclosing one another

and adding up to a three dimensional structure with the shape of a sphere. This

extensive branching is at the origin of the bush-like structure of glycogen,

referred to as β-particle, and has the advantage of maximising the number of

end points available to glycogen metabolising enzymes (Shearer & Graham,

2002).

12

Figure 1.6 The revised version of the Meyer and Bernfeld model

commonly referred to as the Whelan model of glycogens

structure. Figure adapted from Melendez and colleagues (1998).

Based on this model and on the β-particle of rabbit muscle

glycogen having a maximal molecular weight of the 107 Da

(Wanson & Drochmans, 1968), several structural parameters for a

mature glycogen β-particle have been inferred as described in

Table 1.1 (Goldsmith et al., 1982).

13

Table 1.1 Structural parameters of a mature β-glycogen. Table adapted

from Goldsmith and colleagues (1982).

Molecular weight of rabbit muscle β-particle

(Reported by Wanson and Drochmans (1968))

107 Da

Total number of tiers 12

Total number of glucose residues 55 000

Average chain length 13 glucose residues

Effective length per tier

(every tier is the same thickness)

1.9 nm x 2 = ~ 3.8 nm

Total diameter of the particle 3.8 nm x 11 tiers = ~ 42 nm

These structural details about glycogen imply that in a full 12-tier glycogen

molecule, the amount of glucosyl residues directly available for phosphorolytic

degradation is 34.6% of the total glucose contained in the particle, which is

approximately 18 500 residues in a fully mature glycogen particle (~34.6% of 53

000). This pattern of glucose availability continues as each tier of the glycogen

is depleted, with 9 200 residues (~34.6% of 26 600) in the 11th tier, 4 600

residues (~34.6% of 13 300) in the 10th tier, and 2 300 (~34.6% of 6 600) in the

9th tier. Accordingly, only approximately 6% of the total glucose of a mature β-

particle is located in tiers 8 to 1 (Melendez et al., 1997).

Although, mathematically, it would be possible for a 13th tier to be added to a

glycogen granule, the molecule itself is sterically limited for further growth

(Melendez et al., 1997). Meléndez and colleagues (1997) reported that in the

12th tier, glucose occupies 26% of the space while in a hypothetical 13th tier,

the space occupied by glucose would be 62%. As the enzymes involved in

glycogen metabolism also occupy space, the density reached in the 13th tier

14

would be such that enzymes would have no room left to attach to the glucosyl

residues (Melendez et al., 1997).

Interestingly, despite a mature 12-tier β-particle having a diameter of 42 nm,

very few particles of this size are found in glycogen extracts. Many studies have

reported glycogen molecules with sizes in the range of 12–40 nm with averages

of ~26 nm (Drochmans, 1962; Scott & Still, 1968; Meyer et al., 1970;

Schmalbruch & Kamieniecka, 1974; Rybicka, 1981b, a; Marchand et al., 2002;

Marchand et al., 2007; Ryu et al., 2009). Research involving carbohydrate

loading of participants have regularly reported marked increases in skeletal

muscle glycogen stores (Price et al., 2000; Arnall et al., 2007; Marchand et al.,

2007; McLay et al., 2007; Rico-Sanz et al., 2008; Barnes et al., 2009), yet,

average glycogen particle sizes are well below the theoretical maximum

diameter (Marchand et al., 2007).

Other structural features of the glycogen molecule are the presence of

phosphate covalently bound to the glucose residues (Lomako et al., 1993b;

Lomako et al., 1994). Glycogen from mammalian muscle contains

approximately 0.064% by weight of phosphate or 0.121% by molar

concentration (Lomako et al., 1993b); however, the mechanism by which the

phosphate is introduced and its role remain unclear (Tagliabracci et al., 2008).

It is important to note that the glycogen molecules in the liver have very high

molecular weights of up to ~106 kDa (Orrell & Bueding, 1964; Geddes et al.,

1977b). Using electron microscopy, Drochmans (1962) reported that these large

α-particles have molecular sizes ranging from 40–200 nm and appear as rosette

clusters formed from connected β-particles. These α-particles are not

15

dissociated by reagents known to disrupt hydrogen or peptide bonds including

1% Triton X-100, 1% SDS, 8 M urea, 8 M guanidine and 8 M lithium bromide,

and are stable from pH 5.0 to 12.0 (Orrell & Bueding, 1964).

Many studies have speculated as to the formation and functional significance of

these α-particles (Drochmans, 1962; Geddes et al., 1977a, b; Takeuchi et al.,

1978; Devos et al., 1983; Rybicka, 1996; Roach, 2002), although the origin of

these particles still remains unclear. Most recently, Sullivan and colleagues

(2010) have provided evidence, through theoretical modelling and experimental

data, that glucosyl chains from multiple β-particles may be covalently linked to

each other to form an α-particle (Sullivan et al., 2010). The authors stress

however, that there exists no obvious enzymatic process to cause this binding,

nor does it explain why α-particles are found exclusively in the liver and not in

skeletal muscles (Sullivan et al., 2010).

1.4 Brief overview of the discovery of the enzymes involved in

the breakdown of glycogen

In 1937, Cori and colleagues demonstrated that glycogen, when added to

dialysed muscle extract, was broken down to form glucose-1-phosphate (G1P)

in the presence of inorganic phosphate. The enzyme responsible was named

“phosphorylase” (Cori et al., 1937) and now referred to as “glycogen

phosphorylase” (GP). They proposed that the formation of G1P was the result

of disruptive phosphorylation of the glycogen molecule, where inorganic

phosphate enters the α-1,4-D-glucosidic bond and liberates a glucosyl residue,

without water being involved (Figure 1.7; Cori et al., 1938), and this reaction

16

was subsequently referred to as “phosphorolysis” instead of “hydrolysis”

(Parnas, 1937).

Figure 1.7 Disruptive phosphorylation of the glycogen molecule by

phosphorylase. Figure modified from Cori and colleagues (1938).

Further research on glycogen breakdown by GP led to the view that GP itself is

not able to completely digest the polysaccharides. Targeting the α-1,4-

glycosidic bonds of the chains, phosphorolysis was found to be halted when the

enzyme approached the α-1,6-glycosidic bonds of the branch points (Cori &

Larner, 1951). Extensive GP digestion would result in a phosphorylase limit

dextrin 34.6% smaller than the original molecule (Hestrin, 1949). A second

enzyme is thus required to remove the 1,6 branch points for phosphorolysis to

continue. It was known that crude muscle extract could completely digest the

polysaccharide (Hestrin, 1949); however no specific “debrancher” had been

identified.

Through exhaustive preparation, Cori and Larner (1951) produced a muscle

extract that was free from amylase and phosphorylase activity, yet maintained

the glucosidase activity of the crude extract. Incubation of this “debranching”

enzyme with the phosphorylase limit dextrin resulted in only a 3% degradation

of the limit dextrin. A close examination of the reaction products revealed only

17

D-glucose had been released via hydrolysis of the α-1,6 branch point, hence the

enzyme was termed amylo-1,6-glucosidase or debranching enzyme (Cori &

Larner, 1951). However, no glycogen breakdown occurred when debranching

enzyme was incubated with either whole glycogen or any partially digested

glycogen that was larger than the phosphorylase limit dextrin, demonstrating the

specificity of debranching enzyme for the single α-1,6-bound glucose residue

exposed by GP digestion (Cori & Larner, 1951). Removal of this branch point by

the debranching enzyme allows the continued release of G1P by GP. Thus GP

and debranching enzyme work together allowing complete digestion of the

glycogen molecule (Figure 1.8; Cori & Larner, 1951). This, then, constituted the

pathway of glycogen degradation, or glycogenolysis, in skeletal muscle (Cori &

Larner, 1951).

1,4-glycogen + P → G1P → G6P → glycolysis

1,6-glycogen + H2O → glucose → glycolysis

Main branch

Side branchSide branch

Limit dextrin

GP GP

Debranching enzyme Debranching enzyme

Figure 1.8 The cooperation of GP and debranching enzyme as required

for complete digestion of glycogen. Figure modified from Cori

and Larner (1951).

18

More recently a second enzymatic activity of debranching enzyme has been

identified. Once GP activity has degraded a α-1,4-bound glucose chain to only

four remaining residues, debranching enzyme, through a transferase activity,

recognises this shortened chain and transfers three of the glucose residues to

another α-1,4-bound chain before removing the single α-1,6-bound glucose

residue a shown above (Liu et al., 1995).

There is little evidence for the debranching enzyme to have any regulatory

properties, and it is not generally considered to be the rate limiting step for

glycogenolysis (Roach, 2002). In contrast, GP is tightly regulated in mammalian

tissue via allosteric regulation and reversible phosphorylation. Allosteric

effectors include for instance G6P, an inhibitor of GP, and adenosine mono-

phosphate (AMP), a potent allosteric activator. The phosphorylation of GP

activates it and is a reaction catalysed by phosphorylase kinase (PhK; Shearer

& Graham, 2002). The enzyme responsible for dephosphorylating GP is a type

1 protein phosphatase, found in skeletal muscle as a glycogen associated

phosphatase (PP1). This enzyme consists of a type 1 catalytic subunit

associated with a glycogen targeting regulatory subunit (GM), with this subunit

occurring predominately in striated muscle and containing a hydrophobic

domain that anchors the enzyme to cellular membranes (Roach, 2002).

Because of the physiological importance of GP in the regulation of glycogen

breakdown, this is a topic that has been thoroughly reviewed and will not be

examined any further here (Meyer et al., 1970; Fletterick & Madsen, 1980;

Jenkins et al., 1981; Johnson, 1992; Roach, 2002; Shearer & Graham, 2002;

Johnson, 2009).

19

1.5 Discovery of the enzymes involved in the synthesis of

glycogen

Originally, it was thought that GP was responsible for both the cleaving and

formation of the α-1,4-glycosidic bonds in the glycogen molecule (Stetten &

Stetten, 1960). Using G1P as a substrate, Cori and colleagues (1939) showed

that GP preparations from tissue extracts were able to synthesise, in vitro, a

polysaccharide that displayed properties similar to those of native glycogen

(Cori et al., 1939). Interestingly, they noted that GP from muscle extract

required glycogen to be present as a primer for synthesis to occur (Cori & Cori,

1939). Also, the polysaccharide produced by highly purified crystalline muscle

GP only consisted of straight chains of glucose, similar to amylose, and of much

greater length then found in native glycogen (Hassid et al., 1943). This

suggested that a second enzyme is required to produce the highly branched

structure of native glycogen.

Although the enzymes from muscle tissues displayed no branching properties,

preparations from other tissues were able to form branched polysaccharides,

due to suspected contamination by a “branching enzyme” (Cori & Cori, 1943).

By adding a liver extract to muscle GP preparations, Larner (1953) was able to

synthesise a branched polysaccharide not dissimilar to glycogen. Isotopic

labelling demonstrated that the 1,6-linked glucose branches had been created

from previously 1,4-linked residues (Larner, 1953). The enzyme that plays this

role, branching enzyme (BE) is a transglucosidase that catalyses the formation

of the α-1,6-glycosidic bonds that form the branch points between glucose

chains (Larner, 1953). Mammalian BE acts on a chain by cleaving an α-1,4-

20

glucosidic bond and removing a segment of residues and reattaching them via

an α-1,6-glycosidic bond (Larner, 1953).

Over the following years, doubts began to grow as to the proposed role of GP in

the synthesis of glycogen in mammalian cells. In 1957, Leloir and Cardini

reported that a partially pure preparation of rat liver extract could synthesise

glycogen without the presence of G1P, instead using UDP-glucose (UDPG) as

a substrate and with glycogen again required as a primer for the reaction (Leloir

& Cardini, 1957). Later, they demonstrated that the incorporation of glucose

from UDPG into the primer led to the formation of α-1,4-glycosidic bound

glucose residues (Leloir & Goldemberg, 1960). Soon after this discovery,

UDPG-transferase activity was reported in muscle tissue, as was UDP-

pyrophosphorylase (Villar-Palasi & Larner, 1960), the enzyme that catalyses the

reaction of G1P with UTP to produce UDPG and di-phosphate. This discovery

led to the proposal of the following glycogen synthesis pathway completely

independent of inorganic phosphate.

glucose → G6P → G1P → UDPG → polysaccharide

By 1969, the enzyme responsible for the UDPG-transferase action, glycogen

synthase (GS), had been extracted from rat liver and, when combined with BE,

would synthesise high molecular weight glycogen in vitro. This synthetic

glycogen, unlike that formed by GP, did not differ significantly from native

glycogen (Parodi et al., 1969). GS and BE are now known to be the two

enzymes that catalyse glycogen’s growth (Shearer & Graham, 2002).

GS attaches a UDPG to the distal end of an existing chain with an α-1,4-

glucosidic bond. This process is continued until the chain is between 10 to 18

21

residues in length (Melendez et al., 1997). GS requires the presence of a very

specific acceptor or primer before it catalyses the 1,4-binding of glucose

residues. When α-dextrins, with singular glucose A-chains are used as

acceptors, rabbit muscle GS selectively add glucose units to the B-chains and

not the single glucose A-chains (Brown et al., 1965). When glycogen or

phosphorylase limit dextrins are used as primers, there is similar asymmetrical

growth of the B-chains over the A-chains (Brown et al., 1965). It has since been

demonstrated that a polysaccharide with a degree of polymerisation of less than

four residues will not serve as a primer for mammalian GS, regardless of UDPG

concentration (Manners, 1991).

GS is controlled by the binding of allosteric ligands, especially G6P, and

covalent phosphorylation (Roach, 2002). Mammalian GS can be

phosphorylated at as many as nine sites by a variety of protein kinases,

resulting in progressive inactivation. This progressive phosphorylation is

hierarchal in that the addition of a phosphate at one site is required for

enzymatic recognition and subsequent phosphorylation of the next site (Roach,

2002). Unlike GP, dephosphorylation of GS leads to an increase in its activity.

The enzyme chiefly responsible for this dephosphorylation is PP1, the same

enzyme that dephosphorylates PhK and GP (Cohen, 1989). The allosteric

binding of G6P not only leads to a marked increase in GS activity but also

serves to protect the enzyme from inactivation (Leloir & Goldemberg, 1960;

Roach, 2002). Given the many factors involved in the regulation of GS, it comes

as no surprise that it has been the subject of a large volume of research and the

subject of several recent reviews and for this reason this topic will not be

22

examined further (Nielsen & Richter, 2003; Hargreaves, 2004; Nielsen &

Wojtaszewski, 2004; Graham, 2009; Jensen & Lai, 2009).

BE also has very specific acceptor requirements, with the mammalian enzyme

displaying no activity towards glycogen with outer chain lengths of six or less

residues; however readily catalyses glycogen with outer chain lengths

exceeding 11 glucose units (Larner, 1953). When GS extends a chain to 11

residues, BE removes approximately 7 glucosyl residues to form an α-1,6-

branch point, thus creating a B chain. GS then attaches further glucosyl units to

the original chain, before BE then creates a second branch point. GS then

continues to elongate the chains to an average of 13 residues. This process is

repeated continuously creating the bush like structure of a mature glycogen

(Shearer & Graham, 2002).

1.5.1 Glycogenin and the initiation of glycogen synthesis de novo

As mentioned before, early attempts to synthesise glycogen from glucose

derivatives (G1P, UDPG) in vitro proved unsuccessful. However, it was possible

to incorporate glucose into pre-existing glycogen molecules (Cori et al., 1939;

Cori & Cori, 1939; Hassid et al., 1943; Hauk et al., 1959). This suggested that

the initiation of glycogen synthesis requires the presence of a primer (Leloir &

Cardini, 1957).

In their attempt to discover this glycogen primer, Krisman (1972) found that

UDP[14C]-glucose could be incorporated into the trichloroacetic acid (TCA)

insoluble fraction of a rat liver extract. The acid-insolubility of this species

suggested it was a glycoprotein (Krisman, 1972, 1973). The fact that the

product was rendered acid-soluble, following de-proteinisation by incubation

23

with pronase, further supported this notion. It was also noted that, in the

presence of glycogen, the formation of TCA-insoluble UDP[14C]-glucose product

was inhibited, while addition of glycogen after the reaction had completed had

no effect (Krisman, 1973). The TCA-insoluble glycoprotein, whose saccharide

moiety was comprised of α-1,4-glucosidic bound glucose residues, still

precipitated in TCA despite treatment with 1% Triton X-100, urea, alkali and

phenol, and also after conditions that cause the β elimination of sugar residues

bound to serine or threonine (Krisman, 1973).

To explain their findings, Krisman and colleagues (1975) proposed that

glycogen originates from a protein back bone, which acts both as an acceptor

for and an initiator of glycogen synthesis. They presented a pathway for

glycogen biosynthesis involving four proteins. Initially, a “glycogen initiator

synthase” would attach UDPG residues, forming short maltosaccharide chains,

to multiple sites on a protein acceptor. This would continue until the glucose

chains of the glycoprotein were of sufficient length to act as a primer for GS.

Then, GS and BE would continue the synthesis of glycogen (Figure 1.9;

(Krisman & Barengo, 1975).

24

Figure 1.9 Initiation of glycogen synthesis by a protein primer as

proposed by Krisman and Barengo (1975).

25

Later, Whelan and colleagues (1985) reported that rabbit skeletal muscle

glycogen, when prepared under conditions known to strip all non-covalently

bound protein, still contained a constant protein fraction of 0.35% by mass,

which could not be decreased (Kennedy et al., 1985). Although the glycogen

was too large for gel electrophoretic analysis, stepwise enzymatic degradation

of the carbohydrate moiety with α-amylase and amyloglucosidase, led to the

appearance of a single 37 kDa protein band. This covalently bound protein was

subsequently named glycogenin (Rodriguez & Whelan, 1985). Glycogenin has

since been found in glycogen extracted from many other tissue types (Aon &

Curtino, 1984; Carrizo et al., 1997) and proved to be highly resistant to

dissociation, even after treatment with detergents (Krisman, 1973; Aon &

Curtino, 1984), urea (Krisman, 1973; Aon & Curtino, 1984) and

mercaptoethanol (Aon & Curtino, 1984).

Glycogenin is an N-acetylated protein of 332 amino acids, with a molecular

weight of 37,284 Da when devoided of glucose (Campbell & Cohen, 1989;

Gibbons et al., 2002). Glycogenin behaves as a glucosyltransferase (Pitcher et

al., 1988) that has three essential roles in the synthesis of glycogen. Initially,

using UDP-glucose as a glucose donor, glycogenin forms a C-1-0 tyrosyl bond

with glucose at a single binding site at residue Tyr-194 (Rodriguez & Whelan,

1985; Smythe et al., 1988). Following this self-glucosylation, glycogenin

continues to add glucose residues via the formation of α-1,4-glucosidic bonds,

to form a chain of approximately 8 glucosyl units. This glucosyltransferase

activity is dependent on Mg2+ or Mn2+ (Pitcher et al., 1987; Pitcher et al., 1988).

Finally, the glucosylated glycogenin then acts as a substrate and acceptor for

26

GS, which continues with the elongation of the glucose chain (Pitcher et al.,

1988).

In fed rabbit skeletal muscle, glycogenin and GS have been reported to co-

precipitate with glycogen in a 1:1 molar ratio (Pitcher et al., 1987; Pitcher et al.,

1988). It was proposed that during early glycogen synthesis, GS and glycogenin

are united, but as glycogen grows, GS eventually disassociates from glycogenin

and move to the outer branches of glycogen to continue adding glucose

residues (Pitcher et al., 1987; Pitcher et al., 1988; Smythe et al., 1990; Roach &

Skurat, 1997). Evidence for the dissociation of GS and glycogenin was provided

when, immediately following in vivo muscle glycogen degradation, free GS and

glycogenin were found with no associated glycogen, and only after extended

incubation did GS and glycogenin re-associate (Figure 1.10; Smythe et al.,

1990). This model of glycogen synthesis resembles the “glycogen initiator

synthase” pathway originally proposed by Krisman and Barengo (1975).

The domain of glycogenin that interacts directly with GS has been identified as

its COOH-terminal end (Roach & Skurat, 1997; Skurat et al., 2006). However,

recently it has been reported in muscle that a single glycogen particle can be

associated with more than one molecule of GS (Prats et al., 2009). This may

indicate that at the initiation of the granule, only one GS molecule is present and

further GS molecules are recruited as the glycogen grows (Graham et al.,

2010).

27

Figure 1.10 The complex formed between glycogenin and GS during the

initiation of glycogen synthesis as proposed by Smythe and

colleagues. Figure adapted from Smythe and Cohen (1991).

28

Currently, the regulation of glycogenin in vivo is not fully understood (Graham et

al., 2010). As glycogenin is found at the core of all glycogen granules in skeletal

muscle and no free, non-glucosylated cellular glycogenin exists under normal

conditions, it has been suggested that there exists a 1:1 ratio between

glycogenin and the number of glycogen β-particles (Smythe et al., 1988;

Tagliabracci et al., 2008). If glycogen is sterically limited to 12 tiers of glucose,

an increase in glycogenin concentration would be expected to lead to a

corresponding increase in cellular glycogen concentration. Studies involving the

over expression of glycogenin in varying tissues have, however, failed to show

any meaningful increase in glycogen storage, signifying that the absolute

amount of glycogenin is not limiting total glycogen accumulation (Hansen et al.,

2000; Shearer et al., 2005a; Wilson et al., 2007).

The discovery of a novel family of glycogenin interacting proteins (GNIP) that

can form a complex with glycogenin in vitro, may have provided a possible

means of glycogenin regulation (Skurat et al., 2002). Currently four iso-forms of

GNIP have been identified and reported to be highly expressed in skeletal

muscle, but also, to a lesser extent in the liver, heart and pancreas (Zhai et al.,

2004).

In liver, however, glycogenin accounts for only 0.0025% of glycogen by mass,

200 times less than the glycogenin content of muscle glycogen (Smythe et al.,

1989). This difference in glycogenin concentration was originally attributed to

the much larger mass of the glycogen α-particles of the liver, suggesting that

only one glycogenin molecule may be associated with each α-particle (Smythe

et al., 1989). However, the differences in structure of muscle and liver glycogen

in humans may be due to a second form of glycogenin, glycogenin-2, expressed

29

mainly in human liver, heart and pancreas (Mu et al., 1997; Mu & Roach, 1998).

Glycogenin-2 and muscle glycogenin (glycogenin-1) are 70% identical for the

sites containing the self-glucosylation and catalytic functions and both exhibit

similar properties (Roach et al., 1998). Although the cause of liver glycogens

distinct structure has not been fully uncovered (Sullivan et al., 2010), the

existence of a tissue specific isoform of glycogenin suggests that the regulation

of the initiation of glycogen synthesis in the liver may be fundamentally different

to that of muscles (Roach, 2002).

1.6 Glycosome

Since the discovery of glycogen, numerous studies have demonstrated that

glycogen is always in an active state of turnover, constantly storing and

releasing glucose (Stetten & Stetten, 1960). Scott and Still (1968), when

analysing the state of glycogen in leukocytes, concluded that “particulate or

native glycogen as visualized in the cell is not a molecule in the ordinary static

sense, but a dynamic organelle” which they referred to as glycosome (Scott &

Still, 1968). The idea of glycogen existing as a distinct organelle where

glycogen is associated with its own regulatory enzymes is now the generally

accepted view of glycogen structure (Rybicka, 1996).

The β-particle of glycogen has long been known to be intimately associated with

protein in the cell (Roach, 2002). When Bernard (1957) originally extracted and

identified glycogen, he noted its association with proteins, and proposed that a

proportion of this protein comprised the enzymes responsible for glycogen’s

post-mortem degradation (Bernard, 1857). Similarly, while investigating the high

molecular weight, particulate glycogen in liver, Lazarow (1942) found a small

30

amount of protein associated with each glycogen particle, and despite its small

quantity, suggested that this association was highly important to glycogen’s

particulate state, as conditions that dispersed the glycogen particulates

markedly altered the associated proteins (Lazarow, 1942). However, the

concept of glycogen existing as a proteoglucan in vivo was still considered

controversial, with some suggesting that it was merely an artefact of extraction,

with glycogen being contaminated with proteins (Manners, 1957). Conclusive

evidence that this was not the case was published in the 1960’s, when GS was

shown to be bound to glycogen as part of an “enzyme-substrate complex”

(Leloir & Goldemberg, 1960) in such a way as to remain attached following

partial α-amylase digestion of the glycogen (Luck, 1961).

Shortly after, GP was also found to be reversibly bound to liver glycogen, with

its association mediated by total glycogen concentration (Tata, 1964). Then,

Meyer and colleagues (1970) demonstrated the association of GP, PhK and

phosphorylase phosphatase with glycogen in skeletal muscle, and confirmed

the association of GS, concluding that this protein-glycogen complex is a

specific functional unit with distinct structural and enzymatic characteristics

(Meyer et al., 1970). Later, debranching enzyme was also identified as part of

the protein-glycogen complex (Nelson et al., 1972). In agreement with these

findings, Cohen and colleagues (1975) analysed the protein-glycogen complex

with acrylamide gel electrophoresis, identifying five major protein bands, namely

GP, PhK α- and β-subunits, debranching enzyme and GS (Taylor et al., 1975).

Since then, other regulatory and structural proteins have been reported to be

bound to the glycosome, including its glycogenin core as well as PP1 and its

subunits (Cohen, 1978; Roach & Skurat, 1997; Roach et al., 1998). AMPK, the

31

key kinase responsible for inactivating GS, also possesses a glycogen-binding

domain (Polekhina et al., 2005), locating AMPK to the glycogen granule and

associated substrates (Figure 1.11; Polekhina et al., 2003; McBride et al.,

2009).

Figure 1.11 Diagrammatic representation of the muscle glycosome with

its associated proteins. Figure modified from Shearer and

Graham (2004).

Four subunits of PP1 are known to associate with the glycosome, including

protein targeting to glycogen (PTG), expressed mainly in insulin-sensitive

tissues such as muscle, liver and adipose tissue; muscle glycogen-binding

regulatory subunit (GM), specifically expressed in skeletal muscle; the liver form

of GM (GL), found mostly in liver; and PPP1R6, expressed in a wide variety of

tissue but mainly in skeletal muscle and heart, act as molecular scaffolding for

the glycosome (Armstrong et al., 1997; Newgard et al., 2000; Lerin et al., 2003).

