The purification details of GDP-bound IRGB10 were introduced in a previous study (26). Briefly, the plasmid containing the IRGB10 gene was transformed into Escherichia coli BL21 (DE3) competent cells. Subsequently, the cells were coated onto plates containing Luria-Bertani (LB) agar and incubated overnight at 37°C. A single colony was inoculated into 5–10 mL of LB medium, transferred to 1 L of LB medium, and incubated at 37°C until the optical density (OD) reached ~0.7. Subsequently, 0.5 mM isopropyl β-D-thiogalactopyranoside was added to the medium to induce protein expression, and the cells were incubated overnight at 20°C. After overnight incubation, cells expressing IRGB10 were collected by centrifugation and suspended in 16 mL of lysis buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, and 5 mM imidazole). Subsequently, the cells were disrupted by sonication on ice. The cell lysates were centrifuged at 10,000 g for 30 min at 4°C to remove the cell debris, and the supernatant was incubated with nickel-nitrilotriacetic acid (Ni-NTA) affinity resin (Qiagen, Hilden, Germany). After incubation, the supernatant was loaded onto a gravity-flow column (Bio-Rad, Hercules, CA, USA) and the resin was washed with 50 mL of washing buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, and 25 mM imidazole) to remove impurities. The target protein was eluted from the resin in the column using elution buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, and 250 mM imidazole). The eluted protein was further purified with size-exclusion chromatography (SEC) using SEC buffer (20 mM Tris-HCl pH 8.0, and 150 mM NaCl). The target protein was eluted at around 13 mL, concentrated to 10–12 mg/mL, and stored for structural and biochemical studies.

The same IRGB10 expression clone that was used for the expression and purification of GDP-bound IRGB10 was used for the expression and purification of nucleotide-free IRGB10. The expression in E. coli and affinity chromatography was performed using the same method as that used for the purification of GDP-bound IRGB10. During the washing step, the resin was washed with 30 mL of the first washing buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl), before transferring the washed Ni-NTA resin to 50 mL of the second washing buffer (20 mM Tris-HCl pH 8.0, 1.5 M NaCl) and incubating for 30 min at room temperature. Subsequently, the incubated Ni-NTA resin was reloaded into a gravity column and washed again with 30 mL of the third washing buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 25 mM imidazole). The target protein was eluted using 3 mL elution buffer applied onto the column, and the eluted proteins were loaded onto the SEC column. A Superdex 200 Increase 10/300 GL column (GE Healthcare, Chicago, USA), which had been pre-equilibrated with the SEC buffer, was used in the SEC experiment. The absence of nucleotides was checked by UV absorbance (A260/A280), as outlined in a previous study (28).

The molar masses of nucleotide-free IRGB10, GppNHp-bound IRGB10, and K81A mutant IRGB10 were determined by multi-angle light scattering (MALS). The purified target protein was injected into a Superdex 200 HR 10/30 gel-filtration column (GE Healthcare) that had been pre-equilibrated in buffer containing 20 mM Tris-HCl pH 8.0 and 150 mM NaCl. The chromatography system was coupled to a MALS detector (mini-DAWN TREOS) and a refractive index detector (Optilab DSP) (Wyatt Technology). The data were collected every 0.5 s at a flow rate of 0.4 mL/min and then analyzed using the ASTRA program.

Crystallization of nucleotide-free IRGB10 was performed at 20°C using the hanging drop vapor diffusion method. Initial crystals were screened using a crystallization screening kit from molecular Dimensions, Hampton Research. The crystals were grown on plates by equilibrating a mixed drop of 1 μL protein solution (8–9 mg/mL protein in SEC buffer) and 1 μL reservoir solution containing 0.1 M Tris-HCl pH 7.0, 2.0 M (NH₄)₂SO_4, and 0.2 M Li₂SO₄ against 0.3 mL reservoir solution. The crystallization conditions were further optimized by experimenting with various concentrations and pH values of (NH₄)₂SO₄. The optimized crystals appeared in the presence of 0.1 M Tris-HCl pH 7.2, 1.8 M (NH₄)₂SO₄, and 0.2 M Li₂SO₄.

