Heme oxygenase-1 promoter region (GT)n polymorphism associates with increased neuroimmune activation and risk for encephalitis in HIV infection

A cohort of 442 HIV-positive (HIV+) and 112 HIV-negative (HIV−) subjects was selected from the National NeuroAIDS Tissue Consortium (NNTC) [20] autopsy cohort for genotyping and analysis of RNA expression within dorsolateral prefrontal cortex tissue (DLPFC). This cohort was assembled by the Texas NeuroAIDS Research Center and has been described in a prior report [21, 22]. The 442 HIV+ cases included subjects who had a pathologically confirmed encephalitis (HIVE+; n = 92). NNTC site neuropathologists rendered nosological diagnoses of HIVE as guided by the criteria of Budka et al. [23]. See Table 1 for summarized demographic and HIV disease parameter data. Further details on the demographic and clinical data of this cohort were described previously [21, 22].

DNA was isolated from the DLPFC using the DNA Extraction kit (Agilent Technologies, catalog #200600) according to the manufacturer’s instruction, and each sample was standardized to a 20 ng/μl concentration. The sequences of the primers were taken from Seu et al. (2011). The forward primer sequence was labeled at the 5′ end with 6-FAM: 5′-FAM-CCAGCTTTCTGGAACCTTCTG-3′. The reverse primer sequence was unlabeled: 5′-GAAACAAAGTCTGGCCATAGGA-3′. The samples were amplified using the KAPA HiFi PCR Kit (Kapa Biosystems, cat# KK2101) on a thermocycler (Eppendorf Mastercycler RealPlex2) using a touchdown PCR protocol. Samples were initially heated to 95 °C for 10 min, then cycled between 95 °C (20 s) for denaturing, 62 °C (20 s) for annealing, and 72 °C (45 s) for elongating. The touchdown protocol started the annealing temperature at 62 °C and dropped 0.5 °C during each subsequent cycle for the first 15 cycles. The remaining 35 cycles maintained the annealing temperature at 55 °C, followed by a final hold of 72 °C for 10 min. Approximately 1 ng of DNA was used for each sample, and a final primer concentration of 250 nM for both forward and reverse primers was used. PCR products were run on a 2% agarose gel to ensure quality control of amplification of the target sequence. The resulting products were run on the 3130XL Capillary Sequencer (Applied Biosystems) by the Penn Genomics Analysis Core (PGAC) in the Perelman School of Medicine at the University of Pennsylvania (Philadelphia, PA). Sizing was analyzed with GeneMapper (Applied Biosystems) and gated for products 180–300 basepairs in length. All samples were run at least twice from independent PCR reactions to ensure accurate and reproducible sizing and were determined to be accurate within ± 1 GT repeats. A(-413)T SNP (rs2071746) genotyping was performed using TaqMan SNP Genotyping Assay (ThermoFisher, Assay ID: C__15869717_10) per manufacturer’s instructions. Briefly, two SNP target-specific oligonucleotide probes, one for the A-allele and one for the T-allele, with distinct reporter dyes at the 5′ end and a non-fluorescent quencher at the 3′end were used in PCR of genomic DNA with forward and reverse primers covering the region containing the SNP. Exonuclease activity of the DNA polymerase cleaved only probes hybridized to their target sequence, thus releasing the reporter dye and increasing reporter fluorescence. The change in fluorescence of each of the two reporter dyes associated with the two probes determined the presence or absence of the A and/or T SNP allele in the genomic DNA sample.