GS, although complexed with glycogenin at the initiation of the glycosome, also

interacts directly with PTG, localising GS to both glycogen and PP1, facilitating

32

activation via dephosphorylation (Fong et al., 2000). PTG also forms complexes

with the PP1 substrates GP and PhK, effectively anchoring the primary

enzymes of glycogen metabolism to the glycosome (Printen et al., 1997; Fong

et al., 2000). The skeletal muscle-specific subunit, GM, is responsible for

targeting both glycogen and PP1 to the sarcoplasmic reticulum (Hubbard et al.,

1990). Similarly, in the liver, GL binds PP1 to glycogen promoting

dephosphorylation of glycogen associated substrates, but the GL subunit does

not possess a membrane targeting domain (Bollen et al., 1998). PPP1R6 is also

able to bind PP1 and glycogen; however does not associate with cellular

membranes and is therefore likely to be restricted to the regulation of bound

PP1 (Armstrong et al., 1997). Despite similar functions, no two of the four

identified PP1 subunits have more than 50% of their sequence in common

(Newgard et al., 2000), allowing for glycosome-specific regulation of PP1 in

response to extracellular signals and intracellular changes in metabolites (Gasa

et al., 2000; Yang et al., 2002; Yang & Newgard, 2003).

Recently, two more important proteins have been found to be associated with

the glycosome, laforin and malin (Graham, 2009). Laforin is a dual-specificity

protein phosphatase with a carbohydrate binding domain that is directly

targeted to glycogen (Wang et al., 2002). As well as its carbohydrate binding

properties, laforin is involved in many protein-protein interactions, including

binding directly with PTG (Fernandez-Sanchez et al., 2003). Interestingly,

laforin does not dephosphorylate any of the proteins involved in glycogen

metabolism, but will remove the phosphate bound to complex polysaccharides

(Worby et al., 2006). Tagliabracci and colleagues (2007) demonstrated that

laforin, in cooperation with debranching enzyme, releases the bound phosphate

33

from skeletal muscle glycogen in vitro. In fact, mutation of the laforin gene leads

to a 4-fold increase in the amount of covalently bound phosphate within the

muscle glycogen granule (Tagliabracci et al., 2007; Tagliabracci et al., 2008).

Malin is an ubiquitin ligase that interacts with and polyubiquitinates a number of

glycogen associated proteins (Gentry et al., 2005). Malin, when over-expressed

together with laforin, completely prevents PTG-induced glycogen accumulation

via laforin-dependent ubiquitination (Worby et al., 2008). As malin is similarly

able to ubiquitinate debranching enzyme and laforin itself, therefore targeting

them for protease degradation, malin together with laforin has been proposed to

be a regulator of glycogen metabolism (Solaz-Fuster et al., 2008; Worby et al.,

2008).

Importantly, both the direct interaction of the glycosomal proteins with the

associated glycogen granule and the specific protein-protein interactions of the

glycosomal proteins allows the discrete metabolic regulation of individual

glycosomes (Graham, 2009). For instance, incubation of glycogenin with a 1:1

molar ratio of the GNIP iso-form, GNIP2, causes a marked increase in the

incorporation of UDG-glucose into glycogenin, indicating that GNIP2 activates

glycogenin self-glucosylation (Skurat et al., 2002). Structural analysis of GNIP

suggests possible binding of GS, possibly regulating the interaction of GS and

glycogenin (Zhai et al., 2004). Should this regulatory role exists, GNIP may be

the “regulatory factor” proposed by Smythe and colleagues (1990) to explain the

slow rate of association of glycogenin to GS in vitro compared to that reported

in vivo (Smythe et al., 1990; Zhai et al., 2004).

34

Another example of protein-protein interaction affecting the activity of

glycosomal proteins is each of the subunits of PP1 which has a distinct effect on

the phosphatase activity of PP1 against its glycogen associated substrates,

particularly GS and GP, in response to glycogenic and glycogenolytic signals

(Newgard et al., 2000; Brady & Saltiel, 2001; Toole & Cohen, 2007). Also, the

activity level of AMPK is also altered by its direct interaction with the glycogen

granule, with potent inhibition of AMPK when glycogen structure approaches

that of the GP limit dextrin (McBride et al., 2009).

The combined activation levels of PP1 and AMPK are strongly associated with

the phosphorylation and therefore activation level of GS (Roach, 2002). The

phosphorylation-dependent cellular location of GS provides further evidence for

the protein mediated targeting of glycosomes, with each sub-cellular glycogen

pool associated with a distinct phosphorylated GS form (Prats et al., 2005; Prats

et al., 2009), possible explaining, at least in part, the selective utilisation of

different glycogen pools in muscle (Marchand et al., 2007; Nielsen et al., 2009;

Prats et al., 2009).

1.7 Cellular distribution of glycogen

Although glycogen is distributed throughout the cell, it does tend to concentrate

near specific structures (Rybicka, 1996; Garcia-Rocha et al., 2001; Marchand et

al., 2002). This localisation allows hormonal and other signals to specifically

target relevant glycosomes (Gasa et al., 2000; Yang et al., 2002; Yang &

Newgard, 2003; Marchand et al., 2007; Graham et al., 2010). For instance,

three distinct pools of glycogen have been identified in skeletal muscles, with

glycogen found beneath the sarcolemma in the sub-sarcolemmal space, the

35

intermyofibrillar region between the myofibrils in close association with the

sarcoplasmic reticulum and mitochondria, and the intramyofibrillar region

located within the myofibrils (Rybicka, 1996; Marchand et al., 2002; Marchand

et al., 2007). At rest, the majority of the cellular glycogen is stored in the

myofibrillar (inter and intra) regions (Marchand et al., 2002), with approximately

three quarters of this glycogen pool found in the intermyofibrillar compartment

associated with the sarcoplasmic reticulum and mitochondria (Nielsen et al.,

2009).

The glycogen is anchored in these sub-cellular locations via proteins of the

glycosome interacting with specific cellular structures (Newgard et al., 2000).

Currently, three glycosomal proteins are known to bind and locate glycogen. As

discussed above, attachment of glycogen to the sarcoplasmic reticulum is via

the muscle specific subunit of PP1, GM, via a hydrophobic COOH-terminal

sequence (Tang et al., 1991; Newgard et al., 2000; Lerin et al., 2003).

Glycogenin, in its glucose-free state, binds directly with the actin cytoskeleton in

vitro, and its location does not changed upon incubation with glucose (Baqué et

al., 1997). Finally, the recently discovered glycogenin binding protein, GNIP,

binds with the sub-sarcolemmal scaffolding protein, desmin (Skurat et al.,

2002). This suggests that the GNIP-glycogenin complex may be specifically

located at the beginning of glycogen synthesis (Zhai et al., 2004).

In response to exercise, different glycogen pools have been shown to be

degraded to an extent that is affected by their sub-cellular location, suggesting

that the different glycogen pools have different roles in muscle contraction

(Marchand et al., 2007; Nielsen et al., 2009). Marchand and colleagues (2007)

reported that, following prolonged exercise, glycogen was preferentially

36

depleted from the myofibrillar over the sub-sarcolemmal region, with

intramyofibrillar glycogen preferred over intermyofibrillar (Marchand et al.,

2007). This is supported by Nielsen and colleagues (2009) who examined the

glycogen content and sub-cellular location in skinned muscle fibres immediately

after exhaustion, reporting that intramyofibrillar and intermyofibrillar glycogen

had been reduced to 7% and 23% of the control fibres. The authors further

speculated that glycogen sub-cellular location may be of greater importance

than total glycogen content (Nielsen et al., 2009).

Although it is still unclear if glycogen granules are able to translocate within the

muscle cell, the individual proteins that associate with the glycosome are known

to translocate in response to physiological stimuli (Graham et al., 2010). In

response to decreasing glycogen levels, GS translocates from the membrane

cellular fraction to the cytoskeleton (Nielsen et al., 2001), and this is

accompanied by actin cytoskeleton remodelling (Prats et al., 2009). As

glycogenin and GNIP are also located at the actin cytoskeleton (Skurat et al.,

2002), this suggests the interaction of GS and actin may coordinate glycogen

re-synthesis (Jurczak et al., 2008). Presumably, GS is moving between

glycosomes of different cellular locations as it is current (Graham, 2009), as

multiple GS molecules have been reported to associate with a single glycogen

β-particle (Prats et al., 2009). The physiological and metabolic importance of

these changes in glycogens sub-cellular compartmentalisation is currently not

fully understood.

In hepatocytes, however, glycogen’s cellular location is very much dependent

on total glycogen concentration, with glycogen molecules being orderly located

as they are synthesised and degraded (Garcia-Rocha et al., 2001; Fernández-

37

Novell et al., 2002; Ferrer et al., 2003; Ros et al., 2009). Guinovart and

colleagues (1997; 2001) demonstrated that, in vivo, GS translocates in

response to glucose, from the cytosol to the actin-rich cortex at the cell

periphery (Fernández-Novell et al., 1997; Garcia-Rocha et al., 2001). Glycogen

synthesis initially occurs only at the periphery of the hepatocyte. However, as

glycogen concentration increases, glycogen particles move progressively

towards the centre of the cell accompanied by GS, enabling glycogen synthesis

to continue at internal sites of the hepatocyte, in addition to the cell cortex

(Fernández-Novell et al., 2002). Glycogen degradation in the liver is also an

ordered process with glycogen granules located in the cytosol degraded

preferentially, in such a way that at low glycogen concentrations the remaining

glycogen granules are located near the cell cortex (Fernández-Novell et al.,

2002).

In both the liver and muscle, glycogen is also found in lysosomes, which are

cellular organelles derived from endosomes and autophagic vesicles, the latter

of which being responsible for the digestion of intracellular materials (Bechet et

al., 2005). Each lysosome possesses a glycogen breakdown pathway which is

required to hydrolyse any glycogen caught within the lysosome after cellular

autophagy because neither cytosolic GP nor debranching enzyme are able to

digest glycogen within the lysosome (Geddes, 1986; Calder & Geddes, 1989b).

In liver, at least 10% of cellular glycogen is found in the lysosomal compartment

(Geddes & Stratton, 1977), with at least as much as 6% of total muscle

glycogen entrapped in lysosomes (Calder & Geddes, 1989a).

The glycogen associated with lysosomes has been reported to have a higher

molecular weight than that in the cytosol, with molecular weights exceeding

38

4x105 kDa (Geddes, 1986). This has been attributed to lysosomal glycogen

being associated with almost double the amount of protein as the cytosolic

glycogen, causing many lysosomal β-glycogen particles to bind together via

protein association (Calder & Geddes, 1986, 1989a). The lysosomal hydrolysis

of glycogen is carried out by α-1,4-glucosidase, with the glucose thus formed

then free to leave the lysosome and enter the cytosol (Geddes, 1986).

1.8 Acid-soluble and acid-insoluble glycogen

As early as 1934, when so little was known about glycogen structure, it was

discovered that intramuscular glycogen exists as two distinct fractions differing

in their acid solubility’s (Willstatter & Rohdewald, 1934). When skeletal muscle

is homogenised in the presence of acid, a portion of the glycogen precipitates

together with proteins and for this reason was referred to as lyoglycogen or

acid-insoluble glycogen (AIG), whereas the glycogen that remains in solution

was known as desmoglycogen or acid-soluble glycogen (ASG; (Willstatter &

Rohdewald, 1934). Given the existence of these two pools of glycogen, their

responses to a range of physiological conditions were the subject of several

studies during the 1950’s and 60’s using rats and rabbits as experimental

models (Willstatter & Rohdewald, 1934; Bloom & Knowlton, 1953; Bloom &

Russell, 1955; Russell & Bloom, 1955, 1956; Kits van Heijningen, 1957; Stetten

et al., 1958). These studies showed that, under basal conditions, approximately

55% of total muscle glycogen in the rat exists as ASG (Bloom et al., 1951;

Bloom & Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1955) and

that this fraction is highly responsive to changes in total glycogen. For instance,

the stimulation of glycogenolysis via subcutaneous injections of epinephrine

was accompanied by a fall in ASG levels that accounted for the decrease in

39

total glycogen (Bloom & Russell, 1955). Similar results were reported in skeletal

muscles subjected to electro-stimulation to deplete their glycogen stores (Bloom

& Knowlton, 1953). When total glycogen levels are at their lowest levels,

however, no ASG can be extracted from tissues (Bloom et al., 1951; Bloom &

Russell, 1955). Finally, under conditions favourable to glycogen synthesis, most

of the increase in total glycogen is accounted for by a rise in ASG levels (Bloom

& Russell, 1955; Russell & Bloom, 1956; Kits van Heijningen, 1957).

At the start of the 60’s, evidence was provided that AIG was an artefact of

glycogen extraction. Roe and colleagues (1961) reported that it was possible to

acid extract all muscle glycogen as ASG provided that the extraction conditions

were harsh enough. This brought an almost 30-year halt to the research in this

field as their findings were taken as evidence that AIG was an experimental

artefact. However, what has been overlooked is that the extraction protocol of

Roe and colleagues (1961) resulted in a significant fall in total glycogen yield

compared to that measured in crude homogenate, with this difference in

glycogen yield almost completely being accounted for by the fall in AIG, thus

suggesting poor recovery of this fraction. Secondly, their results failed to explain

the different physiological responses attributed to the two pools of glycogen.

Although the aforementioned findings suggest that ASG represents the most

physiologically active fraction of glycogen (Bloom et al., 1951; Bloom &

Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1956; Kits van

Heijningen, 1957), isotopic labelling experiments performed in a variety of

tissues reveal that AIG most readily incorporates new glucose residues, and

hence is a highly metabolically active fraction (Stetten et al., 1958; Stetten &

40

Stetten Jr, 1958; Krisman, 1973; Krisman & Barengo, 1975; Curtino et al., 1979;

Aon & Curtino, 1984; Lacoste et al., 1990; Huang et al., 1997).

Further examination into the time course of glucose incorporation into the AIG

and ASG fractions pointed to a possible precursor-product relationship between

the two glycogen fractions. Working with bovine retina, Curtino and colleagues

(1979; 1984; 1990) reported that when retina membrane was incubated with low

concentrations of UDP-[14C]glucose, radioactivity incorporation was seen almost

completely in the AIG fraction before reaching a plateau as available UDP-

[14C]glucose was exhausted, with further incubation with unlabeled glucose

causing the transfer of the labelled glucose to the acid-soluble fraction of

glycogen (Curtino et al., 1979; Aon & Curtino, 1984; Lacoste et al., 1990). At

higher concentrations of UDP-[14C]glucose, radioactivity was initially

incorporated into the AIG fraction, but despite an excess of UDP-[14C]glucose,

the AIG incorporation of radioactivity still plateaued. This coincided with a

marked increase in the rate of radioactivity incorporation into ASG which

continued to rise, surpassing that of AIG (Curtino et al., 1979; Lacoste et al.,

1990).

Further evidence that AIG is a metabolically active pool of glycogen is illustrated

by the work of Huang and colleagues (1997) who reported different rates of [3-

3H]glucose incorporation into skeletal muscle AIG and ASG in rats administered

insulin in vivo. They reported that at lower rates of insulin infusion, radioactivity

was incorporated exclusively into the AIG fraction. As insulin infusion rates

increased, so did the level of incorporation of radioactivity into both the AIG and

ASG fractions (Huang et al., 1997). Interestingly, despite the absolute increase

in radioactivity of the AIG fraction, the concentration of AIG in the muscle

41

remained constant (Huang et al., 1997). However, despite these important

findings the mechanisms underlying the different acid-solubility of AIG and ASG

remained for several years without an answer.

One unlikely explanation to account for the different responses of AIG and ASG

to changes in muscle glycogen levels is that AIG correspond to the pool of

lysosomal glycogen. In support of this view, the proportion AIG originally

reported by Lomako and colleagues (1991a; 1991b) was similar to that reported

for lysosomal glycogen (Geddes & Chow, 1994). Moreover, the levels of

lysosomal glycogen in skeletal muscles are also largely unaffected by rapid

changes in muscle glycogen levels (Geddes & Chow, 1994) as is the case for

AIG (Bloom et al., 1951; Bloom & Knowlton, 1953; Bloom & Russell, 1955;

Russell & Bloom, 1955; Kits van Heijningen, 1957). However, the possibility that

AIG corresponds to lysosomal glycogen is challenged by the proportion of AIG

reported by most studies (> 40%) generally exceeding by far that of lysosomal

glycogen (Calder & Geddes, 1989a). More importantly, the pool of AIG and not

that of lysosomal glycogen is highly metabolically active as discussed above,

particularly when muscle glycogen levels are changing (Curtino et al., 1979;

Aon & Curtino, 1984; Lacoste et al., 1990; Huang et al., 1997).

Another mechanism proposed to explain the acid-insolubility of AIG is that it

may correspond to a sub-fraction of glycosomes tethered to the cell membranes

or cytoskeleton via an acid resistant bond and therefore pellets with the cell

debris. Many proteins associated with the glycosome are able to bind with

cellular structures and act as molecular scaffolding to allow glycogen

metabolism occur (Newgard et al., 2000; Cid et al., 2005; Prats et al., 2009).

This hypothesis is further supported by the observation that incubation of the

42

AIG pellet, from rat liver and bovine retina membrane, with the non-specific

protease, pronase, renders AIG soluble in acid (Krisman, 1972, 1973; Curtino et

al., 1979; Aon & Curtino, 1984, 1985; Curtino & Lacoste, 2000). If this

explanation were to hold, the AIG and ASG fractions could thus represent

physiologically distinct pools of glycogen.

It was in the 1990’s that the differences in acid solubility between ASG and AIG

were alleged to have been explained at the molecular level (Lomako et al.,

1991a; Lomako et al., 1991b; Lomako et al., 1993a). Lomako and colleagues

(1991a; 1991b; 1993a) showed that glycogen does not exist as a continuum of

molecular sizes from glycogenin to mature glycogen. With the help of gel

electrophoresis, they found that AIG is comprised mainly of low molecular

weight glycogen particles of approximately 400 kDa, which they named

proglycogen (PG), whereas ASG was referred to as macroglycogen (MG). The

differences in acid solubility between PG and MG were explained by their

different protein to glucosyl residue ratios. Since each glycogen granule is

covalently bound to a glycogenin core, smaller glycogen molecules have a

higher protein to glucosyl residue ratio, thus are less acid-soluble due to the

poor solubility of proteins such as glycogenin in acid. Hence, the 400 kDa PG

species, consisting of about 10% protein, are insoluble in acid (Lomako et al.,

1991b; Lomako et al., 1993a). In contrast, due to its larger size of 10 000 kDa,

MG is acid soluble, as it has a low protein-to-glucosyl ratio, with as little as

0.35% protein (Lomako et al., 1991b; Lomako et al., 1993a). Since Lomako and

colleagues (1993a; 1995) were also unable to detect any glycogen molecules

with a molecular weight less than 400 kDa in fresh muscle tissue from fed

rabbits, they concluding that PG is not simply the acid-insoluble fraction of a

43

continuum of glycogen sizes, but a distinct type of glycogen (Lomako et al.,

1993a; Alonso et al., 1995).

Given the small size of PG, Lomako and colleagues undertook to examine

whether PG is an intermediate in the synthesis of mature glycogen from

glycogenin. Using brain astrocytes, Lomako and colleagues (1991b; 1993a)

provided evidence that PG is an intermediate between glycogenin and fully

mature glycogen. Indeed, their work showed that the accumulation of PG

precedes the appearance of MG, with a proportional decrease in PG

accompanying an increase in MG (Lomako et al., 1991b; Lomako et al., 1993a).

They also concluded that PG is the rate limiting step in glycogen synthesis

(Lomako et al., 1991a; Lomako et al., 1993a; Lomako et al., 1995). However,

what remained unclear from their work were the mechanisms whereby only a

portion of PG is fully converted to MG, as full conversion could greatly increase

glycogen stores.

The existence of a precursor-product relationship between PG and MG had the

effect of stimulating a renewed interest in the physiology of AIG and ASG.

Indeed, Adamo and Graham (1998) published a method, based on that

developed by Jansson (1981), for separating AIG and ASG from small samples

of human muscle, and they referred to these fractions as PG and MG,

respectively (Adamo & Graham, 1998). The Adamo and Graham (1998)

protocol differs from earlier ones in that it does not involve a homogenisation

step. Briefly, small pieces of freeze-dried muscles are incubated in a glass tube

containing perchloric acid and then they are pressed against the wall of the tube

with a plastic rod and left to stand for 20 minutes before being centrifuged to

separate AIG from ASG (Adamo & Graham, 1998). Using this protocol, they

44

reported that when muscle glycogen concentrations are low, ASG accounts for

only 6-10% of total glycogen in human skeletal muscle, but contributes to

approximately 40% under conditions of elevated glycogen concentrations

(Adamo & Graham, 1998).

In order to examine how PG and MG in skeletal muscle contribute to the

synthesis and breakdown of glycogen, the homogenisation-free extraction

protocol of Adamo and Graham (1998) has been adopted in several studies to

investigate the response of AIG and ASG to a range of physiological conditions

in humans. For instance, in response to exercise ranging from low to high

intensity aerobic work or repeated sprint exercise, AIG has been reported to be

the most responsive fraction as it accounts for most of the changes in total

glycogen (Adamo et al., 1998a; Adamo et al., 1998b; Asp et al., 1999; Derave

et al., 2000; Shearer et al., 2000; Graham et al., 2001; Shearer et al., 2001;

Rosenvold et al., 2003; Battram et al., 2004; Shearer et al., 2005a; Shearer et

al., 2005b; Wee et al., 2005; Devries et al., 2006; Marchand et al., 2007; Wilson

et al., 2007).

During recovery from exercise, the rise in AIG levels in skeletal muscle

accounts for most of the early increase in total glycogen concentration. No

significant change in ASG is evident during that time despite the consumption of

a carbohydrate rich diet (Adamo et al., 1998b; Derave et al., 2000). However,

after several hours of a high intake of carbohydrates, AIG synthesis is

eventually blunted, and the increases in ASG concentrations account to a large

extent for any further increase in total glycogen levels, but without a

concomitant fall in AIG concentrations (Adamo et al., 1998b; Derave et al.,

2000; Battram et al., 2004). It is noteworthy that ASG levels do not increase

45

until total glycogen concentration reaches approximately 250 mmol kg-1 dry

weight in human muscles (Adamo et al., 1998b; Battram et al., 2004). Under

extreme conditions, such as recovery from a marathon, AIG requires up to 48

hours to reach pre-exercise levels, whereas over 7 days are required for ASG to

reach pre-exercise levels (Asp et al., 1999). Although AIG is in general present

in excess of ASG, when total glycogen concentrations reach supranormal

levels, ASG contributes up to 40% of total glycogen (Adamo & Graham, 1998;

Derave et al., 2000).

On the basis of the aforementioned findings, it has been argued that AIG

represents an intermediate pool of glycogen that is made available for

immediate use and may be an important site for the regulation of muscle

glycogen metabolism. Under extreme conditions, ASG is also mobilised and

can contribute significantly to the fall in total glycogen (Asp et al., 1999; Graham

et al., 2001; Shearer et al., 2001). More specifically, the delay in both ASG

synthesis and contribution to the increase in total glycogen has been explained

under the PG/MG model of Lomako and colleagues (1991), with PG being

synthesised first before it is converted to MG. Interestingly, these differences in

the pattern of responses of ASG and AIG have led many to hold the view that

the synthesis of AIG and ASG are the object of different enzymatic regulation

(Adamo et al., 1998a; Asp et al., 1999; Prats et al., 2002; Battram et al., 2004).

The adequacy of the PG/MG model to explain the patterns of response of AIG

and ASG to changes in muscle glycogen levels has been challenged on the

grounds that the existence of a distinct, 400 kDa proglycogen species has

recently been questioned (Skurat et al., 1997; Roach, 2002; Katz, 2006; James

et al., 2008). Indeed, the original report of a discrete 400 kDa species may have

46

been an artefact caused by the inappropriate use of discontinuous gel

electrophoresis, as two-dimensional gel electrophoresis yielded a smooth

continuum of glycogen sizes (Skurat et al., 1997). Also, all the studies that have

examined the pattern of molecular size distribution of total glycogen have

reported that glycogen exists as a normally distributed continuum of glycogen

particle sizes (Drochmans, 1962; Scott & Still, 1968; Wanson & Drochmans,

1968; Meyer et al., 1970; Schmalbruch & Kamieniecka, 1974; Rybicka, 1981b,

a; Skurat et al., 1997; Marchand et al., 2002; Shearer & Graham, 2004;

Marchand et al., 2007; Ryu et al., 2009). For these reasons, the term

“proglycogen” is now being used to refer to the acid insoluble fraction of

glycogen believed to represent a sub-population of small glycogen particles

(Marchand et al., 2002; Marchand et al., 2007). It is noteworthy, however, that

none of the recent studies based on a homogenisation-free protocol to extract

AIG and ASG have analysed the molecular weights of AIG and ASG to

determine if they do indeed correspond to glycogen fractions of different

molecular sizes.

1.9 Homogenisation-free extraction of acid-soluble and acid-

insoluble glycogen: artefact of tissue extraction?

One major problem overlooked by almost all recent studies on ASG and AIG is

that their findings contradict those of earlier studies performed in the 1950’s as

well as the findings of Lomako and colleagues (1991). Indeed, as discussed

above, these earlier studies showed that most of muscle glycogen is generally

found as ASG and that this is the most responsive glycogen fraction to changes

in total glycogen concentration (Willstatter & Rohdewald, 1934; Bloom &

47

Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1955, 1956; Kits van

Heijningen, 1957; Stetten et al., 1958). Moreover, Lomako and colleagues

found that AIG accounts for only about 15% of total glycogen (Lomako et al.,

1991b), and that when total glycogen concentration increases, AIG levels

remain stable or fall, whereas ASG concentrations rise markedly (Lomako et al.,

1995). In contrast, all recent studies based on the homogenisation-free protocol

of Adamo and Graham (1998) or other homogenisation-free protocols

(Cheetham et al., 1986; Nevill et al., 1989; Gaitanos et al., 1993) have reported

that AIG is not only the major fraction of glycogen, but also the most responsive

to changes in total muscle glycogen, except when the amounts of stored

glycogen are elevated (Adamo & Graham, 1998; Adamo et al., 1998a; Asp et

al., 1999; Derave et al., 2000; Shearer et al., 2000; Graham et al., 2001;

Shearer et al., 2001; Brojer et al., 2002a; Marchand et al., 2002; Shearer &

Graham, 2002; Rosenvold et al., 2003; Battram et al., 2004; Shearer et al.,

2005a; Shearer et al., 2005b).

