Grouper iridovirus GIV66 is a Bcl-2 protein that inhibits apoptosis by exclusively sequestering Bim

Programmed cell death or apoptosis is a critical mechanism for the controlled removal of damaged or infected cells, and proteins of the Bcl-2 family are important arbiters of this process. Viruses have been shown to encode functional and structural homologs of Bcl-2 to counter premature host-cell apoptosis and ensure viral proliferation or survival. Grouper iridovirus (GIV) is a large DNA virus belonging to the Iridoviridae family and harbors GIV66, a putative Bcl-2–like protein and mitochondrially localized apoptosis inhibitor. However, the molecular and structural basis of GIV66-mediated apoptosis inhibition is currently not understood. To gain insight into GIV66's mechanism of action, we systematically evaluated its ability to bind peptides spanning the BH3 domain of pro-apoptotic Bcl-2 family members. Our results revealed that GIV66 harbors an unusually high level of specificity for pro-apoptotic Bcl-2 and displays affinity only for Bcl-2–like 11 (Bcl2L11 or Bim). Using crystal structures of both apo-GIV66 and GIV66 bound to the BH3 domain from Bim, we unexpectedly found that GIV66 forms dimers via an interface that results in occluded access to the canonical Bcl-2 ligand–binding groove, which breaks apart upon Bim binding. This observation suggests that GIV66 dimerization may affect GIV66's ability to bind host pro-death Bcl-2 proteins and enables highly targeted virus-directed suppression of host apoptosis signaling. Our findings provide a mechanistic understanding for the potent anti-apoptotic activity of GIV66 by identifying it as the first single-specificity, pro-survival Bcl-2 protein and identifying a pivotal role of Bim in GIV-mediated inhibition of apoptosis.

Programmed cell death or apoptosis is a crucial mechanism to remove damaged, unwanted, or infected cells and to maintain tissue homeostasis (1). The B-cell lymphoma 2 (Bcl-2) 2 protein family is a critical regulator of the intrinsic or mito-chondrially mediated apoptotic pathway (2). The family is characterized by the presence of 1-4 conserved Bcl-2 homology (BH) domains (BH1-4), and is subdivided into pro-apoptotic or pro-survival proteins (3). Pro-survival proteins harbor all four BH domains and in mammals comprise Bcl-2, Bcl-x L , Bcl-w, Mcl-1, A1, and Bcl-b. The pro-apoptotic family members are further subdivided into multidomain proteins, including Bax, Bak, and Bok, and family members that only feature the BH3 domain and are consequently referred to as the BH3-only proteins. Pro-apoptotic Bax and Bak are critical for apoptosis to proceed (4) and trigger the release of pro-apoptogenic factors, including cytochrome c from mitochondria, by forming oligomeric pores to perforate the mitochondrial outer membrane. Bax is found predominantly in the cytosol and translocates to the outer mitochondrial membrane after an apoptotic stimulus, whereas Bak is constitutively anchored to the outer mitochondrial membrane via a C-terminal transmembrane anchor.
Pro-apoptotic BH3-only proteins include Bim, Bid, Puma, Noxa, Bmf, Bik, Bad, and Hrk in mammals and induce apoptosis either indirectly by neutralizing pro-survival Bcl-2 or directly by interacting with Bax and Bak (5). This activity is mediated by the helical BH3 domain, which is able to bind to a canonical hydrophobic ligand-binding groove on both pro-survival and pro-apoptotic multidomain Bcl-2 proteins (6). In healthy cells, BH3-only proteins act as sentinels of cellular well-being and are up-regulated in response to cellular insults, including growth factor deprivation, exposure to cytotoxic drugs, or viral infections, leading to the activation of cell death mechanisms (2).
Whereas the intrinsic pathway of apoptosis is highly conserved from the worm to mammals (7), certain differences exist in the apoptotic machinery between mammals and fish. Zebrafish lack the pro-survival Bcl-2 proteins Bcl-w and A1 as well as pro-apoptotic Bak and Hrk; however, the interplay between the different Bcl-2 members and the mechanisms underlying control of apoptosis appear to be conserved (8).
The Iridoviridae represent another family of large DNA viruses and are subdivided into four genera, iridovirus, choriridovirus, ranavirus, and lymphocystivirus (25). Iridoviruses and chloriridoviruses commonly infect invertebrates, whereas ranaviruses and lymphocystivirus target vertebrates, including amphibians, fish, and reptiles. The sequencing of the grouper iridovirus genome revealed the presence of a putative Bcl-2like protein (26), GIV66, and functional studies established that GIV66 locates to the outer mitochondrial membrane and is able to inhibit UV-induced apoptosis in grouper kidney cells (27). However, the molecular and structural basis of apoptosis inhibition by GIV66 remains unclear.

