NMR studies on the interactions between yeast Vta1 and Did2 during the multivesicular bodies sorting pathway

As an AAA-ATPase, Vps4 is important for function of multivesicular bodies (MVB) sorting pathway, which involves in cellular phenomena ranging from receptor down-regulation to viral budding to cytokinesis. The activity of Vps4 is stimulated by the interactions between Vta1 N-terminus (named as Vta1NTD) and Did2 fragment (176–204 aa) (termed as Did2176–204) or Vps60 (128–186 aa) (termed as Vps60128–186). The structural basis of how Vta1NTD binds to Did2176–204 is still unclear. To address this, in this report, the structure of Did2176–204 in complex with Vta1NTD was determined by NMR techniques, demonstrating that Did2176–204 interacts with Vta1NTD through its helix α6′ extending over the 2nd and the 3rd α-helices of Vta1NTD microtubule interacting and transport 1 (MIT1) domain. The residues within Did2176–204 helix α6′ in the interface make up of an amino acid sequence as E192′xxL195′xxR198′L199′xxL202′R203′, identical to type 1 MIT-interacting motif (MIM1) (D/E)xxLxxRLxxL(K/R) of CHMP1A180–196 observed in Vps4-CHMP1A complex structure, indicating that Did2 binds to Vta1NTD through canonical MIM1 interactions. Moreover, the Did2 binding does not result in Vta1NTD significant conformational changes, revealing that Did2, similar to Vps60, enhances Vta1 stimulation of Vps4 ATPase activity in an indirect manner.

The residues D193-G204, corresponding to the residues marked in red and green in Fig. 1F, whose cross peaks finally disappeared during titration, were highlighted. (E) The backbone view of the ensemble of 20 lowestenergy free Did2 176-204 structures. (F) Ribbon representation of free Did2 176-204 , the residues whose cross-peaks disappeared swiftly, disappeared gradually and almost unchanged were highlighted in red, green and grey colors, respectively. report, as shown in Fig. 1A, could enhance Vps4 ATPase activities as the full-length Did2 27,29 . In this report, to investigate how Did2 interacted with Vta1NTD, we first measured the binding affinity of Vta1NTD to Did2 176-204 by isothermal titration calorimetry (ITC) assay (K D = 12.8 ± 1.0 μ M, the number N = 1.16 ± 0.0266) (Fig. 1B), and then determined the solution structure of Vta1NTD in complex with Did2 176-204 . This structure revealed that Did2 176-204 bound to Vta1NTD through canonical type 1 MIT-interacting motif (MIM1) interactions.

Results and Discussion
The suggested MIM region of Did2 176-204 forms an α-helix conformation. Secondary structure prediction and perceived structural homology to ESCRT-III protein Vps24/CHMP3 suggest that Did2 176-204 corresponds to the 6 th helix within the Did2 structure 37 . This helix conformation was confirmed by our following NMR studies. We first assigned its NMR signals based on two-dimensional (2D) 1 H-1 H TOCSY and NOESY, three-dimensional (3D) 15 N-edited HSQC-TOCSY spectra. In its 2D NMR 1 H-15 N HSQC experiment (Fig. 1C), the cross peaks belonging to 27 residues were assigned, except the N-terminal residues N 176 and P 178 (without amide proton). We then determined its NMR solution structure using 191 distance restraints from NOE and 26 hydrogen bonds. Finally, 20 structures with the lowest-energy could be well fitted (Fig. 1E) with the RMSD values of 0.15 ± 0.06 Å for the backbone atoms, and of 0.88 ± 0.09 Å for all heavy atoms in secondary structure region. The Ramachandran plot displays 93.1% of the residues in the most-favored regions, 4.0% residues in additionally allowed regions, and 2.9% residues in generously allowed regions, indicating that the structures are acceptable. The solution structures demonstrate that the region of Did2 187-204 aa forms an α -helix, consistent with the observation in the crystal structure of complex Did2-Ist1 27,29 .
