Introduction
The absence of clinical progression in some HIV-1-infected individuals and the lack of a detectable HIV-1 genome despite multiple and repeated exposures to this virus in some groups of people are noteworthy phenomena when considering the development of preventative and therapeutic means to HIV infection [1-3]. There are individuals who show strong HIV-1 antigen-specific T-lymphocyte responses and HIV-1-reactive mucosal IgA production despite the absence of detectable plasma HIV-1 RNA and HIV-1 cDNA from peripheral blood mononuclear cells (PBMC) [4-6]. HIV-1-neutralizing activity exerted by the IgA isolated from some HIV-1-exposed but uninfected individuals (EUI) [7-9] has suggested a possible contribution of the host immune responses to the resistance against HIV infection. However, genetic factors that may influence the observed T-cell priming and the production of anti-HIV-1 IgA without the establishment of HIV replication are currently unknown.
Host genetic factors influencing viral entry and replication and antiviral immune responses have been extensively studied in mouse models of retroviral infections, among which the best analyzed is Friend mouse leukemia virus complex (FV) [10-13]. Host gene loci that control the entry and replication of FV in the target cells have been identified [14-17]. In addition, MHC class II loci directly restrict the T-helper cell recognition of the viral antigens [18-20], while a class I locus influences the production of cytokines from virus-specific T cells [21]. Another locus that has been mapped to chromosome 15 strongly influences the persistence of viremia after FV infection [12,22-25]. However, the possible relationship between the above persistence of viremia and production of virus-neutralizing antibodies has not been directly examined. Here we have performed linkage analyses on a mouse locus that influences the production of virus-neutralizing antibodies upon FV infection. An extension of this mouse study unexpectedly led us to find human chromosomal markers that are associated with the presence of HIV-1-reactive immune responses in HIV-uninfected individuals.
Methods
Mice, virus, and assays for neutralizing antibodies
B10.A and A/WySn mice were purchased from Japan SLC, Inc., Hamamatsu, Japan and The Jackson Laboratory, Bar Harbor, Maine, USA, respectively. The F1 crosses and backcross mice were bred and maintained at Rakuno Gakuen University and Kinki University School of Medicine under specific pathogen-free conditions. The following experiments were approved by and performed under guidelines of each university. FV was prepared and inoculated as described [18,20-27]. FV-neutralizing antibodies were titered by immuno-enzymatically visualizing foci of virus infection as described [20,26,27].
Analyses of simple sequence length polymorphisms (SSLP) and linkage mapping in mice
Genomic DNA was prepared from the tail tip of each mouse using DNeasy Tissue Kit (QIAGEN GmbH, Hilden, Germany). A pair of oligonucleotide primers for each microsatellite locus was designed based on the sequence information listed in the Genetic and Physical Maps of the Mouse Genome site (http://www-genome.wi.mit.edu/cgi-bin/mouse/) and ordered from QIAGEN GmbH. Fifty nanogrmas of each template DNA was subjected to 35 cycles of PCR amplification using a recombinant Taq polymerase (Invitrogen Life Technologies, Carlsbad, California, USA). PCR products were separated by electrophoresis in 4% agarose gel and visualized by ethidium bromide staining. Correlation between genotypes at each examined locus and the presence or absence of virus-neutralizing antibodies was analyzed by Pearson's χ2 test. Map orders of the chromosomal loci and log-of-the-odds (LOD) scores were determined by multipoint analyses using MAPMAKER/EXP version 3.0b (The Whitehead Institute, Massachusetts, USA).
EUI and HIV-1-infected individuals
Forty-two heterosexual couples discordant for HIV-1 serostatus were enrolled. The female partner was HIV-1-infected in 32 couples, whereas the male partner was HIV-1-infected in the remaining 10 couples. The diagnosis of HIV-1 infection was made based on the detection of plasma HIV-1 RNA before the initiation of antiretroviral drug treatment and significant titers of serum anti-HIV-1 IgG antibody as described in the following section. The inclusion criteria for the EUI group were a history of multiple unprotected sexual episodes for >4 years with an average of eight reported unprotected sexual contacts per year (range 5 to >40) at the time of inclusion, and at least three episodes of at-risk intercourses within 4 months prior to the study point. Forty-two of the 49 HIV-1-infected individuals studied here are the steady and reportedly monogamous partners of the above EUI individuals. In all the infected individuals the diagnosis of HIV-1 infection was made during their chronic phase, and thus unprotected sexual intercourses had been initiated long before their diagnosis. Mean CD4 cell count of the infected partners at the time of this study was 370 × 106/l (range 36 × 106-850 × 106/l). Seven additional age- and sex-matched HIV-1-infected individuals were added to the study, and their HIV-related phenotypes were within the ranges of the above infected partners. All the EUI and HIV-1-infected individuals and 47 uninfected, age- and sex-matched healthy volunteers were enrolled from the Santa Maria Annunziata Hospital, Firenze, and the Luigi Sacco Hospital, Milano. All of the enrolees are Caucasians from the Toscana region. The ethics committees of the above hospitals have approved the research protocols. The genotyping analyses were approved by Kinki University School of Medicine. Written informed consent was obtained from all enrolees, and samples were anonymized and analyzed in a blinded fashion.
