An Extended Stem-Loop 1 Is Necessary for Human Immunodeficiency Virus Type 2 Replication and Affects Genomic RNA Encapsidation

Genomic RNA encapsidation in lentiviruses is a highly selectiveand regulated process. The unspliced RNA molecules are selectedfor encapsidation from a pool of many different viral and cellularRNA species. Moreover, two molecules are encapsidated per viralparticle, where they are found associated as a dimer. In thisstudy, we demonstrate that a 10-nucleotide palindromic sequence(pal) located at the 3' end of the ${psi}$ encapsidation signal iscritical for human immunodeficiency virus type 2 (HIV-2) replicationand affects genomic RNA encapsidation. We used short-term andlong-term culture of pal-mutated viruses in permissive C8166cells and their phenotypic reversion to show the existence ofa structurally extended SL1 during HIV-2 replication, formedby the interaction of the 3' end of the pal within ${psi}$ with a motiflocated downstream of SL1. The stem extending HIV-2 SL1 is structurallysimilar to stem B described for HIV-1 SL1. Despite the highdegree of phylogenetic conservation, these results show thatmutant viruses are viable when the autocomplementary natureof the pal sequence is disrupted, but not without a stable stemB. Our observations show that formation of the extended SL1is necessary during viral replication and positively affectsHIV-2 genomic RNA encapsidation. Sequestration of part of thepackaging signal into SL1 may be a means of regulating its presentationduring the replication cycle.

INTRODUCTION

Top

Introduction

The 5'-untranslated regions of retroviral RNAs contain severalstructured motifs that are involved in various steps in viralreplication, including transcriptional transactivation, splicing,encapsidation, dimerization, and initiation of reverse transcription(for a review, see reference 3). Encapsidation of the genomicRNA in lentiviruses is a very specific process. Two moleculesof unspliced RNA are selected for encapsidation over a vastexcess of different spliced viral RNA species and cellular RNA.The historical model of retroviral RNA encapsidation has beenthat viral structural proteins like Gag recognize and bind specificRNA motifs located downstream of the major splice donor site,and thus only present in the unspliced viral RNA, and transfera pair of these molecules into one newly forming viral particle(for a review, see references 11 and 20).

However, most recent studies suggest that the encapsidationsignal is not limited to a single RNA element. The human immunodeficiencyvirus type 1 (HIV-1) encapsidation signal is multipartite, withstructural elements located across the 5'-untranslated and Gag-encodingregions (4, 6, 31). Similarly, HIV-2 and simian immunodeficiencyvirus (SIV) RNA encapsidation motifs have been localized eitherdownstream of the major splice donor site (38), upstream (16,32, 39, 46), or both (1, 37).

A hallmark of retroviral RNA encapsidation is that two moleculesof genomic RNA are encapsidated per viral particle (for a review,see references 14, 34, and 41). Gentle extraction and deproteinizationof virions revealed that the two encapsidated RNA moleculesform a dimer through noncovalent RNA-RNA interactions (28).Electron microscopy studies with increasing denaturation identifiedthe 5' ends of the genomic RNA molecules as the location ofthe strongest RNA-RNA interaction in dimers isolated from viralparticles; this interaction was named the dimer linkage structure(see reference 2 for the original study and reference 19 fora study with HIV-1). In HIV-1, a strong candidate involved inthe formation of the dimer linkage structure was identifiedin vitro upstream of the major splice donor site. This elementwas named the dimerization initiation site or stem-loop 1 (SL1)(18, 25, 33, 35, 45). Besides its role in genomic RNA dimerization,the SL1 sequence was shown to be involved in other stages ofHIV-1 replication, notably encapsidation (see Table 1 in reference34).