To study how Vta1NTD interacts with Did2 176-204 , we performed NMR titration experiments, in which Vta1NTD was titrated into Did2 176-204 at mole ratios (Vta1NTD vs Did2 176-204 ) of 0.1:1, 0.2:1, 0.4:1, 0.6:1, 0.8:1 and 1.2:1 in NMR buffer conditions. The cross peaks belonging to residues R 200 A 201 L 202 disappeared even at mole ratios of 0.1:1 and 0.2:1, while the cross peaks belonging to D193-L199 and R 203 G 204 became weaker and weaker upon the concentration of Vta1NTD being increased. At the mole ratio of 1.2:1, the cross peaks belonging to residues V177-E192 almost unchanged (Fig. 1D). These observations suggest that the residues D193-G204 are involved in the interactions, and that the MIM sequence of Did2 176-204 bind to Vta1NTD. This hypothesis was consistent with our observation in the complex structure (discussed below), where the residues in the region of 187-204aa interact with Vta1NTD MIT1 domain.
NMR structural determination of complex Vta1NTD-Did2 176-204 . Using two basic sets of NMR samples: 1) 13 C and 15 N isotope double labeled or 13 C, 15 N and 70% 2 H triple-labeled Vta1NTD mixed with unlabeled Did2 176-204 at the stoichiometric ratio of 1:1.2, 2) 13 C and 15 N isotope labeled Did2 176-204 mixed with unlabeled Vta1NTD at stoichiometric ratio of 1:1.2, and by running a series of 2D and 3D NMR experiments, in total, more than 94% NMR signals of the main-chain and 95% NMR signals of the side-chain atoms of the residues in the complex were assigned. The inter-molecule NOEs were correctly assigned by confirming signals observed in 3D 13 C-F1 edited, 13 C/ 15 N-F3 filtered NOESY spectra acquired on both complex samples. The NMR chemical shift changes of Vta1NTD backbone atoms 1 H and 15 N in the absence of and in the presence of Did2 176-204 reveal that Did2 176-204 addition mainly induces amide 15 N and 1 H chemical shifts variation of the residues in Vta1NTD MIT1 ( Fig. 2A), suggesting that Did2 176-204 binding sites locate in this region. This observation accords with the previous analysis on chemical shift mapping of CHIMP1B-binding site on LIP5(MIT) 2 35 , and with the analysis of electrostatic surface of Vta1NTD, which shows that Vta1NTD MIT1 is more negatively charged and more hydrophobic than its MIT2 (Fig. 2B), suitable for positively charged and hydrophobic Did2 176-204 binding.
The solution structure of Vta1NTD-Did2 176-204 complex was then determined by a conventional heteronuclear NMR method using 15 N-labeled or 13 C/ 15 N-labeled proteins. In total, 2730 distance restraints from NOE (including 36 inter-molecular NOEs), 236 hydrogen bonds and 676 dihedral angle restraints for backbone ϕ and ψ angles were used to calculate solution structure. The best-fit superposition of the ensemble of the 20 lowest-energy structures represented in Fig. 2C was displayed with the RMSD values of 0.82 ± 0.17 Å for the backbone (N, C α and CO) atoms and 1.14 ± 0.15 Å for all heavy atoms in the well-ordered second structure regions. The Ramachandran plot displays 97.8% of the residues in the most-favored regions and 2.1% residues in additionally allowed regions (Table 1), indicating the structures are reasonable.