These discrepancies might be explained, in part, on the grounds that the

fraction of AIG produced using a homogenisation-free acid extraction protocol

might be heavily contaminated with ASG (James et al., 2008). Indeed, the main

problem with this extraction protocol is that in the absence of a homogenisation

step to disrupt extensively all muscle cells, some of the glycogen is likely to

precipitate not because of its acid insolubility, but simply because it is trapped

by the remnants of undisrupted muscle myofibrils that co-precipitate glycogen

during centrifugation, thus resulting in a gross overestimation of the proportion

of AIG. This could explain why glycogen appears to accumulate in the AIG

fraction during early synthesis. The increase in ASG seen as total glycogen

48

concentrations reach maximal levels may be attributed to higher concentration

enabling a larger fraction of the glycogen to avoid entrapment within the

myofibril remnants. In support of this interpretation, all earlier studies in the

1950’s and 60’s performed extensive homogenisation of their muscle samples

before centrifugation, and, as a result, reported that ASG rather than AIG is the

most abundant glycogen fraction (Bloom et al., 1951; Bloom & Knowlton, 1953;

Bloom & Russell, 1955; Russell & Bloom, 1955, 1956; Stetten et al., 1958).

However, no attempt has been made so far to compare the extraction yield of

ASG between both extraction protocols.

It is important to stress that all earlier studies were also performed on fresh

muscles from rabbits or rats but not human muscle; whereas most recent

studies have been performed on small freeze-dried muscle samples in humans.

It is possible, therefore, that the different patterns of ASG and AIG responses

between earlier and more recent studies are not the result of the inclusion or not

of a homogenisation step, but are the consequence of either the use of freeze-

dried as opposed to fresh tissues or interspecies differences. However, this

latter possibility is not supported by the literature, since the extraction of muscle

glycogen from rats using the Adamo and Graham (1998) protocol also results in

a large proportion of AIG (Adamo & Graham, 1998; Brojer et al., 2002b),

whereas ASG is the dominant fraction in homogenised muscles from rats

(James et al., 2008). Nevertheless, the possibility remains that different

extraction protocols may extract different glycogen populations with gentler

extraction protocols, such as when a homogenisation step is omitted, extracting

a physiologically significant labile glycogen species that is only loosely bound

within the cell.

49

1.10 Statement of the problem

Most of the recent studies on the physiology of ASG and AIG have adopted a

homogenisation-free extraction protocol to separate these two pools of

glycogen. Given the evidence that the protocol adopted to acid-extract glycogen

might affect AIG and ASG responses to changes in muscle glycogen levels, it

was the first objective of this thesis (Chapter 2) to determine whether this is the

case in response to exercise and re-feeding in humans. We hypothesised that

the pattern of change in AIG and ASG levels is highly sensitive to the protocol

of glycogen extraction, with ASG being the most responsive fraction when a

homogenisation step is included, but AIG when glycogen is extracted without a

homogenisation step. Given the current view that the acid-solubility of glycogen

is determined by its size, with AIG corresponding to a glycogen population of

low molecular weight known as PG, our next objectives (Chapter 3) were to

develop a protocol to extract AIG and to compare the molecular sizes of AIG

and ASG extracted from rat muscles using a homogenisation-free and

homogenisation-dependent protocol. In agreement with the PG/MG model, we

hypothesised that AIG has a much smaller average molecular size than ASG

when homogenisation is performed to extract glycogen, but not when

homogenisation is omitted because of the expected heavy contamination of AIG

with ASG. Finally, although the responses of AIG and ASG to changes in

muscle glycogen levels have been extensively examined, the extent to which

the molecular sizes of AIG and ASG respond to changes in glycogen levels still

remains to be determined. Given published evidence that ASG accounts for

most of the changes in muscle glycogen levels, our third and last objective

50

(Chapter 4) was to test the hypothesis that the average molecular size of ASG

compared that that of AIG is the most responsive to exercise and re-feeding.

By improving our understanding of the molecular mechanisms underlying the

different responses of AIG and ASG to changes in muscle glycogen levels, it is

expected that our work should stimulate future research on this aspect of

glycogen biochemistry given the possibility that the behaviour of these glycogen

fractions might reflect an important but poorly understood aspect of glycogen

metabolism in health and disease.

51

Chapter 2

Homogenisation-dependent responses

of acid-soluble and acid-insoluble glycogen to exercise

and re-feeding in human muscles

52

2.1 Introduction

Since early last century, glycogen in skeletal muscle has been shown to exist as

two distinct fractions on the basis of its solubility in acid, namely acid-soluble

(ASG) and acid-insoluble (AIG) glycogen (Willstatter & Rohdewald, 1934). In

the 1950’s, these two types of glycogen were the subject of several studies

which found that ASG accounts for at least 40% of total glycogen at rest and

that it is the most responsive fraction to changes in total glycogen concentration

(Bloom & Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1955,

1956; Kits van Heijningen, 1957; Stetten et al., 1958). Then, in the early 1990’s,

Lomako and colleagues (1991a; 1993a) reported that AIG particles are much

smaller than ASG, with both being covalently bound to a 37 kDa protein,

glycogenin, the gene and promoter structure of which were subsequently

characterized by our laboratory (Van Maanen et al., 1999a, b). It is the high

protein (glycogenin) to glucosyl ratio of the AIG particle that was then proposed

to be responsible for its low acid solubility (Lomako et al., 1991a; Lomako et al.,

1993a). Just as importantly, Lomako and colleagues (1993a) also provided

some evidence that AIG was an intermediate along the synthesis of ASG, and

for this reason referred to these fractions as PG and MG, respectively (Lomako

et al., 1993a). Not surprisingly, their findings were at the origin of a renewed

interest in the physiology of ASG and AIG, with several recent studies

examining the responses of these glycogen fractions to a range of physiological

conditions (Bogdanis et al., 1995; Bogdanis et al., 1996; Adamo et al., 1998b;

Bogdanis et al., 1998; Asp et al., 1999; Derave et al., 2000; Shearer et al.,

2000; Graham et al., 2001; Shearer et al., 2001; Brojer et al., 2002b; Marchand

et al., 2002; Shearer & Graham, 2002; Battram et al., 2004; Shearer & Graham,

53

2004; Shearer et al., 2005a; Shearer et al., 2005b; Marchand et al., 2007;

Wilson et al., 2007; James et al., 2008). In contrast to the earlier findings of the

1950’s, these recent studies and some earlier ones in humans have reported

that AIG is in general not only the major fraction of glycogen, but also the most

responsive to changes in total muscle glycogen levels, except when the

amounts of stored glycogen are elevated (Jansson, 1981; Cheetham et al.,

1986; Nevill et al., 1989; Gaitanos et al., 1993; Adamo et al., 1998b; Derave et

al., 2000; Hansen et al., 2000; Shearer et al., 2001; Battram et al., 2004;

Marchand et al., 2007).

It is important to stress that most recent studies on ASG and AIG have adopted

glycogen extraction protocols that do not include a homogenisation step

(Jansson, 1981; Cheetham et al., 1986; Nevill et al., 1989; Gaitanos et al.,

1993; Adamo & Graham, 1998; Adamo et al., 1998b; Derave et al., 2000;

Hansen et al., 2000; Shearer et al., 2001; Battram et al., 2004; Marchand et al.,

2007). For instance, in one of the most commonly used protocols, glycogen is

extracted by pressing with a plastic rod some freeze-dried muscle samples

submerged in perchloric acid (PCA). The extract is then centrifuged, with AIG

and ASG found in the pellet and supernatant, respectively (Jansson, 1981;

Adamo & Graham, 1998). Another homogenisation-free protocol uses

powdered freeze-dried muscle tissues for the acid-extraction of glycogen

(Cheetham et al., 1986; Nevill et al., 1989; Gaitanos et al., 1993; Bogdanis et

al., 1995; Bogdanis et al., 1996; Bogdanis et al., 1998). Recently, however, we

provided some evidence that some of the glycogen extracted from rat muscles

without a homogenisation step may precipitate in the presence of acid not

because of its acid insolubility per se, but because it is trapped by the remnants

54

of undisrupted muscle myofibrils with which it co-precipitates, thus causing a

marked overestimation of the proportion of AIG (James et al., 2008). In this

regard, it is interesting to note that the studies on ASG and AIG in the 1950’s

were preformed on homogenised muscle extracts and have consistently

reported higher proportions of ASG compared with AIG (Bloom et al., 1951;

Bloom & Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1955, 1956;

Stetten et al., 1958; Hultman, 1967). Moreover, recently we compared for the

first time the effect of homogenisation-free and -dependent protocols on the

acid extraction of glycogen in rats muscle and showed that ASG rather than AIG

is the most abundant glycogen species when extracted with a homogenisation-

dependent protocol (James et al., 2008). However, it is important to note that

ASG and AIG responses to changes in muscle glycogen concentrations were

not compared between extraction protocols not only in that study (James et al.,

2008), but also in all of the previous research on AIG and ASG.

Given that most of the studies that have examined the patterns of response of

AIG and ASG to changes in muscle glycogen levels in humans have been

based on homogenisation-free acid extraction protocols, it is unclear to what

extent the use of such protocols results in patterns of change in glycogen levels

that are similar to those obtained using a homogenisation-dependent protocol

because, as mentioned above, such a direct comparison has never been

performed before. Moreover, although we have compared the effect of these

different acid extraction protocols on the proportion of ASG and AIG in rat

muscles, one cannot assume that such a comparison in humans would yield

similar results given that glycogen levels in human muscles are 3- to 6-fold

higher than those in rats. Considering that most studies on the pattern of

55

change in ASG and AIG levels in humans have been concerned with the

breakdown and re-synthesis of muscle glycogen in response to exercise and re-

feeding (Bogdanis et al., 1995; Adamo et al., 1998b; Asp et al., 1999; Derave et

al., 2000; Graham et al., 2001; Battram et al., 2004), our goal was to determine

the extent to which a homogenisation-dependent protocol results in findings

similar to those obtained without homogenisation by comparing the effect of

exercise and re-feeding on ASG and AIG levels using the same muscle

samples. In so doing, this study re-examines the physiological significance of

the past studies on ASG and AIG.

56

2.2 Materials and methods

2.2.1 Materials

Calibration gases (ß special gas mixture) were purchased from BOC Gases

Australia Ltd, Australia. Polycose was purchased from Abbott Nutrition,

Columbus, Ohio. Anaesthetic was purchased from AstraZeneca, Australia.

2.2.2 Participants

Eight male participants from the student population of the University of Western

Australia volunteered for the study. All were made fully aware of the

experimental procedure before they gave full written consent in accordance with

University ethics policy. The descriptive characteristics of the participants were

as follows: age 20.0 ± 2.4 years, weight 84.7 ± 10.0 kg, V�O2 peak 58.3 ± 11.2 ml

kg–1 min–1. All participants were healthy, recreationally active non-smokers, and

were required to complete a physical activity readiness questionnaire (Thomas

et al., 1992) to ensure they were not currently on medication or receiving

treatment for any pre-existing medical condition or injury. This study was

approved by the Human Rights Committee of the University of Western

Australia and conformed to the Declaration of Helsinki.

2.2.3 Exercise and re-feeding protocol

Prior to testing, participants were subjected to a familiarisation session with

equipment and personnel. During this session, V�O2 peak was measured and

anthropometric data collected. No earlier than one week after this familiarisation

session, participants were required to attend the laboratory for the experimental

57

trial. On the day prior to testing, participants fasted overnight (minimum of 12

hours) and were also required to refrain from heavy physical activity, caffeine

and alcohol for the preceding 48 hours. On the day of testing, participants were

asked to perform a 5-minute warm up. A biopsy and blood samples were taken

before cycling for 1 hour at a workload corresponding to 70% V�O2 peak.

Immediately after exercise, a second biopsy and blood sample were taken.

Each participant was then asked to consume for 2 hours the equivalent of 0.6

grams of carbohydrate per kilogram of body mass per half hour by ingesting a

20% (w/v) maltodextrose solution (Polycose, Abbott Nutrition, Columbus, Ohio)

every 30 minutes for 2 hours. This intake of carbohydrate was chosen on the

basis that it has been shown to maximise the rate of muscle glycogen synthesis

post-exercise (van Loon et al., 2000). Following this initial 2-hour intake of

carbohydrates, a third biopsy and blood samples were taken, and the

participants sent home with a supply of carbohydrate and asked to restrict their

food intake to that provided by us. While at home, they were required to ingest a

total of 10 grams of carbohydrate per kilogram of body mass, mainly in the form

of maltodextrose, before the end of the day. The participants were also required

to keep a food record until the end of the experiment and to refrain from any

physical activity, caffeine or alcohol immediately after the testing session and

for the following 24 hours. Twenty four hours after the end of exercise protocol,

participants returned to the laboratory for a fourth muscle biopsy (Figure 2.1).

58

Figure 2.1 Experimental design of the study.

Exercise

Muscle Biopsies

0 h 24 h 4 incisions Pre 2 h

Recovery Pre-exercise

Warm-up Carbohydrate Consumption

59

2.2.4 Anthropometric data and ��O2 peak measurement

Prior to V�O2 peak testing, the body mass of each participant was determined. To

this end, each participant was asked to remove his shoes and any excess

clothing, and was weighed to the nearest 0.05 kg on an electronic scale. The

V�O2 peak of each participant was then determined on a front access cycle

ergometer, with the subject breathing through a mouthpiece connected to a

Hans-Rudolf valve which was attached to Collins tubing (inside diameter of 32

mm). All inspired and expired air passed through an on-line gas analysis system

comprised of a Morgan Ventilation Monitor, Ametek S-3A Oxygen Analyser and

CD 3A Carbon Dioxide Analyser. The Morgan ventilometer was calibrated by

using a one litre syringe to pump ten litres of air through the Hans-Rudolf valve.

The gas analysers were then calibrated with a gas of known composition (O2 =

16.09 %, CO2 = 4.19 %). Electrical signals from the analysers were continuously

integrated by a 286 personal computer, and values for V�O2 (ml min-1) and V�O2

(ml kg-1 min-1) were calculated every 15 seconds. Following the completion of

any test, the system was re-calibrated to adjust for any drift encountered during

the testing procedure.

2.2.5 Muscle biopsies

An area of skin approximately 20 cm in length and 10 cm in width was shaved

over the vastus lateralis muscle on both legs of each participant. A local

anaesthetic (1-2% Xylocaine, epinephrine free) was applied to the skin prior to

the incisions. Four incisions were performed, with the two incisions on each

vastus lateralis separated by 10 cm. The incisions were then closed up with

steri strips until the prescribed biopsy time. All biopsies were taken from the

60

mid-thigh level of the vastus lateralis using an improved version (Hennessey et

al., 1997) of the percutaneous needle biopsy technique developed by

Bergström (1962), with suction applied manually. Incisions were closed with

stitches after the biopsy. Each muscle biopsy was immediately freeze-clamped

in liquid nitrogen and stored at –80°C for the later enzymatic analysis of muscle

glycogen content.

2.2.6 Acid extraction of muscle glycogen

Acid extraction was preformed as previously described by James and

colleagues (2008) with only minor changes. Briefly, a previously freeze-dried

muscle biopsy sample was broken into small pieces in a mortar pre-cooled in

liquid nitrogen and then freeze-dried for 48 hours. Once dried, fat, blood and

any non-muscular connective tissue were dissected free from the muscle

sample. This sample was then placed in a pre-weighed 2 ml micro centrifuge

tube prior to being weighed to determine tissue sample mass. Small pieces (~1

mg each; 6 mg total) of freeze-dried muscle samples were mixed thoroughly,

and part of these samples were homogenised in the presence of ice-cooled 1.5

M PCA (200 µl per 3 mg of sample) in a 2 ml micro centrifuge tube using an IKA

Labortechnik T-8 homogeniser (Staufen, Germany). Then, the homogenate was

centrifuged at 2700 g for 10 minutes before the supernatant was removed and

the pellet re-suspended and re-homogenised with ice-cooled 1.5 M PCA (100 µl

per 3 mg of sample) in a 2 ml micro centrifuge tube. After another

centrifugation, the pellet was collected and supernatants were combined.

Other pieces of the same muscle were also extracted using the protocol

outlined in Adamo and Graham (1998). This protocol involved the freeze-drying

61

of small muscle pieces, which after being dissected free of visible blood and

connective tissue were placed in a glass tube in the presence of 1.5 M PCA.

The muscle samples were pressed against the tube with a plastic rod and left to

stand for 20 minutes, then centrifuged at 2700 g for 10 minutes before the

supernatant was removed.

2.2.7 Glycogen determination

The supernatants obtained above were vortexed before a 100 µl sample was

removed for the determination of ASG and a 200 µl sample for free-glucose

analysis. Then, 2 M hydrochloric acid was added to the pellet and supernatant

samples. Both samples were vortexed, and tube weights recorded. The tubes

were then placed in a 90°C water bath for 2 hours to hydrolyse glycogen into

glucose, with the tubes being vortexed after 1 hour to aid digestion. After

incubation, the samples were vortexed, and a 400 µl aliquot was removed and

neutralized by the addition of 2 M potassium carbonate. The resulting extracts

were assayed for glucosyl units and corrected for free glucose according to

Bergmeyer (1974). For the determination of total muscle glycogen, one aliquot

of uncentrifuged 1.5 M PCA muscle extracts prepared as described above was

incubated in the presence of 2 M hydrochloric acid to hydrolyse glycogen, and

another aliquot was used for the assay of free glucose. The resulting hydrolysed

extracts were assayed for glucosyl units and corrected for free glucose

according to Bergmeyer (1974). Finally, in some samples, glycogen levels

measured following acid hydrolysis were compared to glycogen determined

enzymatically as described in Adamo and Graham (1998).

62

2.2.8 Expression of results and treatment and analysis of data

All of the glycogen results are expressed as millimole glucosyl units per

kilogram dry weight tissue. The results obtained from the exercise/re-feeding

experiment were analyzed using a 2-way ANOVA with repeated measures with

time and treatments as independent variables followed by a Fisher LSD post

hoc test. All analyses were performed using SPSS (Chicago, IL) version 12 and

all data is presented as mean ± standard error of the mean.

63

2.3 Results

2.3.1 Glycogen yield of homogenisation-dependent and

independent protocols

There was a positive linear relationship (Figure 2.2; r = 0.97) and no significant

difference between total glycogen determined using our homogenisation-

dependent protocol and that determined using a well established

homogenisation-independent protocol (Adamo & Graham, 1998), the latter of

which being a protocol that has already been extensively validated against other

glycogen assay methods, including those based on enzymatic digestion of

glycogen (Jansson, 1981; Adamo & Graham, 1998). The coefficient of

variations (CV) for ASG, AIG and total glycogen determined as described here

were 4.5, 4.0 and 4.1%, respectively, and within the published range (Adamo &

Graham, 1998).

The proportion of ASG extracted by the homogenisation-free extraction protocol

adopted here was significantly lower than that achieved by our homogenisation-

dependent protocol (Figure 2.4 and Figure 2.5; p < 0.05). Although all glycogen

determinations were corrected for free glucose levels, these levels were less

than 1% of total glycogen as reported previously (Essen, 1978; Jansson, 1981),

with for instance resting free glucose levels of only 1.7 ± 0.6 and 2.4 ± 0.6 mmol

kg-1 d.w. for the homogenisation-free and homogenisation-dependent extraction

protocols, respectively.

64

Figure 2.2 A comparison of total glycogen in human muscle determined

using a homogenisation-free protocol and a homogenisation-

dependent protocol. The values shown are expressed in

millimoles glucosyl units per kilogram dry tissue weight. Solid line,

linear regression analysis: y = 0.994x + 4.276, r = 0.971, n = 18;

dashed line, line of identity with slope = 1.

0

200

400

600

800

0 200 400 600 800

Ad

am

o &

Gra

ha

m (

mm

ol

kg

-1d

.w.)

Homogenisation-dependent (mmol kg-1 d.w.)

65

2.3.2 Effect of exercise and re-feeding on ASG and AIG levels in

human muscles

In response to one hour of exercise at an average power output of 203.7 ± 12.9

W, total glycogen levels decreased significantly (Figure 2.3; p < 0.05). During

the first 2 hours of recovery, glycogen concentrations increased significantly (p

< 0.05), yet remained below pre-exercise levels (p < 0.05). After 24 hours of

recovery during which the participants ingested the equivalent of 10.9 ± 0.6 g

kg–1 of carbohydrate, total glycogen reached levels significantly higher than

those prior to exercise (Figure 2.3; p < 0.05).

The responses of AIG and ASG to exercise were significantly different between

the homogenisation-free protocol and the homogenisation-dependent protocols

(p < 0.05). In response to exercise, there was a fall in both ASG and AIG

concentrations extracted without a homogenisation step (p < 0.05), with AIG

accounting for most of the fall in total glycogen (Figure 2.4; p < 0.05). In

contrast, ASG determined using a homogenisation-dependent extraction

protocol accounted for the entire fall in total glycogen during the exercise bout

(p < 0.05), while the AIG fraction remained at stable and low levels (Figure 2.5;

p < 0.05).

66

Figure 2.3 Pattern of response of total muscle glycogen to exercise and

recovery. The values shown represent means ± S.E.M. (n = 8)

and are expressed in millimoles glucosyl units per kilogram dry

muscle. a, significantly different from pre-exercise (p < 0.05). b,

significantly different from 0 hour (p < 0.05). c, significantly

different from 2 hours (p < 0.05).

0

100

200

300

400

500

600

Pre 0 2 24

Gly

cog

en

(m

mo

l k

g–

1d

.w.)

Recovery time (h)

aa,b

a,b,c

67

During the first 2 hours of recovery, the responses of AIG and ASG to re-

feeding were different between the two extraction protocols. The AIG

determined using the homogenisation-free extraction protocol accounted for the

entire increase in total glycogen during the first 2 hours of recovery (p < 0.05),

while ASG remained at stable and low levels (Figure 2.4; p > 0.05). In contrast,

using the homogenisation-dependent extraction protocol, AIG remained at low

and stable levels during the first 2 hours of recovery (p > 0.05), whereas the

change in ASG levels accounted for the increase in total glycogen levels (Figure

2.5; p < 0.05).

During the 2 to 24 hour recovery period, the responses of AIG and ASG were

also affected by the protocol of glycogen extraction. The AIG and ASG

determined using the homogenisation-free extraction protocol contributed

significantly to the increases in total glycogen (Figure 2.4; p < 0.05). In contrast,

using the homogenisation-dependent protocol to extract glycogen, the rise in

ASG concentration accounted for all the increase in total glycogen

concentrations (p < 0.05), whereas the levels of AIG remained unchanged and

were not significantly different from either pre- or post-exercise levels (Figure

2.5; p > 0.05).

68

Figure 2.4 Effect of exercise and recovery on (A) the pattern of response

of ASG and AIG using a homogenisation-free protocol and

(B) changes in concentrations of ASG and AIG. The values

shown represent means ± S.E.M. (n = 8) and are expressed in

millimoles glucosyl units per kilogram dry tissue weight. a,

significantly different from pre-exercise (p < 0.05). b, significantly

different from 0 hour (p < 0.05). c, significantly different from 2

hours (p < 0.05). d, significantly different from corresponding

glycogen level at 0 to 2 hours (p < 0.05). e, significantly different

from AIG of same time interval (p < 0.05).

0

100

200

300

400

500

600

Pre 0 2 24

Gly

cog

en

(m

mo

l k

g–

1d

.w.)

Recovery time (h)

ASG

AIG

A

a

a

a

a,b

b,c

a,b,c

-200

-100

0

100

200

300

400

Gly

cog

en

(m

mo

l k

g–

1d

.w.)

Pre-0 0-2 2-24

Time intervals (h)

ASG

AIGe e

d

B

d

69

Figure 2.5 Effect of exercise and recovery on (A) the pattern of response

of ASG and AIG using our homogenisation-dependent

protocol and (B) changes in concentrations of ASG and AIG.

The values shown represent means ± S.E.M. (n = 8) and are

expressed in millimoles glucosyl units per kilogram dry tissue

weight. a, significantly different from pre-exercise (p < 0.05). b,

significantly different from 0 hour (p < 0.05). c, significantly

different from 2 hours (p < 0.05). d, significantly different from

ASG at 0 to 2 hours (p < 0.05). e, significantly different from AIG

of same time interval (p < 0.05).

0

100

200

300

400

500

600

Pre 0 2 24

Gly

cog

en

(m

mo

l k

g–

1d

.w.)

Recovery time (h)

ASG

AIG

A

a

a,b,c

b

-200

-100

0

100

200

300

400

Gly

cog

en

(m

mo

l k

g–

1d

.w.)

Pre-0 0-2 2-24

Time intervals (h)

ASG

AIGe

e

d,eB

70

2.4 Discussion

Almost 75 years ago, muscle glycogen extracted in the presence of acid was

shown to exist as ASG and AIG forms. In recent years, there has been a

considerable volume of research aimed at elucidating the physiological

significance and interrelationship of these two fractions of glycogen in skeletal

muscle, with AIG levels shown to be higher and more responsive than ASG to

change in glycogen levels, except when total muscle glycogen levels are

elevated. Unfortunately, these studies have adopted homogenisation-free

glycogen extraction protocols that might have resulted in the incomplete

extraction of ASG, thereby resulting in the contamination of AIG by ASG. Here,

for the first time, the effects of homogenisation-dependent and -independent

acid extraction protocols on the patterns of change in ASG and AIG levels in

human muscles were compared. We show that the use of a homogenisation-

free glycogen extraction protocol markedly underestimates the proportion of

ASG, and that with more thorough conditions of acid extraction, most of the

glycogen in human muscles is extracted as ASG rather than AIG. More

importantly, ASG levels in homogenised muscle extracts account for most of the

changes in total glycogen levels in response to exercise and re-feeding post-

exercise, but AIG when a homogenisation-free extraction protocol is adopted.

Altogether, these findings show that the pattern of change in ASG and AIG

levels in response to changes in total muscle glycogen concentrations is

dependent on whether muscles are homogenised to acid-extract glycogen and

raise the issue of the physiological significance of the many studies on ASG and

AIG.

71

Although the patterns of change in ASG and AIG levels with exercise and re-

feeding found here differ greatly between extraction protocols, they are

consistent with those reported in previous studies. In muscle samples extracted

using a homogenisation-free protocol; our results show that there is a significant

decrease in both ASG and AIG levels during exercise, with most of the fall in

total glycogen levels being accounted for by the decrease in AIG levels. During

the first 2 hours of recovery, the rise in AIG levels accounts for the increase in

total glycogen levels while ASG remains at stable levels. However, during the 2

to 24 hour period post-exercise, both ASG and AIG contribute to the increases

in total glycogen levels. These patterns of AIG and ASG responses to exercise

and recovery are similar to those reported in the studies based on a

homogenisation-free acid extraction protocol (Adamo et al., 1998b; Graham et

al., 2001; Shearer et al., 2001; Battram et al., 2004; Shearer et al., 2005a;

Shearer et al., 2005b; Marchand et al., 2007; Wilson et al., 2007). Also, we

show that when muscles are homogenised to extract glycogen, ASG rather than

AIG accounts for all the changes in total glycogen, with AIG remaining at stable

and low levels throughout both exercise and re-feeding. These latter findings

corroborate earlier work from this laboratory in starved-to-fed rats (James et al.,

2008) and those of other studies using homogenisation-dependent protocols to

acid-extract glycogen, where ASG levels have been reported to be the most

responsive to a wide range of conditions affecting muscle glycogen levels, such

as adrenaline administration, electro-stimulation and starvation (Bloom et al.,

1951; Bloom & Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1955,

1956; Stetten et al., 1958; James et al., 2008). It is noteworthy that although

muscle glycogen levels in humans are much higher (3-6 fold) than those

reported in rats (James et al., 2008), the concentrations of AIG extracted here

72

with a homogenisation-dependent protocol in humans are similar to those

measured previously in rats using a similar protocol (James et al., 2008).