Phenotype definitions
Plasma HIV-1 load was quantified by using the AMPLICOR HIV Monitor test (Roche Diagnostic Systems, Nutley, New Jersey, USA) as described [4,6]. Possible presence of HIV-1 cDNA in PBMC and in cells isolated by urethral swabbing or uterine cervical brushing was analyzed by a reverse transcription-PCR method [4,6]. For the detection of mucosal anti-HIV-1 IgA, 500 μl of mucus was collected from each enrolee by swabbing from the urethra or vagina [4,6]. Titers of serum anti-HIV antibodies were determined by an enzyme-linked immunosorbent assay (EIA) using Abbot HIV-1/2+ test (Abbott Laboratories, Abbott Park, Illinois, USA). This assay detects HIV-specific IgG and IgM [7]. Titration of HIV-1-specific IgA in the mucosal secretions was performed by an isotype-specific EIA using the HIV EIA test (Calypte Biomedical Corp., Berkeley, California, USA) with modifications [4-9]. HIV-1-reactive memory T cells in the peripheral blood were enumerated by an enzyme-linked immunospot assay for interferon (IFN)-γ as described [6].
Analyses of human SSLP markers
Five-hundred nanograms of genomic DNA extracted from PBMC of each examined individual was used as the template for 40 cycles of PCR amplification using the flanking primer sets designed based on the sequence data compiled in the Ensembl Genome Browser (http://www.ensembl.org/) and ordered from QIAGEN GmbH. Each forward primer was labeled with a florescent dye, and 50-100 fmol of PCR amplified fragments were applied onto an ABI 3100 DNA sequencer (Applied Biosystems, Foster City, California, USA) with appropriate size markers. Peak identification and size measurements were done with the GeneScan software (Applied Biosystems). To determine absolute fragment sizes, PCR products obtained from two or more homozygotes for each locus were cloned into pCR2.1-TOPO vector (Invitorogen Life Technologies) and sequenced by using the M13 forward primer until six or more identical clones were observed for each allele.
Population genetic analyses and detection of linkage disequilibrium (LD)
Genotypic data were analyzed for possible population differentiation and LD between pairs of loci by using the Arlequin 2.001 (Genetics and Biometry Laboratory, University of Geneva, Switzerland). A population-pairwise genetic distance test using pairwise FST and extended exact test were performed to examine possible population differentiation. A likelihood ratio test was performed to examine the possible LD between pairs of loci. A total of 100,172 permutations on 10 initial conditions were performed by the expectation maximizing algorithm for each pair of loci.
Statistical analyses
Distributions of allele frequencies at each examined locus were compared between each pair of the three phenotypic groups by a Monte-Carlo approach using the CLUMP software [28]. T2 statistic was chosen because many cells in the contingency tables contained values ≤ 2.
To examine the possible presence of a dominant allele having different frequencies between the three phenotypic groups, mathematical analyses were performed based on the assumption that the number of individuals possessing each genotype had a multinominal distribution. Since the number of candidate dominant alleles was more than one, multiple comparisons were taken into consideration. The test statistics for alleles i and j, ti and tj, respectively, can be strongly correlated especially when most of the individuals having allele i or j are of the genotype i/j. Therefore, the typically used Bonferroni correction may be too conservative. A universally applicable method for overcoming this problem is a closed testing procedure [29], where ti is based on a well-acquainted variance stabilizing transformation and the test statistic for a common hypothesis is based on the maximization of ti's. To calculate the P-values, we applied a parametric bootstrap [30] based on the asymptotic null distribution of ti's. Details of the mathematical methods are described in the Appendix section.
