going deep into the acute phase protein response
TRANSCRIPT
General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
You may not further distribute the material or use it for any profit-making activity or commercial gain
You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from orbit.dtu.dk on: Oct 17, 2021
Going deep into the acute phase protein response
Leuchsenring, Anna Barslund
Publication date:2020
Document VersionPublisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA):Leuchsenring, A. B. (2020). Going deep into the acute phase protein response. DTU Bioengineering.
1
Division for Immunology and Vaccinology, DTU Vet
Department of Biotechnology and Biomedicine, DTU Bioengineering
Technical University of Denmark
Going deep into the acute phase protein response:
Use of targeted mass spectrometry for antibody independent serum amyloid A isoform profiling Ph.D. thesis Anna Barslund Leuchsenring April 2020
2
Supervisor Professor Peter M. H. Heegaard Innate Immunology Group, Veterinary Institute National, DTU Vet Department of Biotechnology and Biomedicine, DTU Bioengineering Currently: Center for Diagnostics, DTU
Assesment committee Professor MSO Maher A. Hachem (chairperson) Institut for Bioteknologi og Biomedicin, DTU Bioengineering Technical University of Denmark
Professor Theo Niewold Senior Consultant
Associate professor Per M. Hägglund Department of Biomedical Sciences University of Copenhagen
Going deep into the acute phase protein response: Use of targeted mass spectrometry for antibody independent serum amyloid A isoform profiling Ph.D. thesis 2020 © Anna Barslund Leuchsenring. Cover: SRM data produced in this Ph.D. project Pig stock photo, obtained from colourbox.com under DTU license
3
Contents
Contents ........................................................................................................................... 3
Preface .............................................................................................................................. 6
Acknowledgments ............................................................................................................. 7
Abbreviations .................................................................................................................... 9
Summary ........................................................................................................................ 11
Sammendrag (Danish summary) ..................................................................................... 13
Chapter 1 ........................................................................................................................ 15
Introduction .................................................................................................................... 15
1.1 Objectives .............................................................................................................. 16
1.2 Outline of Ph.D. thesis .......................................................................................... 17
Chapter 2 ........................................................................................................................ 18
Background ..................................................................................................................... 18
2.1 The acute phase response ...................................................................................... 18
2.2 Acute phase proteins ............................................................................................. 19
2.2.1 Serum Amyloid A............................................................................................ 23
Chapter 3 ........................................................................................................................ 33
Technical aspects ............................................................................................................ 33
3.1 Mass Spectrometry ................................................................................................ 34
3.2 Liquid Chromatography-Mass Spectrometry (LC-MS) .......................................... 36
3.3 Matrix-assisted laser desorption ionization (MALDI) ............................................ 37
3.4 Electrospray ionization (ESI) ................................................................................ 39
3.5 Mass analyzers ....................................................................................................... 40
4
3.6 Quadrupole Mass Analyzers .................................................................................. 41
3.7 Time of Flight (TOF) mass analyzers ................................................................... 42
3.8 Tandem MS (MS/MS)........................................................................................... 44
Chapter 4 ........................................................................................................................ 46
Discovery and Targeted proteomics ................................................................................ 46
4.1 Parallel Reaction Monitoring (PRM) .................................................................... 47
4.2 Selected Reaction Monitoring (SRM) .................................................................... 48
4.3 Selection of target peptides .................................................................................... 51
4.4 SRM transitions..................................................................................................... 53
4.5 Quantification by SRM ......................................................................................... 58
4.6 Tandem Mass Tag (TMT) Mass Spectrometry ..................................................... 59
4.7 Data-independent acquisition and SWATH-MS .................................................... 61
Chapter 5 ........................................................................................................................ 64
Materials and Methods ................................................................................................... 64
5.1 Biological samples .................................................................................................. 64
5.1.1 Staphylococcus aureus (Sa) pig infection models ............................................. 65
5.1.1.1 Sa intravenous infection model ................................................................. 65
5.1.1.2 Implant-Associated Osteomyelitis (IAO) Sa infection model .................... 66
5.1.2 Actinobacillus pleuropneumoniae (Ap) pig infection models ........................... 66
5.2 SAA ELISA ........................................................................................................... 67
5.3 Triton X-114 cloud point extraction (CPE) .......................................................... 67
5.4 TCA Precipitation ................................................................................................. 69
5.5 2D gel electrophoresis ............................................................................................ 69
5.6 MALDI-TOF ......................................................................................................... 70
5.7 SRM method development .................................................................................... 70
Chapter 6 ........................................................................................................................ 71
Paper I ............................................................................................................................ 71
6.1 Summary of Paper I .............................................................................................. 72
5
6.2 Paper 1 as submitted to Journal of Proteomics ..................................................... 73
Chapter 7 ...................................................................................................................... 103
Results and discussion .................................................................................................. 103
7.1 Two-dimensional gel electrophoresis and MALDI-TOF ....................................... 104
7.2 Relative SAA concentrations in porcine serum samples ...................................... 108
7.3 Absolute concentrations of SAA isoforms ............................................................ 111
7.4 Conclusions .......................................................................................................... 117
7.5 Perspectives ......................................................................................................... 117
References ..................................................................................................................... 119
Supplementary .............................................................................................................. 138
6
Preface
The work presented in this thesis was carried out from November 2014 to April 2020
(interspersed with two separate maternity leaves of around 12 months each). From
November 2014 until April 2019 the Ph.D. project was performed at the National
Veterinary Institute of DTU (DTU Vet) and from May 2019 until April 2020 the Ph.D.
project was performed at the Department of Biotechnology and Biomedicine, DTU
Bioengineering. The project was financed with internal funding and supervised by
Professor Peter M. H. Heegaard. This Ph.D. has been done in collaboration with Johan
Malmström, Department of Infection Medicine, Lund University, where all SRM
experiments took place.
7
Acknowledgments
Through all the years working on this project, I have had the pleasure to work with many
wonderful people; all of whom have helped, supported, and guided me. The first and
biggest acknowledgment goes to my supervisor Peter Heegaard who gave me the
opportunity to work with this exciting project. We’ve had many obstacles thrown at us,
but Peter has always been supportive and found a way through. I am very grateful for
your patience, your guidance, and your great knowledge – I hope at least some of it has
rubbed off on me.
I also want to thank Christofer Karlsson who has taught me the practical proteomics skills
I needed and always answered every question with detailed and thoughtful answers no
matter how small or big. His calm presence and patience guidance helped me through
many a stressful situation and convinced me to keep going, even when it got tough. I am
also very grateful to Johan Malmström for the cooperation and for the opportunity to
work in his lab where I’ve learned so much and felt like part of the group from my very
first day. I want to thank Anahita Bakochi for being a wonderful friend and colleague and
for welcoming me so warmly into the Infection Medicine group along with Emma Åhrman
and Lotta Happonen.
Thanks should also be given to Flemming Jessen and Hanne Lilian Stampe-Villadsen at
DTU Food for helping me learn the many steps of 2D gel electrophoresis and more.
8
Louise Brogaard and I have shared an office for most of my time at DTU, and I want to
thank her for making it joyous to come to work and for being like a wiser older Ph.D.
sister whom I could look up to, ask all the stupid (or clever) questions and tag along to
see how to get things done the right way – and for accepting my crazy Christmas
decorations. I am going to miss her very much.
Last, but definitely not least, I want to thank my entire family for all of their support
throughout the years. Especially my husband Mikkel has patiently listened to my joys and
sorrows and has been my rock when everything seemed impossible. He has taken care of
me and our kids while I have been working like crazy to finish this Ph.D. project. I love
you and I could not have done this without you.
A very special thanks goes to my mother, who has supported my dreams since I could first
express them, and who have always done her absolute best to help me realize them. While
writing this Ph.D. thesis she has helped taking care of my kids, and I can surely say that
it would not have been finished if it hadn’t been for her. Thank you from the bottom of
my heart.
Anna Barslund Leuchsenring, April 2020
9
Abbreviations
2DE Two-dimensional gel electrophoresis
Ap Actinobacillus pleuropneumoniae
APPs Acute phase proteins
APR Acute phase response
BW Bodyweight
CFU Colony-forming unit
CID Collision-induced dissociation
CPE Cloud point extracted
CRP C-reactive protein
DDA Data-dependent acquisition
DIA Data-independent acquisition
DP Declustering potential
ELISA Enzyme-Linked Immunosorbent Assay
ESI Electrospray ionization
FPR Formyl peptide receptor
HDL High-density lipoprotein
HPLC High-Pressure Liquid Chromatography
IEF Isoelectric focusing
IFN Interferon
IL Interleukin
10
IAO Implant-Associated Osteomyelitis
LC Liquid chromatography
LC-MS Liquid Chromatography MS
M/z Mass-to-charge ratio
MALDI Matrix-assisted laser desorption/ionization
MS/MS Tandem mass spectrometry
MS Mass spectrometry
pI Isoelectric point
PI Post infection
PRM Parallel reaction monitoring
RAGE Receptor for advanced glycation end-products
rt Retention time
RT Room temperature
Sa Staphylococcus aureus
SAA Serum amyloid A
SR-BI Scavenger receptor class B type I
SRM Selected reaction monitoring
SWATH-MS Sequential Windowed Acquisition of All Theoretical Fragment Ion
MS
TCS Trichloroacetic Acid
TGF Transforming growth factor
TLR Toll-like receptor
TMT Tandem mass tags
TNF Tumor necrosis factor
TOF Time-of-flight
11
Summary
Tissue perturbation and infection are accompanied by a generalized set of immediate host
reactions at the site of disturbance as well as an array of systemic reactions known as the
acute phase response (APR). The APR leads to major changes in circulating
concentrations of a subset of blood proteins called acute phase proteins. Serum amyloid A
(SAA) is a well-described acute phase protein in vertebrates synthesized in the liver and
non-hepatic tissue. Most species have four genes each encoding a specific isoform of SAA,
which have found to be differentially regulated depending on infection and tissue. This is
a unique feature of SAA not found in other acute phase proteins and makes circulating
SAA and its isoforms very interesting biomarker candidates. However, SAA isoform
analysis is cumbersome, non-quantitative and no antibodies specific for SAA isoforms have
been developed, making SAA isoform distribution at the protein level less characterized.
Using mass spectrometry (MS)-based methods may overcome these problems as MS is
more specific and adequately sensitive. In this Ph.D. project, the native composition and
presence of SAA and its isoforms in pig serum was first investigated through 2D gel
electrophoresis (2DE) and the MS-based method Matrix-assisted laser
desorption/ionization (MALDI) with a time-of-flight (TOF) mass spectrometry.
Subsequently a new targeted quantitative MS method was developed based on selected
reaction monitoring (SRM) mass spectrometry. 2DE and MALDI-TOF confirmed the
presence of an APR in the investigated serum samples, but only SAA2 and SAA4 could
be identified. As this experimental set-up is both time demanding and not the most
suitable for biomarker development, the SRM method developed was used to discriminate
12
and quantify the porcine SAA isoforms with high specificity and sensitivity in porcine
serum samples from two separate and previously performed infection studies. For improved
quantitative performance, isotopically labeled SAA peptides were used to calibrate the
method. In Paper I, the SRM method was applied to sequential serum samples from a
previous experiment in which pigs were infected with Staphylococcus aureus (Sa) where
all circulating porcine SAA isoforms were detected. The SAA isoform profiles correlated
with the total SAA response previously established by ELISA, and a close correlation
between the relative hepatic SAA isoform mRNA abundance and SAA isoform profiles in
the circulation was also found. This proved that antibody independent MS-based
quantification is possible at high specificity and sensitivity allowing discrimination between
a set of very similar circulating SAA variants directly in serum samples. The results
unequivocally establish SAA2 as the main circulating porcine SAA isoform. Porcine serum
samples from pigs infected with Actinobacillus pleuropneumoniae (Ap) in another previous
experiment were also analyzed, and while all SAA isoforms could be detected, only relative
SAA concentrations could be measured as the selected proteotypic peptides did not
perform well enough to provide absolute quantification. An additional set of porcine serum
samples from a previous experiment on Sa osteomyelitis in pigs were analyzed in which all
SAA isoforms were detected and quantified. Both relative and absolute SAA
concentrations were measured and correlated well with each other in this sample set.
Relative and absolute SAA isoform concentrations were compared across the three sample
sets, and each set showed a different SAA isoform composition supporting the concept
that different types of infections - for example having different tropism as with Ap and Sa
- will result in infection-specific SAA isoform profiles.
The prospect of using SAA isoform profiling to achieve increased diagnostic power from
acute phase protein biomarkers, and to discriminate between different infections and/or
infections with different tissue tropism, is of great importance. It would be very interesting
to investigate further the potential of this approach for a detailed characterization of mass
variants of circulating proteins and the possible links to their tissue origin.
13
Sammendrag (Danish summary)
Kroppens forsvar mod infektion, vævsskade og patogener varetages af en lang række
immunologiske mekanismer. Den første reaktion på en infektion el.lign. er det innate
immunforsvar, som igangsætter en kaskade af systemiske reaktioner og opregulering af
blandt andet proteiner kendt som akut fasereaktanter. Serum amyloid A protein A (SAA)
er en vigtigt og velbeskrevet akut fasereaktant som produceres både i leveren og i ikke-
hepatisk væv, hos hvirveldyr. Hos de fleste undersøgte arter findes der fire gener, der hver
koder for en bestemt isoform af SAA, som har vist sig at være afhængig af væv og
infektionstype. Dette er et unikt træk ved SAA, der ikke findes i andre akutte fase-
proteiner og gør cirkulerende SAA og dets isoformer meget interessante
biomarkørkandidater. Den gængse metode til at analysere SAA-isoform ekspression er
tidskrævende, ikke kvantitative og de antistoffer, der er tilgængelige for SAA, kan ikke
skelne mellem isoformerne, hvilket betyder at SAA isoform-fordeling på proteinniveau har
været svær at analysere og er derfor ikke afklaret. Massespektrometri (MS) -baserede
metoder kan overvinde disse problemer, da massepektrometri er meget mere specifik og
lige så sensitiv som de traditionelle metoder.
I dette Ph.D. projekt blev kompositionen af nativ SAA først undersøgt vha. 2D
gelelektroforese (2DE) og den MS-baserede metode ”Matrix-assisted laser
desorption/ionization” (MALDI) med et ”time-of-flight” (TOF) massespektrometer, og
derefter ved hjælp af en ”targeted” kvantitativ MS-metode, der er udviklet i dette ph.d.-
projekt, kaldet ”Selected reaction monitoring” (SRM). 2DE og MALDI-TOF bekræftede
tilstedeværelsen af et akut fase respons i serumprøverne, men kun SAA2 og SAA4 kunne
14
identificeres. Da denne eksperimentelle opsætning både er tidskrævende og ikke den bedst
egnede til biomarkørudvikling, blev SRM-metoden, der blev udviklet, brugt til at
diskriminere og kvantificere de porcine SAA-isoformer med høj specificitet og følsomhed i
porcine serumprøver fra to separate, tidligere udførte infektionsundersøgelser. Isotopisk
mærkede SAA-peptider blev også benyttet til at kalibrere metoden for at gøre den
kvantitativ.
I Paper I blev SRM-metoden anvendt på sekventielle serumprøver fra et tidligere
eksperiment hvor svin blev inficeret med Sa og det var muligt at påvise alle cirkulerende
porcine SAA-isoformer. SAA-isoform-profilerne korrelerede fint med det totale SAA-
respons, tidligere bestemt med ”enzyme-linked immunosorbent assay” og en tæt korrelation
mellem mængden af SAA-isoform mRNA og SAA-isoform-profilerne i cirkulation blev også
fundet. Dette beviste, at antistof-uafhængig MS-baseret kvantifikation med høj sensitivitet
og specificitet er mulig og kan bruges til at diskriminere mellem et sæt af næsten identiske
cirkulerende SAA-isoformer direkte i serumprøver. Resultaterne viste desuden også, at
SAA2 uden tvivl er den primære cirkulerede SAA-isoform hos grise.
I dette Ph.D.-projekt blev eksisterende porcine serumprøver fra svin inficeret med Ap også
analyseret, og mens alle SAA-isoformer kunne påvises, kunne kun relative SAA-
koncentrationer måles, da de valgte proteotypiske peptider ikke var gode nok til absolut
kvantificering. Et yderligere sæt eksisterende porcine serumprøver fra svine med
eksperimentel Sa osteomyelitis blev analyseret, hvor alle SAA-isoformer blev påvist og
kvantificeret. Både relative og absolutte SAA-koncentrationer blev målt og korrelerede
godt med hinanden i dette prøvesæt. Relative og absolutte SAA-isoformkoncentrationer
blev sammenlignet på tværs af de tre prøvesæt, og hvert sæt havde en separat SAA-
isoformsammensætning. Dette underbygger tidligere fund, der antyder infektionsafhængig
vævstropisme af SAA-isoformer. Det vil være meget interessant at videreundersøge
potentialet af denne fremgangsmåde for at kunne opnå detaljeret karakterisering af
forskellige masse-varianter af andre cirkulerende proteiner og afdække mulige forbindelser
til deres oprindelsesvæv.
15
Chapter 1
Introduction
This Ph.D. project started with the idea that acute phase proteins, in particular, SAA (see below)
might have an untapped potential as biomarkers for neurodegenerative diseases like Alzheimer’s
disease and Parkinson’s disease [1-3], obesity and obesity-related disorders [4-6] and autoimmune
diseases [7-9], that all share the same inflammatory components [9-11], and are not caused by a
single identifiable extraneous component. These diseases are on a rise on a global scale and there
is an urgent need for an early and specific diagnosis, making the development of reliable biomarkers
extremely important.
The early stages of these diseases do not allow their use as biomarkers for differential diagnosis
and/or staging of disease, but the shared inflammatory component is important, as circulatory
serum acute phase proteins (APPs) are synthesized as part of the acute phase response (APR) in
response to inflammation [12-14]. The APPs constitute a very powerful set of inflammation
biomarkers [15-17], as these proteins by definition change their serum concentrations by >25% in
response to pro-inflammatory cytokines and are highly sensitive indicators of inflammation [16,
18, 19]. However, they are most commonly used as generalized markers of inflammation/infections.
Serum Amyloid A (SAA) is one of the most dramatically induced APPs, sometimes reaching a
more than 1000 fold increase in serum concentration [20-22]. It has an increased potential as a
disease-specific biomarker, as four SAA genes, exist in most species [23-25], and there seems to be
16
a preferential difference in hepatic expression as opposed to peripheral expression between isoforms
[26-28]. Interestingly, gene expression evidence shows that the specific isoforms of SAA seem to be
regulated differently depending on tissue and infection [29], and this tissue-specific differential
regulation of isoforms is a unique feature of SAA making it an exciting biomarker candidate and
thus the focus of this Ph.D. project. There is a growing amount of data on the differential increase
of various APPs, not only SAA, and by using MS-based methods with high specificity and
sensitivity, the biomarker potential of APPs such as SAA may be bigger than previously assumed
[15].
This Ph.D. project was based on porcine serum samples from two separate previously performed
infection studies; samples from a study performed by Olsen et al. (2013) [29] where pigs were either
intravenously infected with Staphylococcus aureus (Sa) or intranasally infected with Actinobacillus
pleuropneumoniae (Ap), and from a study by Jensen et al. (2016) [30] that used an Implant-
Associated Osteomyelitis (IAO) model where a steel implant was inserted into the tibia bone along
with a dose of Sa. Pigs were chosen as the porcine model permitting highly controlled challenges
to be performed, and disease development to be monitored closely in an organism that closely
mirrors human beings regarding immune response. Pigs were inoculated with porcine pathogenic
strains of either Sa or Ap, thereby creating an innate immune response from which the levels of
APPs have been detected [29]. The serum samples were analyzed with several MS-based methods
such as MALDI-TOF and SRM MS. The latter MS method was developed and validated in this
work specifically for the detection of porcine SAA isoforms and used to analyze SAA levels in the
porcine serum samples.
1.1 Objectives
The hypotheses for this Ph.D. project can be summarized as follows:
I. SAA has potential as a biomarker with increased diagnostic specificity by using targeted
MS to quantify specific SAA isoforms.
17
II. Quantification of tissue and infection specific SAA isoforms during infection will show that
SAA isoforms are differentially regulated depending on tissue origin and infection type.
III. Based on the resulting isoform profiles, targeted MS of SAA isoforms can be used to
diagnose specific infection types and specific types and locations of inflammation.
The aim was thus to develop a quantitative mass spectrometry-based method for accurate
quantification of SAA isoforms, since a method for accurate quantification of SAA isoforms on the
protein level is not currently available. Such a method will be an extremely important tool to
investigate the hypothesis that the accurate quantification of SAA isoforms in patients’ blood
samples can be used for the early, specific diagnosis of autoimmune, neurodegenerative and other
diseases having an inflammatory component. Furthermore, accurate mapping of SAA isoform
fluctuations will contribute significantly to the understanding of the role of this highly regulated
host response protein in health and disease. The project specifically relied on selected reaction
monitoring (SRM) mass spectrometry (MS) for antibody-free quantification of specific proteins in
complex biological samples.
