Structural and functional insights into alphaviruspolyprotein processing and pathogenesisGyehwa Shina,1,2, Samantha A. Yosta,1, Matthew T. Millera, Elizabeth J. Elrodb, Arash Grakouib,and Joseph Marcotrigianoa,3

aCenter for Advanced Biotechnology and Medicine, Department of Chemistry and Chemical Biology, Rutgers University, Piscataway, NJ 08854; and bEmoryVaccine Center, Division of Microbiology and Immunology, Department of Medicine, Division of Infectious Diseases, Emory University School of Medicine,Atlanta, GA 30322

Edited by Robert A. Lamb, Northwestern University, Evanston, IL, and approved August 31, 2012 (received for review June 19, 2012)

Alphaviruses, a group of positive-sense RNA viruses, are globallydistributed arboviruses capable of causing rash, arthritis, encephali-tis, and death in humans. The viral replication machinery consists offour nonstructural proteins (nsP1–4) produced as a single polypro-tein. Processing of the polyprotein occurs in a highly regulatedmanner, with cleavage at the P2/3 junction influencing RNA tem-plate use during genome replication. Here, we report the structureof P23 in a precleavage form. The proteins form an extensive in-terface and nsP3 creates a ring structure that encircles nsP2. TheP2/3 cleavage site is located at the base of a narrow cleft and is notreadily accessible, suggesting a highly regulated cleavage. ThensP2 protease active site is over 40 Å away from the P2/3 cleavagesite, supporting a trans cleavage mechanism. nsP3 contains a pre-viously uncharacterized protein fold with a zinc-coordination site.Known mutations in nsP2 that result in formation of noncyto-pathic viruses or a temperature sensitive phenotype cluster atthe nsP2/nsP3 interface. Structure-based mutations in nsP3 oppo-site the location of the nsP2 noncytopathic mutations prevent ef-ficient cleavage of P23, affect RNA infectivity, and alter viral RNAproduction levels, highlighting the importance of the nsP2/nsP3interaction in pathogenesis. A potential RNA-binding surface,spanning both nsP2 and nsP3, is proposed based on the locationof ion-binding sites and adaptive mutations. These results offerunexpected insights into viral protein processing and pathogenesisthat may be applicable to other polyprotein-encoding viruses such asHIV, hepatitis C virus (HCV), and Dengue virus.

Alphaviruses are small, enveloped RNA viruses that infectvarious vertebrates, including humans, horses, birds, and

rodents (1). Transmission occurs via an invertebrate vector,usually mosquitoes. In humans, typical alphavirus infection canresult in rash, arthritis, encephalitis, and death. Within the pastdecade, infection by Chikungunya virus (CHIKV), a member ofthe genus alphavirus, caused massive epidemics in Africa andSoutheast Asia, with positive cases reported in Europe and theUnited States, raising the possibility of an emerging threat (2).Additionally, CHIKV often presents with nonclassic symptomsthat go undiagnosed, making global pathogenicity difficult tomonitor (3).The alphavirus genome consists of single-stranded, positive-sense

RNA of ∼9–11 kb with a 5′ cap structure and 3′ polyadenosine tail.The genome contains two cistrons. The first, located in the 5′ two-thirds of the genome, encodes the viral replication machinery,termed nonstructural proteins (nsPs), whereas the structuralproteins that form virus particles are encoded in the secondcistron. The genomic RNA is used as an mRNA for the trans-lation of the nsPs and as a template for the synthesis of com-plementary negative-sense RNA. The negative-sense strand is atemplate for subgenomic RNA containing the second cistronand progeny, genomic RNA. Because genomic and subgenomicRNAs are made in vast excess, the negative-sense RNA strand isconsidered a replication intermediate.The alphavirus replication machinery is composed of four

nonstructural proteins (nsP1 to -4), which are expressed as one

of two polyproteins (P123 or P1234). P1234 is expressed as aread-through of an opal termination codon at the end of nsP3.These precursor polyproteins are cleaved by a protease withinnsP2 (1, 4). After translation of P1234, cleavage at the P3/4junction occurs either in cis or trans, followed by the P1/2 junc-tion, which occurs in cis only (4, 5). Both P123+nsP4 and nsP1+P23+nsP4 preferentially synthesize negative strand viral RNA(6, 7). The final cleavage event between P23 produces fully maturensPs and switches RNA template for synthesis of positive-sensegenomic and subgenomic RNAs. The correlation between P23cleavage and the switch from negative- to positive-sense RNAproduction is poorly understood.In mammalian cells, alphavirus infection inhibits host gene

expression, causes severe cytopathic effects (CPEs), disrupts thecellular response to IFN, and leads to cell death (8–11). MaturensP2 contains an amino-terminal RNA helicase domain, a centralprotease domain that catalyzes all cleavages between the non-structural proteins, and an inactive RNA methyltransferase-like(MT-like) domain (12). During infection, a portion of nsP2localizes to the nucleus (13, 14) and plays a role in shutting offhost-cell transcription and viral cytopathogenicity (9, 10). Twogenetically distinct groups of alphaviruses termed New Worldand Old World have emerged because geographic isolation (8).In Old World alphaviruses, such as Sindbis virus (SINV) andSemliki Forest virus (SFV), transcriptional shutoff is nsP2-de-pendent (15). Even in the context of a replicon lacking thestructural proteins, the nsPs of SINV and SFV alone can effi-ciently shut off host transcription and induce severe CPE (16).Mutations that cause alphaviruses to establish a persistent in-

fection with minimal or no CPE have been identified (16–20). Akey feature of these mutants is their inability to inhibit host-celltranscription (10, 15). The best-characterized noncytopathicmutations map to nsP2 proline 726 (P726) within the MT-likedomain (10, 16, 17, 21, 22). The P726 mutation has been shownto attenuate the cytotoxic effect caused by the expression of wild-type SINV nsP2 protein alone (10). Noncytopathic mutations ofnsP2 P726 also reduce viral RNA replication levels and are unableto shut off host transcription and translation processes, showing

