principles of the mitochondrial fusion and fission cycle...

40
1 Principles of the mitochondrial fusion and fission cycle in neurons Michal Cagalinec 1 , Dzhamilja Safiulina 1 , Mailis Liiv 1 , Joanna Liiv 1 , Vinay Choubey 1 , Przemyslaw Wareski 1 , Vladimir Veksler 1,2,3 , Allen Kaasik 1 1 Department of Pharmacology, Centre of Excellence for Translational Medicine, University of Tartu, Ravila 19, Tartu, Estonia, 2 INSERM, U-769, Châtenay-Malabry F-92296, France, 3 Univ Paris-Sud, Châtenay-Malabry F-92296, France. Running title: Principles of the mitochondrial fusion and fission cycle To whom correspondence should be addressed: Allen Kaasik, Department of Pharmacology, Centre of Excellence for Translational Medicine, University of Tartu, Ravila 19, Tartu, Estonia. Tel: 372 7374361; Fax: 372 7374352; Email: [email protected] © 2012. Published by The Company of Biologists Ltd. Journal of Cell Science Accepted manuscript JCS Advance Online Article. Posted on 22 March 2013

Upload: lamhanh

Post on 25-May-2018

216 views

Category:

Documents


2 download

TRANSCRIPT

1

Principles of the mitochondrial fusion and fission cycle in neurons

Michal Cagalinec1, Dzhamilja Safiulina1, Mailis Liiv1, Joanna Liiv1, Vinay Choubey1,

Przemyslaw Wareski1, Vladimir Veksler1,2,3, Allen Kaasik1

1Department of Pharmacology, Centre of Excellence for Translational Medicine, University of

Tartu, Ravila 19, Tartu, Estonia,

2INSERM, U-769, Châtenay-Malabry F-92296, France,

3Univ Paris-Sud, Châtenay-Malabry F-92296, France.

Running title: Principles of the mitochondrial fusion and fission cycle

To whom correspondence should be addressed: Allen Kaasik, Department of Pharmacology,

Centre of Excellence for Translational Medicine, University of Tartu, Ravila 19, Tartu, Estonia.

Tel: 372 7374361; Fax: 372 7374352; Email: [email protected]

© 2012. Published by The Company of Biologists Ltd.Jo

urna

l of C

ell S

cien

ceA

ccep

ted

man

uscr

ipt

JCS Advance Online Article. Posted on 22 March 2013

2

Abstract

Mitochondrial fusion-fission dynamics play a crucial role in many important cell processes.

These dynamics control mitochondrial morphology, which in turn influences several important

mitochondrial properties including mitochondrial bioenergetics and quality control, and it

appears to be affected in several neurodegenerative diseases. However, an integrated and

quantitative understanding of how fusion-fission dynamics controls mitochondrial morphology

has not yet been described. Here, we took advantage of modern visualisation techniques to

provide a clear explanation of how fusion and fission correlate with mitochondrial length and

motility in neurons.

Our main findings demonstrate that: 1) the probability of a single mitochondrion fissing

is determined by its length; 2) the probability of a single mitochondrion fusing is determined

primarily by its motility; 3) the fusion and fission cycle is driven by changes in mitochondrial

length and deviations from this cycle serves as a corrective mechanism to avoid extreme

mitochondrial length; 4) impaired mitochondrial motility in neurons overexpressing 120Q Htt or

Tau suppresses mitochondrial fusion and leads to mitochondrial shortening whereas stimulation

of mitochondrial motility by overexpressing Miro-1 restores mitochondrial fusion rates and

sizes.

Taken together, our results provide a novel insight into the complex crosstalk between

different processes involved in mitochondrial dynamics. This knowledge will increase

understanding of the dynamic mitochondrial functions in cells and in particular, the pathogenesis

of mitochondrial-related neurodegenerative diseases.

Keywords: mitochondrial dynamics, mitochondrial fusion, neurons, neurodegeneration.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

3

INTRODUCTION

Mitochondria are dynamic organelles that constantly fuse with each other and then split apart

(i.e., undergo fission). Fusion serves to mix and unify the mitochondrial compartment whereas

fission generates morphologically and functionally distinct organelles.

The balance of these two processes determines organelle shape, size and number and is

critical for organelle distribution and bioenergetics. The latter is particularly important in

neurons, which have a unique bioenergetic profile due to their dependence upon energy from

mitochondria and their specialised, compartmentalised energy needs. Beyond the control of

morphology, the mitochondrial fusion-fission cycle appears to be also critical in regulating cell

death and mitophagy. Mitochondrial fission contributes to quality control by favouring removal

of damaged mitochondria via mitophagy (Gomes, et al., 2011) and may facilitate apoptosis in

conditions of cellular stress (Suen, et al., 2008). Failure of mitochondrial fusion-fission dynamics

has been linked to several diseases. Perturbations of the mitochondrial fusion-fission cycle cause

autosomal dominant optic atrophy and Charcot-Marie-Tooth type 2A and appear to be involved

in the pathogenesis of several neurodegenerative diseases (Westermann, 2010).

The molecular mechanisms underlying mitochondrial fusion and fission events are

relatively well-known. The key molecules responsible for fusion are Mitofusin-1 and -2 (Mfn1

and 2, respectively), which are located in the outer mitochondrial membrane, and OPA1, which

is located in the inner mitochondrial membrane. The outer membrane protein Fis1 and the

cytoplasmic protein Drp1 are responsible for fission events (for a review, see Knott et al., 2008).

Recent papers have also revealed interplay between fusion and fission; in most cases, they

appear to be sequential, cycle-forming events (Twig et al., 2008; Wang 2012). Nevertheless,

despite these advances, there remains no clear understanding of how mitochondrial fusion and

fission events interact at the cellular systems level. If the fusion and fission events are cyclic then

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

4

how do they sense each other? How do the fusion or fission events sense mitochondrial size? Do

these events have feedback mechanisms to maintain the size of the mitochondrial population? To

what extent do they depend upon mitochondrial motility or can they regulate mitochondrial

motility themselves? Perhaps most relevantly, to what extent is this network of events affected in

neurons known to be most vulnerable to mitochondrial impairment?

To answer to these questions, using living neurons, we examined the dynamics in

morphology of mitochondrial populations together with fusion and fission characteristics. We

reveal key rules governing mitochondrial dynamics and show the role that mitochondrial length

and motility play in the feedback that regulates mitochondrial homogeneity. Finally, we show

that in models of neurodegenerative diseases, we can rescue an impaired mitochondrial

phenotype by correcting mitochondrial motility.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

5

MATERIALS AND METHODS

Neuronal culture. Primary cultures of rat cortical cells were prepared from neonatal Wistar rats.

Briefly, cortices were dissected in ice-cold Krebs-Ringer solution (135 mM NaCl, 5 mM KCl, 1

mM MgSO4, 0.4 mM K2HPO2, 15 mM glucose and 20 mM HEPES, pH = 7.4) containing 0.3%

BSA and then trypsinised in 0.8% trypsin for 10 min at 37°C. The cells were then triturated in a

0.008% DNase solution containing 0.05% soybean trypsin inhibitor. Cells were resuspended in

Basal Medium Eagle with Earle’s Salts (BME) containing 10% heat-inactivated FBS, 25 mM

KCl, 2 mM glutamine, and 100 µg/ml gentamicin and then plated onto 35 mm glass-bottom

dishes (MatTek, MA, USA), which were pre-coated with poly-L-lysine, at a density of

approximately 106 cells/ml (2 ml of cell suspension per dish). After incubating for 3 h, the

medium was changed to NeurobasalTMA medium containing B-27 supplement, 2 mM

GlutaMAXTM-I, and 100 µg/ml gentamicin. Over 60% of the cells showed a neuronal

phenotype when the described procedure was followed.

