Structure of the Rhesus Monkey Trim5α PRYSPRY Domain, the HIV Capsid Recognition Module

Structure of the rhesus monkey TRIM5α PRYSPRY domain, the HIV capsid recognition module

Nikolaos Birisa, Yang Yangb, Alexander B. Taylora, Andrei Tomashevskia, Miao Guoa, P. John Harta,c, Felipe Diaz-Grifferob, and Dmitri N. Ivanova,d,1

aDepartment of Biochemistry and dCancer Therapy and Research Center, University of Texas Health Science Center at San Antonio, San Antonio, TX 78229; bDepartment of Microbiology and Immunology, Albert Einstein College of Medicine, Bronx, NY 10461; and cGeriatric Research, Education, and Clinical Center, Department of Veterans Affairs, South Texas Veterans Health Care System, San Antonio, TX 78229

Edited by* Stephen P. Goff, Columbia University College of Physicians and Surgeons, New York, NY, and approved June 15, 2012 (received for review February 28, 2012)

Tripartite motif protein TRIM5α blocks retroviral replication after pandemic when the simian immunodeficiency virus (SIV) passed cell entry, and species-specific differences in its activity are deter- from chimpanzees into a human host (1, 8, 9, 14). mined by sequence variations within the C-terminal B30.2/ TRIM5α binds to the assembled capsid of the mature viral core PRYSPRY domain. Here we report a high-resolution structure of a rather than the monomeric capsid protein, suggesting that TRI- TRIM5α PRYSPRY domain, the PRYSPRY of the rhesus monkey TRI- M5α may act as a pattern-recognition molecule (4, 8, 9). Remark- M5α that potently restricts HIV infection, and identify features in- ably, an EM investigation revealed that the purified tripartite volved in its interaction with the HIV capsid. The extensive capsid- motif of TRIM5α forms hexagonal arrays that match the symme- binding interface maps on the structurally divergent face of the try of the assembled retroviral capsid (15, 16). This observation protein formed by hypervariable loop segments, confirming that suggested a model of TRIM5α–capsid interaction, in which the TRIM5α evolution is largely determined by its binding specificity. hexagonal assembly of TRIM5α would juxtapose the PRYSPRY Interactions with the capsid are mediated by flexible variable loops domains with the regularly spaced epitopes on the surface of the via a mechanism that parallels antigen recognition by IgM antibo- assembled capsid, leading to specific, high-affinity binding of dies, a similarity that may help explain some of the unusual func- TRIM5α to the retroviral core. Mutations that interfere with tional properties of TRIM5α. Distinctive features of this pathogen- TRIM5α self-association also disrupt capsid cosedimentation BIOPHYSICS AND recognition interface, such as structural plasticity conferred by the confirming the importance of TRIM5α multimerization and the COMPUTATIONAL BIOLOGY mobile v1 segment and interaction with multiple epitopes, may avidity effect in capsid recognition (15, 17–19). Such multivalent, allow restriction of divergent retroviruses and increase resistance high-avidity interactions pose significant experimental chal- to capsid mutations. lenges. The binding of the individual PRYSPRY domains to the capsid surface may be very weak, which may be one of the – etroviral restriction factors are important components of in- reasons why direct PRYSPRY capsid interactions have not yet Rnate immunity defenses that protect higher organisms against been demonstrated by biochemical, biophysical, or structural retroviral pathogens. The splicing variant alpha of tripartite motif means despite the extensive mutagenesis and evolutionary data – five (TRIM5α) is particularly remarkable because of the potent suggesting a PRYSPRY capsid interface. activity that the TRIM5α of rhesus monkey (rhTRIM5α) displays The arrangement of the HIV capsid protein in the mature against HIV-1 (1). TRIM5α is a member of the tripartite motif retroviral core is well-characterized, and the atomic-resolution model of the entire assembled structure is now available (20); (TRIM) family of proteins increasingly recognized for their role α in innate immunity (2–4). All TRIM proteins share a conserved in contrast, structures of the primate TRIM5 PRYSPRY N-terminal tripartite domain motif consisting of a RING domain, domains have remained elusive, limiting our insight into capsid recognition by TRIM5α. Here we describe the structure of the followed by one or two B-box domains and then by a coiled-coil α segment. The composition of the C-terminal part of TRIMs var- rhesus TRIM5 PRYSPRY domain determined by a hybrid ies, and about one half of approximately 100 TRIM proteins in experimental approach that combines NMR spectroscopy and X-ray crystallography. The structure, NMR titration experiments, the human genome contain a C-terminal PRYSPRY domain (also – known as B30.2 domain), a protein-protein interaction module and site-directed mutagenesis suggest an extensive capsid (2, 3, 5). PRYSPRY interface dominated by the highly mobile v1 loop Rhesus TRIM5α is a cytoplasmic protein that normally blocks of the PRYSPRY domain. The capsid recognition mechanism, HIV replication after cell entry but prior to completion of reverse which is reminiscent of antigen recognition by the natural and the transcription (1). Viral determinants of susceptibility to TRIM5α- early immune response antibodies because it also involves mobile variable loops and high-avidity binding, may facilitate restriction mediated restriction are located within the capsid protein (6, 7), α and the restriction potency correlates with the ability of the of divergent retroviruses and increase resistance of TRIM5 to cytosolic TRIM5α to cosediment with the assembled viral capsid capsid mutations. (8, 9), strongly suggesting that direct interactions of TRIM5α with the viral capsid are required for restriction. The PRYSPRY do- Author contributions: N.B., Y.Y., A.B.T., A.T., M.G., F.D.-G., and D.N.I. designed research; main of TRIM5α is believed to form most of the capsid–TRIM5α N.B., Y.Y., A.B.T., A.T., M.G., F.D.