A brain-enriched Drp1 isoform associates with lysosomes, late endosomes, and the plasma membrane

Dynamin-related protein 1 (Drp1) constricts mitochondria as a mechanochemical GTPase during mitochondrial division. The Drp1 gene contains several alternative exons and produces multiple isoforms through RNA splicing. Here we performed a systematic analysis of Drp1 transcripts in different mouse tissues and identified a previously uncharacterized isoform that is highly enriched in the brain. This Drp1 isoform is termed Drp1ABCD because it contains four alterative exons: A, B, C, and D. Remarkably, Drp1ABCD is located at lysosomes, late endosomes, and the plasma membrane in addition to mitochondria. Furthermore, Drp1ABCD is concentrated at the interorganelle interface between mitochondria and lysosomes/late endosomes. The localizations of Drp1ABCD at lysosomes, late endosomes, and the plasma membrane require two exons, A and B, that are present in the GTPase domain. Drp1ABCD assembles onto these membranes in a manner that is regulated by its oligomerization and GTP hydrolysis. Experiments using lysosomal inhibitors show that the association of Drp1ABCD with lysosomes/late endosomes depends on lysosomal pH but not their protease activities. Thus, Drp1 may connect mitochondria to endosomal–lysosomal pathways in addition to mitochondrial division.

Drp1 is a dynamin-related GTPase that mediates the constriction of mitochondria during mitochondrial division (1)(2)(3)(4). Drp1 is a soluble protein in the cytosol and recruited to mitochondria through interactions with the mitochondrial outer membrane proteins such as Mff, Mid49/51, and Fis1 (5,6). On mitochondria, Drp1 oligomerizes into spiral filaments that circle mitochondria. Oligomerization-stimulated GTP hydrolysis by Drp1 regulates constriction of the oligomers and initiates mitochondrial division.
Drp1-driven mitochondrial division is critical for controlling the number and size of mitochondria through a dynamic balance in collaboration with the opposite activity, mitochondrial fusion mediated by mitofusins and Opa1 (4). Demonstrating the physiological importance of Drp1, complete knockout of Drp1 leads to embryonic lethality in mice (7,8). Brain-specific knockout of Drp1 causes neurodegeneration with accumulation of enlarged mitochondria, which results from unopposed mitochondrial fusion in the absence of division (8 -12). These enlarged mitochondria become ineffective in turnover and transport in axons and dendrites because of increases in their size. Similar to neurons, loss of Drp1 in cardiomyocytes in the heart results in mitophagy defects and leads to lethal heart failure (10,13,14). When Drp1 is deleted in oocytes, enlarged mitochondria entangle with other organelles, such as the endoplasmic reticulum, leading to female infertility (15). In addition to these physiological roles, Drp1-mediated mitochondrial division regulates mitochondrial shape during oncogenic transformation, and this dynamic shape change is critical for tumorigenesis (16,17).