Author contributions: G.S., S.A.Y., and J.M. designed research; G.S., S.A.Y., and J.M. per-formed research; G.S., S.A.Y., E.J.E., A.G., and J.M. contributed new reagents/analytictools; G.S., S.A.Y., M.T.M., and J.M. analyzed data; and G.S., S.A.Y., M.T.M., and J.M. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank, www.pdb.org (PDB ID code 4GUA).1G.S. and S.A.Y. contributed equally to this work.2Present address: Science and Engineering Campus, Korea University, Seoul, Republic ofKorea, 136-701.

3To whom correspondence should be addressed. E-mail: jmarco@cabm.rutgers.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1210418109/-/DCSupplemental.

16534–16539 | PNAS | October 9, 2012 | vol. 109 | no. 41 www.pnas.org/cgi/doi/10.1073/pnas.1210418109

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

021

the overall importance of nsP2 to viral replication, as well ashost-cell interaction (11, 16).The function(s) of nsP3 in alphavirus replication remains un-

resolved. Based on sequence conservation, nsP3 is organized intothree domains: an amino-terminal macro or X domain (23), acentral alphavirus-specific region, and a hypervariable-sequencecarboxyl terminus. Macro domains have been shown to bind ADPribose (24) with certain viral macro domains also retaining ADPribose-1′′ phosphatase activity (25, 26). Recent work in SFVshows that residues just after the macro domain of nsP3 play arole in positioning of the P23 cleavage site (27). Additionally, thecarboxyl-terminal region of nsP3 contains numerous phosphor-ylation sites targeted by yet unknown cellular kinases (28, 29).Here, we present the structure of an uncleaved P23 precursor

protein spanning the nsP2 protease domain through to thecentral, zinc-binding domain (ZBD) of nsP3 from SINV (Fig.1A). The P23 cleavage site is located in a narrow cleft formedbetween nsP2 and nsP3 that is inaccessible for proteolysis.Surprisingly, all of the previously reported nsP2 noncytopathicmutants lie at the interface between nsP2 and nsP3. Muta-genesis of residues in nsP3 opposite of nsP2 P726 showed in-efficient cleavage of P23, delayed onset of pathogenesis and hostgene shutoff, and altered viral RNA synthesis. Because P23pro-zbd

copurified with large amounts of RNA, we propose a potentialRNA-binding surface that extends across both nsP2 and nsP3and hypothesize that cleaving the P2/3 junction may alter theRNA-binding surface, thereby contributing to template switchingby the replication machinery. Our results provide a better un-derstanding of the mechanism of viral polyprotein processingand pathogenesis.

ResultsAlphavirus P23pro-zbd Structure. The SINV P23 precursor proteinextending from the nsP2 protease domain to the nsP3 centraldomain was shown to be fully intact by SDS/PAGE and massspectrometry despite the presence of a competent nsP2 proteaseactive site (Fig. S1 A and B). Sequence alignment of the centraldomain of nsP3 shows numerous absolutely conserved cysteineresidues (Fig. S2), suggesting the presence of a metal ion–bindingsite. Various purified preparations of the middle region of nsP3were analyzed by quantitative X-ray fluorescence for commontransition metals found in metalloproteins (30, 31) and shown tohave one zinc ion per molecule. Crystals of P23pro-zbd diffractedto 2.85-Å resolution and the structure was determined by mo-lecular replacement with structures of nsP2pro (12) and nsP3macro domain (32) (Table S1). The resulting electron densityallowed for an unambiguous trace of the entire polypeptidechain, including the previously uncharacterized nsP3 central re-gion containing the zinc-binding site. The location of the zinc ionwas confirmed by anomalous difference maps (Table S1 andsee Fig. S4B).SINV P23pro-zbd shows a highly compact structure, with the

four domains (protease and MT-like from nsP2; macro and ZBDfrom nsP3) arranged with each domain occupying a vertex of arectangle (Fig. 1 B–D). The polypeptide chain progresses aroundthe perimeter of the rectangle such that the nsP2 protease andnsP3 ZBD are adjacent to one another. The carboxyl terminus ofthe P23pro-zbd construct is pointing toward the protease active site(Fig. 1 B–D). A previously described SINV mutant (ts7) defectivein RNA synthesis at nonpermissive temperatures contains a F312to serine mutation in nsP3 (33). This residue is located in the hy-drophobic core and surrounded by residues from the macro, ZBD,and MT-like domains (Fig. 1E). Mutation of this residue wouldlikely destabilize the core and disrupt domain interaction, givingrise to the RNA− phenotype at the nonpermissive temperature.In a previously described nsP2pro structure, the six carboxyl-

terminal amino acids were disordered (12). In the P23pro-zbd

structure presented here, the entire polypeptide between nsP2

and nsP3 is ordered, permitting modeling of the P2/3 cleavagesite. The P2/3 cleavage site is located at the base of a narrowcleft about 11–13 Å wide formed by the MT-like domain of nsP2and the macro domain of nsP3 (Fig. 1 F and G). The cleavage siteis 40 Å away from the nsP2 protease active site, supporting atrans cleavage mechanism (5, 34, 35). Although the 6 aa sur-rounding the scissile bond are solvent-exposed, attempts to positionthe protease domain onto the P2/3 cleavage site using the pro-posed model (36) resulted in numerous steric clashes.