View this table:
[in this window]
[in a new window]

TABLE 1. Changes in the ${psi}$ -SL1 region occurring during long-term culture of strong and weak pal-defective viruses

We previously presented evidence that the ${psi}$ encapsidation signalinterferes with SL1-mediated HIV-2 RNA dimerization in vitro(23). The sequence responsible is a 10-nucleotide (nt) palindromicsequence (pal) located at the 3' end of ${psi}$ . The pal motif wasalso shown to mediate RNA dimerization upon binding of an antisenseoligonucleotide directed against SL1 (23). We and others showedthat the dimerization of HIV-2 leader RNA is regulated in vitroby a long-distance intramolecular interaction between nucleotideslocated at the 5' end of U5 and nucleotides encompassing thegag translation initiation site (10, 24). Thus, several short-and long-distance interactions influence the presentation ofthe ${psi}$ -SL1 region in HIV-2 leader RNA in vitro.

Since the pal motif is located within the ${psi}$ encapsidation signal,is phylogenetically conserved (see Discussion), and can influencedimerization of HIV-2 leader RNA fragments in vitro, we haveanalyzed in this study the function of HIV-2 pal during replicationin permissive C8166 cells. We also used long-term culture ofdefective viruses and their phenotypic reversion to gain insightinto the necessity of the palindromic sequence and functionallyimportant secondary structure interactions. In particular, weshow that the defects in viral replication for some pal-mutatedviruses can be compensated for by nucleotide changes in or closeto the mutated pal element. Our results provide evidence forthe existence of a structurally extended SL1 during HIV-2 replication,formed by the interaction of the 3' end of the 10-nt pal elementin ${psi}$ with a motif located downstream of SL1. Our observationsalso suggest that formation of the extended SL1 during viralreplication positively affects HIV-2 genomic RNA encapsidation,although a palindromic sequence is not strictly required forviral viability.

MATERIALS AND METHODS

Top

Materials and Methods

Plasmid construction. To engineer mutations in the 5'-untranslated region, we useda plasmid derived from a modified pROD10 plasmid containingfull-length HIV-2 ROD (17, 42). The modified plasmid pROD10was provided by the EU Programme EVA/MRC Centralised Facilityfor AIDS Reagents, NIBSC, United Kingdom. The AatII-XhoI fragmentfrom modified pROD10 digestion was subcloned into pGEM7-Zf+(Promega). This plasmid contains the 5' long terminal repeatand most of the gag coding region (Fig. 1) (22). The numberingis based on the genomic RNA sequence of an HIV-2 ROD isolate(GenBank accession no. M15390). Most mutations were introducedusing a QuikChange II XL site-directed mutagenesis kit (Stratagene)and the appropriate primers. The reversion-associated mutationsin the ${psi}$ -SL1 region were introduced using a QuikChange II XLkit with plasmids bearing the original mutations. The AatII-XhoIfragments of plasmids bearing the mutations were then reinsertedin the modified pROD10 backbone. All constructs were checkedby DNA sequencing.

View larger version (27K):
[in this window]
[in a new window]

