Shortening of the Symptom-Free Period in Rhesus Macaques Is Associated with Decreasing Nonsynonymous Variation in the env Variable Regions of Simian Immunodeficiency Virus SIVsm during Passage

Abstract

During six blood passages of simian immunodeficiency virus SIVsm in rhesus macaques, the asymptomatic period shortened from 18 months to 1 month. To study SIVsm envelope gene (env) evolution during passage in rhesus macaques, the C1 to CD4 binding regions of multiple clones were sequenced at seroconversion and again at death. The env variation found during adaptation was almost completely confined to the variable regions. Intrasample sequence variation among clones at seroconversion was lower than the variation among clones at death. Intrasample variation among clones from a single time point as well as intersample variation decreased during the passage. In the variable regions, the mean number of intrasample nonsynonymous nucleotide substitutions decreased from the first passage (5.26 × 10⁻² ± 0.6 × 10⁻² per site) to the fifth passage (2.24 × 10⁻² ± 0.4 × 10⁻² per site), whereas in the constant regions, the mean number of intrasample nonsynonymous nucleotide substitutions differed less between the first and fifth passages (1.14 × 10⁻² ± 0.27 × 10⁻² and 0.80 × 10⁻² ± 0.24 × 10⁻² per site). Shortening of the asymptomatic period coincided with a rise in the Ks/Ka ratio (ratio between the number of synonymous [Ks] and the number of nonsynonymous [Ka] substitutions) from 1.080 in passage one to 1.428 in passage five and mimicked the difference seen in the intrahost evolution between asymptomatic and fast-progressing individuals infected with human immunodeficiency virus type 1. The distribution of nonsynonymous substitutions was biphasic, with most of the adaptation of env variable regions occurring in the first three passages. This phase, in which the symptom-free period fell to 4 months, was followed by a plateau phase of apparently reduced adaptation. Analysis of codon usage revealed decreased codon redundancy in the variable regions. Overall, the results suggested a biphasic pattern of adaptation and evolution, with extremely rapid selection in the first three passages followed by an equilibrium or stabilization of the variation between env clones at different time points in passages four to six.

The discovery of lymphomas in macaques previously housed with sooty mangabeys or African green monkeys (7, 16, 18, 22) led to the isolation of the first simian immunodeficiency virus (SIV) isolates and characterization of the wasting diseases that they caused in macaques. Later, it was found that African feral monkeys were commonly SIV infected (22%) (14, 24, 30) and that SIV infection was endemic in sooty mangabeys housed at some centers. At the Tulane Regional Primate Research Center, tissue-derived inoculum from a sooty mangabey was used to inoculate several macaques intravenously; one died of a wasting syndrome 18 months later (rhesus macaque B670) (7). The transmission of this mangabey virus (SIVsm) to Asian macaques resulted in an infection characterized by a loss of CD4⁺ T cells, persistent serum antigenemia, and trapping of virions in the follicular dendritic cell foot processes (6), all hallmarks of human immunodeficiency virus (HIV) infections.

Studies with molecular clones have shown that single nucleotide substitutions in env of HIV and SIV can cause changes in the biological phenotype, neutralizing antibody escape, and growth kinetics (1, 15, 25). The env nucleotide substitutions seen during HIV infection are concentrated in the variable regions. The predominance of nonsynonymous over synonymous substitutions is believed to be due to immune pressure (5, 11) on various viral proteins as well as on env. During cross-species transmission of a lentivirus, adaptation occurs for accommodation to the newly encountered nonhost environment. This process allows the viral proteins needed in the infective processes to adapt to the cellular composition of the new host.

The sequencing of multiple env clones at a particular time point in infection gives an approximation of the quasispecies present in the blood at that time. A comparison of the variation within these sequences gives an idea of the relative intrasample variation taking place at that time in the gene sequenced (env). An inverse correlation between virus variation and length of the immunocompetent period has been shown for asymptomatic carriers of HIV and for individuals progressing to AIDS (31). Progression to AIDS following HIV infection is known to be load dependent, with the more rapidly replicating and syncytium-inducing phenotypes being the most efficient in the pathway of events leading to immunodeficiency (26, 39).