However, the proportion of AIG is much lower in humans due to the large

excess of ASG which accounts for the large difference in total muscle glycogen

levels between humans and rats.

Except when muscle glycogen levels are elevated, the low proportion of ASG

obtained from glycogen extracted using a homogenisation-free protocol raises

the question of the factors explaining such a low and variable relative extraction

yield. If the only factor determining the extraction yield of ASG was the size of

the glycogen particle relative to glycogenin, as originally proposed by Lomako

and Lomako (1991a), the levels and proportions of ASG and AIG would not be

affected by the protocol of glycogen extraction. However, against this

interpretation is the compelling evidence that AIG is not a discrete species of

low molecular weight (Skurat et al., 1997; Skurat & Roach, 2004) and the recent

evidence that ASG and AIG have a similar molecular weight (James et al.,

2008). As mentioned in an earlier study in rats (James et al., 2008), it is

possible that the poor yield of ASG using a homogenisation-free acid extraction

protocol might be due to some of the glycogen precipitating not because of its

poor-acid solubility per se, but simply because it is trapped within the dense

mesh of undisrupted myofibrils that precipitate during centrifugation in the

presence of acid. This, in turn, could result in the contamination of AIG by ASG

and in a large overestimation of the proportion of AIG and underestimation of

ASG levels. When total muscle glycogen levels are low or moderate, extracting

glycogen without a homogenisation step would liberate only a small proportion

of the pool of ASG from the mesh of poorly disrupted muscle cells, with the

73

resulting AIG contaminated by ASG accounting for most of the change in total

muscle glycogen levels. However, when total glycogen increases to levels that

exceed the capacity of this mesh of muscle myofibrils to trap glycogen as

effectively, a disproportionate and marked rise in the release of ASG would be

expected to occur with an increase in glycogen content as reported here and

other studies (Adamo & Graham, 1998; Asp et al., 1999; Derave et al., 2000;

Battram et al., 2004; Marchand et al., 2007), with ASG appearing to contribute

substantially to the changes in total glycogen levels. Since our results show that

there are no marked changes in ASG levels when glycogen concentrations are

below 200 mmol kg–1 d.w., this suggests that the limit of the proposed capacity

of myofibrils to trap glycogen is somewhere between 200-400 mmol kg–1 d.w.

under our experimental conditions.

Although the above interpretation implies that the patterns of change in ASG

and AIG levels obtained using a homogenisation-free extraction protocol could

be the result of an artefact of tissue extraction, the results obtained using such a

protocol might still be highly physiologically significant. Indeed, this would be the

case if the ASG and AIG fractions thus obtained and their patterns of response

to changes in glycogen levels were to reflect the behaviours of distinct and

labile sub-populations of glycogen that are vulnerable to homogenisation-

dependent extraction. For instance, since each glycogen particle binds a

number of proteins including those involved in its synthesis and degradation to

form a complex known as glycosome (Rybicka, 1996), with some of these

glycosomes being associated with the sarcoplasmic reticulum (s.r.) and with

some proteins whose binding (e.g. glycogen synthase) is affected by factors

such as glycogen levels (Prats et al., 2005), it is possible that AIG and ASG

74

correspond to distinct protein/s.r.-associated glycosomes. The disruption of

these structures and associated fall in protein to glycogen ratio when muscles

are homogenised could result in an increase in the proportion of glycogen

extracted as ASG. Another possibility is that AIG and ASG extracted without

homogenisation may reflect, at least in part, glycogen from different locations as

suggested by the uneven distribution of glycogen between the sub-sarcolemmal

compartment, the intra- and inter-myofibrillar spaces and the newly discovered

intracellular cytoskeleton-associated compartment (Prats et al., 2005) where

glycogen differs in concentration and is metabolised at different rates

(Marchand et al., 2007).

One alternative and popular explanation to explain the unique patterns of

change in ASG and AIG levels when muscle glycogen levels are changing and

extracted without a homogenisation step is based on the PG/MG paradigm, with

PG and MG corresponding to AIG and ASG, respectively (Adamo et al., 1998b;

Derave et al., 2000; Hansen et al., 2000; Shearer et al., 2001; Battram et al.,

2004; Marchand et al., 2007). In response to exercise and re-feeding, PG has

been proposed to exist as a population of low molecular weight glycogen and to

be the most metabolically active glycogen species when total muscle glycogen

levels are low to moderate, with most changes in glycogen levels being

explained by the inter-conversion between high and low molecular weight PG.

In contrast, when muscle glycogen levels are elevated, the conversion of PG

into MG has been proposed to contribute to the increase in ASG and total

muscle glycogen levels (Adamo et al., 1998b; Asp et al., 1999; Battram et al.,

2004; Marchand et al., 2007).

75

Against this latter interpretation, however, are not only our findings suggesting

that AIG obtained using a homogenisation-free extraction protocol is

contaminated with ASG, but also the compelling evidence against the existence

of PG as a discrete species of glycogen. Skurat and colleagues (1997; 2004)

showed not only that total glycogen separated by electrophoresis exists as a

continuum covering a broad range of molecular weights, but also that the

discovery of PG by Lomako and colleagues was probably the result of an

artefact of their electrophoresis analysis (Skurat et al., 1997; Skurat & Roach,

2004). More recently, Marchand and colleagues (2002; 2007) reported using

transmission electron microscopy that the sizes of glycogen particles in resting

human muscle approximate a normal distribution. Although it has been

proposed that AIG might correspond to a population of low molecular weight

glycogen of varying sizes (Shearer & Graham, 2002, 2004), the absence of two

populations of glycogen with different average sizes (Marchand et al., 2002)

does not support this view. More importantly, our recent work using size

exclusion gel chromatography suggests that ASG and AIG have a similar

molecular weight (James et al., 2008). Clearly more work is required to explain

the factors determining the pattern of AIG and ASG extraction obtained under

conditions where muscles are not subjected to a homogenisation step in order

to assess the physiological significance of these glycogen fractions and explain

their responses to homogenisation.

Under conditions where muscle glycogen is extracted using a homogenisation-

dependent protocol, the low levels and absence of changes in AIG levels raise

the question of the factors underlying the behaviour of this glycogen pool. One

possibility is that AIG is a less metabolically active and responsive pool of

76

glycogen. In this respect, glycogen levels in the sub-sarcolemmal space has

been reported to be unresponsive to exercise of sub-maximal intensity (Friden

et al., 1985), thus suggesting that AIG corresponds to this glycogen fraction.

Against this interpretation, however, is the recent work of Marchand and

colleagues (2007) that showed that glycogen levels in this and other cellular

compartments in skeletal muscles are markedly depleted in response to

exercise of sub maximal intensity. Another possibility is that AIG corresponds,

at least in part, to the fraction of muscle glycogen entrapped inside lysosomes

(Calder & Geddes, 1989a). Indeed, the fact that lysosomal glycogen is not

metabolised by glycogen phosphorylase or synthase (Hers & Van Hoof, 1973)

might explain, in part, the absence of rapid changes in the levels of AIG in

response to exercise and re-feeding. The problem with this interpretation,

however, is that the work of Huang and colleague (1997) shows using

isotopically labelled glucose in rats that AIG obtained from homogenised

extracts is a highly metabolically active glycogen pool (Huang et al., 1997), thus

unlikely to represent lysosomal glycogen. Their findings also highlight the very

important point that the absence of net changes in AIG levels does not exclude

the possibility that the turnover rate of this glycogen fraction might be elevated,

with AIG synthesis and breakdown occurring at similar and high rates. Finally, it

is possible, as discussed previously, that AIG obtained after homogenisation

may represent a distinct metabolically active sub-fraction of glycosome with a

distinct complement of proteins and cellular location compared not only to ASG

but also AIG obtained without homogenisation.

In conclusion, this study shows that the levels and patterns of response of AIG

and ASG to changes in glycogen concentrations in human muscles are highly

77

dependent on the protocol used to acid-extract glycogen. Also, it highlights the

fact that although the findings of the many studies on ASG and AIG could be

physiologically meaningful, none of these studies including this one has

excluded the possibility that their reported patterns of change in AIG and ASG

levels could be the result of an artefact of tissue extraction, particularly those

based on homogenization-free extraction protocols. Clearly, more work is

required to elucidate the mechanisms underlying the acid solubility of muscle

glycogen across extraction conditions (with or without homogenisation) in order

to establish once for all the physiological significance of the findings of the large

number of studies performed since the start of last century on AIG and ASG.

78

Chapter 3

Molecular size distribution

of acid-soluble and acid-insoluble glycogen

and the effect of extraction protocol

79

3.1 Introduction

Glycogen is a branched glucose polymer that serves as a rapidly available but

limited source of fuel. Glycogen stores, 50-80% of which are located in skeletal

muscles, are known to influence whole body fuel homeostasis, exercise

performance, the onset of muscle fatigue, and metabolic diseases such as

diabetes mellitus (Roach, 2002; Shearer & Graham, 2002). Although the

metabolic pathways by which glycogen is synthesised and degraded as well as

the regulation of these processes have been thoroughly investigated, several

aspects of glycogen’s structure, sub-cellular location and association with

proteins remain to be elucidated (Graham, 2009). In particular, it has been

known for over 75 years that in the presence of acid, muscle glycogen can be

separated into an acid insoluble glycogen (AIG) and acid soluble glycogen

(ASG) fraction (Willstatter & Rohdewald, 1934), but the molecular structure of

these glycogen fractions has remained elusive.

We showed in Chapter 2 that the separation of AIG and ASG in skeletal

muscles and their responses to changes in muscle glycogen levels are highly

dependent on the acid extraction protocol (Chapter 2; Barnes et al., 2009).

When repeated muscle homogenisation is performed, most of the glycogen is

extracted as ASG, and we showed in Chapter 2 that this is the most responsive

fraction to exercise and re-feeding post-exercise, with AIG levels remaining

stable under these conditions (Chapter 2; James et al., 2008; Barnes et al.,

2009). These findings corroborate those of the many studies that have adopted

homogenisation-dependent protocols to acid-extract glycogen, with ASG being

the most responsive fraction to a wide range of conditions affecting muscle

glycogen levels, such as adrenaline administration, electro-stimulation,

80

starvation and re-feeding after a prolonged fast (Bloom et al., 1951; Bloom &

Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1955, 1956; Stetten

et al., 1958; James et al., 2008).

When a homogenisation step is not included to acid-extract muscle glycogen,

which is a protocol that has been used extensively in recent years, we and

others have shown that AIG rather that ASG is the most abundant and

metabolically responsive fraction of glycogen, except when total muscle

glycogen levels are elevated (Barnes et al., 2009). For instance, Chapter 2

shows that AIG accounts for most of the changes in muscle glycogen levels in

response to exercise and recovery, but not when muscle glycogen levels are

elevated at which point ASG accounts to a far greater extent for the changes in

glycogen levels (Adamo et al., 1998b; Graham et al., 2001; Shearer et al., 2001;

Battram et al., 2004; Shearer et al., 2005a; Shearer et al., 2005b; Marchand et

al., 2007; Wilson et al., 2007).

Although the different patterns of AIG and ASG responses to changes in muscle

glycogen levels have been taken as evidence that they correspond to two

physiologically distinct glycogen pools, the mechanism underlying their different

solubilities in acid has remained elusive for several years. Almost two decades

ago, however, glycogen behaviour in acid was alleged to have been explained,

with AIG reported to correspond to a distinct small 400 kDa glycogen fraction

named proglycogen (PG; Lomako et al., 1991; Lomako & Lomako, 1991). As

each glycogen granule in skeletal muscle is covalently bound to the protein

glycogenin, the low acid solubility of PG/AIG was attributed to the resulting high

protein to glucosyl ratio of the smaller glycogen particles (Lomako et al., 1991a).

In contrast, the larger ASG with sizes of up to 10 000 kDa was termed

81

macroglycogen (MG), and its acid solubility was proposed to result from its low

glycogenin to glucosyl ratio (Lomako et al., 1991a). Some evidence was also

provided that PG was a discrete intermediate along the pathway of MG

synthesis (Lomako et al., 1993a), thus the origin of the terms PG and MG. As a

result, these findings sparked renewed interest in the physiology of AIG and

ASG, with many recent studies adopting the PG/MG paradigm to explain the

responses of AIG and ASG to a range of physiological conditions (Adamo et al.,

1998a; Adamo et al., 1998b; Asp et al., 1999; Derave et al., 2000; Shearer et

al., 2000; Graham et al., 2001; Shearer et al., 2001; Brojer et al., 2002a;

Marchand et al., 2002; Rosenvold et al., 2003; Battram et al., 2004; Shearer et

al., 2005a; Marchand et al., 2007).

One problem with the PG/MG model is the evidence against the existence of a

distinct 400 kDa PG species (Roach, 2002; Katz, 2006; James et al., 2008), as

the original report of such a discrete glycogen species may have been the result

of an experimental artefact caused by the inappropriate use of discontinuous

gel electrophoresis (Skurat et al., 1997). Moreover, many studies have reported

that total glycogen exists as a normally distributed continuum of glycogen

particles of different sizes, with no evidence for the existence of two populations

of glycogen differing in their molecular weights (Drochmans, 1962; Scott & Still,

1968; Wanson & Drochmans, 1968; Meyer et al., 1970; Schmalbruch &

Kamieniecka, 1974; Rybicka, 1981b, a; Skurat et al., 1997; Marchand et al.,

2002; Marchand et al., 2007). For this reason, it has been proposed that PG

and MG or AIG and ASG in skeletal muscles correspond to populations of low

and high molecular weight glycogen, respectively. However, whether this is the

case in skeletal muscle has never been examined. It is important to do so

82

because of the compelling evidence from the use of isotope labelling work that

AIG and ASG are physiological distinct glycogen pools (Stetten et al., 1958;

Krisman, 1973; Krisman & Barengo, 1975; Curtino et al., 1979; Aon & Curtino,

1984; Pitcher et al., 1987; Lacoste et al., 1990; Huang et al., 1997).

One factor explaining the lack of information about the molecular weight of AIG

and ASG in skeletal muscle might have to do with the difficulty of extracting AIG

from the acid-insoluble pellet. Although James and colleagues (2008) from our

laboratory extracted AIG with KOH digestion, this might partially degraded the

glycogen granules that may not have been truly representative of the intact AIG

(James et al., 2008). Also, the chromatography gel adopted to compare the

molecular weight of AIG and ASG was chosen to resolve glycogen particles

with large differences in molecular weight such as that between PG and MG,

but not glycogen with much smaller size differences. Finally, only glycogen

extracted using homogenisation-dependent protocol was compared in that

study, with no comparison between AIG and ASG extracted without

homogenisation. For these reasons, the primary purpose of this study was firstly

to develop a high yield extraction protocol to isolate AIG and secondly to

determine the molecular weight of the AIG and ASG fractions obtained using

both homogenisation-dependent and homogenisation-free extraction protocols

to examine whether AIG and ASG have different molecular sizes as proposed in

the literature.

83

3.2 Experimental procedures

3.2.1 Materials

Acarbose was purchased from Lomb Scientific Pty Ltd, Australia. Pronase was

obtained from Roche Diagnostics, USA. Sephacryl S-400 HR was purchased

from GE Healthcare, Australia. Carbon coated 150-mesh copper grids for

transmission electron microscopy were purchased from ProScitech, Australia.

Finally, uranyl acetate was purchased from BDH Chemicals Ltd, England.

3.2.2 Animals

All experiments were performed on adult male albino Wistar rats weighing

between 290 and 380 grams and obtained from the Biological Sciences Animal

Unit at the University of Western Australia (n = 5). Male rats were used in

preference to females to avoid the physiological changes associated with the

oestrous cycle (4-6 days). The rats were kept at approximately 20°C on a 12

hour light / 12 hour dark photoperiod and had unlimited access to water and a

standard laboratory chow diet (Glen Forrest Stockfeeders, Glen Forrest, W.A.,

6071: 55% digestible carbohydrate, 19% protein, 5% lipid and 21% non-

digestible residue by weight).

3.2.3 Tissue sampling

In the morning, rats were anaesthetised under halothane prior to sampling the

mixed portion of their gastrocnemius muscles. Anaesthesia was induced with

4% isoflurane in 96% oxygen, the level of halothane being subsequently

reduced to 1.5% once the animal was anaesthetised (Ferreira et al., 1998).

84

After removal, each muscle sample was immediately freeze-clamped and

frozen in liquid nitrogen and stored at -80°C for the subsequent measurement of

the content and molecular size of its glycogen.

3.2.4 Acid extraction of muscle glycogen

Acid extraction was performed as previously described by Barnes and

colleagues (2009). Freeze-dried muscle samples were dissected free of fat,

blood, and any other visible non-muscular connective tissue, then homogenised

in the presence of ice-cooled 1.5 M PCA (200 µl per 3 mg of sample) using an

IKA Labortechnik T-8 homogeniser (Staufen, Germany). The homogenate was

centrifuged at 2700 g for 10 minutes before the supernatant was removed, the

pellet re-suspended and re-homogenised with ice-cooled 1.5 M PCA (100 µl per

3 mg of sample), and centrifuged as before. This procedure was then repeated

for a total of 3 homogenisations. After the last centrifugation, the pellet was

collected and supernatants were combined.

Other pieces of the same muscle were also extracted using the protocol

outlined in Adamo and Graham (1998). Freeze-dried muscle samples separated

in small pieces were dissected free of visible blood and connective tissue and

placed in a glass tube in the presence of 1.5 M PCA. The muscle samples were

pressed against the tube with a plastic rod and left to stand for 20 minutes.

Then the extract was centrifuged at 2700 g for 10 minutes before the

supernatant was removed.

85

3.2.5 Molecular size distribution analysis using transmission

electron microscopy

Transmission electron microscopy analysis was performed as previously

described by Parker and colleagues (2007). Glycogen samples were

appropriately diluted up to 10-fold with the following buffer: 50 mM

Tris(hydroxymethyl)aminomethane pH 7.4, 125 mM NaCl. Strong-Carbon

coated 150-mesh copper grids (ProSciTech, Australia) were hydrophilised by

glow discharging in air. Diluted glycogen was applied to the grid within 15

minutes of glow discharging. One minute after application, excess sample was

drawn off with filter paper and the grids stained by the addition of 2 µl of 2%

(w/v) uranyl acetate. The samples were examined using a JEOL 2100

Transmission Electron Microscope operating at 120 kV. Five images were

recorded digitally using an 11 megapixel Orius digital camera for each sample

and measurements of particle diameter were recorded using the Image J image

analysis software. To ensure reliability of particle analysis, each image was

assigned a randomly generated 8 to 10 digit number. Analysis was performed

on each image with only the randomly assigned number available to the tester.

After analysis of all conditions, sample data was tabulated by the images’

corresponding number.

3.2.6 Glycogen determination

The combined ASG supernatants were vortexed before a 100 µl sample was

removed for the determination of ASG and a 200 µl sample for free-glucose

analysis. Then, 2 M hydrochloric acid was added to the AIG-rich pellet and ASG

supernatant samples for a final concentration of 1.9 M. Both samples were

86

vortexed, and tube weights recorded. The tubes were then placed in a 95°C

block heater for 2 hours to hydrolyse glycogen into glucose, with the tubes

being vortexed after 1 hour to aid digestion. After incubation, the samples were

vortexed, and a 400 µl aliquot was removed and neutralised by the addition of 2

M potassium carbonate. The resulting extracts were assayed for glucosyl units

and corrected for free glucose. Glucose levels were assayed according to

Bergmeyer (1974).

3.2.7 Expression of results and treatment and analysis of data

All glycogen concentrations are expressed as millimole glucosyl units per

kilogram dry weight tissue unless otherwise stated. Glycogen molecular size

distributions are expressed in diameters and divided into continuous classes as

a percentage of total particles. All statistical analyses were performed using

SPSS (Chicago, IL) version 17 and all data is presented as mean ± standard

error of the mean.

87

3.3 Results

3.3.1 Optimisation of glycogen extraction: effect of repeated

homogenisation of glycogen on its molecular size

determination by gel filtration chromatography and

transmission electron microscopy

To examine if the repeated homogenisation steps of the homogenisation-

dependent extraction protocol of AIG and ASG alter the molecular weight of

glycogen, gel filtration chromatography using Sephacryl S-400 HR (GE

Healthcare) was performed as described in James and colleagues (2008).

Sephacryl S-400 HR was chosen on the grounds that it is reportedly suited for

the separation of dextrans of molecular weights 1 × 104 to 2 × 106 Da. The

Sephacryl S-400 HR column (100 × 1.5 cm) was equilibrated with 50 mM 4-(2-

hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 2 mM 3[(3-

cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), pH 7.5 at a

constant flow rate of 0.5 ml min-1. Then, approximately 10 to 12 grams of fresh

gastrocnemius muscle was ground to a powder under liquid nitrogen using a

mortar and pestle before being transferred to pre-cooled vials for weight

determination. Ten volumes of ice cold 1.5 M PCA was added, and ASG was

extracted using a mortar and pestle pre-cooled in liquid nitrogen and subjected

to the homogenisation-free protocol outlined above. The ASG fraction thus

obtained was removed and half of the supernatant was repeatedly

homogenised (3 × 1 minute) as described earlier for the homogenisation-

dependent extraction of glycogen, whereas the other half remained untreated.

To concentrate the ASG thus obtained before application to the column, the

88

pooled supernatants were first dialysed to remove PCA using 10 kDa molecular

weight cut off cellulose dialysis tubing (this size exclusion < 10 kDa is 40-fold

smaller than the reported size of PG) against 3 changes (1 L) of double distilled

water per 30 ml of sample over 24 hours. After dialysis, the samples were

freeze-dried and then dissolved in 1 ml of chromatography buffer. The samples

were applied to the column at a flow rate of 0.5 ml min-1, and fractions of

approximately 5 ml were collected and assayed for glycogen. The resulting

elution profiles for glycogen with or without repeated homogenisations were

then compared.

The effect of repeated homogenisation on glycogens molecular size distribution

was also examined using transmission electron microscopy. Approximately 30

mg of freeze-dried gastrocnemius muscle was dissected free of fat, blood, and

any other visible non-muscular connective tissue and transferred to micro

centrifuge tubes. Ice-cooled 1.5 M PCA (200 µl per 3 mg of sample) was added,

and ASG was extracted in a similar manner as the homogenisation-free

protocol outlined above. As for the gel chromatography experiment described

above, the homogenate was centrifuged at 2700 g for 10 minutes and the ASG

supernatant was removed. Half of the supernatant was then repeatedly

homogenised (3 × 1 minute) as described earlier for the homogenisation-

dependent extraction of ASG, whereas the other half remained untreated. The

ASG was then precipitated from the supernatants by the addition of absolute

ethanol to a final concentration of 66% (v/v) and left to precipitate overnight at

4°C. The precipitate, collected by centrifugation for 10 minutes at 2700 g, was

washed in 66% (v/v) ethanol and again centrifuged for 10 minutes at 2700 g.

100% acetone was then added to the glycogen precipitate and allowed to

89

evaporate at room temperature before being re-suspended in 50–200 µl of the

following buffer: 50 mM Tris(hydroxymethyl)aminomethane pH 7.4, 125 mM

NaCl, for analysis. The molecular size distributions of the two samples were

then analysed using transmission electron microscopy.

Both gel filtration chromatography and electron microscopy indicate that

repeated extensive homogenisation has no effect on glycogen molecular size.

Using gel filtration chromatography, the elution profile of the repeatedly

homogenised ASG sample and the untreated ASG control were similar with the

elution peak for both samples occurring at almost identical elution volumes

(Figure 3.1). This was corroborated by electron microscopy analysis where the

homogenisation treatment had no significant effect on glycogen molecular size

distributions with average molecular sizes of 32.1 ± 0.17 nm and 31.4 ± 0.22 nm

(Cohen’s d = 0.2) for the repeatedly homogenised and the untreated control

sample, respectively (Figure 3.2).

90

Figure 3.1 Effect of extensive homogenisation on glycogen molecular

size distribution using gel filtration chromatography. The

values shown are expressed as a percentage of total glycogen

applied to the column in each fraction.

Figure 3.2 Effect of extensive homogenisation on glycogen molecular

size distribution using transmission electron microscopy.

The values shown for each particle diameter are expressed as a

percentage of total number of particles measured for each

condition.

0%

5%

10%

15%

20%

0 50 100 150 200 250

% o

f to

tal g

lyco

ge

n

Elution volume (ml)

Homogenised

Control

0%

5%

10%

15%

20%

25%

0 10 20 30 40 50 60

% o

f to

tal p

art

icle

s

Particle size (nm)

Homogenised

Control

91

3.3.2 Optimisation of AIG extraction

In order to optimise the extraction of AIG from the acid-insoluble muscle extract

pellets, some pellets obtained with the homogenisation-dependent protocol

described above were neutralised with 2 M NaOH and incubated in the

following buffers to determine which one resulted in the highest extraction yield:

1. 20 mM Tris/HCl pH 7.5 + 50%(w/v) urea

2. 0.1 M Tris/HCl pH 7.5 + 0.5%(w/v) SDS

3. 0.1 M Tris/HCl pH 7.5 + 0.5%(w/v) SDS + 0.1% (w/v) pronase

Since urea can disrupt non-covalently bound proteins and extract AIG from the

liver under neutral conditions (Meyer & Lourau, 1956), part of the neutralised

acid-insoluble pellet was re-suspended in 1.5 volumes of 50% urea and

incubated at 45°C for three hours whilst shaking as described by Meyer and

Lourau (1956). Then, the homogenate was centrifuged for 10 minutes at 2700 g

and the supernatant collected. The resulting pellet was re-suspended in one

volume of 50% urea, incubated for 12 hours without shaking, and centrifuged

for 10 minutes at 2700 g. After this, the combined supernatants and the pellet

were assayed for glycogen.

