Correspondence Aikichi Iwamoto, Division of Infectious Disease, Advanced Clinical Research Center, Institute of Medical Science, University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo, 108-8639, Japan. Tel: +81 3 5449 5359; fax: +81 3 6409 2008; email: firstname.lastname@example.org
HLA class I allele types have differential impacts on the level of the pVL and outcome of HIV-1 infection. While accumulations of CTL escape mutations at population levels have been reported, their actual impact on the level of the pVL remains unknown. In this study HLA class I types from 141 untreated, chronically HIV-1 infected Japanese patients diagnosed from 1995–2007 were determined, and the associations between expression of individual HLA alleles and level of pVL analyzed. It was found that the Japanese population has an extremely narrow HLA distribution compared to other ethnic groups, which may facilitate accumulation of CTL escape mutations at the population level. Moreover while they uniquely lack the most protective HLA-B27/B57, they commonly express the alleles that are protective in Caucasians (A11:10.4%, A26:11.55%, B51:8.6% and Cw14:12.7%). Cross-sectional analyses revealed no significant associations between expression of individual alleles and the level of the pVL. The patients were then stratified by the date of HIV diagnosis and the analyses repeated. It was found that, before 2001, B51+ individuals displayed significantly lower pVL than the other patients (median: 5150 vs. 18 000 RNA copies/ml, P= 0.048); however thereafter this protective effect waned and disappeared, whereas no changes were observed for any other alleles over time. These results indicate that, at a population level, some HLA alleles have been losing their beneficial effects against HIV disease progression over time, thereby possibly posing a significant challenge for HIV vaccine development. However such detrimental effects may be limited to particular HLA class I alleles.
polymerase chain reaction-sequence specific oligonucleotide probes
plasma virus loads
HIV-1 is the causative agent for AIDS. Since the discovery of HIV-1 in 1983, although a myriad of studies focusing on the immunopathogenesis of HIV-1 infection have been conducted, a number of questions remained unanswered, hampering development of HIV/AIDS vaccines.
As the HIV-1 epidemic has continued, it has become evident that the rate of decline in CD4+ T cells varies considerably between infected people, and that untreated individuals with larger pVL during the asymptomatic phase of infection progress to AIDS more rapidly than those with lower pVL (1, 2). Host genetics, host innate and adaptive immune responses, and viral sequence variations have all been suggested as possible factors influencing the level of viremia and disease outcome (3–5). Amongst host genetic factors, HLA class I types are recognized to be the most influential with respect to disease progression (6–9), indicating that the effects of HLA class I molecules on HIV-1 specific CTL responses play a major role in controlling viremia. A number of studies have reported differential impacts of HLA class I allele expression on the level of the pVL and/or disease outcome: HLA-B27, B51 and B57 are associated with lower pVL and better clinical outcome (7, 10–12), whereas HLA-B*3502/3503 and B53 have a detrimental effect on these parameters (6, 8, 13, 14).
However, such studies have been performed either in Western countries, such as the United States (6, 7, 11), or in South Africa (12), where Caucasians and/or Africans dominate over other ethnic groups; accordingly information from Asian countries is largely lacking, although an estimated 5.0 million people were living with HIV/AIDS in Asia in 2007, accounting for 15% of the world total (15). Because people living in Asia have distinct patterns of HLA class I profiles, the known associations between HLA class I allele expression and HIV disease outcome may be applicable only to a limited geographical area on the globe. In order to design globally effective HIV vaccines that aim to induce CTL responses restricted by HLA class I molecules, it is crucial to identify the differential ability of HLA class I alleles to control viremia in different parts of the world.
Of importance, CTL escape mutations have been shown to accumulate in populations (16, 17), suggesting that we have been losing targeting epitopes. However, the actual impact on the level of pVL or clinical outcome has yet to be determined. There is an urgent need to investigate whether or not accumulations of CTL escape mutations at a population level increase the virulence of HIV-1 infection.