-G., and D.N.I. performed research; N.B., Y.Y., A.B.T., interface as species-specific sequence variations within the A.T., M.G., P.J.H., F.D.-G., and D.N.I. analyzed data; and N.B., A.B.T., P.J.H., F.D.-G., and D.N.I. wrote the paper. PRYSPRY domain account for differences in the viral specificity of the TRIM5α-mediated restriction (10–12). In fact, the TRI- The authors declare no conflict of interest. M5α PRYSPRY domains contain some of the most rapidly chan- *This Direct Submission article had a prearranged editor. ging protein segments within primate genomes, an illustration of Data deposition: The crystallography, atomic coordinates, and structures factors have been deposited in the Protein Data Bank (PDB), www.pdb.org (PDB ID codes 3UV9 and 2LM3). how the evolutionary antagonism between retroviruses and their The chemical shift data have been deposited in the Biological Magnetic Resonance Data primate hosts accelerates remodeling of the host-pathogen inter- Bank (BMRB code: 18097). face (13). Most notably, recent evolution of the human TRIM5α 1To whom correspondence should be addressed. E-mail: [email protected]. PRYSPRY domain resulted in the variant that has poor affinity This article contains supporting information online at www.pnas.org/lookup/suppl/ for the HIV capsid, the vulnerability that contributed to the AIDS doi:10.1073/pnas.1203536109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1203536109 PNAS Early Edition ∣ 1of6 Downloaded by guest on October 2, 2021 Results contacts. The pairwise interproton distance restraints obtained Structure and Dynamics of the rhTRIM5α PRYSPRY. Production of the from the NOE data limit the number of conformations that the recombinant TRIM5α PRYSPRY domain in bacterial expression v1 region can possibly adopt. The resulting hybrid X-ray/NMR systems is impeded by the low total protein expression and by the structure is shown in Fig. 1B. The v1 loop protrudes from the low soluble-to-insoluble protein ratio. Two features of our protein otherwise globular PRYSPRY domain and can adopt multiple expression strategy were important for high yields of soluble re- divergent conformations without violating the NOE-derived recombinant protein: (i) codon optimization of TRIM5α sequences straints. These findings are consistent with the mobility of the and protein expression in the BL21 Rosetta 2 (EMD/Novagen) v1 segment apparent from the relaxation measurements. bacterial strains and (ii) use of the N-terminal solubility enhancement tag derived from the B1 domain of protein G (GB1) (21). Structural Evolution of PRYSPRY Domains. Comparison of the TRI- Detailed description of protein production and crystallization can M5α PRYSPRY structure to other known PRYSPRY structures be found in SI Experimental Procedures. offers insights into the molecular evolution of the primate TRI- The wild type version of rhTRIM5α PRYSPRY was refractory M5α and the expansion of PRYSPRY-containing TRIM proteins to crystallization; therefore, we used NMR to identify mobile seg- within the genomes of higher organisms. TRIM21 is the closest ments within PRYSPRY that could interfere with protein crystal evolutionary cousin of TRIM5α for which the PRYSPRY struc- packing. Measurement of the relaxation parameters of the pro- ture is known, and the interactions with its binding partner, the tein backbone amides revealed that the v1 variable loop of the invariant Fc segment of IgG, have been biochemically and struc- PRYSPRY domain was indeed highly mobile (Fig. 1A). Analysis turally characterized (24, 25). Structural comparison of the of the relaxation data and modeling of the TRIM5α structural PRYSPRY domains of TRIM5α and TRIM21 reveals that, core guided production of a v1-deleted rhTRIM5α PRYSPRY whereas the core beta sandwich architecture and the packing construct (designated Δv1 elsewhere in the text). In this con- of the hydrophobic core residues are essentially the same for the struct, the 24-residue v1 segment of the rhPRYSPRY (aa 326– two proteins, their surfaces have diverged. The structural diver- 349) was replaced by a two-amino-acid linker, Ala-Gly. The Δv1 gence between TRIM5α and TRIM21 PRYSPRY is primarily PRYSPRY variant produced well-diffracting crystals at multiple manifested in the distinct backbone conformations within the so- crystallization conditions enabling determination of the crystal called variable loop regions v1 through v4, which display low se- structure at 1.55 Å resolution. Crystallographic statistics are given quence homology between different PRYSPRYs. Remarkably, in Table 1, and the unusual packing of the protein in the crystal is the structural differences are clustered on one face of the protein discussed in Experimental Procedures. (Fig. 2A). This structurally divergent protein face forms the bind- The v1 loop is critical for retroviral restriction. Numerous ing interface in TRIM21,Gustavus, and SPSB1-4 proteins, the mutations within the loop are known to disrupt restriction, and interactions of which with their binding partners have been struc- conversely, mutations within the v1 of the human TRIM5α can turally characterized (24, 26, 27). In primate TRIM5α, the resi- restore HIV restriction with potency approaching that of the dues that display strong positive selection in recent evolution as rhesus protein (12, 22, 23). Moreover, the composition of the well as sequence insertion sites map to the same protein surface v1 loop has been under strong positive selection during recent (10, 11, 13), offering strong evidence that the TRIM5α PRYSPRY primate evolution (13). We analyzed the Nuclear Overhauser uses the same interface for binding to the retroviral capsids Effect (NOE) data from the 15N- and 13C-dispersed NOE–het- (Fig. 2B). The observation that the structural divergence between eronuclear single quantum coherence (HSQC) experiments per- closely related PRYSPRY domains is confined to the interaction formed on the wild type version of rhTRIM5α PRYSPRY to interface illustrates that the binding specificity of the PRYSPRY calculate v1 conformations consistent with the observed NOE domain has conferred enhanced fitness and shaped the evolution