Overall complex structure. The complex Vta1NTD-Did2 176-204 structure shows that the bound Vta1NTD still has two MIT domains, each of them is composed of three α -helices (MIT1: helices α 1, α 2 and α 3; MIT2: helices α 5, α 6 and α 7; respectively), almost similar to those observed in its free state and in its complex with Vps60 128-186 29,33-36 . The backbone atoms belonging to MIT1 and MIT2 regions of bound Vta1NTD have RMSD values of 1.7 Å and 1.8 Å with those of free Vta1NTD (Fig. 2D), respectively, indicating that Did2 176-204 binding does not induce overall major conformational changes in Vta1NTD. The linker (64-85aa) between MIT1 and MIT2 domains is well ordered, where residues 66-69 become an α -helix, and residues 73-84 adopt a longer helical structure (here we called it as helix α 4) ( Fig. 2E) 33,34 . The helix α 4 is much longer in Vta1NTD-Did2 176-204 than that in free Vta1NTD crystal structure, but similar to the observation in NMR structure of Vta1NTD bound to Vps60 128-186 [33][34][35] . This observation is consistent with our previous secondary structure prediction of free Vta1NTD using the programs CSI 38 and TALOS 39,40 based on the assignments of NMR signals belonging to the backbone atoms of Vta1NTD. In contrast, this linker in free Vta1NTD crystal structure adopts largely random-coil structure with only a one-turn α -helix occurring at residues 80-84. Particularly, residues 65-75 appear to be disordered in the structure of free Vta1NTD. Thus, this conformational change might be caused by the interactions between helix α 4 (73-84) of Vta1NTD and Vps60 128-186 or Did2 176-204 , or by stacking during free Vta1NTD crystallization.
SCIeNTIfIC REPORTS | 6:38710 | DOI: 10.1038/srep38710 Within our expectation, in the current complex structure, the Did2 176-204 folds into one α -helix (denoted as α 6′ helix hereafter), which is involved in the interaction with the first MIT domain of Vta1NTD (Fig. 2E). The bound Did2 176-204 adopts an overall rod-like helix structure with a flexible loop in its N-terminus (similar to free Did2 176-204 with an RMSD value of 0.5 Å for the backbone atoms in secondary structure region). The helix consists of residues 187′ -203′ , a little longer than that (residues 184′ -198′ ) observed in LIP5NTD-CHIMP1B  The helices are numbered. In (C) and (E), the MIT1 and MIT2 domains of Vta1NTD, the linker between MIT1 and MIT2, and Did2 176-204 were displayed in blue, green, red and orange colors, respectively. complex structure 36 . The Did2 176-204 α -helix sits on the surface groove formed by the 2 nd and the 3 rd helices of Vta1NTD MIT1 in a mode similar to that observed in Vps4-CHMP1A complex structure 20 and that observed in human LIP5NTD-CHMP1B and LIP5NTD-CHMP1B-CHMP5 complex structures 36 . The Vta1-Did2 176-204 complex buries a total of approximately 1039 Å 2 surface area at the interface, close to that (1115 Å 2 ) observed in the complex structure of Vps4-CHMP1A 20 , but much larger than that (~624 Å 2 ) observed in the complex structures of LIP5NTD-CHMP1B and LIP5NTD-CHMP1B-CHMP5 36 .
Mutations were introduced to these observed binding sites to test the importance of the residues to the overall stability of the complex. As shown in Fig. 3F, in vitro GST pull-down experiments demonstrate that all the single alanine or glycine substitutions of Vta1NTD residues Y25, E33, L36, L53, I56, E57, F59 and K60 have obvious effects on the Did2 176-204 binding, confirming the energetic importance of all those residues. Consistent with the relatively small buried interface area of Vta1NTD-Did2 complex (compared to the surface area at the interface of Vta1NTD-Vps60 128-186 , approximately 3608 Å 2 ), each single site mutation dramatically decreases all Did2 176-204 binding, so that the binding affinities of all Vta1NTD variants to Did2 176-204 were non-detectable.

