Natural Infection of a Homozygous Δ24 CCR5 Red-capped Mangabey with an R2b-Tropic Simian Immunodeficiency Virus

The preparation of virus stocks HIV-1 JR-FL, HIV-1 Z00979, HIV-2 A198411, HIV-2 A227011, SIVmac239, SIVcpzGab1, SIVmac1A11, and SIVagmTYO1 has been described previously (25, 31). SIVrcmGab1 was obtained from RCM-009, a naturally infected household pet in Gabon (9). HIV-2 A198411 and HIV-2 A227011 strains were provided by Dr. Beatrice Hahn (36) and were primary isolates from West Africa that were propagated and titered in human PBMCs in our laboratory as previously described (31). Luciferase reporter viruses were prepared by cotransfecting 293T cells with pNL-Luc-Env⁻ or pSIV-Luc-R⁻E⁻ and vectors expressing different SIV or HIV-1 Envs, as previously described (16, 25).

All animal care and use was approved by the institutional animal care committees.

RCM PBMCs were prepared by Ficoll gradient separation, as previously described (7). Human osteosarcoma HOS.CD4 and glioma U87.CD4 cells, which stably express human and non-human primate chemokine receptors, have been described previously (25). GHOST.CD4 cl.34 cells are human osteosarcoma cells expressing human CD4 and containing the green fluorescence protein (GFP) gene controlled by the HIV-2 LTR promoter as the indicator of lentivirus infection. GHOST.CD4.Hu-BOB and GHOST.CD4.Hu-Bonzo cells, which express human coreceptors BOB (GPR-15) and Bonzo (STRL33), were provided by V.N. KewalRamani of New York University Medical Center.

Fecal specimens from free-living mangabeys in Gabon were collected fresh by spreading plastic sheets under a tree being used for sleeping by a wild mangabey troop. The mangabeys were observed through binoculars in late afternoon, to verify the species of the individual depositing the specimen. The specimen was collected immediately after it dropped to the plastic. In the field, 35 ml PBS was added to 10 g of fresh feces and suspended by shaking and vortexing. The fecal suspension was layered onto 15 ml of lymphocyte separation medium (Organon-Teknika, Durham, NC). Using a portable centrifuge, the mixture was spun at 3,000 rpm for 20 min. Cells were harvested at interface, in a volume of ∼15 ml. If the cell layer was diffuse after a single centrifugation, medium above and below the interface was collected. The ficoll separation step was repeated up to five times, using 3.0 ml of lymphocyte separation medium, until the cell layer was clearly defined. Cells were suspended in 10 ml PBS, washed twice, and stored in a portable liquid nitrogen tank.

The full length RCM-CCR5 was amplified as previously reported (25) for rhesus CCR5, using primers that flank the entire CCR5 coding region. PCR primers that hybridized 5′ and 3′ to a ccr5 region flanking the Δ24 deletion were as follows: 5′ primer (nucleotides 318–343 in the CCR5 coding region): 5′-GCT CTA TTT TAT AGG CTT CTT CTC TG-3′, and 3′ primer (510–530): 5′-GTG TAA TGA AGA CCT TCT CTC TGA GAT CTG-3′. Using this primer pair, 189 or 213 bp ccr5-specific fragments with and without the 24-bp deletion were amplified from genomic DNA (0.5 μg). The PCR reaction was according to the manufacturer's specifications (Boehringer Mannheim, Indianapolis, IN) in 100 μl containing 10 mM Tris-HCl (pH 8.5), 50 mM KCl, 1.5 mM MgCl₂, 0.1% Triton X-100, 200 μM each of dATP, dGTP, dCTP, and dTTP, 20 pmol of each primer, and 0.5 U of DNA Taq polymerase (Boehringer Mannheim). Amplification cycles were 95°C for 2 min followed by 35 cycles of 95°C for 15 s, 55°C for 20 s, and 72°C for 20 s plus the last extension of 72°C for 10 min. RNA was extracted from cells using Triazol (GIBCO BRL, Gaithersburg, MD) and treated with 10 U RNase-free DNase (Promega Corp., Madison, WI). cDNA was prepared from 5 μg RNA using an oligo-dT primer Superscript reverse transcriptase (GIBCO BRL), and resuspended in 80-μl 10 mM Tris, pH 8.0, 1 mM EDTA (TE). The cDNA was amplified over 35 cycles using the primer pair specific for the ccr5 fragment flanking the Δ24 deletion as described above.