15 15 1 Fig. 1. PRYSPRY structure and dynamics. (A) The NT1,T2 and f N- Hg NOE measurements of the rhTRIM5α PRYSPRY. (B), The structure of the rhTRIM5α PRYSPRY domain. Ten lowest-energy conformations of the v1 loop calculated using NOE restraints are shown as colored ribbons. The rest of the structure determined by X-ray crystallography is shown as a light green cartoon. The strands forming the beta sandwich PRYSPRY fold are numbered, and the variable loop regions are labeled and highlighted in dark green.

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1203536109 Biris et al. Downloaded by guest on October 2, 2021 Table 1. Data collection and refinement statistics of surface PRYSPRY residues result only in partial reduction of the restriction activity, therefore we chose the v1C-5A mutant X-ray statistics of the ΔV1 rhSPRY structure (3UV9) because it is severely impaired in HIV restriction, but is never- Data collection theless structurally intact (SI Text). The characteristic enhanced Space group R3 (hexagonal setting) broadening of the NMR signals observed for the wild-type rhesus Cell dimensions PRYSPRY is not present in v1C-5A mutant (Fig. 2C), suggesting a, b, c (Å) 86.6, 86.6, 77.3 α β γ that the PRYSPRY–capsid interaction observed by NMR corre- , , (°) 90, 90, 120 α Wavelength 0.979 lates with the restriction activity of TRIM5 . Resolution (Å) 30–1.55 The binding between CA-NTD and PRYSPRY is very weak, be- R μ sym* 0.044 (0.456) cause even when the two proteins are mixed at 200 Meachthe I∕σI 31.4 (4.5) binding is not close to saturation. By extending the NMR titration Completeness (%) 98.7 (98.4) experiments to higher CA-NTD concentrations we can estimate Redundancy 6.4 (6.4) that the dissociation constant of the PRYSPRY/CA-NTD binding Refinement is approximately 410 90 μM(SI Text). This remarkably weak Resolution (Å) 26.9–1.55 No. reflections 29851 interaction of the isolated PRYSPRY domain with CA-NTD con- R ∕R 0 148∕0 196 firms the importance of the avidity effect in TRIM5α recognition work free . . No. Atoms of the mature viral cores. The fact that the binding occurs in the Protein 1462 intermediate NMR exchange regime (see above) even for such k Solvent 155 weak interaction indicates that the on rate for the binding is slow, B-factors consistent with the observation that the binding is accompanied by Protein 27.6 significant conformational changes. Solvent 39.5 We do not observe NMR signals for some backbone segments R.m.s deviations Bond lengths (Å) 0.005 at the binding interface, raising the possibility that the residues Bond angles (°) 0.949 not visible by NMR may still be important for capsid binding. In- NMR statistics of the V1 loop structure(2LM3) deed, site-directed mutagenesis revealed that features of the structurally variable face of the PRYSPRY domain outside of the Distance constraints for V1 loop v1 and v2 loops are important for restriction (Fig. 2D). For ex- Total number of NOE 259 ample, alanine substitution of residue E410 in the v3 region, BIOPHYSICS AND Intraresidue 103 α Sequential 102 which is important for MLV restriction by the human TRIM5 COMPUTATIONAL BIOLOGY Medium-range NOEs (1 < ji-jj < 4)49 (28), and the neighboring V412, displayed as much attenua- Long-range NOEs (ji-jj > 5)5 tion of HIV-1 restriction by rhTRIM5α as did the point mutations Deviations from idealized covalent geometry in the v1 loop. Mutation of R484 in the v4 loop reduced restric- Bonds (Å) 0.0033 ± 0.0001 tion potency from 100-fold to 21-fold. Thus, although the v1 and Angles (°) 0.5829 ± 0.080 v2 loops undergo the most dramatic and observable conforma- Impropers (°) 0.4307 ± 0.0137 tional changes upon capsid binding they are not the sole deter- Ramachandran statistics minants of HIV restriction on the variable face of TRIM5α Most favored region (%) 16.7 D Additionally allowed region (%) 38.9 PRYSPRY (Fig. 2 ). Together these results suggest an extended Generously allowed region (%) 33.3 interaction interface with multiple epitopes akin to the one ob- Disallowed region (%) 11.1 served in the TRIM21-IgG Fc complex. *Values in parentheses are for the highest-resolution shell. Discussion Here we describe the high-resolution structure of the rhesus of primate TRIM5α, and probably of other PRYSPRY-containing TRIM5α PRYSPRY domain and map its interactions with the TRIM proteins. HIV capsid, offering insight into the structural basis of retroviral capsid binding. We find that capsid recognition by the PRYSPRY Mapping of the PRYSPRY–Capsid Interface. To further characterize domain displays three distinctive features: (i) the capsid-binding the surface involved in the PRYSPRY–capsid interactions we per- interface of the PRYSPRY domain is located on the structurally 15 formed NMR titrations of the N-labeled rhPRYSPRY with the divergent face of the protein formed by hypervariable loops de- unlabeled N-terminal domain of the HIV capsid (CA-NTD). The signated v1 through v4, (ii) the binding surface is dominated by a NMR spectra revealed that the addition of CA-NTD to PRYSPRY mobile v1 segment that undergoes considerable structural rear- produced broadening/weakening of the PRYSPRY NMR signals rangements upon capsid binding , and (iii) the interaction of the as a function of the CA-NTD concentration (SI Text). The rates isolated PRYSPRY domain with the capsid is weak and likely to of signal broadening were not uniform, but rather were most pro- involve multiple capsid epitopes. nounced for the signals of the v1 and v2 segments and for the The structural features of the TRIM5α PRYSPRY do- residues in the N-terminal segment preceding the v1 loop (Fig. 2C main reveal striking similarities between capsid recognition by and D). The broadening of the NMR signals is indicative of TRIM5α and antigen recognition by germ line antibodies. Germ the conformational changes in the v1 and v2 segments of the line or near germ line amino acid sequences of the variable do- PRYSPRY structure upon CA-NTD binding. The fact that the mains are frequently observed in the natural and early IgM anti- broadening rather than peak shifting is observed is characteristic bodies (29), which are commonly polyreactive (capable of binding of the intermediate exchange regime on the NMR timescale, when structurally unrelated epitopes), bind antigens relatively weakly, the exchange rates between the bound and the unbound forms are and display high-avidity binding owing to the presence of 10 anti- comparable to frequency differences between NMR signals of the gen-binding sites in a single pentameric IgM macromolecule. two states. These properties make IgM distinct from the affinity-matured To test whether the PRYSPRY–capsid interaction observed by IgG antibodies that bind their antigens with very high affinity NMR is physiologically relevant we performed the same titration and specificity. Structural and biophysical studies of antigen bind- with the v1C-5A mutant of the rhesus PRYSPRY, which contains ing by the germ line versus affinity-matured antibodies revealed a five-alanine substitution of the NFNYC (aa 345–349) segment that the antigen-binding loops of the germ line antibodies and the in the C-terminal portion of the mobile v1 loop. Point mutations antibodies of the early immune response are very mobile and

Biris et al. PNAS Early Edition ∣ 3of6 Downloaded by guest on October 2, 2021 Fig. 2. The capsid-binding surface of PRYSPRY. (A) rhTRIM5α PRYSPRY (green) superimposed on the structure of TRIM21 PRYSPRY (blue) in complex with the IgG Fc chain (white). Structurally divergent variable loops (v1 through v4) are highlighted and labeled. (B) The view of the predicted capsid-interacting surface of the rhTRIM5α PRYSPRY. Residues under positive evolutionary selection are colored red and labeled (13). The mobile surface formed by the v1 loop is colored yellow, whereas v2, v3, and v4 segments are highlighted in dark green. (C) The broadening rates of the 15N TROSY HSQC signals plotted for all assigned residues of the wild-type rhTRIM5α PRYSPRY (blue) and the v1C-5A PRYSPRY mutant (green). (D) The view of the capsid-binding surface illustrating the results of NMR titrations (Fig. 2C) and mutagenesis. Backbone segments perturbed by HIV CA-NTD titrations are colored in red, unaffected in green, and unassigned in gray. Residues tested by point mutagenesis are shown in magenta, whereas the segment changed to five alanines in the v1C-5A mutant is shown in cyan. HIV-1 restriction activity measurements for mutations located on the variable face of the rhTRIM5α PRYSPRY are listed in the table.

