Report

Contrasting Genetic Influence of CCR2 and CCR5 Variants on HIV-1 Infection and Disease Progression

See allHide authors and affiliations

Science  15 Aug 1997:
Vol. 277, Issue 5328, pp. 959-965
DOI: 10.1126/science.277.5328.959

Abstract

The critical role of chemokine receptors (CCR5 and CXCR4) in human immunodeficiency virus–type 1 (HIV-1) infection and pathogenesis prompted a search for polymorphisms in other chemokine receptor genes that mediate HIV-1 disease progression. A mutation (CCR2-64I) within the first transmembrane region of theCCR2 chemokine and HIV-1 receptor gene is described that occurred at an allele frequency of 10 to 15 percent among Caucasians and African Americans. Genetic association analysis of five acquired immunodeficiency syndrome (AIDS) cohorts (3003 patients) revealed that although CCR2-64I exerts no influence on the incidence of HIV-1 infection, HIV-1–infected individuals carrying theCCR2-64I allele progressed to AIDS 2 to 4 years later than individuals homozygous for the common allele. BecauseCCR2-64I occurs invariably on a CCR5-+–bearing chromosomal haplotype, the independent effects ofCCR5-Δ32 (which also delays AIDS onset) andCCR2-64I were determined. An estimated 38 to 45 percent of AIDS patients whose disease progresses rapidly (less than 3 years until onset of AIDS symptoms after HIV-1 exposure) can be attributed to theirCCR2-+/+ or CCR5-+/+ genotype, whereas the survival of 28 to 29 percent of long-term survivors, who avoid AIDS for 16 years or more, can be explained by a mutant genotype forCCR2 or CCR5.

The nexus of chemokine immunobiology and AIDS pathogenesis has revealed untapped avenues for resolving patterns of HIV-1 disease progression, for clarifying epidemiologic heterogeneity, and for design of therapies (1-6). Identification of the CC-chemokines, RANTES, MIP1α and MIP1β, as suppressor factors produced by CD8 cells that counter infection by certain HIV-1 strain infections (7) previewed the critical identification of two chemokine receptor molecules, CXCR4 (formerly named LESTR/fusin) and CCR5 (formerly CKR5), as cell surface coreceptors with CD4 for HIV-1 infection (8-13). Additional chemokine receptors CCR2 and CCR3 also have been implicated as HIV-1 coreceptors on certain cell types (12-14). HIV-1–infected patients harbor predominantly macrophage-tropic HIV-1 isolates during early stages of infection, but accumulate increasing amounts of T cell–tropic strains just before accelerated T cell depletion and progression to AIDS. The identification of “dual”-tropic HIV-1 strains over the course of infection suggests that such strains may represent an intermediate between macrophage- and T cell–tropic populations (11-13, 15). This tropic transition indicates that viral adaptation from CCR5 to CXCR4 receptor use may be a key step in progression to AIDS (16).

A common 32–base pair (bp) deletion mutation in the CCR5gene that causes truncation and loss of CCR5 receptors on lymphoid cell surfaces of homozygotes was recently described (17-19). Genotype analysis of more than 4000 individuals from multiple AIDS cohorts demonstrated that (CCR5-Δ32/Δ32) deletion homozygosity was not uncommon (1 to 5%) among exposed uninfected individuals, but exceedingly rare (<0.1%) among infected individuals, indicating theCCR5-Δ32/Δ32 homozygotes strongly resist HIV-1 infection (19-23). Further, the onset of AIDS was postponed 2 to 4 years in individuals heterozygous forCCR5-Δ32 and the normal CCR5-+ allele in several large AIDS cohort studies (19, 21,22).

Despite the CCR5 data plus several HLAassociations that influence HIV-1 exposure outcome (24-27), identified genetic factors can account for only a small proportion of the “long-term survivors” who continue to resist AIDS-defining illness 10 to 20 years after HIV-1 infection. For example, 80% of highly exposed uninfected individuals in these studies are not CCR5-Δ32/Δ32homozygotes (21), and more than 60% of long-term survivors are homozygous for the common allele (CCR5-+/+) (19,21, 22).

