Genetic Studies of the {beta}-Hairpin Loop of Rous Sarcoma Virus Capsid Protein

The first few residues of the Rous sarcoma virus (RSV) CA proteincomprise a structurally dynamic region that forms part of aGag-Gag interface in immature virus particles. Dissociationof this interaction during maturation allows refolding and formationof a ß-hairpin structure important for assembly ofCA monomers into the mature capsid shell. A consensus bindingsite for the cellular Ubc9 protein was previously identifiedwithin this region, suggesting that binding of Ubc9 and subsequentsmall ubiquitin-like modifier protein 1 (SUMO-1) modificationof CA may play a role either in regulating the assembly activityof CA in immature particles or mature cores or in controllingpostentry function(s) during the establishment of infection.In the present study, mutations designed to eliminate the consensusbinding site were used to dissect the potentially overlappingfunctions of these residues. The resulting replication defectscould not be traced to a failure to form particles of normalcomposition but, rather, to a deficit in genome replication.Genetic suppressors of two detrimental ß-hairpin mutationsimproved infectivity without restoring the consensus site orcreating a novel one elsewhere. Optimal restoration of infectivityto a Lys-to-Arg mutant required a combination of secondary changes,one on the surface of each domain of CA. Rather than arguingfor a critical role of Ubc9 and SUMO in RSV replication, thesefindings provide strong support for a structural role of theN-terminal residues and a particularly striking example of long-rangeinteractions between regions of CA in achieving a functionalcore competent for genome replication.

INTRODUCTION

Top

INTRODUCTION

The CA (capsid) protein, the major structural protein of allinfectious orthoretroviruses, plays critical roles in both theassembly and postentry phases of the replication cycle. As adomain of the Gag protein, CA contributes to the structuralframework of the immature particle (19, 63, 64). Once liberatedfrom Gag by proteolytic processing late in the assembly process,CA condenses to form a protein shell around the genome-proteincomplex, consisting of viral RNA, tRNA, the nucleocapsid protein(NC), and the reverse transcriptase (RT) and integrase (IN)enzymes. Structural integrity of the CA shell in this maturecore appears to be critical for activity of the reverse transcriptioncomplex in replicating the viral genome. An extensive geneticanalysis of the CA protein of human immunodeficiency virus type1 (HIV-1) demonstrated a strict correlation between normal capsidmorphology and infectivity (59). In the Rous sarcoma virus (RSV),mutations that perturb the integrity of CA and the capsid havepotent effects on infectivity due a failure to reverse-transcribethe viral genome (7, 12). Numerous other studies have describedCA mutations that cause loss of reverse transcription and infectivitywithout detectable effects on particle formation (1, 13, 17,18, 33, 43, 44, 52, 54, 55, 60). These genetic studies suggestthat the properly assembled CA plays a critical role duringthe early stages of the virus infection cycle. This interpretationis consistent with the demonstration that CA protein remainsassociated with cytosolic core-derived structures of HIV andof murine leukemia virus (MLV) during the process of reversetranscription (4, 14, 15, 36) and with the activity of mammaliancell restriction factors that block retrovirus infection bytargeting CA in the incoming core (22, 40, 51, 57). Still, themechanisms through which CA contributes to early replicationevents remain unexplained.

The first few amino acids of the CA protein comprise a structuralelement with an important role in the core maturation processesthat yield the properly assembled capsid. The amino terminusof the CA domain exists in an unfolded conformation in the Gagproteins of HIV and RSV and in the latter, at least, forms partof an extended interface between Gag molecules (39, 53). Cleavageof Gag by the viral protease PR to form the amino terminus ofmature CA allows refolding of this region into a ß-hairpinstructure. In HIV, this refolding is known to be accompaniedby structural changes throughout the N-terminal domain (NTD)of CA that appear to facilitate formation of NTD-NTD interactionsand maturation of the core (39, 53, 58). Mutations that preventthe hairpin from being anchored to the N-terminal domain ofCA have been shown to prevent core formation and infectivityin HIV and MLV (44, 58).

