Genetic Restriction of HIV-1 Infection and Progression to AIDS by a Deletion Allele of the CKR5 Structural Gene

The chemokine receptor 5 (CKR5) protein serves as a secondary receptor on CD4+ T lymphocytes for certain strains of human immunodeficiency virus-type 1 (HIV-1). The CKR5 structural gene was mapped to human chromosome 3p21, and a 32-base pair deletion allele (CKR5Δ32) was identified that is present at a frequency of ∼0.10 in the Caucasian population of the United States. An examination of 1955 patients included among six well-characterized acquired immunodeficiency syndrome (AIDS) cohort studies revealed that 17 deletion homozygotes occurred exclusively among 612 exposed HIV-1 antibody-negative individuals (2.8 percent) and not at all in 1343 HIV-1-infected individuals. The frequency of CKR5 deletion heterozygotes was significantly elevated in groups of individuals that had survived HIV-1 infection for more than 10 years, and, in some risk groups, twice as frequent as their occurrence in rapid progressors to AIDS. Survival analysis clearly shows that disease progression is slower in CKR5 deletion heterozygotes than in individuals homozygous for the normal CKR5 gene. The CKR5Δ32 deletion may act as a recessive restriction gene against HIV-1 infection and may exert a dominant phenotype of delaying progression to AIDS among infected individuals.

Streller and G. Wingsle, Planta 192, 195 (1994). 20. Fast neutron-mutagenized Isdl seed was grown under SD conditions. Ten M2 seed lots from -200 Ml individuals were screened for wild-type plants after shift to LD conditions at 2 weeks of age. Lesion initiation in young seedlings is lethal. Loci defined by these suppressor mutants are named PHOENIX (phx), after the mythological bird risen from its own ashes. M3 progeny of survivors were retested, and suppressors were categorized by segregation of the Isdl phenotype in Fl individuals and F2 populations of backcrosses to Ws-0 and Isdl. These analyses established that phx21 is unlinked to Isdl, recessive to its wild-type allele, and has no obvious phenotype in combination with LSD1 (R. A. Dietrich et aL., unpublished data). NBT staining after shift to LD conditions was also abolished (9). 21 Genetic Restriction of HIV-1 Infection and Progression to AIDS by a Deletion Allele of the CKR5 Structural Gene The chemokine receptor 5 (CKR5) protein serves as a secondary receptor on CD4+ T lymphocytes for certain strains of human immunodeficiency virus-type 1 (HIV-1). The CKR5 structural gene was mapped to human chromosome 3p21, and a 32-base pair deletion allele (CKR5A32) was identified that is present at a frequency of -0.10 in the Caucasian population of the United States. An examination of 1955 patients included among six well-characterized acquired immunodeficiency syndrome (AIDS) cohort studies revealed that 17 deletion homozygotes occurred exclusively among 612 exposed HIV-1 antibody-negative individuals (2.8 percent) and not at all in 1343 HlV-1-infected individuals. The frequency of CKR5 deletion heterozygotes was significantly elevated in groups of individuals that had survived HIV-1 infection for more than 10 years, and, in some risk groups, twice as frequent as their occurrence in rapid progressors to AIDS. Survival analysis clearly shows that disease progression is slower in CKR5 deletion heterozygotes than in individuals homozygous for the normal CKR5 gene. The CKR5A32 deletion may act as a recessive restriction gene against HIV-1 infection and may exert a dominant phenotype of delaying progression to AIDS among infected individuals.
In all well-characterized epidemics there are individuals in the population that respond differently to the infectious agent (1,2). Although resistance to infection is the most common variable phenotype, variation in disease outcomes has also been observed. Epidemiologic studies have shown that inherited factors are involved in the risk of mortality from infectious agents (3,4). The HIV-1 epidemic presents a critical challenge for the application of current genetic techniques to the study of host genetic variation for infection and susceptibility to infection. This problem is confounded in the studies of HIV-1 by the rapid rate of evolution of the virus (5-7). However, a number of groups have shown that specific alleles of the human lymphocyte antigen (HLA) locus are associated with different rates of progression from infection to an AIDS diagnosis (8). Yet little evidence for non-HLA loci regulating HIV-1 infection or AIDS progression has been reported, although it does seem likely that other host genetic factors would play a role in AIDS epidemiology (8,9).
