Genetic Acceleration of AIDS Progression by a Promoter Variant of CCR5

See allHide authors and affiliations

Science  04 Dec 1998:
Vol. 282, Issue 5395, pp. 1907-1911
DOI: 10.1126/science.282.5395.1907


The CCR5 gene encodes a cell surface chemokine receptor molecule that serves as the principal coreceptor, with CD4, for macrophage-tropic (R5) strains of human immunodeficiency virus–type 1 (HIV-1). Genetic association analysis of five cohorts of people with acquired immunodeficiency syndrome (AIDS) revealed that infected individuals homozygous for a multisite haplotype of the CCR5regulatory region containing the promoter allele, CCR5P1, progress to AIDS more rapidly than those with other CCR5promoter genotypes, particularly in the early years after infection. Composite genetic epidemiologic analyses of genotypes bearingCCR5P1, CCR5-Δ32,CCR2-64I, and SDF1-3′A affirmed distinct regulatory influences for each gene on AIDS progression. An estimated 10 to 17 percent of patients who develop AIDS within 3.5 years of HIV-1 infection do so because they are homozygous forCCR5P1/P1, and 7 to 13 percent of all people carry this susceptible genotype. The cumulative and interactive influence of these AIDS restriction genes illustrates the multigenic nature of host factors limiting AIDS disease progression.

The humanCCR5 and CCR2 chemokine receptor genes, which serve as coreceptors with CD4 for HIV, are tightly linked on chromosome 3p21-22, separated by 20 kb (1–4). Common allelic variants in both genes are associated with slower progression to AIDS after infection (1, 5–11). The protective influences ofCCR5-Δ32 and CCR2-64I are independent in AIDS cohorts, and the two mutations have never been found on the same chromosome haplotype (5, 10). The physical proximity of CCR2 and CCR5, the equivalent functional efficiency of alternative CCR2 allelic products as chemokine or HIV-1 coreceptors (9), and the rarity of HIV-1 strains that use the CCR2 receptor (4) have led to the speculation that CCR2-64I may be hitchhiking (or tracking by linkage disequilibrium) with an undiscoveredCCR5 variant, perhaps in the cis-regulatory region, that is directly responsible for the CCR2-64I protective effect (5, 9–11).

The promoter region of CCR5 has been characterized, and transcription is regulated by two domains, the stronger of which is located within the region that includes intron 1, exon 2, and part of exon 3 (12–14). Four common allelic variants (CCR5P1–P4) were detected in a denaturing high-pressure liquid chromatography (DHPLC) screen of AIDS patients (Fig. 1) (15). Six rare alleles (CCR5P5–P10) were discovered as heterozygotes upon subsequent single-strand conformation polymorphism (SSCP) screening of five AIDS cohorts (15). Sequence analysis of theCCR5 promoter region of individuals homozygous for theCCR5P1–P4 variants and heterozygotes of the six rare variants revealed 10 polymorphic nucleotide positions that describe the 10 CCR5 promoter haplotype alleles, referred to as promoter alleles (Fig. 1).

Figure 1

Map of CCR2, CCR5,CCR5 promoter region, and nucleotide variants. SSCP and DHPLC screens (15) of two segments of the CCR5promoter region (12) (positions 18 to 252 and 558 to 841) revealed 10 nucleotide positions with genetic variation. Combinations of the 10 polymorphic nucleotide residues specify 10CCR5 promoter region haplotype alleles designatedCCR5P1 through CCR5P10. Dots indicate nucleotides identical to those of CCR5P1; dashes represent sites in rare haplotypes that were not determined. The mutantsCCR5-Δ32 and CCR2-64I are invariably associated in linkage disequilibrium with CCR5P1. Sites 208, 627, 676, and 684 were also described in (12).

