A membrane-inserted structural model of the yeast mitofusin Fzo1

Mitofusins are large transmembrane GTPases of the dynamin-related protein family, and are required for the tethering and fusion of mitochondrial outer membranes. Their full-length structures remain unknown, which is a limiting factor in the study of outer membrane fusion. We investigated the structure and dynamics of the yeast mitofusin Fzo1 through a hybrid computational and experimental approach, combining molecular modelling and all-atom molecular dynamics simulations in a lipid bilayer with site-directed mutagenesis and in vivo functional assays. The predicted architecture of Fzo1 improves upon the current domain annotation, with a precise description of the helical spans linked by flexible hinges, which are likely of functional significance. In vivo site-directed mutagenesis validates salient aspects of this model, notably, the long-distance contacts and residues participating in hinges. GDP is predicted to interact with Fzo1 through the G1 and G4 motifs of the GTPase domain. The model reveals structural determinants critical for protein function, including regions that may be involved in GTPase domain-dependent rearrangements.


Supplementary
Supplementary Figure 2 | Target-template alignment using the Clustal Omega method without the first one hundred N-terminal residues from Fzo1. The set of 43 sequences from the cyanobacteria (see Methods) were aligned using Clustal Omega 46 . Subsequently, the generated multiple alignment was merged with the sequence from the target Fzo1 without its first one hundred N-terminal residues, using M-Coffee 45 .
Supplementary Figure 3 | Target-template alignment using T-coffee considering the whole sequences. The set of 43 sequences from the cyanobacteria (see Method) were aligned using T-coffee 47 . Subsequently, the generated multiple alignment was merged with the sequence from the target Fzo1, using M-Coffee 45 .
Supplementary Figure 4 | Target-template alignment using T-coffee without the first one hundred N-terminal residues from Fzo1. The set of 43 sequences from the cyanobacteria (see Methods) were aligned using T-coffee 47 . Subsequently, the generated multiple alignment was merged with the sequence from the target Fzo1 without its first one hundred N-terminal residues, using M-Coffee 45 .  Table  3). The transmembrane segment is indicated in gray . The values are computed on the alpha carbon atoms with respect to the model after the minimization. In every plot the structure is colored accordingly and represents the last frame from the corresponding simulation time. In the upper panels the structures are depicted in tube representation of varying thickness as a function of the Root Mean-Square Fluctuation (RMSF). Fzo1.III 4.6 (6.5) ± 2.68

Supplementary
For each protein segment the number of residues is indicated. Statistics are computed on the Cα atoms for each fragment and the values relative to the segment C are calculated without the contribution of the transmembrane domain. The model after the minimization was used as a reference structure. Values are in Å: Mean (last frame) ± standard deviation.

Supplementary Figure 8 | (a)
Ab initio prediction for the TM helical dimer in Fzo1 using the PREDDIMER server. The models are ordered from the left according to their Fscor computed by PREDDIMER 24 , 3.113, 3.100 and 2.647, respectively, with associated crossing-angle χ of 119.7°, 175.1° and -129.7°, respectively. The glycines within the motif GxxxG are in the space-filled representation and residues Lys716 and Ser746 are depicted in stick form. (b) Prediction of the BDLP template orientation with respect to a membrane. The crystal structure PDB-Id 2J68 16 has been submitted to the PPM web server 67 . The structure is represented in ribbon mode, the GDP nucleotide in a space-filled representation. The N-(res 2-571) and C-terminal (res 607-695) regions exposed outside of the membrane are in cyan and orange , respectively. The paddle region (res 572-606) is depicted in yellow . The membrane layer is represented by dummy atoms. The predicted embedded residues are 574, 577, 581 and 583, suggesting that BDLP may be a peripheral membrane protein. The Table shows for each monomer in the transmembrane segment which are the corresponding partners between the two TM helices. For each residue position the number of contacts identified are summed and the persistence along each trajectory is indicated. The analysis is conducted without considering the H and the polar atoms N and O. The single common interaction between the replicas is highlighted. TM1 (res 706-726); TM2 (res 737-757). Residues interacting through H-bonds over 50% of persistence are indicated. The common interactions between the replicas are highlighted. POPE, palmitoyl-oleoyl-phosphatidylethanolamine; POPC, palmitoyl-oleoyl-phosphatidylcholine.