Fig. 1. Overview of the alphavirus replication machinery and P23pro-zbd

structure. (A) Schematic representation of the alphavirus replication ma-chinery with emphasis on domain organization of nsP2 and nsP3. Thecrystallization construct consists of four domains: protease (blue) andmethyltransferase-like (teal) domains of nsP2 and the macro (yellow) andzinc-binding (red) domain of nsP3, encompassing amino acids 1011–1675 ofthe P1234 polyprotein. Amino acid numbering provided from the aminoterminus of nsP2 and nsP3 are also noted in parentheses. (B–D) Ribbon dia-gram of P23pro-zbd colored according to A. A gray sphere denotes the positionof the zinc ion. The P2/3 cleavage site, nsP2 protease active site, and ADPribose–binding site are labeled with an arrow, asterisk (*), and filled circle,respectively. (C) Ribbon diagram of P23 rotated 90° about a vertical axisfrom the view in B. (D) Ribbon diagram of P23 rotated 180° about a verticalaxis from the view in B. Box shows location of the view in E. (E) Previouslydescribed nsP3 temperature-sensitive mutant ts7 (F312S) highlighted ingreen with the immediately surrounding amino acids located in the macro,ZBD, and MT-like domains. nsP2 residues are shown in italics. (F and G) TheP2/3 cleavage site is located at the base of a narrow cleft formed by the MT-like and macro domains. Three amino acids on either side of the scissile bond(arrow) are shown in stick representation. The view in F is similar to that in B,whereas that in G is rotated 90° about a horizontal axis from F, lookingdirectly into the cleft.

Shin et al. PNAS | October 9, 2012 | vol. 109 | no. 41 | 16535

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

021

ZBD in nsP3 Is Essential for Viral Replication.After the P2/3 cleavagesite, the polypeptide chain continues into the nsP3 macro do-main. The ADP ribose–binding site within the macro domain issolvent-exposed and points away from the other four domains.The polyprotein exits the last helix of the macro domain adjacentto the P2/3 cleavage site and continues into a stretch of ∼40 aathat lacks secondary structure but forms a twist (Fig. S3). Theconserved nsP3 central domain contains an antiparallel α-helicalbundle, two parallel β-strands, and a previously unknown zinc-coordination site (Fig. S4B). The zinc ion is coordinated byC263, C265, C288, and C306 of nsP3. C263 and C265 are locatedin the loop between the last two α-helices, whereas C288 andC306 are at the carboxyl termini of the two parallel β-strands(Fig. S3C). Each of the zinc coordinating cysteines is invariantamong all alphavirus sequences (Fig. S2). Structural comparisonof the nsP3 ZBD against all known folds in the Protein DataBank using PDBe Fold (37) and Dali servers (38) failed to pro-duce any statistically significant hits, and no metalloprotein foldswere identified. The spatial arrangement of secondary structuralelements that contribute to nsP3 zinc-binding site, i.e., two cys-teines in a loop between alpha helices and the other two cysteinesat the end of two parallel β-strands, represents a structural scaffoldfor zinc-ion coordination (39).Site-directed mutagenesis was performed to examine the

functionality of the four conserved cysteine residues involved inzinc coordination in nsP3 (Fig. S4A). All four SINV mutants(C263A, C265A, C288A, and C306A) failed to produce pro-ductive infection in BHK cells, as evaluated by infectious centerand plaque assays (Fig. S4C). A conserved cysteine residue innsP3 (C247) unrelated to the zinc-binding site served as a controland C247A mutant produced infectious particles to near wild-typelevels. Zinc-coordination mutants failed to express the non-structural polyprotein and replicate virus, thereby supporting thestructural role of the zinc ion (Fig. S4D).The coordination site forms a distorted tetragon due to the

presence of two highly conserved serine residues (S289 and S290)below the zinc atom (Fig. S2). Mutation of these serine residuesdid not alter viral replication levels. Furthermore, a doubleserine to alanine mutant (S289A+S290A) resulted in decreasedviral RNA infectivity, but overall viral titers were comparable towild type (Fig. S4C).

Importance of P23 Interface for Polyprotein Processing and ViralPathogenesis. The interface between the macro and ZBD ofnsP3 has a buried surface area of 570 Å2, whereas the linkerconnecting the two domains is extended, making nsP3 into aring-like structure with an inner diameter of 15–18 Å (Fig. 2A).Together, nsP2 and nsP3 share an extensive interface with 3,000Å2 of buried surface area. nsP3 encircles the MT-like domain ofnsP2, with residue R781 of nsP2 protruding through the opening(Fig. 2C). The interface between nsP2 and nsP3 is charged, withthe nsP2 surface being mostly basic, whereas nsP3 is generallyacidic (Fig. 2 B and C).Several groups have identified mutations in the carboxyl-ter-

minal portion of nsP2 that result in noncytopathic virus andpersistent infection (16–20). These mutations reduce viral RNAreplication levels with minimal effects on host gene expressionand viral yields. All of the noncytopathic mutations map to thesurface of nsP2 at the nsP3 interface (Fig. 2D). The mutationsfollow the encircling path made by the macro to ZBD linker.Mutagenesis of P726 in SINV to amino acids with large sidechains (Phe, Tyr, Leu, Arg, and Gln) caused a dramatic decreasein RNA replication and were noncytopathic, whereas substitutionswith small side chains (Ala or Val) behaved like wild-type virus,with Gly, Ser, and Thr substitutions giving an intermediate effect(16). P726 is located in the loop connecting α9 with β11 andmakes direct contact with nsP3 linker just after the macro do-main (Fig. 2E). There is a striking, direct correlation between