An SSCP-based genotype survey of CCR5Palleles among 2603 individuals enrolled in five AIDS cohorts (1,5, 16, 17) indicated that CCR2-64I was always found on a CCR5P1-bearing haplotype and thatCCR5-Δ32 was consistently found on aCCR5P1 haplotype as well. This conclusion is derived from the observation that CCR2-64I/64I homozygotes were always CCR5P1/P1 homozygotes (N = 43) and CCR2-+/64Iheterozygotes (N = 559) were invariably heterozygous or homozygous for CCR5P1. Similarly,CCR5-Δ3232 homozygotes were always CCR5P1/P1 homozygotes (N = 18), whereas CCR5-+32 heterozygotes were invariably heterozygous or homozygous for CCR5P1(N = 298). Finally, none of 657 individuals who lacked the CCR5P1 allele had either theCCR5-Δ32 or the CCR2-64I allele (P < 0.0001). Because of these associations, the entire CCR2-CCR5 complex can be considered as a six-allele genotype system, based on the composite [CCR2(+/64I).CCR5P(P1–P4).CCR5(+32)] haplotype (CCR5P denotes promoter region alleles;CCR5 refers to the coding region alleles). The haplotypes using the same locus order are [+.P1.+], [64I.P1.+], [+.P1.Δ32], [+.P2.+], [+.P3.+], and [+.P4.+]. The six respective haplotype frequencies observed in Caucasians (N = 1383) were 0.358, 0.098, 0.104, 0.085, 0.14, and 0.354; in African Americans (N = 1006) the respective frequencies were 0.258, 0.155, 0.018, 0.229, 0.191, and 0.147 (18).

The frequency distribution of CCR5 promoter alleles and alternative genotypes was compared among 474 uninfected individuals from “at risk” cohorts and 1353 HIV-1–infected patients stratified by ethnic group. No significant differences in CCR5P allele or genotype frequencies were observed in Caucasians or African Americans or in a more stringent analysis of documented high-risk uninfected individuals from the MACS and MHCS cohorts (17,18).

The influence of CCR5 promoter alleles on the rate of AIDS progression among HIV-1–infected Caucasian (N = 700) and African American (N = 162) seroconverters (individuals whose infection date can be estimated as the time between the last negative HIV-1 antibody test and the first positive test) was examined in a survival analysis. A Cox proportional hazards model (19, 20) was used to compare progression to AIDS among genotype combinations of commonCCR5P alleles for combined and separate cohorts, including those CCR5, CCR2, and SDF1 genotypes previously shown to confer resistance to AIDS progression (20, 21). Four AIDS endpoints reflecting advancing morbidity were evaluated: (i) CD4 < 200 cells/mm2; (ii) AIDS-1993, as defined by the U.S. Centers for Disease Control and Prevention (22) (that is, HIV-1 infection plus AIDS-defining illness or decline of CD4 T lymphocytes to <200 cells/mm3), or death; (iii) the more stringent AIDS-1987 definition (22) (HIV-1 infection plus AIDS-defining illness), or death; and (iv) death during follow-up for an HIV-1–infected patient.

Aside from the previously demonstrated CCR2-64I–,CCR5-Δ32–, andSDF1-3′A/3′A–mediated delay in AIDS, the only consistent genotypic association was an accelerated AIDS progression among individuals homozygous for the CCR5P1/P1genotype. This rapid progression was highly significant for AIDS-1987 among combined Caucasian cohorts [N = 700; relative hazard (RH) = 1.53, P = 0.005], and was significant for both homosexual (MACS) and hemophiliac (MHCS) cohorts for AIDS-1993 (P = 0.06 and 0.003, respectively) and AIDS-1987 endpoints (P = 0.03 and 0.002, respectively) (18).

To assess the role of CCR5P1 in the context ofCCR2-64I and CCR5-Δ32, we performed a survival analysis considering composite [CCR2.CCR5P.CCR5] genotypes (Fig. 2, A and B). The survival analysis for combined Caucasian cohorts shows an accelerated rate to AIDS among [+.P1.+]/[+.P1.+] homozygotes for each endpoint, relative to patients with other genotypes (RH = 1.5,P = 0.002 to 0.005). The curves reflecting AIDS protection due to CCR5-+32,CCR2-+/64I, or SDF1-3′A/3′Areaffirm the protective effects of these alleles. Survival curves representing the other genotypes (individuals with CCR5P2,P3, or P4, who typed negative forCCR5-Δ32, CCR2-64I, orSDF1-3′A/3′A) progressed at an intermediate rate between the [+.P1.+]/[+.P1.+] homozygotes and the CCR2, CCR5, or SDF1 protective genotypes. Relative to unadjusted RH, the RH forCCR5P1/P1 adjusted for the protective effects ofCCR5-Δ32, CCR2-64I, andSDF1-3′A/3′A is diminished (Table 1 and Fig. 2, A and B), reflecting the mix of protective and nonprotective genotypes that contribute to the nonaccelerating (referent) group compared with theCCR5P1/P1 homozygotes alone.