Results
It was shown previously that GIV66 is localized at the outer mitochondrial membrane and is able to inhibit Bcl-2-mediated apoptosis in UV-irradiated GK cells (27); however, no detailed mechanism was proposed to rationalize these findings. To understand the molecular mechanism of action and structural basis of GIV66-mediated inhibition of apoptosis, we recombinantly expressed and purified GIV66 to perform isothermal titration calorimetry (ITC) with peptides spanning the BH3 motif of pro-apoptotic Bcl-2 proteins. Peptides were selected from pro-apoptotic Bcl-2 proteins identified in grouper fish (Epinephelus coioides) as well as from zebrafish (Danio rerio) where grouper homologs were unavailable (Fig. 1E). Unexpectedly, GIV66 only displayed measurable affinity for zebrafish encoded Bcl-2-like 11 (Bcl2L11 or DR_Bim, K D ϭ 887 nM), whereas all other BH3-motif peptides tested showed no detectable affinity (Fig. 1). Furthermore, GIV66 did not show any detectable affinity for peptides from human pro-apoptotic Bcl-2 family members (data not shown). These findings suggest that GIV66 is the first single-specificity pro-survival Bcl-2 protein.
Considering the unusual ligand-binding profile of GIV66, which is reminiscent of the restricted pro-apoptotic Bcl-2 ligand-binding profiles observed for homodimeric viral Bcl-2 (vBcl-2) homologs (21), we examined the oligomeric state of GIV66 in solution using size-exclusion chromatography (SEC) (Fig. 2). This revealed that GIV66 exists as a dimer in solution. SEC analysis at GIV66 concentrations of 1.5 and 30 mg/ml did not reveal a difference in elution volume, indicating that, within this concentration range, dimerization is not concentration-dependent (data not shown). Similarly, ITC analysis where GIV was titrated into buffer revealed no significant heat changes, suggesting that even at low concentrations, the GIV66 dimer does not dissociate (Fig. 1C).
We next examined the structural basis of GIV66-mediated apoptosis by determining the crystal structure of GIV66 (Table  1). GIV66 adopts the conserved Bcl-2 fold comprising eight ␣-helices that form a globular ␣-helical bundle (Figs. 3A and 4A). Helices 2-5 form the canonical hydrophobic ligand-binding groove found in all pro-survival Bcl-2 proteins that is used to engage BH3-motif sequences of pro-apoptotic Bcl-2 family members. A DALI analysis (28) revealed that Mcl-1 bound to Noxa BH3 (PDB code 4G35) (29) is most similar to GIV66, with an r.m.s.d. value of 2.0 Å over 129 C␣ atoms and a sequence identity of 19%. The closest vBcl-2 structural homolog to GIV66 is murine ␥ herpesvirus M11 bound to Beclin-1 (PDB code 3BL2) (30), with an r.m.s.d. value of 2.3 Å over 131 C␣ atoms and sequence identity of 12%. In the asymmetric unit, only a single chain of GIV66 was found, despite the observation that GIV66 is dimeric in solution (Fig. 2). A PISA analysis (31) did not identify a potential interface and suggested that GIV66 is monomeric. However, examination of the crystal packing revealed a conspicuous interface, where helix 3 from a neighboring symmetry-related molecule packs into the canonical hydrophobic ligand-binding groove formed by helices ␣3 and ␣4, burying 720 Å 2 of solvent-accessible surface in the process (Fig. 3B). An alternative configuration where the back faces of two GIV66 chains meet appears to be a less likely dimeric configuration. Interestingly, the groove-to-groove dimeric arrangement orients the C termini of both chains in the same direction and thus allows insertion of the putative transmembrane regions into the outer mitochondrial membrane. Considering the observation that GIV66 is able to bind DR_Bim BH3, the dimeric configuration would have implications for the ability of GIV66 to engage DR_Bim BH3. Utilization of the canonical ligand-binding groove would require dissociation of the dimer to unblock the groove for DR_Bim binding. Alternatively, GIV66 could remain dimeric and utilize an alternative non-canonical ligand-binding site, such as the one previously observed for Bax (32). To address this, we performed analytical SEC with the GIV66 -DR_Bim BH3 complex. The retention volume of GIV66 -DR_Bim BH3 compared with GIV66 on its own indicates that the complex is smaller in solution compared with the GIV66 dimer and corresponds to a 1:1 complex of GIV66 and DR_Bim BH3 (Fig. 5).
To understand the structural basis of DR_Bim binding to GIV66, we determined the crystal structure of the GIV66 -Bim complex (Figs. 3C and 4B). A single 1:1 GIV66 -DR_Bim complex was present in the asymmetric unit. In the complex structure, the DR_Bim BH3 motif is bound to the conserved hydrophobic ligand-binding grove in an overall configuration similar to that previously observed for other mammalian and viral prosurvival proteins, such as Mcl-1 (Fig. 3D) and BHRF1 (Fig. 3E). The Bim BH3 motif from zebrafish is bound to the canonical GIV66 hydrophobic groove formed by ␣ helices 2-5, burying 851 Å 2 of surface-accessible area in the process. DR_Bim engages four hydrophobic pockets in the GIV66 ligand-binding groove using four conserved hydrophobic residues (Val-124, Leu-128, Ile-131, and Phe-135) (Fig. 6). Furthermore, a con-