Crystallization of the GppNHp-bound IRGB10 was performed at 20°C using the hanging drop vapor diffusion method. Just before crystallization, 10 mM GppNHp and 2 mM MgCl₂ were added to 11 mg/mL nucleotide-free IRGB10 protein sample and incubated for 20 min. After incubation, the mixture was screened using a crystallization screening kit. Initial crystals were grown on a reservoir solution containing 0.1 M Tris-HCl pH 8.5, 20% (w/v) polyethylene glycol (PEG), 2,000 monomethyl ether (MME), and 0.2 M trimethylamine n-oxide. The diffraction data sets were collected at the BL-5C beamline of Pohang Accelerator Laboratory (PAL) (Pohang, Republic of Korea). Data processing and scaling were conducted using the HKL2000 package.

The structures of nucleotide-free and GppNHp-bound IRGB10 were determined by the molecular-replacement (MR) phasing method using the Phaser program in the PHENIX program (29). The previously solved IRGB10 GDP-bound structure (PDB ID: 7C3K) was used as the search model. Model building and refinement were conducted by COOT (30) and Refmac5 (31), respectively. Water molecules were added using the ARP/wARP function in Refmac5. The geometry was inspected using PROCHECK and was found to be acceptable. The quality of the model was confirmed using MolProbity (32). All structure figures were created using PyMOL (33).

Oligomerization of IRGB10 was assessed using turbidity measurement (34, 35). Assembly of the IRGB10 oligomer was determined by measuring the absorbance at 350 nm UV using a Nanophotometer NP80 (IMPLEN, Munich, Germany) at 37°C. Purified proteins were concentrated to 100 μM ~ 500 μM and placed in quartz cuvettes. The protein only was placed in cuvettes before starting the measurement. After 500 s, 10 μL of the GTP and MgCl₂ mixture was added. After finishing the measurement, the protein samples were centrifuged at 10,000 g for 10 min at 4°C to remove aggregates. The remaining solution was loaded onto a SEC column, which had been pre-equilibrated with buffer containing 20 mM Tris-HCl pH 8.0, 150 mM to determine the dimer form of IRGB10.

Self-oligomerization of IRGB10 due to GTPase activity was monitored by native-PAGE using a Phast system (GE Healthcare). Pre-made 8%–25% acrylamide gradient gels (GE Healthcare) were used for this experiment. The shifted bands on the gel were stained with Coomassie Brilliant Blue. Purified nucleotide-free IRGB10 was mixed and incubated with different concentrations of GTP and MgCl₂ mixtures at 37°C for 30 min, before loading the mixture onto native gels.

A tentative structural change of IRGB10 caused by GTPase activity was detected using CD measurements. A J-1500 spectropolarimeter at the Korea Basic Science Institute (Osong, South Korea) was used for the CD experiment. The spectra were obtained from 200 to 260 nm at 25°C in a 1-mm pathlength quartz cuvette using a bandwidth of 1.0 nm, a 100 nm/min speed, and a 5-s response time. Three scans were accumulated and averaged. The concentration of nucleotide-free IRGB10 and K81A mutant IRGB10 in the SEC buffer was 0.3–0.4 mg/mL. Next, 2 mM GTP and 0.2 mM MgCl₂ mixture was added to the protein to generate a nucleotide-free IRGB10 + GTP sample. The mixture was incubated at 25°C for 30 min just before injecting the sample into the spectropolarimeter.

The atomic coordinates and structure factors of nucleotide-free and GppNHp-bound IRGB10 were deposited in the Protein Data bank under accession numbers 8JQY and 8JQZ, respectively.