To study the relationship between viral variation and length of the asymptomatic period, we passaged SIVsm in Asian macaques. The cross-species transmission was carried out with the Delta B670 SIV strain (7) as the primary inoculum followed by four serial intravenous inoculations with peripheral blood mononuclear cells (PBMC) taken from animals at the symptomatic stage of infection. The viral quasispecies present in the monkeys at seroconversion and death were sampled, and multiple env clones were sequenced. We report here on the shortening of the asymptomatic period from 18 months to a few weeks and the concomitant reduction of intrasample and intersample variations. Finally, we observed that the rate of nonsynonymous nucleotide substitutions during env adaptation of SIVsm in rhesus macaques was variable and that the changes seen occurred almost entirely in the env variable regions, where limited codon usage was found.

MATERIALS AND METHODS

Virus.

The Delta B670 SIV strain (3, 6, 7, 9, 31a, 33, 41) is an SIV originally found in a sooty mangabey presenting with a cutaneous lepromatous lesion. When tissue from this animal was used to infect rhesus macaques, it closely reproduced AIDS in humans (33). The virus inoculum was proven to be free of type D retrovirus, which can cause such symptoms as well (28).

Passage.

A full history of the passage is currently being prepared (unpublished data). Briefly, six Asian rhesus macaques, all 2 years of age, were used for the purpose of experimental infection with an SIVsm strain. The first virus sampled (P1) was the Delta B670 SIV stock from the macaque inoculated with the sooty mangabey virus (33). The second monkey (P2) was infected intravenously with 5 × 10² infectious doses of the Delta B670 SIV strain. The monkeys then were intravenously inoculated in a serial fashion with 2 × 10⁶ PBMC taken at the symptomatic stage. Passage five symptomatic-stage PBMC were used to infect two monkeys (passages six A and six B). The time to death postinfection (tdpi) and moment of sampling (ms) for the animals were as follows: P1—tdpi, 18 months, and ms, 18 months; P2—tdpi, 12 months, and ms, 3 and 7 months; P3—tdpi, 9 months, and ms, 2 and 4.1 months; P4—tdpi, 4 months, and ms, 1 and 1.8 months; P5—tdpi, 2 months, and ms, not done, as no sample was available; P6A—tdpi, 2 months, and ms, 2 months; and P6B—tdpi, 2 months, and ms, 2 months. Animals were euthanatized upon evidence of undue suffering. PBMC used for serial passages were not cultured or cocultivated.

RNA isolation and RT-PCR.

Viral RNA was harvested with silica in the presence of a chaotropic agent (10) from the sera of experimentally infected Asian macaques and was used as a template in a reverse transcriptase (RT) PCR (RT-PCR). Viral RNA was isolated from 20-μl volumes of sera, resuspended in 20 μl of RNasin-containing H₂O (1 U/μl), and used in an RT reaction consisting of a mixture of 5 μl of viral RNA, 250 μM each deoxynucleoside triphosphate (dNTP), 2 ng of 3′ RT-PCR primer (SIV4Not1: TTATATGCGGCCGCCTACTTTGTGCCACGTGTTG) per μl, 2.5 mM Mg²⁺, 1 U of RNasin (Promega) per μl, 10 U of Super Script I (Gibco-BRL), and 1× reaction buffer (37) in a 20-μl volume. The components were assembled at 37°C and incubated at that temperature for 90 min. The PCR mixture consisted of 250 μM each dNTP, 2 ng of 3′ and 5′ (SIV1HIII: GTAGACAAGCTTGGGATAATACAGTCACAGAAC) PCR primers per μl, 1× reaction buffer, 2.25 mM Mg²⁺, and 1.5 U of Taq polymerase (Perkin-Elmer Cetus) in a final volume of 100 μl including 5 μl of RT reaction mixture. The PCR mixture was overlaid with paraffin and heated to 95°C for 5 min followed by 35 cycles of 1 min at 95°C, 1 min at 55°C followed by 1 min at 72°C, and finally 10 min at 72°C in a Perkin-Elmer Cetus DNA Thermocycler. RT reactions and PCRs were carried out in duplicate for each sample to prevent mispriming and to preserve the fidelity of the virus genotypes sampled.