1.2 Outline of Ph.D. thesis
This Ph.D. thesis will first present relevant background knowledge in chapters 2, 3, and
4, followed by a review of materials and methods in chapter 5.
After introduction, background, technical aspects, and materials and methods chapters (1-
5), results are presented in the form of a submitted paper in chapter 6, and in chapter 7,
additional results are presented along with a discussion of all results, a conclusion, and
perspectives on future work.
18
Chapter 2
Background
2.1 The acute phase response
The innate immune system directs the initial response to immunological stress initiated
by tissue and cell destruction and by pattern recognition receptors on the surface of innate
immune cells, recognizing pathogen-associated molecular patterns [13, 18]. This initial
response is the acute phase response; a prominent systemic reaction comprised of a number
of responses [12, 13, 18]. Following an inflammatory stimulus, the local release of pro-
inflammatory cytokines leads to the acute phase response (APR), activation of the
vascular system and activation of further inflammatory cells, which then leads to the
production of more cytokines and other inflammatory mediators that diffuse to the
extracellular fluid compartment and circulate into the blood [12, 18]. The cytokines induce
the liver to produce acute phase proteins (APPs) which, together with cytokines and other
inflammatory mediators generate a systemic response (fig. 1) [12, 14].
19
During an APR, the APPs are released, very quickly, into the circulation and their serum
concentrations increase substantially while the APP concentration in healthy animals and
humans vary from a baseline level to almost undetectable [31, 32]. The type and magnitude
of APP expression vary between the species, but in general, the APP concentration reaches
a peak within 24–48 hours and if no new stimulus appears and inflammation ceases, APPs
regulation will limit the response within 4 to 7 days after the challenge. In the case of
chronic inflammation, a longer and slightly increased serum concentration of APPs
compared to acute inflammation is observed [38-42].
Fig. 1 An inflammatory stimulus such as an infection or tissue injury, activates monocytes and macrophages that in turn release cytokines. These cytokines induce the production of acute phase proteins by the liver that collaborates with the cytokines to generate a systemic response. IL-1/6: Interleukin 1/6, TNF-α: tumor necrosis factor α, IFN-γ: Interferon γ, TGF-β: Transforming growth factor β, SAA: Serum Amyloid A, CRP: C-reactive protein.
2.2 Acute phase proteins
The APPs are broadly divided into two categories – the positive APPs whose circulating
levels increase in response to an inflammatory stimulus, and the negative APPs whose
circulating levels decrease in response to an inflammatory stimulus [31]. Profound
interspecies’ variations in the hepatic response have been observed, reflected by the serum
levels of individual APPs. In general terms it can be said that C-reactive protein (CRP),
Serum Amyloid P, Serum Amyloid A (SAA), α1-Acid Glycoprotein, Haptoglobin, α 1-
20
Proteinase Inhibitor, α2-Macroglobulin, Ceruloplasmin, Complement component C3, and
Fibrinogen behave as positive APR proteins, while Albumin and Transferrin respond as
negative APPs [31, 33-36] (fig 2).
Fig. 2 Shown are peak concentration change during a prototype APR of selected APPs in Swine, Cattle, Dog, Cat, Human being, Mouse, Rat, and Rabbit. NEG: Decrease in concentration (10–30%); 0: No change; I: 50% to 100% increase; II: Between 100% and 10 times increase; III: More than 10 times increase. * Normal level and acute phase concentration changes of Serum amyloid P differ strongly between different mouse strains. ?: Nothing reported about the reaction of the protein indicated in the literature [35].
Synthesis of positive APPs is induced mainly by cytokines such as Interleukins IL-6, IL-
8, IL-1β, IL-22, and tumor necrosis factor α (TNFα) [2, 13–16]. Simultaneously negative
APPs are downregulated. The most typical APPs and their general function are
summarized in table 1. The early and significant onset of certain APPs have made them
interesting candidates of inflammation biomarkers for diagnostic purposes in human and
veterinary medicine. APPs have poor specificity when it comes to detecting the cause for
a disease, but they have high sensitivity in broadly detecting the different conditions that
can cause subclinical infection or inflammation [37-39]. Neurodegenerative diseases like
Alzheimer’s disease and Parkinson’s disease [1-3], obesity-related disorders [4, 5], and
obesity itself [4, 6], and autoimmune diseases [7-9] have all been associated with increased
levels of APPs.
21
APPs are have several important applications as they can provide:
- Alternative means of monitoring animal and human health
- Objective information about the presence and extent of ongoing lesions in humans
and animals
- Information regarding the presence of a disease
Table 1. Classical positive APPs upregulated during an APR - exact classifications and kinetics are species-specific Acute phase protein Known mail role(s)
C-Reactive protein Binding pathogens and damaged cells, opsonin, complement
activation [13, 40, 41]
Serum amyloid P Binding pathogens, opsonin, complement activation,
regulation of innate defense and repair [42, 43]
Serum amyloid A
Opsonin, regulation of innate defense, induction of
extracellular matrix degrading enzymes, chemoattractant,
apolipoprotein with HDL, regulation of lipid metabolism [13,
44-47]
Lipopolysaccharide
binding protein
Pattern recognition, binding, and disaggregation of
lipopolysaccharides, opsonin [48, 49]
Complement factors Pathogen recognition, destruction, chemotaxis, opsonin,
vasoregulation [13]
α1 Acid glycoprotein Binding of plasma proteins and mediators, chaperone,
regulation of innate defense [13]
Fibrinogen Clot formation
Haptoglobin Scavenging of hemoglobin, antioxidants, angiogenesis,
chaperone [13, 50]
Additional references [13, 16, 29, 51, 52]
22
- Prognostic information, using the magnitude and duration of the APR as a
reflection of the severity of the disease.
- A practical means to monitor the effectiveness of antibiotic therapy in the
treatment of bacterial infections
In addition, the APPs can help monitor optimal animal growth by detection of small
changes in APPs concentrations, using an acute phase index [35, 39, 53-56].
In humans, CRP indicates low-grade inflammation associated with osteoarthritis and the
risk for cardiovascular events [57, 58]. In most healthy beings, plasma levels of CRP are
usually 1mg/L with normal being defined as <10 mg/L (0.06-8 mg/L), and during an APR
the CRP concentration exponentially increases (100-500 mg/L), doubling every 8-9 hours
[39]. CRP is used widely as diagnostic and prognostic markers for infectious diseases,
severe inflammation, sepsis, acute kidney injury, impaired lung function, cardiovascular
risk, inflammatory bowel diseases, Crohn’s disease, and more [16, 40, 59-61]. CRP
measurements are generally performed by immunoassays using species-specific CRP
antibodies [39].
SAA is normally found at 20-50 mg/L, and the levels rise up to 1000-fold within the first
24h of an APR, and studies indicate SAA involvement in chronic inflammatory conditions,
atherosclerosis, lung disorders [62-67]. SAA has been proposed as a possible serum
biomarker for many types of tumors such as ovarian, lung, renal, endometrial, uterine,
and melanoma [68-76]. For SAA, monoclonal antiserum against human SAA has been
used in a sandwich ELISA, and can be used successfully for other species as well [39].
APPS may also be measured by several methods using mass spectrometry.
Numerous studies indicate different patterns of APP increase, not solely explained by one
common factor as the levels of circulating APPs depend on several factors [13, 37, 77-79].
In humans, CRP is more elevated during bacterial pneumonia than in severe viral acute
respiratory syndrome. In the latter case, alpha 1-antitrypsin, SAA, and Haptoglobin is
instead highly increased [15, 79]. In different cancer types, a dramatic rise in SAA
23
concentration is seen while other acute inflammatory signs are still clinically silent. A
differential expression of SAA isoforms depending on tissue and infection type, have also
been observed [29, 80, 81]. These differences in APP levels might be due to different
costimulatory cytokines, that can be secreted depending on the underlying failure and in
that way cause the varying hepatic expression rates, kinetics, and nascent storage size [13,
15, 82], or it could be due to a differential metabolism and clearance of APPs, differentially
regulated by the underlying alteration [44, 83-85].
2.2.1 Serum Amyloid A
Serum amyloid A (SAA) is a small (9 to 14 kDa) highly regulated acute-phase protein [22,
44, 86]. It is present in the blood circulation at low levels at homeostasis and is, during
the APR accompanying inflammation in most vertebrates induced several 100-fold,
becoming the predominant apolipoprotein of high-density lipoprotein, replacing
apolipoprotein A1 [22, 44, 86]. SAA has been found involved in many processes such as
binding of, transporting and scavenging cholesterol to the liver during inflammation [87],
inhibition of phagocyte oxidative burst [54, 88], inhibition of platelet aggregation [35],
binding to Gram-negative bacteria and opsonization [89, 90], inhibition of lymphocytes
and endothelial cells proliferation [91] and detoxification of endotoxin by binding to
lipopolysaccharide [83, 91]. SAA binds to high-density lipoprotein (HDL) [92, 93] and
during inflammation, increased incorporation of SAA into HDL is observed, which appears
to accelerate the removal of these particles from the circulation [31, 34]. SAA is found to
function as a chemoattractant for human monocytes, neutrophils, and T lymphocytes
recruiting inflammatory cells into inflammatory lesions [31, 44, 94, 95]. SAA is therefore
widely accepted as important during the early phases of inflammation [96, 97]. The
predominant site for synthesis of SAA is the liver [25, 29, 98], but SAA has been detected
24
in many different tissue types; liver, lymph node, spleen, tonsils, and blood leukocytes [99,
100], lung, salivary gland and diaphragmatic muscle [101] and mammary tissue [102, 103].
Despite the small size of SAA, its 3-dimensional structure(s) has been difficult due to its
association with HDL and subsequent poor aqueous solubility despite hydrophobic residues
not prominent in the sequences. In 1986, Turnell et al. proposed a two α-helices and a β-
sheet region model based on algorithms, and a SAA monomer structure by
multiwavelength anomalous dispersion using a selenomethionine derivative to 2.2-Å
resolution was determined by Lu et al. (2014) and supported by Derebe et al. [104-107].
The monomer comprises four α helices in a cone-shaped array with the C-terminal residues
wrapped around the bundle (fig. 3 A). Consistent with earlier findings Lu et al. found a
hexamer formed by two identical trimers, and potential HDL binding sites were predicted
to overlap near the residues R-1, R-62, and H-71 in helix h4 [106, 108-111].
With this structural model, Frame and Gursky noted that there are both hydrophobic and
hydrophilic faces in helices 1 and 3, and that the hydrophilic faces interact with helices 2
and 4 leaving the hydrophobic domains able to bind the lipid surface of HDL. SAA can
thereby serve as a “hub”, binding HDL on one side and interacting with
glycosaminoglycans, cystatin C, retinoic acid, and cell receptors on the other side, with
domains specifically involving the C-terminus (fig. 3 B) [106, 112].
25
Fig. 3 (A) Atomic structure of hSAA1.1 (PDB ID 4IP8). Blue: h1, residues 1–27, teal: h2, 32–47, green: h3 50–69, orange: h4, 73–88, red: CT tail 89–104 including a 3/10 helix h0, 90–96. Asterisks mark predicted amyloidogenic segments. Indicated is also the highly conserved GPGG motif [112, 113]. (B) Structural model of SAA bound to an HDL particle. The SAA monomer binds to HDL on the concave apolar surface site of helices h1 and h3 (blue and green rectangles), thereby protecting the amyloidogenic segments from initiating the misfolding. The C-tail, cleaved in AA amyloidosis, binds cell receptors and ligands, bridging them together and mediating interactions. When inflammation is resolved and SAA plasma levels drop, SAA is released from HDL and this dynamic network dissociates. Figure modified from [112].
One precise function of SAA and its role during the acute phase response has not been
established, but its participation is clear. A recent comprehensive review by George H.
Sack Jr provides an excellent insight into what is known regarding the function of SAA,
and gather information from numerous SAA studies into a detailed and comprehensive
overview of SAA function [62]. Well-described characteristics include the ability of
circulating SAA protein to bind to high-density lipoprotein (HDL) [92, 93] and to directly
induce pro-inflammatory cytokines and chemokines in monocytes and granulocytes [114].
It has chemotactic effects in various cell types through binding to formyl-peptide receptor-
1 (FPR1), formyl-peptide receptor-2 (FPR2), and Toll-like receptor-2 (TLR2) [115-118].
In cultured human neutrophils, SAA was found to cause a dramatic increase in TNFα, IL-
1β, and IL-8 when incubated with the cells [119].
26
Fig. 4 A wide selection of SAA interactions through receptors has been determined. HDL: high-density lipoproteins, SR-BI: scavenger receptor class B type I, FPR2: formyl peptide receptor 2, P2X7: An ATP receptor, TLR2/4: Toll-like receptor 2/4, RAGE: receptor for advanced glycation end-products. SAA from blood and tissue is shown separately to illustrate their different functions [116].
Most recently cytokine-like properties have been ascribed to SAA, being intimately
implicated in the intestinal control of Th17, immunity itself being directly induced in
epithelial cells of mucosal tissue by IL22, which can be controlled by the microbiota in the
absence of inflammation [120-122]. In very brief summation:
- SAA participates in the APR
- SAA helps link the complex network of cells (macrophages, hepatocytes, and others)
and proteins mediating inflammation
- SAA functions as a cytokine-like protein in cell-cell communication as well as
feedback in pathways that are immunologic, neoplastic, and protective during
inflammation [62].
Most of these interactions are summarized in fig.4 [116].
27
Across species, the sequence of SAA and its isoforms has been very well conserved
throughout evolution (fig.5) and this, together with its rapid and huge acute phase
response, also implies that SAA proteins probably have important functions during the
response to tissue perturbations [23, 123]. The SAA superfamily consists of four very
similar SAA genes, each coding for a specific isoform of SAA (SAA1-4). They have been
demonstrated having the same four-exon three-intron structure in all species [38, 44] with
an 18 amino acid long signal sequence is present in the initial transcript, which is removed
in the resulting serum proteins. The human SAA gene cluster is comprised of four gene
loci (i.e. saa1, saa2, saa3, and saa4), clustered within a 160 kb region of chromosome
11p15.1 [124-126]. Within the species, SAA1 and SAA2 are highly homologous, while
SAA3 is less homologous to the other SAA genes and SAA4 is the most divergent isoform
[127-130]. The evolutionary relationships of the SAA superfamily members show that all
of them have undergone concerted evolution on the nucleotide level. This means that there
are no strict evolutionary homologs between the mammalian SAAs known, as they have
been homogenized within the species, e.g. human SAA1, SAA2 and SAA3p are more
related to each other than to their murine orthologous, as the murine genes are more
related to each other than to the human orthologous [23, 131]. Functions and disease
associations can therefore not be presumed to be the same for one type of SAA in one
species as in another [23].
28
Fig.5 Shown are available sequences of SAA isoform aligned across species, and categorized within SAA isoform type. Each sequence is shown next to its UniprotID1. Human (Homo sapiens), pig (Sus scrofa), mouse (Mus musculus), mink (Neovison vison), hamster (Mesocricetus auratus), and rabbit (Oryctolagus cuniculus). Colored squares and * indicate identical amino acids across sequences.
The detection of the individual SAA isoforms remains a difficult process, and isoforms
have in many cases been wrongly identified. Many of the early reports of SAA did not
distinguish among SAA isoforms at all, others have used outdated nomenclature and in
1 Due to updates, the sequences for porcine SAA2 and SAA4 are currently moved to the Uniparc archive, and can be found at uniprot.org/uniparc with the following ID: SAA2: UPI0001E86A58, SAA4: UPI0001E86A59
29
general, the field has been quite confusing and not always easy to navigate. Purification
and detection of SAA isoforms are difficult, as SAA tends to either form large protein
aggregates [15] or precipitate [13] when separated from HDL, and the fact that the SAA
isoforms have very similar sequences, make antibodies for SAA isoform distinctions very
difficult to develop. No solid reports of SAA isoform antibody have been produced, except
for one substantiated recent report demonstrating a selective antibody which does not
react with porcine SAA4, but do also not discriminate between porcine SAA2 and SAA3,
nonetheless (SAA4 being the most divergent of the SAA isoforms) [132]. Immuno-based
SAA assays are most often not based on purified species-specific SAA but instead based
on calibration and standards consisting of recombinant SAA, heterologous SAA, or pooled
acute phase serum [133-137]. Using species-specific calibration material from the native
purified protein is important to obtain precise measurements of SAA isoforms, but
currently mainly size-exclusive chromatography and isoelectric focusing (IEF) are used
[138-140], using delipidation steps [140-142] like ultracentrifugation or hydrophobic
interaction chromatography [143, 144] to obtain SAA, not in complex with HDL [137].
In IEF, molecules are separated by their isoelectric point (pI) in a gel with a pH gradient,
making proteins in a pH region below its pI positively charged and migrating toward the
cathode. As it migrates through the gradient of increasing pH, the overall charge will
decrease, and when the protein reaches a pH region corresponding to its pI, there is no net
charge and migration stops, thereby focusing the protein at a point in the pH gradient
corresponding to its pI. IEF is followed by Western Blotting where antibodies are used to
detect the protein of interest. SAA isoform detection at the protein level is possible with
IEF as the different isoforms of SAA have different pI due to their different amino acid
sequences, making isoelectric focusing the primary method used for detection of the SAA
isoforms on the protein level for many years. IEF is not a precise method, with the results
relying on manual inspection of more or less slurred blot as presented in fig. 6. The earliest
information regarding the SAA isoforms on the protein level was based on IEF [130, 140,
145-149]. The method, however, is not very precise and as mass spectrometry-based
30
methods in the field of proteomics have improved, they may provide the sensitivity and
specificity needed for SAA isoform detection.
Fig. 6 Human SAA isoforms extracted from acute phase sera were electro-focused between pI 5-8 and detected by Coomassie blue staining (track I) and by electroblotting to PVDF membranes and immunoenzymatic detection (tracks 2-4) [149].
IEF results have been the base of most knowledge regarding SAA isoforms, and a very
interesting finding was that some SAA isoforms are present locally in e.g. colostrum. SAA3
has been found in bovine, equine, and bovine colostrum, and Sack et al. found SAA1 in
human colostrum [102, 150]. In several species, SAA3 has been proposed as the main
isoform of local inflamed tissue, as it is the only SAA isoform detected locally in e.g. joints
in horses [27], dogs [151], bovine udders [28] and adipose tissue in mice [152]. SAA3 has
also been found upregulated in lung tissue of mice during an APR induced by inhalation
of zinc oxide nanoparticles [153] and in porcine bone in pigs with implant-associated
osteomyelitis [154].
During the acute phase response, the predominant site for synthesis of SAA is the liver.
Generally speaking, SAA1 and SAA2 are expressed preferentially in the liver while SAA3
is found in hepatic as well as non-hepatic tissues during an APR [25, 29, 98]. In the pig,
31
SAA gene expression is also induced both hepatically and extrahepatically during an APR
[99, 155] and elevated levels of circulating SAA proteins have been found [17, 156]. Murine
saa3 is induced in several different tissue types, whereas there have been no reports of in
vivo transcription of the human homolog SAA3p [98, 128, 157, 158]. Murine saa4
expression has only been found in liver tissue, but human SAA4 has been found expressed
in both liver and extrahepatic tissue [129, 130, 159]. The SAA isoforms (SAA1, SAA2 and
SAA3) have been recognized at the transcriptional, as well as protein level, in many other
species than human and mice [44, 160], where the main isoform of local inflamed tissue is
most often found to be SAA3 [27, 28, 80]. A porcine SAA3 gene transcript has been
characterized [100] and SAA3 is proposed to be the most prominent circulating SAA
isoform in pigs [101, 161], making the pig unique as SAA3 has not been detected as a
circulating isoform in other species examined [27, 28, 80, 102, 152, 162-164].
Olsen et al. [29] studied the relative expression of porcine genes encoding SAA1, SAA2,
SAA3, and SAA4 in tissue samples from lung, liver, and spleen, respectively, from pigs
experimentally infected with either the Gram-negative swine specific pathogen
Actinobacillus pleuropneumoniae (Ap), or a porcine strain of the Gram-positive pathogen
Staphylococcus aureus (Sa). They identified two previously uncharacterized porcine SAA
transcripts as SAA2 and SAA4, and found no transcript corresponding to porcine SAA1,
concluding that porcine SAA1 is most likely a non-expressed pseudogene. They found
that both infection types induced increased hepatic expression of SAA2, SAA3, and SAA4
and found elevated pulmonary and splenic expression of SAA3 [29]. An interesting detail
was that different types of infection models differentially induced the expression of the
SAA isoforms. In both experiments a marked increase in the expression of SAA2 was found
in liver tissue samples from infected pigs compared with control pigs. In lung tissue from
Ap-infected pigs an increased SAA2 expression was observed, and while the change was
negligible in the lung tissue from the Sa-infected pigs, in spleen the SAA2 regulation was
negligible in both infection experiments [29].