Supplementary Figure 9 | (a) Cartoon representation of the Fzo1 model and its functional domains. (top)
Residue numbers delimiting the domains, (bottom) secondary structure elements are annotated with the Fzo1 mutations performed in this study. The Fig. 4 in main text has been replicated in (a) and extended with data on available mutants across Fzo1 functional domains from the literature 9,11,13,15,33,86 . The point mutations performed in this study are highlighted by larger font size at the very bottom. The color code is cyan , loss of function (LOF), maroon , yeast phenotype is analogous to the wild-type, orange , point mutations that cause a severe LOF only when associated; magenta , residue involved in post-translational modification. Residues considered for the charge swap strategy are connected by a bar. Putative hinge regions are indicated by blue arrows. N-and C-terminal halves are indicated above the secondary structure plot ( green and pink , respectively). The topology diagram was generated with the HERA program 83 . (b) Level of structuration discussed in this study for the Fzo1 model. The structures refer to the model after the equilibration phase presented also in Fig. 1d. Left, the annotation according to the N-(res 1-415, pink ) and C-terminal (res 416-855, green ) halves. Right, subdomains across the hinge regions considered in this study.   Distances are in Å. The residue at position a on one chain is packed against the corresponding residue at position d on the other chain to form a-d packing. The standard deviation is indicated.  HR2 and α18 ( white ). The latter does not exhibit heptad periodicity and in our model is in close contact with the HR1 towards its a-d positions ( green surface). The PCOILS algorithm 53 has been used to predict heptad periodicity (see Methods). Donor-acceptor distances are calculated with LigPlot+ 52 using a distance cut-off of up to 3.5 Å.  Homologous residues were identified in the initial target-template alignment (see Methods) as well as after the superposition of the structures indicated. The Fzo1 model represents the centroid of Fzo1.I (see the main text). Residues were subdivided in the three categories discussed in the text. In blue , the analogous interactions, in cyan , interactions retained on the same GDP atom. H-bond were identified using UCSF Chimera 61 (distance cut-off of 3.5 Å and up to 30 degree off-axis angle).

Supplementary Figure 14 | Coordination of the bound magnesium in the GDP binding site.
( a ) Fzo1 model after the minimization phase, in which the Ser201 (Ser89 in human Mfn1) directly coordinates the cation with a distance of 1.96 Å, as suggested also in the fragment crystal structure from Mfn1 19 . Similarly, the homologous Ser41 in dynamin was proposed to act in concert with Lys44 (Lys200 in Fzo1) and Thr65 (Thr221 in Fzo1) in coordinating the bound magnesium, to stabilize the developing charge in the transition state of GTP hydrolysis 37, 87 . We thus added a magnesium ion in the nucleotide binding site of the Fzo1 model (see Methods in main text), which revealed that Ser201 of Fzo1 possibly plays a role in Mg 2+ binding. ( b ) The structure represents the result of the cluster analysis from the trajectory Fzo1.I. During the simulation time the Ser201 oxydril group was subsequently stabilized over the average of 4.2 ± 0.19 Å between the trajectories. In particular, we observed a change in the coordination from 3 oxygens (1 from α and 2 from β phosphates, see a) to 2 oxygens (1 from α and 1 from β phosphates, see b) with the remaining coordinations being supplied by water molecules. Furthermore, analysis of minimum distances in combination with the number of contacts showed that the water molecules coordinating the Mg 2+ at the equilibration phase, remained in place over the simulation time for each replica. Although the placement of the bound magnesium was suggested by available crystal structures (see Method), a template to correctly position the Mg 2+ ion for the mitofusin Fzo1 is currently lacking. However, our results suggest a role for this cation in ligand accommodation within the binding pocket as recently suggested for human Mfn1 18, 19 .