Many GTPases, including the GTPase domain-containing dynamin family, function appropriately by altering their structure and stoichiometry dependent on their GTP/GDP binding state and GTPase activity (36, 37). To reveal the accurate working mechanism of IRGB10 in the process of pathogen membrane disruption, whose function might be dependent on the state of nucleotide binding and hydrolysis capacity, we attempted to solve the structures of the nucleotide-free IRGB10 and IRGB10/GTP complexes. As we observed that endogenous GDP in E. coli was automatically incorporated into IRGB10 during the purification step, we used an additional high-salt washing step during an affinity chromatography step, which has been used previously to remove nucleotides from binding proteins, to obtain nucleotide-free IRGB10 (38). The absence of nucleotides was checked by UV absorbance (A260/A280), as has been outlined previously (28). This experiment showed that the absorbance value of nucleotide-free IRGB10 was 0.67~0.64, while that for GDP-bound IRGB10 was 1.02~1.36, indicating that GDP was washed out during the purification step ( Supplementary Figure 1 ). Next, purified nucleotide-free IRGB10 was applied to SEC, with a GDP-bound IRGB10 sample used for size control. Comparison of the SEC profiles indicated that the main elution peak of nucleotide-free IRGB10 was moved to the monomer size position, although the overall generated peaks on the SEC profiles were similar ( Figure 1A ). The molecular size of nucleotide-free IRGB10 was accurately determined by MALS, which was then used to calculate the absolute molecular mass of the protein particle. The results of MALS showed that the molecular weight of the tentative monomeric peak from nucleotide-free IRGB10 was 53.65 kDa (± 0.7%), whereas the molecular mass of the dimeric GDP-bound IRGB10 was 102.72 (± 1.8%) ( Figure 1B ). These results indicated that the dimeric GDP-bound IRGB10 became monomeric when IRGB10 lost its GDP. Purified nucleotide-free IRGB10 was successfully crystallized, which allowed the structure of the monomeric nucleotide-free IRGB10 to be solved. The crystallographic data and refinement statistics are summarized in Table 1 . Unlike dimeric GDP-bound IRGB10, a single molecule of IRGB10 was detected in the crystallographic asymmetric unit (ASU). The overall structure of monomeric nucleotide-free IRGB10 was almost identical to that of GDP-bound IRGB10, which was composed of a helical domain formed by N-terminal, C-terminal, and GTPase domains, which is a typical domain composition of the IRG family ( Figures 1C, D ). The GTPase domain consisted of six β-sheets (S1–S6) and six α-helices (H4–H9), while the helical domain consisted of eleven α-helices, including H1~H3 from the N-terminus region and H10~H17 from the C-terminus region. The model of nucleotide-free IRGB10 was constructed from residue 16 to residue 406. The LEH residues at the C-terminus, which were from the plasmid construct, were included in the final model. The electron density of the N-terminus residues and several loops, including switches I and II in the GTPase domain, were not visible in the model ( Figure 1C ). These parts of the structure could not be constructed due to poor electron density. The unconstructed N-terminus and several loops in the GTPase domains around these structures were also observed in structural studies of IRGA6 and dimeric GDP-bound IRGB10; this indicates that the N-terminus loop, containing around 13–15 residues, and several loops, including switches I and II at the GTPase domain, are extremely flexible and unstructured regions (25). As we removed GDP from IRGB10 during the purification step and found that nucleotide-free IRGB10 became a monomer in solution, using the current structure, we first sought to investigate whether GDP or GTP is in the GTPase domain. The electron density search revealed no traceable electron density for GDP in the typical nucleotide-binding site of the GTPase domain ( Figure 1E ).

Next, we compared the structure of the nucleotide-free IRGB10 to that of the GDP-bound IRGB10 (PDB ID: 7C3K) to analyze any structural changes that might occur by the loss of nucleotides in IRGB10. Pair-wise structural alignments between nucleotide-free and GDP-bound IRGB10 showed that the overall structures were similar to each other, with a RMSD between the two structures of 1.3 Å ( Figure 1F ). However, close-up analysis showed that the H2 and H3 helical domains formed by the N-terminal part of IRGB10 were dislocated from the positions of the H2 and H3 regions of GDP-bound IRGB10 ( Figure 1G ). In contrast, the last part of the helical domain that was constructed by the C-terminal part of IRGB10 was identical to that in GDP-bound IRGB10, indicating that binding nucleotide or GTP hydrolysis causes a slight structural alteration of IRGB10. Although the structures of the GTPase domain of each structure were almost identical, the positions of several loops were not. The structures of switches I and II, both of which are critical for GTPase activity, were unconstructed in nucleotide-free IRGB10, while only switch I was unconstructed in GDP-bound IRGB10 ( Figure 1H ). Interestingly, the P-loop, which is critical for nucleotide binding and GTP hydrolysis, was well constructed in both structures, indicating that the formation and location of the proper positioning of the P-loop is independent of nucleotide binding, contrary to what we have argued in a previous structural study of GDP-bound IRGB10 (26).

