Inner membrane fusion mediates spatial distribution of axonal mitochondria

In eukaryotic cells, mitochondria form a dynamic interconnected network to respond to changing needs at different subcellular locations. A fundamental yet unanswered question regarding this network is whether, and if so how, local fusion and fission of individual mitochondria affect their global distribution. To address this question, we developed high-resolution computational image analysis techniques to examine the relations between mitochondrial fusion/fission and spatial distribution within the axon of Drosophila larval neurons. We found that stationary and moving mitochondria underwent fusion and fission regularly but followed different spatial distribution patterns and exhibited different morphology. Disruption of inner membrane fusion by knockdown of dOpa1, Drosophila Optic Atrophy 1, not only increased the spatial density of stationary and moving mitochondria but also changed their spatial distributions and morphology differentially. Knockdown of dOpa1 also impaired axonal transport of mitochondria. But the changed spatial distributions of mitochondria resulted primarily from disruption of inner membrane fusion because knockdown of Milton, a mitochondrial kinesin-1 adapter, caused similar transport velocity impairment but different spatial distributions. Together, our data reveals that stationary mitochondria within the axon interconnect with moving mitochondria through fusion and fission and that local inner membrane fusion between individual mitochondria mediates their global distribution.


Inducible knockdown
We knocked down dOpa1 at different developmental stages of Drosophila using RU486 inducible elav-Gal4 driver (switch-elav > dOpa1RNAi) 1,2 and characterized mitochondrial morphology in segmental nerves of third instar larvae. Expression of UAS transgenes reached the highest level in ~21 hours after feeding 2 . For inducible gene expression, crosses were carried out on normal food, and 2 nd or early 3 rd instar larvae were collected and transferred for feeding of food containing 15ug/ml RU486. We fed 2 nd or early 3 rd instar larvae for 3 or 2 days, respectively, before wandering 3 rd instar larvae were collected for TUNEL staining, in vivo imaging, or transmission electron microscopy (TEM). Flies under induced dOpa1 knockdown died at late pupa stage or on adult day 1 or day 2. For imaging mito-GFP under induced dOpa1 knockdown, female switch-elav Gal4 was crossed with male UAS-mito-GFP/CyO to generate a control. Male UAS-mito-GFP/+; switch-elav Gal4/+ was further crossed with UAS-dOpa1RNAi female to generate UAS-mito-GFP/UAS-dOpa1RNAi; switch-elav Gal4/+. For TEM analysis, elav-Gal4 or switch-elav Gal4 female was crossed with UAS-dOpa1RNAi male. Age-matched UAS-dOpa1RNAi larvae were used as a control.

Transmission electron microscopy
Drosophila third instar larvae were fixed in 3% paraformaldehyde with 1% glutaraldehyde at 4°C for 24 hours. After washing in 3 rounds of PBS, the samples were placed in a 1% osmium tetroxide solution buffered with PBS for one hour, followed by 3 round of 3 washing with water. The brain and nerve fibers were dissected from the body. The samples were dehydrated in a series of ethanol solutions of increasing concentration (50%, 70%, 95%, and 100%). Propylene oxide was used as a transitional solvent, and the samples were placed in a 1:1 mixture of Spurr resin and propylene oxide, and stored overnight in a desiccator. The following day, the Spurr and propylene oxide mixtures were removed and replaced with 100% Spurr resin.
The samples were infiltrated with the Spurr resin for an additional 8 hours, placed in flat embedding molds, and polymerized for 48 hours at 60ºC. The samples were re-embedded in Spurr resin in an orientation with the cerebral hemispheres at the base of the embedding mold.
Thick sections of 2 microns were taken for approximately 50 microns. At that distance, thin (100nm) sections were cut using a DDK diamond knife on a Reichert-Jung Ultracut E ulatramicrotome. The sections were stained with lead citrate for 1 minute, and were observed for section and stain quality. If the quality was adequate, the samples were sectioned an additional 50 microns deeper into the specimen, and another set of thin sections were collected and stained.
This procedure was continued until the nerve fibers had been sectioned at a distance of several hundred microns from the brain. The grids were viewed on a Hitachi H-7100 transmission electron microscope (Hitachi High Technologies) operating at 75 keV. Digital images were collected using an AMT Advantage 10 CCD Camera System (Advanced Microscopy Techniques) and inspected using NIH ImageJ software.