DLPFC was dissected from Brodmann areas 9 or 10. RNA was prepared using the RNeasy Lipid Tissue Mini Kit (Qiagen) as previously described [24, 25]. Briefly, 100 mg of brain tissue was dissected on dry ice and homogenized in a minibead beater. RNA was extracted with chloroform and centrifuged in RNeasy mini spin columns, washed, and eluted. Expression of brain mRNAs was quantified after cDNA synthesis from mRNA samples using Taq-Man Universal PCR Master Mix (Applied Biosystems) by qPCR as previously described [24]. Duplicate qPCR reactions were run and relative mRNA expression was calculated using the ΔΔC_t method (compared to GAPDH mRNA expression) using commercially available primer and probe sets from Applied Biosystems: CD19, Hs99999192_m1; CD38, Hs01120071_m1; CD163, Hs01016661_m1; CD68, Hs00154355_m1; CD8A, Hs01555600_m1; GAPDH, Hs99999905_m1; GZMB, Hs01554355_m1; IRF1, Hs00971959_m1; ISG15, Hs00192713_m1: MX1, Hs00182073_m1; PECAM1, Hs00169777_m1; VWF, Hs00169795_m1.

In the majority of cases, blood and CSF samples were obtained within 6 months of subject death. Plasma and CSF viral loads were determined with the Amplicor HIV-1 Monitor test v1.1 through v1.5 (Roche) and expressed as HIV copies per milliliter. Brain parenchyma HIV RNA was quantified from brain-extracted RNA as previously described and expressed as HIV copies per gram of tissue [24, 25]. Briefly, 1 μg microgram of brain RNA and 1 μM of antisense primer 84R were used in 20 μl reaction (iScript cDNA Synthesis Kit, Bio-Rad Laboratories). Four microliters of cDNA was used for 25 μl quantitative real-time PCR (qPCR) by using JumpStart Taq ReadyMix for Quantitative PCR (Sigma) and SmartCycler (Cepheid). Results were standardized against a known brain secondary standard. CD4+ T-lymphocyte counts were determined by flow cytometry and performed at each NNTC site’s Clinical Laboratory Improvement Amendments (CLIA)–certified, or CLIA equivalent, medical center laboratory.

All group quantifications are expressed as median ± 95% confidence interval. Comparisons of distributions of categorical variables were determined by chi square analysis. Comparisons of continuous variables between groups were performed using the non-parametric Mann-Whitney test or the Kruskal-Wallis test. Analyses of linear trends were performed by Spearman rank correlation. Statistical input support was provided by the Biostatics and Data Management Core, Center for AIDS Research, Perelman School of Medicine, and University of Pennsylvania.

To determine the potential role for the HO-1 promoter region (GT)n repeat length polymorphism (referred to hereafter as HO-1 (GT)n) in HIV-associated neuropathogenesis, we obtained DNA samples from a well-characterized autopsy cohort from the National NeuroAIDS Tissue Consortium (NNTC) (Table 1) [21, 22]. This cohort included HIV-negative individuals (HIV−, n = 112) and HIV-positive individuals with and without the post-mortem diagnosis of HIV encephalitis (HIVE) (HIVE+, n = 92; HIV+/HIVE−, n = 350). From the DNA obtained, we successfully genotyped the HO-1 (GT) repeat lengths in 549 of 554 (99.1%) of the cohort subjects.

In this composite cohort, the HO-1 (GT)n repeat lengths ranged from 13 to 44, with a trimodal distribution of peaks at 23, 30, and 39 repeat lengths (Fig. 1a). This is consistent with previous population cohort studies of HO-1 (GT)n repeat lengths [13]. Based on this trimodal distribution in our cohort, we assigned HO-1 (GT)n repeat length-based allelic genotypes, defined as short ‘S’ (27), medium ‘M’ (27–34), or long ‘L’ (> 34) (GT)n repeat lengths (Fig. 1a). Given our hypothesis that shorter HO-1 (GT)n repeat lengths would be protective against HIV CNS disease, we determined associations with the HO-1 (GT)n polymorphism by two distinct approaches: (i) examining associations with the presence of a short ‘S’ allele, that is having a high HO-1 expressing/inducing promoter, and (ii) examining correlations with the repeat length of an individual’s shortest HO-1 (GT)n allele, that is the length of the repeat in their most expressive/inductive HO-1 promoter. Using these two approaches, we found no significant associations between the HO-1 (GT)n polymorphism and age at death, post-mortem interval, gender, ethnicity, or HIV status (data not shown).