Results
Linkage mapping of a mouse locus controlling FV-neutralizing antibodies
When (B10.A × A/WySn)F1 and A/WySn mice that share FV-susceptible Fv-1b/b, Fv-2s, and H2a/a genotypes were tested for their production of virus-neutralizing antibodies, none possessed a detectable level of neutralizing antibodies at post-infection day (PID) 10. Neutralizing antibodies remained undetectable at PID 15 and 20 in parental A/WySn mice. In contrast, all the infected (B10.A × A/WySn)F1 mice possessed a significant neutralizing titer at PID 15, and the titers increased toward PID 20 (Fig. 1a). Therefore, possible segregation of neutralizing titers in (B10.A × A/WySn) × A/WySn backcross mice was examined by testing them at PID 15, 17, and 21. Virus-neutralizing antibodies were not detectable in 63 (44%) of the 143 backcross mice at PID 15 (Fig. 1b), suggesting that a single locus is involved in the production or lack of production of neutralizing antibodies. For linkage analyses, we concentrated genotyping on chromosome 15, because initial analyses performed by using 43 separate backcross individuals showed significant correlation between virus-neutralizing titers at PID 17 and genotypes at four loci in chromosome 15 (data not shown). The results of linkage analyses performed by using the 143 backcross mice indicated a strong correlation between genotypes at marker loci in chromosome 15 and titers of virus-neutralizing antibodies at PID 15, with the strongest correlation (χ2 = 74.0, P = 1.17 × 10-7) observed at the D15Mit71 locus (Fig. 2). Linkage mapping with MAPMAKER/EXP located a single locus determining the presence or absence of virus-neutralizing antibodies at PID 15 between the D15Mit71 and D15Mit171 loci. Further mapping was performed by genotyping the backcross animals that possessed a critical recombination between the D15Mit28 and D15Mit171 loci. As a result, eight backcross mice that possessed reciprocal recombination within this region were identified (Fig. 2). Since a significant correlation (P = 0.029 by two-tailed Fisher's exact test) between genotypes at the D15Mit71, D15Mit2, D15Mit214, D15Mit69, and D15Mit70 loci and the production of virus-neutralizing antibodies at PID 15 was observed in these recombinant animals, it is conceivable that the locus controlling the production of FV-neutralizing antibodies is located within the region telomeric to the D15Mit1 and centromeric to D15Mit118 loci at the widest estimation.
Genetic analyses of HIV-1-exposed and uninfected Italians
The above region of mouse chromosome 15 harbors previously mapped genes that are known or likely to affect immune cell development and/or activation and retroviral replication. Therefore, we next explored the possibility that a putative ortholog of the above mouse locus might influence immune responsiveness in HIV-1 infection. Because of the route of transmission of HIV-1 and resultant rarity of multicase families, linkage analyses comparing affected and unaffected siblings are impossible. Thus, we performed a simple association study by comparing genotypes between the exposed but uninfected and HIV-1-infected individuals, hypothesizing that efficient anti-HIV-1 immune responses are associated with the presence of a dominant genetic factor which might be an ortholog of the above mouse locus conferring the ability to produce FV-neutralizing antibodies. Thus, the three phenotypically distinct groups of individuals (Table 1) were genotyped at the loci shown in Fig. 2. The examined groups were not different to each other when tested for population-pairwise genetic distance (P > 0.17), in accordance with their all being Caucasians enrolled from the Toscana region of Italy. When allele frequencies were compared among the three phenotypic groups, their distribution at the D22S277 locus differed between the EUI and healthy control groups at P = 0.0396. No significant difference was observed at the other loci. When likelihood ratio tests were performed for all possible pairs of the examined loci, a highly significant LD of an exact P < 0.0004 level was observed in all three groups between the D22S284, D22S423, and D22S299 loci, reflecting their close physical locations (Fig. 2 and Fig. 3). A similarly significant LD (P < 0.00002) was observed between the D22S418 and D22S1166 loci in all three groups, confirming their close genetic locations. Interestingly, a highly significant LD (P < 0.00002) was observed between the D22S276 and the above surrounding loci in both the HIV-1-infected and healthy control groups, but this was not observed in the EUI group (Fig. 3).
When frequencies of individuals possessing a particular allele at a given locus were compared among the three phenotypic groups by adopting a dominant model, objective mathematical analyses revealed multiple loci with significant differences (Table 2). These individual differences were further examined for possible false rejection of a single equal-frequency hypothesis due to multiple comparisons by using the closed testing procedure. As a result, frequencies of individuals possessing either the allele 156 or 158 at the D22S277 locus were significantly different between the EUI and two other groups, those of the individuals possessing the allele 134 at the D22S272 locus were significantly different between the EUI and healthy control groups, and those of individuals possessing the allele 229 at the D22S423 locus also differed significantly between the EUI and HIV-infected individuals.
Discussion
In the present study we have demonstrated that the presence or absence of virus-neutralizing antibodies in FV-infected (B10.A × A/WySn) × A/WySn backcross mice at PID 15 is closely associated with their genotypes at the chromosome 15 loci. The linkage mapping data indicated that a single gene controlling the production of virus-neutralizing antibodies was located near the D15Mit71 locus, colocalizing with the previously mapped Rfv-3 locus [23,24]. Since the Rfv-3-associated phenotypes were defined by clearance of viremia by 35-40 days after FV infection [22-25], and neutralizing antibodies were detectable at as early as PID 15 in mice possessing the B10.A-derived dominant allele (Fig. 1), it is conceivable that early production of virus-neutralizing antibodies is associated with early clearance of viremia.