The extraction of AIG from acid-insoluble pellets was also performed in the

presence of sodium dodecyl sulphate (SDS) to strip non-covalently bound

proteins from the glycogen molecule (Pitcher et al., 1987). The suitability of this

protocol is suggested by the work of others who reported that SDS solubilises

AIG obtained from bovine retina via the breakdown of acid-insoluble cellular

membranes that localise glycogen through hydrophobic interactions (Miozzo et

al., 1989; Lacoste et al., 1990). Using the SDS extraction buffer: 0.1 M Tris/HCl

92

pH 7.5 + 0.5%(w/v) SDS, as described by Roche Applied Science (Pronase,

#10165921001, product instructions), the acid-insoluble pellet was incubated for

one hour at 37°C before being centrifuged for 10 minutes at 2700 g, and both

supernatant and pellet were subsequently assayed for glycogen.

The third extraction medium tested here contained pronase, since pronase is

not only the most suitable protease cocktail for extensive proteolysis for use in

the presence of carbohydrate (Spiro, 1966), but also has been shown to liberate

AIG in a range of tissue types such as rat liver and bovine retina (Krisman,

1972; Curtino et al., 1979; Aon & Curtino, 1984; Curtino & Lacoste, 2000).

Using the pronase buffer described above (Roche Applied Science, Pronase,

#10165921001, product instructions), the acid-insoluble pellet was incubated for

one hour at 37°C, centrifuged for 10 minutes at 2700 g, and both supernatant

and pellet were subsequently assayed for glycogen.

Of the three AIG extraction buffer described above, the highest AIG yield was

achieved when extraction was performed after incubation with pronase. Indeed,

incubation with pronase for 1 hour at 37°C liberated significantly more AIG (99.1

± 1.3%), than with either SDS alone (80.8 ± 1.9%), or urea (59.3 ± 6.8%; Figure

3.3).

93

Figure 3.3 Incubation of AIG pellet with various extraction buffers. The

values shown represent means ± S.E.M. (n = 3) and values are

expressed as a percentage of total glycogen. a, significantly

different from urea and SDS (p < 0.05).

0%

20%

40%

60%

80%

100%

Urea SDS SDS + pronase

Extr

act

ion

yie

ld (

%)

Soluble

Insoluble

a

a

94

3.3.3 Acid solubility of extracted AIG

In order to ascertain indirectly whether the proteins associated with AIG are

responsible for its insolubility in acid. The AIG extracted using the pronase

buffer was analysed for its solubility in the presence of acid. To this end, the

supernatant from the pronase digestion was acidified by the addition of an equal

volume of 3.0 M PCA, left on ice for 20 minutes, and centrifuged for 10 minutes

at 2700 g. Under these conditions, all of the glycogen remained in the

supernatant after centrifugation (Figure 3.4), with no glycogen remaining in the

pellet. As the addition of acid resulted in the noticeable precipitation of many

proteins, this step was included in preparing samples for electron microscopy as

it provided much cleaner samples.

Figure 3.4 Acid solubility of pronase-extracted AIG. The values shown

represent means ± S.E.M. (n = 3) and are expressed in millimoles

glucosyl units per kilogram dry muscle. No significant difference

between pronase solubilised and acid soluble, pronase solubilised

(p > 0.05).

0

10

20

30

40

50

60

70

80

90

100

Pronase solublised Acid soluble, pronase solublised

Gly

cog

en

(m

mo

l k

g-1

d.w

.)

95

3.3.4 Effect of pronase treatment on molecular size distribution of

glycogen

In order to ensure that pronase treatment had no effect on the molecular size of

glycogen extracted due to the presence of contaminating glucosidase, 50 µl

aliquots of ASG prepared as described above for transmission electron

microscopy, were exposed to the following treatments:

1. No treatment

2. Addition of 500 µl of pronase-free SDS incubation buffer and left on ice for

20 minutes

3. Addition of 500 µl of pronase-free SDS incubation buffer followed by a 1

hour incubation at 37°C in the presence of 0.1% (w/v) pronase final

concentration

4. Addition of 500 µl of pronase-free SDS incubation buffer followed by a 4

hour incubation at 37°C in the presence of 0.1% (w/v) pronase final

concentration

5. Addition of 500 µl of pronase-free SDS incubation buffer followed by a 1

hour incubation at 37°C in the presence of 0.1% (w/v) pronase and 10 mM

acarbose final concentration

6. Addition of 500 µl of pronase-free SDS incubation buffer followed by a 4

hour incubation at 37°C in the presence of 0.1% (w/v) pronase and 10 mM

acarbose final concentration

After being subjected to these treatments, all samples were centrifuged,

acidified and concentrated as described above for transmission electron

96

microscopy analysis. Then, the molecular size distributions were compared

across treatments.

The glycogen molecular size distributions of the untreated and non-pronase

incubated controls as well as the combined pronase and acarbose digested

samples reveal a very similar molecular size distribution (Figure 3.5A). In

contrast, both the 1- and 4-hour pronase only treatments caused a more

pronounced trailing in the distribution profile, with the presence of a second

peak of smaller size glycogen particles becoming apparent after 4 hours of

digestion (Figure 3.5B).

97

Figure 3.5 The effect of pronase digestion on the glycogen molecular

size distribution A) without the inclusion of acarbose and B)

in the presence of acarbose. The values shown for each particle

diameter are expressed as a percentage of total number of

particles measured for each condition.

0%

5%

10%

15%

20%

25%

0 10 20 30 40 50

% o

f to

tal g

lyco

ge

n p

art

icle

s

Particle size (nm)

No treatment

Pronase-free

Pronase 1hr

Pronase 4hr

A

0%

5%

10%

15%

20%

25%

0 10 20 30 40 50

% o

f to

tal g

lyco

ge

n p

art

icle

s

Particle size (nm)

No treatment

Pronase-free

Pronase + Acarbose 1hr

Pronase + Acarbose 4hr

B

98

3.3.5 Molecular size distribution of ASG and AIG extracted with

homogenisation-free and homogenisation-dependent

protocols

In order to compare the molecular size distribution of AIG and ASG using

electron microscopy analysis, gastrocnemius muscles from fed rats

(approximately 35 mg dry weight) were separated into AIG and ASG fractions

using both the homogenisation-dependent (Figure 3.7) and homogenisation-

free (Figure 3.9) protocols outlined above. For electron microscopy analysis of

ASG, this glycogen species was prepared from the acid supernatant as

described above.

To solubilise the AIG for analysis, the acid insoluble pellet was neutralised with

the addition of 2 M NaOH as described above, and incubated for 2 hour at 37°C

in a Tris buffer containing pronase (0.1 M Tris-HCl pH 7.5, 0.5% (w/v) SDS, 10

mM acarbose, 0.1% (w/v) pronase). Then, the sample was centrifuged at 2700

g for 10 minutes, the supernatant acidified with an equal volume of 3.0 M PCA,

left for 20 minutes on ice, and centrifuged at 2700 g. The resulting supernatant

was collected, and the glycogen was precipitated by the addition of absolute

ethanol to a final concentration of 66% (v/v) and left to precipitate overnight at

4°C. The precipitate, collected by centrifugation for 10 minutes at 2700 g, was

washed in 66% (v/v) ethanol and again centrifuged for 10 minutes at 2700 g.

100% acetone was then added to the glycogen precipitate and allowed to

evaporate at room temperature before being re-suspended in 50–200 µl of Tris

buffer (50 mM Tris(hydroxymethyl)aminomethane, 125 mM NaCl, pH 7.4) for

analysis.

99

The proportion of ASG extracted using the homogenisation-free and

homogenisation-dependent protocols were 20.3 ± 1.1% and 66.4 ± 2.7%,

respectively (Figure 3.6).

Figure 3.6 Extraction of ASG and AIG for TEM size distribution analysis.

The values shown represent means ± S.E.M. and are expressed

in millimoles glucosyl units per kilogram dry muscle. a,

significantly different to homogenisation-dependent (p < 0.05).

0

50

100

150

200

Homogenisation-dependent Homogenisation-free

Gly

cog

en

(m

mo

l k

g-1

d.w

.)

ASG

AIG

a

a

100

The molecular size distributions of ASG and AIG extracted using the

homogenisation-dependent protocol was very similar with both having a peak

molecular size class of 32-34 nm and average molecular size of 32.2 ± 0.22 nm

and 31.7 ± 0.18 nm (Cohen’s d = 0.1; Figure 3.8) for ASG and AIG,

respectively. These glycogen fractions had also similar median sizes (32.5 nm

and 32.1 nm for ASG and AIG, respectively), with ASG and AIG being normally

distributed. The molecular size distributions of ASG and AIG obtained without

any homogenisation step were also similar, with both having a peak particle

size class of 30-32 nm and a similar average molecular size of 31.2 ± 0.16 nm

and 30.8 ± 0.20 nm, respectively (Cohen’s d = 0.08; Figure 3.10). Here as well,

both glycogen fractions were normally distributed.

101

Figure 3.7 Electron microscopy of purified A) AIG and B) ASG extracted

using the homogenisation-dependent protocol.

A

B

102

Figure 3.8 Glycogen molecular size distributions of ASG and AIG

extracted using the homogenisation-dependent protocol

expressed by size frequency. The values shown for each

particle diameter are expressed as average percentage ± S.E.M.

of total number of particles measured for each condition.

0%

5%

10%

15%

20%

25%

5 15 25 35 45 55

% o

f to

tal p

art

icle

s m

ea

sure

d

Particle size (nm)

ASG-homogenisation-dependent

AIG-homogenisation-dependent

103

Figure 3.9 Electron microscopy of purified A) AIG and B) ASG extracted

using the homogenisation-free protocol.

A

B

104

Figure 3.10 Glycogen molecular size distributions of ASG and AIG

extracted using the homogenisation-free protocol expressed

by size frequency. The values shown for each particle diameter

are expressed as average percentage ± S.E.M. of total number of

particles measured for each condition.

0%

5%

10%

15%

20%

25%

5 15 25 35 45 55

% o

f to

tal p

art

icle

s m

ea

sure

d

Particle size (nm)

ASG-homogenisation-free

AIG-homogenisation-free

105

3.4 Discussion

It was established in the previous Chapter that muscle glycogen extracted in the

presence of acid gives rise to AIG and ASG, with the yield of each glycogen

fraction being dependent on whether a homogenisation step is included or not

in the extraction protocol. In the early nineties, it was proposed that the

structural feature underlying the different acid solubilities of these two glycogen

fractions was their sizes, with AIG proposed to correspond to a population of

lower molecular weight glycogen particles compared to ASG. Although this view

is generally accepted in the scientific literature (see Chapter 1), surprisingly the

molecular size distributions of AIG and ASG in skeletal muscles have never

been compared. Here, for the first time, not only we describe a protocol for the

almost complete extraction of AIG in muscle, but also we compare the pattern

of glycogen molecular size distribution between AIG and ASG obtained using

homogenisation-free and homogenisation-dependent protocols. Our results

show that ASG and AIG have a similar average molecular size and share a

similar pattern of molecular size distribution, irrespective of the extraction

protocol. In addition, the pattern of molecular size distribution is comparable to

that of total glycogen reported by others (Marchand et al., 2007). Finally, we

also show that AIG is no more acid-insoluble after protease treatment, thus

suggesting that the different acid solubilities of AIG and ASG might be related to

different complements of proteins associated with these fractions, with the

identity of these proteins still remaining to be determined.

Prior to comparing the molecular size distribution of AIG and ASG, a number of

important precautions were taken in this study. Firstly, since the

homogenisation-dependent protocol requires the repeated extensive

106

homogenisation of muscle extracts, we examined the possibility that this may

physically damage the glycogen granule and release some glycogen fragments

in solution, thus resulting in an overestimation of the proportion of ASG and a

leftward shift of its molecular size distribution. To test whether this is the case,

we examined the effect of repeated homogenisation on the molecular size of

glycogen obtained under mild extraction conditions. For this reason, ASG

obtained from the homogenisation-free extraction protocol was used here to test

the effect of repeated homogenisation on glycogen’s structure. Using gel

filtration chromatography to evaluate the effect of repeated homogenisation on

the molecular weight of glycogen, we found that repeated homogenisation had

no effect on the elution profile of glycogen, thus suggesting that its structure

was not affected by this treatment (James et al., 2008). However, since the gel

chromatography conditions adopted here were suitable only for the detection of

large differences in glycogen molecular weight and provided no information

about the molecular size distribution of the different glycogen particles,

glycogen obtained with or without repeated homogenisation were compared

using transmission electron microscopy. In agreement with our gel

chromatography results, repeated homogenisation had no effect on the average

size and molecular size distribution of glycogen.

The next challenge was to develop a protocol to extract most of the AIG from

the acid-insoluble pellet so that the AIG thus obtained was representative of

total AIG. To this end, acid-insoluble pellets were incubated with strong physical

dissociative reagents, such as SDS or urea (Aon & Curtino, 1984). In the

presence of SDS or urea, only 81% and 60% of AIG were extracted,

respectively. Such a resistance of skeletal muscle AIG to physical dissociation

107

has also been reported by others attempting to extract AIG from other tissues,

such as rat liver (Krisman, 1972). However, there have been previous reports of

higher yields in other tissues, possibly due to the incubation time or temperature

adopted or the tissue itself as none was performed on skeletal muscle. As this

study was primarily concerned with preserving the molecular integrity of

glycogen, incubations where carried out at 37°C for two hours whereas those

previous studies have used significantly higher temperatures or longer

incubation periods to maximise AIG extraction from tissues such as retina

(Meyer & Lourau, 1956; Miozzo et al., 1989; Lacoste et al., 1990).

A better way to extract a high yield of AIG compared to SDS or urea treatment

is to re-suspend and expose the AIG-rich pellet to pronase as our results show

that this results in the extraction of over 98% of the AIG in skeletal muscle. This

finding is in agreement with those obtained by others investigating tissues such

as liver and retina (Krisman, 1973; Curtino et al., 1979; Aon & Curtino, 1984,

1985; Curtino & Lacoste, 2000). In addition, our results suggest that the acid-

insolubility of AIG may be attributable to an interaction between AIG and

specific cellular proteins, since AIG is no longer insoluble in acid after pronase

treatment. A similar loss of acid-insolubility of AIG has also been seen in bovine

retina following AIG extraction with pronase digestion (Aon & Curtino, 1984) and

by boiling with SDS (Miozzo et al., 1989). Given the evidence that acid solubility

is mediated, at least in part, by the proteins bound to glycogen, it follows that

the different acid solubilities of AIG and ASG might have to do with different

complements of proteins being associated with these fractions, but with the

identity of these proteins still remaining to be determined.

108

One factor that cannot be overlooked is the possibility that pronase might be

contaminated with enzymes capable of degrading glycogen (Aon & Curtino,

1984). That this is an important factor to consider is shown by our results that

glycogen incubation with pronase results in the partial digestion of glycogen,

possibly by reversibly denatured glucosidases or impurities in the pronase

preparation (Aon & Curtino, 1984). This interpretation is further supported by

our observation that the inclusion of acarbose in the incubation buffer protects

the molecular integrity of glycogen, with no detectable changes in molecular

size even after four hours of incubation.

By adopting the many precautions described here not only to minimise changes

in glycogen structure during the extraction of ASG and AIG, but also to optimise

the extraction yield of both AIG and ASG, the molecular size distributions of AIG

and ASG were compared. Our findings show that the molecular size

distributions of AIG and ASG as well as their average sizes are almost identical,

and this finding holds irrespective of whether glycogen is extracted with or

without a homogenisation step. In addition, the average size and molecular size

distribution of glycogen in these fractions were similar to those published on

total glycogen (Drochmans, 1962; Scott & Still, 1968; Wanson & Drochmans,

1968; Meyer et al., 1970; Schmalbruch & Kamieniecka, 1974; Rybicka, 1981b,

a; Skurat et al., 1997). These findings not only corroborate the work of others

(Skurat et al., 1997; Katz, 2006; James et al., 2008) that AIG does not

correspond to the discrete 400 kDa PG species originally proposed by Lomako

and colleagues (1991a; 1991b), but also that AIG doe not correspond to a

population of glycogen of low molecular sizes compared to ASG as has been

proposed in recent years (Marchand et al., 2007; Graham, 2009; Graham et al.,

109

2010). It follows from these findings that the use of the PG/MG model to explain

the presence of AIG and ASG in skeletal muscles should now be abandoned.

Whether this should also be the case in other tissues remains to be determined.

It is also noteworthy that since repeated homogenisation has no effect on the

molecular size distribution of glycogen particles, this provides further support to

the interpretation that the higher yield of ASG obtained using homogenisation-

dependent extraction protocol compared to that obtained with no

homogenisation is not an artefact whereby extensive homogenisation results in

the fragmentation of the glycogen particles. On the contrary, as discussed

previously in Chapter 2, it is possible that the lower ASG yield obtained when

muscles are extracted without a homogenisation step is the result of an artefact

of tissue extraction. Indeed, the poor yield of ASG using such a protocol might

be due to some of the glycogen precipitating not because of its poor acid

solubility per se, but simply because it is trapped within the dense mesh of

undisrupted myofibrils that precipitate during centrifugation in the presence of

acid. This, in turn, could result in the contamination of AIG by ASG and in a

large overestimation of the proportion of AIG and underestimation of ASG

levels. Alternatively, it is possible that glycogen extracted without a

homogenisation step results in a highly labile glycogen-protein fraction that

carries a distinct physiological role. Arguably, more work is required to elucidate

the physiological importance of the AIG and ASG obtained without

homogenisation.

It is important to note that despite AIG and ASG fractions having a similar

molecular size distribution, there is plenty of evidence that these fractions

obtained after extensive homogenisation correspond to physiologically distinct

110

pools of glycogen. For instance, we have shown in humans that changes in

ASG levels account for the fall and increase in muscle glycogen levels in

response to exercise and re-feeding, respectively, whereas AIG remains at a

constant concentration (Chapter 2; Barnes et al., 2009). Similarly, ASG

accounts for most of the changes in muscle glycogen levels in response to

fasting and re-feeding in rats (James et al., 2008) and, as discussed in Chapter

1, ASG is generally the most responsive fraction to changes in glycogen levels

when muscle extraction is performed with a homogenisation step. It does not

follow from the lesser responsiveness of AIG to changes in glycogen levels that

this fraction is not physiologically important, since AIG has been shown to

readily incorporate isotopically labelled glucose (Stetten et al., 1958; Krisman,

1973; Krisman & Barengo, 1975; Curtino et al., 1979; Aon & Curtino, 1984;

Pitcher et al., 1987; Lacoste et al., 1990; Huang et al., 1997).

Given the evidence that AIG and ASG obtained from homogenised muscle

extracts correspond to different glycogen pools, this raises the question of the

mechanisms underlying their different solubilities in acid. Our findings do not

support the notion that a high glycogenin to glucosyl ratio in glycogen is

responsible for the acid insolubility of glycogen (Lomako et al., 1991a). This is

because this protein is typically embedded inside the glycogen molecule, thus

preventing protease treatment of glycogen from affecting its glycogenin core

(Aon & Curtino, 1985; Curtino & Lacoste, 2000). Moreover, if glycogenin were to

play the key role in determining the solubility of glycogen, AIG should still

remain acid-insoluble following pronase digestion. Alternatively, since each

glycogen particle is non-covalently bound to a number of proteins to form a

complex known as glycosome, AIG and ASG might correspond to different sub-

111

fractions of glycosomes, each with a distinct complement of proteins (Rybicka,

1996; Skurat & Roach, 2004). Moreover, given that a large fraction of muscle

glycosomes is in turn non-covalently bound to membranes such as the SR via

the muscle glycogen-binding regulatory subunit of PP1 (Rybicka, 1996) maybe

this is a factor that affects glycogen solubility. Finally, ASG and AIG may

represent glycogen particles located in different compartments inside the

muscle cell as proposed recently (Marchand et al., 2002; Marchand et al.,

2007), but this is unlikely to involve lysosomes since only a small fraction of

glycogen (~6%) in skeletal muscle is found in this compartment (Calder &

Geddes, 1989a).

In conclusion, here we show in rats that AIG and ASG exist as a range of

glycogen particles of similar average size and comparable molecular size

distribution, confirming that AIG does not correspond to a population of low

molecular size glycogen previously referred to as PG. Our findings also show

that the higher extraction yield of ASG using a homogenisation-dependent

extraction protocol is not the result of an artefact of homogenisation whereby

the glycogen particles are fragmented in smaller particles. Finally, we provide

evidence that the acid insolubility of AIG is determined, at least in part, by the

complement of proteins with which it is associated. Further investigation is

required, however, to uncover the mechanisms responsible for the different acid

solubilities of AIG and ASG.

112

Chapter 4

Effect of exercise and re-feeding on the

molecular size distribution of acid-soluble and

acid-insoluble glycogen in skeletal muscle

113

4.1 Introduction

Glycogen in skeletal muscle is an important yet limited source of fuel,

particularly when energy demand is high (Shearer & Graham, 2002). Such is

the importance of glycogen that it has been the object of a large volume of

research. Despite this, some aspects of its biochemistry are still without an

answer, such as the observation made nearly 75 years ago that when extracted

in the presence of acid, glycogen separates in an acid insoluble glycogen (AIG)

and acid soluble glycogen (ASG) fraction (Willstatter & Rohdewald, 1934).

In Chapter 2, we reported and published that the proportions of AIG and ASG

and their responses to changes in muscle glycogen levels are highly dependent

on the extraction protocol (Barnes et al., 2009). When a homogenisation step is

not included, as is the case for almost all studies performed in recent years,

most of the glycogen is extracted as AIG, with this glycogen fraction being the

most responsive to changes in muscle glycogen levels unless these levels are

elevated (Adamo et al., 1998b; Graham et al., 2001; Shearer et al., 2001;

Battram et al., 2004; Shearer et al., 2005a; Shearer et al., 2005b; Marchand et

al., 2007; Wilson et al., 2007). In contrast, when repeated homogenisations are

performed, ASG is the most abundant and also responsive fraction to changes

in muscle glycogen levels, with AIG remaining at near stable levels under these

conditions (Bloom et al., 1951; Bloom & Knowlton, 1953; Bloom & Russell,

1955; Russell & Bloom, 1955, 1956; Stetten et al., 1958; James et al., 2008).

Although it is possible that the AIG extracted without a homogenisation step is a

physiologically significant pool of glycogen easily mechanically disrupted when

muscle extracts are homogenised, it is possible that the higher proportion and

114

responsiveness of AIG under these conditions are the result of an artefact of

tissue extraction (Chapters 2, 3; James et al., 2008; Barnes et al., 2009). This is

because the high proportion of AIG obtained using such a protocol could be due

to some of the glycogen precipitating not because of its poor acid solubility per

se, but simply because it is trapped by the dense mesh of undisrupted

myofibrils that precipitate during centrifugation in the presence of acid. The

resulting contamination of AIG by ASG would thus be expected to result in a

large overestimation of the proportion of AIG and underestimation of ASG levels

(James et al., 2008; Barnes et al., 2009).

Until almost two decades ago, the molecular mechanism underlying the

difference in acid solubility between AIG and ASG in extracts subjected to

extensive homogenisation remained elusive. However, in the early nineties,

glycogen behaviour in acid was alleged to have been explained with evidence

that AIG corresponds to a small 400 kDa glycogen species named proglycogen

(Lomako et al., 1991a). Since each glycogen particle in skeletal muscle is

covalently bound to glycogenin and that proteins are in general insoluble in

acid, the low acid solubility of AIG compared to the much larger ASG was

proposed to be the result of its high glycogenin to glucosyl residues ratio

(Lomako et al., 1991a). However, the existence of proglycogen as a discrete

glycogen species was challenged in subsequent years (Skurat et al., 1997;

Roach, 2002; Katz, 2006; James et al., 2008), with the finding that muscle

glycogen exists as a normally distributed continuum of glycogen particles of

different sizes (Drochmans, 1962; Scott & Still, 1968; Wanson & Drochmans,

1968; Meyer et al., 1970; Schmalbruch & Kamieniecka, 1974; Rybicka, 1981b,

a; Skurat et al., 1997; Marchand et al., 2002; Shearer & Graham, 2004;

115

Marchand et al., 2007; Ryu et al., 2009). For this reason, it was proposed that

AIG and ASG correspond to subpopulations of glycogen of low and high

molecular weight, respectively (Marchand et al., 2002; Marchand et al., 2007).

This claim, however, was challenged for the first time by our findings in Chapter

3 where we show that both AIG and ASG in fed animals have the same

molecular size distribution, thus indicating that molecular size does not explain

the different behaviours of AIG and ASG in acid as had been previously

proposed (Lomako et al., 1991a; Lomako et al., 1991b).

The similar molecular size distribution of AIG and ASG raises the question of

whether this is also the case when muscle glycogen levels are changing. Given

that, as described above, most of the changes in total glycogen concentration

measured from homogenised extracts have been attributed to ASG and that it

has recently been reported that an increase in glycogen level is associated with

a rightward shift in its molecular size distribution (Marchand et al., 2007), it is

possible that the molecular size distribution of ASG responds in a way similar to

that of total glycogen to changes in glycogen levels, but not AIG because its

levels change little under these conditions. However, since AIG is also a

metabolically active glycogen pool (Stetten et al., 1958; Krisman, 1973; Krisman

& Barengo, 1975; Curtino et al., 1979; Aon & Curtino, 1984; Pitcher et al., 1987;

Lacoste et al., 1990; Huang et al., 1997) that can incorporate glucose even

without net changes in AIG concentration when muscle glycogen levels are

changing, a remodelling of the molecular size distribution of AIG is also

theoretically possible under these conditions. The aim of this study, therefore,

was to examine the effect of changes in glycogen levels brought about by

116

exercise and post-exercise re-feeding on the molecular size distributions of AIG

and ASG.

4.2 Experimental procedures

4.2.1 Materials

Acarbose was purchased from Lomb Scientific Pty Ltd, Australia. Pronase was

purchased from Roche Diagnostics, USA. Carbon coated 150-mesh copper

grids for transmission electron microscopy were purchased from ProScitech,

Australia. Uranyl acetate was purchased from BDH Chemicals Ltd, England.

4.2.2 Animals

All experiments were conducted on adult male albino Wistar rats weighing on

average 384 ± 22 grams and obtained from the Biological Sciences Animal Unit

at the University of Western Australia. Male rats were used in preference to

females to avoid the physiological changes associated with the oestrous cycle

(4-6 days). The rats were kept at approximately 20°C on a 12 hour light / 12

hour dark photoperiod and had unlimited access to water and a standard

laboratory chow diet (Glen Forrest Stockfeeders, Glen Forrest, W.A., 6071: 55%

digestible carbohydrate, 19% protein, 5% lipid and 21% non-digestible residue

by weight).