cytopathogenicity and the level of RNA replication with the sizeof the mutant side chain. Molecular modeling analyses demon-strated that mutating P726 to Phe, Tyr, Leu, Arg, or Gln resultedin numerous steric clashes with nsP3, whereas Ser, Thr, Ala, andVal had fewer conflicts. Moreover, the P726G mutation coulddestabilize the α9-β11 loop, causing it to interfere with P23interaction. Attempts to purify P23pro-zbd with a P726S orP726L mutation resulted in poorly soluble and aggregatedprotein, supporting the importance of this position.To investigate how destabilizing P23 interface may contribute to

changes in viral pathogenesis and replication, site-directed muta-genesis to the corresponding surface of SINV nsP3 in close prox-imity to nsP2 P726 were created (Fig. 2E). The side chains of V162and L165 of nsP3 are ∼4.6 and 3.3 Å away from nsP2 P726 sidechain, respectively (Fig. 2F). V162 and L165 of nsP3 were mutatedto one of the following: Ala, Asp, Glu, Phe, Arg, or Trp. nsP2 P726to Ser, Gly, or Leu mutants were used for comparison. RNA in-fectivity measured by infectious center assay on BHK cells werereduced in P726L, L165D, L165E, and L165R mutants comparedwith the wild-type SINV (Fig. 3A). P726L mutation in nsp2 re-duced RNA infectivity to only 2% of wild-type levels, whereas thethree nsP3 L165 mutants averaged an RNA infectivity at 8% ofwild type. These four mutants also displayed a small size plaquephenotype in infectious center assay (Fig. S5 A–D). In addition,

Fig. 2. nsP2 and nsP3 interface and the location of nsP2 noncytopathicmutants. (A) Solvent-accessible surface of nsP3. (B and C) Surface of nsP3 (B)or nsP2 (C) colored for electrostatic potential at ±5 kT/e; blue (basic), red(acidic), and white (neutral) with ribbon diagram of nsP2 (B) or nsP3 (C). (D)Molecular surface of nsP2 highlighting (white) the location of nsP2 non-cytopathic mutations. The numbering corresponds to SINV nsP2 sequence.(E) Location of P726 in nsP2, highlighting interactions with the nsP3 linker.A portion of the last helix in the nsP3 macro domain is shown in yellow.(F) Surface of nsP2 P726 and surrounding area colored for electrostatic po-tential at ±4 kT/e with nsP3 linker region in stick format. Residues labeled initalics are located in nsP2. Arrow indicates P2/3 cleavage site. Domain col-oring in each panel is identical to Fig. 1.

16536 | www.pnas.org/cgi/doi/10.1073/pnas.1210418109 Shin et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

021

L165A, L165D, L165E, and L165R mutants produced lower viraltiters at 24 h after electroporation, only reaching an average of34% of wild-type levels (Fig. 3B). Of the nsP3 mutants generated,L165D had the lowest titers, only producing 26% compared withwild-type titers. All others, including all three P726 mutants,

reached wild-type levels of virus titers. All V162 mutants behavedsimilarly to wild type in both RNA infectivity and viral titer andwere not pursued further. None of the nsP3 mutants displayeda noncytopathic phenotype in the TSG/puromycin N-acetyl-tranferase (PAC) replicon background (16), although onset ofCPE in L165 to D, E, and R mutants was delayed.Radiolabeling experiments were performed to evaluate both

viral and host protein and RNA production over the course ofinfection and to analyze efficiency of host-cell transcriptional andtranslational shutoff caused by SINV nsP3 mutations. Wild-type SINV is able to halt host translation completely within 6 hpostinfection, whereas P726S and P726G mutants were unableto do so even up to 12 h postinfection, as indicated by persistentactin production (Fig. S6). nsP2 P726L and the four selectednsP3 L165 mutants (Asp, Glu, Arg, Trp) were able to shut offhost translation, although at a slightly delayed time postinfectioncompared with wild type (Fig. S6).Tritium-labeled uridine was used to track viral and host RNA

production over the course of infection. In the presence of acti-nomycin D, both genomic and subgenomic viral RNA productioncan be clearly observed in wild-type SINV starting as early as 2 hpostinfection and remaining stable until 12 h (Fig. 3C). Levels ofsubgenomic and genomic RNA in P726G and P726S mutants arereduced compared with wild type. nsP2 P726L and the four nsP3L165 mutants exhibit a slow initial rate of transcription, espe-cially of genomic RNA; however, these mutants reach wild-typelevels by 4 h postinfection. Thereafter, the levels of RNA repli-cation in the L165 mutants continue to increase, and even sur-pass, wild-type RNA replication, especially subgenomic RNA(Fig. 3C). RNA labeling in the absence of actinomycin D showedcomplete inhibition of host cellular transcription by 8 h post-infection as demonstrated by decreasing 28S and 18S cellularrRNA (Fig. 3D). P726G and P726S mutants exhibited a lack ofhost-cell transcriptional shutoff with rRNA persisting up to 12 hpostinfection. The nsP2 P726L mutant and the four nsP3 L165mutants showed a delay in host transcription inhibition buteventually achieved complete shutoff 12 h postinfection (Fig. 3D).Given that residuesR159 andE163 of SFVnsP3havebeen shown