Structural and biochemical characterization of GIV66
served ionic interaction between Arg-65 GIV66 from the GIV66 BH1 motif and Asp-133 Bim is present. Additional hydrogen bonds are formed by Asn-54 GIV66 and Arg-129 Bim as well as Asn-47 GIV66 and Glu-121 Bim . A comparison of the ligandbound and ligand-free structures of GIV66 reveals that no significant rearrangements of helices are required to accommo-date DR_Bim. GIV66 superimposes with GIV66 -DR_Bim with an r.m.s.d. value of 1.3 Å over 115 C␣. The bulk of the structural changes are located in the ␣3 helix, which shifts by 1.6 Å to allow the opening of the hydrophobic ligand-binding groove and the connecting loop between ␣2 and ␣3, which becomes shortened and results in a longer ␣2 helix.

Structural and biochemical characterization of GIV66
We then performed SEC in-line with small-angle X-ray scattering (SAXS) to examine the oligomeric state of GIV66 in the presence and absence of DR_Bim, as well as the topology of dimers of GIV66 (Table 2). GIV66 on its own as well as in complex with DR_Bim were measured. GIV66 was injected at a concentration of 5 mg/ml (Fig. 7 (A and B) and Fig. S1), and SAXS data were collected while GIV66 was fractionated on the column. The chromatogram revealed that in solution, GIV66 displayed a small amount of aggregation, as shown by a small peak at the void volume. However, the scattering profile extracted from the well-resolved main peak (Fig. 7A) conforms to a straight line in the low q region on a Guinier plot (Fig. 7C). The molecular mass calculated from the forward scattering intensity (I(0)) (33) for this peak corresponds to a species of ϳ27 kDa, indicating that in solution, GIV66 is a dimer (Fig. 7D). We then injected GIV66 -DR_Bim at a concentration of 5 mg/ml (Fig. 7, A and B). The size-exclusion chromatogram revealed one major peak preceded by a smaller shoulder of higher molecular mass (Fig. 7A). The calculated molecular mass for the main peak corresponds to a species of ϳ19 kDa, indicating that although in solution the GIV66 -DR_Bim complex exists as a mixture of GIV66 -DR_Bim heterodimers and GIV66 dimers that is not fully resolved by size-exclusion chromatography, the main peak corresponds to a GIV66 -DR_Bim heterodimer (Fig.  7, A-D). Because crystal structures of both GIV66 on its own as well as of GIV66 -DR_Bim are available, we compared the experimentally obtained scattering data with the available models using CRYSOL. From those analyses, it is clear that GIV66 scatters commensurate with a dimeric configuration (Fig. 7E), with a 2 of 5.76 and 1.26 for a back-to-back and groove-to-groove dimer arrangement, respectively, and not as a monomer ( 2 of 27.60). In contrast, the main-peak scattering data collected for GIV66 -DR_Bim fits a GIV66 -DR_Bim heterodimer model comprising a single chain of GIV66 and a single chain of Bim (Fig. 7F) with a 2 of 1.60, compared with a 2 of 17.47 and 12.71 for back-to-back and groove-to-groove dimers, respectively. We next attempted to identify which of the GIV66 crystallography dimers was present in solution using a rigid-body modeling approach. Modeling was carried out with CORAL using the monomeric GIV66 structure and generating a groove-to-groove as well as a back-to-back dimer observed previously in the GIV66 crystal packing. A model for a GIV66 back-to-back dimer fits the experimental scattering data poorly ( Fig. 8 and Fig. S1), with a 2 of 4.70. In contrast, a groove-to-groove dimer model resulted in an excellent fit of the scattering curves ( Fig. 8 and Fig. S1) with

