Research Article

The Influence of CCL3L1 Gene-Containing Segmental Duplications on HIV-1/AIDS Susceptibility

See allHide authors and affiliations

Science  04 Mar 2005:
Vol. 307, Issue 5714, pp. 1434-1440
DOI: 10.1126/science.1101160

Abstract

Segmental duplications in the human genome are selectively enriched for genes involved in immunity, although the phenotypic consequences for host defense are unknown. We show that there are significant interindividual and interpopulation differences in the copy number of a segmental duplication encompassing the gene encoding CCL3L1 (MIP-1αP), a potent human immunodeficiency virus–1 (HIV-1)–suppressive chemokine and ligand for the HIV coreceptor CCR5. Possession of a CCL3L1 copy number lower than the population average is associated with markedly enhanced HIV/acquired immunodeficiency syndrome (AIDS) susceptibility. This susceptibility is even greater in individuals who also possess disease-accelerating CCR5 genotypes. This relationship between CCL3L1 dose and altered HIV/AIDS susceptibility points to a central role for CCL3L1 in HIV/AIDS pathogenesis and indicates that differences in the dose of immune response genes may constitute a genetic basis for variable responses to infectious diseases.

Duplicated host defense genes that are known to have dosage effects are thought to contribute to the genetic basis of some complex diseases, although direct evidence for this is lacking. We surmised that a hotspot for segmental duplications on human chromosome 17q might be relevant to immunity against infectious diseases such as HIV-1 because it encompasses two CC chemokine genes, CC chemokine ligand 3-like 1 (CCL3L1; other names, MIP-1αP and LD78β) and CCL4L1 (MIP-1β-like), which represent the duplicated isoforms of the genes encoding CCL3 and CCL4, respectively (13). As a consequence of these duplications, the copy number of CCL3L1 and CCL4L1 varies among individuals (2, 3) (fig. S1A). This is important because CCL3L1 is the most potent known ligand for CC chemokine receptor 5 (CCR5), the major coreceptor for HIV, and it is a dominant HIV-suppressive chemokine (3).

In light of this relationship between CCL3L1 and its in vitro effect on HIV infection, we selected HIV infection as a model system in which to test our hypothesis that segmental duplications causing dosage effects of host defense genes are associated with phenotypic effects in vivo. To test this hypothesis, we determined the distribution of chemokine gene–containing segmental duplications in 1064 humans from 57 populations and 83 chimpanzees (4). We next analyzed 4308 HIV-1–positive (HIV+) and HIV-1–negative (HIV) individuals from groups with different geographical ancestries (e.g., Africans and Europeans) to determine if the risk of acquiring HIV and the rate at which HIV disease progressed were sensitive to differences in the dose of CCL3L1 gene–containing segmental duplications (4) [supporting online material (SOM) section 4.1].

Nonrandom distribution of CCL3L1-containing segmental duplications. African populations possessed a significantly greater number of CCL3L1 gene copies than non-Africans (Fig. 1 and fig. S1B). The geographic region of origin explained nearly 35% of the total variation in the distribution of CCL3L1 gene copies (analysis of variance: F = 94.41, df = 6, 1037; P = 1.23 × 10–94). Corroborating this, in separate cohorts of HIV subjects, there were significant interindividual and interpopulation differences in CCL3L1 copy numbers. The median copy number in HIV Argentinean children was two, and in HIV African-American (AA), European-American (EA), and Hispanic-American (HA) adults, it was four, two, and three, respectively (Fig. 2, A to D, open bars, and fig. S2).

Fig. 1.