32
In both infections model, severe infectious states were observed with clear systemic
affection, but during the Ap infection, it was observed that the lowest increase in relative
SAA3 expression was found in spleen tissue when compared to liver and lung, while the
opposite was the case in the Sa infection experiment. This tissue-specific expression pattern
of SAA3 mRNA makes it a potential marker of tissue-specific involvement of acute
inflammation in pigs [29]. In the Ap-infected pigs, when compared to control pigs, an
average increase of 8.6-fold was observed in the SAA4 hepatic transcription levels, but in
the Sa-infected pigs only a slight increase was observed when comparing infected pigs and
control pigs. In samples from lung and spleen tissue, SAA4 transcription was below
detection [29]. The tissue-specific differential of isoforms during different diseases seems to
be a unique feature of SAA and makes it a very interesting biomarker candidate.
It is possible to imagine a form of SAA fingerprint where specific SAA isoform expression
profiles indicate the presence of a specific infection or disease, improving diagnosis and
discovery early during the disease or infection; it might also improve the possibilities of
early treatment.
As one of the major acute phase proteins, SAA is implicated in a wide range of diseases
and has been proposed as a biomarker for many of these diseases [15, 62-64, 165]. For
example, local SAA1 and SAA4 production have been found in the colon suggesting a role
in colonic tumorigenesis and may have both prognostic and therapeutic applications, as
no SAA was detected in normal-looking colonic epithelium [166]. In ovarian carcinomas,
the SAA1 and SAA4 genes are overexpressed when compared with normal ovaries, and
SAA levels increase when epithelial cells progress through the benign and borderline
adenomas onto the primary and metastatic adenocarcinomas. The high serum levels of
SAA strongly correlated with the ovarian tumor marker CA-125, suggesting an
involvement of SAA in ovarian tumorigenesis [167]. Overexpression of SAA has been
reported in renal-cell, uterine serous papillary, and endometrial carcinomas [68, 69, 73,
77]. Local SAA isoform levels may be used as a biomarker for cancer aiding early detection
and diagnosis.
33
Chapter 3
Technical aspects
As mentioned above, others and we have shown by transcriptional analysis, that SAA
isoforms are differently expressed depending on tissue location and type of trigger [29].
Until now, however there has been no way of quantifying circulating SAA protein isoforms,
as no rapid and reproducible method exists for this purpose. We therefore developed a
specific, quantitative and antibody-free mass spectrometry-based method for quantifying
circulating porcine SAA isoforms. Porcine serum samples from two separate previously
performed infection studies were analyzed; one with pig infected with either Sa or Ap, and
one with pigs infected with Sa in an IAO model, allowing us to correlate the tissue-specific
expression of SAA isoforms with SAA isoforms in the circulation. This is of interest not
only for elucidating possible specific biological functions of SAA isoforms, but also for
describing the possible role of SAA isoforms as differential biomarkers of infection. The
phases of experimental work performed in this Ph.D. project are depicted in fig.7 and the
technical aspects of the mass spectrometry methods used will briefly be described in the
following section.
34
Fig. 7 Experimental parts of the Ph.D. project simplified into four groups. The project started with SAA ELISA to measure total SAA levels in the serum samples, then 2DE, and MALDI-TOF was performed to investigate native SAA isoforms and subsequently a SRM method was then developed to further analyze the samples. Future work of interest would use methods such as Parallel reaction monitoring (PRM), SWATH-MS as a method for data-independent acquisition (DIA), and tandem mass tags (TMT).
3.1 Mass Spectrometry
Mass spectrometry (MS) is a valuable tool for the analysis of proteins with a wide range
of variations and applications and is frequently used within biomarker research and
veterinary research of APPs [168-170].
A mass spectrometer has six essential interlinked components: a system for sample
delivery, a vacuum system, an ionization chamber (ion source), a mass analyzer where
ions are separated according to mass, a detector, and a computer for data analysis (fig.8)
[171, 172]. The basis of mass spectrometry can be summarized as follows: First, the sample
is introduced into the ion source through a series of ion lenses and ion optics that focus
and guide the ion beam to a mass analyzer. A high vacuum is required to prevent the ions
from reacting with or being scattered by air molecules. Liquid chromatography (LC) is
often used here, to separate sample molecules by their interaction with absorbent material
in an LC column, which causes different flow rates as they pass from the column into the
35
ion source where they are ionized. Ionization can be done by e.g. electrospray ionization
(ESI) or Matrix-assisted laser desorption ionization (MALDI). Regardless of the ionization
method applied, a successful ionization requires an energy transfer to the analyte, which
is extremely important since the applied energy and the way it is applied to determine the
produced mass spectrum. The ions are subsequently directed into the mass analyzer where
they are sorted.
Fig. 8. Schematic overview of a mass spectrometer. Through a sample delivery system, the sample is introduced into the ion source through a series of ion lenses and ion optics that guide the ions to the mass analyzer, where they are sorted according to their mass-to-charge ratios and recorded by a detector. The recorded signals are converted into mass spectra by a computer. A series of vacuum and turbo pumps maintain the vacuum.
The sorting can be according to their mass-to-charge ratios (m/z), or based on the time
the ion take to fly from the ion source to the detector (time-of-flight) [171-173]. The ions
reach the detector at different times and this information is collected and gathered into a
MS spectrum. The spectrum is recorded as a bar graph where ion intensities are normalized
so that the largest peak in the mass spectrum, called the base peak, is set to an intensity
of 100 units, and all other ion abundances are expressed as a percentage of this. An
36
additional stage of ion fragmentation may be included before detection to obtain structural
information in a technique known as tandem MS [172, 173].
3.2 Liquid Chromatography-Mass Spectrometry (LC-MS)
Samples for MS analysis may be solid, gas phase, or in liquid. Each type of sample has a
designated sample delivery system. In this Ph.D., liquid samples have been used, and focus
is therefore chosen to be on this type of sample delivery system i.e. Liquid Chromatography
MS (LC-MS) [171, 172]. LC-MS combines the physical separation capabilities of LC with
the mass analysis capabilities of MS, and with its high molecular specificity and detection
sensitivity it can provide a structural identity of the individual components [174, 175].
Electrospray ionization (ESI) is commonly used to couple High-Pressure Liquid
Chromatography (HPLC) with mass spectrometry, by facilitating the transition from a
high-pressure environment to high-vacuum conditions needed at the MS analyzer [174,
175]. HPLC uses pumps to pass the sample mixture and solvent as a pressurized liquid
through a column filled with solid adsorbent material to separate, identify, and quantify
each component in a mixture. The components in the sample interacts with the adsorbent
material in a slightly different manner, which causes different flow rates for the different
components and thereby separates the components as they flow out of the column [171,
172].
The components of the system being analyzed are termed analytes. As mentioned earlier,
they elute at specific time points from the chromatographic column and enter the MS
ionization chamber. A time delay of elution of the different analytes is caused by
physicochemical interactions between the stationary phase on the chromatographic
column, the analytes, and the mobile phase. The intensity of the detectable m/z is
registered inside the MS detector, its value representing the approximate amount of ionized
molecules of each m/z detected at the exact retention time (rt) [176]. A typical dataset
37
from an LC-MS experiment is represented as a set of data points in a three-dimensional
space defined by the rt, mass-to-charge ratio (m/z), and intensity (counts), as shown in
fig. 9. At a specific rt in the mass spectrum, one “slice” of the 3d data set is selected, as
an accumulation of all the detections from the MS detector during a very short time-period
[176].
Fig. 9. 3D view of a LC-MS mass chromatogram with intensity, retention time and mass-to-charge ratio making up the mass spectrum from the recorded signals in an LC-MS run [177]
3.3 Matrix-assisted laser desorption ionization (MALDI)
MALDI is one of the most commonly used protein ionization techniques. It was first
described in 1985 by Michael Karas et al. and successfully used for protein ionization in
1988 by Tanaka et al. [178, 179]. The method is characterized by easy sample preparation
and large tolerance to contamination by salts, buffers, detergents, etc. It is a soft ionization
technique with minimal fragmentation yielding intact molecular ions.
38
Different types of samples such as cells, tissues, and biological fluids can be analyzed by
MALDI. The extracted and purified protein or peptide mixture is mixed with a matrix –
an organic compound with a strong optical absorption at the laser wavelength, usually in
the UV region, that will facilitate the ionization by absorbing the laser energy and
transferring it to the sample analytes. The sample/matrix mixture is placed on a target
plate, usually a stainless steel plate with multi-well support that allows for a spatial
separation of multiple samples. The sample mixture airdry whereby the sample co-
crystallizes with the matrix. The sample-matrix crystals are then irradiated by a laser
pulse, inducing desorption of the crystals surface i.e. the sample analytes and matrix
molecules desorb and sublimates resulting in the formation of gas-phase ions. In this vapor
phase, proton transfer from the matrix to the analyte molecule occurs. (fig. 10) [175, 180,
181]. The ions that are detected first are the smallest, as ions with different m/z are
accelerated to different velocity inversely proportional to their mass, at a fixed kinetic
energy. This is known as linear mode and the mass range achievable here is ranging from
tens to hundreds of thousands of Daltons [175, 180, 181]. In the resulting MALDI mass
spectrum, the dominant peaks will mostly correspond to the (singly) protonated molecule.
Multiply charged ions, doubly, and triply charged forms, are at low abundance [175, 181].
Fig. 10. The analyte is embedded in a large excess of a matrix compound deposited on the target plate. A brief laser pulse irradiates the spot thereby desorbing and ionizing the sample. The energy of the laser is absorbed by the matrix which carries the analyte molecules into the gas phase. The analyte molecules are ionized by being protonated with the matrix molecules [182].
39
3.4 Electrospray ionization (ESI)
Another soft ionization technique is ESI. Ions are generated by applying a strong electric
field, under atmospheric pressure, to a liquid passing through a capillary tube with a low
flow rate (normally 1– 10 µL/ min). The high ion spray voltage applied to the electrode
tip of the emitter results in the formation of a Taylor cone containing positively or
negatively charged ions depending on machine mode, depending on the HPLC buffer
system and molecules being analyzed. Charged parent droplets are released from the
Taylor cone and as the solvent molecules evaporate, the charges are brought closer
together until they reach the threshold for ionic repulsion and a columbic fission occurs.
The charged ions enter the MS through a heated capillary, which evaporates the remaining
solvent molecules. Both singly-charged and multiply-charged ions are generated (fig.11).
Depending on the flow rate of the electrosprayed liquid, ESI can be subdivided into
microelectrospray [183] and nanoelectrospray [184]. Microelectrospray ion sources maintain
flow rates over 500 nL/min, whereas nanoelectrospray is of approximately 30 nL/min.
Fig. 11. The sample is introduced from HPLC to ESI and positively charged ions are generated as the solvent-analyte mixture passes through the spraying nozzle. The solvent evaporates resulting in columbic fission of the charged droplets, forming charged ions that are detected by the mass analyzer as described above.
40
3.5 Mass analyzers
The mass analyzer of the mass spectrometer separates the ionized atoms and molecules
according to their mass-to-charge ratio (m/z) by the use of electrical or magnetic fields to
apply a force to charged particles to induce separation or differential focusing of ions into
the detector [171]. The resolution of a mass analyzer is an important instrumental
characteristic that measures the instrument’s ability to separate ions i.e. the ability to
distinguish two peaks of slightly different mass-to-charge ratios in a mass spectrum. A
larger resolution indicates a better separation of peaks, and higher resolution also leads to
higher absolute mass accuracy [173]. The resolution varies from instrument to instrument
as well as from instrument combinations such as MALDI-TOF, Q-TOFs, QQQs, Q-
Orbitrap etc. The specifications and abilities of the instruments are changing fast, as they
are continuously improved. It is therefore important to investigate the resolution of the
instrument one is considering to use for an experiment and determine whether the
resolution and specifications are suited for the experiment in question. There are several
different types of mass analyzers as presented in table 2, and as mentioned several
hyphenated mass analyzers where different mass analyzers are coupled together, have also
been developed.
Tabel 2. Overview of selected mass analyzers and their separation principle
Mass analyzer m/z separation principle
Quadrupole Stability of ion trajectory
TOF Velocity of ions
Orbitrap Oscillating along the axis
Ion trap Resonance frequency
41
3.6 Quadrupole Mass Analyzers
The quadrupole mass analyzers are ion beam analyzers, comprised of four parallel rods
with cylindrical geometry arranged in a specific three-dimensional fashion around the path
of the ion beam between the ion source and the detector (fig. 12).
A DC voltage is applied across the rods so that two rods have positive charges and two
have negative charges - opposing rods carry the same charge. An alternating current is
applied to both pairs of rods, and when the ions enter the electric field created within the
rods, they oscillate, and only ions with a stable oscillation will have sufficient kinetic
energy along the length of the filter to reach the detector. Depending on the voltage and
frequency of the quadruple, only certain ions will make it through, while other ions oscillate
with increasing amplitude until they collide with the rods and fail to reach the detector
(fig. 12) [171, 172].
Fig. 12. Schematic diagram of a quadrupole mass analyzer. Four parallel rods are arranged in a 3D fashion around the path of the ion beam between the ion source and the detector. Only ions with stable oscillation (“correct m/e”) will have sufficient kinetic energy along the length of the filter to reach the detector [171].
42
The quadrupole mass analyzer can operate in different modes depending on voltage and
frequency settings:
- It may allow ions with an m/z within a certain interval to be transmitted
simultaneously
- It may allow ions of a specific m/z ratio through, whereas ions with different m/z
ratios collide with the rods and are lost
- It may allow ions of increasing m/z to pass through in small intervals (e.g. 0.1 to
0.2 Da). In this operation mode, the quadrupole mass analyzer is said to be
scanning.
Because of these filtering options, a quadrupole mass analyzer is commonly referred to as
a mass filter [172]. The quadrupole mass analyzers are commonly known as Qs, and if
quadrupole mass analyzers are coupled together with other mass analyzers, they may be
called Q-TOFs, QQQs, or Q-MALDIs etc.
3.7 Time of Flight (TOF) mass analyzers
In TOF mass analyzers, ions are produced over a very short period using e.g. pulsed laser
desorption to expel the ions from the ion source. The generated charged ions are
accelerated by a constant electric field and acquire a similar kinetic energy. They
subsequently fly at constant, ion-specific speed towards the detector through a free-field
region called ‘flight tube’ being separated according to their m/z (fig.7). As the ions have
different masses, their velocity is different from each other and thus each ion traverses a
fixed distance to the detector in a different time [171, 172]. In theory, there is no upper
limit to the mass range of the TOF instrument, making them very useful for the analysis
of large molecules. TOF mass analyzers also have fast cycles as they technically transmit
43
all m/z ions (full scan mode) [172]. Using ion mirrors known as reflectrons may increase
the resolution of the TOF instrument as the reflector increases the ion drift path length
by bouncing ions off it (fig. 13) [180].
Fig. 13. Basic principles of TOF MS with a reflectron. The reflectron reflects the ion beam with a constant electrostatic field, towards the detector. Ions with a high kinetic energy (smaller dots) penetrate deeper into the reflector thereby taking a slightly longer path to the detector than ions with lower kinetic energy (bigger dots circles).
The reflector is made up of an electric field with a rising electrode potential, composed of
a circular electrode that directs the ions by changing the direction of the field. Ions pass
through the reflector until they lose their kinetic energy, their flight paths varying
according to the differences in their kinetic energy so ions with high kinetic energy stay
longer in the reflector than low energy ions. This leads to the correction of a flight time
by the reflector greatly improving the resolution of the TOF analyzer. The TOF mass
analyzers are very often coupled with a MALDI source.
44
3.8 Tandem MS (MS/MS)
While single-stage mass spectrometry does play a role in protein identification, many
protein identifications are performed by tandem mass spectrometry (MS/MS) of peptides
derived from protein digests [185, 186]. By coupling two or more mass analyzers together
with a reaction region between them, separate MS functions can be applied to analyze the
sample mixture using sequential processes [171, 186]. A variety of instrument setups can
be used for MS-MS, but the basic concept is that precursor ions are selected in the first
MS. The ions then move into a second MS that acts as a reaction collision cell, fragmenting
the precursor ions into product ions (also called daughter ions). The product ions are
focused by the third MS and detected as individual ions or as spectra [171, 186].
Fig. 14. Cleavage of the amide bonds in the peptide backbone generates fragment ions. a, b, and c ions are generated from the N-terminus while x, y, and z ions are generated from the C- terminus [187].
Each MS can be set to scan a mass range or select one or more individual ions [171]. Fig.
14 illustrates the ion pairs that can be generated from backbone bond fragmentation when
MS/MS is performed on a peptide. Fragmentation of the backbone from a singly-charged
45
peptide ion produces an ion and a molecule. N-terminal fragments give rise to a, b, or c
ions, and C-terminal ions are called x, y, or z ions. The most common cleavage at the
amide bond results in b or y ions i.e. tryptic peptides with a C-terminal basic residue
generally exhibit predominant y ions [187, 188]. The peptide fragment ions produced in
the collision cell provide sequence information that can be searched against protein
database, using algorithms to match the measured peptide MS/MS, and thereby enabling
protein identification [189].
Collision-induced dissociation (CID) is a slow activation method, and the most widely
used MS/MS technique. The precursor ions are accelerated to higher kinetic energy and
allowed to collide with neutral gas atoms or molecules such as helium, nitrogen, or argon.
This collision results in a conversion of some of the ions kinetic energy into internal
vibrational energy which subsequently results in breakage of covalent bonds in the peptide
[189, 190]. In addition to CID, there are other fragmentation techniques such as electron
transfer dissociation and high collision dissociation, and the fragmentation can be done in
a variety of ways [189].
46
Chapter 4
Discovery and Targeted proteomics
In the early years of proteomics, the focus was mostly on identifying individual proteins
or protein complexes, and while proteomic analysis is a perfect tool for discovery and
qualitative identification of high numbers (<1000) of proteins in cells or from other
biological samples, there is also a need to quantify these proteins.
Within the fields of proteomics, a distinction is made between discovery proteomics and
targeted proteomics. Discovery proteomics intends to identify as many proteins as possible
across a broad dynamic range, wherein targeted proteomics, the analysis focuses on a
subset of proteins of interest in a sample – one or more targets are chosen and focused on.
Discovery proteomics is often used as a way to inventory as many proteins as possible in
a sample or to detect the differences in the abundance of proteins between samples. In
contrast, targeted proteomics are used in pharmaceutical and diagnostic settings to
quantifying proteins in complex samples. Targeted proteomics may follow discovery
proteomics to quantify proteins of interest found during the discovery experiments with
high precision, sensitivity, specificity, and throughput. Especially within the field of
47
biomarker research targeted proteomics are valuable tools. Most biomarkers are found in
blood plasma, as it is accessible and assumed to contain a quantifiable amount of proteins
that reflect the physiological and pathological state of the human body [191, 192]. The
blood-based biomarkers, however, are complex mixtures with a wide dynamic range of
protein concentrations [193].
One of the most common approaches for verification and validation of biomarker
candidates is the sandwich Enzyme-Linked Immunosorbent Assay (ELISA), a method with
high specificity but restricted to the availability of antibodies for novel candidate proteins.
The targeted LC-MS methods Parallel Reaction Monitoring (PRM) and Selected Reaction
Monitoring (SRM) are excellent methods for protein quantification as they are capable of
detecting low abundance proteins in highly complex sample mixtures with high selectivity,
reproducibility, and sensitivity.
4.1 Parallel Reaction Monitoring (PRM)
Parallel Reaction Monitoring (PRM) is a targeted MS method that simultaneously
analyzes all fragment ions of a pre-selected peptide of interest. PRM uses a setup with two
quadrupoles in connection with an Orbitrap or TOF mass analyzer allowing for a targeted
proteomics strategy that can monitor all transitions of a given targeted precursor peptide
simultaneously with high resolution and high mass accuracy. The operation is as described
above regarding MS/MS; the precursor ion is selected in Q1, the precursor ion is
fragmented in the collision cell Q2 and then all product ions are scanned with high
resolution and high accuracy in the Orbitrap. All transitions are in this way co-detected
and distinguished from one another as well as from the background, in the final mass
analysis stage (fig.15) [194, 195].
48
Fig. 15. The principles of PRM. The precursor ion is selected in Q1, the precursor ion is fragmented in the collision cell Q2 and then the Orbitrap scans all product ions with high resolution and high accuracy. Figure modified from [196]
PRM has quantitative analysis capabilities as well as an increased qualitative ability. It is
highly specific since all potential product ions of a peptide upon fragmentation are
available to confirm the identity of the peptide instead of just 3–5 transitions as in SRM.
Since PRM monitors all transitions, the prior knowledge of target transitions before
analysis is eliminated, however, precursor peptide selection before MS analysis is still
required, and the demands for proteotypicity and quantotypicity are just as strict for PRM
as SRM. PRM can thereby provide a higher tolerance for co-isolated background
peptides/species with the numerous ions available for identification and quantitation,
which makes the presence of interfering ions in a full mass spectrum less disruptive than
in a narrow mass range as in SRM. A PRM method is also easier to develop as much of
the effort required to determine optimal transitions and further optimize a traditional
SRM assay is eliminated [194, 197, 198].