A crystallographic study of the CA NTD from MLV (37) demonstratedthat the ß-hairpin and first three ${alpha}$ -helices of CAform the interfaces that hold the NTDs of six CA monomers togetherin a hexameric ring structure that forms a prominent repeatingfeature of the capsid outer surface. The six ß-hairpinsline the wall of the central cavity of the hexameric ring (37).Neighboring hexamers are linked via dimeric C-terminal domain(CTD)-CTD interactions involving helix 9 that have been wellcharacterized in HIV (31). Cryoelectron microscopy of in vitroassembled HIV CA (31) and analyses of RSV, HIV, and MLV CA proteinsassembled on lipid monolayers (2, 20, 35) suggest that all orthoretrovirusesuse similar principles of protein organization to assemble afunctional core. In addition to the well-documented NTD-NTDand CTD-CTD interactions, the presence of an NTD-CTD interfacein assembled, mature HIV CA, but not in the immature Gag particles,has been demonstrated by hydrogen-deuterium exchange and proteinchemical cross-linking (29, 30). Involvement of the interdomaininterface is supported by genetic studies of RSV from our laboratory(5, 29, 30) but has not yet been confirmed by high-resolutionmethods in any virus.

Intriguingly, a consensus amino acid sequence (IKTE) was identifiedin the ß-hairpin of the RSV CA protein (50) whichcould potentially act as a binding site for the cellular regulatoryprotein Ubc9. The conjugation activity of Ubc9 is known to catalyzethe modification of the target protein by covalent attachmentof the small ubiquitin-like modifier protein 1, SUMO-1, to alysine residue in or near the Ubc9-binding site (23, 45). Proteinsumoylation is known in numerous systems to be involved in thecontrol of a variety of nuclear events including entry, subnuclearstructure formation, and the modulation of transcriptional activity,although the mechanisms of action are not well understood (23,49). Numerous examples of sumoylation of viral regulatory proteins(23), but only a small number of structural proteins, have beendocumented. This group includes at least two hantavirus NP proteins(27, 32) and two retroviral CA proteins (61, 65). The CA proteinsof Mason-Pfizer monkey virus and MLV have been shown by yeasttwo-hybrid analysis and in vitro interaction studies to bindUbc9 (61, 65). Lysine substitution in the Ubc9 binding sitein MLV blocked both in vitro sumoylation of CA and the establishmentof infection after the step of viral DNA synthesis, suggestingthat SUMO modification of CA may be important for an unidentifiedearly step of virus infection near or in the nucleus. Thus,Ubc9 and SUMO-1 represent attractive candidates as regulatorsof CA function in reverse transcription and/or nuclear eventsduring the establishment of retroviral infections. In RSV thelocation of the IKTE motif within the ß-hairpin raisesa further possibility that SUMO and/or Ubc9 could serve to regulateCA protein folding and/or capsid assembly.

In a previous study, replacement of the lysine residue in theIKTE motif with arginine dramatically reduced RSV infectivitybut did not affect particle production, suggesting a possiblerole for the sumoylation machinery in the maturation and/orearly postentry phases of replication (50). To evaluate thesignificance of the IKTE motif for the viral life cycle anddissect potentially overlapping functions in this region, anexamination of the replication defect in the arginine substitutionmutant and a more detailed genetic analysis, including isolationof second site suppressors, was undertaken. The findings supportthe importance of the region in the assembly of a functionalcore structure but do not identify a critical role for Ubc9and SUMO-1 in RSV replication.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Plasmids. CA mutations were created in the Prague C gag gene derived fromthe proviral vector pATV-8 (26, 48) and cloned into either pJD100,a replication-competent proviral vector which bears the srcgene, or pRS.V8.eGFP, in which src is replaced by the enhancedgreen fluorescent protein (EGFP) gene (8). The K244R substitutionwas transferred to pJD100 using the two SacII sites in gag.New mutations, T245I, E246A, and E246D, were generated by PCRmutagenesis; the threonine codon ACA was replaced with ATA,and the glutamate GAG was replaced with either GCG (alanine)or GAC (aspartate). The mutant alleles were then transferredinto pRS.V8.eGFP using SstI and FseI, thereby replacing the5' half of gag. R325C, C431R, R325C/C431R, and N343D substitutionswere similarly created by changing the arginine codon CGC toTGC (cysteine), the cysteine TGC codon to CGC (arginine), andasparagine AAT to GAT (aspartate). The C431R mutation was transferredinto pRS.V8.eGFP or pRS.K244R.eGFP using SdaI in gag and BstXIin pol. R325C and N343D mutations were transferred into variousproviral plasmids using SstI and SdaI, replacing the N-terminalhalf of gag. All clones were confirmed by DNA sequencing.

Budding assays. Assembly and release of virus particles by the wild-type andmutant Gag proteins was analyzed by transfection of QT6 cells(38) with proviral plasmids as described previously (7). Particlesreleased from [³⁵S]methionine-labeled cells were immunoprecipitatedwith rabbit anti-RSV serum, separated by sodium dodecyl sulfate-polyacrylamidegel electrophoresis (SDS-PAGE), and then quantified by phosphorimageanalysis. Budding efficiency was calculated by normalizing theamount of radiolabeled CA released into the medium for the amountof Gag expression measured in a parallel plate of cells; activityof the mutants was expressed relative to the wild type.

