Background
Heme oxygenase-1 (HO-1) is a sentinel, cytoprotective enzyme that has emerged as a critical effector for limiting oxidative stress, inflammation, and cellular injury within the central nervous system (CNS) and other tissues. The cytoprotective functions of HO-1 have been linked to its degradation of heme and the subsequent generation of carbon monoxide, biliverdin, and bilirubin, which have immunomodulatory and anti-oxidative properties [
1]. The expression of HO-1 is rapidly induced during acute cellular injury states in part through transcriptional upregulation by the transcriptional factor Nrf2. Several common genetic variations including (GT)n dinucleotide repeats and single nucleotide polymorphisms (SNPs) within the HO-1 promoter region modulate the level of promoter transcriptional activity, which suggests the likelihood of host variability in executing effective HO-1-driven protective responses to inflammation and cellular injury.
We previously demonstrated that HO-1 protein expression is decreased in the brains of HIV-infected individuals diagnosed with HIV-associated neurocognitive disorders (HAND) and that this reduction of HO-1 is associated with CNS viral load and markers of neuroimmune activation, including type I interferon responses [
2]. We also previously identified two potential mechanisms for this loss of brain HO-1 including (i) transcriptional downregulation of HO-1 expression in HIV-infected macrophages, a major CNS HIV reservoir, and (ii) IFNγ-accelerated degradation of HO-1 by the immunoproteasome in astrocytes [
3,
4]. Finally, we demonstrated that HIV infection of macrophages not only decreases macrophage HO-1 expression, but that this loss of HO-1 augments neurotoxin production from the infected macrophages [
2,
5], thus suggesting that decreased HO-1 expression within the CNS may promote neuronal injury and dysfunction through enhanced neurotoxin production.
Because of the observed relationships between brain HO-1 loss, immune activation, and neurocognitive dysfunction and because of the known associations between HO-1 promoter polymorphisms and increased risk of disease progression in numerous inflammatory diseases, we examined the relationship between HO-1 promoter polymorphisms and CNS disease state in a HIV-infected autopsy cohort. We studied two polymorphisms in the 5′-flanking region of the human HO-1/
HMOX1 gene: a (GT)n dinucleotide repeat that shows length polymorphism, and a single nucleotide polymorphism (SNP), A(-413)T (rs2071746), each of which is known to modulate HO-1 gene transcription under certain conditions. These polymorphisms may also regulate alternative splicing within the 5′ untranslated region of the HO1- gene and contribute to translational regulation [
6].
The best characterized of these polymorphisms is the HO-1 promoter region (GT)n dinucleotide repeat polymorphism. The length of this (GT)n repeat typically varies from 12 to 45 repeats with a bimodal or trimodal distribution; the most common alleles have 23 and 30 repeats [
7‐
11]. In studies using luciferase promoter constructs in transfected cell lines, the length of the (GT)n repeat has been demonstrated to modulate HO-1 promoter activity. Specifically, shorter (GT)n repeat lengths produce higher basal promoter activity and higher promoter inducibility in response to various stimuli, including oxidative stress [
7,
8]. Additional studies using primary cells and lymphoblastoid cell lines established from subjects with known (GT)n repeat lengths have confirmed these findings [
12‐
14]. Furthermore, other studies have shown that cells with shorter (GT)n repeats are more resistant to oxidative stress-induced apoptosis and have higher oxidative stress-induced HO-1 expression and enzymatic activity [
12]. Clinical cohort studies have demonstrated an association between HO-1 promoter region (GT)n repeat length and disease progression in a variety of disease states. For example, shorter (GT)n repeats are associated with better clinical outcomes in patients with emphysema [
7], coronary artery disease [
15], necrotizing pancreatitis [
16], sepsis [
17], pneumonia [
18], ischemic stroke [
19], and rheumatoid arthritis [
14], among others. Many of these disease states involve inflammation and oxidative stress, and the association of HO-1 promoter region (GT)n repeat length with disease progression further highlights the potentially significant role for HO-1 in modulating inflammation and oxidative stress in varied pathological conditions, including HIV infection.