To identify alterations in chemokine receptor genes implicated as HIV-1 coreceptors (8, 12-14,28-30), we screened the entire CCR2gene for variants by means of the single-strand conformation polymorphism (SSCP)/heteroduplex mobility assay (31). A G-to-A nucleotide substitution was detected at position 190 (counting from the ATG start codon) that substitutes the CCR2-+ amino acid residue valine at position 64 to isoleucine (CCR2-64I), a conservative change located within the first transmembrane domain of the CCR2 receptor. That domain has a completely conserved amino acid sequence identity with CCR5, which suggests functional constraints on mutational variation. The mutant CCR2-64I allele has a transmembrane sequence that is identical to the CCR5 normal allele. Using both SSCP and sequence-directed polymerase chain reaction–restriction fragment length polymorphism (PCR-RFLP) assay (32), we determined the allele and genotype frequencies of 3003 individuals enrolled in five prospective AIDS cohorts. The CCR2-64I alteration was common in all ethnic groups with the following allele frequencies: 0.098 in Caucasians (n = 1847 individuals); 0.151 in African Americans (n = 899); 0.172 in Hispanics (n = 207), and 0.250 in Asians (n = 40).

There were no significant differences in CCR2 allele or genotype frequencies in comparisons of exposed (or high-risk) uninfected (HIV–) versus infected (HIV+) patients in any of five clinically defined AIDS cohorts (33-39). A collection of 58 extremely high-risk, exposed uninfected individuals (those with documented receipt of HIV-1–contaminated clotting factor, or frequent unprotected sexual encounters with high-risk partners) (19, 21, 37, 38) also showed CCR2allele/genotype frequencies not significantly different from those of HIV-1–infected individuals (40). The CCR2genotype frequencies in each cohort and each HIV-1 infection category conformed to expectations of Hardy-Weinberg equilibrium, further excluding any significant effect of CCR2-64I on HIV infection.

A subgroup of 891 seroconverter patients (those whose date of HIV-1 infection could be estimated precisely, because they enrolled in the cohort before converting from HIV-1 antibody-negative to HIV-1 antibody-positive) from four cohorts (MACS, SFCC, MHCS, and ALIVE) (41) was analyzed by comparing the rate of progression to AIDS among different CCR2 genotypes, using a Cox proportional hazards model (42). Three endpoints or AIDS definitions, two of which had been stipulated by CDC (43) (reflecting increasing morbidity) were considered: (i) AIDS-1993 definition includes HIV-1 infection plus AIDS-defining illness, decline of CD4-T-lymphocytes to ≤200 cells/mm3 or death; (ii) the more stringent AIDS-1987 definition includes HIV-1 infection plus development of AIDS-defining illness or death; and (iii) death during follow-up for an HIV-1–infected patient (97% of these had AIDS-1993). The results of these analyses are illustrated in Fig.1 and tabulated in Table1.

Figure 1

Kaplan-Meier survival curves demonstrate the dependence of progression to AIDS-1993 on CCR2 genotype in (A) MHCS and (B) combined “All” cohort analyses among seroconverters (33, 41).CCR2-+/+ genotype survival is compared toCCR2-+/64I plus CCR2-64I/64I individuals, because CCR2-64I/64I individuals comprise few (∼1%) members of each cohort. Number of patients (n), statisticalP value (P), and relative hazard (RH) based on the Cox proportional hazard models (35) are given. Summary statistics for each cohort for each AIDS endpoint is presented in Table1. (C) Kaplan-Meier survival plots of time to AIDS-1993 are shown for MHCS seroconverters genotyped for CCR2-64I andCCR5-Δ32 polymorphisms. Curves represent compound genotypes of the CCR2/CCR5 compound locus as defined in text. (D) Survival analysis of combined “All” cohorts for AIDS-1993, based on compound locus gentoypes as in (C); (E) survival plots for MACS seroconvertors time-to-death, based on compound locus genotypes as in (C); and (F) survival analysis of combined “All” cohorts time-to-death, based on compound locus genotypes as in (C) are shown. Summary statistics for each cohort and each AIDS endpoint are presented in Table 2 (CCR2 and CCR5 genotypes separated). (C) through (F) are for Caucasians only.

