New interfaces on MiD51 for Drp1 recruitment and regulation

Mitochondrial fission is facilitated by dynamin-related protein Drp1 and a variety of its receptors. However, the molecular mechanism of how Drp1 is recruited to the mitochondrial surface by receptors MiD49 and MiD51 remains elusive. Here, we showed that the interaction between Drp1 and MiD51 is regulated by GTP binding and depends on the polymerization of Drp1. We identified two regions on MiD51 that directly bind to Drp1, and found that dimerization of MiD51, relevant to residue C452, is required for mitochondrial dynamics regulation. Our Results have suggested a multi-faceted regulatory mechanism for the interaction between Drp1 and MiD51 that illustrates the potentially complicated and tight regulation of mitochondrial fission.


INTRODUCTION
It is well established that Drp1 either in the presence of GTP or GMP-PNP 23 forms oligomeric assemblies 19,20 . To further clarify that whether the apparent 24 greater binding between MiD51 and Drp1 in the presence of GTP or GMP-PNP 25 relies on the oligomerization of Drp1, we made a series of Drp1 mutations and 6 tested their effects on binding to MiD51. In previous studies 21-25 , Drp1 residues 1 G350, G363, R376, A395 and G401 play important roles in its polymerization. We 2 designed a series of mutants where these residues were changed to Asp and 3 monitored their ability to polymerize using gel filtration. We found that G350D and 4 A395D behaved differently from wild-type Drp1 in gel filtration (Fig. 1D), 5 suggesting defective oligomerization. Similar defects in oligomerization were also 6 observed for the AAAA mutant form of Drp1 ( 401 GPRP 404 →AAAA), 23 . In addition, 7 the compound mutants G350D/A395D, G350D/AAAA and A395D/AAAA showed 8 a severe reduction in oligomerization. Our results indicate that the A395, G350 9 and GPRP (401-404) residues are involved in the polymerization of Drp1, 10 consistent with previous studies. We next assessed the interactions between 11 MiD51 133-463 and Drp1 mutants. We found that compared to wild type，Drp1 12 oligomerization mutants G350D, A395D, AAAA, G350D/A395D, G350D/AAAA 13 and A395D/AAAA have reduced affinity for MiD51   (Fig. 1B), confirmed by 14 quantification (Fig. 1C), and the mutant proteins generally have the same binding 15 affinity for MiD51  in the presence of different nucleotides (Fig. 1E). We also 16 designed other Drp1 mutant proteins, targeting residues not responsible for 17 oligomerization, such as K38A, and phosphorylation-mimic mutants S579D and 18 S600D (related to S616 and S637 in Drp1 isoform 1). We found that these three To understand how MiD51 interacts with Drp1 during mitochondrial fission, 25 we performed crystal structure studies and solved two types of MiD51 crystal 26 structures under apo conditions (Table S1). Type I contains the cytosolic domain 27 7 MiD51 129-463 , and the crystal space group is P4 1 2 1 2 with one molecule per 1 asymmetric unit. Type II contains the fragment MiD51 133-463 , which was expressed 2 as a C-terminal 6×His fusion protein, and the crystal space group is P1 with two 3 molecules per asymmetric unit. The overall structure consists of a central β-strand 4 region flanked by two α-helical regions ( Fig.2A) and looks similar to NTPase 5 family crystal structures published by two groups 17,18 . The C domains of the Type 6 I and Type II crystal structures are almost identical, but there is a tiny structural 7 conformation change in the N domain (Fig. S1B), with a RMSD (root mean square 8 deviation) variation of 1.14 Å for 329 aligned C α atoms. By comparison, we found 9 that all of the released crystal structures of MiD51 from PDB (Table S2), which 10 include different nucleotide forms (Apo, ADP or GDP), lack distinct conformational 11 changes when compared to the Type I and Type II crystal structures, with RMSD 12 variations ranging from 0.97 to 1.88 Å ( Fig. S1C and Table S3). We are not sure 13 about the significance of such a small conformational change, which is probably 14 due to different constructs, crystallization conditions and crystal packing. And the 15 ADP/GDP binding sites are almost identical, which implies the structural rigidity of 16 MiD51. It was reported that MiD51 can form dimers primarily based on the crystal 17 packing of MiD51, but we did not observe such packing in our two crystal types 18 (Fig. S1D). Further studies are needed to determine the oligomer state of MiD51, 19 and we describe these studies in a later section. binding. Therefore, we performed a systematic analysis of MiD51 mutants (Table   23 S4) to identify which region is involved in the interaction with Drp1. Initially we 24 designed a series of mutant proteins, each containing a cluster of three or four 25 mutated residues. We then used a pull-down assay to test the affinity of each 26 MiD51 mutant for Drp1. These assays indicated that six MiD51 mutations disrupt 8 the interaction with Drp1 ( Fig. S2A and B). Next, we did a second round of point 1 mutations of MiD51. We found eight mutant proteins with decreased affinity for 2 Drp1 ( Fig. 2C and D, Fig. S2C and D). There was almost no conformational 3 change in the mutant proteins compared to wild type MiD51 based on circular 4 dichroism (CD) spectroscopy and thermal shift assay ( Fig. S2E and F). 5 We analyzed the distribution of these sites and found that the eight mutations 6 are located in two areas. The first area contains four residues, R234, Y240, F241 7 and R243, which are located on an exposed loop between β4-α4 (Fig. 2B). When 8 these residues are substituted with alanine or glutamate (R234E, Y240A, F241A 9 and R243E), the resulting mutant proteins have modest or serious decreases in 10 Drp1 binding affinity, as confirmed by quantitation ( Fig. 2C and D). This suggests 11 that the exposed loop is a main determinant for Drp1 binding. We name this area 12 DBS1 (Drp1 Binding Site One) (Fig. 2B), which is consistent with previous studies 13 17,18 . The second area contains the amino acids E420, D444, Y448 and Y451 (Fig.   14   2B). Mutation of these residues by substituting with alanine, or by substituting 15 aspartate and glutamate with arginine (E420R, D444R, Y448A and Y451A), 16 results in a more dramatic effect on the ability of MiD51 to bind Drp1, and in some 17 cases even abolishes binding ( Fig. 2C and D). We define this area as DBS2 (Drp1 18 Binding Site Two), which is located on α12 and α13 in the C domain and forms a 19 surface for Drp1 binding (Fig. 2B). Therefore, MiD51 requires DBS2, a surface in 20 the C domain, to cooperate with Drp1 binding. An amino acid sequence alignment 21 of MiD51 and MiD49 proteins from different species reveals that these eight DBS1 22 and DBS2 residues are highly conserved ( Fig. 2E and Fig. S2G). Based on the 23 crystal structure of MiD49 18 , these eight residues also form an exposed loop in the 24 N domain and a surface in the C domain for Drp1 binding. Mitochondrial fission receptors, such as Fis1 and Mff, form dimers to perform 1 their functions in mitochondrial fission 12 . A previous study reported that MiD51 2 could form a dimer under non-reducing conditions 26 , suggesting that MiD51 may 3 form a dimer via an intermolecular disulfide bond between cysteines in the region 4 of residues 49 to 195 26 . But based on the crystal structure, another study found 5 that MiD51 forms a dimer via electrostatic interactions in the N-terminal helix, and 6 the dimerization is very important for its function in mitochondrial fission 18 .
7 Surprisingly, we did not observe a similar surface mediating the dimerization of 8 MiD51 in our crystal packing. Therefore we experimentally determined whether 9 MiD51 forms dimers. Using a time course assay where the level of dimer 10 formation was quantified every twenty-four hours, we determined that MiD51 133-463 11 does form dimers and that the level of dimerization continues to increase over 12 time ( Fig. 3A and B). These results correlate well with the results of Zhao et al 26 .  To determine which residue mediates MiD51 dimerization, we analyzed the 17 MiD51 sequence and found that there are seven cysteines, but only two, C165 18 and C452, are exposed on the protein surface. When C452 was substituted by showed dimerization similar to wild type (Fig. 3E, F and G). This suggests that 23 C452, not C165, is the residue that forms the disulfide bond. We also found that