4.2 Selected Reaction Monitoring (SRM)
SRM is also a targeted MS/MS technique for the detection and quantification of specific,
predetermined analytes. The SRM method has an increased sensitivity as data collection
49
is limited to a low number of pre-selected protein-derived tryptic peptides each having a
set of characteristic fragment ions ideal for antibody-free quantification of specific proteins
in complex biological samples, including tissue samples and serum [199-201]. SRM is most
effectively used in LC-MS systems, where the capillary LC column is connected to the ESI
source of the mass spectrometer [200]. SRM has a high sensitivity, a large dynamic range,
high specificity, and reproducibility and it can be multiplexed to increase the throughput
[199, 202-204].
In SRM, a mass spectrometer with a triple quadrupole (QQQ) mass analyzer is used to
filter out unrelated ions both at the tryptic peptide stage and at the fragment ion stage
according to preselected mass to change rations [205, 206]. The first quadrupole (Q1) is
set to only admit a predetermined m/z value of ionized peptides. The admitted peptides
are then fragmented in the second quadrupole (Q2) by CID and then only specific ions
will reach the third quadrupole (Q3) that also has a fixed m/z filter allowing the passage
of a fixed (low) number of ions with defined m/z ratios only. Each Q1 and Q3 pair is
termed a transition. For a given peptide, several transitions will occur, as a result of the
fragmentation of the peptide. This is outlined in fig. 16 where the general SRM method is
depicted at the top, and a peptide with possible fragmentations marked is shown above a
table of what is called an SRM assay. The SRM assay is the data fed into the MS,
controlling which ions are allowed through the mass filter. As can be seen in fig. 16, the
mass filter in Q3 changes, and only the associated fragment ion is detected.
The two levels of m/z filtering in SRM result in high selectivity and low background
signals and thereby an increased signal-to-noise ratio, as fewer ions will reach the detector.
The workflow of SRM-based proteomics is comprehensive, but the time and effort needed
to establish the SRM method (SRM assay) is well spent considering the high sensitivity
and selectivity, which are the hallmarks of SRM. Once the SRM assay is established, it
can be used indefinitely in any study with the same protein of interest [199].
50
Fig.16. Representation of the triple quadrupole mass spectrometer used in SRM. Ions are introduced into the first quadrupole mass analyzer (Q1) and the precursor ions are filtered from predetermined m/z. In the second quadruple (Q2) the ions are fragmented and then filtered once more in the third quadrupole (Q3) that also has a fixed m/z filter. The top part of the figure (SRM principle) is modified from [201], the bottom part of the figure is made for this Ph.D. thesis.
51
4.3 Selection of target peptides
As mentioned, the workflow of developing an SRM method is comprehensive with several
time-consuming steps as shown in fig. 17.
Fig. 17. The protein of interest is used as a base for finding the best proteotypic peptides, which are then validated and used in SRM to optimize the SRM method. When proteotypic peptides have been determined and the optimal transitions have been found, the SRM method can be used for quantitative sample analysis [199].
The first step of the SRM experiment workflow is the selection of the protein of interest.
Once the target protein is determined, the target peptides unique to the target protein are
selected. The choice of peptides is crucial as it determines the sensitivity of the SRM assay.
The peptides should be proteotypic peptides easily detectable by mass spectrometry [207,
208]. A proteotypic peptide is always observed for a specific peptide, is unique for this
peptide, and should fragment well in a mass spectrometer such that they can be confidently
identified [208-210].
52
A first step is therefore to determine the peptides resulting from e.g. tryptic digestion of
the protein in question. These peptides should then be evaluated for their ability as
proteotypic peptides. In principle, all possible peptides can be tested, but using previous
information from e.g. discovery MS would reduce the time spent on testing. Results from
MS experiments performed for several organisms have been deposited in online repositories
such as PeptideAtlas [211] and is a good place to start. In these databases, information of
peptides, which have been reproducibly detected and thereby are likely to be associated
with the most intense signals, can be found. For peptides not found in these, there are
several computational tools developed for evaluating proteotypic peptides [208, 211-215]
Proteotypic peptides suitable for quantification are termed quantotypic peptides. Not all
proteotypic peptides meets the higher standards for quantotypic peptides, but all
quantotypic peptides are proteotypic. The abundance of a quantotypic peptide must
correlate with the abundance of the parent protein [208-210]. Where proteotypic peptides
need “only” to be unique, continuously observed, and fragment well in a mass spectrometer,
more demands are set for quantotypic peptides, as have been compiled in table 3. As
issues affecting the quantitative accuracy cannot be completely avoided in praxis, it is
important to monitor multiple well-behaving peptides for each targeted protein [199, 200].
Table 3. Properties of quantotypic peptides and corresponding rationale
Property Rationale
Unique amino acid sequence Ensures that peptide is diagnostic for a single protein
Avoid short hydrophilic and long
hydrophobic peptides
m/z value of the peptides should be within mass-range of
the instrument. Peptides of 8–25 amino acids are preferred
Amino acid modifications Can alter the cleavage pattern of peptides. Avoid peptides
susceptible to chemical modification
No methionine or tryptophan Side chains of these amino acids are prone to oxidation
No glutamine or asparagine Side chains of these amino acids are prone to deamidation
53
4.4 SRM transitions
In Q1 and Q3 a selection of specific m/z settings is made. The combination of these two
m/z settings is referred to as a transition (fig.15). The m/z value in Q1 is determined by
the mass and the predominant charge state of a peptide and the m/z of the fragment ion
of the peptide for Q3 is easily calculated. The intensities of the individual fragments may
differ substantially, so to obtain high sensitivity it is essential to select the transitions
characterized by the most intense fragment [199].
For sensitive detection, it is essential to target the predominant precursor charge state,
which can be deduced through experiments on other ESI instruments, but one should be
aware of possible differences in instruments, flow rate, solvent and background as these
No glutamate or N-terminal
glutamine
Glutamate and N-terminal glutamine residues are quickly
transformed to pyro-glutamate under acidic conditions.
No N- or C-terminal peptides Sequences in the native proteins are susceptible to
degradation
No sites of known or putative post-
translational modification
Post-translational modification may split the analyte signal
into modified and unmodified variants, which compromise
quantification.
No dibasic context Dibasic motifs tend to be cleaved at one or the other of the
basic residues. Signal from the analyte is then split between
the two variants, but the standard maps to only one of them
Not susceptible to missed cleavage Prevents stoichiometric release of the peptide from the
native protein and protein-level standard.
Detectable in a LC–MS experiment The shape of chromatographic peak of a peptide should be
symmetrical with narrow width. The peptide should ionize
efficiently and provide a stable and intense signal.
The information compiled in this table is found in the following references [199, 200, 208-210, 212-215].
54
can change the charge state distribution [199, 216]. Ion source parameters can also impact
ionization as the process of dissolvement and dissociation of ion clusters are supported by
the declustering potential (DP). The DP is a voltage applied to the orifice (i.e. the opening
where ions enter the mass spectrometer) that helps to prevent the ions from clustering
together, and this can slightly differ from instrument to instrument, depending on the
manufacturer of the ionization device. The individual transitions of a peptide will have
the same DP optimum when they are derived from the same precursor charge state,
making it important to pay attention to instrument-specific impacts when developing an
SRM method [199]. Regarding fragmentation conditions, in a linear collision cell singly
charged y-ions are the predominant type of fragments generated by CID. Only small b-
ions are usually observed here, and it is important to avoid fragments with m/z values
close to the precursor as these transitions usually are “noisy” i.e. they have low intensity
and add to the background noise of the instrument. Instead, fragments with m/z values
above the precursor should be chosen, as they have the highest selectivity since the singly-
charged chemical background does not result in fragments with higher m/z than the
precursor [199].
The collision energy is also important to take into account. As the collision energy
increases, more precursor ions are fragmented and the ion intensity will increase until it is
overcompensated by losses due to secondary fragmentation events. The optimal collision
energy can be predicted as it is an approximately linear correlation with the precursor
mass of a given charge state, but some peptides or fragments may deviate from this
predicted optimal collision energy, which may need to be taken into account. Ionization
and fragmentation conditions can be optimized by testing the possible transitions in direct
infusion mode and plot the parameters or through add-ons for acquisition software [199].
55
Fig. 18. Validation of transitions for peptide VFAQFSSFVDSVIAK. (A) Peak 1 and peak 2 with their co-eluting transitions are seen at 37.5 and 43.3 min, respectively. They have the same precursor mass thereby complicating correct peak identification. The five different transitions are traced in different colors. (B) MS/MS spectra from SRM of peak 1 (upper panel) and 2 (lower panel). Y ion matches are colored in red. At 43.3 min the SRM transition intensities are higher but the MS/MS spectra show that the targeted peptide is eluting at 37.5 min [199].
Even though SRM analysis has high specificity, a particular transition may not be specific
for a targeted peptide in a complex sample. As seen in fig.18 there may be unspecific
signals from other peptides with closely related sequences resulting in precursor/fragment
ion pairs of similar masses. Unspecific signals might be above the detection limit and more
intense than the signal for the targeted peptide, so these signals can be mistakenly
identified as derived from the targeted peptide and lead to mis-quantifications. Validation
is therefore important [199, 200, 214]. One way of validating the transitions is a so-called
SRM-triggered MS/MS scanning where the QQQ instrument used, obtain a full fragment
ion spectrum whenever a signal for a particular transition is detected, and them comparing
the MS/MS spectra with the predicted peptide fragments. In this way, the major MS/MS
peaks are matched and the detected SRM signals are confirmed as being derived from the
targeted peptide (Fig. 18).
An alternative to acquiring MS/MS spectra for validation is to use heavy isotope-labeled
peptides spiked into the sample prior to analysis. Using heavy labeled peptides with
incorporated 15N and 13C labeled amino acids that match the sequence of the targeted
56
peptide will co-elute with the non-labeled target peptides and allow for distinguishing
between the detection of low-abundant peptides from unspecific signals. The signal
intensity ratios of the transitions are identical for the labeled and unlabeled peptides, but
stable isotope-labeled peptides are expensive and may not be a possibility in every study
design [199, 200].
When running SRM, the instrument repeatedly cycles through the specified list of
transitions spending a defined time on each transition, called the dwell time. The number
of transitions that are measured per cycle and the dwell time are both mutually dependent
at a fixed cycle time since targeting ten peptides with five transitions, 20ms dwell time
will add up to a cycle time of 1s (10*5*20ms), meaning that every second an intensity
value is recorded for each of the transitions. To achieve high sensitivity, the dwell time
therefore has to be long enough to accumulate sufficient signal. Cycle time can also be
adjusted to analyze a higher number of transitions at a fixed dwell time, but if not enough
data points are being recorded over the chromatographic elution time of a peptide, the
accuracy will decrease leading to peak shape deterioration, where the peak might be too
broad or split (fig. 19) [199, 200]
Fig. 19. Shown are the resulting peak shape using varying dwell and cycle time. Using cycle times of 10 or 20s peak height cannot be estimated correctly despite the excellent accuracy of the individual data points at 500ms dwell time. Reducing the dwell time from 50 to 5ms decreases accuracy, but at 50ms dwell time and 2s cycle time, a near-perfect peak is produced. Changes in dwell time do not affect absolute signal intensity as it is normalized as ‘counts per second’ [199].
57
At least eight data points should be obtained across the elution peak to ensure precise LC-
MS quantification, and a compromise between dwell time and cycle time is therefore
necessary. This is achieved by mainly two strategies;
- Dwell times for each transition are specifically adjusted to the expected
concentration of targeted peptides. In this way, more time is spent on low
abundance peptides and shorter dwell times are chosen for high abundance peptides.
This necessitates that the concentration of a peptide is known beforehand.
- Scheduling transitions so that the acquisition of particular transitions is restricted
to a window around the elution (retention) time of the corresponding peptide and
thereby using the full cycle time to detect and quantify the expected eluting
peptides in a given time window.
The detection limit will however always depend on the instrument used. A detection limit
of around 10-50 amol should be expected for a well-ionizing peptide with state-of-the-art
QQQ instruments in combination with nano-LC [199, 200].
If many proteins are targeted, numerous transitions in a single experiment are required
which reduces the dwell time of the individual transition, the total number of targeted
peptides that can be quantified is therefore limited. As an example, if every protein should
be quantified by two peptides with two transitions each, 50 proteins can be quantified in
total. Fig. 20 show how the number of transitions pr. peptide correlates with dwell-time
settings. The best 2-4 transitions per peptide is therefore usually selected for the SRM
assay, based on data from previously performed experiments like discovery proteomics or
experimentally determined on a QQQ instrument. It may also be based on calculated
fragment ion masses that are then experimentally tested on a QQQ instrument which
yields the most reliable selection, but can be very time-consuming depending on the
number of peptides selected [199, 200]. Performing scheduled SRM would surpass this, as
transitions for a particular peptide are only acquired in the time window around the
expected rt. This increases the number of detected peptides (and associated proteins) that
58
can be quantified in a single LC-MS experiment. Using scheduled SRM <1000 transitions
can be quantified with high sensitivity and reproducibility.
Fig. 20. Depending on the cycle and dwell time, the number of transitions will dictate how many peptides can be quantified [199].
4.5 Quantification by SRM
SRM can be used to quantify targeted proteins using the transitions for specific peptides,
validated, and optimized as described above. In some cases it will be sufficient to determine
the relative changes of protein amounts based on the absolute signal intensity of the
individual samples. However, performing precise label-free quantification can be a
challenge as signal intensities vary from one LC-MS analysis to another and within the
LC-MS analysis itself. There may be fluctuations in ionization efficiency, as the intensities
recorded for a peptide will depend on the peptide sequence, the background, the sample,
and several experimental factors that cannot be precisely controlled. Normalization, with
the assumption that the majority of protein/peptide concentrations in the sample are not
changing, can only correct for global shifts and do not compensate for local suppression or
enhancement effects on individual peptides [199, 200, 214, 217].
Label-free quantification can be successfully performed under well-controlled settings, but
adding a stable isotope-labeled internal standard can efficiently overcome the problems of
fluctuations in signal intensity. In this case, the relative quantification is based on the
relative intensities of the sample analytes compared to the internally labeled standard.
Absolute quantification is in this way possible with SRM.
59
By using individual isotopically labeled reference peptides or proteins spiked into the
samples, it is possible to calculate an accurate ng/ml concentration of a protein in e.g.
blood from the signal intensity of the light/heavy transitions as the amount of these
peptides/proteins is precisely determined [199, 217-220]. Selection and validation of the
transitions is performed as described above, and the SRM assay is then supplemented with
the corresponding light or heavy transitions. In this way, a pair of light/heavy transitions
is produced for every precursor/fragment combination, and each peptide is then quantified
relatively to its matching heavy-labeled peptide. To minimize technical variability, the
isotopic label should be introduced as early in the sample preparation as possible [199,
221].
The sample amount is often based on total protein mass that is determined by an assay
e.g. Lowry, Bradford or BCA, but these assays are often not sufficiently accurate, so an
alternative can be to normalize the samples on basis of stable proteins, that are expected
to be constant throughout the experiment. This could be so-called “housekeeping” proteins
that are more abundant and constantly present at similar levels in all samples. The
proteins are included in the SRM assay and quantified together with the target protein(s),
and normalization is then performed after acquisition during the data analysis [199, 222-
224]. The stable isotope-labeled peptides can be ordered by several suppliers and are often
referred to as AQUA peptides, but they are pricey which may be a problem if a large
number of peptides are required [199].
4.6 Tandem Mass Tag (TMT) Mass Spectrometry
Using tandem mass tags (TMT) is a non-targeted MS/MS labeling strategy for relative
quantification. It uses isobaric tags typically composed of a mass reporter, a mass
normalizer, and an amine reactive group (fig.21) [225]. The mass reporter and mass
normalizer are used to incorporate stable isotopes in multiple configurations allowing the
60
mass reporter’s mass to be resolved in a MS/MS spectrum, while the intact mass of each
tag variant is the same [225]. These tags are designed so that on analysis by CID, the
TMT fragment is released to give rise to an ion with a specific m/z ratio and the intensities
recorded in the resulting MS/MS spectrum allow for relative peptide quantification [225,
226]. Using TMT up to ten samples can be compared simultaneously, as the proteins are
labeled with separate tags, pools, co-eluted, and then analyzed by LC-MS/MS. This
multiplexing ability allows for an experimental design very suitable for biomarker research
[227]. The isobaric and chemically identical tags allow identical peptides labeled with
different tags to have the same chromatographic elution profile, greatly simplifying the
interpretation of peaks. This leads to a more accurate quantification and improves the
signal-to-noise ratio of quantitative measurements by enabling quantification at the
MS/MS level [225]. The co-migration of TMT-labeled species results in an MS signal for
each peptide pair which is not split into two peaks, as in conventional isotope labeling,
thereby improving sensitivity in the MS mode [226].
Fig. 21 Structure of TMT reagents. A: A TMT reagent contains a mass reporter, a mass normalizer, and an amine reactive group. Between the mass reporter and mass normalizer, a HCD cleavable linker is placed. B: Variants of TMT tags. Heavy labeled 12C and 13N are indicated by asterisks. 10 total variants are possible. Figure modified from [225].
61
4.7 Data-independent acquisition and SWATH-MS
Data-dependent acquisition (DDA) methods, while having high sensitivity and being
powerful methods for detecting peptides representing proteins of interest with high
quantitative accuracy, these approaches are not suitable for discovery-based applications
[228]. Data-independent (DIA) acquisition methods that allow for accurate peptide
quantification without being limited to profiling predefined peptides of interest, have also
been developed. DIA does not require any kind of labeling, and all information from a
sample is recorded in contrast to DDA MS/MS analysis where only a limited number of
precursor ions are selected according to the intensities of precursor ions [228]. Data-
independent acquisition (DIA) performs predefined MS/MS fragmentation and collect data
regardless of the sample content, which results in a more sensitive and accurate protein
quantification compared to DDA. SRM and PRM are types of targeted DIA, and SWATH-
MS is an untargeted DIA method [229]. In SRM a single fragment is monitored with high
sensitivity and in PRM all fragments are recorded with high mass resolution. In both
cases, all intact precursor ions are measured with a wide precursor window (between 2 and
200 m/z) for fragmentation that usually contains several peptide precursors. All obtained
fragment ions are measured from this precursor isolation window, before moving to the
next precursor isolation window and measuring MS/MS fragmentation here, continuing
until the entire MS1 mass range is covered [229].
In SWATH-MS the derived data depends on the defined m/z window, and for each cycle
the instrument focuses on this narrow mass window and acquires MS/MS data from all
precursors detected here. The mass window is changed so it ultimately reaches across the
entire mass range, systematically collecting MS/MS data from every mass and all detected
precursors (fig.22) [228, 230, 231].
62
Fig. 22 SWATH-MS measurements are performed using a quadrupole as the first mass analyzer and a TOF or Orbitrap as the second mass analyzer. A single precursor ion (MS1) spectrum is recorded, followed by a series of fragment ion (MS2) spectra with wide precursor isolation windows. Data is recorded through repeated cycling of consecutive precursor isolation windows with a defined mass range and includes continuous information of all detectable fragment and precursor ions. Extracted ion chromatograms (XIC) can then be generated on both MS1 and MS2 level. Figure modified from [230].
In the excellent review by Ludwig et al. (2018) an overview of the pro and cons for
SWATH-MS, DDA and targeted DIA (SRM and PRM) have been composed, and is seen
here in fig.23 [230].
As can be seen, each method excels in several categories and lacks in others, underlining
the importance of understanding which type of experiment and what type of output is
wanted from sample analysis.
63
Fig. 23. Performance benefits and limitations of DDA, SWATH-MS, SRM, and PRM as presented in [230]. *Least optimal performance, **Medium performance, ***Best performance.
64
Chapter 5
Materials and Methods 5.1 Biological samples
Pigs were chosen as it is a very suitable animal model for studying the innate immune
response. The porcine model allows for highly controlled challenges to be performed, and
disease development to be monitored closely in an organism that closely mirrors humans
regarding many aspects including antiviral immune response. All animal procedures were
conducted according to protocols approved by the Danish Animal Experiments
Inspectorate. Samples from [29] have License Numbers 2001/561-350 and 2008/56121465
for the Actinobacillus Pleuropneumoniae (Ap) and Staphylococcus aureus (Sa) infection
studies, respectively. Samples from [30] have License No. 2013/15-2934-00946.
For the 2DE and MALDI-TOF experiments and first parts of SRM development, two
porcine serum pools were made with additional samples available from previous infection
studies performed at DTU Vet, both from pigs undergoing an acute phase response, were
used.
65
5.1.1 Staphylococcus aureus (Sa) pig infection models From two separate infection studies, samples were obtained. The study performed by
Olsen et al. (2013) [29] studies pigs intravenously infected with a porcine strain of the
Gram-positive pathogen Staphylococcus aureus (Sa) and the study by Jensen et al. (2016)
[30] used a model based on the insertion of a small steel implant into the right tibial bone
with either Sa or saline. Based on inoculum, pigs were allocated into three groups which
were high-inoculum dosed (104 colony-forming unit (CFU); Group A), low-inoculum dose
(102 or 103 CFU; Group B), and controls (Saline; Group C). Samples analyzed in this
Ph.D. were from Group A and Group C.