Distribution of CCL3L1 gene–containing segmental duplications in human populations and CCL3L ortholog(s) in chimpanzees (SOM section 1). The cloning and characterization of the chimpanzee orthologs is described in fig. S3. (A) The human populations are labeled below the figure, and their geographic affiliations are shown at the top. CCL3L1 copy number was determined by real-time Taqman polymerase chain reaction assays (SOM sections 2.1 and 4.2). (B) Distribution of CCL3L1 copy numbers in African and non-African human populations and chimpanzee orthologs. The mean, variance, standard deviation (SD), median, and interquartile range (IQR) of the copy numbers are shown here and in fig. S1B. (C) Cumulative frequency curves of the distribution of CCL3L1 copies in human populations from different geographic regions and chimpanzee. The order of the abbreviations [geographic regions shown in (A) and chimpanzee (CH)] matches the order of the cumulative frequency curves from left to right.

Fig. 2.

CCL3L1 dose and risk of acquiring HIV-1. (A to D) Histograms and the cubic-spline smoothed frequency curves (insets) show that the distribution of the CCL3L1 copy numbers (x axis) in HIV+ (red bars or red lines in inset) versus HIV (open bars or black line in inset) individuals is markedly different (χ2 and P values above insets; n = number of individuals in each group). Vertical green arrow indicates the switch point (copy number at which the HIV+/HIV ratio switched from >1 to ≤1) (SOM section 5.1). The cohort of Argentinean children is composed of children exposed perinatally to HIV (4). The HIV+ adults from the indicated ethnic/racial groups (noted on the right) are from the Wilford Hall Medical Center (WHMC) cohort (14) and are compared with a control group from the general population that is matched for ethnicity/race (4). (E to H) Risk of acquiring HIV relative to the population-specific median [horizontal arrow; odds ratio (OR) = 1] was determined by multivariate logistic regression analyses. *, Jewell correction (4); #, CCL3L1 gene copy number; CI, confidence interval; P, significance value. (I) Distribution of CCL3L1 copies in the indicated subject groups (Poisson means and exact 95% CI). #, group number. Arg., Argentinean. SOM section 1.1 provides details of these study groups. In the HIV WHMC cohort, as HAs were categorized with EAs, they were placed within a single group (WHMC EA + HA) and compared with subjects from the HIV+ WHMC cohort that are matched for ethnicity/race. (J) Results of Poisson and logistic regression models in the study groups indicated in (I) (#1 to 6) for the association between CCL3L1 copies and risk of acquiring HIV infection were comparable (table S1B). RR, relative risk.

The duplicated region encoding human CCL3L1 had an ancestral correlate in chimpanzee (Fig. 1 and fig. S3). Together, these results demonstrated that there were significant differences between species and among human populations in the frequency of chemokine gene–containing segmental duplications (Fig. 1, B and C). Despite these differences, the dispersion around the average copy number was similar in both human populations and chimpanzees (Fig. 1B and fig. S1B). On the basis of these observations, we hypothesized that it is not the absolute copy number per se, but rather the gene dose relative to the average copy number in each population that confers HIV/AIDS susceptibility.

CCL3L1 gene dose and HIV/AIDS susceptibility. Several lines of evidence, from four different human populations and in the setting of two different modes of acquiring HIV (i.e., mother-to-child and adult-to-adult), indicated that possession of a low CCL3L1 copy number was a major determinant of enhanced HIV susceptibility among individuals. Individuals with a low CCL3L1 copy number were overrepresented among the HIV+ compared with HIV subjects (shift to the left in Fig. 2, A to D, and figs. S2 and S4). On the basis of the consistency, strength, and significance of the differences in the distribution of CCL3L1 copy numbers in the HIV+ and HIV individuals in each of the cohorts studied, we rejected the null hypothesis of no association between risk of acquiring HIV and CCL3L1 copy number (Fig. 2, A to D, and fig. S2).

