Alleles of the gene encoding IL-1[alpha] may predict control of plasma viraemia in HIV-1 patients on highly active antiretroviral therapy

Objective: Some HIV patients treated with highly active antiretroviral therapy (HAART) do not resolve their plasma viraemia or HIV RNA can reappear after a period of virological control. We investigate whether polymorphisms in cytokine genes affect the control of plasma HIV RNA over 5 years on HAART.

Design: The study utilized adult HIV-infected patients in Western Australia. Plasma HIV-RNA levels were assessed from commencement of HAART in patients who had a CD4 T-cell count less than 100 cells/μl before HAART and achieved immune reconstitution assessed by CD4 T-cell counts.

Results: Control of plasma viraemia could be predicted from carriage of allele 2 at position -889 in the IL1A gene (IL1A-889*2). This was significant when assessed by the proportion of patients with a plasma HIV-RNA level of 400 copies/ml or less (P = 0.002). At 48 months post-HAART, proportions were approximately 0.76, 0.51 and 0.32 for IL1A (1,1), (1,2) and (2,2) patients, respectively. The outcome was independent of the patients' CD4 T-cell counts before or on therapy, drug regimen or age. Polymorphisms in IL6, TNFA, IL1B or IL12B had less significant effects, which became marginal when IL1A was included in the statistical model. IL1A-889 was in linkage disequilibrium with a non-synonymous polymorphism at IL1A+4845.

Conclusion: Alleles carried at IL1A-889 or IL1A+4845 may predict the control of HIV replication in previously immunodeficient patients responding to HAART.

Introduction

A proportion of HIV patients treated with highly active anti-retroviral therapy (HAART) do not resolve their plasma viraemia or HIV RNA can reappear after a period of virological control. Virological failure may follow interruptions to therapy, the emergence of drug-resistant virus or non-adherence to therapy [^1,2] and may be resolved by restarting or changing the HAART regimen. However, some patients experience virological failure despite good adherence, suggesting that other mechanisms are involved. These may include genetic factors that affect antiretroviral drug metabolism or HIV disease pathogenesis, but no clear candidates have emerged to date. A single nucleotide polymorphism at position +3435 (exon 26) of the multidrug-resistance transporter 1 (MDR1) gene had no effect on virological outcome [³], or promoted a trend towards early virological failure [⁴]. Some studies have associated the CCR5Δ32 genotype or CCR5 promoter polymorphisms with early virological failure [⁵], whereas others found no association with CCR5 or the CX3CR1 genotype [^4,6].

In this study we address associations between polymorphisms in genes encoding pro-inflammatory cytokines and virological outcome after HAART, because immune activation contributes to HIV disease pathogenesis and persists in patients with satisfactory responses to HAART [^7,8]. The study utilizes patients who began therapy with low CD4 T-cell counts, as they often present critical issues in medical management. The patients' immunological response to HAART confirms their physicians' assessment of reasonable adherence to therapy.

Materials and methods

Patients

Adult HIV-infected patients attending Royal Perth Hospital (Western Australia) and affiliated clinics were studied. Royal Perth Hospital is the state referral centre for patients with HIV/AIDS. The demographic characteristics of patients are similar to other Australian cities, with the majority of patients acquiring HIV infection through male to male sexual activity. Plasma viral loads were assessed from the commencement of an initial triple antiretroviral therapy regimen, including a protease inhibitor (PI) or a non-nucleoside reverse transcriptase inhibitor (NNRTI) with various nucleoside reverse transcriptase inhibitors (NRTI). Analyses were restricted to patients who had a CD4 T-cell count less than 100 cells/μl before HAART, and achieved immune reconstitution assessed by an increase in their CD4 T-cell count by greater than fourfold or to more than 200 cells/μl. The patients attended clinic regularly and were monitored for changes in the plasma HIV-RNA level and for the emergence of antiretroviral-resistant HIV. Their therapy was adjusted in response to viral breakthrough or drug toxicities in accordance with Australian treatment guidelines.

Quantitation of T-cell subsets and HIV RNA

T-lymphocyte subsets were quantitated by standard flow cytometry. Plasma HIV RNA was assayed by Amplicor version 1.0 (standard protocol, 400-750 000 copies/ml) and later by version 1.5 (ultrasensitive protocol, 50-75 000 copies/ml) (Roche Diagnostic Systems Inc., Branchburg, NJ, USA) by a routine clinical laboratory. A threshold of 400 copies/ml was used for statistical analyses.