Did2 interacts with Vta1NTD in a classic MIM1 mode. The MIT domain is a versatile protein-protein
interaction domain identified in proteins that have a role in vesicle trafficking, including Vps4, Vta1, AMSH and UBPY, where they mediate interaction within the ESCRT-III complex 41 . The MIT domain recognizes sequence motifs called the MIMs primarily within the ESCRT-III subunits. It has been implicated that the interaction between MIT and MIM acts in regulating the disassembly of ESCRT-III as well as targeting specific proteins to the site of ESCRT functions. As we summarized in previous report 33   hydrophobic residues L29, V32, L36, L53 and I56 (which present an overall hydrophobic surface), and hydrophilic residues E33, T46, D54, E57 and K60 in Vta1NTD MIT1 play more important roles in the Vta1-Did2 interactions. These residues correspond to the residues Y34, L37, M41, L64, A67, D38, R57, E68 and K71 in Vps4 MIT domain, all of which are involved in its interactions with CHMP1A MIM1 (Fig. 4E). On the other hand, in Vta1NTD-Did2 176-204 complex structure, the MIM region utilizes conserved hydrophobic residues L195′ , L199′ , and L202′ and hydrophilic residues R198′ , R203′ to interact with MIT. These residues within Did2 176-204 helix α 6′ in the interface make up of an amino acid sequence as E 192 ′ xxL 195 ′ xxR 198 ′ L 199 ′ xxL 202 ′ R 203 ′ , nearly identical to CHMP1A 180-196 MIM1 motif (D/E)xxLxxRLxxL(K/R) 19,20 . Thus, Did2 interacts with Vta1NTD through a classic MIM1 mode. Interestingly, although the binding mode of Did2 and Vta1NTD resembles that of LIP5NTD-CHMP1B, the extent of their further stimulation for Vps4 activity diverge from each other, which suggests different mechanism for Did2 or CHMP1B involved in MVB pathway.
In addition, as shown in Fig. 4G, the crystal structure of Ist1NTD-Did2 MIM1 complex indicated that Did2 MIM1 interacts with the groove made up by Ist1NTD helices α 1, α 3 and α 5 through hydrophobic residues L195′ , L199′ , L202′ and positively charged residues R198′ and R203′ in helix α 6′ 42 . This observation suggested that Did2 MIM1 could not simultaneously interact with Vta1NTD and Ist1NTD. Moreover, the binding affinity (K D ) of Ist1 to Did2 MIM1 is close to 1 μ M, much stronger than that (K D = ~39 μ M) of Vps4 MIT domain binding to Did2 MIM1, and that (K D = ~28 μ M) Vps4 MIT domain binding to Vps2 MIM1 domain, and that (K D = 12.8 μ M) of Vta1NTD binding to Did2 MIM1. This observation reveals that Did2 MIM1 may prefer bind to Ist1NTD due to stronger binding affinity than that to Vta1NTD, and that the interaction between MIT and MIM1 domains in ESCRT-III system is not significantly specific. This analysis may interpret why MIT domain can interact with different subunits in ESCRT-III containing MIM1 domain.
Either Did2 or Vps60 enhances Vta1 stimulation of Vps4 in a specifically indirect manner. The dynamic assembly and disassembly of the ESCRT-III polymer play a critical role in ESCRT-mediated membrane deformation events, and the alterations of Vps4 ATPase activity. To address how Vps60 and Did2 binding enhance Vta1 stimulation of Vps4 ATPase activity, two models were previously presented 27 . One is that their binding to MIT domains results in the conformation changes of Vta1; the other is that the interaction between Vta1 and Did2 or Vps60 increases the local concentration of Vta1-Vps4 in vitro. It was reported that removal of the two Vta1 tandem MIT domains (Vta1 165-330 ) does not enhance the basal activation of Vps4 by Vta1, implying that Vta1 MIT domains does not autoinhibit Vps4 activation 27 . The NMR structures of complex Vta1NTD-Vps60 128-186 33-35 and Vta1NTD-Did2 176-204 provided evidences that the Vps60 or Did2-binding did not induce overall conformational changes in the N-terminus of Vta1. These observations suggested that either Did2 or Vps60 did not allosterically regulate Vta1NTD and thus could not potentiate its ability to directly activate Vps4. Recently, at C-terminus of Vta1, a novel short amino acid sequence, called as Vps4 stimulatory element (VSE), was identified to be released to stimulate Vps4 ATPase activities, upon Vta1NTD interacting with ESCRT-III Did2 or Vps60 43 . VSE activity is auto-inhibited in a manner dependent upon the unstructured linker region, which joints the N-terminal MIT domains and the C-terminal VSL domain. Thus, although structural studies on Vta1NTD-Vps60 128-186 and Vta1NTD-Did2 176-204 provided no direct evidences of how Vps60 and Did2 function, Vps60 or Did2 binding to Vta1NTD might lead to further structural arrangement in the C-terminal domain of Vta1. Either Did2 or Vps60 enhances Vta1 stimulation of Vps4 in a specifically indirect manner.