3D confocal microscopy
To characterize mitochondrial morphology in the thick bundle of segmental nerves, we used 3D confocal microscopy. Three dimensional z-stacks of mito-GFP within the segmental nerve bundle of control and inducible dOpa1 knockdown larvae were collected on a spinning disk confocal microscope with an EMCCD camera (Andor Technology). The effective pixel size was 0.105 µm in the x and y dimension, and 0.2 µm in the z dimension. Mitochondria in each image were segmented by custom software in 2D using the difference of Gaussian filter followed by adaptive thresholding, as previously described 3 . The 3D volume of each mitochondrion was determined by stacking their 2D segmentations.

TUNEL staining
Brains of third instar larvae containing intact CNS and ventral ganglia were dissected in standard HL3 media. Positive control was included in each round of assay. For each round and each genotype, 3~4 brains were collected in 500 µl tubes. Brains were fixed in 4% PFA in PBS overnight in 4 ºC and were washed in 0.3% Triton-X100, 0.1% sodium citrate (pH 6) for 1 hour.
Then they were rinsed in PBS with 0.1% Triton-X100 (PBST) 3 times. Positive controls were treated with DNaseI (Sigma) for 10 mins. Brains were blocked with 5% BSA diluted in PBST for one hour at room temperature. TUNEL staining was carried out following manufacturer's protocol (Roche). Brains were incubated with TUNEL staining reagents for 2 hours at room temperature in darkness. After washing with PBST, brains were stained with 10ug/ml Hoechst 33342 (Invitrogen) for 1 hour. TUNEL signals were imaged using a Nikon Eclipse Ti-E inverted microscope with a CoolSNAP HQ2 camera (Photometric) and a 20× objective lens. We took 1.8 μm thick stacks through the whole sample. TUNEL positive signals were quantified from the maximal projection of collected z-stacks. Positive signals were detected by the Otsu segmentation function in ImageJ.
To measure axon growth, primary neurons were treated with 500 nM mitoTracker Red for 5 mins on DIV3. Images were taken on a Nikon Eclipse TE2000-U inverted microscope with an Andor EMCCD camera and a 60×/1.40 NA oil objective lens. Axon arbor was imaged under DIC and mitochondria were imaged using a Cy3 filter set. Axons were traced manually. Axon traces were smoothed using polynomial fitting in MATLAB.
To measure mitochondrial membrane potential, primary larval neurons were treated with JC1 dye at 10ug/ml for 20 mins on DIV 3. JC1 dye was diluted in full culture media and centrifuged to remove pellet. Images were taken by a CoolSNAP HQ2 camera. JC-1 monomer (green channel) was imaged using a FITC filter set; while JC-1 aggregate (red channel) was imaged using a TRITC filter set. The intensity of the red emission is proportional to mitochondrial membrane potential. To measure the ratio of red/green intensity, we first used image segmentation after Gaussian filtering to identify individual mitochondria. We then calculated the ratio between red and green channel intensities for each identified mitochondrion.

Western blot analysis
Twenty brains of Drosophila third instar larvae were collected in ice cold PBS and homogenized in RIPA (Life Technologies) supplemented with PMSF (Sigma) and protease inhibitor cocktails (Sigma). Proteins were separated on 10% polyacrylamide gel by electrophoresis in Tris-Glycine buffer, and transferred to nitrocellulose membrane. Membranes were blocked in 5% non-fat milk for 1 hour at room temperature. Blocked membranes were blotted with mouse anti-GAPDH (Santa Cruz) or mouse anti-OPA1 (Abnova) overnight at 4 ºC.
On the following day, after washing with 0.1% Tween TBS, membranes were blotted with HRP conjugated secondary antibody (Thermo). SuperSignal West Pico Substrate (Thermo) was used for detection. Membranes were imaged using a LAS-3000 imager (GE Healthcare).

Bootstrap testing of control velocity data
Control data were collected for each experiment. Their analysis results were pooled into a single dataset to ensure that the sample size was sufficiently large for reliable determination of statistical distribution of mitochondrial properties such as transport velocity, size, and aspect ratio. Specifically, a total of 8 control experiments were conducted, with an average of 12 time-  The stationary mitochondrion changed its shape dynamically in the fusion. This was identified as a visual cue in verification. Movies S1 and S2 are the corresponding time-lapse movies for (A) and (B), respectively.  Kymograph of a representative time-lapse movie from a wild-type larva expressing mito-GFP (sg26 > UAS-mito-GFP). Five stationary mitochondria were marked by green arrows at the top.
The intensity change of each mitochondrion was revealed by its corresponding trace in the kymograph. Four out of the five mitochondria showed intensity recovery, and the starting time points of their abrupt intensity increase were marked by red arrows. The average intensity recovery of the four mitochondria was 54 ± 23% (mean±SD, n = 4). No gradual intensity recovery indicative of soluble mito-GFP uptake was observed.