4.2.3 Exercise protocol

Since rats are natural swimmers, exercise protocols based on swimming are

widely used, the intensity of the exercise being determined by the amount of

lead weight attached to the tail (Brau et al., 1997). The advantage of this

117

exercise protocol over one that uses a treadmill is that a prolonged training

period is not required for the animal to exercise to near maximal intensity.

Immediately before swimming, each animal was weighed and a lead weight

equivalent to 9% of body mass was attached to the base of the tail (McArdle &

Montoye, 1966; Brau et al., 1997). Swimming lasted for three minutes and took

place in a 30-cm diameter plastic tank filled with water (48-cm deep) at 34°C. In

order to exercise the rats to near exhaustion, the size of the lead weight was

progressively reduced, on each occasion by a third, as the animals tired until

two thirds of the weight was removed (Brau et al., 1997). Rats were randomly

assigned to one of six groups (n = 7 per group). The first group of rats were

sacrificed at rest and had been given unlimited access to a standard laboratory

chow diet. All other rats were fasted for a period of 24 hours. One group of

fasted rats were sacrificed at rest (group 2) and another one immediately upon

completion of the 3-minute swim (group 3). After exercise, each of the other two

groups was allowed to recover individually in separate cages without access to

food before being sacrificed after 15 (group 4) and 60 minutes of recovery

(group 5). Finally, each animal of the last group was allowed to recover

individually in separate cages without access to food for 60 minutes, after which

the rats were given unlimited access to a standard laboratory chow diet and

allowed to recover for 24 hours before being sacrificed (group 6, Figure 4.1).

For TEM analysis of the glycogen molecular size distribution, only groups 2, 3

and 6 were used as they exhibited the most extreme changes in total glycogen

concentration.

118

Figure 4.1 Exercise and muscle sampling protocol.

Exercise

Rats sacrificed

0 h 24 h Pre 1 h

Recovery

Laboratory chow No food Fasting 24 h

15 min Fed

119

4.2.4 Tissue sampling

Rats were anaesthetised under halothane prior to sampling their gastrocnemius

muscles. Anaesthesia was induced with 4% isoflurane in 96% oxygen, the level

of halothane being subsequently reduced to 1.5% once the animal was

anaesthetised (Ferreira et al., 1998). After removal, each muscle sample was

immediately freeze-clamped in liquid nitrogen and stored at -80°C for the

subsequent enzymatic analysis of its glycogen content or for microscopy

analyses. The rats were then killed by cardiac excision.

4.2.5 Acid extraction of muscle glycogen

Acid extraction was preformed as previously described in Chapter 2 (Barnes et

al., 2009). Briefly, freeze-dried muscle samples were dissected free of fat,

blood, and any other visible non-muscular connective tissue, and were

homogenised in the presence of ice-cooled 1.5 M PCA (200 µl per 3 mg of

sample) using an IKA Labortechnik T-8 homogeniser (Staufen, Germany). The

homogenate was centrifuged at 2700 g for 10 min before the supernatant was

removed and the pellet re-suspended, re-homogenised with ice-cooled 1.5 M

PCA (100 µl per 3 mg of sample) and centrifuged as before. This was then

repeated, for a total of 3 homogenisations. After the last centrifugation, the

pellet was collected and supernatants were combined.

4.2.6 Extraction of AIG

In order to extract AIG from the acid-insoluble pellets, the protocol developed in

Chapter 3 was adopted here. Each pellet was re-suspended in 500 µl of 0.1 M

Tris-HCl pH 7.5 and neutralised with 50µl of 2 M NaOH and made up to a final

120

volume of 200 µl per 3 mg of muscle tissue with the following buffer at a final

concentration of: 0.1 M Tris-HCl pH 7.5, 0.5% (w/v) SDS, 10 mM acarbose,

0.1% (w/v) pronase and incubated for 2 hours at 37°C. Acarbose and pronase

were included to prevent glycogenolysis and promote proteolysis, respectively.

Following incubation, the sample was centrifuged at 2700 g for 10 minutes and

the supernatant acidified with an equal volume of 3.0 M PCA to precipitate

proteins (pronase), left for 20 minutes on ice and then centrifuged at 2700 g for

10 minutes before the supernatant was collected and prepared for molecular

size analysis of glycogen.

In order to estimate the extent to which ASG might have contaminated the AIG

pellet due to trace amounts of ASG in the supernatant remaining in the

insoluble pellet after centrifugation, the acid-insoluble pellets of 6 samples (8.2

± 0.02 mg) were incubated at 45°C and allowed to evaporate in a ducted fume

hood. The sample weights were measured before and at regular intervals

during incubation until the samples reached a stable weight, and this was used

to calculate the extent to which ASG contributed to the AIG level determined in

the pellet, which here corresponded to an average of only 0.85 ± 0.13%

contamination.

4.2.7 Molecular size distribution analysis using transmission

electron microscopy

For transmission electron microscopy (TEM) analysis, glycogen was

precipitated from the supernatants by the addition of absolute ethanol to a final

concentration of 66% (v/v) and left to precipitate overnight at 4°C. The

precipitate, collected by centrifugation for 10 minutes at 2700 g, was washed in

121

66% (v/v) ethanol and again centrifuged for 10 minutes at 2700 g. Then, 100%

acetone was added to the glycogen precipitate which was allowed to evaporate

at room temperature before being re-suspended in 50-200 µl of the following

buffer: 50 mM Tris(hydroxymethyl)aminomethane pH 7.4, 125 mM NaCl, for

TEM analysis.

TEM analysis was performed as previously described by Parker and colleagues

(2007). Glycogen samples were appropriately diluted up to 10-fold with the

following buffer: 50 mM Tris(hydroxymethyl)aminomethane pH 7.4, 125 mM

NaCl. Strong-Carbon coated 150-mesh copper grids (ProSciTech, Australia)

were hydrophilised by glow discharging in air. Diluted glycogen was applied to

the grid within 15 minutes of glow discharging. One minute after application

excess sample was drawn off with filter paper and the grids stained by the

addition of 2 µl of 2% (w/v) uranyl acetate. The samples were examined using a

JEOL 2100 Transmission Electron Microscope operating at 120 kV. Five

images were recorded digitally using an 11 megapixel Orius digital camera for

each sample and measurements recorded using the Image J image analysis

software. To ensure reliability of particle analysis, each image was assigned a

randomly generated 8 to 10 digit number. Analysis was performed on each

image with only the randomly assigned number available to the tester. After

analysis of all conditions, sample data was tabulated by the images’

corresponding number.

4.2.8 Glycogen determination

For the determination of ASG, the combined supernatants were vortexed before

a 100 µl sample was removed and combined with 10 volumes of 2 M HCl for the

122

assay of glycogen and a 200 µl sample for free-glucose analysis. For the assay

of AIG, 1000 µl of 2 M HCl was added to the pellet. After the addition of HCl,

AIG and ASG samples were vortexed and tube weights recorded. The tubes

were then placed in a 95°C block heater for 2 hours to hydrolyse glycogen, with

the tubes being vortexed after 1 hour to aid digestion. After incubation, the

samples were vortexed, and a 400 µl aliquot was removed and neutralised by

the addition of 2 M potassium carbonate. The resulting extracts were assayed

for glucosyl units and corrected for free glucose. Glucose levels were assayed

according to Bergmeyer (1974).

4.2.9 Expression of results and treatment and analysis of data

All glycogen concentrations are expressed as millimole glucosyl units per

kilogram dry weight tissue unless otherwise stated. Glycogen molecular size

distributions are expressed in diameters and divided into continuous classes as

a percentage of total particles. All statistical analyses were performed using

SPSS (Chicago, IL) version 17 and all data is presented as mean ± standard

error of the mean.

123

4.3 Results

4.3.1 Effect of exercise and re-feeding on the levels of glycogen,

AIG and ASG.

In response to a 24-hour fast, total muscle glycogen levels decreased

significantly (p < 0.05). In response to exercise, there was an additional

significant fall in muscle glycogen concentration. During the first hour of

recovery without food, muscle glycogen levels increased significantly, and in

response to subsequent feeding total glycogen levels increased further,

reaching above pre-exercise fed values after 24 hours.

In response to an overnight fast and exercise, ASG levels decreased

significantly and accounted almost completely for the decrease in total glycogen

concentrations, with AIG levels changing little or remaining at relatively stable

levels (Figure 4.2). During the first hour of recovery without food and

subsequent recovery period with food, ASG levels increased significantly to

reach levels higher than those measured in pre-exercised fed animals. The

increase in ASG levels during that time accounted almost completely for the rise

in total glycogen levels, with only a marginal rise in AIG levels (Figure 4.2).

124

Figure 4.2 Changes in ASG and AIG in response to a 3-minute bout of

intense exercise and recovery. The values shown represent

means ± S.E.M. (n = 7) and are expressed in millimoles glucosyl

units per kilogram dry tissue weight. a, significantly different to

Fed levels (p < 0.05). b, significantly different to Post-ex levels (p

< 0.05). c, significantly different to Pre-ex levels (p < 0.05). d,

significantly different to 15 min rec levels (p < 0.05). e, significantly

different to 60 min rec levels (p < 0.05).

0

50

100

150

200

250

Fed Pre-ex Post-ex 15min rec 60min rec 24h rec

Gly

cog

en

(m

mo

l k

g-1

d.w

.)

ASG

AIGa

a

a,b

a,c

a,b,c,d,e

b

125

4.3.2 Effect of exercise and re-feeding on the molecular size

distribution of AIG and ASG.

Prior to exercise, the molecular size distributions of AIG and ASG were similarly

distributed with average sizes of 27.4 ± 0.49 nm (skewness = -0.42 ± 0.24) and

28.8 ± 0.53 nm (skewness = -0.55 ± 0.24), respectively (Figure 4.3A).

Immediately following exercise, the molecular size distribution of AIG was found

to be relatively unaffected (skewness = -0.41 ± 0.24), with only a small

significant decrease in the average particle size to 24.4 ± 0.54 nm (Figure

4.3B). In contrast, the ASG molecular size distribution was significantly shifted

to the left (skewness = 0.11 ± 0.24), with a high and low molecular size peaks at

10-12 nm and 22-24 nm, and a higher proportion of smaller glycogen particles

than before exercise. This was accompanied by a marked decrease in average

molecular size to 20.0 ± 0.65 nm (Figure 4.3B). After 24 hours of recovery, the

molecular size distribution of AIG was associated with a small increase in the

proportion of larger glycogen particles (skewness = -0.67 ± 0.24), and with a

significant increase in average molecular size to 30.7 ± 0.60 nm (Figure 4.3C).

Similarly, the average molecular size of ASG increased significantly to 32.2 ±

0.55 nm, this being higher than average molecular size before exercise, with a

molecular size distribution profile similar to that of the AIG fraction (skewness =

-0.62 ± 0.24; Figure 4.3C).

126

Figure 4.3 Molecular size distributions of A) AIG and B) ASG pre-

exercise, post-exercise and after 24 hours of recovery. The

values shown for each particle diameter are expressed as an

average percentage ± S.E.M. of total number of particles

measured for each condition.

0%

5%

10%

15%

20%

25%

0 10 20 30 40 50 60

% o

f to

tal g

lyco

ge

n p

art

icle

s

Particle size (nm)

AIG Pre-ex

AIG Post-ex

AIG 24h rec

A

0%

5%

10%

15%

20%

25%

0 10 20 30 40 50 60

% o

f to

tal g

lyco

ge

n p

art

icle

s

Particle size (nm)

ASG Pre-ex

ASG Post-ex

ASG 24h rec

B

127

4.4 Discussion

Given that ASG and AIG in resting skeletal muscle have both a similar average

molecular size and pattern of molecular size distribution, and that ASG from

homogenised muscle extracts is by far the most responsive fraction to changes

in glycogen levels, we examined whether the molecular size distributions of AIG

and ASG also respond differently to changes in muscle glycogen levels. This

was achieved by examining the effect of exercise and recovery on the

molecular size distributions of AIG and ASG in rats. Here we show that although

the molecular size distributions of both AIG and ASG were similar prior to

exercise, that of ASG shifted markedly towards glycogen particles of smaller

sizes immediately after exercise due to the conversion of larger into smaller

glycogen particles, whereas the molecular size distribution of AIG changed little.

After 24 hours of recovery with food, the molecular size distribution of ASG

shifted to the right, with no marked changes in that of AIG, a finding consistent

with the observation that most of the changes in total glycogen levels during

that time were accounted for by ASG. Such different responses of AIG and ASG

to changes in glycogen levels suggest that these glycogen species correspond

to physiologically distinct populations of glycogen.

Although this is the first study to examine the effect of exercise and recovery on

the molecular size distribution of AIG and ASG, our results for ASG share some

similarities with those on total glycogen by Marchand and colleagues (2007).

Since in their study there were no data on resting participants prior to exercise,

this prevents any comparison with our pre-exercise results. However, the

molecular size distribution and average molecular size were similar to those

reported in the literature (Drochmans, 1962; Scott & Still, 1968; Meyer et al.,

128

1970; Schmalbruch & Kamieniecka, 1974; Rybicka, 1981b, a; Marchand et al.,

2002; Ryu et al., 2009). In agreement with our findings, Marchand and

colleagues (2007) reported an increase in the average molecular size of total

glycogen during recovery from exercise together with a rightward shift in the

molecular size distribution of glycogen. There were, however, some noticeable

differences between both studies, with the average particle sizes reported in

their study being considerably smaller after exercise than reported here

(Marchand et al., 2007). This may be due either to the use of human

participants instead of rats as some interspecies differences may exist or to the

much longer duration and intensity of their exercise protocol leading to a larger

depletion of muscle glycogen stores.

The responses of ASG and AIG levels to exercise and recovery reported here

corroborate those published by others and are consistent with the patterns of

change in the molecular size distribution of these glycogen fractions. Indeed,

under conditions where repeated homogenisations are performed to extract

glycogen, we found in Chapter 2 that ASG is the most responsive fraction to

exercise and recovery in humans, with AIG remaining at near stable levels

(Barnes et al., 2009). Similar findings have also been reported for a broad range

of conditions affecting muscle glycogen levels, such as adrenaline

administration, electro-stimulation and fasting (Bloom et al., 1951; Bloom &

Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom, 1955, 1956; Stetten

et al., 1958; James et al., 2008). Also, consistent with the marked changes in

ASG levels and near stable concentrations of AIG reported here during and

after exercise, the changes in ASG concentration were accompanied by

corresponding changes in ASG average molecular sizes and molecular size

129

distribution. The relatively stable concentration of AIG across all conditions was

accompanied by only minor shift in the molecular size distribution of this

glycogen fraction.

It is important to stress that the absence of marked changes in the levels and

molecular size distribution of AIG in response to exercise and recovery does not

imply that this glycogen population is metabolically inert compared to ASG.

Indeed, the many studies that have examined the pattern of isotopic labelling of

ASG and AIG in tissues incubated in the presence of radio-labelled glucose

have reported that the AIG fraction is highly active, incorporating more rapidly

new glucose residues than ASG (Stetten et al., 1958; Krisman, 1973; Krisman &

Barengo, 1975; Curtino et al., 1979; Aon & Curtino, 1984; Pitcher et al., 1987;

Lacoste et al., 1990; Huang et al., 1997) although AIG levels change little under

these conditions. To explain these findings, it has been proposed that the AIG

fraction may be in a constant state of flux, with glycogen molecules migrating

from AIG to form ASG or vice versa as glycogen is synthesised or degraded,

respectively. This means that glucose would be initially incorporated into AIG

before glucose or AIG is transferred to the ASG fraction. This would enable the

AIG fraction to remain stable despite actively incorporating new glucose while

ASG is increasing in size. This interpretation is supported by the observation

that the AIG-incorporated glucose can translocate to the ASG fraction (Curtino

et al., 1979; Aon & Curtino, 1984; Lacoste et al., 1990).

Given the evidence that the levels and molecular size distributions of AIG and

ASG respond differently to exercise and re-feeding, this raises the obvious

question of the mechanisms underlying these differences. Since each glycogen

particle is non-covalently bound to a number of proteins to form a complex

130

known as glycosome, AIG and ASG might correspond to different sub-fractions

of glycosomes, each with a distinct complement of proteins (Rybicka, 1996;

Skurat & Roach, 2004) and a different response to signals promoting the

synthesis or breakdown of glycogen. Moreover, given that some glycosomes

are in turn non-covalently bound to membranes such as those of the

sarcoplasmic reticulum (SR) via the muscle glycogen-binding regulatory subunit

of PP1 (Rybicka, 1996) maybe this is a factor that explains the different

responses of AIG and ASG to changes in glycogen levels. Finally, since distinct

sub-cellular glycogen stores have been reported to respond differently to

changes in glycogen levels based on their location and cellular associations

(Prats et al., 2005; Marchand et al., 2007; Nielsen et al., 2009; Prats et al.,

2009), this may also contribute to the differences between ASG and AIG

responses. Clearly, the structural feature explaining the difference in behaviours

between ASG and AIG remains to be determined.

In conclusion, this study corroborates our earlier findings in Chapters 2 and 3

that ASG accounts for most of the changes in glycogen concentration and that

both AIG and ASG have a similar molecular size and distribution at rest. What

we have demonstrated here for the first time is that it is the molecular size

distribution and average molecular size of ASG that respond the most to

changes in muscle glycogen levels, with AIG fraction being little affected.

Clearly, further investigation is required to improve our understanding of the

mechanism underlying the different responses of AIG and ASG to changes in

muscle glycogen levels, an important issue given the possibility that the

behaviour of these glycogen fractions might reflect an important but poorly

understood aspect of glycogen metabolism in health and disease.

131

Chapter 5

General Discussion

132

5.1 General discussion

Since the pioneering work of Claude Bernard on glycogen over one and a half

centuries ago, much has been learned about glycogen’s structure, its ultra

structural organisation, and the regulation of its metabolism in health and

disease. As described in Chapter 1, this journey was not without several

challenges and hurdles, with many features of glycogen’s structure and

functions taking several decades before their exposition. One such challenge

was the elucidation of the mechanisms underlying the observation made by

Willstatter and Rohdewald in 1934 that glycogen separates into an AIG and

ASG fraction when extracted in the presence of acid (Willstatter & Rohdewald,

1934). It took almost 60 years before this phenomenon was explained at the

molecular level. Now, it is generally accepted that AIG corresponds to a

population of very small glycogen particles with their low acid-solubilities

explained by the high protein (glycogenin) to glucosyl ratio of the glycogen

granules (Lomako et al., 1991a; Lomako et al., 1993a; Shearer et al., 1999;

Graham et al., 2001; Shearer et al., 2001; Marchand et al., 2002; Marchand et

al., 2007). Since it was also proposed that AIG is an intermediate along the

pathway of ASG synthesis, AIG and ASG were referred to as PG and MG,

respectively (Lomako et al., 1993a). Not surprisingly, this led to a considerable

volume of research aimed at elucidating the physiological significance and

interrelationship between PG and MG in skeletal muscles (Adamo et al., 1998a;

Adamo et al., 1998b; Asp et al., 1999; Derave et al., 2000; Shearer et al., 2000;

Graham et al., 2001; Shearer et al., 2001; Rosenvold et al., 2003; Battram et

al., 2004; Shearer et al., 2005a; Shearer et al., 2005b; Wee et al., 2005; Devries

et al., 2006; Marchand et al., 2007; Wilson et al., 2007). Although the PG/MG

133

model has been extensively used to explain both the different acid-solubilities of

glycogen and the different responses of AIG and ASG to changes in muscle

glycogen levels, this thesis shows that this model does not hold anymore for

muscle glycogen and must be abandoned, thus not only leaving open the

question of the molecular mechanisms underlying the acid solubility of muscle

glycogen, but also making this issue again one of the oldest unresolved puzzle

in glycogen biochemistry.

The possibility that the PG/MG model may not be adequate was raised for the

first time following our close examination of the literature which revealed that

the acid-solubility of glycogen might be dependent on the experimental

condition adopted to extract glycogen. Indeed, in all studies where skeletal

muscles are homogenised to acid-extract glycogen, ASG is the predominant

fraction, and it is the most responsive fraction to changes in muscle glycogen

levels, with AIG remaining at near stable levels under most conditions (Bloom et

al., 1951; Bloom & Knowlton, 1953; Bloom & Russell, 1955; Russell & Bloom,

1955, 1956; Stetten et al., 1958; James et al., 2008). In contrast, in all studies

that do not include a homogenisation step to acid-extract muscle glycogen, AIG

rather that ASG is the most abundant and metabolically responsive glycogen

fraction, except when total muscle glycogen levels are elevated. It stands to

reason, therefore, that if the acid-solubility of a glycogen particle were to be

governed solely by its size relative to that of glycogenin as proposed by the

PG/MG model, the extraction yields of muscle AIG and ASG should not be

affected by the extraction protocol. However, since the aforementioned studies

were performed on different animal species and by different research teams,

with no study comparing directly the effect of both acid-extraction protocols on

134

the yields of AIG and ASG, it was the first aim of this thesis to determine

whether the extraction yields of AIG and ASG and their patterns of response to

changes in muscle glycogen levels are affected by the protocol adopted to acid-

extract glycogen.

In agreement with our interpretation of the literature, one major finding of this

thesis is that the extraction yields of AIG and ASG are highly dependent upon

the protocol adopted to extract glycogen. We showed in Chapter 2 that when

repeated muscle homogenisations are performed, most of the glycogen is

extracted as ASG. Also, we found that this is the most responsive fraction to

changes in muscle glycogen levels such as those brought about by exercise

and re-feeding post-exercise, with AIG levels remaining stable under these

conditions (Barnes et al., 2009). In contrast, when a homogenisation step is not

included to acid-extract muscle glycogen, AIG rather that ASG is the most

abundant and metabolically responsive fraction of glycogen, with AIG

accounting for most of the changes in muscle glycogen levels in response to

exercise and recovery, but not when muscle glycogen levels are elevated at

which point ASG accounts to a far greater extent for the changes in glycogen

levels (Chapter 2).

If the only factor determining the extraction yields of ASG and AIG was the size

of the carbohydrate moiety of glycogen relative to glycogenin, as originally

proposed by Lomako and colleagues (1991a), the levels of AIG and ASG and

their responses to changes in muscle glycogen levels should not be affected by

the extraction protocol unless the glycogen particles themselves are altered.

Indeed, the use of extensive homogenisation may physically damage the

glycogen granule, releasing acid-soluble fragments of glucosyl residues that

135

may cause an overestimation of ASG. This, however, was not the case here, as

our results in Chapter 3 reveal that the elution profile as well as the molecular

size distribution of glycogen molecules are unaffected by repeated

homogenisations, reaffirming our interpretation that the extraction yields and

patterns of response of AIG and ASG are highly dependent on the extraction

protocol. Thus, it follows that at least one of the extraction protocols examined

here most probably separates glycogen on the basis of factors other than

molecular size differences.

Since the study of Lomako and colleagues (1991a) at the origin of the concept

that glycogen solubility in acid is primarily determined by its size was performed

using homogenised tissue extracts (Lomako et al., 1991a), and that repeated

homogenisations do not affect the molecular size of glycogen (Chapter 3), these

observations might be taken as evidence that only the results obtained from

muscles subjected to repeated homogenisations are likely to be explained by

the PG/MG model. This leaves unanswered the question about the nature of

AIG and ASG obtained when glycogen is extracted without a homogenisation

step. We proposed in Chapter 2 that the poor yield of ASG when a

homogenisation step is not included to extract glycogen might be due to some

of the glycogen precipitating not because of its poor-acid solubility per se, but

simply because it is trapped within the dense mesh of undisrupted myofibrils

that precipitate during centrifugation in the presence of acid. This, in turn, could

result in the contamination of AIG by ASG and a serious overestimation of the

proportion of AIG and corresponding underestimation of ASG levels (James et

al., 2008; Barnes et al., 2009). In addition, the patterns of change in ASG and

AIG levels using this extraction protocol could be explained on the grounds that

136

when total muscle glycogen levels are low or moderate, the acid extraction of

glycogen without a homogenisation step could result in the liberation of only a

small proportion of the pool of ASG from the mesh of poorly disrupted muscle

cells, with the resulting AIG contaminated by ASG accounting for most of the

changes in total muscle glycogen levels. However, when total glycogen

increases to levels that exceed the capacity of this mesh of muscle myofibrils to

trap glycogen as effectively, a disproportionate rise in the release of ASG is

expected to occur with an increase in glycogen content as is reported here and

other studies (Adamo & Graham, 1998; Asp et al., 1999; Derave et al., 2000;

Battram et al., 2004; Marchand et al., 2007). It is important to note, however,

that our findings do not exclude the possibility that the AIG and ASG obtained

without a homogenisation step could still represent physiologically relevant

pools of glycogen particles with different average molecular weight. However,

unless the molecular sizes of AIG and ASG are known, the issue of which of the

two extraction protocols compared in Chapter 2 results in the separation of

glycogen in a manner which is dependent on size will remain unclear.

Given that prior to the work described in this thesis the molecular sizes of AIG

and ASG had never been examined in skeletal muscles, in part because of the

difficulty of extracting AIG from acid-insoluble muscle pellet, it remained to be

determined which of the two aforementioned extraction protocols generates an

ASG and AIG fraction that behaves as described by the PG/MG model. For this

reason, the next main objective of this thesis was to develop a protocol for

removing AIG from the acid-insoluble protein pellet while keeping glycogen’s

structure intact for molecular size analyses. As described in Chapter 3, this was

successfully achieved by neutralising the protein pellet and incubating it with the

137

non-specific protease, pronase, to disrupt this pellet and liberate over 99% of

the AIG. It was also necessary to include in the incubation buffer a glucosidase

inhibitor, acarbose, to prevent partial glycogen digestion during the pronase

digestion step (Aon & Curtino, 1984). As a result, we showed that this extraction

protocol is without any effect on the molecular size distribution of glycogen and

provides high quality AIG and ASG preparations suitable for the analyses of

their molecular sizes by transmission electron microscopy.

Using transmission electron microscopy analyses to compare ASG and AIG, we

made the important but unexpected finding in Chapter 3 that regardless of

whether a homogenisation step is performed to acid-extract muscle glycogen,

AIG and ASG in resting muscles have a similar average molecular size normally

distributed over a similar range of particle sizes. This is an important finding

because it indicates that the acid-insolubility of AIG is not due to its molecular

size, as assumed in the recent scientific literature. Moreover, this finding not

only corroborates the work of Skurat and colleagues (1997) who reported that

AIG is not the discrete low molecular weight 400 kDa glycogen particle originally

proposed by Lomako and colleagues (1991a), but also challenges the currently

held view that AIG corresponds to a population of low molecular weight

glycogen species (Marchand et al., 2002; Marchand et al., 2007). Although an

earlier study from our laboratory also provided evidence that AIG and ASG

obtained from homogenised muscle extracts have a similar molecular weight,

the gel chromatography method adopted in that study was only adequate for the

detection of large differences in molecular weight and unsuitable for the

detection of small differences, thus making the current study to first one to have

compared the molecular size distribution of AIG and ASG.