In summary, we determined NMR solution structure of Vta1NTD-Did2 176-204 , which provided the molecular basis of how Did2 interacts with Vta1NTD. Structural comparison and sequence alignment suggest that Did2 binds to Vta1NTD in a classic MIM1 mode. Both Vps60 128-186 and Did2 176-204 stimulate Vps4 activities by releasing VSE through interaction with Vta1NTD.

Methods
Cloning, expression, and purification. DNA fragments encoding yeast Vta1 and Did2 were amplified from the S. cerevisiae genomic DNA. Vta1NTD and Did2 176-204 were expressed in Escherichia coli BL21(DE3) using a modified pET28b vector with a SUMO protein inserted between a His 6 -tag and the Vta1NTD or Did2 176-204 coding region, respectively. To correctly estimate the concentration of Did2 176-204 during its purification, an extra residue tryptophan was inserted in the N-terminus of the peptide during constructing the plasmid. To obtain pure Vta1NTD, the His 6 -tagged SUMO-Vta1NTD was purified by Ni 2+ -NTA affinity chromatography (GE Healthcare, USA) following standard procedures. ULP1 protease was then added to remove the His 6 -SUMO tag and the protein mixture was passed over a second Ni 2+ -NTA column and was further purified by anion exchange chromatography on a Resource Q column (GE Healthcare, USA). The Vta1NTD variants were purified in the same way as native proteins. To prepare pure Did2 176-204 , the His 6 -tagged SUMO Did2 176-204 was first purified by Ni 2+ -NTA affinity chromatography and by anion exchange chromatography on a Resource Q column, respectively. Then ULP1 protease was added to remove the His 6 -SUMO tag, and the protein mixture was passed over a gel-filtration chromatography Superdex 75 column (GE Healthcare, USA). The concentration of Did2 176-204 was finally obtained from the absorbance at 280 nm with an absorption coefficient of 5500 M −1 cm −1 . The peptide solution was lyophilized for future usage.
For isotope labeling NMR sample (either Vta1NTD or Did2 176-204 ), M9 minimal medium was used supplemented with 15  complex with 1.8 mM unlabeled Vta1NTD. All NMR experiments were performed at 20 °C on a Varian Unity Inova 600 NMR spectrometer (with cryo-probe) equipped with triple resonances and pulsed field gradients, or on a Bruker Avance III-800 MHz NMR spectrometer (with cryo-probe) equipped with four channels and z-axis pulsed-field gradient. The standard suite of experiments for assigning the 1 H, 13 C and 15 N backbone and side chain chemical shifts of 13 C and 15 N double labeled Vta1NTD in complex with unlabeled Did2 176-204 , or of 13 C and 15 N double labeled Did2 176-204 in complex with unlabeled Vta1NTD, and for the collection of NOE-based distance restraints were measured 44,45 , including the two-dimensional (2D) 13 C-edited HSQC in both aliphatic and aromatic regions, and 15 N-edited HSQC; the three-dimensional (3D) HNCA, HNCO, HN(CO)CA, HNCACB, CBCA(CO)NH, 15 N-resolved HSQC-TOCSY, HCCH-TOCSY in both aliphatic and aromatic regions, 15 N-resolved HSQC-NOESY, 13 C-resolved HSQC-NOESY for both aliphatic and aromatic resonances, 2D (H β ) C β (C γ C δ )H δ and (H β )C β (C γ C δ C ε )H ε spectra for correlation of C β and H δ or H ε in aromatic ring used in aromatic protons assignment 46 . The intermolecular NOEs between isotope labeled Vta1NTD or Did2 176-204 peptide and unlabeled Did2 176-204 peptide or Vta1NTD were obtained by analyzing 3D 13 C-F1 edited, 13 C/ 15 N-F3 filtered NOESY spectra, respectively.