Table 1

Survival analysis for progression to AIDS among HIV-1–infected individuals as a function of CCR2 genotype as in Fig. 1. Seroconverters of all racial groups were analyzed for the ALIVE, MACS, MHCS, and SFCC cohorts and the combination of all the cohorts (41). HGDS was excluded since this cohort has only seroprevalent individuals. A χ2 (1 df),P, and relative hazard (RH) were calculated for each variable in the analysis of AIDS outcomes. Time to AIDS-1993, AIDS-1987, and death were calculated from the midpoint of their last HIV-1–negative and first antibody-positive test (41, 53). Seroconvertors were analyzed in a Cox proportional hazard analysis (42). Log Survival Time versus Log Time plots were examined for proportionality and they were reasonably parallel especially after the first few years of outcomes. The significance of the Kaplan-Meier log-rank Chi-square results obtained for the analyses were essentially the same as Cox proportional hazards models (54). Analyses were age adjusted for those individuals <30, 30 to 40, or >40 years old.

View this table:

For three cohorts (MACS, MHCS, and SFCC) plus the combined cohort analysis, a consistent 2- to 3-year postponement in median time to AIDS outcome (when the median was reached by each definition) was observed for CCR2-+/64I plusCCR2-64I/64I genotypes compared with patients homozygous for the normal CCR2 +/+ allele. TheCCR2 genotypic protection was statistically significant for combined analysis with AIDS-1987, AIDS-1993, and death endpoints. The ALIVE cohort, which is composed of 94% African Americans, did not show a CCR2 genotype association in the survival analysis (44). When Caucasian participants alone were examined, the combined analysis showed significant (or highly significant) postponement of AIDS among CCR2-+/64I plusCCR2-64I/64I genotypes for each clinical AIDS endpoint definition. The significant relative hazards for Caucasian seroconverters ranged from 0.69 to 0.74, indicating that individuals with a CCR2-+/+ genotype progress to AIDS 40% more rapidly than patients carrying the CCR2-64I allele.

The protective effect exerted by the CCR2-64I allele and included genotypes is also apparent from a defined disease category analysis, which allows the inclusion of both seroconverters and HIV-1 seroprevalent individuals (Fig. 2). Each individual cohort plus combined cohorts were divided into relatively rapid progressors versus slow progressors and nonprogressors on the basis of the midpoint of their survival distribution. TheCCR2-64I allele frequency was consistently lower among rapid progressors to AIDS than in the slow or nonprogressor group who avoid AIDS (by each clinical definition) for greater than 6 to 12.5 years after infection. This analysis included 1746 patients, and in every case, the cohorts showed a 30 to 80% increase in CCR2-64Iallele frequency in the slow and nonprogressor category (Fig. 2). Similarly, when the frequency of CCR2-+/64I genotypes is compared among disease categories, the heterozygote genotype frequency is greater in the slow and nonprogressor categories for 11 of 12 comparisons (four cohorts, three AIDS outcomes; Fig. 2) (45).

Figure 2

Analysis of CCR2-64I allele and genotype frequencies (all races) and CCR2/CCR5 compound locus genotypes (Caucasians only) with reference to progression to AIDS in each cohort and in all patients based on three AIDS endpoints. Seroconverters who progressed to designated outcomes before the cutoff time (n = 640) were compared with seroconverters plus seroprevalents (n = 660; those who were HIV-1 antibody-positive when enrolled in the cohort) who survived outcome-free for at least that long. Cutoffs, in years, were chosen as the time where approximately half the seroconverters had progressed to the outcome. Times for the outcomes and cohorts were : (for AIDS-1993) MACS, 6 years; MHCS, 9 years; SFCC, 12.5 years; and All, 8 years; (for AIDS-1987) MACS, 7.5 years; MHCS, 11.5 years; SFCC, 14.5 years; and All, 10 years; (for death) MACS, 8 years; MHCS, 11.5 years; SFCC, 14 years; and All, 12 years (34-38). Number of patients in each disease category is listed below the bar graph. The χ2 analyses of alleles and genotypes had one and two degrees of freedom, respectively. P values are indicated by * (<0.05), ** (<0.01), and **** (<0.0001). For CCR2allele and genotype (left panel), all patients are included, regardless of CCR5 genotype. Slow or nonprogressors to AIDS had higherCCR2-64I allele frequencies than did more rapid progressors in every cohort-outcome combination. Similarly, in all comparisons, theCCR2+/64I and 64I/64I genotypes are more frequent in the longer-term survivors who have a binomial sign test,p = 6 × 10−8. CCR2-+/+;CCR5-+/Δ32 (right panel) equals [+/Δ32] compound genotype prevalence in defined disease categories and CCR2-+/64I; CCR5-+/+equals [64I/+] compound genotype prevalence in defined disease categories. CCR2-64I andCCR5-Δ32 allele frequencies were also elevated in slow progressors in analyses of individual and combined cohorts (54).