Structural and biochemical characterization of GIV66
a value of 2 of 1.16, suggesting that in solution, the canonical ligand-binding groove of GIV66 is buried and thus inaccessible.
Having established a putative dimerization mechanism, we next attempted to disrupt the dimer via structure-guided mutagenesis to verify that GIV66 dimers form via the proposed interface. Inspection of the GIV dimer interface suggested that four residues, Thr-38, Ala-41, Phe-42, and Asn-54, are involved in the dimerization (Fig. 9, A and B). Whereas single mutants of these residues displayed elution volumes on SEC identical to wildtype GIV66 (data not shown), a T38Y/A41Y/F42E triple mutant eluted at a volume commensurate with a monomeric state (Fig. 9C). Furthermore, the GIV66 T38Y/A41Y/F42E triple mutant retained the ability to bind Bim, suggesting that despite the mutations, the Bcl-2 fold was maintained (Fig. 9D).

Discussion
Numerous large DNA viruses have been shown to subvert host-cell apoptotic defenses by utilizing virus-encoded homologs of pro-survival Bcl-2 proteins (3). We now show that grouper iridovirus encodes for a pro-survival Bcl-2 protein, GIV66, that is unusually specific and only shows detectable affinity to DR_Bim. These findings suggest that GIV66 is the first single-specificity pro-survival Bcl-2 protein identified to date (34). In contrast to GIV66, mammalian Bcl-2 proteins display considerable promiscuity and bind the vast majority of BH3-only proteins as well as Bax and Bak. Bcl-b is the sole mammalian pro-survival Bcl-2 protein with a restricted ligandbinding profile and only engages Bim and Bax (35). Other highly specific pro-survival Bcl-2 proteins include the worm CED-9, which only binds the BH3-only proteins EGL-1 and CED-13, and the executor CED-4 (36), whereas the sponge BHP2 has only been shown to engage the BH3 motif of a related sponge Bak-like molecule (37). However, in both cases, the underlying apoptosis regulatory machineries are substantially less complex and involve only a limited number of pro-apoptotic effector proteins.
Among viruses, considerable diversity exists among vBcl-2 proteins regarding mechanism of action and ability to engage pro-apoptotic Bcl-2 proteins. Myxoma virus M11L was shown to inhibit apoptosis by Bax and Bak sequestration (18), whereas vaccinia virus F1L only requires Bim sequestration (17). EBV BHRF1 employs a hybrid mode where both Bim (38) and Bak (39) are targeted. The diversity of mechanisms of action is also reflected in the differing abilities of vBcl-2 homologs to directly bind pro-apoptotic host Bcl-2 proteins. African swine fever virus A179L is ultra-promiscuous and able to bind to all major pro-apoptotic Bcl-2 (15). Similarly, fowlpox virus FPV039 binds all pro-apoptotic Bcl-2 proteins except Bok, whereas sheeppox virus SPPV14 engages all BH3-only proteins except Bad, Bik, and Noxa (23). GIV66 is notable for its unique single-ligand specificity, which is in marked contrast to all other vBcl-2 homologs examined to date. The only vBcl-2 homologs with highly restricted ligand-binding profiles are vaccinia virus F1L (17,40) and deerpox virus DPV022 (21), which bind Bim, Bax, and Bak, and variola virus F1L, which binds Bid, Bax, and Bak (16). All three proteins adopt a domain-swapped homodimeric configuration that can engage two BH3 domain ligands simultaneously, whereas GIV66 utilizes the canonical ligand-binding groove for dimerization and consequently requires breaking of  yellow sticks). B, electron density map encompassing the hydrophobic binding groove of CNP058 in complex with Bim BH3. GIV66 is shown as yellow sticks, whereas DR_Bim BH3 is shown as cyan sticks. The electron density maps are shown as a blue mesh contoured at 1.