Figure 2

(A andB). Kaplan-Meier survival curves of seroconverters. Time to AIDS-1993 (A) and AIDS-1987 (B) was examined in Caucasians from all the cohorts. The dotted lines are survival curves for individuals with the [CCR2-CCR5P-CCR5]: [+.P1.+]/[+.P1.+] genotype. The combined protective effects observed in individuals carrying the [+.P1.Δ32] and [64I.P1.+] haplotypes (or both), with or without theSDF1-3′A/3′A protective genotype, are shown by solid lines. Survival curves representing the other genotypes (individuals with CCR5P2, P3, or P4, who typed negative for CCR5-Δ32,CCR2-64I, or SDF1-3′A/3′A) are shown by dashed lines. RH values for genotypic protection by [+.P1.+]/[+.P1.+] versus all other genotypes and CCR5-+32,CCR2-+/64I, or SDF1-3′A/3′A(or any combination of these) versus all other genotypes are also shown. The RHs and statistical significance for [+.P1.+]/[+.P1.+] rapid progression to AIDS, computed by Cox regression analysis, become diminished when they are adjusted to consider the protective influence ofCCR5-+32, CCR2-/64I, and SDF1-3′A/3′A genotypes (in parentheses; see text). (C and D) Frequencies of the susceptible [+.P1.+]/[+.P1.+] genotype, in six time classes of progression to AIDS outcomes of AIDS-1993 (C) and AIDS-1987 (D). The genotypic frequencies in each time category reflect only seroconverters before 10 years; after 10 years, time to failure or in study includes both seroprevalents and seroconverters (5,16). (E and F) Frequencies of theCCR2 and CCR5 protective genotypes [64I.P1.+]/[+.P4.+], [+.P1.Δ32]/[+.P4.+] in the same six time classes of AIDS progression for AIDS-1993 (E) and AIDS-1987 (F). χ2 and P values reflect a Mantel-Haenszel test for a linear association between genotypic frequencies and increased time to AIDS-1993 and AIDS-1987 (36).

Table 1

Survival analyses of progression to AIDS outcomes by Caucasians with the CCR5P1/P1 genotype, with (RHadj) and without (RHu) adjustment forCCR5, CCR2, and SDF1 protective genotypes.

View this table:

The pattern of [+.P1.+]/[+.P1.+] acceleration observed in Fig. 2, A and B, suggested that the strength of the promoter genotype effect changed over time after infection, as has been observed previously for SDF1-3′A/3′A(16). To test this hypothesis, we examined theCCR5P1/P1 protection with the use of an RH model partitioned over 2-year intervals (23). This analysis indicated that RHs for this genotype were statistically significant for intervals less than 4 years with CD4 < 200, AIDS-1987, and AIDS-1993 endpoints, and for intervals less than 6 years with death as an outcome. After this point, the CCR5P1/P1influence was no longer evident.

Table 1 summarizes this time dependence by comparing the entire period of follow-up (up to 19 years) with a separate analysis for theCCR5P1/P1 effect early (<5 years after HIV-1 infection) and late (>5 years after HIV-1 infection). These analyses yielded two separate RH values (for combined cohorts with four AIDS outcomes): (i) adjusted RH (RHadj), where the effects ofCCR2-64I, CCR5-Δ32, andSDF1-3′A/3′A are considered as covariables; and (ii) unadjusted RH (RHu), where individuals with protective genotypes are included but protection is not treated as a covariable. The results affirm the marked influence of the [+.P1.+]/[+.P1.+] genotype in promoting rapid progression to AIDS, as well as the time dependence of the association. The statistical association is strongest in the early stages of HIV infection and is reduced or absent after 5 years for the three early AIDS endpoints and after 6 years for death (that is, although RH > 1.0, the P values for late stages are all >0.05). The overall and time-dependent effect ofCCR5P1/P1 is apparent in Caucasian homosexual and hemophiliac cohorts analyzed separately as well as when multiple cohorts are combined (Table 2). Further, the nonpromoter protection (that conferred by theCCR5-Δ32, CCR2-64I, andSDF1-3′A/3′A genotypes) explains part, but not all, of the CCR5P1/P1 effect, as the adjusted RHs are only 10 to 17% lower than the unadjusted RH and remain highly significant (RH >> 1.0, P ≤ 0.01; Table 1) even when the effects of protective genotypes are removed. The African American cohort fails to show the [+.P1.+]/[+.P1.+] association with rapid progression for any AIDS outcome (N = 160,P = 0.17 to 1.0), and when added to the Caucasians, the RHs for combined ethnic groups decrease compared with those for Caucasians (18).