Previous studies have identified Drp1 isoforms that are enriched in particular organs and cell types. For example, two Drp1 isoforms (termed Drp1-101 and Drp1-001 in the original study (18) and Drp1 BD and Drp1 D here (Fig. 1, A and D)) are highly expressed in immune cells and associate with microtubules in addition to mitochondria (18). This unique localization is determined by the two alternative exons: the combination of the absence of exon C and the presence of exon D in the variable domain (also known as the B insert) (18). Another example is the isoform of Drp1 that contains exons B, C, and D (Drp1 BCD ; Fig. 1, A and C), and this isoform is mainly expressed in the brain (18 -20). Exon B is located in the GTPase domain. Drp1 BCD has a relatively lower propensity of oligomerization at the basal level but is strongly activated by the major Drp1 receptor/effector protein Mff (19). This property of Drp1 BCD may enable robust control of mitochondrial division in neurons, in which mitochondrial division and fusion are highly regulated.
In this study, we tested whether there are additional Drp1 isoforms. We extensively examined Drp1 transcripts in multiple mouse organs and identified an isoform that is enriched in the brain and associates with lysosomes and late endosomes in addition to mitochondria. Interestingly, this new isoform appeared to be concentrated at contact sites between mito- The amino acid (aa) sequences of these exons are shown. The number of amino acids comprising each exon is shown in parentheses. B, the location of exons A and B (red) in the 3D structure of the GTPase domain of Drp1 ABCD (1-362 amino acid residues) and S-Drp1 (1-343) are predicted using I-TASSER (35) and visualized using Jmol (www.jmol.org; please note that the JBC is not responsible for the long-term archiving and maintenance of this site or any other third party-hosted site). A short protein domain unique to the mitochondrial division dynamin, termed the 80-loop, is shown in yellow (21,27). C and D, expression analysis of Drp1 transcripts. mRNAs isolated from the indicated mouse organs and MEFs are used as templates to PCR-amplify Drp1 cDNAs. ϳ100 Drp1 cDNAs were analyzed by DNA sequencing for each organ. 10 isoforms were identified (C). Different tissues contain different combinations of these 10 isoforms (D). E, production of antibodies specific to the junction of exons A and B. A peptide derived from 12 amino acids (underlined in red in A) was used as an antigen to raise antibodies in rabbits. Antibodies were affinity-purified using the original peptide. To remove antibodies that recognize exon B alone, antibodies were preincubated with purified recombinant His 6 -Drp1 BCD overnight. We also incubated the antibodies with nitrilotriacetic acid-agarose beads coupled to His 6 -Drp1 BCD and removed antibodies that bind exon B alone by centrifugation. F, Drp1-KO MEFs individually expressing each of Drp1 CD , Drp1 ACD , Drp1 BCD , or Drp1 ABCD were subjected to Western blotting using anti-exon AB antibodies in the presence (AB) or absence (crude AB) of overnight preincubation with His 6 -Drp1 BCD . We also used antibodies to Drp1 (pan-Drp1) and the mitochondrial outer membrane protein Tom20. G, Drp1-KO MEFs individually expressing all Drp1 isoforms were analyzed by Western blotting using antibodies against AB, Drp1 (pan-DRP1), Tom20, and GAPDH. H and I, immunoblotting of mouse organs (H) and subregions of the brain (I) using antibodies to AB, Drp1 (pan-Drp1), the mitochondrial protein pyruvate dehydrogenase (PDH), GAPDH, and actin. Equivalent amounts of total proteins were loaded in each lane. chondria and lysosomes/late endosomes. We also characterized mechanisms that target this Drp1 isoform to the lysosomal, endosomal, and plasma membranes.