FIG. 1. 5' leader region of HIV-2 ROD genomic RNA and mutations in the encapsidation domain. (A) The landmark sequences with known functions are indicated by boxes, with their names indicated above. TAR, polyA signal, PBS, ${psi}$ , SL1, SD, and gag represent the transactivation region, the poly(A) signal domain, the primer binding site, the encapsidation signal, stem-loop 1, the major splice donor site, and the 5' end of the Gag protein coding region, respectively. The secondary structure of nucleotides 380 to 444 is represented, with gray boxes highlighting the 10-nt palindromic sequence pal and the 6-nt palindromic sequence in the loop of the SL1 element. The short thick line below nt 380 to 408 represents the encapsidation signal ${psi}$ characterized in cell culture (15). (B) Replication kinetics of wild-type (open squares), ${psi}$ -negative (open triangles), and pal-negative (closed circles) viruses in C8166 cells. DNA plasmids were transfected into COS-7 cells, and viruses were isolated from the supernatant. A standardized amount of viral particles (10 ng of p27 capsid, as determined by ELISA) was used to infect permissive C8166 cells. Aliquots of supernatants were then assayed for reverse transcriptase activity. (C) (Top) Sequences of the wild-type and scrambled pal 392-401 regions for the clones used in this experiment. SCR1 was designed to both destroy the palindrome and disrupt a proposed ${psi}$ -SL1 interaction (23, 24). SCR2 is the complementary sequence to SCR1. The nucleotides different from the wild-type sequence are represented with lowercase letters. (Bottom) Replication kinetics of wild-type (open squares), SCR1 (open triangles), and SCR2 (closed circles) pal viruses in C8166 cells. DNA plasmids were transfected into COS-7 cells, and viruses were isolated from the supernatant. A standardized amount of viral particles (10 ng of p27 capsid, as determined by ELISA) was used to infect permissive C8166 cells. Aliquots of supernatants were then assayed for p27 capsid by ELISA. (D) (Top) Sequences of the wild-type and mutated pal 392-401 regions for the clones used in this experiment. Weak and strong pal sequences were chosen from in vitro studies in which they were shown to form weaker and stronger RNA-RNA duplexes, respectively, than the wild-type sequence (36). (Bottom) Replication kinetics of wild-type (open squares), weak (open triangles), and strong (closed circles) pal viruses in C8166 cells. Viral replication was followed as described for panel C.

Cell culture and transfection. COS-7 cells were maintained in Dulbecco's modified Eagle's mediumsupplemented with 10% fetal calf serum, penicillin, and streptomycin(Invitrogen). Transient transfection of COS-7 cells was performedusing a Trans-IT-COS transfection kit (Mirus). Cells and mediumwere harvested at 2 days posttransfection. HIV-2 capsid proteinlevels in the medium were quantified by a p27-specific enzyme-linkedimmunosorbent assay (SIV p27 ELISA; Zeptometrix). The infectivityof viruses in the medium was quantified using P4-CCR5 indicatorcells. This cell line expresses the CD4 and CCR5 receptors andcontains a beta-galactosidase gene under the control of theHIV long terminal repeat, thus detecting synthesis of the transactivatingTat protein after productive infection of the cells by the inputviruses (5). After 2 days of culture postinfection, the cellswere stained for beta-galactosidase activity, and the blue infectedcells were counted. The numbers were then normalized first tothe amounts of input viruses, as determined by reverse transcriptaseactivities, and second to the level of infection by wild-type(wt) virus to give the infectivities of viruses relative tothe wild type.

Cell culture and infection. C8166 cells (NIH AIDS reagent 404) were maintained in RoswellPark Memorial Institute (RPMI) 1640 medium supplemented with10% fetal calf serum, penicillin, streptomycin, and glutamine(Invitrogen). Medium from transfected COS-7 cells containing10 ng of HIV-2 p27 capsid protein, as determined by ELISA, wasused to infect 2 x 10⁴ C8166 cells. This corresponds to a multiplicityof infection of 0.01 for the wild-type virus, as determinedwith P4-CCR5 indicator cells (data not shown). After 12 h, cellswere washed twice and resuspended in RPMI medium-fetal calfserum. An aliquot of medium was then taken and served as a reference(p27 ELISA). Aliquots of medium were taken daily and spun toremove cells, and the supernatants were frozen at –80°C.Viral replication was followed for 1 or 2 weeks by quantifyingthe level of either CAp27 protein or reverse transcriptase activityin the supernatant. The CAp27 protein concentration was determinedby enzyme-linked immunosorbent assay (Zeptometrix). The levelof reverse transcriptase activity was quantified using the AmershamBiosciences Quan-T-RT assay system.