to affect P2/3 cleavage (27), we assayed the effect of the SINV nsP3mutations on cleavage activity. nsP3 linker region mutants L165Dand L165E revealed drastic inefficiency in P23 cleavage that was notseen in L165WandL165Rmutants (Fig. 3E andF andFig. S7). Thepresence of uncleaved P23 in the nsP3 L165D and L165E mutantsstarted 6 h postinfection and persisted up to 12 h, whereas P726mutants, as well as nsP3 L165R and nsP3 L165W, showed sub-stantially lower amounts of uncleaved P23. Western blots probedwith anti-nsP3 antibodies were used to confirm the presence ofuncleaved P23 (Fig. 3F). nsP3 appears as a smear, indicativeof alternate phosphorylation states and possibly other post-translational modifications (40). The nsP2 P726 mutants may alsoaccumulate precleavage forms but at very low concentrations thatdo not build up over time, as in the case of the nsP3 linkermutants.

DiscussionThe structure of alphavirus precursor polyprotein P23pro-zbd pre-sented here reveals the arrangement of four domains (protease,MT-like, macro, and ZBD) in a precleavage form. Based on thestructure of P23pro-zbd, several inferences can be made about themechanism of alphavirus polyprotein cleavage. First, the car-boxyl terminus of the P23pro-zbd structure is pointing toward thensP2 protease active site about 40–45 Å away. It is possible thatthe large, flexible carboxyl region of nsP3 could span the dis-tance, placing the P3/4 junction in the protease active site for ciscleavage (34). Second, the 40 Å distance between the P2/3cleavage site and the nsP2 protease active site supports the currenttrans model of cleavage (5, 34, 35). Third, the inaccessibility ofthe P23 cleavage site itself indicates that access is tightly regulated.It is possible the activator sequence present in the extreme

Fig. 3. In vivo effects of nsP3 linker region mutations. (A and B) Viral RNAinfectivity [infectious units (IU)/μg RNA] as measured by infectious centerassay (A) and viral titers [plaque-forming units (pfu)/mL] (B) 24 h afterelectroporation of BHK cells, as determined by plaque assay over two ormore independent experiments. Error bars represent the SEM. (C and D) Cellswere infected with wild-type or mutant SINV at a multiplicity of infection(MOI) of 10. At indicated time points postinfection, RNAs were labeled for3 h in the presence (C) and absence (D) of actinomycin D. Total RNAs wereharvested and analyzed by denaturing agarose gel electrophoresis. Theposition of genomic (G), subgenomic (SG), 28S, and 18S RNAs are labeled. Mindicates mock-infected cells labeled for 3 h. (E and F) Analysis of P23cleavage of nsP2 P726 and nsP3 L165 mutants in vivo. Total protein washarvested from wild-type and indicated mutant SINV-infected BHK cells atindicated hours postinfection and analyzed by Western blot using anti-nsP2(E) and anti-nsP3 (F) antibodies.

Shin et al. PNAS | October 9, 2012 | vol. 109 | no. 41 | 16537

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

021

amino terminus of nsP2, which becomes exposed after cleavageat the P1/2 site, may be responsible for this regulation (5).During the course of alphavirus infection, a portion of nsP2 is

localized to the nucleus, whereas nsP3 remains cytoplasmic (13, 14,41). At the moment, it is unclear how nsP2 would dissociate post-cleavage from the extensive, encircling grasp of nsP3. Comparisonof the domains in P23pro-zbd with the isolated structures of nsP2pro

(12) and nsP3 macro domain (32) demonstrated that the proteaseand macro domains were highly similar with a root mean squaredeviation (rmsd) of <0.6 Å, whereas the MT-like domains have anrmsd of 1.3 Å for similar α carbon. A superposition of the nsP2structures shows the discrepancy between nsP2pro and P23pro-zbd islimited to specific regions within theMT-like domain (Fig. 4A andB). α-Helices α8, α10, and α11, as well as β-strands β11, β12, andβ13, between the two structures superimpose well with an rmsd of<1 Å, although the β13-α10 loop moves 5 Å because of inter-actions with the nsP3 linker. In addition, the nsP3 linker interactswith α9, causing the helix to rotate 8° with the carboxyl-terminalend of the helix moving 3.8 Å to form a single continuous helixwith α8. This movement of the α9 helix causes the β-sheet to be-come more concave. The distortion of the MT-like domain in theP23pro-zbd becomes more pronounced the further the secondarystructures are from α9, with α7 and β9 moving about 4 Å fromtheir corresponding positions in nsP2pro. Perhapsmovement of the

MT-like domain facilitates separation of nsP2 and nsP3 post-cleavage, which is triggered by binding viral proteins or RNA to theconcave side of the nsP2 MT-like β-sheet.Previously described noncytopathic mutations at the surface

of nsP2 contacting the nsP3 linker region appear by molecularmodeling to destabilize P23 interaction, decreasing efficiency of theRNA replication complex during virus infection. Altering the cor-responding nsP3 residue (L165), which contacts nsP2 P726, did notproduce virus with the same noncytopathic phenotype. In contrastto the nsP2 P726G and P726S mutants, nsP3 L165 mutants hadenhanced RNA replication levels, indicating the apparent stabili-zation of the replication complex. Despite the large, bulky nature ofa tryptophan substitution at L165, the virus remained similar inphenotype to wild type. The L165D and L165E mutants, however,exhibited the most extreme phenotype in terms of reduced RNAinfectivity and inefficient P23 cleavage. The surface of nsP2 sur-rounding P726 is both basic and hydrophobic (Fig. 2F), suggestingthat the P23 interface may be stabilized by the introduced acidicamino acid at L165 through interaction with the surface of nsP2. Thisstabilization could cause the interface to become “locked in,”preventing the flexibility needed for conformational change and,thus, disallowing nsP2 access to the P2/3 cleavage site.The buildup of uncleaved P23 is evidence that the nsP3 linker