Table 2

Analysis of protection against AIDS outcomes by theCCR5-P1/P1 genotype in separate homosexual and hemophiliac cohorts.

View this table:

The antagonistic effects of the recessive [+.P1.+]/[+.P1.+] genotype to accelerate AIDS progression and the dominant [64I.P1.+] and [+.P1.Δ32] heterozygous genotypes to delay AIDS progression (1, 5–11) were examined further in a categorical comparison of genotype frequency distribution across time intervals of AIDS outcomes (1, 5, 16) (Fig. 2, C to F). This approach, termed a defined disease category analysis, allows the inclusion of seroprevalent patients (those who enter cohort studies already HIV-1 antibody–positive) in the slow/nonprogressor category. In the absence of CCR2-64I orCCR5-Δ32 genotypic protection, the susceptible genotype [+.P1.+]/[+.P1.+] shows a significant drop in frequency over increasing periods of HIV-1 infection (P = 0.001 to 0.003; Fig. 2, C and D). Alternatively, individuals heterozygous forCCR5-+32 or CCR2-+/64Iprotective genotypes show a significant rise in incidence among patients who avoid AIDS for longer periods (P = 0.0003 to 0.002; Fig. 2, E and F). As the two protective alleles are dominant and are invariably both carried on alternative CCR5P1haplotypes ([64I.P1.+] and [+.P1.Δ32]), an increase in protective heterozygote frequency in slow/nonprogressors suggests that dominant protective alleles are stronger in effect than the cis-associatedCCR5P1 promoter allele, at least as heterozygotes. When homozygotes for CCR5-P1/P1 occur in individuals that are also heterozygous for protection ([+.P1.+]/[64I.P1.+] or [+.P1.+]/[+.P1.Δ32]), their effects are offset by each other, particularly in later years (7 to 20 years after seroconversion), because the frequencies of the combinedCCR5P1/P1 susceptible andCCR2/CCR5 protective genotypes are not significantly different in patients who survive AIDS for longer periods (P = 0.15 and 0.20 for AIDS-1993 and AIDS-1987, respectively) (18). The epidemiologic canceling ofCCR5P1/P1 AIDS acceleration byCCR5/2 genotype protection implies that the two effects are roughly equivalent (albeit opposite in direction) to promote or impede AIDS pathogenesis, respectively.

Kostrikis et al. (10) recently reported a variant at position 927 (Fig. 1) within intron 2 of the CCR5gene, which was in linkage disequilibrium with the AIDS protective variant CCR2-64I. Their suggestion thatCCR2-64I–associated protection might be explained by theCCR5P-927 variant was examined in the present cohort by typing 984 individuals for CCR2 and CCR5P-927variant (T substituted for C at position 927). AllCCR2-64I–bearing haplotypes [64I.P1.+] contained the CCR5P-927T type, confirming the strong linkage disequilibrium association of the two sites. The CCR5P-927Tvariant was also found infrequently on a [+.P1.+] haplotype (16/230; 7%). Nine of these individuals were seroconverters, too few for a robust Cox analysis. However, a preliminary analysis of these individuals hinted toward AIDS protection, although none were significant [for CD4 < 200, RHu = 0.49 (P = 0.31), RHadj = 0.41 (P= 0.21); for AIDS-1993, RHu = 0.57 (P = 0.34), RHadj = 0.49 (P = 0.22); for AIDS-1987, RHu = 0.82 (P = 0.78), RHadj = 0.70 (P = 0.61); for death, RHu = 1.35 (P = 0.67), RHadj = 1.14 (P = 0.86)]. In contrast, an independent study ofCCR2-64I protection (24) showed that a few patients carrying the CCR5P-927T variant (CCR2-+/+) progressed to AIDS faster than did patients who had the common 927C allele (18). The data affirm the association of CCR2-64I andCCR5P-927T, but neither support nor refute the idea thatCCR5P-927T adequately explains CCR2-64Iprotection. The opposite and offsetting effects of CCR5P1and CCR2 64I alleles (respectively to accelerate and delay AIDS; Fig. 2) would imply that their influences involve different mechanisms.