undergo significant structural rearrangements upon antigen In a functional parallel to IgM polyreactivity, primate binding, whereas in the affinity-matured antibodies amino acid TRIM5α proteins are remarkable for their ability to restrict di- changes acquired via somatic hypermutation and clonal selection vergent retroviruses that share few similarities in the amino acid lock these loops in well-defined conformations and thereby trans- composition of their capsids. Rhesus TRIM5α PRYSPRY, for ex- form antigen binding into a high-affinity, lock-and-key mode of ample, displays potent restriction activity against HIV-1, but is interaction (30–32). The structural plasticity of the antigen bind- also moderately active against MLV (34). Conformational plas- ing surface explains the relatively low antigen affinity of germ line ticity of the PRYSPRY domain is the most likely explanation antibodies and is essential for their polyreactivity, a distinctive of the broad specificity displayed by some TRIM5α variants, ana- property viewed as critical for diverse biological functions of logous to the mechanism of the IgM polyreactivity. IgM (29). Similarities to IgM should be taken into consideration when The long and mobile v1 loop that protrudes from the otherwise using the high-resolution PRYSPRY structure reported here to globular structure is a characteristic feature of the TRIM5α interpret extensive mutagenesis data available for TRIM5α PRYSPRY. The mobility of v1 in TRIM5α and the associated loss and to design future studies. Just as in the case of antigen–anti- of conformational entropy upon capsid binding almost certainly body interactions, TRIM5α mutations that affect binding are not contribute to the weak affinity of the isolated PRYSPRY domain necessarily altering direct contacts with the ligand, but rather may for the capsid. In contrast, structural and biophysical character- act by limiting the conformational space accessible to the v1 vari- ization of TRIM21-IgG interactions revealed that in the TRIM21 able loop segment. For example, v1 dynamics may help explain PRYSPRY the v1 loop is rather rigid, flanking an extensive pre- how point mutations within the v1 of the human TRIM5α can formed binding interface and resulting in a significantly tighter alter its viral specificity profile or restore its activity against HIV interaction with its target, the Fc chain of IgG (25). Such differ- (12, 22, 23, 35, 36). ences in v1 mobility resemble differences between germ line and It is informative to consider the relative sizes of the protein affinity-matured antibodies and most likely reflect different bio- assemblies and binding interfaces involved in the TRIM5α– logical functions of TRIM5α and TRIM21. TRIM21 is thought to capsid interaction. Fig. 3 shows one possible arrangement of function as a cytosolic IgG receptor (33) that binds to the struc- TRIM5α at the surface of the HIV mature core proposed in turally invariant Fc segment, whereas TRIM5α binds directly to a the EM study of TRIM5α hexagonal assemblies (15). We still lack viral structural element; thus, the mobility of the v1 region may be the intermolecular PRYSPRY–capsid distance restraints that beneficial in evolutionary terms because it would allow TRIM5α would allow identification of the PRYSPRY binding sites, so to adapt to mutations on the surface of the capsid. The relatively the placement of the PRYSRPY domains on the capsid surface low affinity of the individual PRYSPRY modules for the capsid is completely arbitrary and many different arrangements could be resulting from v1 mobility is compensated for by the avidity effect envisioned. The figure, however, illustrates two important points. produced by higher-order multimerization of TRIM5α into hex- First, the placement of the PRYSPRY domains is relatively sparse agonal arrays that match the symmetry of the assembled retrovir- and TRIM5α uses only a small fraction of the capsid monomers in al capsids (15) (Fig. 3A). the core for the binding (Fig. 3A). Second, the variable face of

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1203536109 Biris et al. Downloaded by guest on October 2, 2021 AB

coiled coil SPRY B-box L1 L2 RING

Fig. 3. Possible arrangement of the TRIM5α PRYSPRYon the surface of the HIV-1 mature core. (A) Domain structure of TRIM5α and one of the several possible relative orientations of the hexagonal arrays formed by TRIM5α and the HIV capsid (adapted from ref. (15)). HIV capsid assembly is derived from PDB ID codes 3DIK (47) and 3H47 (48). Only the N-terminal domains of the capsid that form the outer surface of the core are shown. Capsid C-terminal domains are omitted for clarity because they face the interior and are not accessible to interact with the PRYSPRY. Capsid monomers within the hexamers are shown in alternating colors (green and blue). (B) The structure of the rhTRIM5α PRYSPRY (orange) with the variable interaction face oriented toward the surface of the assembled capsid. The figure is only meant to illustrate the relative sizes of the PRYSPRY, CA-NTD, and the hexagonal arrays formed by the TRIM5α and the HIV capsid.