5.1.1.1 Sa intravenous infection model
Details of experimental design and procedures can be found in Jensen et al. [232] and
Leifsson et al. [233]. Briefly, three clinically healthy 8-week-old female SPF Danish
landrace/Yorkshire crossbred pigs with a bodyweight (BW) of 20–25 kg were inoculated
intravenously with a bolus of 108 CFU/kg BW of a porcine pathogenic strain of Sa. After
12h, the pigs received a second bolus equal to the first. The pigs were euthanized after
24h, and the postmortem examination was immediately done. From the same herd three
control pigs were sham-infected intravenously with sterile saline and managed in parallel
with the infected pigs. Blood samples were taken at 0h (before inoculation) and 24h post-
infection (PI) at euthanasia, wherein SAA protein concentration was measured.
66
5.1.1.2 Implant-Associated Osteomyelitis (IAO) Sa infection model
Details of experimental design and procedures are found in [30]. Briefly, 42 female Danish
Landrace pigs obtained from specific pathogen-free herds at the age of 3 months (30 kg
BW) or 8 months (67-77 kg BW) were included. Under anesthesia, an implant cavity was
drilled in the tibial bone followed by insertion of a small implant made of stainless steel,
2x15mm (pigs of 30 kg BW) or 2x20mm (pigs of 80 kg BW), and simultaneous inoculation
of Sa bacteria (n=32) or saline (n=10) into the implant cavity. The inoculum of Sa was
prepared as previously described [234] and diluted with sterile 0.9% isotonic saline to
obtain inoculation doses/volumes of 104 CFU in 10 mL. All animals were euthanized 5
days after the insertion of implants. Blood samples were taken on the days of surgery and
euthanasia, and evaluation of the systemic acute phase response by measurement CRP
and SAA respectively, by enzyme-linked immunosorbent assay [162, 235].
5.1.2 Actinobacillus pleuropneumoniae (Ap) pig infection models
Experimental design and procedures have been described in detail by Skovgaard et al. [99,
155]. Briefly, three clinically healthy 8–10 weeks old castrates of Danish
landrace/Yorkshire/Duroc crossbreeds from a specific pathogen-free herd were inoculated
intranasally with Ap pleuropneumoniae serotype 5b, isolate L20, a highly pathogenic
porcine bacteria causing severe acute necrotizing pleuropneumonia as described previously
by Skovgaard et al. [236]. Cultivation of bacteria was performed as previously described
[237]. Inoculation was performed by dripping bacterial solution in each nostril during
inhalation (1 ml McFarland suspension mixed 1:1 with Brain Heart infusion Broth + 0.5%
NAD, containing approximately 0.7 107 CFU/ml) [236]. Three non-inoculated clinically
healthy pigs from the same herd served as control pigs. The pigs were euthanized 14–18h
after inoculation and necropsied immediately. SAA protein concentration was measured
67
in serum sampled immediately prior to inoculation as well as at euthanasia for verification
of initiated acute phase response [99].
5.2 SAA ELISA
SAA protein concentration was measured in blood, sampled using a commercially available
sandwich ELISA (Phase SAA assay, Tridelta Development Ltd.). The assay is based on
antihuman monoclonal antibodies in a sandwich set-up, as originally described by
McDonald et al. [162]. Samples were tested according to the manufacturer’s instructions
and all samples including standards were determined in duplicate. Sample values were
calculated from the curve fitted to the standard.
5.3 Triton X-114 cloud point extraction (CPE)
Two separate pools of porcine serum samples were prepped for 2D gel electrophoresis and
MALDI TOF analyzed using Triton X-114 cloud point extraction [238]. Two separate
set-ups were used (summarized in table 4). Firstly, duplicates of the porcine serum sample
pools were either treated with the protocol as described below or treated with the same
protocol, but where all amounts had been cut in half. In total 8 2D gels were run in this
setup. The second setup was a dilution experiment where the porcine serum sample pools
were diluted prior to TX-114 CPE, in total 8 2D gels were run in this setup.
Table 4. Summation of the two separate 2DE setups
Setup 1
Gel 1 & 2 Duplicates of porcine serum pool 1 Full amounts during TX-114 CPE
68
Gel 3 & 4 Duplicates of porcine serum pool 2 Full amounts during TX-114 CPE
Gel 5 & 6 Duplicates of porcine serum pool 1 Half amounts during TX-114 CPE
Gel 7 & 8 Duplicates of porcine serum pool 2 Half amounts during TX-114 CPE
Setup 2
Gel 1 Pool 1 Undiluted prior to TX-114 CPE
Gel 2 Pool 1 2 times diluted prior to TX-114 CPE
Gel 3 Pool 1 3 times diluted prior to TX-114 CPE
Gel 4 Pool 1 4 times diluted prior to TX-114 CPE
Gel 5 Pool 2 Undiluted prior to TX-114 CPE
Gel 6 Pool 2 2 times diluted prior to TX-114 CPE
Gel 7 Pool 2 3 times diluted prior to TX-114 CPE
Gel 8 Pool 2 4 times diluted prior to TX-114 CPE
400 µl sample was diluted with 1350 µl cold (0-4°C) Tris-buffer (50 mM Tris (pH 7.4), 150
mM NaCl) in a conical tube and 750 µl cold (0-4°C) 10 % (w/v) TX-114 was added. The
tube was turned several times for gentle mixing and put on ice for 30 min. The sample
was then carefully overlaid on 2 ml of Sucrose-TX-114 (6% (w/v) sucrose and 0.06% (w/v)
TX-114 in buffer A) in a conical tube, incubated 5 min in a water bath at 37°C for 5 min.
and then centrifuged at room temperature (RT) in a swing-out rotor; 5 min, 5000 g. After
centrifugation the supernatant constituting the aqueous phase of the sample and the
sucrose-containing layer were removed carefully, leaving the detergent phase at the bottom
of the tube. The tube was put on ice and the detergent phase (sample) was re-dissolved
in 2250 µl Tris-buffer. When the solution has cleared up, it was overlaid on 2 ml of Sucrose-
TX-114, incubated at 37°C for 5 min, and centrifuged (5000g) for 5 min at RT. The
supernatant was discarded, and the detergent phase was re-dissolved in 1000 µl cold (0-
4°C) Tris-buffer and stored at -20°C.
69
5.4 TCA Precipitation
The cloud point extracted (CPE) sample was subsequently treated with a Trichloroacetic
Acid (TCA) Precipitation by transferring it to an Eppendorf tube, and 100 µl 65% (w/v)
TCA was added to a final concentration of 15%. The sample was mixed, incubated on ice
for 30 min and centrifuged (15.000g) for 15 min at 4°C. Supernatant was carefully removed
and the precipitate was washed twice with acetone (one wash consisted of the addition of
1 ml acetone (90%), mixing, centrifugation (15.000g) for 5 min at 4°C, and removal of
supernatant). Washing was repeated and following supernatant removal, the pellet was
left to dry at room temperature for about 30 min.
5.5 2D gel electrophoresis
Immobiline Drystrips (GE Healthcare, 18cm, pI 3-10) were used for the first-dimensional
protein separation. The air-dried TCA precipitated sample was solubilized in 450µl 1D
buffer (8M urea, 2M thiourea, 50mM DTT, 1.5% Chaps, 2% Pharmalyt 3–10, and 10 mM
Tris-HCl (pH 8.3)) with Orange G as a dye. All 350 µl was loaded on an Immobiline
Drystrip by reswelling overnight and the gel was run approx. 18h, with 4 phases (0.01h at
100V 1mA, 5h at 100V 1mA, 5h at 3500V 1mA, and then staying at 3500V 1mA until
stopped (all steps were at 5W) and stored at -80°C until the second dimensional SDS-
PAGE.
Immediately after thawing, the Immobiline Drystrip was incubated for 20 min in 10ml
equilibration buffer (6M urea, 50mM Tris-HCl (pH 8.8), 30% (w/v) glycerol, and 2% SDS)
with 1% (w/v) DTT; followed by alkylation for 20min in 10ml equilibration buffer with
4.5% iodoacetamide. The Immobiline Drystrip was placed and sealed by 0.5% (w/v)
agarose, 25mM Trizma base, 192mM glycine, and 0.1% SDS on top of an acrylamide gel
70
(T = 12%), run on a Hoefer DALT system (Amersham Biosciences) with 90V, 400mA for
17,5h followed by 17h at 40V, 40mA, at 15°C. After fixation for 60min in 40% (v/v)
ethanol and 10% (v/v) acetic acid, the gel was stained by Colloidal Coomassie Brilliant
Blue [ref 22]. The gels were digitized using a charge-coupled device camera (Camilla,
Raytest, Straubenhardt, Germany). Progenesis SameSpots software (version 4.5,
Nonlinear Dynamics, Newcastle, UK) was used for image analysis with the built-in
algorithms of Progenesis SameSpots for alignment, matching, and quantification of the
spots, as well as for normalization of spot volumes based on a spot volume abundance
ratio between gels.
5.6 MALDI-TOF
From the 2DE gels, spots of interest were excised, reduced, and alkylated (using DTT and
iodoacetamide, respectively) and digested overnight with trypsin. Digested samples were
put with the matrix (4-HCCA) on a MALDI plate and analyzed using an Ultraflex II
TOF/TOF mass spectrometer (Bruker-Daltonics, Bremen, Germany). Ionization was
performed in positive mode and mass spectra were analyzed by FlexAnalysis. The obtained
mass spectra were then used as input to MASCOT MS/MS Ion searches of the porcine
subset of the NCBI database. These searches were performed assuming the formation of
single-charged peptides, carbamidomethylation of cysteine residues, possible oxidation of
methionine residues, and up to 1 missed cleavage.
5.7 SRM method development
The details regarding SRM development – sample preparation, SRM assay construction,
MS analysis, and data analysis can be found in Paper I.
71
Chapter 6
Paper I
Targeted mass spectrometry for Serum Amyloid A (SAA) isoform
profiling in sequential blood samples from experimentally
Staphylococcus aureus infected pigs
Anna Barslund Leuchsenring, Christofer Karlsson, Louise Bundgaard, Johan Malmström,
Peter M. H. Heegaard.
72
6.1 Summary of Paper I
In Paper I, an SRM method was developed and applied to sequential serum samples from
an infection experiment with Sa. It was possible to detect all three porcine SAA isoforms.
The porcine SAA isoform profiles correlated with the previously established total SAA
response [29] and a close correlation between the SAA isoform hepatic mRNA abundance,
and SAA isoform profiles in the circulation was also found. This proved that antibody
independent MS-based quantification is possible at high specificity and sensitivity,
allowing discrimination between a set of very similar circulating SAA variants directly in
serum samples. In addition to this, the results unequivocally establish that SAA2 is the
main circulating porcine SAA isoform, being induced around 10 times during the porcine
APR to Sa infection. A discrepancy was observed between the total SAA serum
concentrations detected in Paper I and those established previously by immunoassay [29],
and the peptides used for SRM did not correlate equally well for every SAA isoform. The
underlying biological reason for this is interesting to investigate. The SRM approach can
be generalized to other circulating biomarkers and MS methods have a great potential for
high-number multiplexing.
Paper I was submitted to Journal of Proteomics on April 24th 2020.
73
6.2 Paper 1 as submitted to Journal of Proteomics
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
Chapter 7
Results and discussion
As described in chapters 1 and 2, SAA is an APP highly upregulated during an APR, and
an interesting biomarker candidate for early diagnosis of infection and diseases with an
inflammatory component such as neurodegenerative diseases, autoimmune diseases,
obesity and obesity-related diseases [15, 62-64, 165].
SAA isoform genes have been found differentially regulated depending on infection and
tissue; a unique feature of SAA not found in other APPs to our knowledge, making SAA
a very interesting biomarker candidate [29, 44, 98, 99, 132, 154]. Investigating this
potential, however, has been difficult, due to methods for SAA isoform detection are
limited as SAA antibodies are incapable of discriminating between the highly similar SAA
isoforms. In addition to detecting the SAA isoforms, quantifying them would provide
important information regarding possible specific biological functions of SAA isoforms and
allowing us to investigate the possible role of SAA isoforms as differential biomarkers of
infection. This Ph.D. project therefore developed a specific, quantitative, and antibody-
free targeted mass spectrometry-based method for detection and quantification of
circulating porcine SAA isoforms in serum samples to provide the rapid and reproducible
method with high sensitivity and specificity, we are currently missing.
104
The developed SRM MS method was used to analyze the porcine serum samples described
in chapter 5 and resulted in Paper I, presented in chapter 6. Additional experiments were
performed in this Ph.D. project in the process of analyzing the porcine samples and
developing the SRM method and the results not shown in Paper I will be presented in this
chapter.
7.1 Two-dimensional gel electrophoresis and MALDI-TOF
In order to examine the presence of native SAA isoforms in porcine serum samples proteins
were obtained by TX-114 cloud point extraction (CPE) of porcine serum sample pools and
analyzed by 2DE as described in chapters 5.3-5.5. In total 16 separate 14% 2D gels were
run within the pI range 3–10. From the first setup, it was concluded that full amounts of
TX-114 CPE provided best results, and from the second setup it was concluded that
samples could easily be 2-3 times diluted while still providing clear spots for identification
(data not shown). After completed 2DE the gels were scanned, and spots of interest were
marked. In fig. 24, scans of a gel with spots from pool 1 and pool 2 respectively, from the
second experimental setup is shown. From pool 1, spots 1-3 were of interest and from pool
2 spots 4-12 were of interest as they were within the Mw range expected for SAA isoforms
(9-14 kDa).
105
Fig. 24. Proteins obtained by CPE of 400µl porcine serum pools (left: pool1, right pool2). The two 2D gels are representative of 16 gels, pretreated with TX-114, and run on 14 % gels with a pI range of 3-10. The numbered spots were identified and identities are given in table 5.
The spots of interest were numbered, excised, and subsequently identified using MALDI-
TOF as described in chapter 5.6. In total 12 spots were analyzed, resulting identifications
are summarized in table 5, and as can be seen, there was a positive identification of SAA2
and SAA4 in several spots, but no identification of SAA3. MALDI-TOF MS and MS/MS
analysis resulted in mass spectra as shown in fig. 25 and a mass list. The software
FlexAnalysis and BioTools (Bruker) were used for peak picking and a database search was
then performed for the identification of the protein. Fig. 24 shows a representative MALDI-
TOF spectrum from analysis using FlexAnalysis, the remaining spectra can be found in
the supplementary chapter.
106
Fig. 25 MALDI-TOF mass spectrum. Following 2DE, gel spots were excised and prepped for MALDI-TOF analysis. The analysis resulted in mass spectra such as this, and a mass list which was used to perform a database search for the identification of the protein.
Table 5. 2D gel spot identifications
Sample Spot number Identification
Pool1 1 Inconclusive
Pool1 2 SAA2
Pool1 3 SAA4
Pool2 4 SAA2
Pool2 5 SAA2
Pool2 6 SAA4
Pool2 7 Hemoglobin α
Pool2 8-12 Inconclusive
The porcine serum samples from pigs with ongoing APR were analyzed by Triton X-114
CPE followed by 2DE. The TX-114-based CPE on serum samples is used to enrich
lipoproteins such as SAA, before MALDI-TOF as the SAA isoforms can be hard to
107
visualize in unfractionated plasma when using 2DGE with no pretreatment [238]. These
procedures are tedious and comprehensive and ultimately resulted in the identification of
the presence of SAA2 and SAA4 in the samples, with no detection of SAA3. The presence
of an APR had been confirmed previously for the samples used so the absence of SAA3
may be due to misidentification due to absence or insufficient reference MS spectra in a
database, or it could propose that the amount of SAA3 present in the samples were simply
too low to be detected.
Serum samples are very complex in terms of peptide and protein composition, and the 10
most abundant proteins constitute almost 90% of the serum proteome by mass and may
mask the more low abundant proteins, making pretreatment of the sample a prerequisite
in order to reduce their complexity [239, 240]. Hemoglobin α was also identified, and with
a Mw of 15,258 kDa, it makes sense to find this subunit of Hemoglobin in serum samples.
SAA3 may be masked by background noise or by higher levels of SAA2, as their molecular
weights are very close to each other (and with 86% sequence similarity). From our results
using SRM, we found that the absolute concentration of SAA3 was much lower than that
of SAA2, supporting this theory.
MALDI-TOF offers a fast analysis with high resolution and a simple setup for sample
preparation, but many aspects impact the MALDI experiments which therefore require
highly standardized sample protocols and pretreatment protocols to ensure good
reproducibility. Furthermore, reproducibility is also influenced by the matrix and
variations in the crystallization process. The crystallization process depends on aspects
such as the contamination of the sample, ratio of matrix to analyze, and atmospheric
conditions such as humidity, temperature, etc. [240]. Matrix preparation also influences
background intensity, resolution, signal strength, and detectability, and the resulting
distribution of the analyte on the sample spot may be uneven and differ from experiment
to experiment [240].
In the collection of MALDI spectra generated from the same sample a substantial
variability in the noise level, baseline and peak intensities will therefore occur and the
108
uneven sample distribution results in so-called ‘hot-spots’ on the target surface where the
ratio of analyte to matrix is optimal. In these ‘hot-spots’ analyte signal is high in relation
to other locations, thereby impacting reproducibility, making it important to find the
optimal sample preparation method and to perform standard calibrations in order to
obtain the highest possible accuracy [197, 240, 241]. This is very time-demanding.
Quantitative analysis can be performed with MALDI-TOF MS, but it is time-consuming
and challenging due to the variations in signal intensity. It requires understanding and
consideration of the sample and instrument conditions as most of the critical problems
stem from sample preparation (e.g. choice of matrix compound, concentration, solvents
and crystallization conditions), ionization chemistry, electric fields, and detector properties
– parameters that are interrelated and adjusting one parameter, changes the optimal
conditions of others. It is therefore also important to acquire and average many single-
shot spectra from several positions within a given sample spot to get representative sample
data [197].
7.2 Relative SAA concentrations in porcine serum samples
SRM was instead chosen as a suitable method, as it is a highly sensitive and selective
method for the targeted approach of peptide detection and the corresponding proteins in
complex biological samples. SRM also allows for absolute quantification when using spiked-
in stable isotope-labeled (‘heavy’) peptides. SRM method development is also somewhat
time-consuming, but the resulting method can be used on porcine samples from endless
experiments once it is developed, and it is easily adapted to e.g. human samples.
As described in Paper I, the SRM assay was first tested on pooled porcine serum samples
with synthetic peptides spiked in, and once it was confirmed that all spiked-in peptides
could be identified by the method, it was used on sequential porcine serum samples from
infections experiments previously performed [29, 30]. The relative SAA concentration in
109
serum samples from pigs infected with either Sa (intravenous infection model and Implant-
Associated Osteomyelitis (IAO) model, respectively) or Ap (intranasal infection model),
was then determined using the SRM method (fig. 26).
In the study by Jensen et al. (2016) [30] a model IAO with low inoculum infection of Sa
was evaluated. The samples showed clear clinical signs of infections with significant results
on all tested parameters and a clear correlation between the number of injected bacteria
and development of IAO was observed [30]. SAA levels were measured by ELISA and used
to evaluate the presence of an APR, and a significant increase in the SAA levels in infected
pigs compared to control animals was found [30].
By the developed SRM method all SAA isoforms (SAA2, 3, and 4) were detected in the
pig IAO samples. Mean relative concentrations were similar when comparing infected pigs
and control pigs, and higher levels of SAA were measured in infected pigs than in control
pigs. While not significantly different, the relative SAA levels did correlate with those
measured by ELISA in Jensen et al. (2016) [30] as the SAA levels were higher in infected
pigs than control pigs.
110
0.00
0.05
0.10
0.15
SAA2
Rela
tive
SAA
leve
ls, P
I
Control pigsIAO pigs
0.00
0.05
0.10
0.15
SAA3
Rela
tive
SAA
leve
ls, P
I
Control pigsIAO pigs
0.00
0.05
0.10
0.15
SAA4
Rela
tive
SAA
leve
ls, P
I
Control pigsIAO pigs
Relative SAA isoform levels in Sa infected and control pigs, IAO model
Infected Controls0.00
0.02
0.04
0.06
0.08
0.10
SAA2
Rela
tive
SAA
leve
ls
PI0h
Infected Controls0.00
0.02
0.04
0.06
0.08
0.10
SAA3
Rela
tive
SAA
leve
ls
Infected Controls0.00
0.02
0.04
0.06
0.08
0.10
SAA4
Rela
tive
SAA
leve
ls
Relative SAA isoform levels in Ap infected and control pigs
Infected Controls0.00
0.02
0.04
0.06
0.08
0.10
SAA2
Rela
tive
SAA
leve
ls
0h 24h
Infected Controls0.00
0.02
0.04
0.06
0.08
0.10
SAA3
Rela
tive
SAA
leve
ls
Infected Controls0.00
0.02
0.04
0.06
0.08
0.10
SAA4
Rela
tive
SAA
leve
ls
Relative SAA isoform levels in Sa infected and control pigs
Fig. 26. Mean relative serum levels of SAA isoforms 2, 3, and 4, respectively, in three sets of porcine serum samples before and 24 hours after infection / sterile saline injection (controls). Columns indicate mean values. In addition, individual concentrations are also shown for infected and control pigs for each SAA isoform. Top: Results from serum samples from the Sa infection experiment (IAO model), note that all results shown here are PI, as no 0h samples were made. Middle: Results from serum samples from the Ap infection experiment (Intranasal infection model). Bottom: Results from serum samples from the Sa infection experiment (Intravenous infection).