We next determined the strength of the association between CCL3L1 copy number and risk of acquiring HIV (Fig. 2, E to H). In our initial analyses, we chose the population-specific median copy number in the uninfected group as a reference point to compute the risk of acquiring HIV (SOM section 5.1). Compared with possession of two copies of CCL3L1, children possessing less than two or more than two copies had significantly higher or lower risks, respectively, of acquiring HIV (Fig. 2E). This association was evident in the analysis of the entire cohort of children with (table S1A) or without (Fig. 2E) adjustments for receipt of zidovudine prophylaxis given to reduce the risk of transmission and for individuals who received no prophylaxis (table S1A). Notably, with each increase in CCL3L1 copy number above the median, there was a dose-dependent, stepwise decrease in the risk for acquiring HIV (Fig. 2E). The findings depicted in Fig. 2, F to H, and those derived from a separate analysis in another cohort of 1133 HIV individuals matched for ethnicity/race (fig. S2), indicated that adults who possessed a CCL3L1 copy number lower than the population-specific median were at a higher risk of acquiring HIV. Thus, in each population, the median number of CCL3L1 copies served as the transition point at which the balance tilted in favor of protection against acquiring HIV.

We also estimated the risk of acquiring HIV across the cline of population-specific high to low CCL3L1 copy numbers (fig. S4). Depending on the study population, each CCL3L1 copy lowered the risk of acquiring HIV by 4.5 to 10.5%, indicating that the population-specific high and low CCL3L1 copy numbers are at different ends of a distribution of HIV susceptibility (SOM section 5.2). Substantiating this, relative to possession of the population-specific high CCL3L1 copy numbers shown in fig. S4, individuals who had a low copy number had between 69 and 97% higher risk of acquiring HIV (fig. S4).

The aforementioned analyses were conducted with logistic regression. Although membership in either the HIV+ or HIV group is not a random outcome, to the extent that these two groups can be thought of as random samples from their respective subsets of a well-defined population, logistic regression on group membership allows estimation of the relative odds of being HIV+ or HIV for two different copy numbers. In this sense, logistic regression permits the determination of the association between CCL3L1 copy number and risk of acquiring HIV. Because the number of CCL3L1 copies followed a Poisson distribution, we also used Poisson regression analysis to determine the association between CCL3L1 copy number and risk of acquiring HIV infection. These analyses assume that copy numbers are conditional on HIV status and show that HIV+ subjects have significantly lower CCL3L1 copy numbers than do HIV subjects (Fig. 2I). Although the true risk of HIV acquisition associated with possession of different CCL3L1 gene copy numbers can theoretically only be estimated from a longitudinal study, in the cohorts we studied, the results of two different statistical approaches demonstrate a strong association between possession of low CCL3L1 copy number and risk of acquiring HIV infection (Fig. 2, E and J, and table S1B).

In addition to influencing HIV acquisition, the number of CCL3L1 copies was associated with variable rates of disease progression (figs. S5 and S6). For example, in the adult HIV+ cohort, a gene dose lower than the overall cohort median or population-specific median was associated with a dose-dependent increased risk of progressing rapidly to AIDS or death (Fig. 3, A and B, and figs. S5). A disease-influencing effect of CCL3L1 dose was not detected in the HIV+ children, suggesting either that the roles of CCL3L1 in HIV+ adults and children differ or that the short follow-up time in the pediatric cohort was insufficient to detect an effect.

Fig. 3.

