Cyclic AMP-dependent Protein Kinase Phosphorylation of Drp1 Regulates Its GTPase Activity and Mitochondrial Morphology*

Mitochondria in cells comprise a tubulovesicular reticulum shaped by dynamic fission and fusion events. The multimeric dynamin-like GTPase Drp1 is a critical protein mediating mitochondrial division. It harbors multiple motifs including GTP-binding, middle, and GTPase effector (GED) domains that are important for both intramolecular and intermolecular interactions. As for other members of the dynamin superfamily, such interactions are critical for assembly of higher-order structures and cooperative increases in GTPase activity. Although the functions of Drp1 in cells have been extensively studied, mechanisms underlying its regulation remain less clear. Here, we have identified cAMP-dependent protein kinase-dependent phosphorylation of Drp1 within the GED domain at Ser637 that inhibits Drp1 GTPase activity. Mechanistically, this change in GTPase activity likely derives from decreased interaction of GTP-binding/middle domains with the GED domain since the phosphomimetic S637D mutation impairs this intramolecular interaction but not Drp1-Drp1 intermolecular interactions. Using the phosphomimetic S637D substitution, we also demonstrate that mitochondrial fission is prominently inhibited in cells. Thus, protein phosphorylation at Ser637 results in clear alterations in Drp1 function and mitochondrial morphology that are likely involved in dynamic regulation of mitochondrial division in cells.

Mitochondria are critical for a number of cellular functions because of their roles in producing energy, buffering calcium, and regulating apoptosis. In contrast to their common depiction as sausage-shaped organelles in schematic diagrams, they form a dynamic reticulum in the cell, continuously dividing and fusing with one another (1)(2)(3)(4). The steady-state equilibrium of these opposing processes is altered during critical cellular events such as apoptosis and cell division, and perturbation of this balance has been directly implicated in a number of inherited neurological disorders including Charcot-Marie-Tooth neuropathy and optic atrophy type 1 (3,5) as well as a neonatal lethal syndrome of microcephaly, abnormal brain development, optic atrophy, and lactic acidemia (6).
The dynamin-related GTPase Drp1 is an evolutionarily conserved protein that mediates mitochondrial division, and its functional impairment results in aggregates of large, interconnected mitochondria within cells. Drp1 is also involved in the mitochondrial scissioning that occurs during apoptosis and cell division (5,7). However, mechanisms by which Drp1 function is regulated during such critical cellular events are less well understood. Frequently, cellular responses to changing conditions are mediated by reversible covalent modifications of existing molecules, and Drp1 is modified post-translationally by protein phosphorylation, ubiquitylation, and sumoylation (7)(8)(9)(10). However, in most cases, neither the direct impact of these modifications on Drp1 functions nor the sites of modification have been thoroughly investigated.
Like many dynamin-related proteins, Drp1 has an N-terminal GTP-binding domain, a middle assembly domain, a short insert (insert B), and a C-terminal interaction domain dubbed the GTPase effector domain (GED); 2 the GED domain folds back to interact in cis with the GTP-binding and middle domains and also interacts intermolecularly in trans with GED domains in other Drp1 proteins (11)(12)(13)(14)(15). Here, we demonstrate phosphorylation of Drp1 by cAMP-dependent protein kinase (protein kinase A; PKA) at a specific serine residue within the C-terminal GED domain of Drp1. We further show that this phosphorylation alters the intramolecular domain associations of the Drp1 GTPase, resulting in decreased GTPase activity and impairment of mitochondrial fission.

EXPERIMENTAL PROCEDURES
Metabolic Labeling-HeLa cells were transfected with pGW1-Myc-Drp1 (splice variant 1; GenBank TM Accession Number NM_012062) as described previously (15). After 24 h, cells were incubated in serum-free Dulbecco's modified Eagle's medium for 3 h and then labeled with 1 mCi/ml [ 32 P]orthophosphate (8500 -9120 Ci/mmol; PerkinElmer Life Sciences) in phosphate-free medium for 3.5 h. Me 2 SO or 20 M forskolin (Sigma-Aldrich) and 10 M 3-isobutyl-1-methylxanthine (IBMX; Sigma-Aldrich) prepared in Me 2 SO were added 20 min before cells were lysed in modified radioimmunoprecipitation buffer supplemented with protease inhibitor mixture (Sigma-Aldrich; P8340) and phosphatase inhibitor mixture (Sigma-Aldrich; P2850). Myc-Drp1 protein was immunoprecipitated using anti-Myc antibodies immobilized on agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) * This work was supported by the Intramural Research Program of the NINDS, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains a supplemental figure. 1  Sheets were again air-dried, and 32 P-labeled phosphopeptides were visualized by autoradiography.