Genotyping

IL1A-889 alleles were assessed as a restriction fragment length polymorphism (RFLP) [⁹]. Briefly; 150 ng genomic DNA was placed in a 50 μl reaction mix containing 800 μM dNTP, 2 mM magnesium chloride, 800 μM primers. DNA flanking IL1A-889 was amplified with 1 U Taq platinum polymerase (Gibco/BRL, UK) [96°C 15 min, (94°C 1 min, 50°C 1 min, 72°C 2 min) × 45 cycles, 72°C 5 min]. Amplicons were digested overnight with Nco1 (37°C), separated by electrophoresis on 15% polyacrylamide or 3% agarose gels and visualized using ethidium bromide. IL1A-889 allele 1 (C) yielded 83 + 16 basepair (bp) bands, whereas allele 2 (T) yielded a 99 bp band.

IL1A+4945 alleles were also determined by polymerase chain reaction-RFLP [¹⁰]. Cycling required 95°C 1 min, (94°C 1 min, 56°C 1 min, 72°C 2 min) × 35 cycles, 72°C 5 min. Amplicons were digested overnight with Fnu4H1 (37°C), separated by electrophoresis in 3.5% agarose gels, and visualized using ethidium bromide. IL1A+4845 allele 1 (G) yielded 124 + 75 + 29 bp bands, whereas allele 2 (T) yielded a 153 + 75 bp band.

Polymerase chain reaction-RFLP assays were used to type polymorphisms at IL6-174, TNFA-308, IL1B +3935 and IL12B (3'UTR) and IL12B (promoter) as previously described [¹¹]. Operators were blinded to clinical outcomes when genotypes were assigned.

Statistical analyses

The proportions of patients with plasma HIV-RNA levels of 400 copies/ml or less were analysed over time from the commencement of HAART using a random effects mixed model. This allowed the proportions of values 400 copies/ml or less to vary flexibly over time, and incorporated correlation caused by the repeated measures on the same individual. Significance levels for overall differences and differences between genotypes were assessed by likelihood ratio tests. Analyses were carried out using the S-PLUS statistical package (S-PLUS 6 for Windows; Insightful Corporation, Seattle, USA). The time from the commencement of HAART to the first HIV-RNA level of 400 copies/ml or less (time to first response) and the time to a rebound above 400 copies/ml for at least 2 months (time to first rebound) were analysed using Kaplan-Meier survival methods.

Results

IL1A-889 genotype is associated with control of plasma HIV RNA on highly active antiretroviral therapy

Eighty-one patients fulfilled the entry criteria (Table 1). Homozygous and heterozygous patients were well matched for baseline HIV RNA, baseline CD4 T-cell count, age and months of dual NRTI therapy before HAART. The average follow-up was similar in the three groups, with 71, 77 and 80%, respectively, having at least one HIV-RNA measurement 5 years post-HAART. Analyses were restricted to a 5-year follow-up, giving 1768 HIV-RNA values, with a mean of 21.8 per patient. All patients received NRTI therapy consisting of stavudine or zidovudine with didanosine (n = 6) or lamivudine (n = 75), plus either a PI (n = 74) or an NNRTI (n = 7). Treatment regimens were similar in the three IL1A groups. No adjustments were made for changes of treatment regimen because these represent outcomes.

Table 1

The proportion of homozygous IL1A (1,1) patients with 400 copies/ml or less HIV-RNA increased over the first 6 months and was then maintained at approximately 70% to 40 months (Fig. 1a), with an increase to over 80% by 60 months. IL1A (1,2) patients showed a similar initial response, but the proportion of patients with undetectable HIV-RNA levels declined to approximately 50% by 24 months [P = 0.003 compared with IL1A (1,1)]. Homozygous IL1A (2,2) patients responded more slowly in the first few months of treatment and had a higher rebound rate, leading to less than 30% with undetectable HIV-RNA levels at 3 years, increasing to 40% after 5 years [P = 0.007 compared with IL1A (1,1); Fig. 1a].

Fig. 1

Times to first response were similar for IL1A (1,1) and (1,2) patients, but were marginally longer for IL1A (2,2) patients (P = 0.22, Fig. 1b). Times to first rebound were shortest for the (2,2) group and similar for IL1A (1,1) and (1,2) patients (P = 0.01, Fig. 1c). Although these analyses are approximate and utilize only first response and rebound, they are consistent with the patterns shown in Fig. 1a. The significant difference between IL1A (1,1) and (1,2) in Fig. 1a suggests that the IL1A (1,1) group had a higher rate of further response after the initial rebound than did the IL1A (1,2) group.