The procedure for constructing expression plasmids was the same as that described for rhesus CCR5 and CXCR4 (25). In brief, virus stocks for the retroviral vectors expressing CCR5 and CXCR4, pBabe/CCR5, and pBabe/CXCR4, respectively, were prepared by cotransfecting 293T cells with the murine leukemia virus (MuLV)-gag/pol vectors pSV-psi-MLV-env⁻ and pcVSV-G. Virus was harvested 48 h after transfection and frozen at −70°C. 293, HOS.CD4, and U87.CD4 cells were infected at the same time with 2 ml of the pBabe virus and were selected 48 h later in culture medium containing 1.0 μg/ml puromycin (Sigma Chemical Co., St. Louis, MO). Puromycin-resistant cells were then used for HIV-1 and SIV infection with replication-competent or luciferase reporter viruses.

To test for viral entry, 3 × 10⁴ cells were seeded in 24-well plates in culture medium and infected with luciferase reporter virus (20 ng p24 or p27) in a total volume of 300 μl containing 1 ng/ml polybrene. After incubation overnight, 1 ml of medium was replaced in each well. After three days at 37°C, 100-μl lysates were prepared and luciferase activity in 20 μl was measured using commercially available reagents (Promega Corp.). For GFP assay, 2 × 10⁴ GHOST cells expressing coreceptors were seeded into 24-well plates. After 24 h, 5 ng p27 of each SIV was inoculated for 12 h with 20 μg/ml polybrene. Input virus was removed by washing, and cells were maintained in DMEM medium with 10% FCS. GFP expression was detected by FACS^® analysis (Becton Dickinson, San Jose, CA) 48 h after infection.

To determine coreceptor usage of different viruses, 2 × 10⁴ HOS.CD4 or U87.CD4 cells stably expressing human or rhesus coreceptors were seeded in 24-well plates. The next day the adherent cells were washed with PBS and infected with 5 ng p27 of SIV or 500 TCID₅₀ of HIV-1. After 12 h, virus was removed by replacing the medium three times. The cells were maintained in medium with 2% FCS for up to 10 d. Virus production was measured by harvesting aliquots of each supernatant and measuring p27 (Coulter, Chicago, IL) or p24 (Abbot, Chicago, IL) by ELISA. After removing 0.5 ml of culture supernatant per well for assay, the same amount of fresh medium was added to the culture.

The assay was performed as previously described (37). In brief, 293 cells expressing RCM-CCR5 or RCM-Δ24-CCR5 were harvested from tissue culture plates using a cell scraper. The cells were washed with tissue culture medium before resuspending them at 10⁶  cells/ml in HBSS containing 0.5% (wt/vol) BSA, 4 mM probenicid (Sigma Chemical Co.) and 4 μg/ml Fluo-3 AM, 10 μg/ml Fura-Red, and 0.02% Plurionic F127 (all from Molecular Probes, Eugene, OR). The cells were incubated at 37°C for 30 min then centrifuged at 1,500 rpm for 5 min before being gently resuspended at 10⁶ cells/ml HBSS containing 0.5% BSA. The cells were left at room temperature for 20 min. All analyses were performed using a FACSCalibur^® flow cytometer with Cellquest software (Becton Dickinson). Equal amounts of the cells were removed, further diluted to 10⁶ per ml in HBSS containing 0.5% BSA and warmed to 37°C for 5 min. The sample temperature was maintained throughout the assay by a water jacket at 37°C. Data were collected over a 3-min period 20 s after adding 200 nM β-chemokines, MIP-1β, or RANTES. A ratio of Fluo-3/Fura-red fluorescence was calculated using FACS Assistant software (Ray Hicks, Cambridge University, Cambridge, UK). Data are displayed as a density plot of time verses fluorescence ratio and also as a line graph of time versus mean fluorescence ratio, which was calculated using FACS Assistant Software.

The complete sequence of RCM-CCR5 was obtained using an automated sequencer (ABI Prism 377 DNA Sequencer; PE Applied Biosystems, Foster City, CA). At least two molecular clones were sequenced from two animals using overlapping primers. The RCM-CCR5 sequence are available from EMBL/GenBank/DDBJ under accession numbers AF084002–AF084004 and AF094752–AF094753.