Yeast Two-hybrid Analyses-Full-length Drp1 and Drp1 truncation fragments were cloned into pGAD10 and pBHA vectors as described previously (15). Yeast two-hybrid tests were performed in the L40 strain, which contains LexA operators upstream of HIS3 and lacZ reporter genes. Interactions were semiquantified based on growth on His/Leu/Trp dropout plates supplemented with 10 mM 3-amino-1,2,4-triazole (Sigma-Aldrich) and time for colonies to turn blue in X-gal filter lift assays from Leu/Trp dropout plates (15), with similar results obtained in at least three experiments. Immunoprecipitates were resolved by SDS-PAGE, and 32 P-labeled Myc-Drp1 was visualized by autoradiography. Migrations of molecular mass standards (in kDa) are indicated to the left. B, tryptic phosphopeptides of 32 P-labeled Myc-Drp1 were resolved by two-dimensional TLC. Origins are identified with filled circles. The major phosphopeptide induced by forskolin and IBMX treatment is indicated with an arrow. C, GST-Drp1 WT and Drp1 S637A (0.5 g of protein/lane) were phosphorylated in vitro using PKA in the presence of [␥-32 P]ATP and resolved by SDS-PAGE. 32 P-labeled GST-Drp1 is indicated by an arrow, and an asterisk identifies autophosphorylated PKA. No GST protein was included in the control shown. D, samples from in vitro Drp1 phosphorylation assays were subjected to phosphopeptide mapping. A prominent phosphopeptide difference between Drp1 WT and Drp1 S637A maps is indicated by arrows. E, in vitro phosphorylated GST-Drp1 WT and Drp1 S637A (0.5 g/lane) were immunoblotted with phosphospecific antibodies against phospho-Ser 637 . Protein phosphatases (PPases) were added as indicated. F, HeLa cells (25 g/lane) were treated with Me 2 SO (control) or else forskolin and IBMX, and total cell lysates were immunoblotted with anti-Drp1 phospho-Ser 637 (pS637) antibodies. Protein phosphatases were added as indicated. Equal protein loading was monitored by immunoblotting for total endogenous Drp1.
GTPase Activity Assays-Calmodulin-binding peptide (CBP)-Drp1 fusion proteins were purified from Escherichia coli BL21 (DE3) expressing CBP-tagged wild-type Drp1, Drp1 S637A , or Drp1 K38A as described previously (15). Affinity-purified CBP-Drp1 fusion proteins were dialyzed against assay buffer (20 mM HEPES (pH 7.2), 2 mM MgCl 2 , and 1 mM dithiothreitol). The reaction mixture for the GTPase assay at 25°C included 1.45 g of dialyzed CBP-Drp1 protein with 0.05% bovine serum albumin and 165 nM [␣-32 P]GTP EasyTides (3000 Ci/mmol; PerkinElmer) in assay buffer. For assay of in vitro phosphorylated CBP-Drp1 fusion proteins, 3.0 g of CBP-Drp1 was first incubated with 50 ng of purified PKA catalytic subunit in 10 mM HEPES (pH 7.0) with 20 mM MgCl 2 and 500 mM ATP for 30 min at 30°C before diluting into GTPase assay buffer. Samples of the reaction mixtures were spotted onto polyethyleneimine cellulose on polyester thin layer chromatography (TLC) plates (Sigma-Aldrich) at various times. Guanine nucleotides were separated by ascending chromatography in 1.0 M formic acid: 1.2 M LiCl. [ 32 P]GDP and [ 32 P]GTP spots were identified, and intensities were quantified using a Storm 860 PhosphorImager with ImageQuant software (GE Healthcare). GTPase activity was expressed as the ratio of GDP to total guanine nucleotides (GDP ϩ GTP) at each time point. All data points represent the averages of at least three independent experiments.