However, as previously described [13], we did observe a significant difference in the HO-1 (GT)n allelic distribution between self-identifying Caucasians and African Americans, the two largest racial groups in this cohort (Fig. 1b; p < 0.001, Kolmogorov-Smirnov test). The most common HO-1 (GT)n allele length of 30 (‘M’ allele) was observed in both Caucasians and African Americans, with frequencies of 46 and 20%, respectively. In contrast, more African American individuals had long ‘L’ alleles in comparison with Caucasians (37 vs 5%, respectively). These data demonstrate that our autopsy cohort expresses HO-1 (GT)n repeat lengths in a distribution expressed in other clinical population cohorts that have been used to assess effects on clinical- and biomarker-associated outcomes in various disease states.

We also genotyped the HO-1 promoter region A(-413)T SNP (rs2071746), which is also reported to affect HO-1 promoter transcriptional activity [11] and affect clinical disease outcomes, albeit to a lesser extent than the HO-1 (GT)n polymorphism [26‐28]. We were able to reliably genotype this HO-1 A(-413)T SNP in 553 or 99.8% of the subjects in our cohort and found an overall frequency of the ‘A’ allele of 51.4% and ‘T’ allele frequency of 48.6% (Fig. 1c). In this cohort, the allelic and genotypic frequencies of the A(-413)T SNP were in Hardy-Weinberg equilibrium, although their frequencies differed significantly between Caucasians and African Americans (Fig. 1c, d). African Americans had a higher prevalence of ‘T’ alleles in comparison with Caucasians (65.3 versus 34.7%, respectively). Notably, the A(-413)T SNP and the HO-1 (GT)n repeat length genotypes were not independently expressed in our cohort. Specifically, the ‘A’ SNP allele associated significantly with medium ‘M’ HO-1 (GT)n alleles, whereas the ‘T’ SNP allele associated significantly with the presence of a short ‘S’ or long ‘L’ HO-1 (GT)n allele (data not shown).

Given our previous observations demonstrating a significant correlation between reduced HO-1 protein expression within the brain frontal cortex and the diagnosis of HIV encephalitis (HIVE) [2], we first asked whether the HO-1 (GT)n repeat length associated with an altered risk for having pathologically confirmed HIVE. We found that presence of at least one short ‘S’ HO-1 (GT)n allele correlated with a significantly lower risk of HIVE (p = 0.04) with an odds ratio of 0.62 (95% CI 0.39–0.98) (Fig. 2a). Notably, the group of individuals with two short alleles, ‘SS’, had the lowest HIVE prevalence (13%) among all HO-1 (GT)n genotypes, while individuals with two long alleles, ‘LL’, had the highest HIVE prevalence (31%; Fig. 2b). There was no correlation between the HO-1 promoter A(-413)T SNP and HIVE prevalence (Fig. 2c). These data support the hypothesis that increased HO-1 promoter inducibility associated with a shorter HO-1 promoter region (GT)n repeat length may be at least partially protective against neuropathogenic processes linked to the development of HIVE.

Previous studies suggested that HO-1 may inhibit HIV infection and/or replication [29], providing a potential mechanism by which differential expression of HO-1 could affect HIV pathogenesis, both in the periphery and CNS. In our previous studies, however, we did not observe any effect of induction or suppression of HO-1 expression, or application of HO-1 enzymatic inhibitors, on HIV viral replication in monocyte-derived macrophages [2]. These in vitro findings are consistent with our failure to observe associations between either the HO-1 (GT)n repeat length or the A(-413)T SNP and plasma viral load, cerebrospinal fluid (CSF) viral load, and brain parenchyma viral load, which is presumed to reflect primarily the macrophage/microglial reservoir (Fig. 3 and Figure S1, see Additional file 1). We also failed to observe correlations between either of the HO-1 polymorphisms examined and blood CD4 T-lymphocyte counts (Fig. 3 and Additional file 1: Figure S1). The lack of correlation between either HO-1 polymorphism and viral load in these compartments, or blood CD4 T-lymphocyte count, also were lacking in any subgroup examined: (i) HIV-infected individuals without HIVE, (ii) HIV-infected Caucasians, or (iii) HIV-infected African Americans (data not shown). These data suggest that HO-1 does not directly modulate HIV replication in vitro or in vivo.