PRYSPRY, oriented toward the capsid surface in Fig. 3B is larger tallization peculiarities not linked to the function of the native than the solvent-exposed surface of the capsid N-terminal do- proteins. We believe the latter to be the case for the Δv1 main monomer and is likely to recognize epitopes spanning more PRYSPRY crystals, as the wild type and the Δv1 PRYSPRY con- than one capsid monomer within the hexamer. Such an extended structs are predominantly monomeric in our NMR samples, and interaction surface with multiple epitopes is consistent with nu- the observed domain-swapped trimerization obstructs protein merous functional studies (28, 35, 37) and with the analysis of the surfaces expected to contact the viral capsid. The coordinate file α MLV mutants that escape restriction by TRIM5 (38), which of the unswapped rhesus Δv1 PRYSPRY monomer with the B showed that point mutations that compromise restriction are factors preserved is available at http://ivanovlab.uthscsa.edu/ BIOPHYSICS AND located almost 30 Å apart on the capsid surface. trim5alpha/. The structure of the monomer is not perturbed by COMPUTATIONAL BIOLOGY In summary, our structural and biophysical results reveal that α domain-swapped oligomerization with the exception of the short the PRYSPRY domain of TRIM5 displays distinctive features of linker region connecting the swapped element to the rest of the an immune recognition module that has evolved to bind patho- structure (in our case the Ala-Gly linker used to substitute the v1 gen-associated molecular patterns displayed on the surface of loop) (SI Text). the retroviral capsid. Dynamics of the v1 segment located on the structurally variable capsid-binding surface of the domain and 15 – NMR Spectroscopy. The NT1 and T2 relaxation parameters and parallels to IgM antigen interactions suggest that mutations that 1 15 affect the mobility of this variable loop may have strong effects on [ H- N]-heteronuclear NOE values were obtained from stan- the affinity and specificity of capsid binding. Conformational dard experiments (45) performed at 298 K on a 700 MHz spectro- 15 plasticity of the binding interface, weak capsid affinity of the iso- meter using an N-labeled approximately 0.3 mM rhPRYSPRY lated PRYSPRY domains and interaction with multiple epitopes sample. are important factors to be considered in the studies of capsid NOE-based distance restraints for the variable loop v1 were recognition by TRIM5α. extracted from 700-MHz three-dimensional NOESY-[1H,15N]- HSQC and 700-MHz three-dimensional NOESY-[1H,13C]- Experimental Procedures HSQC spectra. No long-range NOEs were observed between the Protein Production and Purification. Detailed description of protein v1 and the protein outside of v1. The final structures of the production and crystallization can be found in SI Experimental rhPRYSPRY were calculated using XPLOR-NIH. Simulated an- Procedures. nealing with NOE-derived distance restraints was applied to the variable loop v1 only (residues 325–349) whereas positions of the X-ray Crystallography. The structure of rhPRYSPRY was deter- rest of the atoms in protein were held fixed at the coordinates mined by the molecular replacement method implemented in determined by X-ray crystallography. Structural statistics shown PHASER (39), using Protein Data Bank (PDB) ID code 2WL1 listed in Table 1 are typical of a disordered loop region in a pro- (40) as the search model. Coordinates for the model were refined tein structure determined by NMR spectroscopy. using PHENIX (41) including simulated annealing with torsion NMR titration of 15N-labeled rhPRYSPRY ( approximately angle dynamics and individual anisotropic displacement para- 0.19 mM) with unlabeled CA-NTD was conducted at 298 K meter refinement, and alternated with manual rebuilding using on a Bruker Avance 700 MHz spectrometer by recording a series COOT (42). Data collection and refinement statistics are shown of 1H-15N Transverse Relaxation Optimized Spectroscopy, in Table 1. 1∶0 TRIM5αΔv1 PRYSPRY is assembled into a domain-swapped TROSY-HSQC spectra containing molar ratios from up 1∶2 5 trimer in the crystal (SI Text). The N-terminal segments (aa 292– to . . For each cross-peak the intensity was evaluated using 325) are cyclically swapped in the monomers within the trimer. NMRView and plotted as a function of the protein concentration R Apparently, the v1 deletion and the high protein concentration ratio. Broadening rates for each peak were determined by at crystallization conditions lead to the formation of the do- fitting signal intensity plots above with simple exponential decays −Rc main-swapped trimer, a stabilized structure that promoted crystal IðcÞ¼Ioe , where c is the protein concentration ratio (CA- growth. Oligomerization by swapping of substructural elements NTD/PRYSPRY). can be functionally significant, especially for proteins occurring The Kd value of the interaction was estimated from a similar in the intrinsically crowded multicopy environments such as titration but extended to higher CA-NTD concentrations (1∶10 viral shells, protein filaments, flagella, etc. (43, 44), but many [PRYSPRY]/[CA-NTD] ratio) (SI Text). The details of the fitting domain-swapped oligomers observed in crystals are simply crys- procedure are described in SI Experimental Procedures.

Biris et al. PNAS Early Edition ∣ 5of6 Downloaded by guest on October 2, 2021 All NMR parameters determined for each residue and plotted titrated by serial dilution on Cf2Th cells to determine the concen- in Fig. 1A and Fig. 2C are listed in SI Text. tration of infectious viruses.