Disease-influencing and functional phenotypic effects associated with number of CCL3L1 copies. (A and B) Kaplan-Meier (KM) survival curves of the development of AIDS in (A) AAs and (B) EAs from the adult WHMC HIV+ cohort who possess a CCL3L1 copy number equal to or lower than the population-specific median (copy numbers noted adjacent to KM curves). As the population-specific median number of CCL3L1 copies was three and four in HIV+ and HIV AAs, respectively, these two copy numbers were used as the reference genetic strata in (A); the reference group in EAs is two copies. P and relative hazard (RH) below the KM curves were determined by Cox proportional hazard models. Overall log-rank significance values and 95% CI for the RHs are shown in fig. S5. (C) Relationship between number of CCL3L1 copies and percentage of CD4+/CCR5+ cells in unstimulated (open bars) or anti-CD3/CD28–stimulated peripheral blood mononuclear cells (black bars). Numbers inside the bars denote the number of individual blood samples studied with the indicated copy numbers. K-W P, overall Kruskal-Wallis test P value. Vertically oriented numbers indicate P values by the Mann-Whitney test for comparison of possession of zero to two versus three to four or five to seven CCL3L1 copies within each experimental condition. (D to F) Second-order polynomial regression curves show that (D) CCL3/CCL3L1 concentrations in supernatants of freshly isolated peripheral blood mononuclear cells [for units, see (4); n = number of individuals], (E) baseline log viral RNA (viral set point), and (F) monthly CD4+ T cell loss have a threshold-type association with CCL3L1 copies (SOM sections 4.5 to 4.7). (D) and (E) depict medians (±1.7 SD of medians), and (F) depicts 95% CI around the point estimates of the regression coefficients obtained by the General Estimating Equations (GEE) method (4). P linear and quadratic (quad) indicate significance values for the linear and quadratic terms in the polynomial regression equation, respectively. (G to L) KM curves of the development of AIDS in HIV+ AAs (red) and EAs (green) who possess a similar or dissimilar CCL3L1 copy number. The disease-influencing effects associated with possession of [(G) and (H)] median, (I) half-median, and (J) low/null CCL3L1 doses were similar in EAs and AAs. However, the disease-influencing effects of possession of (K) two copies in AAs (half-median dose in HIV AAs) and EAs (median dose) or (L) three copies in AAs (median in HIV+ AAs) and one copy in EAs (half-median in EAs) were not equivalent [see (A) regarding differences in median copy numbers in HIV and HIV+ AAs]. Numbers adjacent to the population designators AA and EA indicate the number of copies (e.g., AA4 indicates four copies in AAs). P values indicate significance value by log-rank test. =, >, or < indicates the direction of the associated effects. (M and N) Direction and magnitude of the rate of change in CD3+, CD4+, and CD8+ T cell counts are similar in HIV+ EAs and AAs who possess a CCL3L1 copy number equal to or lower than the population-specific median (error bars indicate 95% CI; table S2). (O and P) Results of discrete-time Markov modeling of the evolution of changes in the frequency distribution of CCL3L1 copy numbers in infinite-sized AA and EA cohorts over 15 years (SOM section 4.8). Numbers adjacent to the curves indicate CCL3L1 copy numbers. (Q) Schema of phenotypic equivalency of the risk of acquiring HIV and disease-influencing effects of population-specific CCL3L1 doses in EAs and AAs.

Mechanistic links between CCL3L1 dose and HIV/AIDS susceptibility. Increasing CCL3L1 copy number was positively associated with CCL3/CCL3L1 secretion and negatively associated with the proportion of CD4+ T cells that express CCR5 (Fig. 3, C and D) (2). Additionally, there was a dose-dependent association between CCL3L1 copy number and the viral set point and rate of change in CD4+ T cell counts, two well-established predictors of clinical outcome (5); low CCL3L1 doses were associated with a higher viral set point and greater subsequent T cell loss (Fig. 3, E and F). These relationships might explain the association between CCL3L1 gene dose and risk of acquiring HIV and disease progression given that (i) chemokines are thought to mediate their HIV-suppressive activity by steric blocking of the interaction between glycoprotein (gp) 120 and CCR5 or ligand-mediated internalization of CCR5, reducing its availability for use by gp 120 (3), and (ii) high CCR5 ligand and/or low CCR5 receptor expression represents a correlate of HIV/AIDS protection (612).

