A Structure Model Explaining the Binding Between a Ubiquitous Unconventional G-protein (OsYchF1) and a Plant-Specific C2-Domain Protein (OsGAP1) from Rice

The unconventional G-protein OsYchF1 plays regulatory roles in plant defense and abiotic stress responses. We have previously resolved the crystal structures of OsYchF1 and its plant-specific regulator, OsGAP1, and determined the residues on OsGAP1 that are essential for its binding to OsYchF1. In this study, we employed site-directed mutagenesis to identify four critical residues on the TGS domain of OsYchF1 that are critical for its binding to OsGAP1. We also generated a docking model of the OsYchF1:OsGAP1 complex to dissect the molecular basis of their interactions. Our finding not only reveals the roles of the key interacting residues controlling the binding between OsYchF1 and OsGAP1, but also provides a working model on the potential regulatory mechanism mediated by a TGS domain, particularly in the class of GTPase of the family. antibodies. Pull-down with HisSUMO tag-only as prey was included as negative control. (C) GST-AtGAP1 was used as bait to pull down MBP-fused AtYchF1 and AtY104 (the preys, ~86.9kDa in size), making use of the MagneGST TM protein purification system (Promega V8603). Pull-down products were subjected to western blot and the presence of MBP-fused prey was detected by anti-MBP antibodies. Results indicated that MBP-AtY104 failed to be detected in the pull-down product and therefore the interaction between GAP1 and YchF1 was specific. Pull-down with MBP-only as prey was included as negative control.


Introduction
We previously solved the crystal structures of an unconventional G protein in rice, OsYchF1, and its regulator, OsGAP1, which is a C2-domain protein [1,2]. OsYchF1 belongs to a group of highly-conserved, ubiquitous P-loop NTPases [3], whereas OsGAP1 is plant-specific and could activate the ATPase/GTPase activities of OsYchF1 [4,5]. OsGAP1 was first characterized as a wounding-inducible gene in the Xa14 rice line (a line harboring the Xa14 resistance gene against the pathogen Xanthomonas oryzae pv. oryzae [Xoo]). The over-expression of OsGAP1 in transgenic rice lines led to enhanced resistance towards Xoo [4]. Through functional characterization of transgenic Arabidopsis expressing OsGAP1 ectopically, OsGAP1 was shown to participate in the well-known plant stress convergent pathway. The transgenic Arabidopsis demonstrated increased resistance towards Pseudomonas syringae pv. tomato DC3000 (Pst DC3000) with increased expressions of both salicylic acid-related (PR1 and PR2) and jasmonic acid-ethylene-related (Thi2.1 and PDF1.2) defense marker genes. In addition, OsGAP1 ectopically expressed in the Arabidopsis npr1-3 mutant line did not show enhanced resistance upon Pst DC3000 inoculation, implying OsGAP1 is positioned upstream of the key plant stress response regulator, NPR1 [4]. Furthermore, OsGAP1 also plays a role in abiotic stresses.
OsGAP1-transgenic Arabidopsis displayed increased salt tolerance with less leaf chlorosis, less reactive oxygen species generation and ion leakage, and higher induction of salt-inducible genes (RD22 and RD29a) upon salt treatment [5]. Meanwhile, OsYchF1 is a negative regulator of both plant defense and abiotic stress responses in plants, with OsGAP1 playing a positive role by down-regulating the biological functions of OsYchF1 [4][5][6]. OsGAP1 could inactivate OsYchF1 through stimulating its GTPase/ATPase activities and turning it into the inactive GDP/ADPbound form. Besides, OsGAP1 could also regulate the subcellular localization of OsYchF1 by recruiting OsYchF1 to the plasma membrane from the cytosol upon wounding [6]. In addition, OsGAP1 was shown to compete for the binding site of 26S rRNA on the TGS domain of OsYchF1 [6]. Therefore, deciphering the molecular interactions between OsYchF1 and OsGAP1 is important in understanding the mechanisms of both plant defense and abiotic stress responses. We previously showed that OsGAP1 binds to the TGS (Threonyl-tRNA synthetase, GTPase, and SpoT proteins) domain of OsYchF1 [6]. Based on the analysis of the primary structures of OsGAP1 plant homologs and using site-directed mutagenesis, we also successfully determined the amino acid residue clusters on one surface of OsGAP1 that are critical for the interaction with OsYchF1, and those residues critical for the interaction with phospholipids on another surface of OsGAP1 [2].
To identify hot spots on OsYchF1 that are also important for the interaction with OsGAP1, we applied the same strategy by aligning the primary structures of 30 OsYchF1 homologs from different plant species to identify the conserved residues that are likely candidates for the binding hot spots with GAP1 due to the structural conservation [7][8][9][10]. We specifically selected the conserved charged and aromatic residues for testing as they would be more electrochemically reactive. Using this approach, we successfully identified the residues of OsYchF1 that are responsible for the interaction between OsYchF1 and OsGAP1. By integrating all structural data, we proposed a high-confidence model to explain the molecular interactions between OsYchF1 and OsGAP1.