Infectivity assays. Virus-spreading assays were performed as previously described(50). Viruses produced by transfection of QT6 cells were normalizedfor RT activity by an exogenous RT assay (12) and used to infectplates of chicken DF-1 cells (46). Infected (i.e., GFP-expressing)cells were detected at postinfection time points by fluorescence-activatedcell sorting. The point at which 50% of cells were GFP positivewas evaluated by visual examination of the data. Focus formationassays were performed as previously described (5, 25) usingthe wild-type and K244R allele in the pJD100 vector to produceparticles in QT6 for inoculation into primary turkey embryofibroblasts.

Microscopy. For subcellular localization of GFP-tagged Gag, QT6 cells oncoverslips were transfected with either wild-type (WT) pGag.GFP(8) or its K244R derivative. After 24 h, cells were treatedwith dimethyl sulfoxide (negative control) or with 10 µg/mlleptomycin B; cells were then fixed with paraformaldehyde andexamined using a Leica TCS SP2 AOBS confocal microscope witha helium-argon laser (47). For electron microscopy (EM), stablyinfected DF-1 cells in 60 mm Permanox dishes (Electron MicroscopySciences, Ft. Washington, PA) were processed for EM as describedpreviously (12).

Core detergent resistance. To examine core stability, radiolabeled virions from stablyinfected DF-1 cells were pelleted through sucrose solutionscontaining either 1% Triton X-100 or no detergent. The CA proteinin each of the resulting supernatant and pellet fractions wasquantified by immunoprecipitation with anti-RSV serum, SDS-PAGE,and phosphorimage analysis and then expressed as a percentageof the total CA (5).

RT and Env incorporation. Virions from [³⁵S]methionine-labeled DF-1 cells stably infectedwith WT or K244R virus were collected by ultracentrifugationthrough a 25% sucrose cushion at 126,000 x g for 40 min at 4°Cin a Beckman TLA100.4 rotor. The RT activity in each resultingvirus pellet was determined by the exogenous RT assay (12).In parallel, CA proteins were analyzed by radioimmunoprecipitationwith anti-RSV serum, SDS-PAGE, and phosphorimage analysis, anda CA:RT ratio was calculated. For Env analysis, unlabeled virionswere collected in similar fashion. Env and CA were visualizedby Western blot analysis using rabbit anti-TM (transmembraneprotein) and rabbit anti-RSV antibodies, respectively. Env andCA bands were quantified by densitometry and data expressedas a ratio.

RNA and DNA isolation. WT and K244R particles from stably infected DF-1 cells werecollected for 24 h, passed through a 0.45-µm-pore-sizefilter, pelleted through 20% sucrose at 20,000 x g for 3 h,resuspended in phosphate-buffered saline, and normalized fortheir exogenous RT activity (12). RNA was extracted from equalamounts of WT and mutant virions using the QIAmp Viral RNA minikit (QIAGEN), treated with DNA-free DNase (Ambion), and analyzedby quantitative reverse transcription-PCR (RT-PCR) (see below).For isolation of viral DNA from infected cells, normalized suspensionsof WT and mutant particles were added to uninfected DF-1 cells,and low-molecular-weight viral DNA was isolated at 14 h postinfection(7, 24). For preparation of endogenous reverse transcription(ERT) products, equal amounts of the WT and mutant viruses wereincubated in standard ERT buffer to activate reverse transcription(7, 41). After 3 h at 42°C, the resulting viral DNA wasextracted using a QIAquick PCR purification kit (QIAGEN) andanalyzed by quantitative PCR.