Phenotypic equivalency of population-specific CCL3L1 gene doses. Human populations differ in their CCL3L1 gene content (Fig. 1). Accordingly, it was important to determine whether an absolute CCL3L1 copy number (e.g., two copies) was associated with similar transmission- and/or disease-influencing phenotypic effects in different populations. To do so, we compared the associated phenotypic effects of similar and dissimilar CCL3L1 copy numbers in HIV+ EAs and AAs (Fig. 3, G to N), and the change in the frequency distribution of copy number in these two populations over time (Fig. 3, O and P). The findings indicated that in HIV+ EAs and AAs, the CCL3L1 copy numbers that corresponded to the population-specific median, half-median, and low doses (i) were associated with comparable rates of disease progression or changes in CD3+, CD4+, or CD8+ T cell counts (Fig. 3, G to N, and table S2), and (ii) had similar trajectories with respect to the changes in their distribution profiles over time (Fig. 3, O and P, and figs. S7 and S8). By contrast, possession of two CCL3L1 copies (i.e., the median and half-median gene dose in EAs and AAs, respectively) was associated with differing rates of disease progression (Fig. 3K). Consistent with this finding, the trajectories of the change in the frequency distribution of individuals possessing two CCL3L1 copies differed over time: increasing in HIV+ EAs, but declining in HIV+ AAs (Fig. 3, O and P). These findings, together with those shown in Fig. 2 and SOM section 5.1, collectively support the concept that different CCL3L1 gene doses among populations are associated with phenotypically similar effects (Fig. 3Q). They also imply that the phenotypic effects associated with CCL3L1 gene dosage cannot be estimated by knowing only the absolute CCL3L1 copy number. This value, in any given individual, is meaningful only if compared with the distribution of CCL3L1 copies in the geographic ancestral population of the given individual (SOM section 5.1).

Distribution of CCL3L1 gene copies under HIV selective pressure. The association between CCL3L1 gene dose and HIV/AIDS susceptibility in adults (Figs. 2 and 3, A and B) predicts that the following pattern should be discernable in a prospective longitudinal cohort in which subjects are recruited at an early stage of infection. Initially, the HIV+ cohort will be enriched for individuals with CCL3L1 copy numbers lower than the population-specific median. Over time, the prevalence of these individuals will decrease because of their rapid progression to AIDS/death. As a result, the prevalence of HIV+ subjects with CCL3L1 copy numbers equal to or greater than the population-specific median will increase. Thus, with increasing follow-up times, the distribution of CCL3L1 copies will begin to resemble that found in HIV subjects. The value of testing this prediction is that it combines into a single analytical model the analyses of (i) the susceptibility to infection in individuals with different numbers of CCL3L1 copies, and (ii) the time to equilibrium between the virus and CCL3L1 genotype–dependent events in the infected host. Our results are consistent with these predictions (Fig. 3, O and P, and figs. S7 to S9). These observations suggest that infection with HIV-1 can exert a negative selective pressure on individuals with low copy numbers that, depending on the strength of this effect in the general population, could change the population-specific distribution of CCL3L1 copy number.

CCL3L1 dose and CCR5 genotypes in HIV/AIDS susceptibility. We and others have shown that CCR5 haplotypes that include CCR5 promoter polymorphisms as well as coding polymorphisms in CCR2 (CCR2-V64I) and CCR532) influence the risk of acquiring HIV and the rate of disease progression (1215). However, CCR5 is part of a complex system in which virus interacts with CCR5 and CCR5 interacts with various ligands. Thus, if gene-gene interactions are not considered, these interactions might complicate analysis of the in vivo contributions of CCR5 genotypes. This concern is made all the more apparent by the observation that CCR5 protein expression levels are influenced not only by variants in CCR5 (16, 17), but also by CCL3L1 (Fig. 3C). Thus, virus × CCR5 × CCL3L1 interactions in vivo and the phenotypic effects associated with CCR5 genotypes could depend, in part, on the genetic background conferred by CCL3L1 dose. To test this hypothesis, we determined the phenotypic effects attributable to CCL3L1 gene dose alone, CCR5 haplotype pairs (genotypes) alone, and their combined effects.