Immunostaining-HeLa cells were grown on glass coverslips and transfected with Myc-Drp1 as described previously (15). Drp1 mutants were produced using the QuikChange method (Stratagene, La Jolla, CA). After 36 -48 h of transfection, cells were washed with PBS and fixed with 4% paraformaldehyde in PBS for 20 min. After three washes with PBS, cells were incubated in blocking buffer (5% horse serum, 0.1% bovine serum albumin, and 0.1% Triton X-100 in PBS) for 1 h. Goat anti-Myc epitope (1:500; Bethyl Laboratories, Montgomery, TX) and mouse monoclonal anti-cytochrome c (1:500; Pharmingen) antibodies were incubated in blocking buffer overnight at 4°C. After three washes with PBS, Alexa Fluor 488 and/or Alexa Fluor 568 secondary antibodies (1:500; Invitrogen) were added for 1 h. After three washes with PBS, coverslips were mounted using Gel/Mount (Biomeda, Burlingame, CA). Images were acquired using a Zeiss LSM-510 confocal microscope. For determination of mitochondrial morphology, three trials were performed, with 100 transfected HeLa cells in each trial classified into normal, elongated and fragmented mitochondrial groups. Where indicated, HeLa cells were incubated with 20 M forskolin and 10 M IBMX for 10, 20, or 30 min before fixation and subsequent immunostaining for cytochrome c.

RESULTS AND DISCUSSION
Drp1 Is Phosphorylated by PKA at Ser 637 -We examined whether Drp1 is a phosphoprotein under basal conditions by labeling HeLa cells overexpressing Myc-tagged Drp1 with [ 32 P]orthophosphate. After immunoprecipitation with anti-Myc antibodies, we clearly detected the Drp1 phosphoprotein at ϳ82 kDa. We tested multiple different kinase activators and found that forskolin and IBMX, activators of PKA, significantly increased phosphorylation levels (Fig. 1A). To determine whether this stimulated phosphorylation occurred at the same or different sites from the basal phosphorylation, we performed two-dimensional tryptic phosphopeptide mapping. This revealed that there was a single novel site of phosphorylation in response to PKA activation that was distinct from sites of basal phosphorylation (Fig. 1B).
To confirm that the phosphorylation was directly mediated by PKA, we performed in vitro phosphorylation assays with [␥-32 P]ATP using purified PKA in combination with GST-Drp1 fusion proteins, and we observed a prominent increase in Drp1 phosphorylation. A number of fusion proteins comprising Drp1 truncation constructs were tested, narrowing down the region sufficient for PKA phosphorylation to residues 490 -646; phosphoamino acid analysis identified the phosphorylated residue(s) as serine (data not shown). One of the sites within this region, Ser 637 , comprised a consensus sequence for PKA phosphorylation (-R(R/K)X(S/T)-). Site-directed mutagenesis of this site, S637A, resulted in loss of PKA-mediated phosphorylation in vitro, both with full-length GST-Drp1 S637A and with GST-Drp1(490 -646) S637A (Fig. 1C and data not shown). After two-dimensional tryptic phosphopeptide mapping, the PKAstimulated phosphopeptide from wild-type GST-Drp1 WT migrated essentially the same as that identified in intact cells. It was selectively abolished in the GST-Drp1 S637A mutant, confirming the PKA phosphorylation site as Ser 637 (Fig. 1D). Finally, we developed anti-phosphopeptide antibodies against phospho-Ser 637 in Drp1; the phosphorylation site was detected in Drp1 WT only after in vitro phosphorylation with PKA, and FIGURE 2. Drp1 S637D phosphomimetic mutation impairs intramolecular Drp1 interactions, but not intermolecular interactions. A, schematic of Drp1 structure; boundary amino acids of constructs used are indicated. Intramolecular interactions between Drp1 fragments were assayed using the HIS3 reporter (10-fold yeast dilutions shown) and ␤-galactosidase activity (assessed by determining the time for colonies to turn blue in X-gal filter lift assays: ϩϩϩ, Ϫ30 min; ϩϩ, 30 -60 min; ϩ, 60 -120 min; Ϫ, no significant activity). N, N terminus; C, C terminus; B, insert B. AH, ␣-helical. B, strengths of intermolecular interactions in the yeast two-hybrid system between fulllength Drp1 WT and Drp1 S637D were assayed as in A.
not at all with Drp1 S637A , confirming the specificity of the antibodies for phospho-Ser 637 . Also, the immunoreactive signal was markedly diminished after treatment with broad spectrum protein phosphatases (Fig. 1E). Similarly, in extracts from HeLa cells treated with activators of PKA, there was appearance of a phospho-Ser 637 band at ϳ82 kDa, representing endogenous phosphorylated Drp1, that was similarly sensitive to protein phosphatase treatment (Fig. 1F).
Importantly, the consensus site for PKA phosphorylation at Ser 637 of human Drp1 is highly conserved in a variety of species (supplemental Fig. S1). This evolutionary conservation suggests that phosphorylation at this site may be a common means of regulation Drp1 function across species.