138

Given that during recovery from exercise, the accompanying rise in glycogen

concentrations is accompanied by an increase in the average molecular size of

glycogen (Marchand et al., 2007), this raised the possibility that the molecular

sizes of AIG and ASG are also affected by changes in muscle glycogen levels.

Since the responses of the molecular sizes of these glycogen fractions to

changes in glycogen levels have never been examined before, our next and last

objective was to examine the responses of the molecular sizes of AIG and ASG

to the changes in muscle glycogen levels brought about by exercise and post-

exercise re-feeding in muscles extracted with repeated homogenisations. With

the help of the AIG extraction protocol developed in Chapter 3, we found that

the relatively stable concentration of AIG during exercise and re-feeding was

accompanied by minimal changes in the molecular size distribution profile of

this glycogen fraction (Chapter 4). In contrast, the marked fall in the

concentration of ASG post-exercise was accompanied by a marked decrease in

the average molecular size of ASG and a leftward shift in its molecular size

distribution towards smaller molecules, whereas the converse was observed

with ASG during recovery (Chapter 4). Given that the average molecular size of

ASG fell below that of AIG during exercise, this response is the exact opposite

to what is predicted by the PG/MG model where AIG rather than ASG

corresponds to the low molecular weight glycogen species. These results thus

challenge further the validity of the PG/MG model.

Given the evidence that the levels and molecular size distributions of AIG and

ASG respond differently to changes in muscle glycogen levels, with both having

similar molecular size prior to exercise, this raises the obvious question of the

mechanisms underlying the different acid solubilities of these pools of glycogen

139

and their different responses to changes in glycogen levels. Although the

mechanisms underlying the different acid-solubilities and behaviours of AIG and

ASG now remain to be determined, our findings do not support the notion that

the size of the glycogen particle via the glycogenin-to-glucosyl ratio in glycogen

determines glycogen’s solubility in acid and its response to changes in glycogen

levels since AIG and ASG have a similar average molecular size, but different

acid-solubilities and responses to changes in glycogen levels (Lomako et al.,

1991a). Moreover, since glycogenin is typically embedded inside each glycogen

particle, this protein should have no or little impact on the acid solubility of

glycogen.

As discussed in Chapter 4, the acid insolubility and absence of marked changes

in the levels and average molecular size of AIG is unlikely the result of this

glycogen pool corresponding to the fraction of muscle glycogen entrapped

inside lysosomes, although lysosomal glycogen is not metabolised by GP or GS

and remains stable in response to rapid changes in muscle glycogen levels

(Hers & Van Hoof, 1973). This is because lysosomal glycogen accounts for only

6% of the glycogen stored in rat muscles (Calder & Geddes, 1989a). Moreover,

the homogenisation protocol adopted here is considerably harsher than the

protocols routinely used to disrupt lysosomes (Calder & Geddes, 1989a).

Finally, there is strong evidence that AIG is a highly metabolically active pool of

glycogen, with new glucose residues being more readily incorporated into the

AIG than the ASG fraction during glycogen synthesis (Stetten et al., 1958;

Krisman, 1973; Krisman & Barengo, 1975; Curtino et al., 1979; Aon & Curtino,

1984; Pitcher et al., 1987; Lacoste et al., 1990; Huang et al., 1997).

140

As discussed in Chapter 4, it is more likely that AIG and ASG correspond to

different sub-fractions of glycosomes, each with a distinct complement of

proteins associated with each glycogen particle (Rybicka, 1996; Skurat &

Roach, 2004) and a different solubility and response to physiological stimuli. In

addition, the binding of these glycosomes to membranes or others ultra

structural components of the cell may also alters both their solubility in acid and

responses to changes in glycogen levels. In support of this view, is the

observation that the SR-glycogen complex in cardiac muscle has been shown

to be resistant to dissociation in the presence of acidic uranyl acetate (Rybicka,

1979, 1981b, a), thus making it a potential AIG candidate. Furthermore, the

proteins of the glycosomes can also bind to cellular structures such as actin and

desmin, thus locating the glycosomes in the cell and affecting their physiological

responses to changes in glycogen levels (Graham et al., 2010). Indeed, distinct

sub-cellular glycogen stores have been reported to respond differently to

changes in glycogen levels based on their cellular location and associations

(Prats et al., 2005; Marchand et al., 2007; Nielsen et al., 2009; Prats et al.,

2009), thus making it likely that ASG and AIG may represent glycogen particles

located in different compartments inside the muscle cell (Marchand et al., 2002;

Marchand et al., 2007). In this regard, it is noteworthy that glycogen degradation

during exercise occurs preferentially near the contractile filaments (Friden et al.,

1985; Marchand et al., 2007), whereas net glycogen synthesis immediately after

exercise is greatest in the sub-sarcolemma region (Marchand et al., 2007).

It is important to note that although the PG/MG model is refuted by our finding,

our results do not exclude the possibility that a precursor-product relationship

exists between AIG and ASG. In support of this view, pulse chase experiments

141

have shown that as glycogen synthesis continues, the glucose residues

incorporated into AIG eventually translocate to the ASG fraction (Curtino et al.,

1979; Aon & Curtino, 1984; Lacoste et al., 1990), thus allowing the absolute

concentration of glucose in the AIG fraction to remain relatively stable whilst

continuing to incorporate new glucose residues (Huang et al., 1997). This

suggests that the AIG fraction may be in a constant state of flux, with glycogen

molecules migrating to and from the AIG fraction as glycogen is synthesised

and degraded, respectively, with changes in molecular size and concentration

restricted mainly to the ASG fraction. In this respect, skeletal muscle glycogen

may be subjected to a structured and ordered process of synthesis and

degradation similar to that reported in liver (Garcia-Rocha et al., 2001;

Fernández-Novell et al., 2002; Ferrer et al., 2003; Ros et al., 2009) whereby

each AIG granule converted to ASG is replaced by smaller glycogen granules

so glucose incorporation can continue at this site without any net increase in

associated glycogen concentrations. Clearly, more work is required to explain

the relationship between AIG and ASG.

In conclusion, this thesis refutes the PG/MG model that AIG and ASG are

glycogen fractions of different molecular sizes corresponding to PG and MG,

respectively, with AIG being comprised of small glycogen particles as well as

being the most abundant glycogen fraction at rest and the most responsive to

changes in total glycogen levels (Lomako et al., 1991a; Lomako et al., 1993a;

Adamo et al., 1998a; Adamo et al., 1998b; Asp et al., 1999; Derave et al., 2000;

Shearer et al., 2000; Graham et al., 2001; Shearer et al., 2001; Rosenvold et

al., 2003; Battram et al., 2004; Shearer et al., 2005a; Shearer et al., 2005b;

Wee et al., 2005; Devries et al., 2006; Marchand et al., 2007; Wilson et al.,

142

2007). In particular, the PG/MG model is challenged by our findings that AIG

does not correspond to a population of low molecular size glycogen particles as

both AIG and ASG have the same average molecular size at rest. The PG/MG

model is further invalidated by the observation that when muscle glycogen

levels are low, it is the average molecular size of ASG that is lower than that of

AIG rather than the converse as predicted by the PG/MG model. Finally, we

show that most of the glycogen in skeletal muscle of humans and rats exists as

ASG and that it is this fraction instead of AIG that is the most responsive to

changes in total glycogen levels. For these reasons, we propose not only that

the PG/MG model should be abandoned, but also that the terms PG and MG

should be replaced by the theory-neutral terms AIG and ASG. It is important to

note that although the refutation of the PG/MG model brings back to life a

puzzle that was believed to have been solved almost two decades ago, this

thesis brings us one step closer to solving it with the finding that the acid

solubility of glycogen is not determined by its size. Clearly, more research is

required to elucidate what is most probably one of the oldest unresolved

enigmas in carbohydrate biochemistry. We believe it is important to do so

considering the possibility that the behaviours of AIG and ASG might reflect

those of important but poorly understood pools of glycogen.

143

Chapter 6

References

144

6.1 References

Adamo KB & Graham TE. (1998). Comparison of traditional measurements with macroglycogen and proglycogen analysis of muscle glycogen. Journal of Applied Physiology 84, 908-913.

Adamo KB, Tamopolsky MA & Graham TE. (1998a). Effects of high and low carbohydrate diet on the synthesis of proglycogen and macroglycogen in human skeletal muscle. In American College of Sports Medicine 45th Annual Meeting, pp. 258. Williams & Wilkins, Orlando, Florida.

Adamo KB, Tarnopolsky MA & Graham TE. (1998b). Dietary carbohydrate and postexercise synthesis of proglycogen and macroglycogen in human skeletal muscle. American Journal of Physiology 275, E229-234.

Akai H, Yokobayashi K, Misaki A & Harada T. (1971). Complete hydrolysis of branching linkages in glycogen by Pseudomonas isoamylase: distribution of linear chains. Biochimica Biophysica Acta 237, 422-429.

Alonso MD, Lomako J, Lomako WM & Whelan WJ. (1995). A new look at the biogenesis of glycogen. FASEB Journal 9, 1126-1137.

Aon MA & Curtino JA. (1984). Evidence for the glycoprotein nature of retina glycogen. European Journal of Biochemistry 140, 557-566.

Aon MA & Curtino JA. (1985). Protein-bound glycogen is linked to tyrosine residues. Biochemistry 229, 269-272.

Armstrong CG, Browne GJ, Cohen P & Cohen PTW. (1997). PPP1R6, a novel member of the family of glycogen-targetting subunits of protein phosphatase 1. FEBS Letters 418, 210-214.

Arnall DA, Nelson AG, Quigley J, Lex S, Dehart T & Fortune P. (2007). Supercompensated glycogen loads persist 5 days in resting trained cyclists. European Journal of Applied Physiology 99, 251-256.

Asp S, Daugaard JR, Rohde T, Adamo K & Graham T. (1999). Muscle glycogen accumulation after a marathon: roles of fiber type and pro- and macroglycogen. Journal of Applied Physiology 86, 474-478.

Baqué S, Guinovart JJ & Ferrer JC. (1997). Glycogenin, the primer of glycogen synthesis, binds to actin. FEBS Letters 417, 355-359.

Barker CC, Hirst EL & Young FG. (1941). Linkage between the repeating units in the starch molecule. Nature 147, 296.

Barnes PD, Singh A & Fournier PA. (2009). Homogenization-dependent responses of acid-soluble and acid-insoluble glycogen to exercise and refeeding in human muscles. Metabolism, Clinical and Experimental 58, 1832-1839.

145

Battram DS, Shearer J, Robinson D & Graham TE. (2004). Caffeine ingestion does not impede the resynthesis of proglycogen and macroglycogen after prolonged exercise and carbohydrate supplementation in humans. Journal of Applied Physiology 96, 943-950.

Bechet D, Tassa A, Taillandier D, Combaret L & Attaix D. (2005). Lysosomal proteolysis in skeletal muscle. International Journal of Biochemistry and Cell Biology 37, 2098-2114.

Bell DJ. (1937). Glycogen; VI. The molecular structure of horse muscle glycogen. Biochemistry 31, 1683-1691.

Bergmeyer HU. (1974). Methods of enzymatic analysis, vol. 4. Academic Press, New York.

Bergstrom J. (1962). Muscle electrolytes in man. Determined by neutron activation analysis on needle biopsy specimens. A study on normal subjects, kidney patients, and patients with chronic diarrhoea. Scandinavian Journal of Clinical Laboratory Investigation 14, 110.

Bernard C. (1843). Du suc gastrique et de son role dans la nutrition. In Bailliere. Paris.

Bernard C. (1855). Sur le mecanisme de la formation du sucre dans le foie. Comptes rendus des seances de l'Academie des Sciences 41, 461-469.

Bernard C. (1857). Nouvelles recherches expérimentales sur les phénomènes glycogéniques du foie. Mémoires de la Société de biologie et de ses filiales 4, 1-7.

Bloom WL & Knowlton GC. (1953). Muscle glycogen fractions during stimulation. American Journal of Physiology 173, 545-546.

Bloom WL, Lewis GT, Schumpert MZ & Shen TM. (1951). Glycogen fractions of liver and muscle. Journal of Biological Chemistry 188, 631-636.

Bloom WL & Russell JA. (1955). Effects of epinephrine and of norepinephrine on carbohydrate metabolism in rat. American Journal of Physiology 183, 356-364.

Bogdanis GC, Nevill ME, Boobis LH & Lakomy HK. (1996). Contribution of phosphocreatine and aerobic metabolism to energy supply during repeated sprint exercise. Journal of Applied Physiology 80, 876-884.

Bogdanis GC, Nevill ME, Boobis LH, Lakomy HK & Nevill AM. (1995). Recovery of power output and muscle metabolites following 30 s of maximal sprint cycling in man. Journal of Physiology 482, 467-480.

Bogdanis GC, Nevill ME, Lakomy HK & Boobis LH. (1998). Power output and muscle metabolism during and following recovery from 10 and 20 s of maximal sprint exercise in humans. Acta Physiologica Scandinavica 163, 261-272.

146

Bollen M, Keppens S & Stalmans W. (1998). Specific features of glycogen metabolism in the liver. Biochemical Journal 336, 19-31.

Brady MJ & Saltiel AR. (2001). The Role of Protein Phosphatase-1 in Insulin Action. Recent Progress in Hormone Research 56, 157-174.

Brau L, Ferreira LDMCB, Nikolovski S, Raja G, Palmer TN & Fournier PA. (1997). Regulation of glycogen synthase and phosphorylase during recovery from high-intensity exercise in the rat. Biochemical Journal 322, 303-308.

Brojer JT, Stampfli HR & Graham TE. (2002a). Analysis of proglycogen and macroglycogen content in muscle biopsy specimens obtained from horses. American Journal of Veterinary Research 63, 570-575.

Brojer JT, Stampfli HR & Graham TE. (2002b). Effect of extraction time and acid concentration on the separation of proglycogen and macroglycogen in horse muscle samples. Canadian Journal of Veterinary Research 66, 201-206.

Brown DH, Illingworth B & Kornfeld R. (1965). Transfer of glucosyl units to oligosaccharides and polysaccharides by the action of uridine diphosphoglucose-α-glucan transglucosylase. Biochemistry 4, 486-495.

Calder PC & Geddes R. (1986). Digestion of the protein associated with muscle and liver glycogens. Carbohydrate Research 148, 173-177.

Calder PC & Geddes R. (1989a). Rat skeletal muscle lysosomes contain glycogen. International Journal of Biochemistry 21, 561-567.

Calder PC & Geddes R. (1989b). Regulation of lysosomal glycogen metabolism: studies of the actions of mammalian acid alpha-glucosidases. International Journal of Biochemistry 21, 569-576.

Campbell DG & Cohen P. (1989). The amino acid sequence of rabbit skeletal muscle glycogenin. European Journal of Biochemistry 185, 119-125.

Carrizo ME, Romero JM, Miozzo MC, Brocco M, Panzetta P & Curtino JA. (1997). Biosynthesis of proteoglycogen: Modulation of glycogenin expression in the developing chicken. Biochemical and Biophysical Research Communications 240, 142-145.

Carter SR & Record BR. (1939). The osmotic pressure of solutions of polysaccharide derivatives. Part II. The osmotic pressure of derivatives of lichenin, inulin, glycogen, xtarch, and starch dextrin. Journal of the Chemical Society, 664-675.

Cheetham ME, Boobis LH, Brooks S & Williams C. (1986). Human muscle metabolism during sprint training. Journal of Applied Physiology 61, 54-60.

147

Cid E, Cifuentes D, Baqué S, Ferrer JC & Guinovart JJ. (2005). Determinants of the nucleocytoplasmic shuttling of muscle glycogen synthase. FEBS Letters 272, 3197-3213.

Cohen P. (1978). The role of cyclic-AMP-dependent protein kinase in the regulation of glycogen metabolism in mammalian skeletal muscle. Current Topics in Cellular Regulation 14, 117-196.

Cohen P. (1989). The structure and regulation of protein phosphatases. Annual Review of Biochemistry 58, 453-508.

Cori CF, Colowick SP & Cori GT. (1937). The isolation and synthesis of glucose-1-phosphoric acid. Journal of Biological Chemistry 121, 465-477.

Cori CF, Schmidt G & Cori GT. (1939). The Synthesis of a Polysaccharide from Glucose-1-Phosphate in Muscle Extract. Science 89, 464-465.

Cori GT, Colowick SP & Cori CF. (1938). The formation of glucose-1-phosphoric acid and in extracts of mammalian tissues and of yeast. Journal of Biological Chemistry 123, 375-380.

Cori GT & Cori CF. (1939). The activating effect of glycogen on the enzymatic synthesis of glycogen from glucose-1-phosphate. Journal of Biological Chemistry 131, 397-398.

Cori GT & Cori CF. (1943). Crystalline muscle phosphorylase IV. Formation of glycogen. Journal of Biological Chemistry 151, 57-63.

Cori GT & Larner J. (1951). Action of amylo-1,6-glucosidase and phosphorylase on glycogen and amylopectin. Journal of Biological Chemistry 188, 17-29.

Craig SAS, McDonald AML, Manners DJ & Stark JR. (1988). The iodine-staining properties and fine structure of some mammalian and invertebrate glycogens. Carbohydrate Research 179, 327-370.

Curtino JA & Lacoste ER. (2000). Two glycogen synthase activities associated with proteoglycogen in retina. Neurochemical Research 25, 129-132.

Curtino JA, Maccioni HJ & Caputto R. (1979). The incorporation of glucose into α-1,4-glucan proteins of bovine membranes. Journal of Neurochemistry 33, 139-147.

Derave W, Gao S & Richter EA. (2000). Pro- and macroglycogenolysis in contracting rat skeletal muscle. Acta Physiologica Scandinavica 169, 291-296.

Devos P, Baudhuin P, Van Hoof F & Hers HG. (1983). The alpha particulate liver glycogen. A morphometric approach to the kinetics of its synthesis and degradation. Biochemical Journal 209, 159-165.

Devries MC, Hamadeh MJ, Phillips SM & Tarnopolsky MA. (2006). Menstrual cycle phase and sex influence muscle glycogen utilization and glucose

148

turnover during moderate-intensity endurance exercise. American Journal of Physiology - Regulatory Integrative & Comparative Physiology 291, 1120-1128.

Drochmans P. (1962). Morphology of glycogen. Electron microscopic study of the negative stains of particulate glycogen. Journal of Ultrastructure Research 6, 141-163.

Essen B. (1978). Studies on the regulation of metabolism in human skeletal muscle using intermittent exercise as an experimental model. Acta Physiologica Scandinavica Supplementum 454, 1-32.

Fernández-Novell JM, Bellido D, Vilaró S & Guinovart JJ. (1997). Glucose induces the translocation of glycogen synthase to the cell cortex in rat hepatocytes. Biochemical Journal 321, 227-231.

Fernández-Novell JM, López-Iglesias C, Ferrer JC & Guinovart JJ. (2002). Zonal distribution of glycogen synthesis in isolated rat hepatocytes. FEBS Letters 531, 222-228.

Fernandez-Sanchez ME, Criado-Garcıa O, Heath KE, Garcıa-Fojeda B, Medrano-Fernandez I, Gomez-Garre P, Sanz P, Serratosa JM & de Cordoba SR. (2003). Laforin, the dual-phosphatase responsible for Lafora disease, interacts with R5 (PTG), a regulatory subunit of protein phosphatase-1 that enhances glycogen accumulation. Human Molecular Genetics 12, 3167-3171.

Ferreira LDMCB, Palmer TN & Fournier PA. (1998). Prolonged exposure to halothane and associated changes in carbohydrate metabolism in rat muscles in vivo. Journal of Applied Physiology 84, 1470-1474.

Ferrer JC, Favre C, Gomis RR, Fernández-Novell JM, García-Rocha M, Iglesia Ndl, Cid E & Guinovart JJ. (2003). Control of glycogen deposition. FEBS Letters 546, 127-132.

Fletterick RJ & Madsen NB. (1980). The Structures and Related Functions of Phosphorylase a. Annual Review of Biochemistry 49, 31-61.

Fong NM, Jensen TC, Shah AS, Parekh NN, Saltiel AR & Brady MJ. (2000). Identification of binding sites on protein targeting to glycogen for enzymes of glycogen metabolism. Journal of Biological Chemistry 275, 35034-35039.

French D. (1964). Structure of glycogen and its amylolytic degradation. In Control of glycogen metabolism, ed. Whelan WJ & Cameron MP, pp. 7-28. J. & A. Churchill, Ltd, London.

Friden J, Seger J & Ekblom B. (1985). Implementation of periodic acid-thiosemicarbazide-silver proteinate staining for ultrastructural assessment of muscle glycogen utilization during exercise. Cell and Tissue Research 242, 229-232.

149

Gaitanos GC, Williams C, Boobis LH & Brooks S. (1993). Human muscle metabolism during intermittent maximal exercise. Journal of Applied Physiology 75, 712-719.

Garcia-Rocha M, Roca A, N. DLI, Baba O, Fernandez-Novell JM, Ferrer JC & Guinovart JJ. (2001). Intracellular distribution of glycogen synthase and glycogen in primary cultured rat hepatocytes. Biochemical Journal 357, 17-24.

Gasa R, Jensen PB, Berman HK, Brady MJ, DePaoli-Roach AA & Newgard CB. (2000). Distinctive Regulatory and Metabolic Properties of Glycogen-targeting Subunits of Protein Phosphatase-1 (PTG, GL, GM/RGl) Expressed in Hepatocytes. Journal of Biological Chemistry 275, 26396-26403.

Geddes R. (1986). Glycogen: a metabolic viewpoint. Bioscience Reports 6, 415-428.

Geddes R & Chow JC. (1994). Differing patterns of carbohydrate metabolism in liver and muscle. Carbohydrate Research 256, 139-147.

Geddes R, Harvey JD & Wills PR. (1977a). Hydrodynamic properties of 2-mercaptoethanol-modified glycogen. European Journal of Biochemistry 81, 465-472.

Geddes R, Harvey JD & Wills PR. (1977b). The molecular size and shape of liver glycogen. Biochemical Journal 163, 201-209.

Geddes R & Stratton GC. (1977). The influence of lysosomes on glycogen metabolism. Biochemical Journal 163, 193-200.

Gentry MS, Worby CA & Dixon JE. (2005). Insights into lafora disease: Malin is an E3 ubiquitin ligase that ubiquitinates and promotes the degradation of laforin. Proceedings of the National Academy of Sciences of the United States of America 102, 8501-8506.

Gibbons BJ, Roach PJ & Hurley TD. (2002). Crystal Structure of the Autocatalytic Initiator of Glycogen Biosynthesis, Glycogenin. Journal of Molecular Biology 319, 463-477.

Goldsmith E, Sprang S & Fletterick R. (1982). Structure of maltoheptaose by difference Fourier methods and a model for glycogen. Journal of Molecular Biology 156, 411-427.

Graham TE. (2009). Glycogen: an overview of possible regulatory roles of the proteins associated with the granule. Applied Physiology, Nutrition, and Metabolism 34, 488-492.

Graham TE, Adamo KB, Shearer J, Marchand I & Saltin B. (2001). Pro- and macroglycogenolysis: relationship with exercise intensity and duration. Journal of Applied Physiology 90, 873-879.

150

Graham TE, Yuan Z, Hill AK & Wilson RJ. (2010). The regulation of muscle glycogen: the granule and its proteins. Acta Physiologica 199, 489-498.

Gunja-Smith Z, Marshall J, Mercier C, Smith E & Whelan W. (1970). A revision of the Meyer-Bernfeld model of glycogen and amylopectin. FEBS Letters 12, 101-104.

Halsall TG, Hirst EL & Jones JK. (1947). Oxidation of carbohydrates by the periodate ion. Journal of the Chemical Society 172, 1427-1432.

Hansen BF, Derave W, Jensen P & Richter EA. (2000). No limiting role for glycogenin in determining maximal attainable glycogen levels in rat skeletal muscle. American Journal of Physiology: Endocrinology and Metabolism 278, E398-404.

Hargreaves M. (2004). Muscle glycogen and metabolic regulation. Proceedings of the Nutrition Society 63, 217-220.

Hassid WZ, Cori GT & McCready RM. (1943). Constitution of the polysaccharide synthesized by the action of crystalline muscle phosphorylase. Journal of Biological Chemistry 148, 89-96.

Hauk R, Illingworth B, Brown DH & Cori CF. (1959). Enzymes of glycogen synthesis in glycogen-deposition disease. Biochimica et Biophysica Acta 33, 554-556.

Haworth WN, Hirst EL & Isherwood FA. (1937). Polysaccharides. Part XXIII. Determination of the chain length of glycoge. Journal of the Chemical Society, 577-581.

Haworth WN, Hirst EL & Smith F. (1939). Polysaccharides. Part XXXVIII. The constitution of glycogen from fish liver and fish muscle. Journal of the Chemical Society, 1914-1922.

Haworth WN & Percival EGV. (1932). Polysaccharides. Part XI. Molecular structure of glycogen. Journal of the Chemical Society, 2277-2282.

Hennessey JV, Chromiak JA, Dellaventura S, Guertin J & Maclean DB. (1997). Increase in percutaneous muscle biopsy yield wih a suction-enhancement technique. Journal of Applied Physiology 82, 1739-1742.

Hers HG & Van Hoof F. (1973). Lysosomes and storage diseases. Academic Press, New York.

Hestrin S. (1949). Action pattern of crystalline muscle phosphorylase. Journal of Biological Chemistry 179, 943-955.

Huang M-T, Lee C-F, Lin R-F & Chen R-J. (1997). The exchange between proglycogen and macroglycogen and the metabolic role of the protein-rich glycogen in rat skeletal muscle. Journal of Clinical Investigation 99, 501-505.

151

Hubbard MJ, Dent P, Smythe C & Cohen P. (1990). Targetting of protein phosphatase 1 to the sarcoplasmic reticulum of rabbit skeletal muscle by a protein that is very similar or identical to the G subunit that directs the enzyme to glycogen. European Journal of Biochemistry 189, 243-249.

Hultman E. (1967). Muscle glycogen in man determined in needle biopsy specimens: method and normal values. Scandinavian Journal of Clinical Laboratory Investigation 19, 209-217.