SIVsm, SIVagm, and SIVcpz are R5-tropic, divergent SIVs found in East, West, and Central Africa (25–27, 30, 31). Because all naturally occurring SIVs described thus far are R5-tropic, the coreceptor use of the new virus, SIVrcmGab1, was first tested on GHOST cells expressing CD4 plus CCR5 or CXCR4 derived from humans and macaques. Unexpectedly, no entry was detected (data not shown). Known HIV-1 and SIV strains were included to demonstrate the appropriate coreceptor function of each cell-type. To further examine SIVrcm coreceptors, human chemokine receptors CCR1, CCR2b, CCR3, CCR4, CCR5, CCR8, CXCR4, STRL33 (Bonzo), and GPR15 (BOB) were tested in the GHOST cell assay (Fig. 1). Among the SIV and HIV strains tested, only SIVrcm entered CCR2b-expressing cells (Fig. 1). This entry was blocked by anti-CD4 antibody Leu-3a (data not shown), indicating CD4 dependency. SIVmac, SIVagm, HIV-1JR-FL, and HIV-1ADA, primate lentiviruses of known coreceptor useage, were included to demonstrate appropriate activity of the cell lines (Fig. 1). SIVrcm also used STRL33 (Bonzo), but with ∼10-fold less efficiency. However, low STRL33 expression could not be excluded as the cause of inefficient STRL33 use, because specific antibodies for testing STRL33 cell surface expression are unavailable. The remaining chemokine receptors, including GPR-15 (BOB), were not used by SIVrcm.

To understand why SIVrcm was uniquely R2b-tropic, but not R5-tropic, among widely divergent HIVs and SIVs, full-length RCM-CCR5 was amplified from the DNA of SIV-infected RCM-009 and completely sequenced as previously described (25). The RCM-009 CCR5 sequence was aligned and compared with the SM consensus amino acid sequence (31), which had been derived from four SMs naturally infected with R5-tropic SIVsm in West Africa (31, 38). The RCM sequence differed from SIVsm-infected SM CCR5 genes at only three amino acid positions, I42V, L146V and P179S. However, a homozygous 24-bp deletion in the CCR5 was identified from the SIV infected RCM-009 (Fig. 2). The deletion encompassed 8 amino acids in frame in the fourth putative transmembrane region, upstream from the Δ32 deletion found in the second extracellular loop of human CCR5 (32). To test for spurious PCR deletion artifacts, five additional CCR5 clones from RCM-009 were sequenced and had the identical deletion (data not shown). Moreover, a limiting dilution of the original genomic DNA of RCM-009 showed that all 20 end point PCR products gave the identical deletion by direct sequencing analysis, ruling out PCR artifacts (data not shown). Identical results were also obtained using a total of three pairs of PCR primers flanking the deletion region. The identical homozygous deletion was also identified from 10 additional RCMs (some are shown in Fig. 2) using the above technique. Since limiting dilution allowed 20 separate amplifications of single molecules, artifacts between DNA products were avoided.

This deletion was not present in the CCR5 consensus sequence obtained from four SMs (Fig. 2) naturally infected in West Africa with four distinct R5 tropic viruses, SIVsmSL92a, -c, -d, and -e (38). These results show that two closely related but geographically isolated mangabey sub-species were naturally infected with R5- or R2b-tropic SIVs that correlated with homozygous full-length or deleted CCR5 genes, respectively. The finding of an R2b-tropic SIVrcm in a natural host having a homozygous CCR5 defect suggested that SIV R2b tropism and defective CCR5 genes may be a related phenomenon.