It is intriguing that genotypes at microsatellite loci located within the segment of human chromosome 22 syntenic to mouse chromosome 15 differed between the HIV-1-exposed but uninfected and HIV-1-infected groups of individuals. The strongest association was observed at the D22S423 locus where the frequency of individuals possessing the allele 229 was significantly higher in the EUI group than in the HIV-1-infected one even after corrections for multiple comparisons were made. This marker locus is located in the middle of the chromosomal segment corresponding to the region of mouse chromosome 15 that harbors the gene locus controlling the production of virus-neutralizing antibodies (Fig. 2). It may also be worth noting that the alleles 156 and 158 at the D22S277 locus that are rare (5.6 and 9.3% per haploid chromosome, respectively) among the Caucasian CEPH population [31,32] were more frequently observed in the EUI group (9.5 and 17.9%, respectively). There were significant differences in the frequency of the allele 156 and that of the allele 158 (P = 0.035 and 0.0061, respectively, by two-tailed Fisher's exact test) when the HIV-1-infected and healthy control groups were combined and compared with the EUI group. The rates of microsatellite mutation are much higher than those of point mutation at coding genes [33], and the most common stepwise mutation is biased toward the reduction of repeat numbers for microsatellites of >20 repeats [34]. Thus, we can justifiably hypothesize that the alleles 156 and 158 at the D22S277 locus (25 and 26 dinucleotide repeats, respectively) are both linked to the same putative allele that is associated with the presence of immune responses to HIV-1 in the uninfected individuals. In this regard, the variance stabilizing analyses performed by assuming that the alleles 156 and 158 are both linked to a single dominant genetic factor resulted in the demonstration of significant differences between the EUI and HIV-1-infected, and the EUI and healthy control groups, and these individual null hypotheses were also rejected (significant difference validated) after the correction for multiple comparisons was made (Table 2). Further, when the same comparison was made between a combined group of the HIV-1-infected and healthy control individuals and the EUI group, the frequency of individuals possessing either the allele 156 or 158 was significantly higher among the EUI (P = 0.0019), and this was highly significant even after the correction for multiple comparisons was made (P = 0.0121). The combination of the HIV-1-infected and healthy control groups was justifiable because neither allele frequency distributions nor frequencies of individuals possessing the allele 156 or 158 were significantly different between the two groups. Thus, genotypes at multiple loci within the segment of human chromosome 22 that is syntenic to mouse chromosome 15 are significantly associated with the presence of strong mucosal and T-cell immune responses against HIV-1 (Table 1) in HIV-uninfected Italians. Furthermore, the multilocus LD spanning from D22S284 to D22S418 which is observed in both the HIV-infected and healthy control groups is disrupted at the D22S276 locus in the EUI group (Fig. 3). This observation is consistent with the hypothesis that in the ancestors of the EUI individuals a possible recombinational or mutational event might have happened in the chromosomal segment surrounding this locus.
Production and class-switching of virus-neutralizing antibodies in FV-infected mice are dependent on CD4 T-cell functions [26,27,35], and the priming of CD4 T-helper cells with a singe-epitope peptide resulted in the early production and class-switching of virus-neutralizing antibodies and strong protection against FV infection [20,27]. Likewise, EUI individuals enrolled into the present study possessed significantly higher amounts of mucosal anti-HIV-1 IgA and larger numbers of HIV-1 envelope-reactive T cells in the peripheral blood in comparison with the HIV-infected individuals (Table 1). IgA antibodies isolated from some EUI individuals have been shown to inhibit the replication of primary HIV-1 isolates [7] and HIV-1 transcytosis across the epithelial cells in vitro [8,36]. Thus, it is possible that efficient priming of T cells with HIV antigens might have resulted in rapid production of HIV-1-reactive IgA antibodies which, in turn, might have been involved in the possible immune protection in the EUI individuals. In this regard, it is noteworthy that IFN-γ production is required for the control of viremia and class-switching of virus-neutralizing antibodies in FV infected mice [37].