region, which encircles nsP2, may function in positioning andrecognition of the P2/3 cleavage site. Given the nsP3 linker resi-dues and nsP2 P726 are within reasonable distance to the P2/3cleavage site (21 and 17 Å from Cα of L165 or P726 to thecleavage site, respectively), it is plausible that mutation of thensP3 linker region shifts local structural features, altering rec-ognition of the cleavage site (Fig. 2F). A previously describedE163R mutation in SFV nsP3 that altered P2/3 cleavage is com-pletely solvent-exposed (Fig. 2E) (27). These results may indicatethat the nsP3 linker residues surrounding nsP2 P726 are recognizedby the nsP2 protease or another factor necessary for P2/3 cleavage.Cleavage between P2/3 dictates template use by the alphavirus

replication machinery. During purification, P23pro-zbd bound ex-tensive amounts of bacterial RNA, which could only be removedby high concentrations of salt (Fig. S1 A and C). The electrostaticpotential of the P23pro-zbd structure identified a large basic surfacethat centers on the zinc-binding site of nsP3 and extends to nsP2(Fig. 4C). Many zinc metalloproteins are involved in nucleicacid–binding and gene regulation (42); therefore, the locationof a basic patch in close proximity to the zinc-binding site suggestsa potential role in RNA binding. Several molecules of sulfateand 2-(N-morpholino)ethane sulfonic acid (MES), used incrystallization, were bound to P23pro-zbd along the basic surface,which may indicate potential binding sites for the phosphategroups of nucleic acids (Fig. 4 C–E). In fact, a molecule of MESoccupies the ADP ribose–binding site in each of the three macrodomains in P23pro-zbd in the asymmetric unit, as well as in themacroH2A1.1 structure (Fig. 4E) (43). The alphavirus genomecontains conserved sequence elements (CSEs) that serve as pro-moters for production of positive- and negative-sense genomicRNA, as well as subgenomic RNA. Mutations in a 51-nt-longCSE in the nsP1-coding region of SINV disrupted the RNAstructure without affecting the coding sequence and demon-strated reduced replication in mosquito cells but not in mam-malian cells (44). Adaptive mutations (E118K of nsP2 and T286Iof nsP3) were shown to restore replication of the mutated CSE.T286 is 2 aa before one of the zinc-coordinating cysteine residues(C288) (Fig. 4F) and is located on the basic surface of the ZBD.T286 is 6.1 Å away from a sulfate ion that is coordinated by threehighly conserved amino acids in nsP3: Y267, R273, and R276(Fig. 4F). Taken together, these data strongly implicate this sur-face as a potential RNA-binding site.Many important human viral pathogens encode a polyprotein

that is cleaved by viral or cellular proteases. Given the wideprevalence of regulated polyprotein processing seen in a number

Fig. 4. (A and B) Superposition of the MT-like domains of P23pro-zbd (teal)and nsP2pro (gray) (12). The nsP3 linker (for simplicity only residues V162-W178 are shown) connecting the macro domain and ZBD is shown in red. Theview in B is rotated 90° about a horizontal axis from A. (C and D) PotentialRNA-binding surface of P23pro-zbd. The location of sulfate andMES ligands arerepresented by spheres. Surface of P23pro-zbd is colored for electrostatic po-tential at ±5 kT/e. D is a 180° rotation about the vertical axis from the view inpanel C. (E) Ribbon diagram of P23pro-zbd in the identical orientation as in C.(F) ZBD of nsP3 showing the four zinc-coordinating cysteines, T286, a sulfateion represented by spheres bound to Y267, R273, and R276, and surroundingresidues. The close-up view corresponds to the box in E.

16538 | www.pnas.org/cgi/doi/10.1073/pnas.1210418109 Shin et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

021

of viral families, there is lack of structural information regardingthe precursor forms. Currently, alphavirus P23pro-zbd and polio-virus 3CD (45) are the only structures of polyprotein precursorfrom a viral replication machinery. Both alphavirus P23pro-zbd

and poliovirus 3CD structures contain the protease domain re-sponsible for cleaving the precursor, which is thought to proceed bya trans mechanism. However, 3CD has no intramolecular con-tacts and a solvent accessible cleavage site that is regulated byintermolecular contacts, whereas P23pro-zbd has an extensive in-terface and an inaccessible cleavage site. The results presentedhere further expand our current understanding of viral poly-protein processing, replication, and cytopathogenesis, which maybe applicable to other positive-sense RNA viruses.

Materials and MethodsComplete materials and methods are provided in SI Materials and Methods.