Materials and methods
Selection of the potential OsYchF1 amino acid residue candidates involved in binding OsGAP1 Using Protein-BLAST, 30 YchF1 homologs from various plant species were selected for alignment analysis by the Consurf server (http://consurf.tau.ac.il/) [7][8][9][10] 5EE9]) previously published by our group [1]. The most conserved amino acid residues in the alignment that are also presented on the surface of the protein molecule based on our 3-D structures, with side chains most likely to be involved in protein-protein interactions, were selected for site-directed mutagenesis. Targeted residues were mutated to alanine to minimize the possible effects on the overall protein structure.

Site-directed mutagenesis
The plasmid pGEX-4T-1-OsYchF (6kb), which contained an OsYchF1 cDNA fragment amplified from the cDNA pool of rice cultivars (SN1033 and JG30) in the pGEX-4T-1 vector, mM NaCl, 500 mM imidazole, pH 7.4). The eluted pull-down products were subjected to western blot analysis with anti-GST antibodies for prey detection.
Meanwhile, GST-AtGAP1 fusion protein was applied as bait to examine the binding with the MBP-fused AtYchF1 and AtY104 mutants with the MagneGST TM protein purification system (Promega V8603). First, all fusion proteins were expressed in E. coli using the same protein expression procedure as above. One milliliter of cell extract in MagneGST TM cell lysis reagent (provided in the kit) obtained from 5mL GST-AtGAP1-expressing DE3 culture was allowed to incubate with 200 μL equilibrated MagneGST beads in 1 mL MagneGST TM binding/wash buffer (provided in the kit) as described in user manual. After three rounds of washing, GST-AtGAP1bound MagneGST beads were used to pull down purified MBP-AtYchF1 or MBP-AtY104 mutant proteins, in which around 10μg MBP fusion proteins purified with SpinClean TM MBP Excellose ® spin kit (Mbiotech 23020) were added as preys. After three further rounds of washing with buffer, final elution was made in 50μL volume. Then, it was subjected to western blot analysis with anti-MBP antibodies for prey detection. All bait protein expressions and extraction efficiencies were examined beforehand to make sure even protein loading in the in-vitro pulldown experiments and western blot analyses. Most procedures were followed as described in the user manuals unless specifically stated above.

Transgenic plant materials
Native rice and Arabidopsis YchF1 homologs (OsYchF1 and AtYchF1) and their corresponding mutants (i.e., OsY104 and AtY104) that were unable to bind the respective rice and Arabidopsis GAP1 proteins were ectopically expressed in tobacco BY-2 cell suspensions, driven constitutively by the cauliflower mosaic virus 35S promoter in the binary vector, V7 or W104 [13]. The BY-2 transformation protocol was as described previously [14]. For the construction of transgenic Arabidopsis, the same binary construct was transformed into AtYchF1-knockdown mutant (from Arabidopsis Biological Research Centre [stock# CS855214]). Protocols of transformation and selection of positive transformants were as previously reported [4,6]. Phenotype characterization was performed after ten days of salt treatment by photo taking and chlorophyll measurement [15].
Pathogen titer determination at 3 days post-inoculation were performed using a plate count method [16]. Expressions of defense marker genes (PR1 and PR2) were examined with samples harvested three days after inoculation.
Gene expression analysis by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) RNA extraction, reverse-transcription and real-time PCR were performed according to a previous report [4]. The relative gene expression was calculated using the 2 -CT method [17] and normalized with the Arabidopsis thaliana UBQ10 gene (AY139999; [18]). Detailed primer information for real-time PCR were previously reported [5,6]. The treatment means were analyzed using one-way analysis of variance (ANOVA) followed by the Games-Howell post hoc test.