Q-PCR. The following primers corresponding to the RSV long terminalrepeat (LTR) and the 5' and 3' untranslated regions were designedto amplify various regions of the viral DNA produced by reversetranscription in infected cells: primer A (5'-GCCATTTGACCATTCACCA-3')and primer B (5'-AATGAAGCCTTCTGCTTCATG-3') for minus-strandstrong stop DNA; primer C (5'-ATTCCGCATTGCAGAGATATTG-3') andprimer B for first-strand transfer; primer A and primer D (5'-GATGGAGACAGGATCGCCAC-3')for second-strand transfer; and primer A and E (5'-CATGTTGCTAACTCATCGTTACCA-3')for two-LTR (2LTR) circles. Additional primers (5'-CCTCCCCCTCTTAACCAAAAC-3'and 5'-TGCTATTTCATCTTTCCCTTGC-3') were designed for the detectionof mitochondrial DNA in the infected cell extracts. FAM/TAMRA(6-carboxyfluorescein/6-carboxytetramethylrhodamine) dual-labeledprobes (Sigma-Genosys) specific for the RSV LTR (5'-CCATCAACCCAGGTGCACACCAATG-3')or chicken mitochondrial DNA (5'-CAGTATAGGCGATAGAAAAGACTACCCCGGC-3')were used for the quantitative-PCR (Q-PCR) reaction using aQuantiTect Probe PCR kit (QIAGEN). The viral DNA products detectedin each sample were adjusted for the total input DNA, as monitoredby the chicken mitochondrial DNA, to account for variationsin DNA recovery during extraction. Quantitative RT-PCR of isolatedviral RNA was performed using the primer pair A and D, the RSVLTR dual-labeled probe, and a QuantiTect Probe RT-PCR kit (QIAGEN).

Isolation of revertants. Cell-free medium from cells stably infected with wild-type RS.V8.eGFPor its K244R or E246A mutant was placed onto uninfected DF-1cells. Cells were passed for one to several weeks before themedium was collected and transferred onto fresh, uninfectedDF-1 cells. The transfer was repeated 12 times to allow ampleopportunity for virus evolution. At various points, the culturemedium was tested for the presence of infectious virus by monitoringthe appearance of RT activity in the medium after transfer touninfected cells. After 12 transfers, virions were collected,and viral RNA was isolated as above and amplified in $~$ 1.4-kbsegments using SuperScript III enzyme (Invitrogen) for RT-PCR.DNA products spanning the 5' LTR through env were then sequencedwith the same primers used for amplification.

RESULTS AND DISCUSSION

Top

RESULTS AND DISCUSSION

The IKTE sequence resembles the accepted consensus Ubc9 bindingand sumoylation site, ${Psi}$ KX(E/D), in which is a hydrophobic residueand X is any residue (3, 23, 45). In the K244R mutant (Fig.1A) substitution of an arginine for the lysine is predictedto block Ubc9 binding and eliminate this lysine as a potentialsite for SUMO addition (50). The replication deficiency of theK244R mutant may be due to the failure of a Ubc9-mediated event(such as sumoylation) that is important for normal infectivityor, alternatively, to structural perturbation of the CA proteinor its Gag precursor that compromises the integrity of eitherimmature or mature virus particles. In an attempt to distinguishbetween these possibilities, a more detailed genetic analysisof this region of CA and an examination of the nature of thereplication block in the K244R mutant were undertaken. If theprimary role of the ₂₄₃IKTE₂₄₆ sequence is to serve as a Ubc9binding and sumoylation site, then replacement of the threoninewith another amino acid or change of the glutamate to aspartateshould have little effect on replication (3, 45). In contrast,replacement of the glutamate with an alanine would be expectedto have a detrimental effect on RSV replication resembling theK244R mutation (50). To test these predictions, T245I, E246D,and E246A substitutions were engineered into the proviral plasmidpRS.V8.eGFP (9).

View larger version (28K):
[in this window]
[in a new window]

FIG. 1. Mutation of the IKTE sequence in RSV CA. (A) Location of the ß-hairpin mutations K244R and E246A and their corresponding suppressors in RSV Gag and CA. The mutant names reflect the Gag amino acid numbering. Arrows connect K244R and E246A to their respective suppressors. SP, spacer peptide. (B) Budding activity of the mutant Gag proteins expressed relative to the wild type. The graph represents the results of four independent experiments. (C) The spreading of GFP-marked viruses in cultures of DF-1 cells. Shown are the results of one typical experiment of three performed. (D) The specific infectivity of the K244R mutant measured by focus formation in primary turkey cells and expressed relative to the wild-type virus. The graph shows the average of three independent experiments.

Virus release and infectivity. Upon expression of each of the three new mutants in quail (QT6)cells, all appeared entirely wild type in particle release (Fig.1B), as did the previously described K244R mutant (50). To analyzeinfectivity, wild-type and mutant particles were harvested fromtransfected QT6 cells, and equal doses were transferred to freshcultures of chicken DF-1 cells (8, 50) that were then analyzedover time for GFP expression. The growth curve for the E246Dmutant was indistinguishable from the wild-type parent witha half-time of spreading of about 8 to 9 days. The E246A mutantresembled K244R with a half-time of spreading almost three timesthat of the wild type (Fig. 1C). T245I had an intermediate effectwith a half-time of spreading about twice that of the wild type.The same relative effects were seen when the infecting doseof each virus was increased 20-fold (data not shown). Thesedata demonstrate the importance of these amino acids for efficientreplication of the virus and are consistent with the possibleinvolvement of Ubc9 and SUMO-1 in virus infectivity.