In the present study, we have examined the impact of HLA class I allele expression on the level of pVL and rate of CD4+ T cell decline in chronically HIV-1 infected Japanese patients who have distinct class I allele expression profiles compared to Caucasians or Africans, in that: (1) they express neither major protective alleles (HLA-B27/B57) nor detrimental alleles (HLA-B*3502/B*3503/B53); and (2) they have a much narrower HLA distribution as represented by around 70% of Japanese people expressing HLA-A24 (18), and thereby likely facilitate accumulation of CTL escape mutations at the population level. In a cross-sectional analysis, we found no significant associations between the level of pVL and individual HLA class I allele expression in this unique Asian population, including HLA-B51 which ranked as the third most protective allele in Caucasians (7). Further analysis revealed that HLA-B51 has been losing its ability to control viremia in this population as the epidemic matures. However this is not the case for the other alleles, suggesting that unfavorable consequences of the accumulation of CTL escape mutations might be limited to particular HLA class I alleles. Nonetheless, these differences still pose a significant challenge for those designing globally effective HIV vaccines.
MATERIALS AND METHODS
In the present study, a total of 141 Japanese subjects who had been diagnosed with HIV-1 infection from 1995 to 2007, and had remained untreated, were enrolled. In order to exclude individuals diagnosed during an acute/early phase of infection, only those who were fully Western blot positive were enrolled, while those with a history of being HIV seronegative within the year prior to their first visit to the clinics were excluded. Written informed consent was obtained from all participants, and the study was approved by the Institutional Review Boards of the Institute of Medical Science, the University of Tokyo (No. 11-2-0329). All the participants were Japanese and all had acquired HIV-1 through sexual intercourse; all but six were men, 96% of whom were MSM.
PVL and rate of decline in CD4+T cell counts
PVL were measured by the Roche HIV Amplicore (Roche Diagnostics, Indianapolis, IN, USA). PVL and CD4+ T cell counts at the first available time points were used for the analyses. The median pVL was 19 000 RNA copies/ml (IQR: 5000–49 000 RNA copies/ml). The median CD4+T cell count was 351/μl (IQR: 273–444/μl) at the corresponding time point for each individual. The rates of decline in CD4+ T cell count (cells/year) were calculated using the values at 6 and 18 months after the first visit to the hospital.
HLA class I allele typing
High resolution HLA genotyping (4–6 digits) was performed as follows: Genomic DNA was extracted from PBMC using the QIAamp DNA Blood Mini Kit (QIAGEN, Valencia, CA, USA) according to the manufacturer's instructions. The genotypes of HLA-A,-B, and -C, were determined by PCR-SSOP using the WAKFlow HLA typing kit (Wakunaga, Hiroshima, Japan) (19) and the Luminex Multi-Analyte Profiling system (xMAP, Luminex Corporation, Austin, TX, USA) (18, 19), according to the manufacturer's instructions. For most of the analyses, we used only 2-digit types.
Comparisons of level of pVL and CD4+ T cell decline between the two groups were performed by the Mann–Whitney U test, and a q-value approach was adopted for multiple comparisons (20). q < 0.2 were considered statistically significant.
The Japanese population has a narrow HLA class I allele distribution lacking the most protective, HLA-B27/B57, yet frequently expressing the third most protective, HLA-B51
In the present study, we aimed to identify HLA class I alleles that are associated with slow or rapid HIV disease progression in the Japanese population, and to investigate changes in the impact of individual HLA class I allele expression on disease progression at the population level over time. To this end, we initially sought to characterize HLA class I allele distribution in the Japanese population as compared to that in Western countries. We expected the Japanese to have a narrower spectrum of HLA class I types, since Japan is geographically isolated and had closed the door to other nations for a long time, as a result having very few immigrants. We reviewed the literature and compared HLA distributions in the general population between Japan and the USA (Fig. 1). We found that the total number of HLA class I alleles with over 1% of allelic frequency in the Japanese population was only 29 (A: 6, B: 15 and Cw: 8, n= 1018, Fig. 1a), which is considerably smaller than that found in European-Americans (total: 46, A: 14, B: 19, Cw: 13, n= 265, Fig. 1b), and in African-Americans (total: 50, A: 16, B: 21, Cw: 13, n= 252, Fig. 1c) (18, 21), confirming that the Japanese population is genetically much less diverse as compared to these other major ethnic groups. Furthermore, we noticed unique features in the Japanese population: (1) over 70% of people express HLA-A24; (2) the major protective alleles against HIV disease progression found in North America and in African countries are rarely seen (B27: 0.05% and B57: 0.0% of allelic- frequency) (18); (3) the major detrimental alleles (B*5802, B*3502/3503 and B53) are not observed at all (18); and (4) HLA-B51, which is widely known to be protective in Caucasians, is common in the Japanese population, almost 20% of people expressing this allele (Fig. 1a). These results indicate that HIV-1 circulating in this unique Asian population has been exposed to a distinct environment in terms of CTL selection pressures as compared to HIV-1 circulating in Caucasian or African populations.