Given the lack of correlation between the HO-1 (GT)n repeat length and markers of systemic HIV disease progression (plasma viral load and CD4 T-lymphocyte count) and the previously identified role for HO-1 in modulating inflammation, we hypothesized that the HO-1 (GT)n repeat length would correlate with CNS markers of immune activation and inflammation. To address this, we quantified RNA expression of selected markers of immune activation and inflammation, including type I interferon (MX1 and ISG15) and macrophage (CD163) markers, which we had previously demonstrated to correlate with decreased prefrontal cortex HO-1 protein expression in HIV-infected individuals [2]. In HIV-infected individuals without HIVE, the presence of a short ‘S’ allele (Fig. 4a–f) and the repeat length of the shortest HO-1 promoter region (GT)n allele (Fig. 5a–f) significantly correlated with lower expression of type I interferon-inducible genes (MX1 and ISG15), type I/II interferon-inducible gene IRF1, and T-lymphocyte activation genes (GZMB and CD38, but not CD8A). The macrophage/microglial marker CD68 was significantly lower (p = 0.04), in HIV-infected individuals with a short ‘S’ allele (Fig. 4g); however, there was no significant correlation between CD68 and shortest HO-1 (GT)n repeat length (Fig. 5g). Additionally, presence of a short ‘S’ allele and the shortest HO-1 (GT)n repeat length did not correlate with expression of CD163, the B-lymphocyte maker CD19, or with the endothelial markers PECAM1 and VWF (Fig. 4 and Fig. 5h–j). Finally, there were no significant associations between the HO-1 promoter region A(-413)T SNP and expression of any of the neuroimmune markers examined (Figure S2, see Additional file 1).

To confirm that the significant correlations observed between type I interferon and T-lymphocyte neuroimmune markers and the HO-1 (GT)n polymorphism were not being driven by differences in racial subgroup proportions, we compared the expression level of each marker that was significant in the full cohort in Caucasians and African American HIV-infected subjects with and without short ‘S’ alleles (Fig. 6). For MX1, the gene most highly significantly different between groups with or without a short ‘S’ allele in the full HIV-infected cohort (Fig. 4a), expression in HIV-infected Caucasian and African American subgroups was significantly lower in individuals with a short ‘S’ allele compared to those without a short ‘S’ allele (Fig. 6a). Moreover, ISG15 expression was significantly lower in African Americans with a short ‘S’ allele, and IRF1 and CD38 were significantly lower in Caucasians with a short ‘S’ allele compared to individuals from the same racial subgroup without a short ‘S’ allele (Fig. 6b–d). For all remaining comparisons in this racial subgroup analysis, expression of each marker was still lower in Caucasians and Africans Americans with short ‘S’ alleles compared to those without short ‘S’ alleles; however, these did not reach statistical significance, potentially as a result of reduced statistical power due to the significantly smaller n in these subgroups (Fig. 6). These subgroup analyses suggest that differences in racial identity (Caucasian or African American) that associate with different HO-1 (GT)n repeat lengths are not driving our observed associations between HO-1 (GT)n repeat length and brain type I interferon and T-lymphocyte neuroimmune marker expression. However, these results do not address or rule out a contributing effect of race on expression levels of these neuroimmune markers within the CNS. Additionally, the significant correlations observed between type I interferon and T-lymphocyte neuroimmune markers and the HO-1 (GT)n polymorphism were highly significant in the full HIV cohort, despite many of these markers positively correlating more significantly and stronger with plasma and CNS viral loads and CD4+ T-lymphocyte count (Table 2). These data suggest that inherent differences in HO-1 promoter activity resulting in higher HO-1 expression may decrease HIV-associated CNS immune activation and neuroinflammation despite significant differences in expression driven by HIV replication and systemic disease progression.