The magnitude of the defect in specific infectivity caused byK244R was measured in primary turkey embryo fibroblasts by afocus formation assay (5, 25) which approximates a single-cycleinfectivity assay. In cultures infected with wild-type JD100virions, small src-transformed foci were first observable atabout 7 days postinfection, whereas detection of foci requiredanother 3 to 5 days for the K244R mutant (data not shown). Byday 16, when foci on all plates were fully formed, K244R mutant-infectedcells yielded twofold fewer foci than those infected with wild-typevirus (Fig. 1D), indicating a moderate specific infectivitydefect of about twofold. Thus, the loss of infectivity in theRSV ß-hairpin mutations resemble the mild (2- to 10-fold)diminishment of infectivity observed in alanine substitutionsin the ß-hairpin of HIV CA (59). A study to identifythe basis for the reduced infectivity of the RSV mutants wasinitiated.

Nuclear trafficking of the K244R mutant Gag. RSV Gag has been shown to enter the nucleus during the processof assembly, and two Gag mutations that interfere with thispathway have dramatic replication defects (9, 47). Since sumoylationis known to regulate nuclear localization of certain proteins(23, 49), the possibility that the K244R mutation interfereswith the ability of Gag to enter the nucleus was examined. QT6cells expressing a wild-type Gag-GFP fusion protein (8) or itsK244R derivative were treated prior to examination by confocalmicroscopy with 10 µg/ml leptomycin B, which specificallyinhibits CRM1-mediated nuclear export and allows wild-type Gagto accumulate in the nucleus. Under these conditions, the K244Rmutant Gag also accumulated in the nucleus, whereas in untreatedcontrol cells both wild-type and mutant proteins were seen inthe cytoplasm and at the plasma membrane (data not shown). Thus,there is no evidence that the IKTE motif is required for nuclearimport of the Gag protein prior to budding.

Characterization of K244R virions. Proteolytic processing of Gag in wild-type particles and thefour mutants was examined by radioimmunoprecipitation of Gag-derivedproteins (7, 12). The processing patterns for all the mutantsproved to be similar to wild type (Fig. 2A). No unusual cleavageintermediates were detected. In particular, the cluster of CAbands (the two mature species of 237 and 240 residues and theCA-spacer peptide cleavage intermediate) that is normally detectedin wild-type virus under these labeling conditions (11, 42)was unaffected by the mutations, indicating that the substitutionsdid not perturb the nearby N-terminal p10-CA cleavage site orthe distal C-terminal cleavage sites.

View larger version (51K):
[in this window]
[in a new window]

FIG. 2. Characterization of mutant virions. (A) Analysis of Gag proteins and cleavage products in cell lysates and medium fractions of QT6 cells expressing wild-type and mutant proviral genomes. Proteins were radiolabeled, immunoprecipitated, and separated by SDS-PAGE as for the budding assays (Materials and Methods). (B) Morphology of wild-type and K244R mutant virus produced by QT6 cells examined by transmission electron microscopy (magnification, x37,000; bar, 100 nm). (C) The stability of wild-type and mutant particle cores evaluated by comparison of the CA protein recovered in the detergent-soluble and particulate fractions after exposure to nonionic detergent. For each treatment condition, CA was immunoprecipitated, separated by SDS-PAGE, and quantified by phosphorimage analysis; the CA protein in each fraction was expressed as a percentage of the total. The graph shows the average of three independent experiments. Supe, supernatant.

To examine virion size and core morphology, DF-1 cells stablyinfected with the wild-type virus or the K244R mutant were embedded,stained, thin-sectioned, and analyzed by EM. K244R particleswere of normal size and morphology with a condensed ribonucleoproteincomplex clearly visible (Fig. 2B). As is often the case withEM studies of RSV, the capsid shell could not be visualizedin either the wild-type or mutant particles.

The stability of the wild-type and mutant cores was evaluatedby pelleting virions through sucrose step gradients in the presenceor absence of 1% Triton X-100. Between 25 and 40% of wild-typeCA is reproducibly found in the pellet in a form believed torepresent intact or partially disassembled cores (5, 7). ManyCA mutations cause increased instability of cores, as detectedby this assay (5, 7, 11, 12); however, the K244R mutant showedonly a very slight and statistically insignificant (P = 0.35)decrease in the amount of detergent-resistant CA compared tothe wild-type parent (Fig. 2C). Thus, the K244R substitutionhad no demonstrable effect on Gag processing, core morphology,particle size, or core sensitivity to detergent.