Identification of a novel Drp1 isoform that is enriched in the brain
We systematically analyzed the expression profile of Drp1 transcripts using mRNAs isolated from different mouse organs and cells, including the brain, heart, skeletal muscle, liver, spleen, testis, lung, and embryonic fibroblasts. We identified 10 different Drp1 isoforms with four alternative exons (exons A, B, C, and D; Fig. 1, A-C). Exons A and B are located in the GTPase domain, whereas exons C and D are present in the variable domain. We found that different tissues contain different combinations of these 10 isoforms (Fig. 1D). Of these, short-Drp1 (S-Drp1), which lacks all of the four alternative exons (Fig. 1, A and C), is the most ubiquitously expressed (Fig. 1D), consistent with previous reports (18 -20). In contrast to the common isoform S-Drp1, two long isoforms, Drp1 BCD , which contains three alternative exons, and a novel isoform, Drp1 ABCD , which contains all of the four alternative exons, were only detected in the brain (Fig. 1D). The brain also expresses four other Drp1 isoforms, S-Drp1, Drp1 BC , Drp1 CD , and Drp1 D . Drp1 BCD and

A novel brain-specific isoform of Drp1
Drp1 CD are major isoforms in the brain and account for ϳ65% of total Drp1 mRNAs (Fig. 1D). Because Drp1 ABCD has not been characterized previously, we focused on this isoform in the rest of the study.
To examine the expression of Drp1 ABCD at the protein level, we generated rabbit polyclonal antibodies that specifically rec-ognize Drp1 ABCD . As an antigen, we used a peptide that corresponds to the junction of exons A and B (six amino acids from each exon) (Fig. 1A) because Drp1 ABCD is the only isoform that carries this sequence (Fig. 1, C and D). Antibodies were affinitypurified using the same peptide (crude exon AB antibodies) (Fig. 1E). To determine the specificity of the crude exon AB A novel brain-specific isoform of Drp1 antibodies, we transduced Drp1-KO MEFs 2 with lentiviruses carrying each isoform and used their whole-cell lysates for Western blotting. Results showed that both Drp1 BCD and Drp1 ABCD were detected by the crude exon AB antibodies (Fig.  1F). To neutralize antibodies that recognize exon B alone, we preincubated the crude exon AB antibodies with recombinant His 6 -Drp1 BCD purified from Escherichia coli (Fig. 1E). Western blotting showed that the preabsorbed anti-exon AB antibodies only detected Drp1 ABCD but not other isoforms (Fig. 1, F and  G). As a control, the expression of these isoforms in Drp1-KO MEFs was confirmed using anti-pan-Drp1 antibodies that recognize all Drp1 isoforms (Fig. 1, F and G).
Using these preabsorbed anti-exon AB antibodies, we examined the expression of Drp1 ABCD in mouse organs. Consistent with the transcript analysis (Fig. 1D), Western blotting showed that Drp1 ABCD is primarily expressed in the brain (Fig. 1H). Anti-pan-Drp1 antibodies showed that the expression level of total Drp1 varies in different organs, with the highest levels in the brain, heart, skeletal muscle, and thymus (Fig. 1H). Further analysis showed that Drp1 ABCD is expressed in most of the subregions of the brain, including the cerebellum, midbrain, hippocampus, striatum, cerebral cortex, and brain stem (Fig. 1I). A relatively low level of Drp1 ABCD was observed in the olfactory bulb, whereas anti-pan-Drp1 antibodies showed similar expression levels of total Drp1 in these brain subregions (Fig. 1I).

Drp1 ABCD functions in mitochondrial and peroxisomal division
To test the function of Drp1 ABCD in mitochondrial division, we individually expressed Drp1 ABCD and other isoforms in Drp1-KO MEFs using the lentivirus transduction system (Fig.  1G). The lentivirus system provided almost 100% transduction efficiency and yielded similar expression levels for each Drp1 isoform in Drp1-KO MEFs ( Fig. 1G) (10). Using these cells, we performed confocal immunofluorescence microscopy with antibodies to Drp1 (pan-Drp1) and the mitochondrial protein Tom20. Drp1-KO MEFs showed an elongated mitochondrial morphology because of defects in mitochondrial division compared with WT MEFs (Fig. 2, A and B). All of the isoforms, including Drp1 ABCD , restored short tubular mitochondria in Drp1-KO MEFs, which look similar to those seen in WT MEFs (Fig. 2, A and B). Consistent with its function in mitochondrial division, Drp1 ABCD formed punctate structures on mitochondria, similar to other Drp1 isoforms ( Fig. 2A). In addition to mitochondrial division, Drp1 ABCD also functions in mitochon-drial respiration. We measured oxygen consumption rates in Drp1-KO MEFs and Drp1-KO MEFs expressing S-Drp1 or Drp1 ABCD . Results showed that the expression of S-Drp1 and Drp1 ABCD similarly increased basal respiration and respiratory capacity (Fig. 2