Restriction Assays. Retroviral vectors encoding wild-type or mu- ACKNOWLEDGMENTS. We thank Dr. Oleg Tsodikov and the staff of Life tant rhTRIM5α proteins were created using the pLPCX vector, Sciences Collaborative Access Team (LS-CAT) sector at the Advanced Photon as previously described (46). Cf2Th canine thymocytes were Source at the Argonne National Laboratory for assistance with collection 5 μ ∕ of the X-ray diffraction data. Support for the NMR Core Facility and the transduced with the different constructs and selected in g mL X-ray Crystallography Core Laboratory is provided by University of Texas of puromycin (Sigma). Recombinant HIV-1 expressing GFP was Health Science Center Executive Research Committee and the Cancer Therapy prepared as described (35). All recombinant viruses were pseu- Research Center (P30 Cancer Center Support Grant from the National Cancer dotyped with the vesicular stomatitis virus G glycoprotein. For Institute CA054174). This research was supported in part by National Insti- infections, 3 × 104 Cf2Th cells seeded in 24-well plates were tutes of Health (NIH) Grants R21 AI068548 (to D.I.), R21 AI084612 (to D.I.), R01 AI087390 (to F.D.-G.), and the Robert A. Welch Foundation grant incubated at 37 °C with virus for 24 h. Cells were washed and re- AQ-1399 (to P.J.H.). The Scholar Award from the Cancer Prevention and turned to culture for 48 h, and then subjected to FACS analysis Research Institute of Texas (CPRIT) (D.I.) and the NIH K99/R00 Pathway to with a FACScan (Becton Dickinson). HIV-1 viral stocks were Independence Award (F.D.-G.) are gratefully acknowledged.