Structural and biochemical characterization of GIV66
the dimer to form a 1:1 heterodimeric complex with a BH3 domain ligand.
Mechanistically, it is highly likely that GIV66 inhibits hostcell apoptosis by targeting the pro-apoptotic BH3-only protein Bim. This unusually high level of specificity suggests that Bim plays a key role during the initial host response to viral infection and is in accord with the observation that Bim is the only universal BH3-only pro-survival Bcl-2 antagonist (35). Whereas little is known about the regulation of intrinsic apoptosis in grouper fish, zebrafish have been used extensively to study apoptosis regulation (41). In zebrafish, intrinsic apoptosis appears to be regulated in an analogous manner to mammals, despite the absence of a small number of Bcl-2 family members (8).
Overexpression of Bim is a potent trigger for apoptosis (42), and this ability is lost after targeted mutations in the Bim BH3 domain. Interestingly, given the high level of specificity of GIV66 for DR_Bim, a comparison between the DR_Bim-bound and free form of GIV revealed minimal movement within the hydrophobic ligand-binding groove, with ␣3 moving by 1.6 Å to accommodate DR_Bim in the binding site. This is reminiscent of Bak

Structural and biochemical characterization of GIV66
BH3 binding to the M11L binding groove, which resulted in a modest shift of 2 Å in the ␣4 helix. In both GIV66 and M11L, the binding groove appears to be more optimized for binding of a smaller set of pro-apoptotic ligands and thus requires fewer and subtler structural changes to accommodate a ligand, whereas the more promiscuous mammalian and worm pro- The dark dots and squares represent the frames used for averaging the scattering profiles based on R g (dots) and I(0) normalized over concentration (squares). B, log-log representation of GIV66 (dark gray) and GIV66 -DR_Bim (light gray) scattering profiles extracted from fractionated peaks obtained from SEC-SAXS. C, Guinier plots of GIV66 (circle) and GIV66 -Bim (diamond) with Guinier range highlighted in purple and pink, respectively. D, oligomerization analyses of GIV66 and GIV66 -Bim based on Guinier range. E, CRYSOL analyses of GIV66 main-peak scattering data against monomer and groove-to-groove and back-to-back dimers. F, CRYSOL analyses of GIV66 -DR_Bim main-peak scattering data against monomer and groove-to-groove and back-to-back dimers.
Although the majority of proteins with a Bcl-2 fold appear to be monomeric, homodimerization has been observed in several instances. Mammalian pro-survival Bcl-2 proteins have been shown to form domain-swapped dimers after exposure to heat or high pH (46) involving helices ␣5 and ␣6, as well as trunca-tion of the loop connecting helices ␣1 and ␣2 (47), leading to a domain swap via ␣1. Similarly, mammalian pro-apoptotic Bcl-2, including Bax and Bak, appears to require domain-swap homodimerization via ␣5 and ␣6 as a prelude to forming large oligomeric complexes that perforate the outer mitochondrial membrane (48,49). vBcl-2 proteins that form dimers via a domain swap of the ␣1 helix include pro-survival F1L from