Drp1 ABCD associates with lysosomes and late endosomes
Interestingly, we noticed that Drp1 ABCD , but not the other nine isoforms, is also located at vesicular structures in the cytosol in addition to mitochondria when we changed focal planes ( Fig. 2A, circled area). To characterize these vesicles, we performed confocal immunofluorescence microscopy with antibodies against Drp1 and Lamp1, a membrane protein associated with lysosomes and late endosomes. We found that Drp1 ABCD -positive vesicles were co-labeled with the anti-Lamp1 antibody (Fig. 3A). To further test whether Drp1 ABCD is associated with lysosomes and/or late endosomes, we stained Drp1-KO MEFs expressing Drp1 ABCD with LysoTracker and then performed immunofluorescence microscopy with Drp1 and Lamp1 antibodies. We found that a subset of Drp1 ABCDpositive vesicles co-labeled with LysoTracker ( Fig. 3B). We also analyzed Drp1-KO MEFs expressing Drp1 ABCD using antibodies against Drp1, Lamp1, and a late endosome marker, Rab7. The results showed that some of Drp1 ABCD -positive vesicles co-localized with Rab7 (Fig. 3C). In contrast, when we used antibodies against EEA1, an early endosomal marker, we found no co-localization with Drp1 ABCD (Fig. 3D). These data suggest that Drp1 ABCD is associated with both lysosomes and late endosomes.
Because lysosomes/late endosomes are associated with microtubules, we tested whether depolymerizing microtubules changes the localization of Drp1 ABCD at these membrane vesicles. We found that a microtubule-depolymerizing drug, nocodazole, does not affect the localization of Drp1 ABCD on lysosomes/late endosomes, indicating that association of Drp1 ABCD with these vesicles is independent of microtubules ( Fig. 3E).
By moving the focal plane close to the bottom of cells, we found that Drp1 ABCD was associated with the plasma membrane (Fig. 3F). We often found fewer Lamp1-positive vesicles

A novel brain-specific isoform of Drp1
in the focal planes, where we observed localization of Drp1 ABCD at the plasma membrane (Fig. 3, F and G). To confirm this plasma membrane localization, we examined Drp1-KO MEFs co-expressing Drp1 ABCD and a plasma membrane marker, Lyn-mCherry (22). We found that Drp1 ABCD co-localized with Lyn-mCherry (Fig. S2) Because Drp1 has the ability to control the size of mitochondria through mitochondrial division, we tested whether Drp1 ABCD also changes the size of lysosomes/late endosomes. We stained WT MEFs, Drp1-KO MEFs, and Drp1-KO MEFs expressing either S-Drp1 or Drp1 ABCD using LysoTracker and analyzed the size of LysoTracker-stained vesicles using confocal microscopy. We found that the size distribution of Lyso-Tracker-stained vesicles was similar in these cells (Fig. 4A). Mean and median areas of LysoTracker-stained vesicles were also indistinguishable (Fig. 4B).
We then tested whether Drp1 ABCD affects the size of larger LysoTracker-stained vesicles by reanalyzing the size distribution for LysoTracker-stained vesicles that have larger area (Ͼ1 m 2 ). Again, we found similar sizes of LysoTrackerstained vesicles in WT MEFs, Drp1-KO MEFs, and Drp1-KO MEFs expressing either S-Drp1 or Drp1 ABCD (Fig. 4D). Thus, Drp1 ABCD does not affect the size of lysosomes/late endosomes. Western blotting of whole-cell lysates showed that the expression levels of S-Drp1 and Drp1 ABCD are comparable (Fig. 4E).