The effect of IL1A was independent of changes in CD4 T-cell count and pre-treatment disease status

When incorporated separately into the analysis, baseline HIV RNA (P = 0.005), previous duration of dual NRTI therapy (P = 0.001), baseline CD4 T-cell count (P = 0.002) and age (P = 0.003) all affected post-HAART HIV-RNA levels, but none abrogated the IL1A effect individually or collectively. Lower baseline HIV RNA, a shorter duration of previous NRTI therapy, higher baseline CD4 T-cell counts and older age were associated with a higher proportion of HIV-RNA levels of 400 copies/ml or less post-HAART. Similar results were obtained when the analysis was restricted to Caucasian male patients (n = 71; 1522 HIV-RNA measurements; IL1A (1,2) versus (1,1), P = 0.05; IL1A (2,2) versus (1,1), P = 0.01).

Corresponding longitudinal analyses of CD4 T-cell counts found no significant differences between IL1A (1,1) and (1,2) (P = 0.1), or between IL1A (1,1) and (2,2) (P = 0.3). To confirm that IL1A genotypes affect HIV RNA rather than CD4 T-cell counts, an expanded group of HIV patients with nadir CD4 T-cell counts below 100 cells/μl, who had received HAART for over 12 months and maintained a plasma HIV-RNA level of 50 copies/ml or less for at least 6 months was selected. The patients were ranked by CD4 T-cell count and genotyped (Table 2). CD4 T-cell counts did not affect the distribution of IL1A alleles in this analysis. The distribution was also similar in HIV-seronegative individuals and an expanded unselected HIV patient pool from Perth and Sydney.

Table 2

Ten patients homozygous for each allele at IL1A-889 and 10 heterozygotes were genotyped at IL1A+4845. One patient carried IL1A-889 (1,2), IL1+4845 (1,1), whereas all others had the same genotype at both alleles. This indicates strong linkage disequilibrium (P < 0.00001).

Other cytokine genotypes have minor effects on plasma HIV RNA

TNFA-308 had no significant associations with the attainment of an HIV-RNA level of 400 copies/ml or less (data not shown; P = 0.26). Despite different early responses, IL6-174 (C,G) and (C,C) patients had a similar probability of attaining an HIV-RNA load of 400 copies/ml or less (P = 0.07; Fig. 2a). IL6-174 (G,G) patients were slightly less likely to have undetectable virus between 6 and 36 months post-HAART [P = 0.03 relative to IL1A-889 (C,G) or (C,C)]. This difference remained significant (P = 0.03) when IL1A-889 alleles were included in the model.

Fig. 2

IL12B (3'UTR) (1,2) (P = 0.03) and (2,2) (P = 0.008) patients were less likely to achieve 400 copies/ml or less HIV RNA than (1,1) patients (Fig. 2b). This effect was reduced when adjusted for IL1A-889 alleles (P = 0.06 and P = 0.03, respectively). IL12B (promoter) (2,2) patients were less likely to have less than 400 copies/ml HIV RNA than (1,1) or (1,2) patients (P = 0.007), and the difference was reduced by the inclusion of IL1A (P = 0.025). IL12B (promoter) (1,1) and (2,2) patients had a similar outcome (P = 0.10) (Fig. 2c). The promoter and 3'UTR alleles of IL12B were in moderate linkage disequilibrium. This was significant in an expanded group of immune reconstituted patients (n = 94, P = 0.025), but not within the group typed for IL1A-889 (n = 79, P = 0.18). Carriage of allele 2 was more common for the promoter polymorphism than for the 3'UTR (70/94 versus 34/94, P < 0.00005).

IL1B+3935 (1,1) patients were slightly more likely to achieve an undetectable viral load than heterozygotes (P = 0.03), with the difference appearing after 3 years on treatment (Fig. 2d). Only two patients carried IL1B+3953 (2,2), so they were excluded from this analysis. The effect of IL1B+3953 was abrogated by the inclusion of IL1A-889 (P = 0.90). The IL1A-889 and IL1B+3953 alleles were in linkage disequilibrium (P < 0.005), so the difference between their effects on viral load is striking.

Discussion

In this study we showed that alleles of IL1A-889 are associated with the attainment and maintenance of undetectable levels of HIV RNA in the plasma of HIV-1 patients receiving HAART. Our study did not investigate HIV resistance to antiretroviral drugs. Reasonable adherence to HAART was described by the physicians and confirmed by their favourable CD4 T-cell responses, but complete adherence could not be guaranteed. However, all patients had regular clinical reviews and were managed according to standard procedures in a single centre. Therefore, the observed effects of IL1A genotypes cannot be explained by the inclusion of a genetically distinct cohort from a centre with different patient management practices. Other cytokine polymorphisms had smaller effects than IL1A, which varied with time on treatment and so have less apparent prognostic value. Our analyses were not adjusted for multiple comparisons because our focus was on the IL1A gene, with subsequent analyses of the other genes for comparison. However, a conservative Bonferoni correction for the six genes considered would place the significance of the effect of IL1A alleles at P = 0.012. Furthermore, the retention of significance for IL1A after comprehensive correction for other factors (including the remaining five genes) reinforces the significance of the differences reported.