HADDOCK docking
The docking of OsGAP1 to OsYchF1 was performed using the data-driven docking server HADDOCK 2.4 (https://wenmr.science.uu.nl/haddock2.4/) [15]. The crystal structures of OsGAP1 (PDB ID: 4RJ9) and OsYchF1 (PDB ID: 5EE1) were used as the protein input templates. To generate the Ambiguous Interaction Restraints (AIRs) for HADDOCK, residues L5, L8, K37, K39, K41, R43, T58, and S60 were input as the active residues of OsGAP1, while E345 alone, or K325, H334, E345 and E354 together was/were input as OsYchF1 active residue(s). Passive residues that are close to the active ones were defined by the HADDOCK server automatically. All HADDOCK runs were performed with 1,000 structures for rigid body docking and 10 rounds of rigid body minimization. The best 200 structures were subjected to semi-flexible refinement. The refined structures were clustered using Fraction of Common Contacts (FCC) with a cut-off of 0.6 and a minimum cluster size of 6.

Molecular dynamics (MD) simulation
The crystal structure of OsYchF1 (PDB ID: 5EE1) and the HADDOCK docking structure of the OsYchF1:OsGAP1 complex were employed to initiate MD simulations. Missing residues in OsYchF1 were modeled using Modeller 9.23 [19] with its crystal structure resolved at pH=7.85 (PDB ID: 5EE0) as a template. MD simulations for the complex were initialized by superimposing OsYchF1 and OsGAP1 (PDB ID: 4RJ9) to the HADDOCK docking result.
Four replicates of 1-μs MD simulations were performed using GROMACS 2020.2 [20] with the CHARMM36 force field [21] for OsYchF1 alone and the OsYchF1:OsGAP1 complex respectively. The proteins were placed in a dodecahedron box and solvated by TIP3P water [22].
Sodium chloride was added at a concentration of 0.1 M to neutralize the system. All systems were subjected to energy minimization followed by a 1-ns NVT equilibration and a 1-ns NPT equilibration with position restraints on the heavy atoms of the protein backbone. In all simulations, van der Waals interactions were smoothly switched from 8 Å to 9 Å. Electrostatic interactions were evaluated employing the particle mesh Ewald (PME) method [23] with a grid spacing of 1.2 Å, a PME order of 4, and a cutoff of 9 Å. The system temperature was kept at 300 K using a velocity-rescaling thermostat [24], whereas the system pressure was kept at 1 bar by Berendsen's algorithm [25]. All bonds with H-atoms were constrained using LINCS algorithm [26]  lost the binding affinity for OsGAP1, compared against native OsYchF1 (Figure 2A-B).
Coincidentally, these residues are all part of the TGS domain. To examine whether the changes in protein-protein interactions were the results of perturbed protein folding of the OsYchF1 mutants, circular dichroism (CD) measurements were carried out. All the OsYchF1 mutants that lost OsGAP1 binding exhibited CD spectra similar to that of the native OsYchF1, indicating that Enhanced salt sensitivity and pathogen susceptibility in transgenic Arabidopsis ectopically expressing native OsYchF1 were previously reported [5,6]. In this study, we compared the RD29a) were quantified with RNA extracted one day after treatment by RT-qPCR, and these two genes were found to have minimal induction after salt treatment in the transgenic lines expressing native OsYchF1 and were even less induced in the lines expressing OsY104, when compared to the wild-type (Col-2). These genes, however, were highly inducible in the AtYchF1knockdown mutant ( Figure 4C). As a quality control, the levels of transgene expressions were examined and those of native OsYchF1 and OsY104 were found to be comparable (Supplementary Figure S2A).
To examine the defense responses of transgenic Arabidopsis ectopically expressing native OsYchF1 and OsY104 against Pst DC3000 infection, 4-week-old seedlings were inoculated with Pst DC3000 via syringe infiltration. Phenotype characterization including disease symptoms ( Figure 5A), pathogen titer ( Figure 5B) and expressions of defense marker genes ( Figure 5C) was carried out at day 0 and day 3 of inoculation. OsY104-expressers exhibited enhanced susceptibility towards Pst DC3000 with more severe disease symptoms, higher pathogen titers  Figure S2B).