To prepare primary cultures of cerebellar granule cells, the cerebella from 8 day-old

Wistar rats were dissociated by trypsinising in 0.25% trypsin at 35°C for 15 min followed by

trituration in a 0.004% DNase solution containing 0.05% soybean trypsin inhibitor. Cells were

resuspended in BME containing 10% FBS, 25 mM KCl, 2 mM glutamine, and 100 µg/ml

gentamicin. Neurons were plated onto 35 mm glass-bottom dishes that were pre-coated with

poly-L-lysine at a density of 1.3 x 106 cells/ml. Ten µM cytosine arabinoside was added 24 h

after plating to prevent the proliferation of glial cells.

The PC12 cell line was maintained in RPMI medium supplemented with 10% horse

serum and 5% FBS on collagen IV (Sigma) -coated plastic dishes. All of the culture media and

supplements were obtained from Invitrogen (Carlsbad, CA, USA).

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

6

Plasmids. Mitochondrial-KikGR1 (mito-KikGR1) was constructed by PCR amplification of the

mitochondrial targeting signal from pdsRed2-Mito (Human COXVIIIa, Nucleotides 597-683)

and cloning in-frame between the EcoRI and HindIII sites of the Kikume vector, which was

obtained from Amalgaam Co. (Tokyo, Japan). The plasmid expressing mito-CFP was obtained

from Evrogen (Moscow, Russia), and mitochondrial-pDsRed2 was obtained from Clontech (CA,

USA). Plasmids expressing shRNA targeted against Drp1, Miro1, and Miro2 were from

SABiosciences (Frederick, MD, USA). Wild-type (wt) and dominant negative Mfn2ΔTM were

generous gifts from Dr. S. Hirose (Honda et al., 2005), Drp1 was from Dr. G. Szabadkai

(Szabadkai et al., 2004), Fis1 was from Dr. J-C. Martinou (Mattenberger et al., 2003), Map2c-

EGFP was from Dr. A. Matus (Kaech et al., 1996), Miro1 was from Dr. P. Aspenström

(Fransson et al., 2003), Tau was from Dr. G. Johnson (Krishnamurthy and Johnson, 2004), and

mutant Htt was from Dr. L. Hasholt (Hasholt et al., 2003).

Transfections. Cultures were transiently transfected on the 2nd day after plating using

LipofectamineTM 2000 (Invitrogen). Briefly, conditioned medium was collected and 120 µl of

Opti-MEM® I medium containing 2% LipofectamineTM 2000 and 1 µg of total DNA (1 µg of

mito-KikumeGR1 in the control; 0.33 µg of mito-KikumeGR1 and 0.67 µg of the desired DNA

in the other groups) were added directly to cells grown on 35 mm glass-bottom dishes, and the

cells were then incubated for 3 h at 37°C in a humidified atmosphere containing 5% CO2/95%

air. At the end of this incubation, 2 ml of fresh (or conditioned medium for granular neurons)

NeurobasalTM medium was added per dish. The cells were then cultured for 3-4 d to enable

expression of the transfected DNA with the exception of 120Q Htt (4-5 d), Tau wt (5-6 d), and

Miro shRNA (8 d). During acquisition, the cells were maintained in Krebs-Ringer solution

supplemented with calcium (1 mM) and glucose (15 mM).

Immunohistochemistry. Neurons were fixed using 4% paraformaldehyde solution in

NeurobasalTM-A containing 5% sucrose for 10 min at 37°C. Fixed cells were permeabilised

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

7

using 0.3% Triton in PBS for 5 min and then blocked using 10% normal goat serum and 3%

BSA for 60 min at room temperature. The neurons were then incubated with the primary

antibodies mouse anti-FLAG (1:500, F1804, Sigma-Aldrich), rabbit anti-Mfn2 (1:100, ab50838,

Abcam, USA), rabbit anti-TTC11 (1:100, ab71498, Abcam), rabbit anti-Myc (1:500, ab9106,

Abcam), mouse anti-DNML (1:250, ab56788, Abcam), rabbit anti-RHOT1 (1:50, HPA010687,

Sigma-Aldrich, Germany), mouse anti-DRPLA-35Q (1:10, MW2, Developmental Studies

Hybridoma Bank at the University of Iowa, USA), or mouse anti-Tau (1:50, 5A6,

Developmental Studies Hybridoma Bank) in the presence of 3% normal goat serum at 4°C for 48

h. After washing, the cells were further incubated with respective Alexa 488- or 594-conjugated

secondary antibodies at room temperature for 1 h and subsequently examined using confocal

microscopy.

Western blotting. For western blotting, cells were lysed in a buffer containing 50 mM Tris-Cl, 1

mM EDTA, 150 mM NaCl, 1% NP-40, 1 mM Na3VO4, 1 mM NaF, 0.25% sodium

deoxycholate, and 5% protease inhibitor cocktail (Roche) for 30 min on ice. Equivalent amounts

of total protein were separated by SDS-PAGE on 10% polyacrylamide gels and then transferred

to Hybond-P PVDF transfer membranes (Amersham Biosciences, UK) in 0.1 M Tris-base, 0.192

M glycine, and 10% (v⁄v) methanol. The membranes were blocked with 5% (w⁄v) non-fat dried

milk in TBS containing 0.1% (v⁄v) Tween-20 at room temperature for 1 h and then probed

overnight with mouse monoclonal anti-β-actin (1:4000, Sigma), mouse monoclonal anti-DNM1L

(1:2000, Abcam), mouse monoclonal anti-RHOT1 (1:2000, Abcam), or rabbit polyclonal anti-

RHOT2 (1:1000, ProteinTech Group Inc.). The membranes were then incubated with appropriate

HRP-conjugated secondary antibodies (1:4000, Pierce, USA) for 1 h at room temperature.

Immunoreactive bands were detected by enhanced chemiluminescence (ECL, Amersham

Biosciences, UK) using medical x-ray film blue (Agfa, Belgium). The probed blots were

analysed by densitometry using MicroImage software (Media Cybernetics, Bethesda, MD).

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

8

Image acquisition. A laser scanning confocal microscope (LSM 510 Duo, Zeiss) equipped with a

LCI Plan-Neofluar 63x/1.3 immersion-corrected DIC M27 objective was used in this study. The

temperature was maintained at 37°C using a climate chamber. Mitochondrial fusion and fission

events were followed using photoconvertable mitochondrial-targeted mito-KikGR1.

Mitochondria labelled with mito-KikGR1 were first visualised using a 488-nm Argon laser line

and BP 505-550 emission filters. Selected green-emitting mitochondria were then

photoconverted to red using a 405-nm Diode laser (50 mW, 1.5% power, 20 iterations). For the

images, which typically comprised 10 - 20 mitochondria per neurite, four to five mitochondria

were photoactivated using two separate bleaching regions to facilitate detection of fusion events.

All mitochondria were then illuminated using the 488 nm Argon laser (for green mitochondria;

30 mW, 1.5% power) and a 561-nm DPSS laser (for red mitochondria; 15 mW, 7% power).

Images were acquired simultaneously to avoid movement distortions during scanning using a BP

505-550 and LP575 filters to separate green- and red-emitting mitochondria, respectively.

For cycle analysis, images were taken every 10 s for 2 h. To study the fusion events, fast

time-lapse experiments were conducted where images taken in ~250 ms intervals were acquired

over 10 mins. To compare the different transfection conditions, images were taken at 10-s

intervals for 10 min. For colocalisation studies of mito- KikGR1 and mito-CFP, mito-CFP

fluorescence after excitation with a 458-nm Argon laser line (1% power) was split using HFT458

and then separated using a BP465-510 filter.

Image processing and analysis. The fate of all photoactivated mitochondria was followed

throughout the time-lapse and the fusion and fission events recorded. Changes in mitochondrial

lengths were measured using LSM5 Duo version 4.2 software (Carl Zeiss MicroImaging Gmbh,

and EMBL Heidelberg, Germany). The resulting database was then used to analyse fusion and

fission cycle parameters, length dependency, and for twin analysis. Mitochondria were further

tracked for motility analysis using Retrack version 2.10.05 (freeware provided by Nick Carter).

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

9

In experiments that compared different conditions, 4 fields per dish were imaged and 4

dishes per condition were used, and all experiments were performed in at least duplicate. Thus,

the mitochondrial population presented always originates from at least 32 different neurons. The

number of fusions, fissions, and contacts with other mitochondria of an photoactivated

mitochondrion was summarised per dish and then averaged over 8-24 dishes. To compare the

mitochondrial velocities, 10-20 mitochondria per neurite (including non-activated mitochondria)

were tracked and the dish medians averaged.