RNA isolation and analysis. The total cellular RNA and the extracellular viral RNA fractionswere purified using a Stratagene Absolutely RNA RT-PCR miniprepkit. To pellet the viral particles, a fraction of the mediumwas centrifuged for 2 hours at 4°C and 12,000 rpm. PurifiedRNAs were used for reverse transcription-PCR (AccuScript high-fidelityRT-PCR system; Stratagene) or an RNase protection assay (RPAIII; Ambion), as described by the manufacturer. Briefly, theRNase protection assay was done using an antisense RNA probecomplementary to the region of positions 401 to 562 of the HIV-2ROD isolate. The antisense region was cloned into the pGEM7Zf(+)vector (Novagen) so that the T7 transcript had around 45 nucleotidesof vector at its 5' end. The non-HIV-2 tail of the probe wasused as a marker of the RNase's digestion efficiency duringthe RNase protection experiment. Up to two protected bands wereexpected, with a 162-nt band corresponding to the 5' end ofthe unspliced genomic RNA and a 70-nt band corresponding toany spliced HIV-2 RNAs (see Fig. 3B). Furthermore, targetingof the region of positions 401 to 562 gives the same profilewith wild-type or pal-mutated viral RNAs.

View larger version (40K):
[in this window]
[in a new window]

FIG. 3. Infectivity and genomic RNA encapsidation are affected by alternative ${psi}$ (pal) mutations and restored through reversion-associated mutations. (A) Single-round infectivity assay with P4-CCR5 reporter cells of viruses produced by transfected COS-7 cells. The mutated viruses used in this experiment are shown in Fig. 2. The infectivity assay is described in Materials and Methods. The error bars represent the standard deviations for duplicate experiments. (B) (Top) RNase protection assay of wild-type (lanes 4), weak (lanes 5), and weak G394A/T397C mutant (lanes 6) pal viruses. The controls were the undigested probe (lane 1) and the digested probe used with in vitro RNA transcript 1-470 (lane 2) or 1-561 (lane 3). (Bottom) Diagram of the radioactive antisense probe used in this experiment and the expected protected fragments. (C) Encapsidation yields of different pal-mutated viruses. The amounts of unspliced genomic RNA in viruses isolated from the medium of transfected COS-7 cells were quantified using an RNase protection assay. The numbers were then normalized first to the amount of viruses in the medium, as determined by p27 capsid ELISA, and second to the level of encapsidation by wild-type viruses, set at 100%, to give the encapsidation relative to that of the wild type. The error bars represent the standard deviations for two or three experiments.

RESULTS

Top

Results

Identification of a conserved 10-nucleotide palindromic sequence in the core ${psi}$ region of HIV-2 leader RNA. The region located upstream of the HIV-2 SL1 element is ableto influence the SL1-mediated dimerization of leader RNA fragmentsin vitro (23). We therefore wanted to assess its functionalimportance in HIV-2 replication in cell culture. We deletedthe 10-nt palindromic sequence (pal) located upstream of SL1and in the downstream part of the encapsidation signal ${psi}$ characterizedby Lever and colleagues (Fig. 1A) (15). Infection of permissiveC8166 cells by both ${psi}$ -negative and pal-negative mutants did notlead to productive infection, suggesting that an important functionalelement overlaps with the 10-nt palindromic sequence (Fig. 1B).

Since deletion of the 10-nt pal greatly affected viral replication,we then analyzed the effects of substitutions that destroy thepalindromic nature of the pal element on viral replication (Fig.1C). Despite robust virus production after transfection in COS-7cells (data not shown), SCR1 and SCR2 mutants were defectivefor replication in permissive C8166 cells (Fig. 1C).

We chose to further test whether the palindromic nature of palmay play a role in virus replication in cell culture. We constructedtwo substitution mutants (Fig. 1D) that we call strong pal (5'-GGGCUGGCCC-3')and weak pal (5'-GGGAGUUCCC-3'). These sequences have been shownto form stronger or weaker RNA-RNA duplexes (compared to thewt sequence) in vitro (36). Infection of permissive C8166 cellsby strong and weak pal mutant viruses did not lead to productiveinfections relative to wild-type infections, with weak pal andstrong pal viruses showing low or no virus production (Fig.1D). These results indicate that the palindromic nature of palis perhaps less important than its primary sequence for HIV-2replication in cell culture.