Drp1 ABCD is located at interfaces between mitochondria and lysosomes/late endosomes
We noticed that the distribution of Drp1 ABCD along the Lamp1-positive membrane is not uniform; even on the same vesicles, there are regions that are associated with Drp1 ABCD whereas other parts are not (see gallery of Drp1 ABCD -associated Lamp1-positive vesicles in Fig. 5A). Interestingly, parts of Lamp1-positive membranes that associated with Drp1 ABCD appear to be close contact with mitochondria labeled with antibodies to the mitochondrial outer membrane protein Tom20 (Fig. 5A, arrowheads). Quantification of these images showed that Lamp1-positive vesicles are significantly more closely A novel brain-specific isoform of Drp1 located to mitochondria where Drp1 ABCD is present (Fig. 5B). To further determine the spatial relationship between Lamp1, Drp1 ABCD , and the mitochondrial outer membrane, we performed a line scan analysis of the fluorescent signal of Lamp1, Drp1, and Tom20. As shown in Fig. 5C, the peak of the Drp1 ABCD signal is located between the peak of Lamp1 and Tom20 signals. It appears that Drp1 ABCD is present between the Lamp1-positive membranes and mitochondria.
To determine whether contacts between mitochondria and Lamp1-positive vesicles increase when Drp1 ABCD is ectopically expressed in Drp1-KO MEFs, we biochemically fractionated mitochondria and examined how many lysosomes/late endosomes co-fractionate using Western blotting with antibodies against Lamp1 and V-type proton ATPase subunit C. The results showed that similar amounts of these two proteins associated with the mitochondrial fraction derived from Drp1-KO MEFs expressing S-Drp1 or Drp1 ABCD (Fig. 5D). These data suggest that Drp1 ABCD is recruited to existing interorganelle interfaces between mitochondria and lysosomes/late endosomes. The peak position of Drp1 is located between those of Lamp1 and Tom20. D, mitochondria were isolated from Drp1-KO MEFs expressing S-Drp1 or Drp1 ABCD using a mitochondrial isolation kit (Thermo, 89874). Whole-cell lysates and mitochondrial fractions were analyzed by Western blotting with antibodies against Drp1, Lamp1, ATP6V1C1, Tom20, and GAPDH.

A novel brain-specific isoform of Drp1 Exons A and B target Drp1 ABCD to lysosomes/late endosomes and the plasma membrane
To determine the mechanisms by which Drp1 ABCD is associated with lysosomes/late endosomes and the plasma membrane, we tested which alternative exons are required for these localizations. We transduced Drp1-KO MEFs with lentiviruses expressing Drp1 carrying different combinations of the four alternative exons: A, B, C, D, CD, AB, ABC, ABD, and ABCD. Immunofluorescence microscopy of these MEFs using antibodies to Drp1 and Lamp1 showed that Drp1 A , Drp1 B , Drp1 C , Drp1 D , and Drp1 CD do not associate with Lamp1-positive vesicles (Fig. 6A), consistent with the earlier data (Fig. 3A). In contrast, all of the Drp1 isoforms that contained both exon A and B (e.g. Drp1 AB , Drp1 ABC , Drp1 ABD , and Drp1 ABCD ) are localized to Lamp1-positive vesicles (Fig. 6A). Therefore, the presence of both exons A and B determines the localization of Drp1 ABCD at lysosomes and late endosomes.
Using the same set of cells, we also examined the localization of these Drp1 proteins at the plasma membrane. Similar to lysosomal/late endosomal localization, we found that exons A and B are also necessary for the plasma membrane localization of Drp1 (Fig. 6B).
To test whether the GTPase domain with the AB insertion or the AB insertion itself is sufficient for localization to Lamp1positive vesicles and the plasma membrane, we performed immunofluorescence microscopy using antibodies against the HA epitope and Drp1 in Drp1-KO MEFs expressing either the HA-GTPase domain or the HA-GTPase AB domain. We found that both of the GTPase domains were present in the cytosol (Fig. 6C). Similarly, GFP fused to the AB insertion (GFP-AB and AB-GFP) localized to the cytosol when expressed in Drp1-KO MEFs (Fig. 6D). Therefore, in addition to the GTPase domain, other regions of Drp1 ABCD are also necessary for targeting Drp1 ABCD to Lamp1-positive vesicles and the plasma membrane.