Structure Determination. P23pro-zbd of the SINV strain Toto1101 was expressedin Escherichia coli as a fusion with GST. The protein was purified by sequentialchromatography over GST affinity, hydroxyapatite, and heparin columns.Crystals of P23pro-zbd were grown by vapor diffusion against 2 M ammoniumsulfate and 0.1 M MES (pH 6.5). Crystals were transferred to a cryoprotectantsolution containing 2 M ammonium sulfate, 0.1 M MES (pH 6.5), 35% (vol/vol)

xylitol, and 5% (vol/vol) trimethylamine N-oxide and then flash-cooled in liquidnitrogen.Diffractiondatawere collectedat the zinc, Kabsorptionedge (1.2824Å)to 2.85 Å resolution at beam line X25 at Brookhaven National Synchrotron LightSource (Table S1). Phases were determined by molecular replacement using thensP2pro (12) and nsP3macro structures (32). The final model contains 3 moleculesof P23pro-zbd, 3 zinc ions, 31 sulfate ions, and 5 molecules of MES. Atomic coor-dinates have been deposited in the Protein Data Bank under PDB ID code 4GUA.

Virological Studies. Infectious center and plaque assays were performedsimilarly to those described in ref. 46. Protein and RNA radiolabeling wereperformed as described in ref. 9. Harvested protein from BHK-J cells at timepoints postinfection were used in Western blotting experiments, and mem-branes were probed with anti-nsP2, anti-nsP3, and anti-GAPDH antibodies.

ACKNOWLEDGMENTS. We thank J. Bonanno for assistance with the dataprocessing; C. Rice and M. MacDonald for providing BHK-J cells, Sindbis virusDNA, and nsP3 antibody; J. Millonig and P. Matteson for the GAPDH antibody;and A. Basant, J. Chiu, F. Jiang, A. Khan, C. Rice, A. Stock, V. Stollar, andJ. Whidby for insightful discussions. S.A.Y. was supported by the NIH-NIAIDtraining Grant 2T32AI007403-18. This study was supported by NationalInstitutes of Health Grants AI080659 (to J.M.) and DK083356 and AI070101 (toA.G.), New Jersey Commission on Cancer Research Grant 10-1962-CCR-EO (toJ.M.), and Yerkes Research Center Base Grant RR-00165 (to A.G). We wish toexpress ourprofoundgratitude to the lateDr.Aaron Shatkin forhis unwaveringsupport, sage advice, and insightful discussions.

1. Griffin DE (2007) Alphaviruses. Fields Virology, ed Knipe DM (Lippincott Williams &Wilkins, Philadelphia), Vol 1, pp 1023–1067.

2. Thiboutot MM, et al. (2010) Chikungunya: A potentially emerging epidemic? PLoSNegl Trop Dis 4:e623.

3. Nkoghe D, et al. (2012) Clinical forms of chikungunya in Gabon, 2010. PLoS Negl TropDis 6:e1517.

4. de Groot RJ, Hardy WR, Shirako Y, Strauss JH (1990) Cleavage-site preferences ofSindbis virus polyproteins containing the non-structural proteinase. Evidence fortemporal regulation of polyprotein processing in vivo. EMBO J 9:2631–2638.

5. Vasiljeva L, et al. (2003) Regulation of the sequential processing of Semliki Forest virusreplicase polyprotein. J Biol Chem 278:41636–41645.

6. Shirako Y, Strauss JH (1994) Regulation of Sindbis virus RNA replication: UncleavedP123 and nsP4 function in minus-strand RNA synthesis, whereas cleaved productsfrom P123 are required for efficient plus-strand RNA synthesis. J Virol 68:1874–1885.

7. Lemm JA, Rümenapf T, Strauss EG, Strauss JH, Rice CM (1994) Polypeptide require-ments for assembly of functional Sindbis virus replication complexes: A model for thetemporal regulation of minus- and plus-strand RNA synthesis. EMBO J 13:2925–2934.

8. Strauss JH, Strauss EG (1994) The alphaviruses: Gene expression, replication, andevolution. Microbiol Rev 58:491–562.

9. Gorchakov R, Frolova E, Frolov I (2005) Inhibition of transcription and translation inSindbis virus-infected cells. J Virol 79:9397–9409.

10. Garmashova N, Gorchakov R, Frolova E, Frolov I (2006) Sindbis virus nonstructuralprotein nsP2 is cytotoxic and inhibits cellular transcription. J Virol 80:5686–5696.

11. Frolova EI, et al. (2002) Roles of nonstructural protein nsP2 and Alpha/Beta inter-ferons in determining the outcome of Sindbis virus infection. J Virol 76:11254–11264.

12. Russo AT, White MA, Watowich SJ (2006) The crystal structure of the Venezuelanequine encephalitis alphavirus nsP2 protease. Structure 14:1449–1458.

13. Rikkonen M, Peränen J, Kääriäinen L (1992) Nuclear and nucleolar targeting signals ofSemliki Forest virus nonstructural protein nsP2. Virology 189:462–473.

14. Rikkonen M, Peränen J, Kääriäinen L (1994) Nuclear targeting of Semliki Forest virusnsP2. Arch Virol Suppl 9:369–377.

15. Garmashova N, et al. (2007) The Old World and New World alphaviruses use differentvirus-specific proteins for induction of transcriptional shutoff. J Virol 81:2472–2484.

16. Frolov I, et al. (1999) Selection of RNA replicons capable of persistent noncytopathicreplication in mammalian cells. J Virol 73:3854–3865.

17. Dryga SA, Dryga OA, Schlesinger S (1997) Identification of mutations in a Sindbis virusvariant able to establish persistent infection in BHK cells: The importance of a muta-tion in the nsP2 gene. Virology 228:74–83.

18. Mayuri, Geders TW, Smith JL, Kuhn RJ (2008) Role for conserved residues of sindbisvirus nonstructural protein 2 methyltransferase-like domain in regulation of minus-strand synthesis and development of cytopathic infection. J Virol 82:7284–7297.