Samples were combined after PCR and size selected on 0.8% agarose gels followed by excision of the 1,151-bp band. The excised band was isolated from the gel slice and digested with NotI and HindIII, followed by agarose gel and silica gel fragment isolation (10). The size-selected, digested, purified RT-PCR product was ligated overnight into plasmid pSP64 (Promega) containing a NotI site. The ligated product was electroporated into electrocompetent Escherichia coli C600, and plasmid DNA from sequencing was isolated with Qiagen columns.

Sequencing and analysis.

Double-stranded plasmid DNA was sequenced with custom labelled dye primers (ABI, Foster City, Calif.) by use of an ABI model 373A automated sequencer and version 1.2.0 software. Clones were assembled and aligned with the Sequence Navigator program (ABI). Nucleotide sequences were aligned with Sequence Navigator and Clustal V (21), with final adjustments being carried out visually. All positions with an alignment gap in at least one sequence were excluded from any pairwise sequence comparisons. p distances, defined as the number of synonymous or nonsynonymous substitutions divided by the total number of synonymous or nonsynonymous sites (34), were used to measure the relative genetic variation between clones. Synonymous and nonsynonymous nucleotide p distances (Ks and Ka, respectively) were calculated with the MEGA program (27). Intrasample calculations are the result of comparisons of clones from the same time point or quasispecies (within a sample); intersample calculations are the result of comparisons of clones one by one from two different time points or quasispecies (between two samples). Samples were named for their passage (P) position (from 1 to 6) and for the time of sampling, either at seroconversion (S) or death (D); e.g., P2S is the passage two seroconversion sample. P6A and P6B were considered one sample (P6) in the data calculations since there was no significant difference between them.

RESULTS

Intrasample genetic variation at seroconversion was lower than that at death.

Declining genetic variation, as measured by p distances, was observed during the first three passages, coincident with a decrease in the sequence variation within the quasispecies (Fig. 1B). The values for the last three passages were not significantly different (0.0155 ± 0.0090, 0.0140 ± 0.0065, and 0.0154 ± 0.0092) and indicated that further adaptation to the host did not occur. The intrasample variation was higher at death than at seroconversion in all passages (Fig. 1A). The intersample variation decreased gradually from 0.0423 ± 0.015 to 0.0185 ± 0.0110 during the first three passages and then stabilized. The number of intrasample variations was one half the number of intersample variations, suggesting continuing replacement of genotypes (high interpassage p distances). Although the intersample variations decreased gradually during the first three passages, they remained higher than the intrasample variations.

FIG. 1.

Intrasample and intersample variations. Intrasample (A) and intersample (B) p distances are shown. Groups of five clones per sample were compared to each other (intrasample) or to the five clones in another sample (intersample). The resulting numbers are plotted against the position of the sample along the passage, from one to six, at seroconversion (S) and death (D). Samples without letters were at death. The symbols (○ and ▵) denote the means of all values, and the variance is shown by the vertical bars.

Nonsynonymous variations decreased during the first three passages and subsequently stabilized.

Synonymous and nonsynonymous substitutions in passage one were almost equal in number, suggestive of a heterogeneous founder population adapting during the introduction of SIVsm to rhesus macaques (Fig. 2A and B). The intrasample nonsynonymous variation decreased threefold in the first three passages (from 0.0325 ± 0.0065 to 0.0110 ± 0.0004). In passages three to six, the synonymous variation was twice the nonsynonymous variation. The intrasample variation was lower at seroconversion than at death during the first two passages and to a lesser extent during the last two passages. The lowest values for synonymous and nonsynonymous variations were at seroconversion in the third passage, with the rate of adaptation leveling off in the following passages. Intersample nonsynonymous and synonymous variations were higher than intrasample variations (Fig. 3A and B), reflecting the adaptation that occurred over time during the infection (intrahost) and between the successive passages (interhost).