Pol, Env, and viral RNA incorporation. To evaluate whether or not K244R virions possess a normal complementof RT, the CA protein detectable in suspensions of WT and mutantparticles was compared to the amount of RT enzymatic activity.Unlike the case of certain HIV CA mutations previously described(54), no obvious effect of the K244R mutation on the CA:RT ratiocould be seen, indicating that the incorporation of the Polprotein and its activation by PR processing to form active RTare normal (data not shown). Similarly, the abundance of Envglycoprotein, determined by Western blotting of standardizedpreparations of unlabeled particles, was indistinguishable betweenwild-type and mutant (data not shown), as was the yield of genomicRNA analyzed by quantitative RT-PCR (Fig. 3A). Thus, the limitedreplication capacity of the K244R mutant could not be linkedto any clear defect in incorporation of essential virion components.

View larger version (24K):
[in this window]
[in a new window]

FIG. 3. Analysis of viral nucleic acids. (A) Viral RNA extracted from standardized preparations of wild-type and mutant virions and measured by quantitative RT-PCR. Duplicate analyses were performed for each RNA sample, and the mutant RNA levels were expressed relative to the wild-type. The average of three independent experiments is displayed. (B) Viral DNA products extracted from wild-type and K244R virus-infected DF-1 cells and analyzed by quantitative PCR for the presence of minus-strand strong-stop, first-strand transfer, second-strand transfer, and 2LTR circle DNA products. The average of five independent experiments is shown. (C) Capacity of extracellular virions to synthesize viral DNA in an ERT reaction. The DNA products were extracted from virions and analyzed by Q-PCR. The data from five independent experiments were averaged.

Analysis of viral DNA products. The reverse transcription capabilities of wild-type and K244Rvirions were compared through the use of Q-PCR to detect DNAsynthesis intermediates in the cytosol of infected DF-1 cells(Fig. 3B). The abundance of minus-strand strong-stop DNA, whichmarks the completion of the first 101 nucleotides of viral DNA,was decreased by $~$ 45% in K244R virus-infected cells comparedto wild type. Thus, the mutation affects synthesis at a veryearly stage, well before the point that entry of the reversetranscribing complexes into the nucleus is thought to occur(6, 56). First-strand transfer and second-strand transfer productswere reduced by $~$ 25 to 35% compared to the wild type, indicatingno further blocks to synthesis at these stages.

The presence of circularized 2LTR circles, considered a hallmarkof nuclear entry of the viral DNA (6, 56), was decreased by $~$ 65% relative to the wild type when analyzed at 14 h postinfection(Fig. 3B). This represents a slightly more severe defect thanseen with the early DNA products. Taken together with the slowerdevelopment of transformation in the focus assay (see above),this could be an indication that either completion of the secondstrand of viral DNA or the entry of the DNA products into thenucleus is slowed by the K244R mutation. Regardless, the finalyield of transformed foci exactly mirrored the twofold deficitin the earliest (e.g., the negative-strand strong-stop) DNAproduct, arguing that most particles that are able to completethe initial phase of genome replication will eventually succeedin infecting the cell. Thus, we conclude that the K244 residueis not critical for the final (nuclear) stages of genome replication.

A defect in DNA synthesis could also be detected in extracellularparticles by the ERT assay (Fig. 3C). The ability of virionsof the K244R mutant to synthesize both minus-strand strong-stopand first-strand transfer DNAs was reduced by $~$ 50% relative tothe wild-type (Fig. 3C) in spite of normal RT and genomic RNAcontent (Fig. 3A). This effect is consistent with the reversetranscription defect in infected cells (Fig. 3B) and with thetwofold reduction in specific infectivity (Fig. 1D). Thus, thereverse transcription defect is not caused by exposure to theintracellular environment early in infection; rather, this arguesthat the quality of the reverse transcription complexes in theextracellular virions is compromised by the K244R mutation,likely due to a subtle structural perturbation.