For assignment of NMR signals belonging to free Did2 176-204 , the isotope 15 N-labeled Did2 176-204 and unlabeled Did2 176-204 were used at the concentration of 1.0 mM in NMR buffer. 2D 1 H-1 H TOCSY and NOESY, as well as 3D 15 N-edited HSQC-TOCSY experiment, were acquired at 20 °C only on the Varian Unity Inova 600 NMR spectrometer (with cryo-probe, as mentioned above).
All spectra were processed with the program NMRPipe 39 and analyzed with the Sparky 3 software 47 . The 1 H chemical shifts were referenced to 2,2-dimethylsilapentane-5-sulfonic acid (DSS), and the 13 C-and 15 N-resonances were indirectly referenced DSS.
NMR structure determination. The structural calculations of free Did2 176-204 and of the complex Vta1NTD-Did2 176-204 were carried out using a standard simulated annealing protocol implemented in the program XPLOR-2.19 (NIH version). The inter-proton distance restraints derived from NOE intensities were grouped into three distance ranges 1.8-2.9 Å, 1.8-3.5 Å and 1.8-6.0 Å, corresponding to strong, medium and weak NOEs, respectively. The dihedral angles phi and psi were derived from the backbone chemical shifts (HN, HA, CO, CA) by the program TALOS 39,40 . Slowly exchanging amide protons, identified in the 2D 15 N-HSQC spectra recorded after a H 2 O buffer was exchanged to a D 2 O buffer, were used in the structure calculated with the NOE distances restraints to generate hydrogen bonds for the final structure calculation, as done in the literature 48 . A total of ten iterations 49 structures in the initial eight iterations were performed. 100 structures were computed in the last two iterations, 20 conformers with the lowest energy are used to represent the 3D structures. In the ensemble of the simulated annealing 20 structures, there was no distance constraint violation more than 0.3 Å and no torsion angle violation more than 3°. The final 20 structures of the complex Vta1NTD-Did2 176-204 or free Did2 176-204 with lowest energy were evaluated with the program PROCHECK-NMR and PROCHECK 50 and summarized in Table 1. All figures were generated using the program PyMOL (http://pymol.org/) and MOLMOL 49 . Isothermal titration calorimetry (ITC) assay. To obtain a direct binding affinity between Vta1NTD and Did2 176-204 peptide, solution of about 0.1 mM Vta1NTD was titrated with 2.0 mM Did2 176-204 peptide using iTC-200 microcalorimeter (GE healthcare, USA) at 25 °C. The protein and peptide were exchanged to a buffer containing 25 mM sodium phosphate, pH 7.0 and 50 mM NaCl by gel-filtration chromatography, centrifuged to remove any particulates, and degassed. The accurate concentrations of Vta1NTD and Did2 176-204 concentration were determined using their A 280 coefficient constants. The obtained data were fitted by a nonlinear least squares approach to the 'one set of sites' binding model from Microcal Origin software, which yielded the association constant (K a ), stoichiometry of binding (n), and the thermodynamic parameters, enthalpy change of binding (ΔH), entropy change of binding (Δ S) and free energy change of binding (Δ G). The ITC experiment was repeated at least two times for validity.

GST pull-down experiments.
The experiments were performed following standard procedures in buffer containing 25 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 5 mM 2-mercaptoethanol. Purified WT Vta1NTD and its mutants were incubated with either GST alone or GST-tagged Did2 176-204 immobilized on glutathione agarose beads for 3 h at 4 °C. The beads were then washed extensively with above buffer three times, and bound proteins were separated on SDS-PAGE and visualized by Coomassie-blue staining. The pull-down experiments were repeated three times with the similar results. The representative results were shown in Fig. 3F. The control GST-tagged Did2 176-204 alone lane and the GST alone lane indicated the cases where the gels were run in the absence of WT Vta1NTD. The two gels were run at different time.