CCR5-Δ32/+ heterozygotes also demonstrate a slower progression to AIDS based on studies of these same and other cohorts (19, 21, 22). Because the CCR2 andCCR5 loci are very tightly linked (∼10 kb apart) on chromosome 3 (19, 29, 30), we examined the co-occurrence and genotypic independence of CCR5 and CCR2alleles among patients from the same cohorts. Analysis of the two locus genotypes for 2916 patients showed that three genotypes (CCR2-64I/64I, CCR5-+/Δ32; CCR2-+/64I, CCR5-Δ32/Δ32; and CCR2-64I/64I, CCR5-Δ32/Δ32) were absent, indicating that the mutant alleles of the two genes are in strong, perhaps complete, linkage disequilibrium with each other (g = 54.01; 4 df; P < 0.0001). This means thatCCR5-Δ32 invariably occurs on a chromosomal haplotype (linked-gene allele combination) that is CCR2-+, whereas CCR2-64I occurs on a haplotype that containsCCR5-+. As a consequence of the tight chromosomal linkage plus alternative haplotypes carrying each mutant allele, the two mutant haplotypes (CCR2-+)-(CCR5-Δ32) and (CCR2-64I)-(CCR5-+) plus the two-locus normal haplotype (CCR2-+)-(CCR5-+) can be considered as three alleles of a combined CCR2/CCR5 compound locus producing three recognizable genotypes designated as: [+/+]; [+/Δ32]; and [64I/+] representing (CCR2-+/+; CCR5-+/+); (CCR2-+/+; CCR5-+/Δ32); and (CCR2-+/64I;CCR5-+/+) compound genotypes, respectively.

A survival analysis of the effect of both CCR2 andCCR5 genotypes on progression to AIDS-1993 and on progression to death for two cohorts (MHCS and MACS, respectively) plus a combined cohort analysis of seroconverters is presented (Fig. 1, C through F). This analysis allows the independent evaluation ofCCR2-64I and CCR5-Δ32 containing genotypes as separate categories compared to [+/+], but removing the mutant-bearing genotypes at one locus from the analysis of the other. The Cox proportional hazards model revealed highly significant postponement of AIDS onset for CCR2-64I–containing genotypes and also for CCR5-+/Δ32 heterozygotes compared to the [+/+] individuals, homozygous for normal alleles at each locus (Fig. 1, C through F).

An analytical summary for each cohort's survival curves is presented (Table 2) forCCR5-+/Δ32 versus CCR5-+/+ for all seroconvertors; compound CCR2-CCR5 locus genotype classes, [+/+] versus [+/Δ32] and [+/+] versus [64I/+] (also illustrated in Fig. 2); and combined analysis of CCR2-+/+; CCR5-+/+ versus all mutant genotypes at either locus. These data reaffirm the protective effects of CCR5 (17-19) and show that theCCR2-64I protection is as strong as and independent of theCCR5-+/Δ32 influence. The CCR2 effect increases when the CCR2-64I/+ andCCR5-+/Δ32 subjects are evaluated as separate categories, because the relative hazards for each cohort in Table 2(CCR2 and CCR5 genotypes separated) are lower than in Table 1, which includes CCR5-Δ32/+heterozygotes in the CCR2-+/+ group. The results are highly significant for combined cohort analyses for each AIDS endpoint (Table2). The CCR5-Δ32 protective effect is also strengthened when CCR2-64I–containing genotypes are evaluated separately (Table 2; CCR2 and CCR5separated versus CCR5 only). When mutant genotypes forCCR2 and CCR5 are combined as a single genetic category (Table 2, CCR2 and CCR5 grouped), the statistical significance for genetic influence reaches as low as 6.3 × 10–6 for combined cohort progression to AIDS-1993 (Table 2). Individuals with a mutant genotype at eitherCCR2 or CCR5 show hazards relative to normal of 0.51 to 0.71 in statistically significant analyses.