Indeed, as we observed that nucleotide binding affects the stoichiometry change of IRGB10, we also investigated the effect of GTPase activity and GTP binding on any oligomeric and structural changes of IRGB10. To accomplish this, we first performed a turbidity assay that has been used previously to analyze the oligomerization of IRGA6 (34). The oligomerization of IRGB10 was detected by checking the absorbance of 350 nm UV light. After adding the GTP/MgCl₂ mixture to GDP-bound IRGB10, UV absorbance was not detected for 1200 s ( Figure 2A ). However, when the GTP/MgCl₂ mixture was added to nucleotide-free IRGB10, a considerable increase in UV absorbance was detected 600 s after GTP addition ( Figure 2B ). This UV absorbance was not detected when a non-hydrolysable GTP analog (GppNHp) was supplied to nucleotide-free IRGB10 ( Supplementary Figure 2 ). The results of these turbidity assays indicated that GTP hydrolysis caused the oligomerization of IRGB10. Moreover, visible IRGB10 oligomeric particles were detected in the tube containing nucleotide-free IRGB10 following GTP addition. After removing those oligomeric particles by centrifugation, the solution was loaded onto SEC to determine the remnants in the solution. As GTPase hydrolyzes GTP to GDP, we speculated that the dimeric form of GDP-bound IRGB10 would be observed if the hydrolyzed product of GDP was incorporated into IRGB10 after hydrolysis. As expected, the SEC profile showed that the last portion of IRGB10 after GTP hydrolysis was a dimeric size and was eluted around the 13–14 position where dimeric GDP-bound IRGB10 was eluted ( Figure 2C ). When the non-hydrolysable GTP analog GppNHp was added to nucleotide-free IRGB10, no oligomeric particles were observed in the tube and the SEC profile showed a monomeric size ( Figure 2C ), indicating that GTP hydrolysis is critical for the dimerization and further oligomerization of IRGB10. The effect of GDP addition on the nt-free IRGB10 was also assessed by performing the same turbidity ( Supplementary Figure 3 ). As shown at the S3 Figure , GDP addition did not produce oligomeric peak although a little absorbance was detected at 200 sec point after addition of GDP/MgCl₂. This indicated that GTP hydrolysis is critical step for IRGB10 oligomerization. The GTP hydrolysis-mediated oligomerization of IRGB10 was confirmed by native PAGE, which is another oligomerization detection assay. As shown in Figure 2D , the newly formed oligomeric band was detected by the addition of GTP, indicating that GTP addition caused IRGB10 oligomerization, which was observed in the turbidity assay. Finally, we attempted to confirm whether GTP hydrolysis of IRGB10 is essential for dimer formation and further oligomer formation by constructing mutants that cannot hydrolyze GTP. To perform this experiment, K81, a catalytically important residue identified from previous study (19), was mutated to alanine to produce the K81A mutant, which is the GTP-locked form of IRGB10. Using this GTP-locked form of IRGB10, we performed a turbidity assay and SEC-MALS. Unlike wildtype IRGB10, K81A did not produce visible oligomeric particles following the addition of GTP. Additionally, K81A did not produce a dimeric peak on the SEC profile when GTP was added to K81A ( Figure 2E ). All three SEC samples, including nucleotide-free K81A, K81A with GTP, and K81A with GppNHp, were eluted at the monomer position at SEC-MALS ( Figures 2E, F ). In addition, K81A did not produced dimeric peak in the presence of GDP ( Supplementary Figure 4 ). These additional experiments confirmed that GTP hydrolysis is essential for dimer formation and further oligomerization of IRGB10, which may be critical for pathogen membrane disruption. Finally, we elucidated the role of the dimer PPI of IRGB10 on the oligomerization of IRGB10. To evaluate this, we used dimer PPI disrupting mutant D185R, which has been identified by our previous study as PPI interfering mutant. The turbidity assay showed that nt-free D185R mutant failed to produce oligomeric peak after addition of GTP/MgCl₂ ( Supplementary Figure 5 ). Based on this experiment, we concluded that dimerization is a seed of further oligomerization of IRGB10.