The recent demonstration that the chemokines RANTES, MIP-lox, and MIP-13 act as natural suppressors of HIV-1 infection (10) has focused attention on the role of these chemokines during HIV-1 infection and clinical pathogenesis. Feng et al. tTo whom correspondence should be addressed. symbol CMKBR5) (16), which serves as the principal cellular receptor for RANTES, MIP-lax, and MIP-1, is an efficient coreceptor for macrophage-tropic isolates of HIV-1 (17)(18)(19)(20)(21)(22). The closely related CKR2B and CKR3 molecules can also act as coreceptors for some HIV-1 strains (20,21). In addition, Dragic et al. (17) observed that CD4+ T cells from some HIV-1-exposed individuals who have remained uninfected are relatively resistant to infection, suggesting that a defect in co-receptors or their expression may protect some individuals from infection.
The C-C or 13 chemokine receptors belong to the heterotrimeric GTP-binding protein (G protein)-coupled receptor superfamily (22)(23)(24). The C-C chemokine receptors consist of seven transmembrane domains and typically contain no introns (I12- 14,16). The genes for the C-C chemokines are linked in a cluster on chromosome 17 (24). In contrast, the genes for the receptors are dispersed in the genome as single genes or in small clusters of genes that have related function and sequence (24).
To genetically map the locus encoding fusin and the CKR5 locus, we used genespecific polymerase chain reaction (PCR) primers designed from the sequences of the genes (13,16) to screen a panel of 90 radiation hybrid (RH) DNA samples (25). The RH panel is designed to retain small segments of the human genome in different combinations so that the map location of new markers is implicated by their concordant occurrence in the panel with previously mapped markers ( Fig. 1A)  related interleukin-8 receptor (IL8RA, IL8RB) genes, and distal to the IL-I and IL-I receptor (ILl, ILIR) gene cluster (12). CKR5 maps to chromosome 3p21, very close to the CKR1 gene, which we also physically mapped in this analysis (Fig. iB). CKR2B has been previously shown to be 18 kb from CKR5 (16) and can therefore also be assigned to 3p2l. These gene-mapping assignments provide additional evidence for the occurrence of chemokine receptor genes in small clusters in different regions of the human genome. We identified alterations in the CKR5 gene by amplifying portions of the entire coding region, digesting the fragments with Hinf I, and resolving the fragments by a combined single-stranded conformation polymorphism-heteroduplex analysis approach (26,27). DNA from more than 600 individuals (healthy controls, HIV-1 at-risk seronegatives, HIV-1-infected, non-AIDS, and AIDS patients) was typed by this procedure, and eight molecular genetic variants were identified. One of the alterations occurred in -10% of the individuals, and this allele was sequenced. This variant contains a 32-base pair (bp) deletion (CKR5A32) that causes a frame shift at amino acid 185. After this manuscript was submitted, Samson et al. (28) and Liu et al. (29) also described the same CKR5 deletion mutation. The CKR5A32 allele is nonfunctional both as a chemokine receptor and an HIV-1 co-receptor.
The other seven variants we found were all rare and were observed in .1% of the individuals studied (30).
Lymphoblastoid B cell lines were established for over 1900 participants who are members of six well-characterized longterm cohorts of hemophiliac, homosexual male, and intravenous drug user risk groups (Table 1) (31)(32)(33)(34)(35)(36). Genomic DNA from 156 cell lines derived from Caucasian participants from the DCG and MACS studies was screened by using 170 defined and mapped polymorphic loci (37), including candidate genes (for example, CD4, chemokine SCYAl, HLA-DQA1, TCRA, TCRB, and CKR5), for distortion of allele and genotype frequency among HIV-1-infected versus HIV-1 antibodynegative individuals at risk for exposure to HIV-1. Loci were selected on the basis of available polymorphisms, potential involvement in retroviral infection or pathogenesis, and their genetic location, producing an average 20-centimorgan interval of markers (37,38). The significance level estimated with a G test (39) for each of the 170 loci for the occurrence of genotypic association between HIV-1infected versus HIV-1 antibody-negative individuals is presented in Fig. 2. With the exception of CKR5, none of the loci tested displayed a significant distortion of genotype frequencies among the infected versus uninfected individuals. The genotypic distribution of the two common alleles of CKR5 [normal or wild type (+) and CKR5A32 deletion] in 738 Caucasian homosexual men displays a highly significant (P = 2.0 X 10-5) departure from genotypic equilibrium when frequencies among HIV-1-infected versus uninfected individuals are examined. Although these 170 markers scan only a portion of the genome for HIV restriction loci, the absence of any association at any of the loci suggests that there is not a major effect on HIV-1 infection from genes close to (within 1 cM of) the test markers. The list includes the HIV receptor (CD4), a A chemokine (SCYAI), and several immune function genes. These markers provide negative controls for the result observed with CKR5.