Isolation of second-site suppressors. The delayed-growth phenotype of the K244R and E246A viruseswas sufficient to allow the selection of revertant viruses withimproved replication capacity by serially transferring virusonto fresh uninfected cells. After six such transfers of theE246A virus, the culture supernatant contained virus that wasvirtually identical to the wild-type in its rate of spread,while the selected K244R virus had improved only slightly comparedto unpassaged K244R virus (data not shown). By transfer 12,both the selected K244R and selected E246A viruses were replicatingat the wild-type level (data not shown). The original E246Aand K244R mutations were detected in the respective culturesby RT-PCR amplification of the viral genome from the releasedparticles and sequencing of the DNA products. New second-sitemutations in the CA sequence were also identified in each culture.The selected E246A virus contained a single new substitutionin CA (N343D, caused by an A-to-G substitution at nucleotide1407) which results in a change from asparagine to aspartatein the NTD (Fig. 1A and 4A). Two new CA mutations were foundin the RNA from the selected K244R population after 12 transfers:R325C (a C-to-T change at nucleotide 1353), which causes anarginine to cysteine change in the NTD, and C431R (a T-to-Cchange at nucleotide 1671), which alters the CTD with a cysteineto arginine substitution. Only the R325C substitution was detectableafter only six transfers. In none of the cultures were any furthernucleotide changes detected elsewhere in gag or in any otherregion of the genome.

View larger version (50K):
[in this window]
[in a new window]

FIG. 4. Second-site suppressors of ß-hairpin mutations. (A) Long-range interactions between ß-hairpin residues and distant residues in the N- and C-terminal CA domains. Ribbon diagrams of the NTD and CTD were created using Protein Data Bank files 1EM9 and 1EOQ. The flexible interdomain linker is illustrated by the dashed line. The K244 and E246 residues in the ß-hairpin are connected by arrows to the genetically interacting residues in the NTD and CTD. The amino acid change(s) resulting from each mutation is indicated at bottom. (B) Suppression of the E246A ß-hairpin mutation by the second-site N343D substitution in the NTD. (C) The effects of the R325C and C431R mutations and the R325C/C431R double mutation on infectivity of a virus bearing the wild-type ß-hairpin sequence. (D) Suppression of the growth deficit due to the K244R ß-hairpin substitution by the R325C/C431R (RC/CR) interdomain combination mutation. For each graph in panels B, C, and D, one representative graph from three replicate experiments is shown.

The suppression activity of the N343D, R325C, and C431R substitutionswas tested by recreating them in the pRS.V8.eGFP proviral plasmidwith and without the original ß-hairpin mutations.The three new mutations individually had no adverse effect onthe ability of the virus to assemble and release virus particlesfrom DF-1 cells (data not shown). When viruses bearing singleand double mutations were evaluated for infectivity, the N343Dsubstitution alone had no significant impact (Fig. 4B), consistentwith the fact that this substitution is a naturally occurringpolymorphism in certain strains of RSV (16). Combined with E246A,however, the N343D substitution was able to restore the infectivityof the E246A mutant to fully wild-type levels (Fig. 4B).

The R325C and C431R substitutions each had a mild detrimentaleffect on the growth of virus that was otherwise wild type,and these effects were additive in the R325C/C431R double mutant(Fig. 4C). Nevertheless, the combination R325C/C431R was ableto strongly suppress the K244R phenotype in the triple mutant,although neither substitution alone could do this (Fig. 4D).In the converse comparison, the K244R substitution was alsoable to improve the growth of the R325C/C431R double mutant;the half-time of spreading for the triple mutant was 6 daysversus 9 days for the double mutant (Fig. 4D). Thus, these threesubstitutions, although positioned far apart from one anotheron the CA structural model (Fig. 4A), are mutually adaptive.

Finally, suppression was found to be allele specific in thatthe N343D substitution had no ability to suppress the K244Rmutation, nor could R325C/C431R improve the growth of the E246Aß-hairpin mutation (data not shown).

A role for Ubc9 and SUMO? Characterization of the K244R mutant has demonstrated no clearlink between the IKTE Ubc9 consensus binding site and eventsof replication known to involve the nucleus, where many activitiesof SUMO-1 are known to occur. The phenotype of the K244R mutationis distinctly different from the CA sumoylation mutants of MLVthat are blocked after DNA synthesis (65). Furthermore, thegenetic suppressors of the K244R and E246A mutations compensatedfor the growth defects caused by the original mutations withouteither restoring a sumoylation consensus site at the originallocation or creating one elsewhere in CA or Gag. Thus, near-normalgrowth can occur independently of any Ubc9 and/or SUMO activitymediated by the IKTE motif, given changes to the amino acidsequence at distant regions of CA. Efforts to detect Ubc9 bindingto RSV Gag were inconclusive (data not shown). We must concludethat, while it remains a possibility that IKTE is recognizedby Ubc9 and perhaps even modified by SUMO addition, any influenceon the replicative capacity of the virus is likely to be indirectdue to influences on CA structure.