Length analysis was performed using MicroImage software (Media Cybernetics,

Bethesda, MD, USA) and also included other non-activated mitochondria from studied axons.

The presented images and videos were 2D deblurred and deconvoluted using the AutoDeblur and

Autovisualise X software package (Media Cybernetics Inc, Bethesda, MD, USA).

Statistics. The D'Agostino-Pearson omnibus test was used to test for normality. The Mann-

Whitney U test, Student’s t-test, one-way ANOVAs followed by Newman-Keuls posthoc tests or

Kruskal-Wallis tests followed by Dunn’s test were used to compare differences between

experimental and control groups. Correlations were calculated using either the Spearman or

Pearson tests. The Χ2 test was used to determine whether the observed distribution was

significantly different from the expected distribution. P values <0.05 were considered

statistically significant.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

10

RESULTS

Visualisation of fusion and fission in neurons. For visualisation of fusion and fission events, we

used mitochondrially targeted KikGR1 protein. Mitochondrial localisation of this protein was

confirmed by its colocalisation with the well-known mitochondrial markers mito-CFP (Figure

1A) or mito-pDsRED2 (data not shown). Mito-KikGR1 demonstrated a relatively high

mitochondrial/cytoplasmic intensity ratio (> 30). The photoconversion properties of KikGR1

remained unaltered after mitochondrial targeting.

The fusion events between red (photoactivated) and green (non-activated) mitochondria

were easily recognizable because the contacting mitochondria turn yellow after exchanging their

matrices (Figure 1B, Video 1). The exchange of matrix occurred quickly and to completion and

typically, was complete in less than 10 s. Fission events occurred where mitochondria split and

the daughter mitochondria moved away from each other. During the 2 h observation period, we

detected several (up to 9) sequential fusion or fission events of photoactivated mitochondria and

their progenies.

Fusion and fission events are cyclic. Mitochondrial fusion and fission rates were estimated in

axons of two morphologically distinct subtypes of neurons: cortical neurons and cerebellar

granule neurons. In both types of cell, the fusion rate matched the fission rate perfectly although

in cortical neurons, these rates (0.023 ± 0.003 fusions/mitochondria/min and 0.023 ± 0.003

fissions/mitochondria/min; n = 17 dishes) were lower than in cerebellar granule neurons (0.045 ±

0.006 fusions/mitochondria/min, p = 0.001, and 0.039 ± 0.005 fissions/mitochondria/min, p =

0.0072; n = 12 dishes).

We also attempted to measure the mitochondrial fusion rate in the soma of cortical

neurons. However, mitochondria are packed very densely in the soma, and in the majority of

cases, the laser beam photoactivated mitochondria located in the vicinity of the mitochondrion of

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

11

interest. Nevertheless, we estimated the fusion rate in soma of cortical neurons, which was

significantly higher (0.149 ± 0.021 fusion/mitochondria/min, n = 10; p < 0.0001) than in axons.

Our next aim was to determine whether the fusion and fission events occurred randomly

or in a regulated manner (for example, cyclic, as suggested by Twig et al., 2008). Theoretically,

each fusion could be followed by fission or by a secondary fusion, or similarly, each fission

could be followed by fusion or a secondary fission (Figure 1C). Analysis of the combinations of

these events demonstrated that in cortical neurons, a fusion was followed by a fission in 86.4%

of cases and by a second fusion in only 13.6% of the cases. In addition, fission was followed by

fusion in 83.5% of cases and by a second fission in 16.5% of cases. We performed the same

analysis using cerebellar granule neurons and obtained similar results (Figure 1C). These data

suggest that fusion and fission events are sequential events that form a cycle rather than

independent, randomly occurring events. However, an alternating fusion-fission cycle was not

consistently observed, and there were deviations from this rule (fission-fission or fusion-fusion

events) in 15% of the cortical neurons and 24% of the cerebellar granule neurons.

We next measured the duration of the two components of the fusion-fission cycle in

cortical neurons. The mean time interval between fusion and fission was 4.7 ± 0.6 min, which

was significantly shorter than the interval between fission and the next fusion (15.3 ± 2.0 min; p

< 0.0001, Mann-Whitney U test). The duration of the entire cycle was approximately 20 min. It

should be noted that this analysis was only applied to the active subpopulation of mitochondria,

which may explain why the cycle duration found here was shorter than that calculated from the

fusion or fission rates of the total mitochondrial population (approximately 43 min).

Nevertheless, these values suggest that mitochondria spend approximately ¼ of their time in

between fusion and fission events and ¾ of their time in between fission and the next fusion

event.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

12

Rules governing mitochondrial fusion and fission. Fusions almost always occurred between

stationary (or passive) and moving (or active) mitochondria. Indeed, of 34 fusion events

analysed, 28 events (82%) showed clearly recognisable active and passive partners, 3 events

(9%) occurred between two active mitochondria, and 3 events (9%) occurred between two

passive mitochondria.

To further study how the mitochondria are juxtaposed at the beginning of fusion, we

performed fast time-course experiments with 250-ms intervals between frames. This approach

enabled accurate examination of the exact positions of two mitochondria relative to each other.

Although we expected that the head of the active partner would collide with the passive partner

to merge the matrices of each mitochondrion, this movement was not common. Such head-on

fusions were only observed in 2% of all cases (2 of 89 analysed events) whereas in most cases,

the active partner passed the passive partner, which lead to side-to-side fusions (43%) or rear-end

fusions (55%). It should be mentioned that the moving head passed the body of the passive

mitochondrion in 88% of all cases (68 out of 77 analysed events) and remained behind the head

of the passive mitochondrion in only 12% of cases (p < 0.0001, Χ2 test).

This complicated behaviour of fusing mitochondria may be related to the localisation

and/or activity of fusion-controlling proteins. We investigated the distribution of Mfn2 in

mitochondria and found that its localization was indeed focal and non-homogenous and localized

to specific spots (Figure 3A; Karbowski et al., 2006).

The passing process itself was often sophisticated and comprised numerous stops

including nudging, turn-aways, and unexpected returns that ended with fusion. The videos show

two examples (Video 2 and 3) where after the initial contact, the active partner moved away

from the passive one and then turned back to make the final contact. The duration of these

dances was very variable with some interactions being very short and others lasting up to a few

minutes (mean value 55 ± 4 s, n = 171). The dance ended with a firm grip between the partners

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

13

that lasted between 10-80 s (mean value 58 ± 4 s, n = 172) and ended with an exchange of

matrix. The duration of the grip appeared to be controlled by proteins responsible for fusion

because in neurons transfected with a dominant negative Mfn2, this parameter was prolonged

(100 ± 14 s, p = 0.001 compared with control).

When fused mitochondria subsequently underwent fission, in most cases, one of the

resulting daughters remained passive whereas the other moved away. Interestingly, the active

daughter almost always followed the same direction as the active parent had. We analysed 43

such events, and in 41 cases, the active daughter continued in the same direction, and in only 2

cases, the direction was reversed (p < 0.0001, Χ2test). These results suggest that the “moving

head” of active mitochondria is conserved before fusion and moves in the same direction after

fission.

Fission rate is length dependent. Both fusion and fission change mitochondrial length. In

consideration of the alternating nature of these events, it is reasonable to suggest that the

negative feedback(s) controlling the fusion and/or fission rates involves mitochondrial length. To

assess this hypothesis, we performed a “twin study”, where daughter mitochondria originating

from the same parent mitochondrion were compared. We tracked 30 “families” up to the moment

when one of the daughters underwent a second fission. The results demonstrated that the average

length of the daughter undergoing a second fission was greater than twice the length of its

nonfissing twin (Figure 2A). In addition, we followed 42 pairs of twins until one of the daughters

fused. The average length of the fusing daughter matched perfectly the length of the non-fusing

daughter (Figure 2B).