25
The changes in Drp1 conformation and oligomerization upon GTP binding, 26 hydrolysis and release, is associated with the procession of mitochondrial fission Although the AAAA mutant has the capacity to bind MiD51, its binding affinity is 10 reduced compared to wild type as we showed ( Fig. 1C and D).

11
We then gained significant insight into the interaction between MiD51 and suggesting that DBS2 is much more important than DBS1 for Drp1 recruitment. 23 We know that the interaction between MiD51 and Drp1 changes during the 24 process of mitochondrial fission, so the interaction may need more than one 25 binding site between MiD51 and Drp1. In addition to the exposed loop in N 26 domain, we have determined that another region in MiD51 makes direct contact 12 with Drp1, and the residues are highly conserved between MiD51 and MiD49. It 1 seems likely that the two binding regions on MiD51 are responsible for the 2 complicated interaction with Drp1 during the process of mitochondrial fission. But 3 the precise role of MiD51 in Drp1 polymerization and mitochondrial fission still 4 remains elusive. 5 We also determined that MiD51 forms dimers via an intermolecular disulfide 6 bond between C452 residues located in the C terminal region, although the and structure refinements are listed in            For thermal shift assay the wild type and mutant 6 His-MiD51 133-463 protein samples were diluted to 1 mg/ml in the buffer containing 20 mM HEPES pH 7.5, 150 mM NaCl. The fluorescent dye SYPRO Orange (Invitrogen) was added into the sample solution by ~1,000 fold of dilution. 20 µl of mixture in a PCR tube was heated up from 25°C to 75°C with the step of 1°C per min. The fluorescence of the mixture was measured by using a RT-PCR device (Corbett 6600). The melting temperature (Tm) was estimated as the temperature corresponding to the minimum of the first derivative of the protein denaturation curve. All the measurements were repeated three times.
For CD spectroscopy assay, the wild type and mutant 6 His-MiD51 133-463 protein samples were diluted to 0.2 mg/ml in buffer containing 10 mM Na 2 HPO4, 1.8 mM KH 2 PO4, 140 mM NaCl, 2.7 mM KCl, 1 mM DTT, pH 7.4. The spectra were recorded over the wavelength from 200 nm to 260 nm with a bandwidth of 1 nm and 0.5 s per step by using CD spectrometer (Chirascan-plus, Applied photphysics). All the measurements were repeated three times and the spectrum data were corrected by subtracting the buffer control.