To examine further the role of the CKR5A32 allele in HIV-1 infection, we determined the distribution of alleles and genotypes with genomic DNA from 1955 individuals from the six cohorts listed in Table 1. Each of these studies represents a long-term epidemiological project designed to characterize important variables in HIV-1 infection and disease progression.
All cohort participants are in high-risk groups for HIV-1 infection; namely, hemophiliacs potentially exposed to contaminated clotting factor before HIV-1 screening and viral inactivation, sexually active homosexual men, and intravenous drug users. Participants include HIV-1-exposed seronegative individuals, HIV-1-infected AIDS patients, and HIV-1-infected individuals studied for various periods who have not progressed to clinical AIDS. Among Caucasian participants in AIDS cohorts, the allele frequency of CKR5A32 was 0.115 (n = 1250), whereas the frequency among Caucasians not part of an HIV-1 high-risk group was 0.080 (n = 143). The frequency among African Americans was lower (0.017; n = 620) (40).
There was no significant difference in the allele frequencies of CKR5 between HIV-1-infected and HIV-1 antibody-negative individuals in any of the cohorts (41); however, a dramatic difference became apparent when the CKR5 genotype distribution was examined ( Table 2). There were 17 homozygotes for CKR5A32 found among the individuals tested, all of which were HIV-1 antibody-negative individuals with a high risk for HIV-1 exposure. The association of the A32/A32 genotype with HIV-1-negative status is highly significant overall (G = 35.0, P = 2.5 x 10-8; Table 2), as well as in each of three individual cohorts (MACS, SFCC, and MHCS, Table 2) (42). Hence, the CKR5A32 allele appears to con-fer a recessive phenotype that is associated with resistance to HIV-1 infection and antibody production.
Although we did not find individuals homozygous for CKR5A32 among 1343 HIV-1-infected patients, there was a sufficient number of heterozygous (+/A32) infected patients (n = 195) for us to examine the cohorts for an association between CKR5 genotype and different rates of progression to AIDS. The homosexual cohorts, but not the hemophilia cohorts, showed greater than twice the percentage of heterozygotes among long-term nonprogressors compared with rapid progressors (Fig. 3A) (43).
Although there are differences in ascertainment and disease group definition between these groups (for example, there are fewer AIDS cases in the hemophilia cohorts), the frequency of the +/A32 genotype was not significantly different between rapid progressors and long-term nonprogressors in the hemophilia cohorts when different AIDS endpoints [for example, 1987 AIDS definition (44), CD4+ T cell counts .200, or death by AIDS] were used or the "middle groups" of patients with intermediate criteria were subtracted. The differential response of hemophiliacs versus homosexual men may be related to different routes of transmission, to exposure levels, or to viral load among individuals in different risk groups. Hemophilia patients received large doses of HIV-1-contaminated clotting factors by intravenous injection, whereas sexual transmission would involve HIV-1 infection of mucosal epithelium.