These results suggest that longer mitochondria enter fission more readily. To confirm this

conclusion, we next divided the total population of mitochondria in cortical and cerebellar

granule neurons based on their length and determined the fission and fusion rates for each

subgroup. Figure 2C demonstrates that in both neuron types, the fission rate was very low in

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

14

short mitochondria; however, the fission rate increased dramatically in long mitochondria.

Therefore, in contrast to the fission rate, the fusion rate appears to be relatively independent of

length.

Globally, the fusion rate exceeded the fission rate in shorter mitochondria whereas the

fission rate exceeded the fusion rate in longer mitochondria. Thus, the fusion-fission balance

shifted towards fusion in shorter organelles and towards fission in longer organelles. Therefore, a

shorter mitochondrial length decreases the probability of fission being the next event and

consequently increases the probability of fusion (Supplementary Table 1). This finding suggests

that the mitochondria of greatest length after fission are more likely to undergo secondary,

“corrective” fission. In contrast, post-fusion mitochondria that are too short are more likely to

undergo secondary fusion. We analysed particular cases to determine whether these “errors”

indeed correlated with mitochondrial size. The results depicted in Figure 3D and E show that

post-fission mitochondria that entered an “erroneous” cycle-breaking second fission were

significantly longer than “normal” mitochondria that entered fusion. In agreement with this

observation, post-fusion mitochondria that entered an “erroneous” second fusion were

significantly shorter that “normal” mitochondria that entered fission. Thus, this type of feedback

mechanism may serve as a quality control mechanism to correct mitochondria that are

excessively short or long.

The next important question concerned the mechanism underlying the fission rate-length

relationship. This relationship was quite steep (Figure 2C), and its non-linearity suggested a

cooperative interaction between factors/molecules responsible for fission provided that the

number of these molecules increased with mitochondrial length. To determine the putative role

of Drp1 in the fission rate-length relationship, we modified Drp1 expression in cortical neurons.

As expected, overexpression of Drp1 together with Fis1 led to mitochondrial shortening whereas

Drp1 silencing was associated with dramatic mitochondrial elongation (Figure 3). Interestingly,

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

15

neither Drp1 silencing nor Drp1 and Fis1 overexpression led to consistent changes in fission or

fusion rates. Analysis of the fission rate-length relationship demonstrated that Drp1 and Fis1

overexpression increased the sensitivity of the fission rate to mitochondrial length whereas Drp1

silencing (using specific shRNAs) led to a dramatic drop in this sensitivity (Figure 4A). The role

of Drp1 was specific because no change was observed following the overexpression of wild-type

(wt) Mfn2 or dominant negative Mfn2.

Impairment of the feedback mechanism in Drp1-suppressed neurons was associated with

high variability in mitochondrial length. The coefficient of variation for length was considerably

higher in Drp1 shRNA-expressing neurons compared with control (Figure 4B). Importantly,

Mfn2 overexpression-mediated elongation of mitochondria did not affect length variability.

Together, these results suggest that the length-fission feedback loop is Drp1-dependent.

Fission rate adapts itself to the fusion rate. The efficiency of this feedback control mechanism

was assessed in experiments that studied the adaptation of the fission rate to a switched fusion

rate and vice versa. We first overexpressed wt Mfn2 and found that it induced a 76% increase in

mitochondrial fusion rate (Figure 3E). This finding was accompanied by a 68% increase in

fission rate but a relatively modest 36% increase in mitochondrial length. Because Mfn2 has also

been implicated in mitochondrial transport (Misko et al., 2010), we also performed a separate

experiment that showed that overexpression of Mfn1 induced similar effects to those of Mfn2

(Figure 3E). Inversely, inhibition of the mitochondrial fusion rate by overexpressing dominant

negative Mfn2 decreased the fusion rate to half that of the control level. This effect was

accompanied by almost perfectly matched changes in the fission rate and a 29% decrease in

mitochondrial length. These results demonstrate that the fission rate adapts to the fusion rate; i.e.,

augmented or decreased fusion rates favour the formation of longer or shorter mitochondria, that

are either more or less likely to enter fission, respectively.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

16

Fusion depends on the contact rate. Fusion can only take place when two mitochondria meet.

However, not every contact results in fusion. In cortical neurons, of 1187 analysed mitochondria

from 66 separate fields, the axonal mitochondria made an average of 0.45 ± 0.02 contacts per

mitochondrion per min, and only 7.0 ± 0.4% resulted in fusion. However, in cerebellar neurons,

axonal mitochondrial made an average of 0.24 ± 0.03 contacts per mitochondrion, of which 20%

resulted in fusion. Higher fusion efficiency in cerebellar neurons may be due to the 1.65-fold

increase in Mfn2 expression (p = 0.014) that was observed in these cells compared with cortical

neurons.

We measured the average number of contacts made by mitochondria and the number of

fusions in a further 232 neurites. Not surprisingly, the number of fusions correlated well with the

number of contacts (Spearman r = 0.33, p < 0.0001), and a higher number of contacts increased

the likelihood of fusion.

Apparently, the number of contacts depends upon the number of partners available, and

the probability of a mitochondrion meeting another is higher in crowded neurites. We compared

the contact rate in different axonal regions with different mitochondrial densities. As expected,

an increase in the density of mitochondria was associated with an increase in contact rate

(Spearman r = 0.144, p = 0.020).

Motility determines fusion rate. It is clear that mitochondrial partners must approach each other

for a contact to occur, and therefore, a contact should depend upon movement characteristics.

Therefore, we measured the motility of individual mitochondria and the number of contacts they

made. As expected, increased motility was associated with an increase in contact rate (Spearman

r = 0.482, p < 0.0001). To confirm the relationship between the motility and fusion rate, we

analysed the velocity of twin mitochondria where only one of the daughters fused. The results

showed that the fusing daughter had a significantly higher velocity than its non-fusing twin

(Figure 5A). In additional experiments, we inhibited mitochondrial motility by suppressing Miro

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

17

proteins, which are responsible for mitochondrial attachment to molecular motors. Suppression

of these proteins led to a proportional inhibition of mitochondrial velocity and contact rate,

which in turn was associated with a reduction in fusion rate (Figure 5B-F). These results showed

that motility, at least in part, determines the rate of fusion (Figure 5B-F). Inhibition of

mitochondrial motility by Miro shRNA also inhibited fusion and decreased the average

mitochondrial length (Figure 5G), an effect that was similar to the effect of dominant negative

Mfn2. Similarly, overexpression of the axonal docking protein syntaphilin inhibited

mitochondrial motility and fusion rate (data not shown).

Mitochondrial motility during the fusion-fission cycle. Considering that mitochondrial velocity is

an important determinant of mitochondrial fusion, we next studied the velocity of single

mitochondria during the course of a fusion-fission cycle. We first traced the velocity of those

mitochondria that later entered fusion. Figure 6A and B show that at approximately 10 min

before a fusion event, both the active and passive partners had the same mean low velocity.

However, later, the velocity of the active partner increased progressively up to the moment of

fusion whereas no change was observed in the velocity of the passive partner. Remarkably, the

fusion event induced a dramatic drop in mitochondrial motility, and the fusion product was

relatively immobile compared with the active parent. In contrast to the pre-fusion state, no

change in mitochondrial velocity was observed in the pre-fission state. Interestingly, after

fission, the velocity of the active daughter that moved away was similar to that of the active

partner prior to fusion (Spearman r = 0.425 and p = 0.0062). However, the high velocity slowed

rapidly but was nevertheless higher than the velocity of the second daughter for a period of time

(Figure 6C and D). Thus, the fusion-fission cycle is associated with clear changes in the velocity

of individual mitochondria.

Fusion-fission dynamics in models of neurodegenerative diseases. Recent reports demonstrate

that overexpression of mutant Huntingtin (Trushina et al., 2004; Chang et al., 2006; Song et al.,

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

18

2011) and Tau (Ebneth et al. 1998) impair mitochondrial dynamics. To test whether the fusion

and fission dynamics are altered in these disease models, we first overexpressed the 1st exon of

mutant Huntingtin, containing 120 polyglutamines (120Q-Htt), in cortical neurons. As expected,

expression of this construct led to a decrease (by 40 ± 7%) in average mitochondrial velocity and

a concomitant reduction (by 35 ± 9%) in the number of contacts between mitochondria.