The mimetic structure of the GTP-bound form of IRGB10 was also solved using GppNHp, which is a non-hydrolyzable GTP analog. The crystallographic data and refinement statistics are summarized in Table 1 . The overall structure and numbers of α-helices and β-sheets were similar to previously revealed nucleotide-free and GDP-bound IRGB10 structures. We detected a clear electron density map at the nucleotide-binding site in the GTPase domain responsible for GppNHp ( Figure 3A ). The P-loop was stably fixed in GppNHp-bound IRGB10, while switches I and II remained unstructured ( Figure 3A ). Additionally, the crystallographic asymmetric unit comprised two identical IRGB10 molecules, molecule A and B ( Figure 3B ). As the stoichiometric and structural changes of the IRG family of GTPases are critical for understanding the working mechanism of the IRG family, we next analyzed the stoichiometry of GppNHp-bound IRGB10 in the solution using MALS. The experimentally calculated molecular weight of GppNHp-bound IRGB10 was 48.93 kDa (± 3.941%), indicating that GppNHp-bound IRGB10 is a monomer in the solution ( Figure 3C ), indicating that GTP bound-IRGB10 without GTP hydrolysis is still a monomer in solution.

To comprehend any structural change caused by nucleotide binding and GTPase activity, we next compared the GppNHp structure with nucleotide-free ( Figure 3D ) and GDP-bound ( Figure 3E ) IRGB10 structures by structural superposition analysis. The results of this structural comparison indicated that the overall GppNHp structure is almost identical to that of the nucleotide-free and GDP-bound forms of IRGB10, with RMSDs of 0.8 Å and 1.2 Å, respectively. However, upon closer examination of the helical domain, the locations of several helices were not identical. Indeed, H2 and H3 of the GppNHp-bound form were tilted by approximately 5° compared to those of the nucleotide-free form of IRGB10 ( Figure 3F ). Moreover, compared to the helical domain of the GDP-bound form, H13 and H14 of the GppNHp-bound form were tilted by approximately 8° compared to those of the GDP-bound form of IRGB10 ( Figure 3G ). The largest structural alteration was detected in the nucleotide binding site ( Figure 3H ). Although no structural changes were detected when the GppNHp-bound structure was compared to the nucleotide-free form ( Figure 3I ), distinct movements of the H4 and H4 connecting loops were detected when the GppNHp-bound structure was superposed with the GDP-bound form ( Figure 3J ). Moreover, by conducting a structural comparison of the GTPase domain of the GppNHp-bound IRGB10 with the GDP-bound form, we found that the loops of the GppNHp-bound form were very flexible and unfixed, and the position of H4 connected to the switch I loop in the GDP-bound form was different from that of the GppNHp-bound form ( Figures 3H, J ). All P-loops structures, which are important for nucleotide binding, of the three structures were identical. Finally, we evaluated the far UV circular dichroic (CD) spectra to determine the tentative structural changes of IRGB10 during GTP hydrolysis. The results of this experiment showed that the spectrum patterns were different when nucleotide-free IRGB10 was treated with GTP ( Figure 3K ). The nucleotide-free IRGB10 alone sample produced a typical CD spectrum pattern of α-helical proteins, exhibiting two pronounced minima at 208 nm and 222nm and a maxima at 200 nm. This pattern was not observed when GTP was provided. Moreover, these changes in the CD pattern were not observed when the GTPase activity defect K81A mutant was treated with GTP. In addition, GDP or GppNHp addition also produced a typical CD spectrum pattern produced by wildtype IRGB10 ( Supplementary Figure 6 ). These CD experiments indicate that GTP hydrolysis might lead to structural changes in IRGB10.