It has been shown that CD4 T cell-dependent early IgA responses against influenza virus infection can be generated in the absence of virus-specific IgM and IgG [38], and costimulatory signals required for mucosal IgA production are strikingly different from those needed for systemic antibody responses [39]. Similarly, mucosal IgA responses to T-dependent HIV-1 antigens might be stimulated without inducing serum IgG production, and putative human homolog of the mouse gene influencing the T cell-dependent production of FV-neutralizing antibodies might be involved in the above activation of mucosal IgA-production in EUI individuals. In fact, the segment of mouse chromosome 15 between the D15Mit1 and D15Mit118 loci and the corresponding segment of human chromosome 22 harbor several genes that are known to be involved in T- and B-cell growth and activation. Expression analyses of these candidate genes both in the mouse model and in humans are currently underway.
None of the previously reported human genes that affect the risk of HIV acquisition are located in chromosome 22, CCR5 and CCR2 being located at 3p21, SDF1 and MBL2 at 10q11.1 and 10q11.2, respectively, HLA including the polymorphic TNF and MIC loci at 6p21.3, KIRs at 19q13.4, IL10 at 1q31-32, and SLC11A1 (NRAMP1) at 2q35 [3,40-52]. In addition, the homozygous CCR5-Δ32 mutation, which results in the lack of the HIV coreceptor [3,40-42], is known to be rare among the EUI individuals in Italy and Thailand [4,9,49], and was not found in the enrolees of the present study, although three of the 42 EUI individuals were heterozygous for this mutation (data not shown). In a very recent analysis of a separate cohort of repeatedly exposed but HIV-1-seronegative individuals in the USA, Liu et al. [53] demonstrated the lack of association between genotypes at the CCR2, SDF1, and RANTES loci and the uninfected status. The homozygous CCR5-Δ32 mutation was also rare (3.2%) among the seronegative individuals. The same authors also noted a significant difference in the frequencies of heterozygosity at the polymorphic DC-SIGN (CD209) locus at 19p13.2 between the exposed but seronegative and HIV-1-infected groups: however, the observed frequency of heterozygotes was 3.2% (3/94), and thus, this genetic skewing could not explain the possible mechanisms that confer HIV resistance to the majority of the seronegative individuals. Altogether, our results have indicated the possible presence in human chromosome 22 of a novel genetic factor that is associated with strong T-cell and mucosal immune responses to HIV-1 antigens.
Acknowledgements
We thank M. P. Gorman and S. Tsuji-Kawahara for critically reading the manuscript.
Sponsorship: This work was supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan, including the High-Tech Research Center Project (2002), from the Ministry of Health, Labor and Welfare of Japan, from the Japan Health Science Foundation, from the Istituto Superiore di Sanita' Programma Nazionale di Ricerca sull' AIDS, from Centro di Eccellenza CISI, from the EMPRO and AVIP EC WP6 Projects, and from the Tuscancy Region, General Drection, Right to Health and Solidarity Policy.
References
1. Rowland-Jones SL, McMichael A. Immune responses in HIV-exposed seronegatives: Have they repelled the virus? Curr Opin Immunol 1995; 7:448-455.
2. Shearer GM, Clerici M. Protective immunity against HIV infection: Has nature done the experiment for us? Immunol Today 1996; 17:21-24.
3. O'Brien SJ, Nelson GW, Winkler CA, Smith MW. Polygenic and multifactorial disease gene association in man: Lessons from AIDS. Annu Rev Genet 2000; 34:563-591.
4. Mazzoli S, Trabattoni D, Lo Caputo S, Piconi S, Blé C, Meacci F, et al. HIV-specific mucosal and cellular immunity in HIV-seronegative partners of HIV-seropositive individuals. Nature Med 1997; 3:1250-1257.
5. Kaul R, Trabattoni D, Bwayo JJ, Arienti D, Zagliani A, Mwangi F, et al. HIV-1 specific mucosal IgA in a cohort of HIV-1-resistent Kenyan sex workers. AIDS 1999; 13:23-29.
6. Biasin M, Lo Caputo S, Speciale L, Colombo F, Racioppi L, Zagliani A, et al. Mucosal and systemic immune activation is present in human immunodeficiency virus-exposed seronegative women. J Infect Dis 2000; 182:1365-1374.
7. Mazzoli S, Lopalco L, Salvi A, Trabattoni D, Lo Caputo S, Semplici F, et al. Human immunodeficiency virus (HIV)-specific IgA and HIV neutralizing activity in the serum of exposed seronegative partners of HIV-seropositive persons. J Infect Dis 1999; 180:871-875.
8. Belec L, Ghys PD, Hocini H, Nkengasong JN, Tranchot-Diallo J, Diallo MO, et al. Cervicovaginal secretory antibodies to HIV type 1 that block viral transcytosis through epithelial barriers in highly exposed HIV-1-seronegative African women. J Infect Dis 2001; 184:1412-1422.