Table 2

Effects of CCR2 and CCR5genotypes on progression to AIDS outcomes in survival analyses of Caucasian seroconverters. Non-Caucasians were excluded because of the absence of CCR5-Δ32 allele in these populations (18-22) plus the potentially differentCCR2 survival (44). Cox proportional hazard analyses were performed as described in Table 1. CCR5;CCR2genotypes were analyzed in three ways: (i) +/Δ32versus +/+ for CCR5 alone; (ii)CCR5 and CCR2 separated into the compound genotypes [64I/+] or [64I/64I], and [+/Δ32] versus [+/+] normal at both loci, illustrated in Fig. 2, C through F; and (iii) CCR5and CCR2 grouped together: [+/+] versus [+/64I], [64I/64I], [+/Δ32], or [64I/Δ32]. Caucasians from the ALIVE cohort are included in the analyses of combined “All” cohorts. Sample sizes for the three analyses described above vary slightly because: (i) analysis of CCR5alone includes some individuals without CCR2 genotypes; (ii) separate analysis of the CCR2 and CCR5 genotypes excludes 17 (CCR2-+/64I;CCR5-+/Δ32) double heterozygotes because too few were available for separate analysis, and combining them with either single locus heterozygote category would be arbitrary; and (iii) group analysis of CCR2 and CCR5includes patients with both CCR2 and CCR5genotypes including (CCR2-+/64I;CCR5-+/Δ32). A completeCCR2 and CCR5 separate analysis that included the 17 double heterozygotes as a separate composite genotypic category was performed as well. Relative hazards (P values) for the combined cohorts of this analysis including four composite genotypes [+/+]; [64I/64I] or [64I/+]; [+/Δ32]; and [64I/Δ32] for three outcomes were: AIDS-1993, RH = 0.72 (P = 0.07); AIDS-1987, RH = 0.81 (P = 0.07); death, RH = 0.79 (P = 0.06). RH and P values for genotypes other than the double heterozygotes were identical to the values listed in theCCR2 and CCR5 separated genotype analysis. The co-occurrence of CCR2 and CCR5 protective variants does not appear to confer additional protection, but the double heterozygote is rare in these groups.

View this table:

A defined disease category analysis using compound two-locus genotypes (Fig. 2) shows that the heterozygote frequencies for bothCCR2 and CCR5 mutant haplotypes were elevated in slow or nonprogressors relative to the rapid progressors. As for the analysis for CCR2 separately, the frequency ofCCR2 and CCR5 mutant genotypes was significantly greater among slow progressors for both AIDS-1993 and death with combined cohort analyses plus for the MACS cohort alone. The mutant alleles CCR2-64I and CCR5-Δ32, and genotypes containing them, are less frequent in the more rapid progressors in all three cohorts and when all three AIDS outcomes (17 of 18 comparisons) (46) are considered, affirming the protective effects shown in the survival analysis.

The cumulative effects of CCR2 and CCR5 mutant alleles over six intervals after HIV-1 infection (Fig.3) show a marked increase of mutant genotypes for both CCR2 and CCR5 among patients that survive without progressing to AIDS over longer periods. The elevated heterozygote frequency for the two loci in long-term survivors (patients surviving more than 16 years and free of AIDS) emphasize the protective effect of both loci. The difference in observed mutant genotype (either CCR2 or CCR5) frequency in patients with different times to AIDS from the frequency in all 849 patients (arrow in Fig. 3) allows an estimate of the fraction of rapid progressors (AIDS in <3.5 years) and long-term survivors (≥16 years without progressing to AIDS) that can be attributed toCCR2/CCR5 genetic factors (47). For long-term survivors, the attributable risk conferred by [+/Δ32] and [+/64I] genotypes is 29% for AIDS-1993; 28% for AIDS-1987 and 28% for death. For rapid progressors, the estimate for [+/+] genotypes is 42% for AIDS-1993; 38% for AIDS-1987, and 45% for death.