Phenotypic reversion of alternative pal mutants in long-term culture. In order to see if the replication of the defective virusescould be enhanced by virus-induced mutations, duplicates ofthe SCR2, strong, and weak pal virus cultures shown in Fig.1C and 1D were passaged for several months. During the firstseveral weeks of culture, no cytopathic effects were observedfor any of the viruses. However, cell death and syncytia wereobserved at around 8 weeks postinfection for the weak pal culturesand at week 14 for one of the strong pal cultures. The supernatantsof these three cultures tested positive for both reverse transcriptaseactivity and the presence of the p27 capsid antigen. However,the SCR2 pal cultures never reverted, and neither viral proteinsnor viral RNA or DNA could be detected outside or inside thecells after 8 months of culture (data not shown).

View larger version (18K):
[in this window]
[in a new window]

FIG. 2. Two nucleotide changes that appeared during long-term culture are solely able to restore viral replication in alternative ${psi}$ (pal) viruses. (A) (Top) Sequences of the wild-type, weak, and weak G394A/T397C mutant pal 392-401 regions for the clones used in this experiment. The reversion-associated mutations are boxed, and nucleotides different from the wild-type sequence are shown with lowercase letters. (Bottom) Replication kinetics of wild-type (open squares), weak (open triangles), and weak G394A/T397C mutant (closed circles) pal viruses in C8166 cells. Viral replication was followed as described in the legend to Fig. 1B. (B) (Top) Sequences of the wild-type and mutated pal 392-401 regions for the clones used in this experiment. (Bottom) Replication kinetics of wild-type (open squares), strong (closed diamonds), strong G394A mutant (closed triangles), strong C400T mutant (open circles), and strong G394A/C400T mutant (closed circles) pal viruses in C8166 cells. Viral replication was followed as described in the legend to Fig. 1B.

Aliquots of infected cells were harvested at week 15 postinfection,and the region corresponding to the leader of the genomic RNAwas sequenced after RT-PCR to check the sequence variabilityin the pool of infected cells for each culture (Table 1). Aliquotsof supernatants were also harvested at week 26 and processedsimilarly. Despite the presence of T310 in all of the originalplasmid DNAs, all sequences showed a majority of T310C nucleotidechanges in the primer-binding site after extended culture (datanot shown), which indicates that a tRNA₃^Lys molecule was usedas the primer for reverse transcription, rather than tRNA₅^Lys(8).

Interestingly, the other nucleotide changes in the 5'-untranslatedregion were located in or close to the mutated pal region. Onereversion to a wild-type nucleotide was seen in all clones:the mutation A394G in the pal domain reverted back to A394 (Table1). Two other nucleotide changes were seen in or downstreamof the mutated pal region: C400 changed to T400 for the strongpal culture by week 15, and a partial or total change of G442to A442 occurred in the weak pal cultures between weeks 15 and26 (Table 1).

Effects of reversion-associated mutations in the weak and strong pal regions on viral replication. To test if the mutations we observed in the mutated pal regionsof the phenotypically reverted clones were solely responsiblefor the reversions, we cloned these changes into the originalweak or strong pal HIV-2 DNA backbone. We tested one of thethree different reversion-associated mutation sets identifiedby the sequencing of individual clones, namely, two nucleotidechanges in the weak pal region (G394A and T397C) (Table 1 andFig. 2A). The medium from the resulting COS-7 transfection wasused to infect C8166 cells, using a constant amount of p27 capsid,as measured by ELISA. The two nucleotide changes in the weakpal domain (G394A and T397C) increased the levels of viral replicationof the mutated weak pal viruses to close to the wild-type level(Fig. 2A). This conclusion held true for the two nucleotidechanges in the strong pal domain, i.e., G394A and C400T (Table1 and Fig. 2B). The changes increased the level of viral replicationof the strong pal G394A/C400T virus to close to the wild-typelevel (Fig. 2B). Interestingly, none of the single changes wasable to compensate for the defective replication of strong palviruses (Fig. 2B). Only when the G394A and C400T nucleotidechanges were present in the same genomic RNA were the mutatedviruses able to replicate at or close to the wild-type level(Fig. 2B). Thus, two nucleotide changes in the weak and strongpal domains were able to restore viral replication efficiency.