Docking of OsYchF1 to OsGAP1
Our mutational studies clearly indicated that the TGS domain, but not the G domain or the helical domain, of OsYchF1 is critical for OsGAP1 binding. We have previously identified two clusters of amino acid residues, namely L5, L8, T58, and S60 on strands 1 and 4, and K37, K39, K41, and R43 on strand 3 of OsGAP1, that serve as the determinants for the interaction with OsYchF1 ( Figure 6A) [2]. Surface rendition of the crystal structure of OsGAP1 revealed that these two clusters lie on the same molecular surface, potentially serving as the interaction OsYchF1, and the root-mean-square deviation (rmsd) of OsGAP1 between the two models was calculated to be 3.2 Å ( Figure 6D). In both docking poses, OsGAP1 docks on the same molecular surface of OsYchF1 formed by the  he 12b and 14 helices of the TGS domain. Since a larger surface area is buried in model 2 than in model 1 (1766±78 Å 2 vs. 1605±120 Å 2 ), we selected model 2 for more detailed analysis and this model is hereafter referred to as the complex model of OsGAP1:OsYchF1.
Nearly all selected active residues in our model are located at the interaction interface between the two proteins except K325 and E354 of OsYchF1 ( Figure 7A). These two residues interact intra-molecularly with the side-chain of K366 and the carbonyl backbone of T373 at the loops flanking strand 11. Such interactions might be important for maintaining and stabilizing the conformation of strand 11 and Y367 which is in direct contact with OsGAP1 ( Figure 7B). In addition, analysis of the crystallographic B-factors, which indicate the dynamic mobility of atoms, of residues within the loops and strand 11 of the OsGAP1 crystal structure revealed higher dynamics of the molecule in this region (Supplementary Figure S4), underscoring the potential need for stabilizing, a task fulfilled by K325 and E354.
Our model reveals that K37, K39 and K41 of OsGAP1 may play the most critical roles in the complex formation. Specifically, the side-chain of K41 forms ionic interactions with the key OsYchF1 residue, E345, and E358 of helix 14 ( Figure 7C). The position of OsGAP1 within the complex is further stabilized by a hydrogen bond with the side-chain hydroxyl group of Y367 and van der Waals interaction with F383 of strand 12 ( Figure 7C). The phenyl group of F383 also mediates van der Waals interaction with the side-chain of K39, the -amine group of which, along with that of K37, form ionic interactions with the side-chain of E314 of strand 9 of OsYchF1 ( Figure 7D). On the other hand, L5, L8, and T58 on strands 1 and 4 of OsGAP1 appear to form a small pocket to accommodate the side-chain of H334 of OsYchF1 ( Figure 7E).
We previously demonstrated that the binding of OsGAP1 activates the NTPase activity of OsYchF1, leading to the activation of defense and salt stress responses. Yet the current study revealed that the binding site of OsGAP1 is distal to the NTP-binding pocket of OsYchF1. We speculate that the effect of their interaction on the NTPase activity is likely mediated in an allosteric manner rather than through direct alteration of the NTP-binding site ( Figure 6D).

Comparison of the ATP and GTP binding affinities between native OsYchF1 and OsY104 using
Mant-nucleotides also illustrated that OsY104 has NTP affinities comparable to those of native OsYchF1 (Supplementary Figure S5).