9. Lo Caputo S, Trabattoni D, Vichi F, Piconi S, Lopalco L, Villa ML, et al. Mucosal and systemic HIV-specific immunity in HIV-exposed but uninfected heterosexual males. AIDS 2003; 17:531-538.
10. Teich N, Wyke J, Mak T, Bernstein A, Hardy W. In: RNA Tumor Viruses, 2nd edn. Edited by Weiss R, Teich N, Varmus H, Coffin J. New York: Cold Spring Harbor Laboratory; 1982: 785-998.
11. Kabat D. Molecular biology of Friend viral erythroleukemia. Curr Top Microbiol Immunol 1989; 148:1-42.
12. Chesebro B, Miyazawa M, Britt WJ. Host genetic control of spontaneous and induced immunity to Friend murine retrovirus infection. Annu Rev Immunol 1990; 8:477-499.
13. Hoatlin ME, Kabat D. Host-range control of a retroviral disease: Friend erythroleukemia. Trends Microbiol 1995; 3:51-57.
14. Best S, Le Tissier P, Towers G, Stoye JP. Positional cloning of the mouse retrovirus restriction gene Fv1. Nature 1996; 382:826-829.
15. Persons DA, Paulson RF, Loyd MR, Herley MT, Bodner SM, Bernstein A, et al. Fv2 encodes a truncated form of the Stk receptor tyrosine kinase. Nat Genet 1999; 23:159-165.
16. Ikeda H, Laigret F, Martin MA, Repaske R. Characterization of a molecularly cloned retroviral sequence associated with Fv-4 resistance. J Virol 1985; 55:768-777.
17. Ikeda H, Sugimura H. Fv-4 resistance gene: a truncated endogenous murine leukemia virus with ecotropic interference properties. J Virol 1989; 63:5405-5412.
18. Miyazawa M, Nishio J, Chesebro B. Genetic control of T cell responsiveness to the Friend murine leukemia virus envelope antigen. Identification of the class II loci of H-2 as immune response genes. J Exp Med 1988; 168:1587-1605.
19. Iwashiro M, Kondo T, Shimizu T, Yamagishi H, Takahashi K, Matsubayashi Y, et al. Multiplicity of virus-encoded helper T-cell epitopes expressed on FBL-3 tumor cells. J Virol 1993; 67:4533-4542.
20. Sugahara D, Tsuji-Kawahara S, Miyazawa M. Identification of a protective CD4+ T-cell epitope in p15gag of Friend murine leukemia virus and role of the MA protein targeting the plasma membrane in immunogenicity. J Virol 2004; 78:6322-6334.
21. Peterson KE, Iwashiro M, Hasenkrug KJ, Chesebro B. Major histocompatibility complex class I gene controls the generation of gamma interferon-producing CD4+ and CD8+ T cells important for recovery from Friend retrovirus-induced leukemia. J Virol 2000; 74:5363-5367.
22. Chesebro B, Wehrly K. Studies on the role of the host immune response in recovery from Friend virus leukemia. I. Antiviral and antileukemia cell antibodies. J Exp Med 1976; 143:73-84.
23. Hasenkrug KJ, Valenzuela A, Letts V, Nishio J, Chesebro B, Frankel WN. Chromosome mapping of Rfv3, a host resistance gene to Friend murine retrovirus. J Virol 1995; 69:2617-2620.
24. Super HJ, Hasenkrug KJ, Simmons S, Brooks DM, Konzek R, Sarge KD, et al. Fine mapping of the Friend retrovirus resistance gene, Rfv3, on mouse chromosome 15. J Virol 1999; 73:7848-7852.
25. Doig D, Chesebro B. Anti-Friend virus antibody is associated with recovery from viremia and loss of viral leukemia cell-surface antigens in leukemic mice. Identification of Rfv-3 as a gene locus influencing antibody production. J Exp Med 1979; 150:10-19.
26. Miyazawa M, Nishio J, Wehrly K, Chesebro B. Influence of MHC genes on spontaneous recovery from Friend retrovirus-induced leukemia. J Immunol 1992; 148:644-646.
27. Miyazawa M, Fujisawa R, Ishihara C, Takei YA, Shimizu T, Uenishi H, et al. Immunization with a single T helper cell epitope abrogates Friend virus-induced early erythroid proliferation and prevents late leukemia development. J Immunol 1995; 155:748-758.
28. Sham PC, Curtis D. Monte Carlo tests for association between disease and alleles at highly polymorphic loci. Ann Hum Genet 1995; 59:97-105.
29. Hsu JC. Multiple comparisons: theory and methods. New York: Chapman & Hall/CRC; 1996.
30. Hall P. The bootstrap and edgeworth expansion. New York: Springer; 1992.
31. Weissenbach J, Gyapay G, Dib C, Vignal A, Morissette J, Millasseau P, et al. A second-generation linkage map of the human genome. Nature 1992; 359:794-801.