GTP hydrolysis and oligomerization of Drp1 ABCD is required for its localization to lysosomes/late endosomes and the plasma membrane
Localization of Drp1 to mitochondria is regulated by its oligomerization and disassembly of oligomers, which is triggered by GTP hydrolysis (5). To further define the mechanism underlying the subcellular localization of Drp1 ABCD , we tested the role of the GTPase activity of Drp1 ABCD and its oligomerization in localization at lysosomes/late endosomes and the plasma membrane. We first determined whether Drp1 ABCD has GTPase activity because Drp1 ABCD has an extra amino acid insertion derived from exons A and B in the GTPase domain. We purified recombinant His 6 -Drp1 ABCD proteins from E. coli and incubated them with GTP (Fig. 7A). GTP hydrolysis was analyzed by measuring the release of phosphate using a mala-chite green assay as we performed previously (23). The results showed that His 6 -Drp1 ABCD hydrolyzes GTP similar to a positive control, His 6 -S-Drp1 (Fig. 7A). As a negative control, we used His 6 -S-Drp1 (K38A), which carries a mutation in the GTPase domain that blocks GTP hydrolysis (24) (Fig. 7A). To assess the role of the GTPase activity of Drp1 ABCD in its localization in cells, we expressed Drp1 ABCD (K38A) in Drp1-KO MEFs and performed confocal immunofluorescence microscopy. We found that Drp1 ABCD (K38A) does not localize to Lamp1-positive vesicles, the plasma membrane, or mitochondria, instead forming aggregate in the cytosol (Fig. 7, B and D). Although we sometimes observed Drp1 ABCD (K38A) aggregates in perinuclear regions, these aggregates were not associated with the Golgi complex (Fig. 7E). Similar aggregation has been observed for S-Drp1 (K38A) and likely results from the defect of disassembly of Drp1 oligomers because GTP hydrolysis is required for this disassembly (5).
Second, we introduced the G350D mutation, which inhibits the oligomerization of Drp1 (25), and expressed Drp1 ABCD (G350D) in Drp1-KO MEFs. Confocal immunofluorescence microscopy showed that Drp1 ABCD (G350D) diffuses in the cytosol and does not associate with Lamp1-positive vesicles, the plasma membrane, or mitochondria (Fig. 7, C and D). Furthermore, combination of the K38A and G350D mutations showed effects on Drp1 ABCD localization similar to the G350D mutation alone, suggesting that oligomerization defects are epistatic to GTP hydrolysis defects (Fig. 7, C and D). Taken together, these data suggest that the localizations of Drp1 ABCD at lysosomes/late endosomes and the plasma membrane require its ability to hydrolyze GTP and homo-oligomerize, suggesting that Drp1 ABCD is associated with these membranes as assembled structures.

The localization of Drp1 ABCD at lysosomes/late endosomes is sensitive to bafilomycin A and chloroquine
To determine whether the localization of Drp1 ABCD at lysosomes/late endosomes depends on the acidity or function of these membrane compartments, we first treated Drp1-KO MEFs expressing Drp1 ABCD with two inhibitors, bafilomycin A and chloroquine, both of which block acidification of lysosomes/endosomes through distinct mechanisms (26). Although bafilomycin A blocks vacuolar H ϩ -ATPase, mild base chloroquine neutralizes the lysosomal/ endosomal acidic pH in these organelles. We found that these inhibitors significantly decreased the association of Drp1 ABCD with Lamp1-positive vesicles (Fig. 8, A and B). This inhibition was reversible because the removal of bafilomycin A from the culture medium restored the localization of Drp1 ABCD at Lamp1-positive vesicles in 2 h (Fig. 8, A and B). These inhibitory effects are specific to Lamp1-positive vesicles; localization of Drp1 ABCD at mitochondria and the plasma membrane was not affected by bafilomycin A or chloroquine (Fig. 8C).
Increased pH could decrease the activity of acid hydrolyases and compromise the function of lysosomes. To find out whether the function of lysosomes is required for the localization of Drp1 ABCD at lysosomes/late endosomes, we treated Drp1-KO MEFs expressing Drp1 ABCD with inhibitors to cysteine proteases (E-64d) and to aspartic acid protease (pepstatin A), both of which inhibit lysosomal proteases (26). We found that these protease inhibitors do not change the localization of Drp1 ABCD at lysosomes (Fig. 8A and B). These data suggest that the acidity of lysosomes, but not their function, is required for the association of Drp1 ABCD with lysosomes/late endosomes.