Effects of reversion-associated mutations in the weak and strong pal regions on genomic RNA encapsidation. We then sought to identify the stage(s) of HIV-2 replicationin C8166 cells responsible for the low replication levels ofthe weak and strong pal viruses and for the wt level of theweak and strong pal viruses containing the two reversion-associatednucleotide changes, i.e., G394A/T397C and G394A/C400T, respectively(Fig. 2). First, the phenotypes observed in C8166 cell cultureswere also observed when the viruses were tested on P4-CCR5 reportercells (Fig. 3A), underscoring the gravity of the defect(s).Second, we used an RNase protection assay to detect the levelof genomic RNA encapsidation, since the pal element is locatedin a sequence, ${psi}$ , that has been shown to play an important rolein encapsidation (15). The encapsidation of ${psi}$ -negative and pal-negativegenomic RNAs was similarly decreased threefold relative to thewt level (Fig. 3C). Indeed, the encapsidation level of weakpal genomic RNA in COS-7-produced virions was also decreasedtwo- to threefold (Fig. 3B and C). Importantly, the genomicRNA encapsidation level for the weak pal G394A/T397C virus wasat or close to the wild-type level (Fig. 3B and C). Similarresults were found for the strong pal and strong pal G394A/C400Tviruses (Fig. 3C). The single changes (G394A or C400T) partiallyincreased the encapsidation levels of the corresponding genomicRNA molecules, suggesting a cumulative effect of both nucleotidechanges when present in the same genomic RNA (Fig. 3C). Thus,two nucleotide changes in the weak and strong pal domains wereable to compensate for the drop in genomic RNA encapsidation.

DISCUSSION

Top

Discussion

An impetus for this study was derived from prior in vitro studiesof functional interactions in the HIV-2 leader RNA. We and othersshowed evidence that several short- and long-distance interactionsin the HIV-2 leader RNA influence the presentation of the SL1dimerization element in vitro (10, 24). Secondary structureprediction suggested a base-pairing interaction between the5' end of SL1 and a sequence immediately upstream, which couldbe responsible for the functional silencing of SL1-mediatedleader RNA dimerization (24). The nature of this silencer sequencewas interesting for two reasons. First, it is located in thedownstream part of the encapsidation signal ${psi}$ characterized byLever and colleagues (15). Second, it is a 10-nt palindromic(or autocomplementary) sequence which is phylogenetically conservedin HIV-2 and macaque and sooty mangabey SIVs (27).

In this work, we used long-term culture of defective virusesand their phenotypic reversion to study the functional importanceof the 10-nt ${psi}$ pal in HIV-2 replication. In particular, we showedthat the defects in viral replication for some pal-mutated virusescould be compensated for by two nucleotide changes in or closeto the mutated pal element. The simplest way to explain thereversion results is to suppose that the 3' end of the pal elementis involved in an extended stem-loop 1 structure. Furthermore,characterization of the mutant viruses suggests that the formationof this extended structure is necessary for HIV-2 replicationand affects genomic RNA encapsidation. The results also suggestthat the 5' end of the pal element plays a related but independentfunction to that of the 3' end in viral replication and encapsidation.Furthermore, the fact that pal-mutated viruses lacking a functionalpalindromic sequence can replicate as well as wild-type palviruses in C8166 cells suggests that the conservation of thepal element may be due more to its role in the formation ofan extended stem-loop 1 structure than to a strict requirementfor a palindrome.