111
When analyzing Ap and Sa samples from Olsen et al. (2013) [3], the relative SAA levels
measured by SRM confirmed the presence of an APR, as the SAA isoform levels were
higher after 24h of infection. For the samples from Ap-infected pigs this was only the case
for SAA4, while SAA2 and SAA3 levels were not elevated. This does not correlate with
the results presented in [3].
In the Olsen et al. study, serum samples in the Ap experiment were taken just before
inoculation (0h) and 24h PI as described previously showing a prominent APR and clinical
signs of infections. Tissue samples obtained from the liver, lungs, and spleen were collected
during necropsy [3] as described previously [232, 233] showing a prominent APR and
clinical signs of infections [29]. No clear reason could be found in the experimental setup
for the discrepancy between results found in this Ph.D. and in [3], implying that the SAA
peptide profile selected may not be adequate to detect SAA peptides in this type of
infection model. While the results did not correlate completely with the results found in
[3], they clearly showed that the SRM method was successfully developed and could detect
all porcine SAA isoforms. The development of this method is important, as it will enable
us to obtain more information on the absolute concentration of SAA isoforms at the
protein level and of the correlation between tissue-specific SAA isoform expression and
circulating SAA isoform profiles.
7.3 Absolute concentrations of SAA isoforms
In addition to providing a picture of the relative SAA levels in the porcine serum samples
the data obtained during SRM without isotopically labeled peptides allowed for selecting
the final quantotypic peptides used for the quantification of SAA isoforms as presented in
Paper I. As mentioned, not all proteotypic peptides are quantotypic, and the abundance
of a quantotypic peptide must correlate with the abundance of the parent protein [205-
207]. Where proteotypic peptides need “only” to be unique, continuously observed and
112
fragment well in a mass spectrometer, quantotypic peptides needs to comply with more
criteria as described in table 3. One criteria is, that the peptide should be detectable in a
LC-MS experiment, and the peptide should ionize efficiently and provide a stable and
intense signal, which were not the case for one peptide for each of the SAA isoforms, as
described in Paper I [205-207]. The remaining two quantotypic peptides were subsequently
used for the absolute quantification for each of the porcine SAA isoforms. Not shown in
Paper I is the concentrations of the SAA isoforms in serum samples from pigs infected
with Sa in the IAO model, which are presented in fig. 27.
0
1
2
3
4
SAA2
Seru
m c
once
ntra
tion,
µg/m
L
ControlsIAO pigs
0.0
0.1
0.2
0.3
0.4
0.5
SAA3
Seru
m c
once
ntra
tion,
µg/m
L
ControlsIAO pigs
0.0
0.1
0.2
0.3
0.4
0.5
SAA4
Seru
m c
once
ntra
tion,
µg/m
LControlsIAOpigs
SAA concentrations at euthanasia in Sa infected pigs - IAO infection model
Fig. 27. Mean serum concentrations of SAA isoforms 2, 3, and 4, respectively, 5 days after implantation of a steel implant along with inoculation with Sa and before and 5 days after sterile saline injection (controls) measured by the SRM MS method.
As seen in fig. 27, serum concentrations of all three SAA isoforms were higher in the pigs
subjected to the IAO model when compared to control pigs at euthanasia; SAA2 reached
around 2 µg/mL, SAA3 and SAA4 reached around 0.3 µg/mL. This correlated well with
the relative SAA concentrations where the largest difference was observed for SAA2 for
IAO pigs after euthanasia when compared to control pigs, and minor differences were
observed for SAA3 and SAA4.
In addition to these results, presented in Paper I are also the absolute concentrations of
the SAA isoforms. The serum concentrations of both SAA2 and SAA3 isoforms were higher
in all infected pigs, 24h after Sa infection when compared to control pigs and 0h
113
measurements in infected and control pigs. At 24h PI, the concentration of SAA2 was 2-
3 µg/mL, SAA3 around 0.1 µg/mL and 0.02-0.03 µg/mL for SAA4. The results were not
as consistent for SAA4 as for SAA2 and SAA3, but a slight elevation was also seen for
this isoform 24h after infection. The serum concentration of SAA3 was much lower than
that of both SAA2 and SAA4 confirming that the major circulating isoform of pig SAA is
SAA2 as originally suggested by Olsen et al. (2013) [29].
Comparing the SAA isoform concentrations in serum samples from pigs with the IAO
model to those of pigs with an intravenous Sa infection model, is interesting, as SAA seems
to be differently regulated depending on tissue and infection. The results presented in fig.
27 show that the SAA concentrations differ and in fig. 28 the mean fold changes for the
SAA isoforms in serum samples of the two Sa infection models (intravenous and IAO)
when compared to control pigs (sterile saline-injected) at 24 hours PI.
SAA2 SAA3 SAA40
5
10
15
Fold
cha
nge
(CTR
L=1)
Intravenous Sa infection model
SAA2 SAA3 SAA40
5
10
15
Fold
cha
nge
(CTR
L=1)
IAO infection model
Fig. 28 Mean fold changes of SAA isoforms in serum samples of Sa-infected pigs compared to control pigs (sterile saline-injected) at 24 hours PI. For comparison, the corresponding fold change found for total SAA as determined by ELISA in samples analyzed in [29] is indicated by the horizontal line. Left: Intravenous SA infection model Right: IAO infection model.
In the IAO samples, serum concentration fold changes from normal controls to Sa–infected
pigs were found to be approximately 15 for SAA2, 6 for SAA3, and 1.5 for SAA4. In
114
comparison, the fold changes from the intravenous Sa-infected pigs were 10 for SAA2, 15
for SAA3, and 2.6 for SAA4. The SAA2 isoform serum concentration fold change found in
here (and presented in Paper I) corresponded well with the hepatic SAA2 mRNA fold
change reported previously in [29] for the same animals. This implies a hepatic origin of
the circulating porcine SAA2 isoform and indicates that the SRM MS method allows to
elucidate the tissue origin of circulating biomarkers, as also noted by Qin et al. (2015)
[242]. Absolute concentrations of SAA isoforms were also different when comparing the
two sample sets, with a higher SAA2 concentration was observed with the intravenous Sa
infection model (2-3 µg/mL compared to around 2 µg/mL), for SAA3 a higher
concentration was observed for the IAO model (around 0.3 µg/mL compared to around
0.1 µg/mL). For SAA4, higher concentrations were found in the IAO model (around 0.3
µg/mL compared to 0.02-0.03 µg/mL).
These differences in SAA concentration and fold changes in different infection models may
be an indication of the tissue-specific differential regulation of isoforms during the APR
observed by Olsen et al. and others at the gene expression level [29, 44, 98, 99, 132, 154].
When comparing the relative SAA levels for the two infection models, the levels are more
similar to the highest levels found in the intravenous Sa model.
In Olsen et al. (2013) [29] an ELISA was used to quantify the total SAA concentrations
in the same serum samples as investigated here. They found the total SAA concentrations
to be below 50 µg/mL for infected pigs before inoculation and in control pigs, while after
PI it reached a mean of approximately 460 µg/mL. This contrasts strongly with the results
found in this Ph.D. and Paper I. SAA concentrations measured here are much lower than
those measured by Olsen et al. [29], although fold changes were similar in both studies.
In the Jensen et al. (2016) study [30] total SAA concentrations were also quantified by
ELISA to around 30 µg/mL for pigs infected with Sa using the IAO model. This is again
much higher values than detected by the SRM method developed in the Ph.D.
115
Several possible reasons exits, including:
- Inaccurate calibration of the commercial standard used for the SAA immunoassay.
According to information from the company (Gallagher, pers. comm. 2020), the purified
human SAA standard used to calibrate the Tridelta Ltd. multispecies SAA ELISA kit was
quantified by amino acid analyses and compared to a purified standard of porcine SAA,
likewise quantified by amino acid analysis. The comparison was done in the ELISA and
used to calibrate the human standard to represent the porcine standard. However, this
procedure may not be as precise as there is a risk of the porcine SAA standard not being
pure, as porcine SAA is notoriously difficult to purify [161]. This can also be suspected for
the primary human standard where any error in quantification would heavily influence
the accuracy of the quantification of the secondary standards, including porcine SAA.
There is therefore no guarantee that the kit standard is accurate for porcine SAA.
- Incomplete tryptic fragmentation.
The SAA isoform quantification of the SRM MS method in this Ph.D. project assumes full
tryptic fragmentation and a subsequent full fragment ionization of all SAA peptides used
for quantification. If this were not the reality, possibly due to SAA being shielded from
full trypsin digestion by its complexation with HDL, the tryptic peptide concentrations
measured would be lower and not correlate to the concentration of the cognate SAA
isoform protein. In addition to this, not all tryptic peptides will become successfully ionized
[243].
- Presence of other SAA isoforms.
The peptides used in this study were based on SAA sequences found in Uniprot. The
entries for SAA2 and SAA4 were un-reviewed and had not undergone manual curation by
a Uniprot biocurators. They were therefore updated automatically following the Ensembl
gene model which resulted in their removal due to an Ensembl update in December 2019.
116
As the sequence for SAA3 found in Uniprot (Q2HXZ9) is based on the porcine SAA cDNA
sequence published by Chang et al. [100], and had been reviewed, it was not deleted. In
general, there still is some confusion regarding nomenclature and sequences for the SAA
isoforms, making it difficult to verify if the protein discovered is in fact SAA2 or SAA3.
In other organisms analyzed SAA2 and SAA3 share the highest degree of homology (86%
in pigs) while SAA4 is the most divergent; a pattern that is also seen with the porcine
SAA sequences. Results in this Ph.D. project and from the study done by Olsen et al.
(2013) [29] clearly show that the porcine SAA proteins with the corresponding peptides
analyzed are present in the porcine serum samples. From our data, Uniprot has now
curated the porcine SAA sequences, which will be publicly available at the UniProt release
on June 17th, 2020 (Release 2020_03) as SAA2 (P0DSO0) and SAA4 (P0DSN9). However,
as presented in Paper I, not all unique SAA peptides were equally good proteo and
quantotypic peptides, raising the question of whether all variants of SAA isoforms are
quantifiable or if the peptides may represent differently regulated specific SAA isoforms.
The results presented in Paper I may indicate the existence of porcine SAA variants that
do result in the anticipated tryptic fragments which, however, may exist in the circulating
porcine SAA population as sub-variants. These are not detected by the SRM method
developed here, but they are detected in immunoassays. Several human SAA subtypes
have been found in blood [75, 150, 165, 244] and a PRM method for quantitation in
depleted human serum samples from non-small cell lung cancer patients compared to
healthy controls have been described [245], and it would be very interesting to explore the
presence of such SAA protein isoforms in the pig. To what extent are they accounting for
the acute phase variations quantified by ELISA? For this a non-targeted, global
quantitative MS method will be needed, such as isobarically labeled samples analyzed by
tandem MS (TMT based MS), or for more accurate quantification methods such as
SWATH-MS [226, 230].
117
7.4 Conclusions
Conclusively, in this Ph.D. project a validated quantitative SRM MS method was
developed and for the first time SAA isoform abundance was directly quantified in crude
porcine serum samples, enabling the correlation between variations in isoform mRNA in
various tissues and isoform protein to be established. This is of great importance for the
further development of pigs as animal models for human inflammation-based diseases. A
direct comparison between separate pig infection models was possible, and the developed
method will be important in answering fundamental questions on the tissue distribution
of SAA isoforms, and their regulation during health and disease to be elucidated.
However, the unique SAA isoform peptides available are not all suitable as quantotypic
peptides and using other methods with higher specificity and the possibility of untargeted
sample analysis in correlation with SRM to analyze samples will be necessary
7.5 Perspectives
Future studies using the developed SRM method on porcine serum samples from additional
infection models such as influenza A, Osteomyelitis, Rheumatoid arthritis, and even
coronavirus, would be of great interest along with studies comparing bacterial infections
with viral infections and studies of diseases with inflammatory components to samples
taken at different phases as the disease progresses. Compiling data regarding SAA isoform
concentrations in the mentioned conditions will provide further support of the tissue-
specific differential regulation observed and allow for a greater understanding of not just
the role of SAA, but also the potential of SAA as a biomarker.
Additional studies of interest therefore include developing an SRM method for SAA
isoform detection and quantification in human serum samples as current methods using
118
acute phase proteins, as biomarkers are often not specific enough to distinguish between
diseases. If the same tissue-specific differential regulation is observed in human samples,
quantified SAA isoform levels could provide this distinction and allow for an early
differential diagnosis of such diseases.
Along with SRM, PRM would be a suitable method of choice for these further
developments, as it offers a higher specificity than SRM, as it obtains a full scan (all
transitions), this also results in an easier and faster method development phase, and a
future method for biomarker measurements should be fast, easy to set up and vary and
with high reproducibility. In addition to this, as all potential product ions of a peptide
upon fragmentation are available to confirm the identity of the peptide and since PRM
monitors product ions with high resolution interfering ions are less of a problem [194, 197,
198].
It would also be interesting to analyze samples using Tandem Mass Tags (TMT). Using
TMT allows for peptides from different samples to be identified by their relative
abundance with greater ease and accuracy than other labeling methods and can
simultaneously determine the identity and relative abundances of peptide pairs. Compared
to other MS type measurements, TMT has a very high signal-to-noise ratio and would
make for a very suitable method to measure the protein changes over time and obtain a
more detailed picture of the dynamics of SAA isoform levels and would be a perfect method
of choice for gathering information regarding the existence of possible unidentified SAA
variants as it is a non-targeted method [225, 226].
119
References
1. McGeer, P.L., M. Schulzer, and E.G. McGeer, Arthritis and anti-inflammatory
agents as possible protective factors for Alzheimer's disease: A review of 17 epidemiologic studies. Neurology, 1996. 47(2): p. 425-432.
2. McGeer, P.L., J. Rogers, and E.G. McGeer, Inflammation, antiinflammatory agents, and Alzheimer's disease: The last 22 years. 2017, IOS Press. p. 11-15.
3. Caggiu, E., et al., Inflammation, infectious triggers, and Parkinson's disease. Frontiers in Neurology, 2019. 10.
4. Nikiphorou, E. and G.E. Fragoulis, Inflammation, obesity and rheumatic disease: common mechanistic links. A narrative review. Therapeutic advances in musculoskeletal disease, 2018. 10(8): p. 157-167.
5. McArdle, M.A., E.B. Kennedy, and H.M. Roche, Linking Inflammation, Obesity, and Diabetes, R.S. Ahima, Editor. 2016, Springer International Publishing. p. 505-524.
6. Cottam, D.R., et al., The Chronic Inflammatory Hypothesis for the Morbidity Associated with Morbid Obesity: Implications and Effects of Weight Loss. Obesity Surgery, 2004. 14(5): p. 589-600.
7. Harpsoe, M.C., et al., Body mass index and risk of autoimmune diseases: a study within the Danish National Birth Cohort. Int J Epidemiol, 2014. 43(3): p. 843-55.
8. Chen, Z., et al., Anti-inflammatory and immune-regulatory cytokines in rheumatoid arthritis. Nature Reviews Rheumatology, 2019. 15(1): p. 9-17.
9. Duan, L., X. Rao, and K. Raj Sigdel, Regulation of Inflammation in Autoimmune Disease. 2019.
10. Fernández-Bergés, D., et al., Metabolic and inflammatory profiles of biomarkers in obesity, metabolic syndrome, and diabetes in a mediterranean population. DARIOS inflammatory study. Revista Espanola De Cardiologia, 2014. 67(8): p. 624-631.
120
11. Lim, S.L., C.J. Rodriguez-Ortiz, and M. Kitazawa, Infection, systemic inflammation, and Alzheimer's disease. Microbes and Infection, 2015. 17(8): p. 549-556.
12. Baumann, H. and J. Gauldie, The acute phase response. Immunology Today, 1994. 15(2): p. 74-80.
13. Gabay, C. and I. Kushner, Acute-phase proteins and other systemic responses to inflammation. New England Journal of Medicine, 1999. 340(6): p. 448-454.
14. Moshage, H., Cytokines and the hepatic acute phase response. Journal of Pathology, 1997. 181(3): p. 257-266.
15. Schrodl, W., et al., Acute phase proteins as promising biomarkers: Perspectives and limitations for human and veterinary medicine. Proteomics Clin Appl, 2016. 10(11): p. 1077-1092.
16. Eckersall, P.D. and R. Bell, Acute phase proteins: Biomarkers of infection and inflammation in veterinary medicine. Vet J, 2010. 185(1): p. 23-7.
17. Heegaard, P.M., et al., Optimal combinations of acute phase proteins for detecting infectious disease in pigs. Vet Res, 2011. 42: p. 50.
18. Ebersole, J.L. and D. Cappelli, Acute-phase reactants in infections and inflammatory diseases. Periodontology 2000, 2000. 23(1): p. 19-49.
19. Germolec, D.R., R.P. Frawley, and E. Evans, Markers of inflammation. Methods in Molecular Biology (clifton, N.j.), 2010. 598: p. 53-73.
20. Sipe, J.D., Induction of the acute-phase serum protein SAA requires both RNA and protein synthesis. British journal of experimental pathology, 1978. 59(3): p. 305-310.
21. McAdam, K.P., et al., Changes in human serum amyloid A and C-reactive protein after etiocholanolone-induced inflammation. The Journal of clinical investigation, 1978. 61(2): p. 390-394.
22. Kushner, I., The phenomenon of the acute phase response. Ann N Y Acad Sci, 1982. 389: p. 39-48.
23. Uhlar, C.M., et al., Evolution of the serum amyloid A (SAA) protein superfamily. Genomics, 1994. 19(2): p. 228-35.
24. Sipe, J., Revised nomenclature for serum amyloid A (SAA). Nomenclature Committee of the International Society of Amyloidosis. Part 2. Amyloid, 1999. 6(1): p. 67-70.
25. Betts, J.C., et al., The human acute-phase serum amyloid A gene family: structure, evolution and expression in hepatoma cells. Scand J Immunol, 1991. 34(4): p. 471-82.
121
26. Meek, R.L., N. Eriksen, and E.P. Benditt, Murine serum amyloid A3 is a high density apolipoprotein and is secreted by macrophages. Proc Natl Acad Sci U S A, 1992. 89(17): p. 7949-52.
27. Jacobsen, S., et al., Serum amyloid A isoforms in serum and synovial fluid in horses with lipopolysaccharide-induced arthritis. Vet Immunol Immunopathol, 2006. 110(3-4): p. 325-30.
28. Jacobsen, S., et al., Kinetics of local and systemic isoforms of serum amyloid A in bovine mastitic milk. Vet Immunol Immunopathol, 2005. 104(1-2): p. 21-31.
29. Olsen, H.G., et al., Organization and biology of the porcine serum amyloid A (SAA) gene cluster: isoform specific responses to bacterial infection. PLoS One, 2013. 8(10): p. e76695.
30. Jensen, L.K., et al., Novel porcine model of implant-associated osteomyelitis: A comprehensive analysis of local, regional, and systemic response. J Orthop Res, 2017. 35(10): p. 2211-2221.
31. Parry, C.L., Acute phase proteins. 2017, CRC Press. p. 677-743. 32. Iliev, P.T. and T.M. Georgieva, Acute phase proteins in sheep and goats – function,
reference ranges and assessment methods: An overview. Bulgarian Journal of Veterinary Medicine, 2018. 21(1): p. 1-16.
33. Watterson, C., et al., A Comparative Analysis of Acute-phase Proteins as Inflammatory Biomarkers in Preclinical Toxicology Studies: Implications for Preclinical to Clinical Translation. Toxicologic Pathology, 2009. 37(1): p. 28-33.
34. Waterson, C.L., Proteins. Animal Clinical Chemistry, A practical handbook for toxicologists and biomedical researchers, 2009: p. 159-181.
35. Petersen, H., J.P. Nielsen, and P.M.H. Heegaard, Application of acute phase protein measurements in veterinary clinical chemistry. Veterinary Research, 2004. 35(2): p. 163-187.
36. Black, S., I. Kushner, and D. Samols, C-reactive Protein. Journal of Biological Chemistry, 2004. 279(47): p. 48487-48490.