Structural and biochemical characterization of GIV66
vaccinia and variola virus as well as deerpox virus DPV022. Interestingly, whereas vaccinia virus-encoded NF-B modulators with a Bcl-2 fold, such as N1 (50), B14, and A52, also form dimers (51), they do not adopt a domain-swap topology and instead utilize an interface formed by helices ␣1 and ␣6. The dimerization mode observed for GIV66 has not been observed previously and reveals an unusual mechanism for Bcl-2mediated homodimerization. Interestingly, our data suggest that the observed dimerization mode of GIV66 impacts BH3 domain binding via the canonical ligand-binding groove, a feature not observed in any of the other Bcl-2 family dimers to date. Indeed, engagement of DR_Bim by GIV66 leads to loss of the dimer, with a resultant 1:1 GIV66 -DR_Bim heterodimeric complex, whereas other dimers, such as F1L (16,17) and DPV022 (21), as well as Bax (49) and Bak (48) remain dimeric upon ligand binding and are able to form 2:2 heterotetrameric complexes with BH3 ligands. Such a ligand-induced disassociation of a Bcl-2 family protein dimer has not been observed previously and raises the question of whether the oligomeric state of GIV66 may impact the efficacy of GIV-mediated inhibition of apoptosis during infection. Importantly, we were able to confirm the observed dimerization mode of GIV66 by structure-guided mutagenesis, with a GIV66 T38Y/A41Y/F42E triple mutant appearing as a monomer rather than a dimer. However, although GIV66 T38Y/A41Y/F42E bound DR_Bim, the affinity measured was lower than what we observed for wildtype GIV66. Furthermore, no affinity for all other BH3-motif peptides was observed.
A detailed structural analysis of the GIV66 -DR_Bim complex reveals that in addition to the canonical salt bridge and use of four hydrophobic residues, DR_Bim also forms additional interactions using Asn-54 GIV66 and Arg-129 Bim as well as Asn-47 GIV66  In the complex of BHRF1 with Bim BH3, a second ionic interaction between Glu-89 BHRF1 and Arg-63 Bim is found, as well as hydrogen bonds between Ser-97 BHRF1 and Asn-70 Bim and between Gly-99 BHRF1 and Asn-70 Bim . The abundance of additional interactions in the Mcl-1 and BHRF1 complexes with Bim and the presence of ionic interactions rather than hydrogen bonds are reflected in the substantially higher affinities of both pro-survival Bcl-2 proteins for Bim compared with GIV66.
Interestingly, the highest ranked vBcl-2 protein in a DALI search with GIV66 was the complex of M11 bound to Beclin-1 (30). However, despite the overall structural similarity with M11, GIV66 does not bind Beclin-1 and appears to not affect autophagy signaling. Consequently, it appears that Beclin-1 binding by vBcl-2 homologs is restricted to the Herpesviridae and Asfarviridae.
In summary, our study reveals that GIV66 is a dimeric Bcl-2-like protein that suppresses host-cell apoptosis by sequestering DR_Bim. Binding of DR_Bim to the canonical GIV66-binding groove triggers the disassociation of a groove-to-groove GIV66 dimer, raising the question of whether the GIV66 oligomeric state impacts the efficacy of GIV66-mediated inhibition of apoptosis. Our data provide a mechanistic basis for GIV66mediated inhibition of apoptosis and will form the platform for understanding how grouper iridovirus subverts host-cell death defenses in an in vivo setting.