To analyze the allelic frequency of Δ24 deletions, genomic DNA was extracted from 14 additional RCMs from widely separate locations in Gabon and an American zoo colony. DNA was analyzed from a total of six zoo-housed RCMs in America, from eight household pets in Gabon, and from one fecal sample collected fresh from a wild, free-living mangabey, RCM-F3. PBMCs from the zoo mangabeys in the USA and all household pets, except for RCM-009, were negative for SIVrcm antibodies by Western blot or culture and PCR as previously described (9). RCM-CCR5 deletions were detected using primers that specifically amplified a 213-bp fragment spanning the Δ24 region (Fig. 2). Four zoo RCMs were homozygous Δ24 and two were heterozygous (Fig. 2). Fig. 2 shows the sequence of the Δ24 and wild-type DNA fragments from the zoo heterozygote, RCM-1208, and two homozygotes from the zoo, RCM-997 and -1049. Of the remaining seven household pets in Gabon, five were homozygous for Δ24 CCR5 and two were heterozygotes. RCM-009, the infected mangabey, was homozygous for the CCR5 deletion (Fig. 2). No RCMs on either continent were homozygous wild-type for CCR5. Fecal DNA from RCM-F3, a wild, free-living RCM, was also homozygous Δ24 CCR5 (data not shown). To calculate an allelic frequency for Δ24, pedigree charts from zoo animals were examined to determine if in-breeding had occurred. Six zoo animals were unrelated. A seventh was a full sibling and was therefore excluded from the analysis. The pedigree charts also showed that the American and African groups could not be related, because of age and birthplace. The household pet RCMs were also unrelated because they were wild-caught by their owners at different locations and kept alone with no known contact with each other. The allelic frequency for Δ24 CCR5 for the 15 RCMs on two continents was 86.6% (Table 1). In marked contrast, none of 34 African pet or wild SMs were homozygous for the Δ24, but 3 were found to be heterozygous, giving an allelic frequency of 4.1%, >20-fold lower than RCMs (P < 0.001). All African SMs were also unrelated because they were young, wild-caught household pets from different parts of two West African countries, Liberia and Sierra Leone. Fig. 2 shows the Δ24 deletion found in an SIVsm-infected captive, SM-941, at the Tulane Regional Primate Research Center. The data showed a correlation between high wild-type and deleted CCR5 allelic frequencies with R5- and R2b-tropic viral infection.

Taking the importance of the Δ32 defect to HIV transmission and AIDS progression in humans into consideration, we screened a Caucasian cohort for the Δ24 CCR5 genotype using the CCR5-specific PCR method that was specific for the deletion as described in Materials and Methods and shown in Fig. 2. In a total of 409 individuals tested, 9 were homozygous for Δ32, 41 were heterozygous for Δ32, and the remaining 306 were wild-type CCR5. The allelic frequency was 7.3% (Table 1). All of the human samples were negative for the 189-bp fragment that is diagnostic for this deletion (Table 1). Therefore, the Δ24 genotype identified in RCMs must be rare or absent in Caucasian populations. However, other human populations must be tested. In addition, we screened nine CPZs, 34 SMs, 7 lunulatus mangabeys, and 35 rhesus monkeys. Besides the three Δ24 heterozygous SMs described above, the other non-human primates tested had neither Δ24 nor Δ32 CCR5 genotypes. However, larger numbers of each simian species must be tested.

To test function of RCM-CCR5 genes, signal transduction was tested by a Ca²⁺ influx assay as previously described (37). Methods used for stable transfection of 293 cells with Δ24 or full length RCM-CCR5 were reported (25). Quantitative CCR5 gene expression was determined by measuring intracellular levels of mRNA transcripts by reverse transcription PCR assay (Fig. 3, A and B). Full length and deleted transcripts were expressed at similar levels. MIP-1β was used to induce signal transduction, because it is specific for CCR5. MIP-1β induced Ca²⁺ influx in 293T cells transfected with RCM-wt-CCR5 (Fig. 3,C, right) but not in RCM-Δ24-CCR5 transfected cells (Fig. 3 C, left). Assays using RANTES gave similar results (data not shown). Therefore, RCM-Δ24-CCR5 has lost physiologic function for signaling, like the Δ32 CCR5 human allele (32).

The coreceptor function of RCM Δ24 and wild-type CCR5 genes, obtained from the zoo heterozygote, were further tested in transfected HOS.CD4 cells. The RCM-CCR5–transfected HOS.CD4 cells were inoculated with single-cycle HIV-1– or SIVmac-based luciferase reporter viruses pseudotyped by various envelopes as previously reported (25). When equal amounts of viruses were added, no viral entry by any of eight R5-tropic viral envelopes was detected in Δ24-RCM-CCR5 cells (Fig. 4,A). However coreceptor function of RCM-wt-CCR5 was intact for each R5-tropic viral envelope (Fig. 4 A).