Phosphorylation at Ser 637 Impairs Drp1 Intramolecular Associations and GTPase Activity-Since the Ser 637 phosphorylation site is within the GED domain that has been shown to mediate both intramolecular and intermolecular interactions (15), we examined whether this modification changed Drp1 interactions using a phosphomimetic substitution, S637D, in yeast two-hybrid tests. We found that this mutation significantly inhibited the intramolecular interactions between GED and GTP-binding/middle domains ( Fig. 2A) but not intermolecular interactions between full-length Drp1 proteins (Fig. 2B).
Since this intramolecular interaction has previously been shown to be important for GTPase activity (15), we examined the effects of Ser 637 phosphorylation on Drp1 GTPase activity. Using in vitro GTPase assays with purified CBP-Drp1 fusion proteins, we found that PKA phosphorylation caused a significant reduction of GTPase activity (Fig. 3A). Similarly, the phosphomimetic GST-Drp1 S637D exhibited decreased GTPase activity as well (Fig. 3B), although not so robust an effect as that seen with the more highly negatively charged phospho-Ser 637 modification. This loss of activity likely occurs through effects of the phospho-Ser 637 modification and S637D mutation on the intramolecular GED domain interaction with GTP-binding/middle domains since there are no effects on oligomerization as assessed by yeast two-hybrid assays (Fig. 2B) and chemical cross-linking studies (data not shown). Indeed, recent studies examining dynamin have suggested that the middle domain is critical for both tetramerization and higher-order self-assembly (16).
Importantly, our data identify a key regulatory mechanism for the GED domain interaction with middle and GTP-binding domains. Although this interaction has been intensively studied for other dynamin superfamily members, phosphorylation at Ser 637 in Drp1 represents the first demonstration of dynamic regulation of this intramolecular interaction with functional changes in GTPase activity.
Phosphorylation of Drp1 at Ser 637 Inhibits Mitochondrial Division-We examined the effects of IBMX and forskolin, which increase cAMP levels and activated PKA, on mitochondrial morphology in HeLa cells 10 -30 min after application but saw no consistent morphology changes. Since cAMP has pleio- tropic effects in cells that may mask any specific effects of Drp1 phosphorylation on mitochondrial morphology, we next evaluated the effects of the phosphomimetic S637D mutation on mitochondrial morphology. Overexpression of this mutant form resulted in mitochondria that were more elongated (Fig.  4). Interestingly, another GED mutation in Drp1 that decreases GTPase activity, K679A, inhibited mitochondrial fission as well (15), reminiscent of our results with Drp1 S637D . Importantly, these effects on mitochondrial morphology are less prominent than those seen upon overexpression of a well characterized Drp1 mutation, K38A, which lacks detectable GTPase activity (15). Thus, effects on mitochondrial fission seem to mirror the extent of changes in GTPase activity for these mutations. Overexpression of Drp1 S637A that cannot be phosphorylated by PKA showed no clear differences in mitochondrial morphology, consistent with a number of reports that found no increase in fission upon overexpression of wild-type Drp1.
A very recent study identified a cdk1/cyclin B phosphorylation site at a nearby residue, Ser 616 in human Drp1 splice variant 1 (Ser 585 in the variant reported in Ref. 7). However, a phosphomimetic substitution, S616D, did not change the intramolecular GTP-binding/middle domain interactions with the GED domain (data not shown), nor did it affect GTPase activity or mitochondrial morphology in cells (7). Thus, Ser 616 phosphorylation likely regulates Drp1 through other mechanisms.
Functional Implications-In this study, we have shown that PKA phosphorylates Drp1 on Ser 637 and inhibits its GTPase activity, likely through changes in the intramolecular associa-tion among the GED and GTP-binding/middle domains of Drp1, although higher-order oligomerization may also be affected. This represents the first reported Drp1 modification to have a direct effect on Drp1 GTPase activity.
Furthermore, in keeping with the impaired Drp1 GTPase activity, we have found that a phosphomimetic substitution, S637D, inhibits mitochondrial division. However, the functional effects of GED domain mutations in general have been controversial in the dynamin superfamily since some studies have seen facilitation of fission in response to changes in conserved amino acid residues for both dynamin and Drp1 (17,18), whereas others have not (15,19). On the other hand, since very recent studies support a two-step model for dynamin function (20), with an early regulatory GTPase-like function preceding a late, assembly-dependent step during which GTPase activity is required for fission, different GED mutations may give different results. In addition, the phosphomimetic substitution is a permanent one that lacks the dynamic aspects of phosphorylation in vivo, and thus, we suspect that the timing of phosphorylation would be critical. Further identification of the temporal aspects of the phosphorylation/dephosphorylation cycle will help unravel the cellular role of this modification.