The HIV+ adult cohort was stratified into four mutually exclusive genetic risk groups (GRGs) based on possession of a population-specific low or high number of CCL3L1 copies (CCL3L1low or CCL3L1high) and disease-accelerating, i.e., detrimental (det) or nondetrimental (non-det) CCR5 genotypes (CCR5det or CCR5non-det) (Fig. 4A). Of the four GRGs, CCL3L1highCCR5non-det and CCL3L1lowCCR5det were at the two extremes of HIV/AIDS susceptibility (Fig. 4, B to I). Relative to possession of CCL3L1highCCR5non-det, CCL3L1lowCCR5det was associated with a ≥ threefold greater risk of progressing rapidly to 8 of 12 AIDS-defining illnesses (Table 1). By contrast, the CCL3L1highCCR5det and CCL3L1lowCCR5non-det genotypes were associated with a ≤ threefold higher risk of progressing to 3 or 4 of these 12 illnesses, respectively (Table 1).

Fig. 4.

Risk of acquiring HIV and disease-influencing effects associated with variations in CCL3L1 and/or CCR5. (A) Genetic stratification system (SOM section 3). In each population (popln), CCL3L1 dose and CCR5 genotypes were dichotomized on the basis of whether they were associated with an accelerated disease course (tables S3 to S5). CCL3L1low and CCL3L1high denote copy numbers < or ≥ population-specific median, respectively (table S3). CCR5det and CCR5non-det denote population-specific, disease-accelerating, i.e., detrimental (det), or nondetrimental CCR5 genotypes, respectively (table S4). Compared with possession of CCL3L1high or CCR5non-det, CCL3L1low or CCR5det was associated with an accelerated disease course (fig. S10). These dichotomized compound genotypes were used to stratify the cohort further into four mutually exclusive GRGs, which reflected (i) the independent disease-accelerating effects associated with population-specific low CCL3L1 gene doses (CCL3L1lowCCR5non-det, orange) or detrimental CCR5 genotypes (CCL3L1highCCR5det, green); or (ii) their combined effects (CCL3L1lowCCR5det, red), all relative to CCL3L1highCCR5non-det (blue). This color code is used in the rest of the panels to indicate the four CCL3L1/CCR5 GRGs. (B) CD4+ and (C) CD8+ T cell changes associated with the GRGs are depicted as 95% CI around the point estimates of the regression coefficients obtained by the GEE method (4). (D) Baseline log viral RNA [viral set point; median (±1.7 SD of the median)] associated with the GRGs. P values reflect significance values for differences between CCL3L1highCCR5non-det and CCL3L1lowCCR5det by Student's t test in (B) and (C) and the Mann-Whitney test in (D). (E and F) KM curves of the development of AIDS in EAs and AAs from the entire (E) or seroconverting portion (F) of the HIV+ adult cohort after stratifying for the GRGs. (Inset) Pie charts depicting frequency distribution of the GRGs. (G) Proportions of individuals within each GRG that developed AIDS. (H and I) Association of indicated GRGs and risk of acquiring HIV infection in (H) adults or (I) children exposed perinatally to HIV. ORs are lowest in GRGs that lack CCL3L1low (green). (J and K) Changes in the frequency distributions of the GRGs and test of linear trend for individuals with varying follow-up times. (L) Differences in the frequency distribution of GRGs between HIV+ and HIV adults. In (H) and (J), to ensure appropriate ethnic/racial matching for the comparisons of the frequency distributions between HIV+ and HIV individuals, these analyses are for the EA, AA, and HA portions of the infected adult cohort (tables S3 and S4; tables also show the genotypes used for the pediatric cohort in (I) (4).

Table 1.

Risk of AIDS-defining illness with CCL3L1/CCR5 GRGs. The reference GRG for statistical analysis is CCL3L1highCCR5non-det (RH = 1). The AIDS-defining illnesses with sufficient events for statistical analyses recorded in the adult HIV+ cohort are shown. CMV, cytomegalovirus; HAD, HIV-associated dementia; MAC, Mycobacterium avium complex; PCP, Pneumocystis carinii pneumonia; PML, progressive multifocal leukoencephalopathy; n, number of individuals with the indicated AIDS-defining illness; values in bold and italic indicate significant association.