Importantly, these changes were associated with decrease in fusion and fission rates (by 57 ± 6%

and 46 ± 10%, respectively). Similar to the reduced motility induced by suppression of Miro

protein, the reduced fusion rate was related to mitochondrial shortening (Figure 7).

To test the causal relationship between reduced motility and fragmentation, we next tried

to increase the motility in 120Q-Htt-expressing neurons using Miro-1. The results depicted in

Figure 7 demonstrate that Miro-1 overexpression restored mitochondrial velocity completely in

addition to the rate of mitochondrial contacts. These changes were followed by a recovery in

fusion rate and an amelioration of the mitochondrial fragmented phenotype.

Finally, we overexpressed Tau protein, which is also known to inhibit mitochondrial

motility in neurons. Similar to the results obtained following 120Q-Htt overexpression, Tau

overexpression reduced mitochondrial motility and led to a decreased contact rate. It was also

associated with a decrease in fusion rate that led to mitochondrial shortening. The co-expression

of Miro-1 and Tau improved mitochondrial motility, contact, and fusion rates and rescued the

mitochondrial phenotypes (Figure 7).

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

19

DISCUSSION

Mitochondrial fission and fusion are often viewed as being in a finely-tuned balance within cells;

however, an integrated and quantitative understanding of how these processes interact with each

other and with mitochondrial motility and morphology has yet to be formulated. Our main

findings demonstrate that:

• the probability of a single mitochondrion fissing is determined by its length;

• the probability of a single mitochondrion fusing is determined primarily by its motility;

• the fusion and fission cycle is driven by changes in mitochondrial length – an increase in

mitochondrial length after fusion increases the probability of fission whereas a decrease in

mitochondrial length after fission reduces this probability;

• deviations from the fusion and fission cycle serve as a corrective mechanism to avoid extreme

mitochondrial length;

• impaired mitochondrial motility in neurons overexpressing 120Q Htt or Tau suppresses

mitochondrial fusions and leads to mitochondrial shortening; stimulation of mitochondrial

motility by overexpressing Miro-1 restores mitochondrial fusion rates and sizes under these

conditions.

Self-regulation of mitochondrial length. Our major novel finding was that mitochondrial length

controls the rate of mitochondrial fission. Longer mitochondria enter fission more readily than

shorter mitochondria. For example, a 5-µm mitochondrion showed a 10-fold higher fission rate

than a 2-µm mitochondrion. Similar results were also observed in our “twin study“ that

minimised the involvement of other factors, such as different DNA or protein composition, or

cytoplasmic environment, which may affect the fission rate independent of length. Regardless of

the molecular mechanism underlying the observed length-fission relationship, this phenomenon

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

20

appears to control mitochondrial length in neurons and also appears to be the main feedback tool

enabling neurons to sense and correct mitochondrial length.

This type of control also explains why the fission rate is related to the fusion rate such

that these two rates are balanced perfectly. None of the interventions used in our experiments

changed the fusion rate/fission rate ratio significantly. On average, mitochondria in axons of

cortical neurons underwent fusion or fission 33 times per day. The slightest imbalance (one

fusion less or more per day) should therefore eventually lead to complete mitochondrial

fragmentation or the formation of a few megamitochondria in the cell; however, this was not the

case. Overexpression of wt Mfn2 doubled the fusion and fission rates and led to only a slight

increase in mitochondrial length. It is most likely that increased fusion activity induced a slight

elongation in mitochondria, which in turn activated the fission machinery that attempted to block

a further increase in mitochondrial length.

In addition, these findings provide a clear explanation as to why manipulations of fission

machinery failed to change the fission rate. For example, Drp1 upregulation did not increase the

rate of fusion although it did lead to mitochondrial shortening. An attempt to increase the fission

rate and thereby shorten mitochondria is likely to be counterbalanced by an inhibition of the

fission machinery by that same mitochondrial shortening. Suppression of the fission rate by

inhibition of Drp1 induced the opposite effect, which leads to mitochondrial elongation that in

turn activates fission machinery. Mitochondrial length is thus self-regulated via Drp1-dependent

fission.

Fusion rate is determined by the availability of partners – “it takes two to tango”. Fusion can

only take place when two mitochondria meet. The number of contacts a mitochondrion makes

depends on its motility and the number of mitochondria in the vicinity. Higher motility increases

the likelihood of a mitochondrion finding a partner and consequently the fusion rate. In turn,

lower motility decreases the number of contacts and the fusion rate. The latter conclusion is

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

21

consistent with previous findings that inhibition of mitochondrial movement by nocodazol and

vasopressin in H9c2 cells also inhibited mitochondrial fusion (Liu et al., 2009). Similarly, a

recent study by Twig et al. (2010) suggested that mitochondrial motility facilitates mitochondrial

fusion in H9c2 and INS1 cells. Our results support this hypothesis with quantitative data

demonstrating that mitochondria entering fusion show two-fold higher motility compared with

their nonfusing sisters. Moreover, our results demonstrate that direct slowing of mitochondrial

motility by suppressing mitochondrial Miro proteins (required for mitochondrial antero- and

retrograde transport (Russo et al., 2009)) or by overexpressing the axonal docking protein

syntaphilin inhibit the rate of mitochondrial fusion.

Obviously, the interaction between mitochondrial motility and fusion/fission dynamics is

rather sophisticated. The velocity of mitochondria that are about to fuse begins to increase

several minutes prior to the fusion and increases progressively to the point of fusion. It is unclear

whether this increase in motility reflects an intrinsic “intention to fuse” or whether the fusion is a

simple consequence of the higher motility. Recent research has demonstrated that Mfn2 interacts

with Miro-2 protein and is required for mitochondrial transport (Misko et al., 2010), which

suggests that the Mfn-Miro interaction is used by mitochondria to inform the transport

machinery of the readiness to fuse.

It is also relevant to note that after the fusion event, the new mitochondrion remains

relatively immobile. This cannot be explained by its increased length because we did not observe

any correlation between mitochondrial length and motility. It is more likely that passive

mitochondria are anchored so strongly that they will also hold the moving mitochondria.

However, after fission, the mitochondria that formed from a previously active segment move

away and slow down. Thus, the velocity of individual mitochondria changes with the phase of

the fusion/fission cycle (Figure 6).

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

22

Another factor that may increase the mitochondrial fusion rate is the number of

surrounding mitochondria. For example, the fusion rate in a neuronal soma, which shows a

higher mitochondrial density, is greater than 5-fold that of axons (where mitochondrial density is

relatively low). This finding could be one of the reasons why the fusion rate changes between

different neuronal compartments and may explain why mitochondria in the soma are 50% (1 µm)

longer than in the periphery. Nevertheless, we cannot rule out the possibility that the increased

fusion rate is also caused by the higher accessibility or activity of the fusion-fission proteins.

The fusion-fission cycle is driven by changes in mitochondrial length. In their paper, Twig et al.

(2008) elegantly demonstrated that fusion and fission, typically, are sequential events that form a

cycle. However, in a recent report, Wang et al. (2012) showed that the fusion and fission events

were not always sequential (35% of events in Hela cells and 40% of the events in MEF cells

were not cyclic). We have shown that in neurons, the fusion and fission events were also not

always sequential: 15% of events in cortical neurons and 25% in cerebellar neurons were not

cyclic. Importantly, the dependency of the fission rate on length discovered in our study provides

a very simple explanation as to why mitochondrial fusion and fission alternate and why these

events are not always cyclic. After each fusion, the mitochondrial length doubles. This doubling

causes an 8-fold increase in the probability of fission and predicts a short lifetime for the post-

fusion mitochondrion. In contrast, fission halves the mitochondrial length and therefore

diminishes the likelihood of a secondary fission and increases the probability of a fusion being

the next event. This mechanism enables alternating fusion and fission events and also explains

why mitochondria spend approximately ¼ of their time in a post-fusion state and ¾ of their time

in a post-fission state. If fusion occasionally leads to the formation of a mitochondrion that is too

short or a fission product that is too long, the cycle has been compromised. Deviations from the

cycle (two consecutive fusions or two consecutive fissions) serve as quality control mechanisms

to correct small or oversized mitochondria. This mechanism may have an important

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

23

physiological relevance in the maintenance of optimal mitochondrial size. Mitochondria that are

too long have been shown to exhibit a compromised bioenergetic capacity (Benard et al., 2007)

and may have difficulties in delivering energy to the cell.