The experiments presented here suggest a model for an essentialfunction of the 3' end of the pal element (GCUCC-3') in theformation of a stem with a 5'-GGAGC element located downstreamof SL1. We call the new stem extending the SL1 structure "stemB" because of its homology to that described for HIV-1 isolates(Fig. 4) (26, 43, 44). Indeed, the extended SL1 structure inHIV-2 was first proposed based on observed encapsidation andviral replication defects when larger deletions were made inthe ${psi}$ -SL1 region (32). Recent computational (21) and biochemical(7) analyses of the structure of the HIV-1 leader region furthersupported the existence of the stem B structure in SL1. Phylogeneticdata provide evidence for the existence and conservation ofstem B in HIV-1 (7, 21) and HIV-2 (Fig. 4D) (27).

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

FIG. 4. Secondary structure models of extended SL1 and stem B structures described in this work. (A) The HIV-2 ROD isolate (GenBank accession no. M15390) SL1 structure is represented with another stem structure that base pairs with the 3' end of the pal element with a 5'-GGAG-containing motif located immediately downstream of the short SL1 structure. We chose the names stem B and stems C1 and C2 to describe the extended HIV-2 SL1 in comparison with the structure of the HIV-1 SL1 element, as described by Laughrea and colleagues (26). (B) Mutations in the pal elements of strong pal viruses disrupt stem B (left structure). An alternative, more stable base pairing is possible by a one-nucleotide shift with a single mismatch in the middle (middle structure). The reversion-associated mutation C400T stabilizes the alternative stem B, and the G394A mutation restores a partial GGA motif (right structure). (C) Mutations in the pal elements of weak pal viruses decrease the stability of stem B (left structure). The reversion-associated mutation T397C or G442A stabilizes the mutated stem B by restoring Watson-Crick base pairing, while G394A reversion restores a partial GGA motif (right structure). The reversion-associated mutations are boxed, and nucleotides different from the wild-type sequence are shown with lowercase letters. (D) The consensus stem B structure from the analysis of 37 HIV-2 and SIV sequences from the HIV Sequence Compendium 2005 is indicated, with the proportions of alternative base pairs shown to the right (27). (E) Secondary structure model of HIV-1 SL1 from the HxB2 isolate (GenBank accession no. K03455). The stems are named according to reference 26.

The formation of stem B is an important step in HIV-2 replication.A less stable or nonexistent stem leads to a very low levelof replication (Table 2) ( ${Delta}$ G values for wild-type, weak, andstrong pal stems were –7.9, –3.7, and –4.1kcal/mol, respectively), and genomic RNA encapsidation appearsto be one of the stages affected by mutations in this region(Fig. 3C). The stem B-stabilizing reversion-associated mutationsoccurred at both wild-type nucleotides (Table 1) (strong palC400T and weak pal G442A mutants) and at one originally mutatednucleotide (Table 1) (weak pal T397C mutant). The strong palC400T reversion exchanged a CA for a UA base pair, increasingthe stem stability from –4.1 to –8.3 (Fig. 4B andTable 2), while the weak pal T397C reversion replaced a wobbleUG with a CG base pair, increasing the stem B stability from–3.7 to –6.2 (Fig. 4C and Table 2).