Figure 3

Frequencies of the protective genotypes ([+/Δ32], [+/64I], [64I/Δ32], and [64I/64I]) in six categories of increasing survivorship during HIV-1 infection in Caucasians. Genotypic frequencies were calculated separately for time to AIDS-1993, AIDS-1987, and death, from seroconverters which were <3.5, 3.5 to <7, and 7 to <10 years. In addition, genotypic frequencies were calculated for seroconverters and seroprevalents whose time to the outcome or in study without developing AIDS was 10 to <13, 13 to <16, and ≥16 years. The number of people observed in each category is shown above each column. The average frequency of these variants in Caucasians (36%) is shown as an arrow for comparison of progression categories. Contingency tests of the three common genotypes ([+/+],[+/Δ32], and [+/64I]) were performed for time to AIDS-1993 [χ2 (10) = 25.44, P = 0.005], AIDS-1987 definition [χ2 (10) = 14.64, P = 0.015], and death [χ2 (10) = 19.27, P = 0.04]. Contingency tests of [+/+] versus all others for time to AIDS-1993 [χ2 (5) = 24.22, P = 0.0002], AIDS-1987 definition [χ2 (5) = 14.21, P = 0.01], and death [χ2 (5) = 19.57, P = 0.002] were also performed.

The protective effect of the CCR2-64I allele in delaying the onset of AIDS among prospective cohorts is of similar magnitude and in addition to the protection observed in CCR5-+/Δ32heterozygotes. Approximately one quarter of the HIV-1–infected long-term survivors who avoided AIDS for more than 16 years can be attributed to their CCR2/CCR5 genotype (Fig. 3). However, there are important differences between CCR2 andCCR5 that bear on their effects and interpretations.

First, the CCR2-64I mutation has no noticeable effect on HIV-1 infection, whereas CCR5-Δ32 homozygotes are strongly resistant (19-23). Second, theCCR5-Δ32 mutation results in a truncated protein removing the HIV-1 coreceptor from cells (17, 48), whereasCCR2-64I encodes a conservative amino acid substitution. Third, the CCR5-Δ32 mutation is unique to the Caucasian ethnic group, whereas CCR2-64I was found in every ethnic group tested. This suggests that the CCR2-64Imutation may be much older in human history than the relatively recentCCR5-Δ32 deletion (49). Fourth, CCR2 and CCR5 bind to different chemokine ligands (CCR2 binds MCP-1, -2, and -3; CCR5 binds RANTES, MIP1α, and MIP1β), although the distribution of CCR2 and CCR5 in cells and tissue is very similar (50, 51).

The mechanism for CCR2-64I delay of AIDS in patients is not immediately apparent. Three possible modes of restriction include: (i)CCR2-64I–bearing individuals stem viral spread and pathogenesis directly by altering the kinetics of HIV-1 infection. This mechanism gains support from the demonstration that CCR2 serves as a critical coreceptor for certain macrophage-tropic, T cell–tropic, and dual-tropic strains of HIV (13, 14, 51). (ii) CCR2 concentration and physiological variation can indirectly affect the availability of CCR5 on target cells for HIV-1. There is precedence for cross-regulation of different cytokine receptors and indirect evidence for the same among chemokine receptors (14, 52). Furthermore, the amount of CCR5 on lymphocytes varies markedly among individuals with the same CCR5 genotype, suggesting that factors other than CCR5 genotype contribute to CCR5 abundance (53). These could be other genes (for example,CCR2), ligand binding, or other chemokine receptors. (iii) The CCR2-64I mutation could be tracking a linked mutation through linkage disequilibrium. This may involve a regulatory or structural mutation in the CCR1-5 gene cluster, but notCCR5-Δ32 because the CCR2-64I mutation travels on a different haplotype. We sequenced the complete coding regions of several CCR5 and CCR2 mutant homozygotes and did not find any other nucleotide changes (31).

The CCR5-Δ32 and CCR2-64I polymorphisms are present in the protective heterozygous or homozygous states in 30 to 40% of people in every ethnic group. The demonstration that having mutant alleles at these genes is protective against progressing to AIDS has important implications for therapy, because chemokine receptors are required cellular ports for HIV-1 cell entry and spread.

  • * These authors contributed equally to this study.

  • To whom correspondence should be addressed. E-mail: obrien{at}ncifcrf.gov

REFERENCES AND NOTES

View Abstract

Navigate This Article