Mitochondrial fusion-fission dynamics in neurodegenerative diseases. Recent reports suggest

that mutant Huntingtin impairs the balance in mitochondrial fission and fusion and thereby

causes neuronal injury. Mutant Huntingtin triggers mitochondrial fragmentation in rat neurons

and fibroblasts from individuals with Huntington's disease (Song et al., 2011). Increases in Drp1

and Fis1 and decreases in Mfn1 and Mfn2 expression have been observed in cortical samples

from HD patients (Shirendeb et al., 2011).

Our experiments demonstrated that mutant Huntingtin impaired mitochondrial fission-

fusion balance by inhibiting fusion activity. These data suggest that mutant Huntingtin may also

induce mitochondrial fragmentation indirectly by inhibiting mitochondrial motility (Trushina et

al., 2004; Chang et al., 2006). This effect could inhibit the number of contacts between

mitochondria and consequently reduce the rate of fusion, which would lead to mitochondrial

shortening. Mutant Huntingtin-mediated mitochondrial fragmentation, defects in the fusion and

fission rates, and a decrease in the number of contacts were all rescued by increasing

mitochondrial motility via the overexpression of Miro1 protein. Interestingly, this effect was not

specific to mutant Huntingtin. Ebneth et al. (1998) reported that overexpression of Tau resulted

in a failure of the cell to transport mitochondria to peripheral compartments, which may be of

relevance to Alzheimer's disease. In our settings, overexpression of wt Tau suppressed

mitochondrial motility and mitochondrial fusion and induced mitochondrial fragmentation.

Similar to mutant Huntingtin, all of these parameters were restored following Miro-1

overexpression. These results suggest that motility likely plays a key role in mitochondrial

fusion-fission dynamics and morphology, and its restoration may constitute a new treatment

option for Huntington's and Alzheimer’s disease.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

24

Together, our results provide a novel insight into the complex crosstalk between

mitochondrial fusion and fission, length, and motility. This knowledge will provide better

understanding of the dynamic mitochondrial function in cellular physiology and the pathogenesis

of mitochondrial-related neuronal diseases.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

25

Acknowledgements. We thank Dr. S. Hirose, Dr. N. Nakamura, Dr. E. Bampton, Dr. J-C.

Martinou, Dr. A. van der Bliek, Dr. G. Szabadkai, Dr. A. Matus, Dr. P. Aspenström, Dr. A.

Fransson, Dr. G. Johnson, and Dr. L. Hasholt for providing the plasmids. We also thank Dr.

Miriam A. Hickey for her assistance with proofreading.

This research was supported by the institutional research founding grant (IUT2-5) from the

Estonian Research Council, grants from the Estonian Science Foundation (grant numbers 7991,

7175, 8810, and 8125), the European Community, the European Regional Development Fund,

and the Estonian-French research program Parrot. M.C. was supported by MOBILITAS

(MJD35).

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

26

References

Benard, G., Bellance, N., James, D., Parrone, P., Fernandez, H., Letellier, T. and Rossignol,

R. (2007). Mitochondrial bioenergetics and structural network organization. J. Cell Sci. 120,

838-848.

Chang, D.T., Rintoul, G.L., Pandipati, S. and Reynolds, I.J. (2006). Mutant huntingtin

aggregates impair mitochondrial movement and trafficking in cortical neurons. Neurobiol. Dis.

22, 388-400.

Ebneth, A., Godemann, R., Stamer, K., Illenberger, S., Trinczek, B. and Mandelkow, E.

(1998). Overexpression of tau protein inhibits kinesin-dependent trafficking of vesicles,

mitochondria, and endoplasmic reticulum: implications for Alzheimer's disease. J. Cell Biol.

143, 777-794.

Fransson, A., Ruusala, A. and Aspenstrom, P. (2003). Atypical Rho GTPases have roles in

mitochondrial homeostasis and apoptosis. J. Biol. Chem. 278, 6495-6502.

Gomes, L.C., Benedetto, G.D. and Scorrano, L. (2011). During autophagy mitochondria

elongate, are spared from degradation and sustain cell viability. Nat. Cell Biol. 13, 589-598.

Hasholt, L., Abell, K., Norremolle, A., Nellemann, C., Fenger, K. and Sørensen S.A. (2003).

Antisense downregulation of mutant huntingtin in a cell model. J. Gene Med. 5,528-538. 446.

Honda, S., Aihara, T., Hontani, M., Okubo, K. and Hirose, S. (2005). Mutational analysis of

action of mitochondrial fusion factor mitofusin-2. J. Cell Sci. 15, 3153-3161.

Kaech, S., Ludin, B. and Matus, A. (1996); Cytoskeletal plasticity in cells expressing neuronal

microtubule-associated proteins. Neuron 17, 1189-1199.

Karbowski, M., Norris, K.L., Cleland, M.M., Jeong, S.Y. and Youle, R.J. (2006). Role of

Bax and Bak in mitochondrial morphogenesis. Nature 443, 658-662.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

27

Knott, A.B., Perkins, G., Schwarzenbacher, R. and Bossy-Wetzel, E. (2008). Mitochondrial

fragmentation in neurodegeneration. Nat. Rev. Neurosci. 9, 505-518.

Krishnamurthy, P.K. and Johnson, G.V. (2004). Mutant (R406W) human tau is

hyperphosphorylated and does not efficiently bind microtubules in a neuronal cortical cell

model. J. Biol. Chem. 279, 7893-7900.

Liu, X., Weaver, D., Shirihai, O. and Hajnoczky, G. (2009). Mitochondrial 'kiss-andrun':

interplay between mitochondrial motility and fusion-fission dynamics. EMBO J. 28, 3074-3089.

Mattenberger, Y., James, D.I. and Martinou, J.C. (2003). Fusion of mitochondria in

mammalian cells is dependent on the mitochondrial inner membrane potential and independent

of microtubules or actin. FEBS Lett. 538, 53-59.

Misko, A., Jiang, S., Wegorzewska, I., Milbrandt, J. and Baloh, R.H. (2010). Mitofusin 2 is

necessary for transport of axonal mitochondria and interacts with the Miro/Milton complex. J.

Neurosci. 30, 4232-4240.

Russo, G.J., Louie, K., Wellington, A., Macleod, G.T., Hu, F., Panchumarthi, S. and

Zinsmaier, K.E. (2009). Drosophila Miro is required for both anterograde and retrograde axonal

mitochondrial transport. J. Neurosci. 29, 5443-5455.

Szabadkai, G., Simoni, A.M., Chami, M., Wieckowski, M.R., Youle, R.J. and Rizzuto, R.

(2004). Drp-1-dependent division of the mitochondrial network blocks intraorganellar Ca2+

waves and protects against Ca2+-mediated apoptosis. Mol. Cell. 16, 59-68.

Shirendeb, U., Reddy, A.P., Manczak, M., Calkins, M.J., Mao, P., Tagle, D.A. and Reddy

P.H. (2011). Abnormal mitochondrial dynamics, mitochondrial loss and mutant huntingtin

oligomers in Huntington's disease: implications for selective neuronal damage. Hum. Mol. Genet.

20, 1438-1455.

Song, W., Chen, J., Petrilli, A., Liot, G., Klinglmayr, E., Zhou, Y., Poquiz, P., Tjong, J.,

Pouladi, M.A., Hayden, M.R., Masliah, E., Ellisman, M., Rouiller, I., Schwarzenbacher, R.,

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

28

Bossy, B., Perkins, G. and Bossy-Wetzel, E. (2011). Mutant huntingtin binds the mitochondrial

fission GTPase dynamin-related protein-1 and increases its enzymatic activity. Nat. Med. 17,

377-382.