Impact of individual HLA class I allele expression on pVL in the HIV-1 infected Japanese population
Given the distinctive HLA distribution in the Japanese population, we sought to find class I alleles associated with slow or rapid disease progression that have never been reported from the Western countries. We performed HLA class I genotyping on specimens from 141 untreated chronically HIV-1 infected Japanese (see Materials and Methods) and examined their impacts on level of pVL, which is known to be closely associated with the rate of HIV disease progression (1, 2). The distribution of alleles in HIV-1 infected Japanese was similar to that of the general Japanese population described above (data not shown). We then compared the level of pVL in terms of presence or absence of individual class I alleles (Table 1), and found that five alleles (HLA-A20, B07, B54, Cw01 and Cw15) were associated with lower or larger pVL, (P < 0.05 by Fisher's exact probability test). However, after determining q-values (20) none of the associations remained significant, indicating that there are no strongly protective or detrimental alleles in this unique Asian population. Notably, in this cross-sectional analysis, expression of HLA-B51, which is the third most beneficial allele after B57 and B27 in Caucasians (7, 22), proved to be not at all protective in Japan; likewise, HLA-A11, A26 and Cw14, which have also been reported to be protective in the USA in a study which controlled for ethnicity (7), did not show any protective effects in Japanese, either. Taken together, these results indicate that alleles which have protective effects in a given population do not necessarily behave similarly in other populations.
Table 1. Association between the level of pVL and expression of individual HLA class I alleles (n= 141)
Absence of allele RNA copies/ml (log10)*
Presence of allele RNA copies/ml (log10)*
*median pVL are shown on a log10 scale.
†The Fisher exact test was performed and P < 0.05 was considered significant.
‡q-values (cut off < 0.2) for seeking strong specificity of the alleles are given.
Level of pVL is not associated with expression of particular HLA class I supertypes or homozygotes for the Bw6 motif of HLA in the Japanese population
An HLA supertype is defined as a group of class I alleles sharing a similar peptide binding motif, thereby being able to present the same CTL epitopes (23). Some HLA class I supertypes have been reported to be associated with pVL in the USA: (B7s with larger pVL, and B27s/B58s with lower pVL) (24). We looked for such associations in the Japanese population by classifying alleles observed in our cohort into eight supertypes according to the literature (i.e., A1s, A2s, A3s, A24s, B7s, B27s, B44s, B62s) (23), and found that there were no significant associations between level of pVL and expression of particular class I supertypes in the Japanese population (data not shown). This finding may be due to the Japanese lacking HLA-B27/B57, which are major contributors to the protective supertypes in the USA (24).
We further assessed the impact on pVL of the Bw4/Bw6 motif of HLA class I molecules, which are known to act as ligands of KIR on natural killer cells and to modulate their activity (25, 26). Homozygosity for Bw6 motif has been reported to be associated with rapid disease progression, whereas the subtype of Bw4, which is carried by various alleles including HLA-B27/B57, is associated with slow disease progression (27, 28). However, there was no difference in the level of pVL between Bw4 and Bw6 homozygotes in the Japanese population (median: 26 000 vs. 20 500 RNA copies/ml, P= 0.976, Fig. 2), indicating that the findings reported from the USA cannot reliably be extended to other populations.