The recessive AIDS accelerating phenotype of the [+.P1.+]/[+.P1.+] genotype is powerful, apparent in multiple and combined cohorts, and equivalent in strength to protection from AIDS progression afforded byCCR5-Δ32 or CCR2-64I heterozygosity (Fig. 1) (1, 5–8). That the acceleration is strongest in the initial 4 to 6 years after infection is consistent with the knowledge that CCR5 is the primary HIV-1 receptor in the early years of infection (25). Potential mechanisms of action would include differential constitutive expression of CCR5 products regulated by various promoter alleles, or, alternatively, differential allele sensitivity to CCR5 promoter binding proteins, which would regulate transcription (12, 13, 26, 27). Quantitative analysis of CCR5 on peripheral blood mononuclear cells (PBMCs) of genotypically CCR5P1/P1,P1/P4, and P4/P4 healthy volunteers studied in two separate laboratories (N.L.M. and R.W.D.) revealed a wide range of CCR5 expression within each genotype but near-equivalent mean concentrations of CCR5, efficiency of promoting luciferase reporter gene, and infectivity by HIV-1 R5-tropic or R5/x4-tropic strains (28). These preliminary results would indicate that quantification of CCR5, ligand signaling, and HIV-1–tropic type usage on different cell subsets susceptible to HIV-1 infection (29), plus identification of requisiteCCR5 promoter binding factors, would be required to more fully account for CCR5 P1/P1 action in vivo.

To date, three distinct genetic sites within a short distance on chromosome 3p21 (CCR2, CCR5P, andCCR5; Fig. 1), plus SDF1 on chromosome 10q11, have been identified as encoding common alleles that independently delimit or increase the rate of AIDS progression among HIV-1–infected carriers (1, 5–11, 16). The frequency of individuals who have one or more protective genotypes (CCR5,CCR2, or SDF1) is substantial: 39.1% of Caucasians and 31.5% of African Americans. CCR5P1represents the first allelic variant to accelerate AIDS progression, and 12.7% of Caucasians and 6.7% of African Americans carry aCCR5P1/P1 genotype. Using data from Fig. 2, C and D, we estimated the protected fraction of theCCR5P1/P1 genotype for rapid (≤3.5 years) progression to AIDS as 10 to 17%, indicating that 10 to 17% of very rapid progressors are in that category because they are homozygous forCCR5P1/P1 (30).

The powerful genetic influence for four chemokine/chemokine receptor loci (CCR5, CCR5P1, CCR2, andSDF1) plus the marked effects of HLA zygosity on AIDS progression (1, 5–11, 16, 31–33), combined with demonstrated functions for their gene products in AIDS pathogenesis (25), provide a compelling example of multigenic influence on HIV-1 disease progression. These findings may provide a basis for the development of therapeutic applications as well as for the resolution of other complex polygenic human diseases, including those that require environmental contingencies (such as viral exposure) for phenotype recognition.

Note added in proof: After this report was submitted, McDermott et al. (34) reported a G/T variant, corresponding to position 303 of the promoter region (Fig. 1), that showed an epidemiological association with rapid progression to AIDS. Alleles at this site were not assessed in the present report but may track the same effect due to linkage disequilibrium withCCR5P1.

  • * For the Multicenter AIDS Cohort Study (MACS).

  • For the San Francisco City Cohort.

  • For the Multicenter Hemophilia Cohort Study (MHCS).

  • § For the Hemophilia Growth and Development Study.

  • || For the AIDS Link to Intravenous Experience (ALIVE) Study.

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


View Abstract

Navigate This Article