Suen, D.F., Norris, K.L. and Youle, R.J. (2008). Mitochondrial dynamics and apoptosis. Genes

Dev. 22, 1577-1590.

Trushina, E., Dyer, R.B., Badger, J.D. 2nd, Ure, D., Eide, L., Tran, D.D., Vrieze, B.T.,

Legendre-Guillemin, V., McPherson, P.S., Mandavilli, B.S., Van Houten, B., Zeitlin, S.,

McNiven, M., Aebersold, R., Hayden, M., Parisi, J.E., Seeberg, E., Dragatsis, I., Doyle, K.,

Bender, A., Chacko, C. and McMurray, C.T. (2004). Mutant huntingtin impairs axonal

trafficking in mammalian neurons in vivo and in vitro. Mol. Cell. Biol. 24, 8195-8209.

Twig, G., Elorza, A., Molina, A.J., Mohamed, H., Wikstrom, J.D., Walzer, G., Stiles, L.,

Haigh, S.E., Katz, S., Las, G., Alroy, J., Wu, M., Py, B.F., Yuan, J., Deeney, J.T., Corkey,

B.E. and Shirihai, O.S. (2008). Fission and selective fusion govern mitochondrial segregation

and elimination by autophagy. EMBO J. 27, 433-446

Twig, G., Liu, X., Liesa, M., Wikstrom, J.D., Molina, A.J., Las, G., Yaniv, G., Hajnóczky,

G. and Shirihai, O.S. (2010). Biophysical properties of mitochondrial fusion events in

pancreatic beta-cells and cardiac cells unravel potential control mechanisms of its selectivity.

Am. J. Physiol. Cell. Physiol. 299, C477-487.

Wang, S., Xiao, W., Shan, S., Jiang, C., Chen, M., Zhang, Y., Lu, S., Chen, J., Zhang, C.,

Chen, Q. and Long, M. (2012); Multi-patterned dynamics of mitochondrial fission and fusion in

a living cell. PLoS One 7:e19879.

Westermann, B. (2010) Mitochondrial fusion and fission in cell life and death. Nat Rev Mol.

Cell. Biol. 11, 872-884.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

29

LEGENDS TO FIGURES

Figure 1. Fusion and fission are sequential events. A. Photoactivation of mito-Kikume-Green.

Mitochondria expressing Kikume-Green were co-transfected with mito-CFP and demonstrated

clear colocalisation of both proteins (left panels). Selected mitochondria were irradiated using a

405-nm laser line (marked with a yellow rectangle) and mito-Kikume-Green was converted into

mito-Kikume-Red. Note that the neighbouring mitochondria retained their green fluorescence. B.

A fusion event between mito-Kikume-Green and photoactivated, mito-Kikume-Red

mitochondria. The fusion product became yellow after mixing of the contents of the red and

green mitochondrial matrices (see also Figure S1). C. Each fusion (left) was followed by a

fission or by a secondary fusion, and each fission (right) was followed by a fusion or a secondary

fission. To determine the extent of fusion and fission events that were sequential, we analysed

the number of event pairs (111 event pairs starting with fusion and 103 starting with fission in

cortical neurons, and 63 pairs starting with fusion and 55 starting with fission in cerebellar

granule neurons). The figure shows the number of each type of event. A Χ2 test was used to

determine whether the observed distribution was significantly different from the expected

distribution.

Figure 2. Mitochondrial fission is length dependent. A,B. Mitochondrial twin analysis.

Comparison of mitochondrial length within families, where one of the daughters underwent

fission (A) or fusion (B) after the primary fission (n = 30 and 45, respectively; the Wilcoxon test

was used to calculate the p value). C. Length dependency of mitochondrial fusion (black circles)

and fission (red circles) rates. Mitochondria from cortical neurons (1277) and cerebellar granule

neurons (413) were sub-grouped based on their lengths, and the fission and fusion rates were

determined for each sub-group. D. Length analysis of the post-fission mitochondria entering

either fusion (normal cycle, n = 86) or fission (erroneous, cycle-breaking, n = 17, Mann-Whitney

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

30

test). E. Length analysis of the post-fusion mitochondria entering either a fission (normal cycle,

n = 96) or fusion event (erroneous, cycle-breaking, n = 15, Mann-Whitney test).

Figure 3. Manipulation of fusion proteins but not fission proteins alters the fusion-fission

dynamics. A. Overexpressed Mfn-FLAG (detected using an anti-FLAG antibody, green) is

located in the ends of mito-DsRed-expressing mitochondria. B. Expression of a specific shRNA-

encoding plasmid suppressed Drp1 levels in PC12 cells by 93%. C. Overexpressed Drp1 (anti-

Dnml) demonstrates classical cytoskeletal localisation. D. Overexpressed Fis1 (anti-TTC11,

green) was homogenously distributed throughout the mitochondrial membrane (mito-DsRed). E.

Fusion and fission rates and lengths of mitochondria in control, wild-type (wt) Mfn2-, dominant

negative (dn) Mfn2-, wt Mfn1-, wt Drp1 + wt Fis 1-overexpressing and Drp1-suppressed

neurons (see also Figure S2). Data are shown as the mean ± SEM. *p < 0.05, **p < 0.01 and ***

p < 0.001 vs. control. The number of dishes analysed (for fusion and fission rate) or the number

of mitochondria analysed (for length) is shown in brackets.

Figure 4. Knockdown of Drp1 abolishes the length dependency of mitochondrial fission. A.

The length dependency of mitochondrial fission in Drp1 shRNA- (red line), Drp1- and Fis 1-

expressing (black line), and control (dotted line) neurons. Mitochondria (513) from the Drp1

shRNA group and Drp1 and Fis 1 (531) groups were analysed. B. The average coefficient of

variability of mitochondrial length in control, Drp shRNA-, and Mfn2-overexpressing neurons.

Coefficients were calculated for individual neurites (27 in control and 12 in other groups) and

then averaged. *** p < 0.001 vs. control.

Figure 5. Fusion is correlated with mitochondrial motility. A. Motility analysis of twin

mitochondria from families where one of the daughters fused during the observation time.

Columns show the mean mitochondrial motility for fusing and non-fusing daughters. B. Both

Miro-1 shRNA (left) and Miro-2 shRNA (right) effectively suppressed the expression of

endogenous Miro-s. C. Kymograms derived from videos obtained from control and Miro-

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

31

silenced neurons. The upper panels show the starting point and the lower panels show the

movement of mitochondria (red or green) over the following 10 min. A fusion event led to the

formation of a yellow signal, and the movement of fission products is designated by the dotted

line. D,E. Parameters of mitochondrial dynamics and morphology in control and Miro-silenced

neurons. Data are shown as the mean ± SEM. *p < 0.05, **p < 0.01, and *** p < 0.001 vs.

control. See also Figure S3.

Figure 6. Mitochondrial motility changes at different stages of the mitochondrial fusion

and fission cycle. A. Mitochondrial velocity starting at 10 min prior to the fusion event up to 10

min after the fusion event. B. The mean mitochondrial velocity measured 10 min prior to the

fusion event, immediately before the fusion event and immediately after the fusion event. C. The

mitochondrial velocity from 10 min before the fission event up to 30 min after the fission event.

D. The mean mitochondrial velocity measured immediately before the fission event, immediately

after the fission event and 20 min after the fission event.