37. Cray, C., J. Zaias, and N.H. Altman, Acute phase response in animals: a review. Comp Med, 2009. 59(6): p. 517-26.
38. Steel, D.M. and A.S. Whitehead, The major acute phase reactants: C-reactive protein, serum amyloid P component and serum amyloid A protein. Immunol Today, 1994. 15(2): p. 81-8.
39. Safi, S., Acute Phase Proteins-Analysis, Clinical Applications and Potentials. Inflammatory Diseases-Immunopathology, Clinical and Pharmacological Bases, 2012.
122
40. Agassandian, M., et al., C-reactive protein and lung diseases. International Journal of Biochemistry and Cell Biology, 2014. 53: p. 77-88.
41. Marnell, L., C. Mold, and T.W. Du Clos, C-reactive protein: Ligands, receptors and role in inflammation. Clinical Immunology (orlando, Fla. Print), 2005.
42. Du Clos, T.W., Pentraxins: Structure, Function, and Role in Inflammation. Isrn Inflammation, 2013. 2013.
43. Cox, N., D. Pilling, and R.H. Gomer, Serum amyloid P: A systemic regulator of the innate immune response. Journal of Leukocyte Biology, 2014. 96(5): p. 739-743.
44. Uhlar, C.M. and A.S. Whitehead, Serum amyloid A, the major vertebrate acute-phase reactant. Eur J Biochem, 1999. 265(2): p. 501-23.
45. Gatt, M.E., et al., Effect of serum amyloid A on selected in vitro functions of isolated human neutrophils. Journal of Laboratory and Clinical Medicine, 1998. 132(5): p. 414-420.
46. Jahangiri, A., High-density lipoprotein and the acute phase response. Current Opinion in Endocrinology, Diabetes and Obesity, 2010. 17(2): p. 156-160.
47. Marhaug, G. and S.B. Dowton, SERUM AMYLOID-A - AN ACUTE-PHASE APOLIPOPROTEIN AND PRECURSOR OF AA AMYLOID. Baillieres Clinical Rheumatology, 1994. 8(3): p. 553-573.
48. Ding, P.H. and L.J. Jin, The role of lipopolysaccharide-binding protein in innate immunity: A revisit and its relevance to oral/periodontal health. Journal of Periodontal Research, 2013. 49(1): p. 1-9.
49. Schröder, N.W.J. and R.R. Schumann, Non-LPS targets and actions of LPS binding protein (LBP). Journal of Endotoxin Research, 2005. 11(4): p. 237-242.
50. Carter, K. and M. Worwood, Haptoglobin: a review of the major allele frequencies worldwide and their association with diseases. International Journal of Laboratory Hematology, 2007. 29(2): p. 92-110.
51. Ray, A., H. Schatten, and B.K. Ray, Activation of Sp1 and its functional co-operation with serum amyloid A-activating sequence binding factor in synoviocyte cells trigger synergistic action of interleukin-1 and interleukin-6 in serum amyloid A gene expression. J Biol Chem, 1999. 274(7): p. 4300-8.
52. Sipe, J.D., Induction of the acute-phase serum protein SAA requires both RNA and protein synthesis. British Journal of Experimental Pathology, 1978. 59(3).
53. Cerón, J.J., P.D. Eckersall, and S. Martínez-Subiela, Acute phase proteins in dogs and cats: Current knowledge and future perspectives. Veterinary Clinical Pathology, 2005. 34(2): p. 85-99.
123
54. Gruys, E., et al., Acute phase reactants, challenge in the near future of animal production and veterinary medicine. Journal of Zhejiang University: Science, 2005. 6(10): p. 941-947.
55. Toussaint, M.J.M., A.M. Ederen, and E. Gruys, Implication of clinical pathology in assessment of animal health and in animal production and meat inspection. Comparative Haematology International, 1995. 5(3): p. 149-157.
56. Wittum, T.E., et al., Haptoglobin response to clinical respiratory tract disease in feedlot cattle. American Journal of Veterinary Research, 1996. 57(5): p. 646-649.
57. Jin, X., et al., CIRCULATING C-REACTIVE PROTEIN IN OSTEOARTHRITIS: A SYSTEMATIC REVIEW AND META-ANALYSIS. 2014, ELSEVIER SCI LTD.
58. Dallmeier, D. and W. Koenig, Strategies for vascular disease prevention: The role of lipids and related markers including apolipoproteins, low-density lipoproteins (LDL)-particle size, high sensitivity C-reactive protein (hs-CRP), lipoprotein-associated phospholipase A2 (Lp-PLA2) and lipoprotein(a) (Lp(a)). Best Practice and Research: Clinical Endocrinology and Metabolism, 2014. 28(3): p. 281-294.
59. Ansar, W. and S. Ghosh, C-reactive protein and the biology of disease. Immunologic Research, 2013. 56(1): p. 131-142.
60. Magro, F., P. Sousa, and P. Ministro, C-reactive protein in Crohn's disease: How informative is it? Expert Review of Gastroenterology and Hepatology, 2014. 8(4): p. 393-408.
61. Salazar, J., et al., C-Reactive Protein: Clinical and Epidemiological Perspectives. 2014.
62. Sack, G.H., Jr., Serum amyloid A - a review. Mol Med, 2018. 24(1): p. 46. 63. FISCHER, K., et al., Serum Amyloid A: a Biomarker for Renal Cancer. Anticancer
Research, 2012. 32(5): p. 1801-1804. 64. Vietri, L., et al., Serum amyloid A: A potential biomarker of lung disorders.
Respiratory Investigation, 2020. 58(1): p. 21-27. 65. Jousilahti, P., et al., The association of c-reactive protein, serum amyloid a and
fibrinogen with prevalent coronary heart disease — baseline findings of the PAIS project. Atherosclerosis, 2001. 156(2): p. 451-456.
66. Ridker, P.M., et al., C-reactive protein and other markers of inflammation in the prediction of cardiovascular disease in women. N Engl J Med, 2000. 342(12): p. 836-43.
67. Zhang, X., J. Chen, and S. Wang, Serum amyloid A induces a vascular smooth muscle cell phenotype switch through the p38 MAPK signaling pathway. BioMed research international, 2017. 2017.
124
68. Cocco, E., et al., Serum amyloid A: a novel biomarker for endometrial cancer. Cancer, 2010. 116(4): p. 843-51.
69. Cocco, E., et al., Serum amyloid A (SAA): a novel biomarker for uterine serous papillary cancer. Br J Cancer, 2009. 101(2): p. 335-41.
70. Edgell, T., et al., Phase II biomarker trial of a multimarker diagnostic for ovarian cancer. J Cancer Res Clin Oncol, 2010. 136(7): p. 1079-88.
71. Findeisen, P., et al., Serum Amyloid A As a Prognostic Marker in Melanoma Identified by Proteomic Profiling. Journal of Clinical Oncology, 2009. 27(13): p. 2199-2208.
72. Moshkovskii, S.A., et al., Ovarian cancer marker of 11.7 kDa detected by proteomics is a serum amyloid A1. Proteomics, 2005. 5(14): p. 3790-7.
73. Paret, C., et al., Inflammatory protein serum amyloid A1 marks a subset of conventional renal cell carcinomas with fatal outcome. Eur Urol, 2010. 57(5): p. 859-66.
74. Sung, H.J., et al., Identification and validation of SAA as a potential lung cancer biomarker and its involvement in metastatic pathogenesis of lung cancer. J Proteome Res, 2011. 10(3): p. 1383-95.
75. Tolson, J., et al., Serum protein profiling by SELDI mass spectrometry: detection of multiple variants of serum amyloid alpha in renal cancer patients. Lab Invest, 2004. 84(7): p. 845-56.
76. Wood, S.L., et al., Association of serum amyloid A protein and peptide fragments with prognosis in renal cancer. Br J Cancer, 2010. 103(1): p. 101-11.
77. Inflammatory Diseases - Immunopathology, Clinical and Pharmacological Bases. InTech.
78. Damman, J., et al., Local renal complement C3 induction by donor brain death is associated with reduced renal allograft function after transplantation. Nephrol Dial Transplant, 2011. 26(7): p. 2345-54.
79. Ray, S., et al., Differential expression of serum/plasma proteins in various infectious diseases: specific or nonspecific signatures. Proteomics Clin Appl, 2014. 8(1-2): p. 53-72.
80. Kjelgaard-Hansen, M., et al., Serum amyloid A isoforms in serum and synovial fluid from spontaneously diseased dogs with joint diseases or other conditions. Vet Immunol Immunopathol, 2007. 117(3-4): p. 296-301.
81. Moshkovskii, S.A., Why do cancer cells produce serum amyloid A acute-phase protein? Biochemistry (Mosc), 2012. 77(4): p. 339-41.
125
82. Bode, J.G., et al., Hepatic acute phase proteins--regulation by IL-6- and IL-1-type cytokines involving STAT3 and its crosstalk with NF-kappaB-dependent signaling. Eur J Cell Biol, 2012. 91(6-7): p. 496-505.
83. Schroedl, W., et al., A novel acute phase marker in cattle: Lipopolysaccharide binding protein (LBP). Journal of Endotoxin Research, 2001. 7(1): p. 49-52.
84. Gollaher, C.J. and L.L. Bausserman, Hepatic catabolism of serum amyloid A during an acute phase response and chronic inflammation. Proc Soc Exp Biol Med, 1990. 194(3): p. 245-50.
85. Li, L., P.A. Thompson, and R.L. Kitchens, Infection induces a positive acute phase apolipoprotein E response from a negative acute phase gene: role of hepatic LDL receptors. J Lipid Res, 2008. 49(8): p. 1782-93.
86. Coetzee, G.A., et al., Serum amyloid A-containing human high density lipoprotein 3. Density, size, and apolipoprotein composition. J Biol Chem, 1986. 261(21): p. 9644-51.
87. Ceciliani, F., et al., Acute phase proteins in ruminants. Journal of Proteomics, 2012. 75(14): p. 4207-4231.
88. Linke, R.P., et al., INHIBITION OF THE OXIDATIVE BURST RESPONSE OF N-FORMYL PEPTIDE-STIMULATED NEUTROPHILS BY SERUM AMYLOID-A PROTEIN. Biochemical and Biophysical Research Communications, 1991. 176(3): p. 1100-1105.
89. Hari-Dass, R., et al., Serum amyloid A protein binds to outer membrane protein A of gram-negative bacteria. J Biol Chem, 2005. 280(19): p. 18562-7.
90. Tothova, C., et al., Acute Phase Proteins as Markers of Diseases in Farm Animals, F. Veas, Editor. 2011, INTECH EUROPE.
91. Murata, H., N. Shimada, and M. Yoshioka, Current research on acute phase proteins in veterinary diagnosis: An overview. Veterinary Journal, 2004. 168(1): p. 28-40.
92. Benditt, E.P. and N. Eriksen, Amyloid protein SAA is associated with high density lipoprotein from human serum. Proceedings of the National Academy of Sciences of the United States of America, 1977. 74(9): p. 4025-4028.
93. Benditt, E.P., N. Eriksen, and R.H. Hanson, Amyloid protein SAA is an apoprotein of mouse plasma high density lipoprotein. Proceedings of the National Academy of Sciences of the United States of America, 1979. 76(8): p. 4092-4096.
94. Badolato, R., et al., Serum amyloid A is a chemoattractant: induction of migration, adhesion, and tissue infiltration of monocytes and polymorphonuclear leukocytes. J Exp Med, 1994. 180(1): p. 203-9.
126
95. Xu, L.L., et al., A NOVEL BIOLOGIC FUNCTION OF SERUM AMYLOID-A - INDUCTION OF T-LYMPHOCYTE MIGRATION AND ADHESION. Journal of Immunology, 1995. 155(3): p. 1184-1190.
96. Morrow, J.F., et al., Induction of hepatic synthesis of serum amyloid A protein and actin. Proc Natl Acad Sci U S A, 1981. 78(8): p. 4718-22.
97. Lowell, C.A., R.S. Stearman, and J.F. Morrow, Transcriptional regulation of serum amyloid A gene expression. J Biol Chem, 1986. 261(18): p. 8453-61.
98. Upragarin, N., et al., Extrahepatic production of acute phase serum amyloid A. Histol Histopathol, 2005. 20(4): p. 1295-307.
99. Skovgaard, K., et al., Rapid and widely disseminated acute phase protein response after experimental bacterial infection of pigs. Vet Res, 2009. 40(3): p. 23.
100. Chang, W.C., et al., The effect of dietary docosahexaenoic acid enrichment on the expression of porcine hepatic genes. Asian-australasian Journal of Animal Sciences, 2007. 20(5): p. 768-774.
101. Soler, L., et al., Serum amyloid A3 (SAA3), not SAA1 appears to be the major acute phase SAA isoform in the pig. Vet Immunol Immunopathol, 2011. 141(1-2): p. 109-15.
102. McDonald, T.L., et al., Elevated extrahepatic expression and secretion of mammary-associated serum amyloid A 3 (M-SAA3) into colostrum. Veterinary Immunology and Immunopathology, 2001. 83(3-4).
103. de Jesus Rodriguez, B., et al., Identification in milk of a serum amyloid A peptide chemoattractant for B lymphoblasts. Bmc Immunology, 2009. 10(1).
104. Meeker, A.K. and G.H. Sack, Jr., A fusion protein between serum amyloid A and staphylococcal nuclease--synthesis, purification, and structural studies. Proteins, 1998. 30(4): p. 381-7.
105. Turnell, W., et al., Secondary structure prediction of human SAA1. Presumptive identification of calcium and lipid binding sites. Mol Biol Med, 1986. 3(5): p. 387-407.
106. Lu, J., et al., Structural mechanism of serum amyloid A-mediated inflammatory amyloidosis. Proc Natl Acad Sci U S A, 2014. 111(14): p. 5189-94.
107. Derebe, M.G., et al., Serum amyloid A is a retinol binding protein that transports retinol during bacterial infection. Elife, 2014. 3(2014).
108. Wang, L., et al., Murine apolipoprotein serum amyloid A in solution forms a hexamer containing a central channel. Proc Natl Acad Sci U S A, 2002. 99(25): p. 15947-52.
127
109. Wang, L., H.A. Lashuel, and W. Colon, From hexamer to amyloid: marginal stability of apolipoprotein SAA2.2 leads to in vitro fibril formation at physiological temperature. Amyloid, 2005. 12(3): p. 139-48.
110. Wang, Y., et al., Serum amyloid A 2.2 refolds into a octameric oligomer that slowly converts to a more stable hexamer. Biochem Biophys Res Commun, 2011. 407(4): p. 725-9.
111. Li, W., et al., Serum Amyloid A Stimulates PKR Expression and HMGB1 Release Possibly through TLR4/RAGE Receptors. Mol Med, 2015. 21: p. 515-25.
112. Frame, N.M. and O. Gursky, Structure of serum amyloid A suggests a mechanism for selective lipoprotein binding and functions: SAA as a hub in macromolecular interaction networks. FEBS Lett, 2016. 590(6): p. 866-79.
113. Das, M. and O. Gursky, Amyloid-forming properties of human apolipoproteins: Sequence analyses and structural insights. Advances in Experimental Medicine and Biology, 2015. 855: p. 175-211.
114. De Buck, M., et al., The cytokine-serum amyloid A-chemokine network. Cytokine and Growth Factor Reviews, 2016. 30(Sp. Iss. SI): p. 55-69.
115. Chen, M., et al., Serum amyloid A1 isoforms display different efficacy at Toll-like receptor 2 and formyl peptide receptor 2. Immunobiology, 2014. 219(12): p. 916-923.
116. Ye, R.D. and L. Sun, Emerging functions of serum amyloid A in inflammation. J Leukoc Biol, 2015. 98(6): p. 923-9.
117. Su, S.B., et al., A seven-transmembrane, G protein-coupled receptor, FPRL1, mediates the chemotactic activity of serum amyloid A for human phagocytic cells. J Exp Med, 1999. 189(2): p. 395-402.
118. Dufton, N., et al., Anti-inflammatory role of the murine formyl-peptide receptor 2: ligand-specific effects on leukocyte responses and experimental inflammation. J Immunol, 2010. 184(5): p. 2611-2619.
119. Furlaneto, C.J. and A. Campa, A novel function of serum amyloid A: a potent stimulus for the release of tumor necrosis factor-alpha, interleukin-1beta, and interleukin-8 by human blood neutrophil. Biochem Biophys Res Commun, 2000. 268(2): p. 405-8.
120. Lucherini, O.M., et al., Critical regulation of Th17 cell differentiation by serum amyloid-A signalling in Behcet's disease. Immunology Letters, 2018. 201: p. 38-44.
121. Sano, T., et al., An IL-23R/IL-22 Circuit Regulates Epithelial Serum Amyloid A to Promote Local Effector Th17 Responses. Cell, 2015. 163(2).
128
122. Ather, J.L., et al., Serum Amyloid A Activates the NLRP3 Inflammasome and Promotes Th17 Allergic Asthma in Mice. J Immunol, 2011. 187(1): p. 64-73.
123. Jensen, L.E., et al., Acute phase proteins in salmonids: evolutionary analyses and acute phase response. J Immunol, 1997. 158(1): p. 384-92.
124. Kluve-Beckerman, B., et al., Localization of human SAA gene(s) to chromosome 11 and detection of DNA polymorphisms. Biochemical and Biophysical Research Communications, 1986. 137(3): p. 1196-1204.
125. Sack, G.H., Jr. and C.C. Talbot, Jr., The human serum amyloid A (SAA)-encoding gene GSAA1: nucleotide sequence and possible autocrine-collagenase-inducer function. Gene, 1989. 84(2): p. 509-15.
126. Sack, G.H., Jr., et al., Molecular analysis of the human serum amyloid A (SAA) gene family. Scand J Immunol, 1989. 29(1): p. 113-9.
127. Lowell, C.A., et al., Structure of the murine serum amyloid A gene family. Gene conversion. J Biol Chem, 1986. 261(18): p. 8442-52.
128. Kluve-Beckerman, B., M.L. Drumm, and M.D. Benson, Nonexpression of the human serum amyloid A three (SAA3) gene. DNA Cell Biol, 1991. 10(9): p. 651-61.
129. Whitehead, A.S., et al., Identification of novel members of the serum amyloid A protein superfamily as constitutive apolipoproteins of high density lipoprotein. J Biol Chem, 1992. 267(6): p. 3862-7.
130. de Beer, M.C., et al., Mouse serum amyloid A protein (SAA5) structure and expression. J Biol Chem, 1994. 269(6): p. 4661-7.
131. Liao, D., Concerted evolution: molecular mechanism and biological implications. Am J Hum Genet, 1999. 64(1): p. 24-30.
132. Huan, B., et al., Porcine serum amyloid A3 is expressed in extrahepatic tissues and facilitates viral replication during porcine respiratory and reproductive syndrome virus infection. Developmental and Comparative Immunology, 2018. 79: p. 51-58.
133. Hansen, A.E., M.K. Schaap, and M. Kjelgaard-Hansen, Evaluation of a commercially available human serum amyloid A (SAA) turbidimetric immunoassay for determination of feline SAA concentration. Veterinary Research Communications, 2006. 30(8): p. 863-872.
134. Hulten, C., et al., A non-competitive chemiluminescence enzyme immunoassay for the equine acute phase protein serum amyloid A (SAA) - a clinically useful inflammatory marker in the horse. Veterinary Immunology and Immunopathology, 1999. 68(2-4): p. 267-281.
129
135. Pepys, M.B., et al., SERUM AMYLOID A PROTEIN (SAA) IN HORSES - OBJECTIVE MEASUREMENT OF THE ACUTE PHASE RESPONSE. Equine Veterinary Journal, 1989. 21(2): p. 106-109.
136. Soler, L., et al., Development and validation of a novel competitive ELISA for the detection of serum amyloid A in pigs. Veterinary Journal, 2011. 190(2).
137. Christensen, M.B., et al., Investigation of the solubility and the potentials for purification of serum amyloid A (SAA) from equine acute phase serum - A pilot study. Bmc Research Notes, 2013. 6(1).
138. Ham, D. and B. Karska-Wysocki, Purification and separation of hydrophobic serum amyloid a precursor isoforms by a one-step preparative method. Journal of Immunological Methods, 2005. 303(1-2): p. 11-18.
139. Hultén, C., et al., The acute phase serum amyloid A protein (SAA) in the horse: isolation and characterization of three isoforms. Veterinary Immunology and Immunopathology, 1997. 57(3-4): p. 215-227.
140. Malle, E., et al., PURIFICATION OF SERUM AMYLOID-A AND ITS ISOFORMS FROM HUMAN PLASMA BY HYDROPHOBIC INTERACTION CHROMATOGRAPHY AND PREPARATIVE ISOELECTRIC-FOCUSING. Electrophoresis, 1992. 13(7): p. 422-428.
141. Hultén, C., et al., The acute phase serum amyloid A protein (SAA) in the horse: isolation and characterization of three isoforms. Veterinary Immunology and Immunopathology, 1997. 57(3): p. 215-227.
142. Strachan, A.F., et al., Identification of three isoform patterns of human serum amyloid A protein. Biochemical Journal, 1988. 250(1): p. 203-207.