Changes in impact of expression of HLA class I alleles on HIV pVL at the population level over time
In the cross-sectional analyses, we did not find any associations between the level of pVL and expression of individual class I alleles, supertypes or Bw motifs in this unique Asian population. Notably, expression of HLA-B51, the third most protective allele in Caucasians, was not associated with lower pVL in this Japanese population. We have previously demonstrated that escape mutations from CTL restricted by HLA-A24, which is the most common allele in Japan (expressed in >70% of Japanese), has been accumulating amongst viral strains circulating in Japan, implying that individuals expressing HLA-A24 have been losing their targeting epitopes (16). Likewise, there is a report that the majority of recently-infected HLA-A02+ individuals in the USA cannot mount CTL responses to the epitopes that had been previously recognized in HLA-A02+ individuals, suggesting that escape mutations from this response have been accumulating in the USA population (29). Moreover, a recent study by Kawashima et al. has demonstrated accumulations of CTL escape mutations for various HLA class I alleles at population levels (17). However, it remains unknown how these accumulations of viral escape mutations in populations affect the course of the disease. We thought that the narrow HLA class I spectrum in the Japanese population might facilitate accumulation of CTL escape mutations, and thereby their influence on disease progression might be more evident in Japan than in other countries. We initially compared level of pVL between individuals diagnosed in the early days of the HIV epidemic and those diagnosed in later years by stratifying the subjects according to the year of HIV diagnosis, regardless of their HLA profiles, but found no difference in the level of pVL between the two phases of the epidemic (Fig. 3a). Next, we focused on HLA-A24, which is shared by over 70% of Japanese people and for which we have previously demonstrated accumulation of CTL escape mutations at the population level (16). However, no difference was observed between the A24+ Japanese diagnosed before 2001 and those diagnosed after 2005 (median: 9650 vs. 23 000 RNA copies/ml, P= 0.379, Fig. 3b). We then performed similar comparisons for the alleles considered protective in Caucasians and commonly expressed in the Japanese (A11: 10.4%, A26:11.6%, B51:8.6% and Cw14:12.7% of allelic-frequency) (7, 18), and observed a trend that individuals expressing HLA-B51 and diagnosed before 2001 had substantially lower pVL than those diagnosed after 2005 (median 5150 vs. 41 500 RNA copies/ml, P= 0.08, Fig. 3c). Moreover, while HLA-B51+ persons displayed significantly lower pVL than B51 negative individuals before 2001 (median 5150 vs. 18 000 RNA copies/ml, P= 0.048), such differences were not observed between people diagnosed after 2005 (Fig. 3c). Given that Kawashima et al. have recently reported a similar trend for HLA-B51 (17), it appears evident that HLA-B51 has been losing its advantage over the other alleles. However, we did not see any changes in pVL over time related to expressions of the other alleles (data not shown), indicating that loss of protective superiority may be limited to particular class I alleles. Although level of pVL is closely associated with the rate of HIV disease progression, it does not measure disease progression directly. We therefore calculated the rate of decline in CD4+ T cell counts (see the Materials and Methods), and investigated their association with HLA allele expression as well, but failed to detect any alterations in the rate of decline as the HIV epidemic matured (data not shown). This may be due to the low statistical power of the present study, therefore larger scale studies are warranted in order to determine to what extent, and for which HLA alleles, such accumulations of CTL escape have been occurring, and how they have been affecting disease progression.
In the present study, we have demonstrated that: (1) there are no individual HLA class I alleles which are strongly associated with the level of pVL in the Japanese population at the current time; (2) the Japanese population has a narrow HLA distribution and lacks in the most protective HLA-B27/B57; (3) the proposed advantage of rare class I supertypes and the disadvantage of homozygotes for Bw6 motif cannot be applied to all ethnic groups across the globe; and (4) HLA-B51 has been losing its dominant effects at the population level over time, whereas this is not the case for the other alleles.
Despite substantial numbers of HIV-1 viremia controllers having been recognized in Japan, this population lacks the well-known protective alleles HLA-B27/B57. We therefore expected to discover novel associations between HIV disease progression and HLA class I alleles which are unique to Asian populations. However, in the cross-sectional analysis, we did not identify any significant associations between the level of pVL and expression of individual class I alleles, indicating that, regardless of the geographical part of the world, the protective effects of HLA alleles are greatly biased to a few of the prominent alleles like HLA-B27/B57. The discordant results for HLA supertypes and homozygosity of the Bw6 motif between Japan and the USA are likely also attributable to the lack of HLA-B27/B57 in the Japanese population. These two exceptional alleles are known to have targeting epitopes within Gag protein (10, 30–35). Likewise it has been suggested that expression of HLA alleles other than B27/57, but having targeting epitopes within Gag protein, are associated with lower pVL (8, 36–40). Therefore it is warranted to confirm that Gag specific CTL responses are associated with lower pVL in Japanese people who lack HLA-B27/57.