Protein expression and purification
Codon-optimized cDNA of wildtype GIV66 (GenBank TM accession number AAV91093.1) lacking the C-terminal 25 residues (Genscript) as well GIV66 T38Y/A41Y/F42E were cloned into the pGEX-6P-3 vector (Invitrogen) and expressed in the Escherichia coli BL21 Star cell line using the autoinduction method (52) in 2YT medium at 25°C. Overnight cultures were harvested by centrifugation at 6000 rpm (JLA 9.1000 rotor, Beckman Coulter Avanti J-E) for 20 min and resuspended in lysis buffer (50 mM Tris, pH 8.5, 150 mM NaCl, and 1 mM EDTA) supplemented with lysozyme (Sigma-Aldrich) and DNase I (deoxyribonuclease I from bovine pancreas, Sigma-Aldrich). A cell disrupter (Constant Systems Ltd.) was used to lyse the cells at a pressure of 35 kilopascals at 4°C for a single cycle. Cellular debris was removed by centrifuging the lysate at 16,000 rpm (JLA 25.50 rotor, Beckman Coulter Avanti J-E) for 20 min, and the supernatant was further filtered using a 0.22-m syringe filter (Millipore). Affinity chromatography was performed using 5 ml of glutathione-Sepharose 4B resin in a gravity flow column (GE Healthcare) using lysis buffer. On-column cleavage was performed overnight in lysis buffer at 4°C using HRV 3C protease to remove the GST fusion tag. Target protein was collected in the flow-through, concentrated to a volume of 1 ml using a centrifugal concentrator with a 3000 molecular weight cutoff (Amicon Ultra 15), and then subjected to preparative size-exclusion chromatography using a Superdex S75 16/60 column mounted on an Ä KTA Pure system (GE Healthcare) equilibrated with 25 mM HEPES at pH 7.5 and 150 mM NaCl, where it eluted as a single peak. The peak fractions were pooled, concentrated using a centrifugal concentrator with 3000 molecular weight cutoff (Amicon Ultra 15) to a final concentration of 31.5 mg/ml, flash-cooled under liquid nitrogen, and stored at Ϫ80°C. Figure 9. Structure-guided mutagenesis to disrupt the GIV66 dimer interface. A, the GIV66 surface is shown in gray, and the floor of the canonical ligand-binding groove is shown in magenta, with the location of three key mutated residues, Thr-38, Ala-41, and Phe-42, shown in green, blue, and cyan, respectively, with the second GIV monomer shown as an orange ribbon. B, wildtype GIV66 (red), GIV66 bound to DR_Bim (black), and GIV66 T38Y/A41Y/F42E (green) were subjected to size-exclusion chromatography on a Superdex 75 3.2/300 column equilibrated with 20 mM Hepes, pH 7.5, and 150 mM NaCl. GIV66 elutes at a retention volume commensurate with a dimeric state in solution, whereas both GIV66 -DR_Bim and GIV66 T38Y/A41Y/F42E elute at a higher retention volume indicative of a monomeric state in solution. C, raw heats of titration for GIV66 T38Y/A41Y/F42E binding to DR_Bim as obtained from isothermal titration calorimetry.

Structural and biochemical characterization of GIV66 Analytical size-exclusion chromatography
Analytical size-exclusion chromatography was performed using a Superdex S200 3.2/300 column mounted on an Ä KTA Pure system (GE Healthcare) equilibrated with 25 mM HEPES at pH 7.5 and 150 mM NaCl. Molecular weight standards used were albumin, carbonic anhydrase, and cytochrome c. Additional proteins used for comparative purposes were dimeric DPV022 (21) and monomeric BHRF1 (39).

Measurement of dissociation constants
Binding affinities were measured using a MicroCal iTC200 system (GE Healthcare) at 25°C using GIV66 or GIV66 T38Y/ A41Y/F42E in 25 mM HEPES, pH 7.5, 150 mM NaCl at a final concentration of 30 M. BH3-motif peptides were used at a concentration of 300 M and titrated using 19 injections of 2.0 l of ligand. All affinity measurements were performed in triplicate. GIV66 was injected into buffer at a concentration of 300 M. Protein concentrations were measured using a Nanodrop UV spectrophotometer (Thermo Scientific) at a wavelength of 280 nm. Peptide concentrations were calculated based on the dry peptide weight after synthesis. The BH3-motif peptides used were commercially synthesized and were purified to a final purity of 95% (Mimotopes). The data were analyzed using the Origin software package (Microcal). Human peptide sequences used were as described previously (21) 43 ; DR_Bad (Q4V925), 88 ALWAAKKYGQQL-RRMSDEFDKGQMKR 113 ; DR_Beclin-1 (F1RCP1), 102 DGGT-MENLSRRLKVTGDLFDIMSGQT 127 . The ability of these peptides to bind to pro-survival Bcl-2 proteins was confirmed by ITC using African swine fever virus-encoded A179L (data not shown).