Structural implications. The appearance of the second-site suppressors at sites scatteredacross the CA protein argues that the original ß-hairpinmutations likely modulate virus infectivity by acting upon proteinstructure and perturbing normal assembly. The IKTE sequencelies within a region that not only contributes to intermolecularinteractions between Gag molecules in immature particles andbetween CA monomers in mature cores but also undergoes dramaticrefolding during the maturation process. Alterations of thehairpin residues, therefore, could conceivably compromise replicationby affecting Gag-Gag interactions, refolding of the first 14residues to make the ß-hairpin, the formation of CA-CAinteractions in the assembled capsid, or some combination ofthese.

In wild-type RSV Gag, T245 appears to be involved directly inthe Gag-Gag interface, while the K244 and E246 residues pointaway from the intermolecular interface and into the aqueousenvironment (39). In the mature CA, all four residues of theIKTE are surface exposed at the top of the hairpin and not apparentlyinvolved directly in CA-CA interactions (28). Direct contactbetween the IKTE residues and those altered by the suppressormutations appears impossible, given the existing structuralmodels for CA monomers, NTD hexamers, and larger arrays of assembledCAs (10, 20, 21, 28, 31, 35, 62). Therefore, it is likely thatsuppression occurs by a more indirect means such as influencingprotein folding/conformation or protein interfaces distant fromthe hairpin residues. The surface location of the suppressorsleads us to favor the latter possibility.

In the allele specificity experiment, failure of the N343D substitutionto suppress the K244R growth defect and of the R325C/C431R combinationto do the same for the E246A substitution indicates that thestructural consequences of the K244R and E246A mutations areunique. It is not possible to correct both detrimental mutationswith the same global strengthening mechanism. This is at leastconsistent with existing structural information. The N343D suppressormaps to the small helix 5 (Fig. 4A, on the "top" of the CA monomer)that is not directly involved in mature CA-CA interactions (29,30, 39, 62) but which has been suggested to be part of the extendedGag-Gag interface in the immature RSV particle (39). On theother hand, the R325C and C431R substitutions (joint suppressorsof K244R) alter helices 4 and 9, respectively, which are knownto participate in the NTD-CTD and CTD-CTD interactions thatform during HIV maturation (29, 30, 62). This raises the possibilitythat the E246A substitution alters Gag-Gag interactions, leadingto secondary effects on cores upon maturation, while the K244Rsubstitution primarily affects interactions within mature CAassemblies.

Other less dramatic examples of long-range interactions havebeen described within the CTD (34) and within the NTD of HIVCA (54). Also, in a previous study, RSV mutations in the highlyconserved major homology region of the CA CTD (5) were foundto be compensated in revertant viruses by secondary changesin either an adjacent ${alpha}$ -helix or by an interdomain interactionwith surface residues in the NTD. The K244R mutation and itstwo cooperating suppressors, however, provide a particularlypowerful demonstration of the importance of long-range structuralinfluences in forming a functional virion core. Unfortunately,the lack of high-resolution structural models for either assembledimmature particles or mature capsids of any retrovirus limitsour ability to explain more fully the long-range structuraleffects of the mutations. All these mutations, however, willultimately serve as important tests of validity as more completemodels of the immature and mature assemblies are developed.

Finally, the lethal major homology region mutations previouslydescribed in RSV (7) and the ß-hairpin substitutionspresented here, as well as others characterized in HIV and MLV(1, 17, 18, 54, 55), are similar in their effects on replication:all affect the capacity of the viral core to carry out an earlystep of viral DNA synthesis. Taken altogether these findingsdemonstrate that some structural feature of the capsid, whichremains unidentified, is critical for either the formation orthe function of reverse transcription complex.

ACKNOWLEDGMENTS

Thanks are due to Judith M. White for the anti-TM antibody,BoYeon Choi Kimmel and Tam-Linh Nguyen for technical assistanceon mutant characterization, Ira Ropson for consultation on proteinstructure, and John Purdy for careful review of the manuscript.Several procedures were performed in the Pennsylvania StateUniversity College of Medicine Core Facilities, where RolandMeyers (EM), Terry Ruger and Dan Krissinger (Q-PCR), and NateSheaffer (fluorescence-activated cell sorting) provided essentialexpertise.

This work was supported by National Institutes of Health grantsCA47482 to J.W.W. and CA100322 to R.C.C.

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology and Immunology, Pennsylvania State University College of Medicine, 500 University Drive, Hershey, PA 17033. Phone: (717) 531-3528. Fax: (717) 531-6522. E-mail: rcraven{at}psu.edu.

Published ahead of print on 8 November 2006.

Present address: Centocor R&D, Inc., 145 King of PrussiaRd., Radnor, PA 19087.

REFERENCES

Top