James AP, Barnes PD, Palmer TN & Fournier PA. (2008). Proglycogen and macroglycogen: artifacts of glycogen extraction? Metabolism, Clinical and Experimental 57, 535-543.

Jansson E. (1981). Acid soluble and insoluble glycogen in human skeletal muscle. Acta Physiologica Scandinavica 113, 337-340.

Jenkins JA, Johnson LN, Stuart DI, Stura EA, Wilson KS, Zanotti G & Wilkie DR. (1981). Phosphorylase: Control and Activity [and Discussion]. Philosophical Transactions of the Royal Society of London Series B, Biological Sciences 293, 23-41.

Jensen J & Lai YC. (2009). Regulation of muscle glycogen synthase phosphorylation and kinetic properties by insulin, exercise, adrenaline and role in insulin resistance. Archives of Physiology and Biochemistry 115, 13-21.

Johnson LN. (1992). Glycogen phosphorylase: control by phosphorylation and allosteric effectors. FASEB Journal 6, 2274-2282.

Johnson LN. (2009). The regulation of protein phosphorylation. Biochemical Society Transactions 37, 627-641.

Jurczak MJ, Danos AM, Rehrmann VR & Brady MJ. (2008). The role of protein translocation in the regulation of glycogen metabolism. Journal of Cellular Biochemistry 104, 435-443.

Katz A. (2006). Glycogenin, proglycogen, and glycogen biogenesis: what's the story? American Journal of Physiology: Endocrinology and Metabolism 290, E757-758.

Kennedy LD, Kirkman BR, Lomako J, Rodriguez IR & Whelan WJ. (1985). Membranes and muscle, ed. Berman MC, Gevers W & Opie LH, pp. 65-84. ICSU Press and IRL Press, Oxford.

Kits van Heijningen AJM. (1957). The influence of insulin on free and fixed glycogen in the diaphragm of the rat. Biochemistry 65, 111-115.

Krisman CR. (1972). A possible intermediate in the initiation of glycogen biosynthesis. Biochemical and Biophysical Research Communications 46, 1206-1212.

Krisman CR. (1973). A possible intermediate in the initiation of glycogen biosynthesis. Annuals of the New York Academy of Sciences 210, 81-89.

152

Krisman CR & Barengo R. (1975). A precursor of glycogen biosynthesis: a-1,4-glucan-protein. European Journal of Biochemistry 52, 117-123.

Lacoste ER, Miozzo MC & Curtino JA. (1990). A glycogen precursor associated with membrane in retina. Biochemical Journal 167, 775-779.

Larner J. (1953). The action of branching enzymes on outer chains of glycogen. Journal of Biological Chemistry 202, 491-503.

Larner J, Illingworth B, Cori GT & Cori CF. (1952). Structure of glycogens and amylopectins. II. Analysis by stepwise enzymatic degradation. Journal of Biological Chemistry 199, 641-651.

Lazarow A. (1942). Particulate glycogen: A submicroscopic component of the guinea pig liver cell; its significance in glycogen storage and the regulation of blood sugar. Anatomical Record 84, 31-50.

Leloir LF & Cardini CE. (1957). Biosynthesis of glycogen from uridine diphosphate glucose. Journal of the American Chemical Society 79, 6340-6341.

Leloir LF & Goldemberg SH. (1960). Synthesis of glycogen from uridine diphosphate glucose in liver. Journal of Biological Chemistry 235, 919-923.

Lerin C, Montell E, Nolasco T, Clark C, Brady MJ, Newgard CB & Gomez-Foix AM. (2003). Regulation and function of the muscle glycogen-targeting subunit of protein phosphatase 1 (GM) in human muscle cells depends on the COOH-terminal region and glycogen content. Diabetes 52, 2221-2226.

Liu W, Madsen NB, Fan B, Zucker KA, Glew RH & Fry DE. (1995). Effects of oligosaccharide binding on glycogen debranching enzyme activity and contormation. Biochemistry 34, 7056-7061.

Lomako J, Lomako WM, Kirkman BR & Whelan WJ. (1994). The role of phosphate in muscle glycogen. BioFactors 4, 167-171.

Lomako J, Lomako WM & Whelan WJ. (1991a). Proglycogen: A low-molecular-weight form of muscle glycogen. FEBS Letters 279, 223-228.

Lomako J, Lomako WM & Whelan WJ. (1995). Glycogen metabolism in quail embryo muscle: The role of the glycogenin primer and the intermediate proglycogen. European Journal of Biochemistry 234, 343-349.

Lomako J, Lomako WM, Whelan WJ, Dombro RA, Neary JJ & Norenberg MD. (1991b). Glycogen biogenesis in the rat-brain astrocyte. FASEB Journal 5, A1014.

Lomako J, Lomako WM, Whelan WJ, Dombro RS, Neary JT & Norenberg MD. (1993a). Glycogen synthesis in the astrocyte: From glycogenin to proglycogen to glycogen. FASEB Journal 7, 1386-1393.

153

Lomako J, Lomako WM, Whelan WJ & Marchase RB. (1993b). Glycogen contains phosphodiester groups that can be introduced by UDPglucose: glycogen glucose 1-phosphotransferase. FEBS Letters 329, 263-267.

Luck DJL. (1961). Glycogen synthesis from uridine diphosphate glucose: The distribution of the enzyme in liver cell fractions. Journal of Biophysical and Biochemical Cytology 10, 195-209.

Manners DJ. (1957). The molecular structure of glycogens. Advances in Carbohydrate Chemistry 12, 261-298.

Manners DJ. (1991). Recent developments in our understanding of glycogen structure. Carbohydrate Polymers 16, 37-82.

Marchand I, Chorneyko K, Tarnopolsky M, Hamilton S, Shearer J, Potvin J & Graham TE. (2002). Quantification of subcellular glycogen in resting human muscle: granule size, number, and location. Journal of Applied Physiology 93, 1598-1607.

Marchand I, Tarnopolsky M, Adamo KB, Bourgeois JM, Chorneyko K & Graham TE. (2007). Quantitative assessment of human muscle glycogen granules size and number in subcellular locations during recovery from prolonged exercise. Journal of Physiology 580, 617-628.

McArdle WD & Montoye HJ. (1966). Reliability of exhaustive swimming in the laboratory rat. Journal of Applied Physiology 21, 1431-1434.

McBride A, Ghilagaber S, Nikolaev A & Hardie DG. (2009). The Glycogen-Binding Domain on the AMPK ² Subunit Allows the Kinase to Act as a Glycogen Sensor. Cell Metabolism 9, 23-34.

McLay RT, Thomson CD, Williams SM & Rehrer NJ. (2007). Carbohydrate loading and female endurance athletes: effect of menstrual-cycle phase. International Journal of Sport Nutrition and Exercise Metabolism 17, 189-205.

Melendez R, Melendez-Hevia E & Cascante M. (1997). How did glycogen structure evolve to satisfy the requirement for rapid mobilization of glucose? A problem of physical constraints in structure building. Journal of Molecular Evolution 45, 446-455.

Melendez R, Meléndez-Hevia E, Mas F, Mach J & Cascante M. (1998). Physical Constraints in the Synthesis of Glycogen That Influence Its Structural Homogeneity: A Two-Dimensional Approach. Biophysical Journal 75, 106-114.

Meyer F, Heilmeyer Jr LMG, Haschke RH & Fischer EH. (1970). Control of Phosphorylase Activity in a Muscle Glycogen Particle. I. Isolation and charactorization of the protien-glycogen complex. Journal of Biological Chemistry 245, 6642-6648.

Meyer F & Lourau M. (1956). Incorporation of radiocarbon furnished by glucose into the lyoglycogen and desmoglycogen of the rat liver. Comptes rendus

154

hebdomadaires des séances de l'Académie des Sciences 243, 1688-1690.

Meyer KH & Bernfeld P. (1940). Reeherehes sur l’amidon V. L’arnylopeetine. Helvetica Chimica Acta 23, 875-885.

Meyer KH & Fuld M. (1941). Recherches sur l'amidon XII. L'arrangement des restes de glucose dans le glycogène. Helvetica Chimica Acta 24, 375-378.

Miozzo MC, Lacoste ER & Curtino JA. (1989). Characterization of the proteoglycogen fraction non-extractable from retina by trichloroacetic acid. Biochemical Journal 260, 287-289.

Mu J & Roach PJ. (1998). Characterization of human glycogenin-2, a self-glucosylating initiator of liver glycogen metabolism. Journal of Biological Chemistry 273, 34850-34856.

Mu J, Skurat AV & Roach PJ. (1997). Glycogenin-2, a novel self-glucosylating protein involved in liver glycogen biosynthesis. Journal of Biological Chemistry 272, 27589-27597.

Nelson TE, White RC & Watts TE. (1972). The action of the glycogen debranching enzyme system in a muscle protein particle. Biochemical and Biophysical Research Communications 47, 254-259.

Nevill ME, Boobis LH, Brooks S & Williams C. (1989). Effect of training on muscle metabolism during treadmill sprinting. Journal of Applied Physiology 67, 2376-2382.

Newgard CB, Brady MJ, O'Doherty RM & Saltiel AR. (2000). Organizing glucose disposal: Emerging roles of the glycogen targeting subunits of protein phosphotase-1. Diabetes 49, 1967-1977.

Nielsen J, Schrøder HD, Rix CG & Ortenblad N. (2009). Distinct effects of subcellular glycogen localization on tetanic relaxation time and endurance in mechanically skinned rat skeletal muscle fibres. Journal of Physiology 587, 3679-3690.

Nielsen JN, Derave W, Kristiansen S, Ralston E, Ploug T & Richter EA. (2001). Glycogen synthase localisation and activity in rat skeletal muscle is strongly dependent on glycogen content. Journal of Physiology 531, 757-769.

Nielsen JN & Richter EA. (2003). Regulation of glycogen synthase in skeletal muscle during exercise. Acta Physiologica Scandinavica 178, 309-319.

Nielsen JN & Wojtaszewski JFP. (2004). Regulation of glycogen synthase activity and phosphorylation by exercise. Proceedings of the Nutrition Society 63, 233-237.

Oakley HB & Young FG. (1936). The osmotic pressure of glycogen solutions. Biochemistry 30, 868-876.

155

Orrell SA & Bueding E. (1964). A comparison of products obtained by various procedures used for the extraction of glycogen. Journal of Biological Chemistry 239, 4021-4026.

Parker GJ, Koay A, Gilbert-Wilson R, Waddington LJ & Stapleton D. (2007). AMP-activated protein kinase does not associate with glycogen α-particles from rat liver. Biochemical and Biophysical Research Communications 362, 811-815.

Parnas JK. (1937). Ergebnisse der Enzymforschung 6, 57-110.

Parodi AJ, Mordoh J, Krisman CR & Leloir LF. (1969). In vitro synthesis of particulate glycogen from uridine diphosphate glucose. Archives of Biochemistry and Biophysics 132, 111-117.

Peat S, Whelan WJ & Thomas GJ. (1952). Evidence of multiple branching in waxy maize starch. Journal of the Chemical Society, 4546-4547.

Peat S, Whelan WJ & Thomas GJ. (1956). The enzymatic synthesis and degradation of starch. Part XXII. Evidence of multiple branching in waxy-maize starch. A correction. Journal of the Chemical Society, 3025-3030.

Pitcher J, Smythe C, Campbell DG & Cohen P. (1987). Identification of the 38-kDa subunit of rabbit skeletal muscle glycogen as glycogenin. European Journal of Biochemistry 169, 497-502.

Pitcher J, Smythe C & Cohen P. (1988). Glycogenin is the priming glucosyltransferase required for the initiation of glycogen biogenesis in rabbit skeletal muscle. European Journal of Biochemistry 176, 391-395.

Polekhina G, Feil SC, Gupta A, O'Donnell P, Stapleton D & Parker MW. (2005). Crystallization of the glycogen-binding domain of the AMP-activated protein kinase β subunit and preliminary X-ray analysis. Acta Crystallographica Section F 61, 39-42.

Polekhina G, Gupta A, Michell BJ, van Denderen B, Murthy S, Feil SC, Jennings IG, Campbell DJ, Witters LA, Parker MW, Kemp BE & Stapleton D. (2003). AMPK β subunit targets metabolic stress sensing to glycogen. Current Biology 13, 867-871.

Prats C, Bernal C, Cadefau JA, Frias JA, Tibolla M & Cusso R. (2002). Glycogen depletion and resynthesis during 14 days of chronic low-frequency stimulation of rabbit muscle. Biochimica et Biophysica Acta 1573, 68-74.

Prats C, Cadefau JA, Cussó R, Qvortrup K, Nielsen JN, Wojtaszewski JF, Hardie DG, Stewart G, Hansen BF & Ploug T. (2005). Phosphorylation-dependent translocation of glycogen synthase to a novel structure during glycogen resynthesis. Journal of Biological Chemistry 280, 23165-23172.

Prats C, Helge JW, Nordby P, Qvortrup K, Ploug T, Dela F & Wojtaszewski JF. (2009). Dual regulation of muscle glycogen synthase during exercise by

156

activation and compartmentalization. Journal of Biological Chemistry 284, 15692-15700.

Price TB, Laurent D, Petersen KF, Rothman DL & Shulman GI. (2000). Glycogen loading alters muscle glycogen resynthesis after exercise. Journal of Applied Physiology 88, 698-704.

Printen JA, Brady MJ & Saltiel AR. (1997). PTG, a protein phosphatase 1-binding protein with a role in glycogen metabolism. Science 275, 1475-1478.

Rico-Sanz J, Zehnder M, Buchli R, Kühne G & Boutellier U. (2008). Creatine feeding does not enhance intramyocellular glycogen concentration during carbohydrate loading: an in vivo study by 31P- and 13C-MRS. Journal of Physiology and Biochemistry 64, 189-196.

Roach PJ. (2002). Glycogen and its metabolism. Current Molecular Medicine 2, 101-120.

Roach PJ, Cheng C, Huang D, Lin A, Mu J, Skurat AV, Wilson W & Zhai L. (1998). Novel aspects of the regulation of glycogen storage. Journal of Basic & Clinical Physiology & Pharmacology 9, 139-151.

Roach PJ & Skurat AV. (1997). Self-glucosylating initiator proteins and their role in glycogen biosynthesis. Progress in Nucleic Acid Research and Molecular Biology 57, 289-316.

Rodriguez IR & Whelan WJ. (1985). A novel glycosyl-amino acid linkage: Rabbit-muscle glycogen is covalently linked to a protein via tyrosine. Biochemical and Biophysical Research Communications 132, 829-836.

Roe JH, Bailey JM, Gray RR & Robinson JN. (1961). Complete removal of glycogen from tissues by extraction with cold trichloroacetic acid solution. Journal of Biological Chemistry 236, 1244-1246.

Ros S, García-Rocha M, Domínguez J, Ferrer JC & Guinovart JJ. (2009). Control of liver glycogen synthase activity and intracellular distribution by phosphorylation. Journal of Biological Chemistry 284, 6370-6378.

Rosenvold K, Essen-Gustavsson B & Andersen HJ. (2003). Dietary manipulation of pro- and macroglycogen in porcine skeletal muscle. Journal of Animal Science 81, 130-134.

Russell JA & Bloom WL. (1955). Extractable and residual glycogen in tissues of the rat. American Journal of Physiology 183, 345-355.

Russell JA & Bloom WL. (1956). Hormonal control of glycogen in the heart and other tissues in rats. Endocrinology 58, 83-94.

Rybicka K. (1979). Glycosomes (protein-glycogen complex) in canine heart. Virchows Archice B Cell Pathology 30, 335-347.

157

Rybicka K. (1981a). Binding of glycosomes to endoplasmic reticulum and to intermediate filaments in cardiac conduction fibers. Journal of Histochemistry and Cytochemistry 29, 553-560.

Rybicka K. (1981b). Simultaneous demonstration of glycogen and protein in glycosomes of cardiac tissue. Journal of Histochemistry and Cytochemistry 29, 4-8.

Rybicka KK. (1996). Glycosomes - the organelles of glycogen metabolism. Tissue & Cell 28, 253-265.

Ryu J-H, Drain J, Kim JH, McGee S, Gray-Weale A, Waddington L, Parker GJ, Hargreaves M, Yoo S-H & Stapleton D. (2009). Comparative structural analyses of purified glycogen particles from rat liver, human skeletal muscle and commercial preparations. International Journal of Biological Macromolecules 45, 478-482.

Sanson MA. (1857). Note sur la formation physiologique du sucre dans l'economie animale. Comptes rendus des seances de l'Academie des Sciences 44, 1323-1325.

Schmalbruch H & Kamieniecka Z. (1974). Fiber types in the human brachial biceps muscle. Experimental Neurology 44, 323-328.

Scott RB & Still WJ. (1968). Glycogen in human peripheral blood leukocytes. II. The macromolecular state of leukocyte glycogen. Journal of Clinical Investigation 47, 353-359.

Shearer J & Graham TE. (2002). New perspectives on the storage and organization of muscle glycogen. Revue Canadienne de Physiologie Appliquée 27, 179-203.

Shearer J & Graham TE. (2004). Novel aspects of skeletal muscle glycogen and its regulation during rest and exercise. Exercise and Sport Sciences Reviews 32, 120-126.

Shearer J, Graham TE, Battram DS, Robinson DL, Richter EA, Wilson RJ & Bakovic M. (2005a). Glycogenin activity and mRNA expression in response to volitional exhaustion in human skeletal muscle. Journal of Applied Physiology 99, 957-962.

Shearer J, Marchand I, Sathasivam P, Tarnopolsky MA & Graham TE. (2000). Glycogenin activity in human skeletal muscle is proportional to muscle glycogen concentration. American Journal of Physiology 41, E177-180.

Shearer J, Marchand I, Tarnopolsky MA, Dyck DJ & Graham TE. (2001). Pro- and macroglycogenolysis during repeated exercise: Roles of glycogen content and phosphorylase activation. Journal of Applied Physiology 90, 880-888.

Shearer J, Marchand I, Tarnopolsky MA & Graham TE. (1999). The influence of muscle glycogen content on pro- and macroglycogen utilization during intense exercise. FASEB Journal 13, A78.

158

Shearer J, Wilson RJ, Battram DS, Richter EA, Robinson DL, Bakovic M & Graham TE. (2005b). Increases in glycogenin and glycogenin mRNA accompany glycogen resynthesis in human skeletal muscle. American Journal of Physiology 289, E508-514.

Skurat AV, Dietrich AD & Roach PJ. (2006). Interaction between glycogenin and glycogen synthase. Archives of Biochemistry and Biophysics 456, 93-97.

Skurat AV, Dietrich AD, Zhai L & Roach PJ. (2002). GNIP, a Novel Protein That Binds and Activates Glycogenin, the Self-glucosylating Initiator of Glycogen Biosynthesis. Journal of Biological Chemistry 277, 19331-19338.

Skurat AV, Lim S-S & Roach PJ. (1997). Glycogen biogenesis in rat 1 fibroblasts expressing rabbit muscle glycogenin. European Journal of Biochemistry 245, 147-155.

Skurat AV & Roach PJ. (2004). Regulation of glycogen synthesis. In Diabetes mellitus: A fundamental and clinical text, ed. LeRoith D, Taylor SI & Olefsky JM, pp. 317-333. Lippincott Williams & Wilkins, Philadelphia.

Smythe C, Caudwell FB, Ferguson M & Cohen P. (1988). Isolation and structural analysis of a peptide containing the novel tyrosyl-glucose linkage in glycogenin. EMBO Journal 7, 2681-2686.

Smythe C & Cohen P. (1991). The discovery of glycogenin and the priming mechanism for biogenesis. European Journal of Biochemistry 200, 625-631.

Smythe C, Villar-Palasi C & Cohen P. (1989). Structural and functional studies on rabbit liver glycogenin. European Journal of Biochemistry 183, 205-209.

Smythe C, Watt P & Cohen P. (1990). Further studies on the role of glycogenin in glycogen biosynthesis. European Journal of Biochemistry 189, 199-204.

Solaz-Fuster MC, Gimeno-Alcaniz JV, Ros S, Fernandez-Sanchez ME, Garcia-Fojeda B, Garcia OC, Vilchez D, Dominguez J, Garcia-Rocha M, Sanchez-Piris M, Aguado C, Knecht E, Serratosa J, Guinovart JJ, Sanz P & de Cordoba SR. (2008). Regulation of glycogen synthesis by the laforin–malin complex is modulated by the AMP-activated protein kinase pathway. Human Molecular Genetics 17, 667-678.

Spiro RG. (1966). Methods in Enzymology, vol. 8. Academic Press, New York.

Staudinger H & Husemann E. (1937). Über hochpolymere Verbindungen. Über die Konstitution des Glycogens. Justus Liebigs Annalen der Chemie 530, 1-20.

Stetten D & Stetten MR. (1960). Glycogen metabolism. Physiological Reviews 40, 505-537.

159

Stetten MR, Katzen HM & Stetten Jr D. (1958). A comparison of glycogens isolated by acid and alkaline procedures. Journal of Biological Chemistry 232, 475-488.

Stetten MR & Stetten Jr D. (1958). Influence of molecular size of glycogen on the phosphorylase reaction. Journal of Biological Chemistry 232, 489-504.

Sullivan MA, Vilaplana F, Cave RA, Stapleton D, Gray-Weale AA & Gilbert RG. (2010). Nature of α and β particles in glycogen using molecular size distributions. Biomacromolecules 11, 1094-1100.

Tagliabracci VS, Girard JM, Segvich D, Meyer C, Turnbull J, Zhao X, Minassian BA, Depaoli-Roach AA & Roach PJ. (2008). Abnormal metabolism of glycogen phosphate as a cause for Lafora disease. Journal of Biological Chemistry 283, 33816-33825.

Tagliabracci VS, Turnbull J, Wang W, Girard JM, Zhao X, Skurat AV, Delgado-Escueta AV, Minassian BA, Depaoli-Roach AA & Roach PJ. (2007). Laforin is a glycogen phosphatase, deficiency of which leads to elevated phosphorylation of glycogen in vivo. Proceedings of the National Academy of Sciences of the United States of America 104, 19262-19266.

Takeuchi T, Iwamasa T & Miyayama H. (1978). Ultrafine structure of glycogen macromolecules in mammalian tissues. Journal of Electron Microscopy 27, 31-38.

Tang PM, Bondor JA, Swiderek KM & DePaoli-Roach AA. (1991). Molecular cloning and expression of the regulatory (RG1) subunit of the glycogen-associated protein phosphatase. Journal of Biological Chemistry 266, 15782-15789.

Tata JR. (1964). Subcellular redistribution of liver α-glucan phosphorylase during alterations in glycogen content. Biochemistry 90, 284-292.

Taylor C, Cox AJ, Kernohan JC & Cohen P. (1975). Debranching enzyme from rabbit skeletal muscle. European Journal of Biochemistry 51, 105-115.

Thomas S, Reading J & Shephard R. (1992). Revision of the physical activity readiness questionnaire (PAR-Q). Canadian Journal of Sport Sciences 17, 338-345.

Toole BJ & Cohen PTW. (2007). The skeletal muscle-specific glycogen-targeted protein phosphatase 1 plays a major role in the regulation of glycogen metabolism by adrenaline in vivo. Cellular Signalling 19, 1044-1055.

van Loon LJ, Saris WH, Kruijshoop M & Wagenmakers AJ. (2000). Maximizing postexercise muscle glycogen synthesis: carbohydrate supplementation and the application of amino acid or protein hydrolysate mixtures. American Journal of Clinical Nutrition 72, 106-111.

160

Van Maanen MH, Fournier PA, Palmer TN & Abraham LJ. (1999a). Characterization of mouse glycogenin-1 cDNA and promoter region. Biochimica Biophysica Acta 1447, 284-290.

Van Maanen MH, Fournier PA, Palmer TN & Abraham LJ. (1999b). Characterization of the human glycogenin-1 gene: identification of a muscle-specific regulatory domain. Gene 234, 217-226.

Villar-Palasi C & Larner J. (1960). Levels of activity of the enzymes of the glycogen cycle in rat tissue. Archives of Biochemistry and Biophysics 86, 270-273.

Wang J, Stuckey JA, Wishart MJ & Dixon JE. (2002). A unique carbohydrate binding domain targets the lafora disease phosphate to glycogen. Journal of Biological Chemistry 4, 2377-2380.

Wanson JC & Drochmans P. (1968). Rabbit skeletal muscle glycogen: a morphological and biochemical study of glycogen β-particles isolated by the precipitation-centrifugation method. Journal of Cell Biology 38, 130-150.

Wee SL, Williams C, Tsintzas K & Boobis L. (2005). Ingestion of a high-glycemic index meal increases muscle glycogen storage at rest but augments its utilization during subsequent exercise. Journal of Applied Physiology 99, 707-714.

Willstatter R & Rohdewald M. (1934). Uder den zustand des glytrogens in der leber, im muskel und in leukocyten. Hoppe-Seyler's Zeitschrift für Physiologische Chemie 225, 103-124.

Wilson RJ, Gusba JE, Robinson DL & Graham TE. (2007). Glycogenin protein and mRNA expression in response to changing glycogen concentration in exercise and recovery. American Journal of Physiology 292, E1815-1822.

Wolfrom ML & O'Neill AN. (1949). Degradation of glycogen to isomaltose. Journal of the American Chemical Society 71, 3857-3858.

Worby CA, Gentry MS & Dixon JE. (2006). Laforin, a dual specificity phosphatase that dephosphorylates complex carbohydrates. Journal of Biological Chemistry 281, 30412-30418.

Worby CA, Gentry MS & Dixon JE. (2008). Malin decreases glycogen accumulation by promoting the degradation of protein targeting to glycogen (PTG). Journal of Biological Chemistry 283, 4069-4076.

Yang R, Cao L, Gasa R, Brady MJ, Sherry AD & Newgard CB. (2002). Glycogen-targeting Subunits and Glucokinase Differentially Affect Pathways of Glycogen Metabolism and Their Regulation in Hepatocytes. Journal of Biological Chemistry 277, 1514-1523.

Yang R & Newgard CB. (2003). Hepatic Expression of a Targeting Subunit of Protein Phosphatase-1 in Streptozotocin-diabetic Rats Reverses

161

Hyperglycemia and Hyperphagia Despite Depressed Glucokinase Expression. Journal of Biological Chemistry 278, 23418-23425.

Young FG. (1957). Claude Bernard and the discovery of glycogen; a century of retrospect. British Medical Journal 1, 1431-1437.

Zhai L, Dietrich A, Skurat AV & Roach PJ. (2004). Structure-function analysis of GNIP, the glycogenin-interacting protein. Archives of Biochemistry and Biophysics 421, 236-242.