Figure 7. Miro-1 restores mitochondrial dynamics in 120Q-Htt and Tau-expressing

neurons. A. Overexpressed 120Q-Htt-EGFP co-localised with staining for an anti-Htt antibody

(red) in the cytoplasm and showed typical inclusion bodies in the nucleus (stained using DAPI,

blue). B. Overexpression of wt Tau (anti-tau 5A6, red) in neurons expressing the microtubule

marker MAP2c-EGFP. C. Overexpression of Miro-1 was confirmed using an anti-Rhot1

antibody (green). Miro-1 was distributed homogenously in mitochondria (mito-DsRed ). D. The

suppression of mitochondrial motility, kiss rate, fusion rate, and mitochondrial length in neurons

overexpressing 120Q-Htt were all reversed following the co-expression of wt Miro-1 (Htt +

Miro). E. The suppression of mitochondrial motility, kiss rate, fusion rate, and mitochondrial

length in neurons overexpressing wtTau wt was reversed following the c0-expression of wt

Miro-1 (Tau + Miro). *p < 0.05, **p < 0.01 and ***p < 0.001 compared with control and # p <

0.05, # # p<0.01, and # # # p < 0.001 compared with Htt or Tau group.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

32

Figure 8. Summarised principles of the feedback that controls mitochondrial length in

neurons. Mitochondrial number and motility positively correlates with the kiss rate, which in

turn, acts as a pacemaker for fusions. The fusion rate may also be affected by the expression or

activity of fusion proteins. Increases or decreases in the fusion rate in turn leads to a

corresponding increase or decrease in mitochondrial length, respectively, which controls the

fission rate. Alterations in the expression or activity of fission proteins also influences

mitochondrial length, which balances the fission rate via feedback. These controls enable the

equilibration of fusion and fission rates.

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

mito-CFP

mito-Kik-Red

mito-Kik-Red

mito-Kik-Green

mito-Kik-Green

merge

merge

before activation

before fusion

after activation

after fusion

2µm

A

B

C FISSIONFUSION

Event pairs Cerebellar neuronsCortical neurons

Fusion - Fission 96

33 86

49

p < 0.0001

p < 0.0001p < 0.0001

p = 0.0017

15

12 17

14Fusion - Fusion

Fission - Fission

Fission - Fusion

2µm

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

FUSION-FISSION FUSION-FUSION

FISSION-FUSION FISSION-FISSION

normal

normal

erroneous

erroneous

Len

gth

(m

m)

normal

normal

erroneous

erroneous

p = 0.000

p = 0.018

Len

gth

(m

m)

0

1

2

3

4

0

1

2

3

4

A

C

B

D

E

Len

gth

(m

m)

Len

gth

(m

m)

non- FUSING

non- FUSING

non-FISSING

non-FISSING

FUSING

FUSING

FISSING

FISSING

p = 0.92

p = 0.000

0

1

2

3

4

0

1

2

3

4

0 2 4 6 80.0

0.2

0.4

0.6

CerebellarCortical

Fu

sio

n o

r F

issio

n e

ven

t p

er

min

Fu

sio

n o

r F

issio

n e

ven

t p

er

min

0 2 4 6 8 10 120.0

0.2

0.4

0.6

0.8

1.0

1.2

FISSIONFUSION

length (mm)length (mm)

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

0.029 ± 0.002 (71) 0.028 ± 0.003 (71) 2.03 ± 0.06 (1857)

0.051 ± 0.003 (27) ***

0.057 ± 0.004 (8) ***

0.047 ± 0.002 (27) ***

0.051 ± 0.006 (8) *

2.77 ± 0.10 (282) ***

2.45 ± 0.10 (178) *

0.014 ± 0.001 (17) *** 0.015 ± 0.001 (17) ** 1.45 ± 0.04 (556) ***

0.029 ± 0.004 (13) 0.032 ± 0.004 (13) 1.31 ± 0.02 (850) ***

0.032 ± 0.003 (16) 0.025 ± 0.003 (16) 7.03 ± 0.52 (201) ***

control

fusion rate (fusion/mito/min)

fission rate (fission/mito/min)

length (µm)

wt Mfn2

wt Mfn1

dn Mfn2

wt Drp1 + wt Fis1

Drp1 shRNA

ctrl

ctrl

Drp1 shRNA

Drp1 shRNA

b-actin

Drp1

Rela

tive D

rp1/b

-act

in r

atioA

C D

B

E

0

50

100

**

2ìm

2ìm

3ìm

10ìm 10ìm

1ìm

Mnf2-FLAG Mitochondria

Fis1 MitochondriaDrp1

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

0 2 4 6 8 100.0

0.2

0.4

0.6 control

Drp1 shRNADrp1, Fis1

contr

ol

DRP s

hRNA

MFN

20.0

0.2

0.4

0.6

length (µm)

Fis

sio

n r

ate

(even

t/m

in/m

ito

)

Co

eff

icie

nt

of

vari

ab

ilit

y0.8

***

A B

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

0

5

10

contr

ol

nonFUSING

FUSING

contr

ol

contr

ol

contr

ol

Miro

shR

NAs

Miro

shR

NAs

Miro

shR

NAs

Miro

shR

NAs

0.0 0.0

0.2

0.4

0.6

0.8

Mit

och

on

dri

al

velo

cit

y (m

m/s

)

0.1

0.2

0.3

0.4

Co

nta

ct

rate

(conta

ct/m

itoch

ondria/m

in)

0.0

0.5

1.0

1.5

2.0

2.5

Mit

oc

ho

nd

ria

l le

ng

th (m

m)

0.000

0.005

0.010

0.015

0.020

5 mm

Fu

sio

n r

ate

(fusi

on/m

itoch

ondria/m

in)

D E F G

*

*

** * ***

0

0.3

0.6

0.9

1.2

Mit

och

on

dri

al velo

cit

y

(m

m/s

)

non- FUSING

FUSING

A B

b-actinb-actin

Miro-2Miro-1

Miro-2 shRNAMiro-1 shRNA

Miro-1 shRNA + Miro-2 shRNA

ctrlctrl

controlC

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

-10 -8 -6 -4 active

active

active

active

passive

passive

passive

passive

fused

fused

immediately before fusion

immediately after fission

10 min beforefusion

time after fusion (min)

time after fission (min)

Avera

ge

mit

oc

ho

nd

ria

l v

elo

cit

y (m

m/s

)

Avera

ge

mit

oc

ho

nd

ria

l v

elo

cit

y (m

m/s

)

Avera

ge m

ito

ch

on

dri

al

velo

cit

y (m

m/s

)

Avera

ge m

ito

ch

on

dri

al

velo

cit

y (m

m/s

)

20 min afterfission

-2 0

0 0 0

1 1 1

2 2 2

3 3 3

2 4 6 8 10

-10 -8 -6 -4 -2 0

0

1

2

3

fusion

fission

5 10 15 20 25 30

0

1

2

3

0

1

2

3

***

*** *** ***

* **

A B

C D

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

D

* *

*

***

###

###

cont

rol

Htt

Htt+

Miro

cont

rol

Htt

Htt+

Miro

cont

rol

cont

rol

Htt

Htt

Htt+

Miro

Htt+

Miro

0.0

0.5

1.0

Mit

och

on

dri

al velo

cit

y (m

m/s

)M

ito

ch

on

dri

al velo

cit

y (m

m/s

)

0.0

0.1

0.2

0.3

0.4

0.5

Co

nta

ct

rate

Co

nta

ct

rate

0.00

0.01

0.02

0.03

0.04

Fu

sio

n r

ate

Fu

sio

n r

ate

(fusi

on/m

itoch

ondria/m

in)

(fusi

on/m

itochondria/m

in)

0

1

2

3

Mit

och

on

dri

al le

gh

th (m

m)

Mit

och

on

dri

al le

gh

th (m

m)

E

cont

rol

Tau

Tau+

Miro

0.0

0.5

1.0

1.5

cont

rol

Tau

Tau+

Miro

0.0

0.1

0.2

0.3

0.4

0.5

cont

rol

Tau

Tau+

Miro

0.00

0.01

0.02

0.03

0.04

cont

rol

Tau

Tau+

Miro

0

1

2

3

** **

*

# # ## ###

10 µm 2 µm

A

5µm 1µm

CB

25ìm

#

(conta

ct /m

itoch

ondria/m

in)

(conta

ct /m

itoch

ondria/m

in)

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t

FUSION RATE

KISS RATE

MOTILITYNUMBER

FISSION RATE

LENGTH

+ wt hMiro1- shRNA Miro1+2

+ Mfn2- nd Mfn2

+Drp1+Fis1- shRNA Drp1 and nd Drp1

= ++

++

(in steady state)

_

Jour

nal o

f Cel

l Sci

ence

Acc

epte

d m

anus

crip

t