32. Gyapay G, Morissette J, Vignal A, Dib C, Fizames C, Millasseau P, et al. The 1993-94 Genethon human genetic linkage map. Nat Genet 1994; 7:246-339.
33. Li Y-C, Korol AB, Fahima T, Beiles A, Nevo E. Microsatellites: genetic distribution, putative functions and mutational mechanisms: a review. Mol Ecol 2002; 11:2453-2465.
34. Whittaker JC, Harbord RM, Boxall N, Mackay I, Dawson G, Sibly RM. Likelihood-based estimation of microsatelliote mutation rates. Genetics 2003; 164:781-787.
35. Super HJ, Brooks D, Hasenkrug K, Chesebro B. Requirement for CD4+ T cells in the Friend murine retrovirus neutralizing antibody response: evidence for functional T cells in genetic low-recovery mice. J Virol 1998; 72:9400-9403.
36. Devito C, Broliden K, Kaul R, Svensson L, Johansen K, Kiama P, et al. Mucosal and plasma IgA from HIV-1-exposed uninfected individuals inhibit HIV-1 transcytosis across human epithelial cells. J Immunol 2000; 165:5170-5176.
37. Dittmer Y, Peterson KE, Messer R, Stromnes IM, Race B, Hasenkrug KJ. Role of interleukin-4 (IL-4), IL-12, and gamma interferon in primary and vaccine-induced immune responses to Friend retrovirus infection. J Virol 2001; 75:654-660.
38. Sangster MY, Riberdy JM, Gonzalez M, Topham DJ, Baumgarth N, Doherty PC. An early CD4+ T cell-dependent immunoglobulin A responses to influenza infection in the absence of key cognate T-B interactions. J Exp Med 2003; 198:1011-1021.
39. Gärdby E, Wrammert J, Schön K, Ekman L, Leandersen T, Lycke N. Strong differential regulation of serum and mucosal IgA responses as revealed in CD28-deficient mice using cholera toxin adjuvant. J Immunol 2003; 170:55-63.
40. Dean M, Carrington M, Winkler C, Huttley GA, Smith MW, Allikmets R, et al. Genetic restriction of HIV-1 infection and progression to AIDS by a deletion allele of the CCR5 structural gene. Science 1996; 273:1856-1862.
41. Liu R, Paxton WA, Choe S, Ceredini D, Martin SR, Horuk R, et al. Homozygous defect in HIV-1 coreceptor accounts for resistance of some multiply-exposed individuals to HIV-1 infection. Cell 1996; 86:367-377.
42. Samson M, Libert F, Doranz BJ, Rucker J, Liesnard C, Farber CM, et al. Resistance to HIV-1 infection in Caucasian individuals bearing mutant alleles of the CCR-5 chemokine receptor gene. Nature 1996; 382:722-725.
43. Martin MP, Dean M, Smith MW, Winkler C, Gerrard B, Michael NL, et al. Genetic acceleration of AIDS progression by a promoter variant of CCR5. Science 1998; 282:1907-1911.
44. Smith MW, Dean M, Carrington M, Winkler C, Huttley GA, Lomb DA, et al. Contrasting genetic influence of CCR2 and CCR5 receptor gene variants on HIV-1 infection and disease progression. Science 1997; 277:959-965.
45. Winkler C, Modi W, Smith MW, Nelson GW, Wu X, Carrington M, et al. Genetic restriction of AIDS pathogenesis by an SDF-1 chemokine gene variant. Science 1998; 279:389-393.
46. Carrington M, Nelson G, Martin MP, Kissner T, Vlahov D, Goedert JJ, et al. HLA and HIV: Heterozygote advantage and B*35-Cw*04 disadvantage. Science 1999; 3:1748-1752.
47. Shin HD, Winkler C, Stephens JC, Bream J, Young H, Goedert JJ, et al. Genetic restriction of HIV-1 infection and AIDS progression by promoter alleles of interleukin 10. Proc Natl Acad Sci U S A 2000; 97:14467-14472.
48. Martin MP, Gao X, Lee J-H, Nelson GW, Detels R, Goedert JJ, et al. Epistatic interaction between KIR3DS1 and HLA-B delays the progression to AIDS. Nat Genet 2002; 31:429-434.
49. Beyrer C, Artenstein AW, Rugpao S, Stephens H, Van Cott TC, Robb ML, et al. Epidemiologic and biologic characterization of a cohort of human immunodeficiency virus type 1 highly exposed, persistently seronegative female sex workers in northern Thailand. J Infect Dis 1999; 79:59-68.