19. Perri S, et al. (2000) Replicon vectors derived from Sindbis virus and Semliki forestvirus that establish persistent replication in host cells. J Virol 74:9802–9807.

20. Petrakova O, et al. (2005) Noncytopathic replication of Venezuelan equine enceph-alitis virus and eastern equine encephalitis virus replicons in Mammalian cells. J Virol79:7597–7608.

21. Weiss B, Rosenthal R, Schlesinger S (1980) Establishment and maintenance of persis-tent infection by Sindbis virus in BHK cells. J Virol 33:463–474.

22. Frolov I, Schlesinger S (1994) Comparison of the effects of Sindbis virus and Sindbisvirus replicons on host cell protein synthesis and cytopathogenicity in BHK cells. J Virol68:1721–1727.

23. Koonin EV, et al. (1992) Computer-assisted assignment of functional domains in thenonstructural polyprotein of hepatitis E virus: Delineation of an additional group ofpositive-strand RNA plant and animal viruses. Proc Natl Acad Sci USA 89:8259–8263.

24. Karras GI, et al. (2005) The macro domain is an ADP-ribose binding module. EMBO J24:1911–1920.

25. Saikatendu KS, et al. (2005) Structural basis of severe acute respiratory syndromecoronavirus ADP-ribose-1″-phosphate dephosphorylation by a conserved domain ofnsP3. Structure 13:1665–1675.

26. Neuvonen M, Ahola T (2009) Differential activities of cellular and viral macro domainproteins in binding of ADP-ribose metabolites. J Mol Biol 385:212–225.

27. Lulla A, Lulla V, Merits A (2012) Macromolecular assembly-driven processing of the 2/3cleavage site in the alphavirus replicase polyprotein. J Virol 86:553–565.

28. Vihinen H, Ahola T, Tuittila M, Merits A, Kääriäinen L (2001) Elimination of phos-phorylation sites of Semliki Forest virus replicase protein nsP3. J Biol Chem 276:5745–5752.

29. Vihinen H, Saarinen J (2000) Phosphorylation site analysis of Semliki forest virusnonstructural protein 3. J Biol Chem 275:27775–27783.

30. Chance MR, et al. (2004) High-throughput computational and experimental techni-ques in structural genomics. Genome Res 14(10B):2145–2154.

31. Shi W, et al. (2005) Metalloproteomics: High-throughput structural and functionalannotation of proteins in structural genomics. Structure 13:1473–1486.

32. Malet H, et al. (2009) The crystal structures of Chikungunya and Venezuelan equineencephalitis virus nsP3 macro domains define a conserved adenosine binding pocket.J Virol 83:6534–6545.

33. Hahn YS, Strauss EG, Strauss JH (1989) Mapping of RNA- temperature-sensitive mu-tants of Sindbis virus: Assignment of complementation groups A, B, and G to non-structural proteins. J Virol 63:3142–3150.

34. Shirako Y, Strauss JH (1990) Cleavage between nsP1 and nsP2 initiates the processingpathway of Sindbis virus nonstructural polyprotein P123. Virology 177:54–64.

35. Hardy WR, Strauss JH (1989) Processing the nonstructural polyproteins of sindbis virus:Nonstructural proteinase is in the C-terminal half of nsP2 and functions both in cisand in trans. J Virol 63:4653–4664.

36. Russo AT, Malmstrom RD, White MA, Watowich SJ (2010) Structural basis for sub-strate specificity of alphavirus nsP2 proteases. J Mol Graph Model 29:46–53.

37. Krissinel E, Henrick K (2004) Secondary-structure matching (SSM), a new tool for fastprotein structure alignment in three dimensions. Acta Crystallogr D Biol Crystallogr 60(Pt 12 Pt 1):2256–2268.

38. Holm L, Sander C (1996) Mapping the protein universe. Science 273:595–603.39. Krishna SS, Majumdar I, Grishin NV (2003) Structural classification of zinc fingers:

Survey and summary. Nucleic Acids Res 31:532–550.40. Lastarza MW, Grakoui A, Rice CM (1994) Deletion and duplication mutations in the

C-terminal nonconserved region of Sindbis virus nsP3: Effects on phosphorylation andon virus replication in vertebrate and invertebrate cells. Virology 202:224–232.

41. Peränen J, Rikkonen M, Liljeström P, Kääriäinen L (1990) Nuclear localization ofSemliki Forest virus-specific nonstructural protein nsP2. J Virol 64:1888–1896.

42. Berg JM (1990) Zinc fingers and other metal-binding domains. Elements for inter-actions between macromolecules. J Biol Chem 265:6513–6516.

43. Kustatscher G, Hothorn M, Pugieux C, Scheffzek K, Ladurner AG (2005) Splicing regu-lates NAD metabolite binding to histone macroH2A. Nat Struct Mol Biol 12:624–625.

44. Fayzulin R, Frolov I (2004) Changes of the secondary structure of the 5′ end of theSindbis virus genome inhibit virus growth in mosquito cells and lead to accumulationof adaptive mutations. J Virol 78:4953–4964.

45. Marcotte LL, et al. (2007) Crystal structure of poliovirus 3CD protein: Virally encodedprotease and precursor to the RNA-dependent RNA polymerase. J Virol 81:3583–3596.

46. Lemm JA, Durbin RK, Stollar V, Rice CM (1990) Mutations which alter the level orstructure of nsP4 can affect the efficiency of Sindbis virus replication in a host-de-pendent manner. J Virol 64:3001–3011.

Shin et al. PNAS | October 9, 2012 | vol. 109 | no. 41 | 16539

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

021