FIG. 2.

Intrasample synonymous and nonsynonymous variations. Intrasample nonsynonymous (A) and synonymous (B) p differences are shown. The five clones of a sample were compared to one another individually, and the synonymous and nonsynonymous p distances were calculated. The symbols (○ and ▵) denote the means of all values, and the variance is shown by the vertical bars.

FIG. 3.

Intersample synonymous and nonsynonymous variations. Intersample nonsynonymous (A) and synonymous (B) p differences are shown. The five clones of a sample were compared to those of the next sample individually, and the synonymous and nonsynonymous p distances were calculated. The symbols (○ and ▵) denote the means of all values, and the variance is shown by the vertical bars.

The Ks and Ka values for the env variable and constant regions followed dissimilar trends.

Figure 4 shows the Ks and Ka values for the variable and constant regions. The Ka values for the variable (Fig. 4A) and the constant (Fig. 4B) regions decreased rapidly from P1 to P3S, when the Ks values increased. The Ks values for the variable and constant regions were similar from P1 to P3S, when the Ka values were more than fourfold higher for the variable regions than for the constant regions. At seroconversion in passage three, there was almost no nonsynonymous variation in the constant regions, when the Ks values were similar (variable, 0.0144; constant, 0.0133) and the Ka values for the variable regions were 13 times the Ka values for the constant regions (0.0187 and 0.0014). Variable-region Ka values from P3S onward remained fourfold those for the constant regions. The Ks values for the variable regions rose while those for the constant regions remained level in the last two passages. The Ks/Ka ratios for the constant regions were always greater than one, while those for the variable regions were less than one in five of the eight samples (Fig. 4A and B).

FIG. 4.

Intrasample variations of constant and variable regions. Ks (synonymous) and Ka (nonsynonymous) substitutions per synonymous or nonsynonymous site were calculated by dividing the C1 to CD4 binding region clones into variable (A) and constant (B) regions and comparing these selected regions one at a time among the clones of each sample.

The variable regions of the envelope gene had a decreased codon redundancy.

The alterations in Ks and Ka values led us to examine the nature of the substitutions and changes in nucleotide concentrations of the variable and constant regions of env (Fig. 5A to C). Differences found in the envelope sequences of SIVsm were compared to SIVmac251 (29) and to 55 HIV type 1 (HIV-1) subtype B gp120 sequences (Los Alamos Sequence Database). Figure 5 shows that env was rich in adenine (A), with increased concentrations in the variable regions compared to the constant regions. The ratios of the four nucleotides remained constant throughout the passages and within the HIV sequences. It has been reported that nucleotide composition affects the evolution of RNA viruses (12) and that individual genes of retroviruses have characteristic base compositions (8). We found that individual regions within retroviral genes, at least env, had characteristic base compositions. The A concentration in the variable regions was higher than that in the constant regions, except at the first nucleotide of SIVmac251 and the third nucleotide of HIV-1 subtype B (Fig. 5A and C). The greatest difference seen in nucleotide concentration was in HIV env, with its 20% discordance between the variable and constant regions. Nucleotides in position 1 of the SIV strains were almost equal between the two regions. In the SIV strains, the mostly synonymous third nucleotide showed higher A levels in the variable regions; HIV-1 subtype B displayed the opposite pattern.

FIG. 5.

Envelope nucleotide concentrations and envelope codon usage. Nucleotide concentrations of the variable and constant regions were calculated for the three positions of each codon (A, B, and C) in env. The numbers indicate the average for the sums of all clones (SIVmac251 [29] and HIV-1 subtype B data were from the Los Alamos Sequence Database). Variation among clones was less than 0.2%. (D) By use of the MEGA program, the numbers of codons used to represent amino acids in the variable and constant regions were calculated for the above-mentioned sequences. The percentage of 61 codons is the number of codons used divided by 61 (the total number of codons, not including those that represent stop signals).