50. Delgado JC, Leung JY, Baena A, Clavijo OP, Vittinghoff E, Buchbinder S, et al. The -1030/-862-linked TNF promoter single-nucleotide polymorphisms are associated with the inability to control HIV-1 viremia. Immunogenetics 2003; 55:497-501.
51. Garred P, Madsen HO, Balslev U, Hofmann B, Pedersen C, Gerstoft J, et al. Susceptibility to HIV infection and progression of AIDS in relation to variant alleles of mannose-biding lectin. Lancet 1997; 349:236-240.
52. Marquet S, Sánchez F, Arias M, Rodríguez J, París SC, Skamene E, et al. Variations of the human NRAMP1 gene and altered human immunodeficiency virus infection susceptibility. J Infect Dis 1999; 180:1521-1525.
53. Liu H, Hwangbo Y, Holte S, Lee J, Wang C, Kaupp N, et al. Analysis of genetic polymorphisms in CCR5, CCR2, stromal cell-derived factor-1, RANTES, and dendritic cell-specific intercellular adhesion molecule-3-grabbing nonintegrin in seronegative individuals repeatedly exposed to HIV-1. J Infect Dis 2004; 190:1055-1058.
Appendix
The possible presence of a dominant allele having different frequencies between the phenotypic groups was examined as follows: Define xij as the number of individuals having the genotype i/j (i ≤ j) for the EUI group, where n = ∑i ≤ j xij is the total number of individuals belonging to this group. Assume that x = (xij)i≤j has a multinomial distribution with the parameter a = (aij)i≤j, where ∑i ≤ j aij = 1. For convenience, let aji = aij. Similarly, define the notations y, b and z, c for the HIV-1-infected and healthy control groups, respectively. The frequency of the individuals having the allele i for the EUI group is expressed as ai = ∑k aik. Similarly, define bi and ci. The hypothesis where the frequencies of the individuals having the allele i for the EUI and HIV-infected groups is the same is expressed as Hi: ai = bi. Similarly, consider the hypothesis ai = ci to compare the EUI with healthy control groups.
We tested whether or not the frequency of individuals possessing a certain dominant allele in the EUI group was different from those in other groups. Since there are multiple candidate alleles at each locus, we must take multiple comparisons into consideration. Let ti be the test statistic for allele i. ti and tj can be strongly correlated, especially when most of the individuals having allele i or j are of the genotype i/j. Therefore, the typically used Bonferroni correction can be too conservative. A universally applicable method for overcoming this problem is a closed testing procedure [29], where ti is based on a well-acquainted variance stabilizing transformation and the test statistic for a common hypothesis is based on the maximization of ti values.
Let
Equation (Uncited)Image Tools
be the closed set consisting of all the intersections of the hypotheses' Hi values. Assume that we can make the reject region with common significance level α for any hypothesis H ∈
Equation (Uncited)Image Tools
. The closed testing procedure says that we can reject H ∈
Equation (Uncited)Image Tools
only after we reject all the hypotheses including H, using the corresponding reject region. Let ti be the standardized test statistic for the hypothesis Hi. The corresponding reject region becomes Wi = {|ti| > ei}. Consider a common hypothesis H. For example, let H be the intersection of H1,…, HI. The corresponding reject region can be defined by W = {maxi=1,…,I|ti| > e}. We used the following variance stabilizing type as the standardized test statistic:
Equation (Uncited)Image Tools
where nx = ∑i ≤ j xij and ny = ∑i ≤ j yij. As an advantage over the commonly used likelihood ratio and Pearson's χ2 tests, the above type enables us to infer that the smaller a P value is the stronger the rejection of the corresponding null hypothesis, because the variances of the arcsine are constant independent of the samples. In view of the closed testing procedure, if the maximal intersection hypothesis H ∈
Equation (Uncited)Image Tools
is rejected, the individual hypothesis corresponding to the minimum P value can automatically be rejected. In addition, if the hypothesis corresponding to the minimum P value alone is rejected among the individual hypotheses, it is the only rejected hypothesis.
The joint distribution of ti values can be approximated by the multivariate normal distribution under the null hypothesis, and the corresponding approximated P values can easily be calculated for the individual hypotheses. The approximated P values for a common hypothesis can be calculated by using the central limit theorem and the parametric bootstrap [30] based on the asymptotic null distribution of ti values. To avoid unnecessary disturbances, we tested only the hypotheses having the estimated frequency ≥ 0.1 when considering the common hypotheses, because alleles with a frequency < 0.1 cannot explain the phenotype of the whole group. Calculations were performed by drawing 100 000 random samples from the approximated multivariate normal distribution for each hypothesis.
© 2005 Lippincott Williams & Wilkins, Inc.