To determine if SIVrcm could replicate in cells expressing the homologous full-length CCR5 derived from its own host species, SIVrcm was inoculated on HOS.CD4 cells expressing RCM-wt-CCR5 (Fig. 4). Replication assays were used because SIVrcm env clones are not yet available for the reporter assay. All six R5-tropic viruses, including HIV-1, HIV-2, and highly divergent strains of SIV, replicated in cells expressing RCM-wt-CCR5 (Fig. 4,B). However, SIVrcm did not replicate in RCM-wt-CCR5 cells (Fig. 4,B). In contrast, SIVrcm was the only lentivirus out of seven tested that replicated in cells transfected with CCR2b (Fig. 4 C). Similar results were obtained using U87.CD4.Hu-CCR2b. Therefore, SIVrcm could not use its own full-length, functional CCR5.

In this study we describe a natural SIV infection of an RCM that was homozygous for a deletion in the CCR5 gene. The naturally occurring SIVrcm was found to be R2b-tropic and not capable of using full-length human or RCM-CCR5. Sequencing of the RCM-CCR5 revealed a unique 24-bp deletion in the fourth transmembrane region. Characterization of the Δ24 CCR5 allele showed that it was common in both wild-caught RCMs in Gabon and in unrelated captives in an American zoo. The overall Δ24 allelic frequency was 86.6%, significantly higher than the 4.1% frequency found in SMs, a closely related species from West Africa that was naturally infected with R5-tropic SIVs (31). The wild-type and deleted CCR5 alleles were cloned from a heterozygous RCM, because no homozygotes for wild-type CCR5 were found. Expression in human cell lines showed that only the full-length wild-type RCM-CCR5 was functional as a coreceptor for R5-tropic HIV-1, HIV-2, SIVsm, and SIVmac strains. The full length wild-type RCM-CCR5 allele was also capable of signaling when tested for MIP-1β–induced calcium influx. Although Δ24 CCR5 mRNA was expressed in transfected cells, the cells were inactive in signal transduction.

Since the 24-bp deletion encompasses eight amino acids in frame, the deleted gene may be synthesized as a truncated protein inside the cell. Stable cell surface expression of the truncated protein is possible, but appears unlikely because signal transduction was not detected (Fig. 3). The deletion may adversely affect expression because it shortens the membrane-spanning domain (Fig. 2) from 26 amino acids to ∼19 amino acids. However, expression of Δ24 CCR5 with limited function is possible because the hydrophylic amino acids (R140 and R168) that determine fourth transmembrane region properties were preserved. Benkirane et al. (39) reported that the 32-bp deletion in the region of the second extracellular loop of the human CCR5 gene resulted in the loss of phosphorylation of the CCR5 gene product. The truncated CCR5 protein remained in the endoplasmic reticulum and was not transported. Because the 24-bp deletion reported here is in-frame, and in the fourth transmembrane domain, it is not possible to predict whether or not the mutation's effect will be similar to that observed by Benkirane et al. for the Δ32 deletion. Additional studies are needed to characterize the 24-bp defect and its effect on function, expression, and transport.

Although only 15 RCMs were tested, the data support a naturally high frequency for the deleted allele. The high frequency was found in two groups of RCMs on different continents, a finding that could not be explained by inbreeding. Furthermore, a high frequency of deleted genes was also present in Gabonese household pets and in one wild RCM. All Gabonese mangabey alleles were derived from wild, free-living populations from different areas of the country. In marked contrast, 34 household pet and wild SMs from Sierra Leone and Liberia showed a significantly lower frequency (4.1%) similar to the prevalence of the Δ32 CCR5 found in certain human populations (40). This cause for the high frequency in RCMs is unknown, but may have been caused by the spread of a small founder population of RCMs. Comparative genetic studies between SMs and RCMs could be used to determine if there was a bottle neck effect on the RCM sub-species that produced the high frequency of deleted CCR5 genes and a unique SIV–host relationship.