The effects of the nucleotide concentrations found were examined with regard to codon usage within the two env domains, constant and variable. Although the nucleotide concentrations varied between viruses, the codon usage frequencies were similar (Fig. 5D). Of the 61 possible codons (there are 64 possible codons with three encoding stop messages not found within the envelope gene), the number used was 11% lower (on average) in the variable regions.

DISCUSSION

The endemic infection of feral monkeys with SIV has no known associated pathology, apparently due to the historic genetic accommodation (13) of the virus. The host adaptation of the virus, or host-pathogen coevolution, leads to asymptomatic infection of most African primate species with their own host-adapted strains (2, 23). No natural SIV infection has been found in Asian primates, and cross-species transmission (e.g., African SIV in Asian macaques) leads to the development of AIDS presenting the common markers of HIV-1 infection (20, 36). A cross-species population passage may cause initially rapid genomic evolution during adaptation to the new immune environment. Large numbers of nonsynonymous changes are due to the env alterations needed to avoid immune attack and/or to maximize the affinity of the viral env protein for efficient binding to and infection of the cells of the new host species. Since natural cross-species transmission occurs via blood contact, rather than by mucosal contact, intravenous serial passage was carried out. Parenteral transmission is one of the major routes of infections with HIV-1 and HIV-2, which is closely related to SIVsm. Recently, it was shown that the route of transmission (mucosal versus blood) does affect virus heterogeneity (4, 34a, 38a), both routes resulting in the transmission of a more homogeneous selection of variants. Similar observations were made by Amadee et al. (using the same strain, Delta B670 SIV, as was used here [3, 4]), who also showed that intravenous, oral, and transplacental transmissions limit virus heterogeneity and select for macrophage tropism, as we found for intravenous passage (39a).

The decrease in the Ka values after the first three passages suggests that the adaptation of SIVsm to rhesus macaques is rapid and punctuated. The decreases in the times to death postinfection and env adaptation follow similar patterns, with the intrasample sequence variation levelling off once the time to death postinfection is less than 4 months (passages four to six A and six B) (Fig. 4A). The decreased intrasample variation remains stable after the third passage, indicating that env variation is restricted, most likely because of the adaptive equilibrium reached. The nearly 50% decrease in env variation indicates that the total adaptation rate is decreasing and that there is positive or purifying selection of a narrow assortment of env genotypes. The lowest Ks value is that for P3S, with almost zero variation of the constant regions and a Ka of 10.0 for the variable regions, since an almost clonal population was present.

The low variation in seroconversion samples compared to death samples is evidence of a strong founder effect, since this outgrowth of a dominant homogeneous viral population emerges prior to detectable immune responses. This finding is in accord with the variation seen in env of HIV-infected individuals progressing to AIDS, in whom homogeneity at seroconversion is higher than that at death (40). The hypothesis that quasispecies homogeneity in HIV infections is caused by single-particle transmissions is not confirmed here, since even after the passage of millions of virus particles, we saw the same patterns of homogeneity at seroconversion and heterogeneity at death. The effect is thus caused by the selective amplification and eventual outgrowth of a dominant viral genome that then diversifies into a quasispecies under the influences of replication competence during adaptation, the immune system, and cell availability.

The fourfold higher Ka values for the variable regions than for the constant regions indicate positive selection (32). Similarity in Ks values and discontinuity in Ka values underscore the difference in the functions of the constant and variable regions and the system governing nucleotide substitutions. If the rate of nucleotide substitutions caused by RT errors is constant over all the nucleotides of a retroviral gene, then there should be equal amounts of variation in all regions. The distinct variations in Ka values reported here are presumably the effects of immune pressure, variations in receptor and coreceptor binding sites on the variable regions, and purifying selection on the constant regions. The similarity of the Ks values for the constant and variable regions reflects the effects of purifying selection on nucleotide substitutions in conserved regions.