AIDS-defining illness nCCL3L1highCCR5detCCL3L1lowCCR5non-detCCL3L1lowCCR5det
RH 95% CI P RH 95% CI P RH 95% CI P
CMV infection 100 1.53 0.71-3.30 0.278 1.60 1.00-2.58 0.051 6.21 3.63-10.63 2.7 × 10-11
Cryptococcosis 33 3.27 0.98-10.87 0.053 2.46 1.00-6.02 0.048 8.11 2.93-22.46 5.6 × 10-5
Cryptosporidiosis 24 1.21 0.27-5.47 0.802 1.21 0.49-3.00 0.686 1.63 0.36-7.37 0.526
HAD 54 2.05 0.82-5.13 0.126 1.65 0.87-3.11 0.124 3.18 1.33-7.60 0.009
Herpes simplex 26 1.78 0.50-6.41 0.375 1.22 0.49-3.04 0.668 1.66 0.36-7.53 0.513
Histoplasmosis 20 3.32 0.83-13.30 0.090 2.81 1.02-7.74 0.045 1.56 0.19-13.01 0.682
Kaposi sarcoma 74 1.76 0.76-4.05 0.186 1.66 0.96-2.86 0.069 3.86 1.90-7.83 2.0 × 10-4
Lymphoma 37 2.87 1.10-7.48 0.031 1.42 0.66-3.08 0.369 3.38 1.21-9.43 0.020
MAC 92 2.22 1.09-4.55 0.029 1.73 1.05-2.87 0.032 5.13 2.79-9.45 1.5 × 10-7
PCP 196 2.13 1.33-3.42 0.002 1.71 1.22-2.39 0.002 2.95 1.84-4.75 7.8 × 10-6
PML 18 1.72 0.36-8.10 0.494 1.27 0.44-3.67 0.657 2.41 0.51-11.43 0.268
Toxoplasmosis 27 1.49 0.32-6.91 0.610 1.69 0.67-4.25 0.268 5.34 1.77-16.07 0.003

The trajectory of the frequency distribution profiles of the four CCL3L1/CCR5 GRGs in individuals with varying follow-up times were also revealing in that they closely paralleled those described previously for a variable number of CCL3L1 copies alone (compare Fig. 4J with Fig. 3, O and P, and fig. S7 to fig. S9). Thus, significant changes occurred only in the frequencies of the two GRGs that contained CCL3L1low and CCL3L1highCCR5non-det, such that over time the distribution of the GRGs in surviving HIV+ subjects approached ever closer to the values observed in the HIV population (Fig. 4, J to L).

Taken together, in the context of a well-characterized prospective cohort composed of HIV+ EAs and AAs, the CCL3L1/CCR5-based genomic signature for HIV/AIDS susceptibility was CCL3L1lowCCR5det > CCL3L1lowCCR5non-detCCL3L1highCCR5det > CCL3L1highCCR5non-det. This observation implied that CCL3L1low may have a stronger effect than disease-accelerating, detrimental CCR5 genotypes in influencing HIV/AIDS pathogenesis in these two populations. Additionally, these findings suggest that a population-specific low CCL3L1 dose provides a permissive genetic background for the full expression of the phenotypic effects associated with detrimental CCR5 genotypes. This was apparent because (i) relative to genotypes that contained only CCR5det, those that contained CCL3L1low with or without CCR5det were associated with a higher risk of acquiring HIV (compare green with orange or red color-coded GRGs in Fig. 4, H and I); and (ii) the maximal disease-accelerating effects associated with detrimental CCR5 genotypes occurred mainly in individuals who also possessed a low number of CCL3L1 copies relative to the population-specific average (compare Kaplan-Meier plots for CCL3L1highCCR5det and CCL3L1lowCCR5det in Fig. 4, E and F).