1. Stremlau M, et al. (2004) The cytoplasmic body component TRIM5alpha restricts HIV-1 25. Keeble AH, Khan Z, Forster A, James LC (2008) TRIM21 is an IgG receptor that is struc- infection in Old World monkeys. Nature 427:848–853. turally, thermodynamically, and kinetically conserved. Proc Natl Acad Sci USA 2. Ozato K, Shin DM, Chang TH, Morse HC, III (2008) TRIM family proteins and their emer- 105:6045–6050. ging roles in innate immunity. Nat Rev Immunol 8:849–860. 26. Filippakopoulos P, et al. (2010) Structural basis for Par-4 recognition by the SPRY 3. McNab FW, Rajsbaum R, Stoye JP, O’Garra A (2011) Tripartite-motif proteins and innate domain- and SOCS box-containing proteins SPSB1, SPSB2, and SPSB4. J Mol Biol immune regulation. Curr Opin Immunol 23:46–56. 401:389–402. 4. Pertel T, et al. (2011) TRIM5 is an innate immune sensor for the retrovirus capsid lattice. 27. Woo JS, Suh HY, Park SY, Oh BH (2006) Structural basis for protein recognition by – Nature 472:361 365. B30.2/SPRY domains. Mol Cell 24:967–976. 5. Han K, Lou DI, Sawyer SL (2011) Identification of a genomic reservoir for new TRIM 28. Perron MJ, Stremlau M, Sodroski J (2006) Two surface-exposed elements of the B30.2/ genes in primate genomes. PLoS Genet 7:e1002388. SPRY domain as potency determinants of N-tropic murine leukemia virus restriction by 6. Owens CM, Yang PC, Gottlinger H, Sodroski J (2003) Human and simian immunode- human TRIM5alpha. J Virol 80:5631–5636. ficiency virus capsid proteins are major viral determinants of early, postentry replica- 29. Notkins AL (2004) Polyreactivity of antibody molecules. Trends Immunol 25:174–179. tion blocks in simian cells. J Virol 77:726–731. 30. Wedemayer GJ, Patten PA, Wang LH, Schultz PG, Stevens RC (1997) Structural insights 7. Cowan S, et al. (2002) Cellular inhibitors with Fv1-like activity restrict human and into the evolution of an antibody combining site. Science 276:1665–1669. simian immunodeficiency virus tropism. Proc Natl Acad Sci USA 99:11914–11919. 31. Yin J, Beuscher AEt, Andryski SE, Stevens RC, Schultz PG (2003) Structural plasticity and 8. Sebastian S, Luban J (2005) TRIM5alpha selectively binds a restriction-sensitive retro- the evolution of antibody affinity and specificity. J Mol Biol 330:651–656. viral capsid. Retrovirology 2:40. 32. Manivel V, Sahoo NC, Salunke DM, Rao KV (2000) Maturation of an antibody response 9. Stremlau M, et al. (2006) Specific recognition and accelerated uncoating of retroviral capsids by the TRIM5alpha restriction factor. Proc Natl Acad Sci USA 103:5514–5519. is governed by modulations in flexibility of the antigen-combining site. Immunity – 10. Ohkura S, Yap MW, Sheldon T, Stoye JP (2006) All three variable regions of the TRI- 13:611 620. M5alpha B30.2 domain can contribute to the specificity of retrovirus restriction. J Virol 33. Mallery DL, et al. (2010) Antibodies mediate intracellular immunity through tripartite – 80:8554–8565. motif-containing 21 (TRIM21). Proc Natl Acad Sci USA 107:19985 19990. 11. Song BW, et al. (2005) The B30.2(SPRY) domain of the retroviral restriction factor TRI- 34. Perron MJ, et al. (2004) TRIM5alpha mediates the postentry block to N-tropic murine M5alpha exhibits line age-specific length and sequence variation in primates. leukemia viruses in human cells. Proc Natl Acad Sci USA 101:11827–11832. J Virol 79:6111–6121. 35. Diaz-Griffero F, et al. (2008) A human TRIM5alpha B30.2/SPRY domain mutant gains 12. Stremlau M, Perron M, Welikala S, Sodroski J (2005) Species-specific variation in the the ability to restrict and prematurely uncoat B-tropic murine leukemia virus. Virology B30.2(SPRY) domain of TRIM5alpha determines the potency of human immunodefi- 378:233–242. ciency virus restriction. J Virol 79:3139–3145. 