Increasing the concentration of a single nucleotide may affect the number of possible codons that can encode amino acids. Amino acids with more than four representative codons (leucine, serine, and arginine have six) can be represented by codons with various nucleotide concentrations, as opposed to methionine and tryptophan, which have single codons. Thus, the result of an increased concentration of one nucleotide is that fewer codons can encode these amino acids when they are present in the variable regions, where the highest adenine (A) concentrations are found (Fig. 5D). Although this concentration effect is only possible with these three amino acids, there is a resulting reduction in the number of codons used in encoding the other amino acids as well. As shown in Fig. 5, the redundancy in the codons used in the variable regions is decreased up to 12% compared to that in the constant regions. Reduced codon usage will lower the redundancy of the encoded amino acids. Presumably, the net effect is greater amino acid variation from a given nucleotide substitution in the variable regions than in the constant regions, which are more buffered by increased codon redundancy. This strategy of reduced redundancy would cause more amino acid substitutions in the variable regions as a selective advantage against humoral and cell-mediated immunities while allowing the constant regions to remain stable during the adaptation reported here.

Reaching a plateau in the evolutionary pace in passages three to six suggests that sufficient env adaptation has occurred for optimal fitness. Exponential gains in the growth kinetics of RNA viruses (35) occur in vitro during a more prolonged series of passages. The short adaptation phase seen here could result from immune system-driven positive selection, since the previous experiments were performed with tissue culture under no such selection. This short adaptation phase could also be and probably is just as likely caused by selection for viruses with greater replication competence in the new species, and not immune selection alone. With the increase in pathogenicity, the shortening of the asymptomatic period falls below the response time for humoral immunity. This effect may lead to decreased adaptation, or evolutionary stasis, of env in the plateau phase from passages four to six. If so, it appears that env gene substitution is selected for by immunocompetence; thereafter, selective amplification and purifying selection control the breadth and direction of variation. Because env is the target for neutralizing antibodies and cell-mediated immunity, this finding is significant because env contains determinants for cell tropism and replication, thus playing a putative role in virulence.

In the first passage, nonsynonymous variations are almost equal to synonymous variations. Since these adaptive variations are not selective, they must be occurring in a very large and heterogeneous quantity of virus to produce the variations in noncoding nucleotides. Decreasing adaptation then continues to the seroconversion in passage three, at which point the lowest nonsynonymous and synonymous variations are seen. According to the competition exclusion principle (17, 19), equilibrium and competition lead to the outgrowth of a very homogeneous population, which we saw at seroconversion in passage three. The P3S quasispecies is dominated by a virus or viruses with a narrowed selection of genomes compared to that seen at passage one. The drop in evolution rate thus signals the end of the adaptation phase and allows competition among the viruses then present, which are of reasonably equal fitnesses. Their competition to achieve the greatest replication competence decides the dominance of the next progeny, not adaptation or immune evasion. This point marks the end of rapid env evolution and of large virulence increases as well.

Shpaer and Mullins have shown that immunogenicity and pathogenicity are presumably linked (38) by the correlation of high rates of amino acid change with increased virulence. The cross-species transmission of SIVsm into rhesus macaques would require the adaptation to the new host of the proteins that are involved in cell binding, entry, replication, immune evasion, and escape. The burst of evolution seen in the first three passages is evidence of this positive selection and its enforcement by the immune response to a foreign pathogen (32), as well as the cell-dependent changes needed for entry. The increased intrasample genetic variation seen between seroconversion and death confirms the notion that competent immune responses drive viral evolution to a certain extent. That nonsynonymous variations are greater than synonymous variations represents positive selection, and not drift, and the adaptation phase of the first three passages is exemplary evidence of this fact. The plateau phase of the later passages is illustrative of the purifying selection that takes place during lentiviral infections. In the absence of an antibody response, the evolutionary rate of the adapted replication-competent virus is close to stasis. The results reported here present suggestive evidence for the relationship of particular env sequences to virulence in rhesus macaques following inoculation with SIVsm.