Public health impact of variations in CCL3L1 and CCR5. In the populations examined, up to 42% of the burden of infection and ∼30% of the accelerated rate of progression to AIDS were attributable to variations in CCL3L1/CCR5 (black bars in Fig. 5 and fig. S11). The largest contributor to the burden of HIV/AIDS was possession of a population-specific low CCL3L1 copy number (Fig. 5, compare combination of red and orange to green bars, and fig. S11). These findings suggest that the contribution of CCL3L1 copy number is comparable to or more than that of the CCR5 genotype in influencing the epidemiology of HIV in the populations examined. These results also substantiate the observation that the disease-accelerating effects associated with variation in CCR5 depend, in part, on the genetic background of CCL3L1 copy number.

Fig. 5.

Attributable fractions of CCL3L1/CCR5 GRGs for risk of acquiring HIV (vertical, mother-to-child; horizontal, adult-to-adult) and rate of disease progression relative to CCL3L1highCCR5non-det in the indicated clinical settings. Vertical bars indicate the point estimate, whereas error bars represent the 95% CI around the point estimate of the attributable fraction.

Discussion. These findings have five major implications. First, they provide a precedent for a link between segmental duplication events leading to changes in the dose of an immune response gene and variability in the phenotypic response to an infectious disease. Recent human-nonhuman primate comparative genomic analyses have led to the prediction that genes embedded within segmental duplications might have enhanced the ability of humans to adapt to their environments (1, 18). Our findings support this prediction.

Second, CCL3L1 gene dose is a previously unrecognized means of buffering against the risk of HIV infection and/or disease progression in the populations examined. CCL3L1 gene doses lower than the population-specific average provide a genetically “unbuffered” state with respect to the risk of HIV/AIDS susceptibility. However, it is important to emphasize that it is not the absolute gene copy number per se, but the copy number within the overall genetic context that confers phenotypic expression. This genetic context varies among populations as a result of their different demographic and evolutionary histories. Thus, an individual's specific CCL3L1 gene dose and CCR5 genotype are associated with susceptibility to HIV/AIDS, but only when viewed in the context of that person's geographical ancestry (Fig. 3Q) (14).

Third, within the populations examined, the Bradford-Hill criteria (19) for causality between CCL3L1 dose and risk of acquiring HIV were met (SOM section 5.3). Thus, by analogy to the genetic studies that established the paradigm of “no CCR5–no HIV-1 infection,” the current findings establish that of “CCL3L1low–enhanced HIV/AIDS susceptibility.” These findings provide strong genetic underpinnings for the substantial body of evidence that CCR5 ligands play an important anti–HIV-1 role in vivo (20). Paradoxically, they also indicate that a network of HIV-suppressive CCR5 ligands (e.g., CCL5) cannot fully compensate for the functional state conferred by CCL3L1low. Therefore, CCL3L1-mediated immune responses may be required to thwart HIV infection and the complications that occur during HIV-induced immune suppression.

Fourth, CCL3L1 gene dose may be an important genetic correlate of vaccine responsiveness. A comparative analysis of the immunological phenotype linked to the GRGs associated with the extremes of susceptibility (i.e., CCL3L1lowCCR5det and CCL3L1highCCR5non-det) could provide key insights into the immune correlates of an effective vaccine. This stems from several vaccine studies in simian models showing that CCR5 ligand production is a true predictor of protection and animals that produce higher levels of chemokines prevaccination exhibit greater protection (2022).

Finally, and of broader import, 5% of the human genome contains duplicated sequences enriched for genes involved in immunity (1), and some of these genes have dosage effects. Thus, the present findings provide both a precedent and a framework for elucidating their relationship to human diseases.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1101160/DC1

Materials and Methods

SOM Text

Figs. S1 to S16

Tables S1 to S7

References and Notes

References and Notes

View Abstract

Navigate This Article