First-Class Control of HIV-1
Andrew J. McMichael1 and E. Yvonne Jones2
Once the major histocompatibility complex (MHC) was established as the most polymorphic mammalian genetic system, searches began for human MHC [or human leukocyte anti- gen (HLA)] associations with infectious disease resistance. Such findings have been rare, though, possibly because susceptibility genes have been deselected over evolution. But the appearance of completely new infec- tions caused by viruses, such as HIV-1, has opened opportunities to look at such selec- tion in the MHC as it happens. On page 1551 of this issue, the International HIV Control- lers Study (2), demonstrates the central role of HLA class I in controlling HIV-1 infection.
Earlier studies had demonstrated the influence of HLA type on slow versus rapid AIDS progression (3). Protection is thought to be mediated by strong T cell (CD8 sub- type) responses to viral peptides presented
The role of infectious disease in driving human genetic variations (poly-morphisms) was first clearly espoused by HLA molecules (4). Polymorphic killer inhibitory immunoglobulin-like receptors (KIRs) expressed by natural killer cells are also associated with good virus control, but only in the presence of the HLA molecules that they bind, which include those associ- ated with slow AIDS progression (HLA types B*57 and B*27) (5, 6).
The first genome-wide association study (GWAS) of an infectious disease (7, 8) identified striking associations between three single-nucleotide polymorphisms (SNPs) and the low level of viremia (which predicts slower disease progression) in the asymptom- atic stage of HIV-1 infection. All three SNPs mapped close to the highly polymorphic HLA A, B, and C class I loci. The strongest associa- tions were in the HCP5 gene near HLA-B, a marker for HLA B*57, and in a noncoding site 35 kb upstream of HLA-C.
The International HIV Controllers Study explored the role of the HLA system in con- trolling HIV-1 infection in more detail. A large cohort of 3622 patients, untreated with antiretroviral drugs, was divided into those who controlled HIV-1 infection well and those who progressed more rapidly to AIDS. Patients of European ancestry showed asso- ciations with HCP5 and HLA-C, SNPs previously associated with a low level of viremia (7, 8). The study then referred to an earlier GWAS of diabetes (9) in which 2767 HLA- typed individuals were mapped for a similar set of SNPs. By imputing the HLA type (10) in the HIV-1–infected cohort, the study found that HLA types B*57:01, B*27:05, and B*14 were protective, whereas types B*35 and C*07 were associated with progression to AIDS. These results confirm and extend pre- vious HLA typing data in HIV-1 infection (3). Examination of all the polymorphic amino acids in HLA class I revealed that amino acid positions 67, 70, and particularly 97, in the peptide binding groove of HLA class I (see the figure) had stronger statistical associa- tions with HIV-1 protection than the whole HLA molecules. Valine, asparagine, and tryp- tophan occur at position 97 in the B*57:01, B*27:05, and B*14 protective haplotypes, respectively, whereas arginine occurs in risk HLA molecules B*35, B*53, and C*07. This implies that these residues influence control of HIV-1 infection independently of the rest of the HLA molecule.
The HLA class I binding groove pres- ents peptides of 8 to 10 amino acids with residue side chains at positions 2 and the car- boxyl terminus serving as anchors, respectively inserting into pockets B and F within the groove. These anchor points are usually supplemented by interactions in the center of the peptide and binding groove. The central- region interactions with different peptides vary greatly, resulting in very different pep- tide conformations. How might individual HLA residues within the binding groove con- fer the HIV protective effect?
Although residue 67 contributes to the specificity of the B pocket, none of the res- idues highlighted by the SNP associations influence the binding characteristics of the F pocket. Indeed, the gross peptide bind- ing motif determined by the B and F pock- ets is rather similar for the protective B*57 and the nonprotective B*35; both methionine 67 (in B*57) and phenylalanine 67 (in B*35) are large hydrophobic side chains contribut- ing to shallow B pockets that bind relatively small side chains at position 2, whereas the F pockets for both alleles bind aromatic resi- dues at the carboxyl terminus. Residue 97 is centrally placed in the floor of the binding groove. Arginine is the most common resi- due at this position and present in the HLA molecules that confer risk for HIV-1 disease. It provides flexibility, in the presence of ser- ine at position 116 (as in the risk HLA types), potentially allowing the binding groove to adapt to fit more peptides (11). This fits with the idea that protective HLA molecules bind fewer peptides, which delete fewer self-reac- tive T cells in the thymus, giving virus-spe- cific T cells broader fine specificity (12). This could reduce virus options to escape by muta- tion (13, 14).
Whatever the mechanism, HIV-1 pep- tides bound to the HLA molecule are critical in protecting against AIDS. Much of the pro- tective effect of B*57 is probably due to ser- endipitous binding of immunogenic peptides that come from the conserved HIV-1 Gag p24 protein. Gag forms the viral capsid structure, and mutations in this protein that are selected by T cells can weaken the virus (15). This could work in concert with the broader T cell specificity to limit escape.
Natural killer cells might also contribute to protection. They express the genetically poly- morphic receptors KIR3DL1 and KIR3DS1, which bind to the peptide-HLA complex with a footprint that covers HLA B residues 77, 80, 81, and 84 as well as the carboxyl termi- nus of the bound peptide. Therefore, the HLA residues associated with effective control of HIV-1 might influence the action of nat- ural killer cells. Weak genetic associations between HIV-1 control and the polymorphic residues 80, 77, and 81 may reflect the relative rarity of the relevant KIR3DL1 andKIR3DS1 receptors in the population. The two GWAS studies (2, 7, 8) demon- strate extreme dominance of the HLA class I region in the genetic association with control of HIV-1 infection. This accounts for nearly 20% of the observed variability in HIV-1 con- trol. Given that both protective and highly sus- ceptible alleles are quite rare, the importance of T cell and/or natural killer cell control is extraordinary and rewards massive efforts to determine the structure and function of these genes and the molecules they encode. The question now is whether those not blessed with these protective genes can be immuno- logically pushed to make similar responses.
References 1. J. Lederberg, Genetics 153, 1 (1999). 2. The International HIV Controllers Study, Science 330,
1551 (2010); 10.1126/science.1195271. 3. M. Carrington, S. J. O’Brien, Annu. Rev. Med. 54, 535
(2003). 4. H.Streecketal.,J.Virol.81,7725(2007). 5. M. P. Martin et al., Nat. Genet. 31, 429 (2002).
6. M.P.Martinetal.,Nat.Genet.39,733(2007). 7. J.Fellayetal.,PLoSGenet.5,e1000791(2009). 8. J.Fellayetal.,Science317,944(2007). 9. W.M.Brownetal.,DiabetesObes.Metab.11(suppl.1),
2 (2009). 10. S.Leslie,P.Donnelly,G.McVean,Am.J.Hum.Genet.82,
48 (2008). 11. K. J. Smith et al., Immunity 4, 215 (1996). 12. A.Kosmrljetal.,Nature465,350(20). 13. W. Fischer et al., PLoS ONE 5, e12303 (2010). 14. E. L. Turnbull et al., J. Immunol. 176, 6130 (2006). 15. J. Martinez-Picado et al., J. Virol. 80, 3617 (2006).