Hypotheses have been proposed that the appearance of the human CCR5 deletion was in response to a past epidemic in which the loss of the CCR5 gene conferred resistance against an epidemic agent (40). Similar forces may have been involved in the selection of the mangabey defect. The deletions in mangabeys and humans may represent convergent evolution, due to similar ecological pressures. However, their temporal occurrence and geographical distributions appear to be highly discordant, based on results thus far. The historic ranges and dispersals of SMs and RCMs containing the Δ24 CCR5 deletion indicate that this genetic defect arose some time before mangabey subspeciation. Subspeciation of these coastal mangabeys developed more than 10,000 yr ago due to natural barriers that formed in their range. The Dahomey gap, which is a 900-km Savannah break in upper and lower guinea forest, is one such barrier. Rivers in the Dahomey gap may also have been barriers that contributed to development of the subspecies (41). This date is significantly older than the 700 yr estimate for the age of the 32-bp CCR5 deletion in humans (40). Why the mangabey CCR5 deletion would be much older than the human deletion remains to be determined. The difference in ages of the two defects suggest that the selective pressures for the deletions were different or were caused by similar events widely separated in time.

The mangabey and human CCR5 deletions are also discordant in geographical distribution. The human Δ32 is most common in Northern Europeans and is absent among populations native to Africa (40). In marked contrast, the mangabey CCR5 deletion was found in equatorial African mangabeys. The discordance in Δ24 CCR5 distributions further indicates that the selective pressures for the emergence of the mutations occurred under quite different circumstances.

Other primate species, including macaques, CPZs, and humans, have the same possible recombination site, GGGTGGTG (Fig. 2), but the deletion was not found in >400 individuals tested from these three species (Table 1). More monkeys, as well as more humans, must be tested to extend these results. The finding of a deleted CCR5 allele in a more readily available research monkey species such as the macaque would be useful for pathogenic studies.

This study also raises questions about the origin of R2b tropism in SIV. This tropism is unknown among highly divergent SIVsm, SIVmac, SIVagm, and SIVcpz, all of which are R5-tropic (25–28, 30, 31). R5-tropic SIV strains originated from different parts of Africa, showing their widespread distribution among naturally infected simian species. Therefore, it is possible that SIVrcm was derived from the more common R5-tropic SIV and switched its tropism to CCR2-based coreceptors in response to the appearance of deleted CCR5 genes at high frequencies in its host. SIVrcm R2b-tropism may have arisen when RCMs became a separate, isolated subspecies in relatively ancient times.

Adaptation of SIV from CCR5 to CCR2b would require an intermediate coreceptor. Such a coreceptor is available for SIVrcm, based on in vitro studies that showed that the virus used STRL-33 (Fig. 1). Because SIVrcm and all other SIVs are dual tropic (Fig. 1 and references 25, 28, 30, 31), R5-tropic ancestors of SIVrcm may have replicated in mangabey hosts using a minor coreceptor and then adapted to CCR2-based coreceptors, the closest chemokine coreceptor to CCR5 (42, 43). This hypothesis can be tested by infecting homozygous deleted mangabeys with R5-tropic SIV. If selective pressure can induce coreceptor switches, then the selection pressure from CCR5-directed drug therapy may cause viral adaptation to other coreceptors resulting in drug resistance. This concern is raised because it has been shown that certain HIV-1 strains used CCR2b as a coreceptor, and changes in coreceptor tropism are known to occur in HIV-1–infected people (42–44).

We thank J. Blanchard of the Tulane Regional Primate Research Center and M. Fernandez of the Audubon Park Zoo for mangabey specimens. We also thank L. Fields for zoological studbook information; D. Littman and V.N. Kewal Ramani for GHOST cell lines; B. Hahn for HIV-2; S.P. Mossman for more env expression vector; N.R. Landau for viral pseudotype vectors; Christine Russo, Amara Luckay, Jeffrey Hays, Linqi Zhang, Yaoxing Huang, Tian He, and Yong Guo for technical assistance; and Theresa Secrist for administrative assistance.

This work was funded by National Institutes of Health grants to P.A. Marx (AI41420 and AI38573) and D.D. Ho (AI40387), and by the The Irene Diamond Fund. Z. Chen is a Charles H. Revson/Norman and Rosita Winston fellow in biomedical research at The Rockefeller University.

AGM/agm

African green monkey

CPZ/ cpz

chimpanzee

GFP

green fluorescent protein

MIP

macrophage inflammatory protein

RCM/rcm

red-capped mangabey

SIV

simian immunodeficiency virus

SM/sm

sooty mangabey