Survival analyses using the clinical 1992 AIDS definition (which includes HIV-1 infection plus either AIDS-defining illness or a decline of CD4+ T cell numbers below 200) (45) were performed for each cohort and for combined cohorts comparing the CKR5 genotypes (+/+ and +/A32). The results demonstrate that +/A32 heterozygotes have a delayed progression to AIDS compared with CKR5 +/+ homozygotes (X2 = 8.1, P = 0.0045) (Fig. 3, B and C, and Table 3). After correction for multiple tests, this result is still significant, and the same result is also found when the stricter 1987 AIDS definition (HIV-1 infection plus AIDS-defining pa-   Table 2. CKR5 genotype distribution among HIV-1-seropositive and HIV-1-compared with overall cohort genotypic frequencies are listed. G tests using seronegative individuals in the same risk group. G tests (39) for departure of only Caucasian individuals are given in parentheses in the G and P columns. three observed CKR5 genotypes within HIV-1-infected versus uninfected NC, no Caucasians.   thologies) is used ( Table 3). The contribution of the individual cohorts to the combined analysis in Fig. 3B is shown in Table 3. The same trend (longer survival of +/A32 individuals) was observed in all cohorts except DCG, which only contributes 43 pa- Further studies of homozygous A32/A32 individuals to determine whether they have a memory T cell response to the virus, which would indicate that they were infected and mounted a successful immune response, would be enlightening. It is possible that A32/A32 individuals could become infected by certain strains of HIV-1 (for example, syncytium-forming viruses or isolates adapted to CD4fusin receptor entry), or by certain routes of infection (for example, transfusion). However, we have observed A32/A32, HIV-1-seronegative individuals in homosexual, hemophiliac, and intravenous drug-use cohorts, suggesting that this genotype confers broad protection against infection.
In addition to CKR5A32, we also found unique SSCP conformers in other patients, several of whom were long-term nonprogressors (30). We speculate that at least some of these alleles also disrupt CKR5 function and inhibit the spread of HIV-1 or progression to AIDS. The entire coding region of CKR5 should be screened in additional HIV-1-seronegative, long-term nonprogressors and in rapid progressors to identify other CKR5 variants. The CKR2B and CKR3 proteins are also capable of acting as receptors for macrophage-tropic HIV-1 strains (19,20). Alterations in these genes, as well as in fusin, may also influence HIV-1 infection or progression.
The identification of CKR5 as a critical molecule for HIV-1 infection suggests alternative antiviral therapies. Native or altered forms of the CKR5 ligands (RAN-TES, MIP-1co, MIP-113) could potentially block or delay infection. Alternatively, fragments of the HIV-1 env protein product that bind to CKR5 may be effective in SCIENCE * VOL. 273 * 27 SEPTEMBER 1996 competing for HIV-1 binding sites on CKR5+ cells. Drugs or gene-targeting constructs that down-regulate or inactivate CKR5 may have therapeutic value. It is also possible that transplantation of bone marrow stem cells from a A32/A32 donor could have therapeutic benefit. Anti-CKR5 therapies may augment the effectiveness of other anti-HIV-1 compounds already in use.
A large difference in the frequency of the CKR5A32 allele was observed between Caucasians (0.1 1) and African Americans (0.017). If widespread screens of African racial groups affirm the absence of the CKR5A32 allele as has been reported in a survey of 124 Africans (29), it may be that CKR5A32 is a recent mutation that occurred on the Caucasian lineage subsequent to divergence from the African-Caucasian ancestor estimated to have occurred 150,000 to 200,000 years ago (48). The frequency observed in African Americans could be entirely due to admixture from Caucasian gene flow in the New World, estimated to be as high as 30% (49). The relatively high prevalence of the inactivating CKR5A32 allele is suggestive of an historic selective pressure among Caucasians, perhaps by another pathogenic virus or parasite, that also used the CKR5 receptor as an entry point. Such historic increases in recessive disease mutations due to epidemiologic agents have been suggested for other hereditary diseases, including sickle cell anemia, thalassemia, and lysosomal storage diseases (50). Liu et al. demonstrated that the wellcharacterized EU2 and EU3 patients (29) are CKR5A32 homozygotes and that no functional CKR5 protein is present on the cell surface. In addition, they showed that cells from CKR5A32 homozygotes fail to respond to MIP-lot in vitro. Samson et al. found no CKR5A32 homozygotes among 723 HIV-1-infected individuals and a rather low frequency for the CKR5i32 allele (0.054) in HIV-1-infected patients, leading them to suggest that +/A32 heterozygotes may be less susceptible to infection than CKR5 +/+ individuals (28). Our results ( Table 2) did not reveal a difference in CKR5A32 allele frequency in HIV-1-infected (0.11) compared with HIV-1-negative exposed (0.12) individuals. Our data suggest that in the heterozygous state, the CKR5A32 allele does not markedly affect susceptibility to infection but does postpone progression to AIDS in infected patients.