Crystallization and structure determination
Crystals of GIV66 were obtained using the sitting-drop method at 20°C in 4.0 M sodium nitrate with 0.1 M sodium acetate, pH 4.6, by mixing 1 l of protein with 1 l of reservoir solution over 500 l of reservoir. The crystals were flash-cooled at Ϫ173°C in mother liquor supplemented with 20% (v/v) ethylene glycol. The crystals belong to space group P4 3 2 1 2 with a ϭ b ϭ 68.03 Å, c ϭ 85.82 Å, ␣ ϭ ␤ ϭ ␥ ϭ 90°and have a solvent content of 63%. Native diffraction data were collected on the MX2 beamline at the Australian Synchrotron using an ADSC Quantum 315r CCD detector (Area Detector Systems Corp., Poway, CA) with an oscillation range of 1.0°/frame and a wavelength of 0.9537 Å. A heavy-atom derivative was obtained by soaking a crystal in mother liquor supplemented with 0.5 mM mercury acetate for 3 h before it was flash-cooled at Ϫ173°C as described above. Derivative diffraction data were collected on the MX1 beamline at the Australian Synchrotron using an ADSC Quantum 210 CCD detector (Area Detector Systems Corp.) with an oscillation range of 1.0°/frame and a wavelength of 1.0064 Å. All diffraction data were processed using xia2 (53) and scaled using AIMLESS (54). Heavy-atom sites and substructure solution were determined using SHELX (55). The model was built in Coot (56) and refined using Phenix (57) to a resolution of 1.5 Å and a final R-factor of 0.183 (R-free of 0.214). A Ramachandran plot indicates that 98% of the residues are in the core region with none in the disallowed region.
A complex of GIV66 with DR_Bim was prepared by mixing protein and peptide in a 1:1.25 molar ratio (58). The reconstituted complex was concentrated to 31.5 mg/ml using a 3000 molecular weight cutoff centrifugal concentrator (Millipore), flash-cooled, and stored under liquid nitrogen. Crystals of GIV66 -DR_Bim BH3 were obtained using the sitting-drop method at 20°C in 0.1 M CHES, pH 9.5, 20% PEG 8000 by mixing 1 l of protein with 1 l of reservoir solution over 500 l of reservoir. The crystals were flash-cooled at Ϫ173°C in mother liquor supplemented with 20% (v/v) ethylene glycol. The crystals belong to space group P6 5 with a ϭ b ϭ 71.8 Å, c ϭ 56.55 Å, ␣ ϭ ␤ ϭ 90°and ␥ ϭ 120°and have a solvent content of 56%. Native diffraction data were collected and processed as described above. The complex structure was solved by molecular replacement with PHASER (59) using the structure of GIV66 as a search model. The final model was manually built using Coot and refined using Phenix to a resolution of 1.75 Å and a final R-factor of 0.160 (R-free of 0.200). 99% of the residues are in the core regions of the Ramachandran plot with no residues in the disallowed regions. All data collection and refinement statistics are summarized in Table 1. All images were generated using the PyMOL Molecular Graphics System, version 1.8 (Schrödinger, LLC, New York). All software was accessed using the SBGrid suite. Coordinate files have been deposited in the Protein Data Bank under the accession codes 5VMN and 5VMO. All raw diffraction images were deposited in the SBGrid Data Bank (60) under 10.15785/SBGRID/502 and 10.15785/SBGRID/501.

In-line SEC small-angle X-ray scattering
GIV66 and GIV-DR_Bim complex samples at 5 mg/ml (75 l) were subjected to size-exclusion chromatography using a Superdex 75 5/150 column equilibrated with 50 mM sodium phosphate, pH 7.6, 50 mM NaCl at a flow rate of 0.2 ml/min. During elution, SAXS analysis was conducted in-line via a coupled coflow sample sheath flow environment run at a fractional sample flow rate of 0.5 (61). SAXS data were acquired on the SAXS/WAXS beamline at the Australian Synchrotron (62), which has been optimized for low instrument background. Data were acquired using a camera length of 1.430 m, providing q-ranges of 0.012-0.602 Å Ϫ1 at 12 keV at a flux of 2.5 ϫ 10 12 photons/s, using continuous exposures with a 1-s integration time using a Pilatus 1M detector. Images were inspected, averaged, and subtracted using PRIMUS from the ATSAS suite of SAXS data analysis tools (63). Data analysis was carried out as described by the 2017 publication guidelines, using the R g and the I(0) normalized over concentration scattering data selection over a given peak (64). Initial data analysis based on

Structural and biochemical characterization of GIV66
reduced data in the form of data files using Guinier plots to calculate the radius of gyration was conducted using AUTORG from the ATSAS suite of SAXS data analysis tools (63). Uncertainties from Guinier fits are 2 S.E. values of the slope of fitted linear regressions of ln(I) versus q 2 . GIV66 and GIV66 -DR_Bim scattering curves of the main peaks were compared with the available PDB models using CRYSOL (65). Rigid-body modeling of GIV66 scattering data with two possible dimeric configurations observed in the crystal structure was carried out using CORAL (63). All data collection and analysis statistics are summarized in Table 2.