36. Maillard PV, Reynard S, Serhan F, Turelli P, Trono D (2007) Interfering residues narrow 13. Sawyer SL, Wu LI, Emerman M, Malik HS (2005) Positive selection of primate TRIM5al- the spectrum of MLV restriction by human TRIM5alpha. PLoS Pathog 3:e200. pha identifies a critical species-specific retroviral restriction domain. Proc Natl Acad Sci 37. Diaz-Griffero F, et al. (2007) Comparative requirements for the restriction of retrovirus USA 102:2832–2837. infection by TRIM5alpha and TRIMCyp. Virology 369:400–410. 14. Kaiser SM, Malik HS, Emerman M (2007) Restriction of an extinct retrovirus by the 38. Ohkura S, et al. (2011) Novel escape mutants suggest an extensive TRIM5alpha binding human TRIM5alpha antiviral protein. Science 316:1756–1758. site spanning the entire outer surface of the murine leukemia virus capsid protein. 15. Ganser-Pornillos BK, et al. (2011) Hexagonal assembly of a restricting TRIM5alpha pro- PLoS Pathog 7:e1002011. – tein. Proc Natl Acad Sci USA 108:534 539. 39. McCoy AJ, et al. (2007) Phaser crystallographic software. J Appl Crystallogr 40:658–674. 16. Diaz-Griffero F (2011) Caging the beast: TRIM5alpha binding to the HIV-1 core. Viruses 40. Weinert C, Grutter C, Roschitzki-Voser H, Mittl PR, Grutter MG (2009) The crystal struc- – 3:423 428. ture of human pyrin b30.2 domain: Implications for mutations associated with familial 17. Diaz-Griffero F, et al. (2007) Modulation of retroviral restriction and proteasome Mediterranean fever. J Mol Biol 394:226–236. inhibitor-resistant turnover by changes in the TRIM5alpha B-box 2 domain. J Virol 41. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macromo- 81:10362–10378. lecular structure solution. Acta Cryst 66:213–221. 18. Li X, Sodroski J (2008) The TRIM5alpha B-box 2 domain promotes cooperative binding 42. Emsley P, Cowtan K (2004) Coot: Model-building tools for molecular graphics. Acta to the retroviral capsid by mediating higher-order self-association. J Virol Crystallogr D Biol Crystallogr 60:2126–2132. 82:11495–11502. 43. Gronenborn AM (2009) Protein acrobatics in pairs—dimerization via domain swap- 19. Diaz-Griffero F, et al. (2009) A B-box 2 surface patch important for TRIM5alpha self- ping. Curr Opin Struct Biol 19:39–49. association, capsid binding avidity, and retrovirus restriction. J Virol 83:10737–10751. 44. Remaut H, Waksman G (2006) Protein-protein interaction through beta-strand addi- 20. Pornillos O, Ganser-Pornillos BK, Yeager M (2011) Atomic-level modelling of the HIV – capsid. Nature 469:424–427. tion. Trends Biochem Sci 31:436 444. 21. Zhou P, Lugovskoy AA, Wagner G (2001) A solubility-enhancement tag (SET) for NMR 45. Palmer AG, III (2001) NMR probes of molecular dynamics: Overview and comparison – studies of poorly behaving proteins. J Biomol NMR 20:11–14. with other techniques. Annu Rev Biophys Biomol Struct 30:129 155. 22. Pham QT, Bouchard A, Grutter MG, Berthoux L (2010) Generation of human TRIM5al- 46. Lienlaf M, et al. (2011) Contribution of E3-ubiquitin ligase activity to HIV-1 restriction pha mutants with high HIV-1 restriction activity. Gene Ther 17:859–871. by TRIM5alpha(rh): Structure of the RING domain of TRIM5alpha. J Virol 23. Yap MW, Nisole S, Stoye JP (2005) A single amino acid change in the SPRY domain of 85:8725–8737. human TRIM5alpha leads to HIV-1 restriction. Curr Biol 15:73–78. 47. Ganser-Pornillos BK, Cheng A, Yeager M (2007) Structure of full-length HIV-1 CA: A 24. James LC, Keeble AH, Khan Z, Rhodes DA, Trowsdale J (2007) Structural basis for model for the mature capsid lattice. Cell 131:70–79. PRYSPRY-mediated tripartite motif (TRIM) protein function. Proc Natl Acad Sci USA 48. Pornillos O, et al. (2009) X-ray structures of the hexameric building block of the HIV 104:6200–6205. capsid. Cell 137:1282–1292.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1203536109 Biris et al. Downloaded by guest on October 2, 2021