143. Havel, R.J., H.A. Eder, and J.H. Bragdon, THE DISTRIBUTION AND CHEMICAL COMPOSITION OF ULTRACENTRIFUGALLY SEPARATED LIPOPROTEINS IN HUMAN SERUM. Journal of Clinical Investigation, 1955. 34(9): p. 1345-1353.
144. Raynes, J.G. and K. McAdam, PURIFICATION OF SERUM AMYLOID-A AND OTHER HIGH-DENSITY APOLIPOPROTEINS BY HYDROPHOBIC INTERACTION CHROMATOGRAPHY. Analytical Biochemistry, 1988. 173(1): p. 116-124.
145. Kluve-Beckerman, B., et al., Characterization of an isoelectric focusing variant of SAA1 (ASP-72) in a family of Turkish origin. Biochemical and Biophysical Research Communications, 1991. 181(181): p. 1097-1102.
146. Hoang, L.T., et al., NOVEL SAA GENE-PRODUCTS IN MUS CAROLI MICE DEFINED BY RAPID AMPLIFICATION OF COMPLEMENTARY-DNA ENDS
130
(3'-RACE AND 5'-RACE). Amyloid-international Journal of Experimental and Clinical Investigation, 1995. 2(2): p. 75-82.
147. Steinmetz, A., et al., ISOPROTEINS AND AN ISOELECTRIC-FOCUSING MUTANT OF HUMAN APOPROTEIN SERUM AMYLOID-A. Electrophoresis, 1990. 11(8): p. 627-630.
148. Beach, C.M., et al., HUMAN SERUM AMYLOID-A PROTEIN - COMPLETE AMINO-ACID-SEQUENCE OF A NEW VARIANT. Biochemical Journal, 1992. 282(2): p. 615-620.
149. Raynes, J.G. and K.P. McAdam, Serum amyloid A isoforms in inflammation. Scand J Immunol, 1991. 33(6): p. 657-66.
150. Sack, G.H., et al., Serum amyloid A1 (SAA1) protein in human colostrum. Febs Open Bio, 2018. 8(3): p. 435-441.
151. Kjelgaard-Hansen, M., et al., Serum amyloid A isoforms in serum and synovial fluid from spontaneously diseased dogs with joint diseases or other conditions. Veterinary Immunology and Immunopathology, 2007. 117(3-4): p. 296-301.
152. Chiba, T., et al., Serum amyloid A3 does not contribute to circulating SAA levels. Journal of Lipid Research, 2009. 50(7): p. 1353-1362.
153. Hadrup, N., et al., Acute phase response and inflammation following pulmonary exposure to low doses of zinc oxide nanoparticles in mice. Nanotoxicology, 2019. 13(9): p. 1275-1292.
154. Lüthje, F.L., et al., The host response to bacterial bone infection involves a local upregulation of several acute phase proteins. Immunobiology, 2020.
155. Skovgaard, K., et al., Hepatic gene expression changes in pigs experimentally infected with the lung pathogen Actinobacillus pleuropneumoniae as analysed with an innate immunity focused microarray. Innate Immun, 2010. 16(6): p. 343-53.
156. Sørensen, N.S., et al., The porcine acute phase protein response to acute clinical and subclinical experimental infection with Streptococcus suis. Veterinary Immunology and Immunopathology, 2006. 113(1-2): p. 157-168.
157. Meek, R.L. and E.P. Benditt, Amyloid A gene family expression in different mouse tissues. J Exp Med, 1986. 164(6): p. 2006-17.
158. Sjoholm, K., et al., A microarray search for genes predominantly expressed in human omental adipocytes: adipose tissue as a major production site of serum amyloid A. J Clin Endocrinol Metab, 2005. 90(4): p. 2233-9.
159. Steel, D.M., et al., Expression and regulation of constitutive and acute phase serum amyloid A mRNAs in hepatic and non-hepatic cell lines. Scand J Immunol, 1996. 44(5): p. 493-500.
131
160. Husby, G., et al., Serum amyloid A (SAA): Biochemistry, genetics and the pathogenesis of AA amyloidosis. Amyloid, 1994. 1(2): p. 119-137.
161. Soler, L., et al., Why working with porcine circulating serum amyloid A is a pig of a job. Journal of Theoretical Biology, 2013. 317: p. 119-125.
162. McDonald, T.L., A. Weber, and J.W. Smith, A monoclonal antibody sandwich immunoassay for serum amyloid A (SAA) protein. J Immunol Methods, 1991. 144(2): p. 149-55.
163. Yamada, T., et al., Measurement of serum amyloid A4 (SAA4): Its constitutive presence in serum. Vol. 1. 2009. 114-118.
164. Sung, H.J., et al., Large-scale isotype-specific quantification of Serum amyloid A 1/2 by multiple reaction monitoring in crude sera. J Proteomics, 2012. 75(7): p. 2170-80.
165. Wu, D.C., et al., Exploring the expression bar code of SAA variants for gastric cancer detection. Proteomics, 2017. 17(11).
166. Gutfeld, O., et al., Expression of serum amyloid A, in normal, dysplastic, and neoplastic human colonic mucosa: implication for a role in colonic tumorigenesis. J Histochem Cytochem, 2006. 54(1): p. 63-73.
167. Urieli-Shoval, S., et al., Expression of serum amyloid a in human ovarian epithelial tumors: implication for a role in ovarian tumorigenesis. J Histochem Cytochem, 2010. 58(11): p. 1015-23.
168. Eckersall, P.D., Review: Proteomic approaches to control lactational parameters in dairy cows. Animal, 2019. 13(1).
169. Zhou, W., E.F. Petricoin, 3rd, and C. Longo, Mass Spectrometry-Based Biomarker Discovery. Methods Mol Biol, 2017. 1606: p. 297-311.
170. Olumee-Shabon, Z., C. Chattopadhaya, and M.J. Myers, Proteomics profiling of swine serum following lipopolysaccharide stimulation. Rapid Communications in Mass Spectrometry, 2020. 34(7).
171. Honour, J.W., Benchtop mass spectrometry in clinical biochemistry. Annals of Clinical Biochemistry, 2003. 40(6): p. 628-638.
172. Smith, J.S. and R.A. Thakur, Mass Spectrometry, S.S. Nielsen, Editor. 2017, Springer International Publishing. p. 165-181.
173. Mellon, F., MASS SPECTROMETRY| Principles and Instrumentation. 2003. 174. Wilm, M., Principles of electrospray ionization. Molecular and Cellular Proteomics,
2011. 10(7). 175. Stroobant, V., E. De Hoffmann, and E. Hoffmann, Mass spectrometry : principles
and applications. 2007: Wiley.
132
176. Urban, J., N.K. Afseth, and D. Štys, Fundamental definitions and confusions in mass spectrometry about mass assignment, centroiding and resolution. TrAC Trends in Analytical Chemistry, 2014. 53: p. 126-136.
177. Norena-Caro, D. LC-MS spectrum 3D analysis. [cited 2020 27-03]; Available from: https://en.wikipedia.org/wiki/File:Liquid_chromatography_MS_spectrum_3D_analysis.png.
178. Tanaka, K., et al., Protein and polymer analyses up tom/z 100 000 by laser ionization time-of-flight mass spectrometry. Rapid Communications in Mass Spectrometry, 1988. 2(8): p. 151-153.
179. Karas, M., D. Bachmann, and F. Hillenkamp, Influence of the wavelength in high-irradiance ultraviolet laser desorption mass spectrometry of organic molecules. Analytical Chemistry, 1985. 57(14): p. 2935-2939.
180. Greco, V., et al., Applications of MALDI-TOF mass spectrometry in clinical proteomics. Expert Review of Proteomics, 2018. 15(8): p. 683-696.
181. Duncan, M.W., et al., Applications of MALDI mass spectrometry in clinical chemistry. Clinical Chemistry, 2016. 62(1): p. 134-143.
182. Šebela, M., Solid mixed matrices and their advantages in matrixassisted laser desorption/ionisation time-of-flight mass spectrometry. Spectroscopy Europe, 2016. 28(1): p. 10-14.
183. Emmett, M.R. and R.M. Caprioli, MICRO-ELECTROSPRAY MASS-SPECTROMETRY - ULTRA-HIGH-SENSITIVITY ANALYSIS OF PEPTIDES AND PROTEINS. Journal of the American Society for Mass Spectrometry, 1994. 5(7): p. 605-613.
184. Wilm, M. and M. Mann, Analytical properties of the nanoelectrospray ion source. Analytical Chemistry, 1996. 68(1): p. 1-8.
185. Hubbard, S.J., Proteome Bioinformatics, ed. S.J.J.A.R.H.S.J.J.A.R. Hubbard. Vol. 604. 2010: HUMANA PRESS INC.
186. Niessen, W.M.A., MS–MS and MSn, in Encyclopedia of Spectroscopy and Spectrometry (Second Edition), J.C. Lindon, Editor. 1999, Academic Press: Oxford. p. 1675-1681.
187. Hao, Z., et al., Current Methods for the Characterization of Posttranslational Modifications in Therapeutic Proteins Using Orbitrap Mass Spectrometry. 2017, wiley. p. 21-34.
188. Baldwin, M.A., Peptides and Proteins Studied Using Mass Spectrometry, in Encyclopedia of Spectroscopy and Spectrometry (Third Edition), J.C. Lindon, G.E. Tranter, and D.W. Koppenaal, Editors. 2017, Academic Press: Oxford. p. 554-562.
133
189. Jones, A.W. and H.J. Cooper, Dissociation techniques in mass spectrometry-based proteomics. Analyst, 2011. 136(17): p. 3419-3429.
190. Johnson, R.S., S.A. Martin, and K. Biemann, Collision-induced fragmentation of (M + H)+ ions of peptides. Side chain specific sequence ions. International Journal of Mass Spectrometry and Ion Processes, 1988. 86: p. 137-154.
191. Zhang, H., et al., Mass spectrometric detection of tissue proteins in plasma. Molecular and Cellular Proteomics, 2007. 6(1): p. 64-71.
192. Liotta, L.A. and E.F. Petricoin, Serum peptidome for cancer detection: Spinning biologic trash into diagnostic gold. Journal of Clinical Investigation, 2006. 116(1): p. 26-30.
193. Anderson, N.L. and N.G. Anderson, The human plasma proteome: history, character, and diagnostic prospects. Molecular and Cellular Proteomics : Mcp, 2002. 1(11): p. 845-867.
194. Peterson, A.C., et al., Parallel reaction monitoring for high resolution and high mass accuracy quantitative, targeted proteomics. Molecular and Cellular Proteomics, 2012. 11(11): p. 1475-1488.
195. Borras, E. and E. Sabido, What is targeted proteomics? A concise revision of targeted acquisition and targeted data analysis in mass spectrometry. Proteomics, 2017. 17(17-18).
196. Proteomics, C. [cited 2020 23-03]; Available from: https://www.creative-proteomics.com/services/parallel-reaction-monitoring-prm.htm.
197. Duncan, M.W., A.L. Yergey, and S.D. Patterson, Quantifying proteins by mass spectrometry: The selectivity of SRM is only part of the problem. Proteomics, 2009. 9(5): p. 1124-1127.
198. Sherman, J., et al., How specific is my SRM?: The issue of precursor and product ion redundancy. Proteomics, 2009. 9(5): p. 1120-1123.
199. Lange, V., et al., Selected reaction monitoring for quantitative proteomics: a tutorial. Mol Syst Biol, 2008. 4: p. 222.
200. Picotti, P. and R. Aebersold, Selected reaction monitoring-based proteomics: workflows, potential, pitfalls and future directions. Nat Methods, 2012. 9(6): p. 555-66.
201. Hüttenhain, R., et al., Perspectives of targeted mass spectrometry for protein biomarker verification. Current Opinion in Chemical Biology, 2009. 13(5-6): p. 518-525.
134
202. Addona, T.A., et al., Multi-site assessment of the precision and reproducibility of multiple reaction monitoring-based measurements of proteins in plasma. Nat Biotechnol, 2009. 27(7): p. 633-41.
203. Onisko, B., et al., Mass spectrometric detection of attomole amounts of the prion protein by nanoLC/MS/MS. Journal of the American Society for Mass Spectrometry, 2007. 18(6): p. 1070-1079.
204. Stahl-Zeng, J., et al., High sensitivity detection of plasma proteins by multiple reaction monitoring of N-glycosites. Mol Cell Proteomics, 2007. 6(10): p. 1809-17.
205. Yost, R.A. and C.G. Enke, SELECTED ION FRAGMENTATION WITH A TANDEM QUADRUPOLE MASS-SPECTROMETER. Journal of the American Chemical Society, 1978. 100(7): p. 2274-2275.
206. Yost, R.A. and C.G. Enke, TRIPLE QUADRUPOLE MASS-SPECTROMETRY FOR DIRECT MIXTURE ANALYSIS AND STRUCTURE ELUCIDATION. Analytical Chemistry, 1979. 51(12).
207. Kuster, B., et al., Scoring proteomes with proteotypic peptide probes. Nat Rev Mol Cell Biol, 2005. 6(7): p. 577-83.
208. Mallick, P., et al., Computational prediction of proteotypic peptides for quantitative proteomics. Nature Biotechnology, 2007. 25(1): p. 125-131.
209. Worboys, J.D., et al., Systematic evaluation of quantotypic peptides for targeted analysis of the human kinome. Nat Methods, 2014. 11(10): p. 1041-4.
210. Simpson, D.M. and R.J. Beynon, QconCATs: design and expression of concatenated protein standards for multiplexed protein quantification. Anal Bioanal Chem, 2012. 404(4): p. 977-89.
211. Deutsch, E.W., H. Lam, and R. Aebersold, PeptideAtlas: a resource for target selection for emerging targeted proteomics workflows. EMBO Rep, 2008. 9(5): p. 429-34.
212. Brownridge, P. and R.J. Beynon, The importance of the digest: Proteolysis and absolute quantification in proteomics. Methods, 2011. 54(4): p. 351-360.
213. Lawless, C. and S.J. Hubbard, Prediction of missed proteolytic cleavages for the selection of surrogate peptides for quantitative proteomics. Omics : a journal of integrative biology, 2012. 16(9): p. 449-456.
214. Holman, S.W., P.F.G. Sims, and C.E. Eyers, The use of selected reaction monitoring in quantitative proteomics. Bioanalysis, 2012. 4(14): p. 1763-1786.
215. Brownridge, P.J., et al., Absolute Multiplexed Protein Quantification Using QconCAT Technology, in Quantitative Methods in Proteomics, K. Marcus, Editor. 2012, Humana Press: Totowa, NJ. p. 267-293.
135
216. Krusemark, C.J., et al., Modifying the charge state distribution of proteins in electrospray ionization mass spectrometry by chemical derivatization. J Am Soc Mass Spectrom, 2009. 20(9): p. 1617-25.
217. Gerber, S.A., et al., Absolute quantification of proteins and phosphoproteins from cell lysates by tandem MS. Proc Natl Acad Sci U S A, 2003. 100(12): p. 6940-5.
218. Barr, J.R., et al., Isotope dilution--mass spectrometric quantification of specific proteins: model application with apolipoprotein A-I. Clin Chem, 1996. 42(10): p. 1676-82.
219. Kirkpatrick, D.S., S.A. Gerber, and S.P. Gygi, The absolute quantification strategy: a general procedure for the quantification of proteins and post-translational modifications. Methods, 2005. 35(3): p. 265-73.
220. Kuhn, E., et al., Quantification of C-reactive protein in the serum of patients with rheumatoid arthritis using multiple reaction monitoring mass spectrometry and 13C-labeled peptide standards. Proteomics, 2004. 4(4): p. 1175-86.
221. Ong, S.E. and M. Mann, Mass Spectrometry–Based Proteomics Turns Quantitative. Nature Chemical Biology, 2005. 1(5): p. 252-262.
222. Smith, P.K., et al., Measurement of protein using bicinchoninic acid. Anal Biochem, 1985. 150(1): p. 76-85.
223. Lowry, O.H., et al., Protein measurement with the Folin phenol reagent. J Biol Chem, 1951. 193(1): p. 265-75.
224. Bradford, M.M., A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem, 1976. 72: p. 248-54.
225. Zhang, L. and J.E. Elias, Relative protein quantification using tandem mass tag mass spectrometry. 2017, Humana Press Inc. p. 185-198.
226. Thompson, A., et al., Tandem mass tags: A novel quantification strategy for comparative analysis of complex protein mixtures by MS/MS. Analytical Chemistry, 2003. 75(8): p. 1895-1904.
227. Lagerstedt, L., L. Azurmendi, and J.-C. Sanchez, Applications of Amine-Reactive Tandem Mass Tags (TMT) in Human Neuroproteomics, E.F.J. Santamaria, Editor. 2017, HUMANA PRESS INC.
228. Doerr, A., DIA mass spectrometry. Nature Methods, 2015. 12(1): p. 35-35. 229. Meyer, J.G. and B. Schilling, Clinical applications of quantitative proteomics using
targeted and untargeted data-independent acquisition techniques. Expert Review of Proteomics, 2017. 14(5): p. 419-429.
136
230. Ludwig, C., et al., Data-independent acquisition-based SWATH-MS for quantitative proteomics: a tutorial. Molecular Systems Biology, 2018. 14(8): p. e8126.
231. Gillet, L.C., et al., Targeted Data Extraction of the MS/MS Spectra Generated by Data-independent Acquisition: A New Concept for Consistent and Accurate Proteome Analysis. Molecular &amp; Cellular Proteomics, 2012. 11(6): p. O111.016717.
232. Jensen, H.E., et al., A non-traumatic Staphylococcus aureus osteomyelitis model in pigs. In Vivo, 2010. 24(3): p. 257-64.
233. Leifsson, P.S., et al., Intravenous inoculation of Staphylococcus aureus in pigs induces severe sepsis as indicated by increased hypercoagulability and hepatic dysfunction. FEMS Microbiol Lett, 2010. 309(2): p. 208-16.
234. Johansen, L.K., et al., A porcine model of acute, haematogenous, localized osteomyelitis due to Staphylococcus aureus: a pathomorphological study. APMIS, 2011. 119(2): p. 111-8.
235. Heegaard, P.M., et al., A robust quantitative solid phase immunoassay for the acute phase protein C-reactive protein (CRP) based on cytidine 5'-diphosphocholine coupled dendrimers. J Immunol Methods, 2009. 343(2): p. 112-8.
236. Skovgaard, K., et al., Validation of putative reference genes for qRT-PCR normalization in tissues and blood from pigs infected with Actinobacillus pleuropneumoniae. Vet Immunol Immunopathol, 2007. 118(1-2): p. 140-6.
237. Kristensen, C.S., et al., Demonstration of airborne transmission of Actinobacillus pleuropneumoniae serotype 2 between simulated pig units located at close range. Vet Microbiol, 2004. 98(3-4): p. 243-9.
238. Jessen, F. and T. Wulff, Triton X-114 cloud point extraction to subfractionate blood plasma proteins for two-dimensional gel electrophoresis. Analytical Biochemistry, 2015. 485: p. 11-17.
239. Luque-Garcia, J.L. and T.A. Neubert, Sample preparation for serum/plasma profiling and biomarker identification by mass spectrometry. Journal of Chromatography a, 2007. 1153(1-2): p. 259-276.
240. Hajduk, J., J. Matysiak, and Z.J. Kokot, Challenges in biomarker discovery with MALDI-TOF MS. Clinica Chimica Acta, 2016. 458: p. 84-98.
241. Wang, C.-C., et al., Critical factors determining the quantification capability of matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry. Philosophical Transactions of the Royal Society A-mathematical Physical and Engineering Sciences, 2016. 374(2079): p. 20150371-20150371.
137
242. Qin, S., et al., Identification of Organ-Enriched Protein Biomarkers of Acute Liver Injury by Targeted Quantitative Proteomics of Blood in Acetaminophen- and Carbon-Tetrachloride-Treated Mouse Models and Acetaminophen Overdose Patients. Journal of Proteome Research, 2016. 15(10): p. 3724-3740.
243. Matsuda, F., A. Tomita, and H. Shimizu, Prediction of Hopeless Peptides Unlikely to be Selected for Targeted Proteome Analysis. Mass Spectrometry, 2017. 6(1).
244. Kiernan, U.A., et al., Detection of novel truncated forms of human serum amyloid A protein in human plasma. Febs Letters, 2003. 537(1-3): p. 166-170.
245. Kim, Y.J., et al., Quantification of SAA1 and SAA2 in lung cancer plasma using the isotype-specific PRM assays. Proteomics, 2015. 15(18): p. 3116-3125.
138
Supplementary
MALDI-TOF mass spectra from all excised 2DE gel spots analyzed.
Spot1: Inconslusive
Spot2: SAA2
139
Spot3: SAA1 or SAA4 (notes not clear = inconclusive)
Spot4: SAA2
140
Spot5: SAA2
Spot 6: SAA4
141
Spot 7: Hemoglobin α
Spot 8: inconclusive
142
Spot 9: inconclusive
Spot 10: inconclusive
143
Spot 11: inconclusive
Spot 12: inconclusive