Discussion
We previously reported that the higher plant-specific OsGAP1 is a regulator of a ubiquitous ancient unconventional G-protein, OsYchF1, in response to biotic and abiotic stresses [2]. In order to study the biological effects exerted by OsGAP1 on OsYchF1, the OsYchF1 mutants that lose the affinity specifically towards OsGAP1 would be valuable genetic materials for functional characterization. In this report, we successfully identified several conserved amino acid residues on the surface of the OsYchF1 protein by taking both sequence homology and three-dimensional structure into account. Most importantly, mutations of these residues did not affect the protein folding of OsYchF1 (Supplementary Figure S1). Among all the residues tested, E345 is of the highest interest due to its specific and critical role in the interaction between OsGAP1 and OsYchF1. To examine its role in GAP1-binding in other plant homologs, AtYchF1, a dicotyledonous model plant YchF1 homolog, was mutated to generate the AtY104 (E345A) mutant. The AtY104 mutant also failed to bind AtGAP1 in the in-vitro pull-down study ( Figure   2B). Thus, we believe that the conserved and solvent-exposed amino acid residue, E345, of the YchF1 protein is the most critical for its interaction with the regulator, GAP1.
To confirm this hypothesis, OsY104 and AtY104 mutants were ectopically expressed in tobacco BY-2 cells and subjected to salt treatment. Both OsY104-and AtY104-expressing transgenic BY-2 cell lines exhibited more salt-sensitive phenotypes when compared to the native YchF1expressing counterparts, which in turn exhibited higher salt sensitivity than wild-type BY-2 cells ( Figure 3). In planta experiments making use of transgenic Arabidopsis ectopically expressing native OsYchF1 and OsY104 also revealed the reduced salt tolerance resulting from OsY104 compared to native OsYchF1 ( Figure 4). Besides, OsY104 ectopic expressers show more severe symptoms upon Pst DC3000 inoculation ( Figure 5). The expression levels of stress response marker genes in OsY104-ectopic expressers were also much lower than those of the native OsYchF1-expressers upon stress treatments (Figures 4 and 5). This result implied that GAP1 binding is important for inactivating YchF1 in order to lower salt sensitivity and pathogen susceptibility in wild-type plants.
Technological advancements in genomic sequencing and mass spectroscopy in the past two decades have revolutionized the fields of plant genomics and proteomics. However, while the number of newly predicted or identified proteins continues to build exponentially, the functional and structural characterization of these novel proteins is significantly lagging behind. Therefore in most cases their biological roles remain elusive. In particular, structural studies of novel proteins using experimental techniques like nuclear magnetic resonance (NMR) spectroscopy, Xray crystallography and cryo-electron microscopy usually require large investments of effort and time with no guaranteed success. Our approach of taking sequence homology and protein structure into account to hunt for amino acid residue(s) critical for interactions and combining the findings with data-driven docking simulations seems to be an effective and efficient strategy for functional and mechanistic studies of novel protein complexes. By aligning the sequence of OsYchF1 with 30 other plant homologs, 173 out of 394 amino acid residues were found to be identical among all 31 homologs. By superimposing conserved amino acids with the 3D protein structure, only those conserved amino acid residues exposed on the protein surface needed to be focused on. This greatly helped to reduce the number of candidates to be screened. In fact, four out of the eleven candidates tested either lost or showed significantly weaker interactions with OsGAP1. Such a high success rate not only helped to minimize workload and economize on materials and to quickly identify four hot-spot residues, it also minimized the efforts and time needed to obtain structural information on the protein complex by providing key information for high-confidence protein-protein complex docking.
In recent years, molecular docking techniques have served as important methods to elucidate the structural details of protein complexes that are difficult to be obtained by experimental techniques. The data-driven HADDOCK is one of the best-performing docking approaches due to its ability to integrate biochemical/biophysical experimental data during the docking process [16]. Using HADDOCK, we were able to generate a docking model of the OsYchF1:OsGAP1 complex to study the molecular basis of their interaction. Different docking attempts using either E345 as the only interacting residue on OsYchF1 or together with other conserved TGS domain We previously reported that OsGAP1 is a wounding-inducible gene in bacterial blight-resistant rice lines harboring the resistant gene, Xa14, and its over-expression leads to enhanced resistance against Xanthomonas oryzae pv. oryzae in rice [6]. Over-expression of AtGAP1 also increased resistance towards Pseudomonas syringae pv. tomato DC3000 and enhanced salt tolerance in Arabidopsis [4,5]. These positive effects of the GAP1 protein upon biotic and abiotic stresses