View this table:
[in this window]
[in a new window]

TABLE 2. Calculated ${Delta}$ G values (kcal/mol) for putative pal-pal and stem B structures for the different sequences described in this work^a

The importance of the 5' end of the pal element is underscoredby several points. First, most, but not all, pal-mutated viruseswith wt stability for stem B replicated at wt levels (compareFig. 1 and 2 to Table 2). Second, strong pal viruses bearingthe single C400T change have a stem B stability similar to thatof the wt, but they did not replicate in C8166 cells. They showedan increased encapsidation level relative to the strong palparent, but it was lower than the wt level (compare Fig. 2Band 3C and Table 2). Lastly, all of the phenotypically revertedviruses showed reversion of the G394 nucleotide back to thewild-type A394 nucleotide (Table 1). We hypothesize that the5' end of pal (5'-GGAG motif) is a purine-rich Gag polyproteinrecognition element, which could be functionally related tothe 5'-GGAG in HIV-1 SL3 (9). Indeed, purine-rich elements havea long history in retroviral RNA encapsidation and dimerizationresearch (for example, see references 30 and 47), although theirexact mode of action is not completely understood. In HIV-1,the purine-rich motifs in or downstream of SL3 play an importantrole in dimerization and encapsidation of the genomic RNA (40).

Remarkably, the phenotypic reversions we observed in our long-termculture experiments show that mutant viruses are viable withouta palindromic sequence (compare Fig. 2 and Table 2), but itis not known if the mutated viruses would be viable in animalinfections in vivo. The palindromic nature of pal is conservedin HIV-2 and SIV: 31 of 37 sequences examined show a T396-containingpal, whereas 6 sequences show an A396-containing pal (Fig. 4D)(27). The A396-containing pal isolates belong to HIV-2 typeAB or B or SIV_SMM (27). The presence of A396, while settingup an AG base pair, is not at odds with the stem B model, asAG base pairs are frequently found at the ends of helices, wherethey form "conformationally unique helix extensions" (12). Indeed,it is possible that an A396-containing stem B could be slightlymore stable than a U396-containing stem B, since the UG basepair in stem B is identified as a class B, "unstacked" pairwhich may not even base pair and, if it does, is probably notin a wobble conformation (13).

Although we cannot rule out that a pal-pal interaction betweentwo HIV-2 RNA molecules exists transiently during viral replication,we now view the palindromic nature of pal as a fortuitous eventin relation to the functions it fulfills: its 5'-GGAG(U) motifis likely to be a Gag polyprotein binding motif, while its GCUCC-3'motif base pairs with another 5'-GGAG(C) motif downstream ofSL1 to form stem B (Fig. 4A). The stem B of HIV-1 SL1 is alsoformed by a similar interaction between CUCG-3' and 5'-CGAGmotifs (compare stems B in Fig. 4A and E).

Finally, the finding of stem B extending the SL1 structure inHIV-2 genomic RNA brings both answers and exciting questions.The next endeavor will be to test if HIV-2 stem B is also involvedin genomic RNA dimerization and reverse transcription, two functionsthat are naturally linked to this region (44). It will be equallyintriguing to ascertain whether the formation of the extendedSL1 during HIV-2 replication affects (or is affected by) thetranslation of the gag gene, since the absence or presence ofthe long-distance interaction involving the gag initiation codonseems to influence the presentation of the pal-SL1 domain invitro (10, 24).

ACKNOWLEDGMENTS

This work was funded by National Institutes of Health grantAI45388 to J.S.L.

Anti-HIV-2 serum and C8166 cells were obtained through the AIDSResearch and Reference Reagent Program, Division of AIDS, NationalInstitutes of Health (no. 1030 and 404, respectively). The plasmidpROD10 was provided by the EU Programme EVA/MRC CentralisedFacility for AIDS Reagents, NIBSC, United Kingdom (grant numbersQLK2-CT-1999-00609 and GP828102). We thank Kathryn Follis andJack Nunberg from the University of Montana for assistance withthe virological experiments. We also thank Tayyba Baig for criticallyreading the manuscript.

FOOTNOTES

* Corresponding author. Mailing address: Division of Biological Sciences, The University of Montana, Missoula, MT 59812. Phone: (406) 243-6393. Fax: (406) 243-4304. E-mail: stephen.lodmell{at}umontana.edu.

Published ahead of print on 17 January 2007.

REFERENCES

Top