In the cross-sectional analysis, we did not identify significant associations between pVL and HLA-A11, 26, B51 or Cw14 expression, all of which have been shown to be protective in Caucasians (7), However, subsequent analysis revealed that HLA-B51, at least, was protective in the past, indicating that there has been loss of targeting epitopes in the viral strains circulating in this population. This result for HLA-B51 is supported by a recent report from Japan which demonstrated the accumulation of HLA-B51 escape mutation (17). Adaptation of HIV to HLA might be occurring at a greater speed in the Japanese population, which has a narrower HLA class I distribution as compared to other ethnic groups. In addition, the discordant rate of accumulation of CTL escape mutations between different populations will pose a significant challenge for designing globally effective HIV vaccines.
An increase in pVL over time was not observed for other alleles, including HLA-A24 for which the accumulation of CTL escape mutations amongst circulating viruses had been previously demonstrated (16). There are a number of feasible explanations for this unexpected observation: loss of viral replicative fitness due to CTL escape mutations may reduce viral burden in vivo (41–46); escape mutations may provide de novo CTL epitopes to the other HLA alleles; CTL restricted by these alleles can do nothing for viremia control from the start, and so on. In order to elucidate the mechanisms for these discordant results, detailed studies on viral sequences and specific CTL responses on an individual epitope basis are required.
We did not see any significant change in the rate of CD4+ T cell decline at the population level over time, though this might have been due to the low statistical power of the current study. Many health care providers have been claiming that recently diagnosed HIV infected individuals appear to progress more rapidly than did those diagnosed in previous years, and Crum-Cianflone et al. have reported significantly lower CD4+T cell counts at the first visit to clinics in individuals diagnosed in recent years (47), which may reflect adaptation of HIV to HLA. It is essential to elucidate whether the recent increase in HIV virulence has been caused by viral adaptation to HLA or to other host factors restricting proliferation of HIV.
There was a little concern that the improvement of the sensitivity of HIV-1 RNA quantification for non-B subtypes might have affected overall results; however, as described in the Materials and Methods section, 96% of studied Japanese were MSM; and in Japan virtually all MSM are considered to be infected with clade B. Therefore, inclusion of non-B infected subjects was extremely limited, and unlikely to affect the overall results.
The present study not only adds considerably to currently available knowledge but is also the first comprehensive study on associations between HLA alleles and HIV disease progression in Asia. However, there were a number of limitations: (1) the scale of the study was relatively small, which may have resulted in overlooking some true associations between HLA and the level of pVL; (2) the observation period might have not been long enough to see changes in pVL at the population level; (3) the incidence and prevalence of HIV infection in Japan might not be sufficiently high to see changes in pVL at the population level; and (4) we did not have viral sequence data, meaning that any accumulation of CTL escape mutations could not be demonstrated, though it is evident from the literature that such accumulation has occurred, at least for HLA-A24 and B51 (16, 17). In order to demonstrate that loss of protective effects of particular HLA alleles are attributable to accumulation of CTL escape mutations in the population, it is necessary to define CTL epitopes restricted by common HLA class I alleles in Japan systematically, and to identify escape mutations from those CTL responses. In spite of these limitations, the present study is valuable in consolidating the loss of predominance of some HLA class I alleles in a given population, and in raising concerns about both designing globally effective HIV vaccines and the future virulence of HIV-1.
The authors declare no conflicts of interest related to this study.
We thank the patients and clinical staff at the Research Hospital of the Institute of Medical Science, University of Tokyo, for their essential contributions to this research study. We also thank M. Motose for technical assistance. This work was supported in part by the Program of Founding Research Centers for Emerging and Reemerging Infectious Diseases of the Ministry of Education, Culture, Sports, Science and Technology (MEXT); Global COE Program (Center of Education and Research for Advanced Genome-Based Medicine) of MEXT; Grants for Research on HIV/AIDS and Research on Publicly Essential Drugs and Medical Devices from the Ministry of Health, Labor, and Welfare of Japan.