IL1A and IL1B alleles affect several immunopathological disorders. IL1A-889*2 is associated with severe juvenile rheumatoid arthritis [¹²]. Without HIV, individuals with severe periodontal disease have an increased frequency of a haplotype comprising allele 2 at IL1A-889 and IL1B+3953 [¹³], which displayed moderate linkage disequilibrium in that cohort [¹⁴]. A more recent study showed that the clearest link was with IL1B+3953, and associated allele 2 with higher IL-1 protein in culture supernatants [¹⁵]. In Western Australian HIV patients, the carriage of IL1A-889*2 was associated with marginally more severe periodontal disease, but this did not reach statistical significance [⁹], and subsequent studies questioned the association in healthy individuals [^16,17].

Here IL1B+3953 had a limited effect on plasma HIV RNA (Fig. 2d), so we suggest the mechanism by which IL1A-889 affects viral load does not involve functions shared by IL-1α and IL-1β. IL1A+4845 is in linkage disequilibrium with IL1A-889 and was associated with atopy and nasal polyposis in asthmatic adults [^18,19]. Linkage disequilibrium between IL1A-889 and IL1A+4845 was confirmed here and should be considered as an explanation for our results. IL1A+4845 changes an alanine [allele 1 (G)] to a serine [allele 2 (T)] at position 114 of IL-1α [¹⁰]. IL-1α is translated as a 31 kDa precursor protein and cleaved by the enzyme m-calpain to form a 17 kDa C-terminal fragment (able to interact with the same IL-1 membrane receptors that can bind IL-1β) and a 16 kDa N-terminal domain. The cleavage site lies between amino acids 117 and 118, close to the residue affected by IL1A+4845. Calpain digestion of other proteins is promoted by the phosphorylation of residues near the active site [^20,21]. Serine can be phosphorylated, so the additional serine residue at position 114 could promote digestion, enhancing the release of the C-terminal fragment from the cell. Cells homozygous for IL1A-889*2 (and therefore usually IL1A+4845*2) release more IL-1α protein [²²]. The N-terminal domain of IL-1α contains nuclear localization signals and can promote oncogenesis [²³] or interact with RNA processing pathways and promote apoptosis [²⁴]. These mechanisms have been demonstrated in mesangial cells [²³], and could be relevant in other cells that produce calpain (such as monocytes and lymphoid cells) [²⁵]. Published studies present several mechanisms by which N-terminal IL-1α may promote HIV disease, including the transformation and enhanced survival of infected T cells and the rapid apoptosis of infected antigen-presenting cells [^24,25]. T-lymphocytes do not usually produce IL-1α but can do so after HTLV-1 infection [²⁶]. This can be tested after HIV infection in vitro.

Two lines of enquiry suggest that IL1A-889 alleles are unlikely to explain our findings. (i) Position -889 does not affect transcription of the IL1A gene. A series of luciferase expression vectors containing progressive deletions of the IL1A promoter found no promoter activity around position -889 [²⁷]. Searches of transcription factor databases suggest IL1A-889 may disrupt a putative binding site for the transcription factor Skn-1, but this was named for its role in keritinocyte differentiation [²⁸] and has not been described in leukocytes. (ii) Luciferase expression vectors containing variants of the IL1A proximal promoter showed only a 5% increase in transcription with the IL1A-889 (2,2) genotype, whereas peripheral blood mononuclear cells from donors with this genotype released more than fivefold more IL-1α protein [²²]. The authors were not aware of linkage disequilibrium between IL1A-889 and IL1A+4845, and so could not explain how a promoter polymorphism (-889) affected protein rather than messenger RNA levels.

In conclusion; we suggest that IL1A genotyping may identify patients at greatest risk of virological failure, so that drug regimens with maximal efficacy and durability can be selected. Although our tests for association incorporate the size of the sample, further investigations are needed to confirm the findings prospectively in larger cohorts from diverse backgrounds [²⁹], and address the roles of adherence, resistance mutations and the HAART regime. Our results also suggest a novel mechanism of regulation of HIV replication through IL-1α, possibly involving the polymorphism at position +4845 affecting the release of the N-terminal fragment, which acts intracellularly. This should be investigated using in-vitro infections because it may lead to the development of novel therapeutic strategies.

Acknowledgements

The authors would like to thank Dr Peter Williamson for access to samples from the Westmead Hospital, Sydney. This is publication 2003-19 of the Department of Clinical Immunology and Biochemical Genetics, Royal Perth Hospital.

Sponsorship: The project was supported by the National Centre for HIV Epidemiology and Clinical Research.