Discussion
In this work, we identified a new Drp1 isoform, Drp1 ABCD , which predominantly expresses in the brain and locates at lysosomes, late endosomes and the plasma membrane in addition to mitochondria. The identification of Drp1 ABCD expands a repertoire of Drp1 isoforms that are differentially expressed in tissues and organs. Our data also suggest that neuronal Drp1 ABCD may function at lysosome/endosome-mitochondria contact sites in endosome-lysosome pathways in addition to mitochondrial division. We currently do not know the exact function of Drp1 ABCD and are planning to address this by specifically knocking down Drp1 ABCD in neurons in our future studies.
The unique localization of Drp1 ABCD to lysosomes, late endosomes, and the plasma membrane depends on the combination of exons A and B. Based on the crystal structure of Drp1, amino acid sequences encoded in these two exons are predicted to be exposed on the surface of this protein (Fig. 1B). These amino acid sequence are necessary, but likely not sufficient, to localize Drp1 ABCD to lysosomes, late endosomes, and the plasma membrane because oligomerization of Drp1 ABCD , which is mediated by the stalk domain, is also essential for these localizations. It appears that Drp1 ABCD assembles into oligomers that are recruited to these membranes. Exons A and B likely help recruit Drp1 ABCD oligomers to lysosomes, late endosomes, and the plasma membrane because we do not find association of Drp1 ABCD (G350D) with these membrane structures. However, it is also possible that Drp1 ABCD oligomerizes onto the membrane structures and helps recruit more Drp1 ABCD molecules in a synergistic manner. The localization signal created by exons A and B is not exclusive, as we observed that a fraction of Drp1 ABCD proteins is associated with mitochondria and peroxisomes.
The variable domain, which is a predicted unstructured loop with ϳ100 amino acids, has been proposed to mediate several types of Drp1 regulation (27). For example, this domain undergoes phosphorylation to regulate the interaction of Drp1 with its receptor/effector proteins such as MFF, Fis1, and Mid49/51 (28,29). Such regulations modulate the recruitment of Drp1 to mitochondria and its assembly on mitochondria. In addition, this domain also participates in interactions with two mitochondrial phospholipids, cardiolipin and phosphatidic acid, that regulate Drp1 function at distinct steps. Cardiolipin promotes the oligomerization of Drp1, whereas phosphatidic acid suppresses oligomerized Drp1 to initiate mitochondrial division (4, 23, 30 -33). Alternative splicing of exons C and D would regulate these protein-protein and protein-phospholipid interactions that fine-tune mitochondrial division. The alternative splicing in the variable domain also changes the association of Drp1 with microtubules and regulates microtubule stability. In addition to the variable domain, our data show that alternative splicing of exons A and B in the GTPase domain also plays a critical role in the regulation of Drp1 beyond its GTP hydrolysis activity. The amino acid sequence derived from exons A and B is located on the surface of this domain, and these two exons are necessary and sufficient alternative axons for lysosomal/endosomal and plasma membrane localization. We therefore speculate that this region likely contributes to a unique protein-protein or protein-lipid interaction for the association of Drp1 ABCD with specific molecules in the endosomal-lysosomal system.

Mice and MEFs
All animal work was done according to guidelines established by the Johns Hopkins University Committee on Animal Care. Drp1-KO MEFs were cultured in Iscove's modified Dulbecco's medium supplemented with 10% fetal bovine serum as described previously (10).

Drp1 transcript analysis
Total RNAs were isolated from mouse brains and MEFs using the RNeasy kit (Qiagen). cDNAs of Drp1 isoforms were prepared from total RNAs using the primers 5-BamH1 Drp1 (AAA GGATCC GCCACC ATGGAGGCGCTGATCCC-GGT) and 3-Not1 Drp1 (AAA GCGGCCGC TCACCAAAGA-TGAGTCTCTC) using the OneStep-RT-PCR kit (Qiagen). Drp1 transcripts in the heart, skeletal muscle, liver, lung, spleen, and testis were PCR-amplified from a mouse cDNA library (Mouse MTC TM Panel I, Clontech). We subcloned ϳ100 Drp1 transcripts from each organ and from MEFs into the pHR-SIN plasmid at BamHI and Not1 sites. We individually sequenced all of these plasmids and identified isoforms. The relative abundance of each isoform in each organ (percent) was calculated by dividing the number of plasmids carrying a specific isoform by the total number of plasmids sequenced.

Production of anti-exon AB antibodies
Anti-exon AB antibodies that specifically recognize Drp1 ABCD were produced using a peptide (amino acid sequence KFQ-SWNPATWKN) as the antigen in rabbits at NeoScientific and affinity-purified using the same peptide. To remove the residual activity of the exon AB antibodies to bind exon B alone, they were preincubated with purified recombinant His 6 -Drp1 BCD overnight prior to Western blotting. In some experiments, we incubated the exon AB antibodies with nitrilotriacetic acidagarose beads coupled to His 6 -Drp1 BCD (molar ratio of anti-AB IgG His 6 -Drp1 BCD ϭ 1:2.5) and used the supernatant after centrifugation. The specificity of the antibodies was tested using Western blotting of Drp1-KO MEFs individually expressing Drp1 ACD , Drp1 BCD , or Drp1 ABCD .

Lentiviruses and plasmids
Lentiviruses were produced as described previously (10). The pHR-SIN plasmid carrying Drp1 isoforms was cotransfected into HEK293T cells with two other plasmids, pHR-CMV8.2⌬R and pCMV-VSVG, using Lipofectamine 2000 (Invitrogen). Two days after transfection, the supernatant of transfected cells containing released viruses was collected. The viruses were quick-frozen in liquid nitrogen and stored at Ϫ80°C.
To generate HA-Drp1 ABCD , full-length Drp1 ABCD was cloned into BamHI and NotI sites of the pEGFP-N1 vector (Clontech) to replace enhanced GFP with Drp1 ABCD . Next, 1ϫ HA was cloned into EcoRI and KpnI sites. To make HA-GTPase, Drp1 ABCD was replaced with the GTPase domain of S-Drp1 (1-343 amino acids) or Drp1 ABCD (1-362 amino acids) at BamHI and NotI sites. For GFP-AB and AB-GFP, AB exons were cloned into the pEGFP-N1 and pEGFP-C1 vectors at XhoI and BamHI sites. The plasmid carrying Lyn-mCherry was a gift from Dr. Hideki Nakamura (Johns Hopkins University).
To quantify the size of Lamp1-positive vesicles, cells were stained with 75 nM LysoTracker Red DND-99 (Invitrogen, L7528) at 37°C for 30 min in the culture medium. Images were converted to binary images using ImageJ, and the area of each vesicle was quantified.

Mitochondrial respiration
Mitochondrial oxygen consumption rates were measured in an XF96 Extracellular Flux Analyzer (Seahorse Bioscience) as described previously described (10,34). Briefly, cells were plated at a density of 5,000 cells/well in an XF 96-well culture microplate and incubated for 24 h. To record oxygen consumption rates, the culture medium was replaced with Seahorse XF base medium supplemented with 25 mM glucose, 1 mM sodium pyruvate, and 2 mM GlutaMAX-I, and then the cells were incubated at 37°C in a CO 2 -free incubator for 1 h. Oxygen consumption rate analysis was initiated with a 13.5-min equilibration step, followed by three cycles of 2 min of mixing and 3 min of measurement. All measurement cycles were performed at 37°C. The basal oxygen consumption rate was determined, followed by sequential injections of 2 M oligomycin, 1 M carbonyl cyanide p-trifluoromethoxyphenylhydrazone, and 0.5 M rotenone plus 0.5 M antimycin A. Oxygen consumption rates were normalized relative to the amount of protein in each well.

GTP hydrolysis by purified Drp1 proteins
The GTPase activity of His 6 -S-Drp1, His 6 -Drp1 ABCD , and His 6 -S-Drp1 (K38A) was measured as we described previously (23). 0.1 M Drp1 was incubated with 1 mM GTP in 150 mM KCl, 2 mM MgCl 2 , 1 mM DTT, and 20 mM MES (pH 7.0). 25-l samples were mixed with 100 l of malachite green solution (Echelon, K-1500) at room temperature for 20 min, and the absorbance at A 620 was measured.