To this end, a recent study by Seu et al. demonstrated that shorter HO-1 promoter region (GT)n repeat lengths in HIV-infected African Americans correlated with decreased plasma viral load and soluble CD14, a marker of monocyte/macrophage activation, suggesting a potential link between HO-1 promoter region (GT)n repeat length and HIV disease progression [
13]. Moreover, this study demonstrated that fewer (GT)n repeats associated with greater HO-1 protein expression in peripheral blood mononuclear cells and CD14+ monocytes in HIV-infected individuals [
13]. The potential role for HO-1 promoter region (GT)n repeat length and neurological disease in HIV infection has not been addressed thus far.
To identify possible associations between the HO-1 promoter region (GT)n repeat length and CNS HIV disease, we quantified the HO-1 promoter region (GT)n repeat lengths in individuals in a large, well-characterized HIV brain autopsy cohort and determined correlations with the pathologic diagnosis of HIV encephalitis (HIVE) and expression of brain tissue-derived markers of immune activation and inflammation. Our data reveal significant associations between the HO-1 promoter region (GT)n repeat length and (i) the pathological diagnosis of HIV encephalitis, (ii) expression levels of type I interferon response genes, and (iii) markers of T-lymphocyte activation in the brains of HIV-infected individuals. These findings suggest that an individual’s HO-1 promoter region (GT)n repeat length serves as genetic determinant of susceptibility to CNS HIV-driven neuropathogenesis associated with neuroinflammation, and they further suggest that HO-1 promoter region (GT)n repeat length genotyping in HIV-infected individuals should be examined as a potential means for identifying an individual’s potential risk for neurocognitive impairment.
Discussion
The relatively high prevalence of symptomatic HAND in virally suppressed antiretroviral (ART)-treated HIV-infected individuals underscores the need for adjunctive therapy for improved neuroprotection in HIV treatment strategies [
30,
31]. Accumulating evidence suggests that CNS inflammation and oxidative stress can persist within the CNS compartment despite the sustained viral suppression provided by ART [
32‐
34], and this suggests that targeting these processes might offer additional neuroprotection. Our previous studies identified reduced brain HO-1 protein expression in HIV-infected individuals with or without ART treatment, and these studies further showed that this decreased HO-1 expression correlated with increased CNS HIV replication, neuroimmune activation, and cognitive dysfunction (HAND diagnosis) [
2]. These results suggest that induction of HO-1 may serve as a protective strategy against neuroinflammation, neuronal injury, and associated neurocognitive impairment in HIV-infected individuals [
2,
4,
5].
We now demonstrate a potentially critical link between HO-1 and an individual’s inherent risk for HIV-mediated CNS neuronal injury by identifying the HO-1 promoter region (GT)n repeat length as a risk factor for the development of neuroinflammation and HIV encephalitis (HIVE). We showed that shorter HO-1 promoter region (GT)n repeat alleles significantly correlated with lower brain expression of type I interferon-inducible genes (
ISG15 and
MX1), type I/II interferon-inducible gene
IRF1, and T-lymphocyte activation genes (
GZMB and
CD38) and a reduced risk of the post-mortem diagnosis of HIVE. Other studies have shown that shorter HO-1 (GT)n repeat lengths are correlated with increased HO-1 gene expression and induction in in vitro model systems and primary cells [
7,
8,
12,
13]. While we did observe that African Americans in our cohort have greater HO-1 (GT)n allele length variation that is shifted towards longer repeats compared to Caucasians, as has been previously described [
13,
35‐
37], subgroup analyses demonstrate that these differences in HO-1 (GT)n allele length distributions between these groups are not driving the primary results observed in the full cohort. The reason for increased frequency of longer HO-1 (GT)n alleles in populations of African ancestry is unclear; however, some studies suggest that longer HO-1 (GT)n repeats may be enriched secondary to a protective effect against malaria [
35,
36]. In contrast to our observations of significant correlations with shorter HO-1 (GT)n repeat lengths, we did not find any significant correlations between the same markers of HIV-driven neuroinflammation and the A(-413)T SNP in the HO-1 promoter, suggesting this SNP may play a limited or secondary role to that of the HO-1 promoter region (GT)n polymorphism. Importantly, we also did not observe a correlation between the HO-1 (GT)n repeat length and either plasma or CNS viral load or CD4+ T-cell count, suggesting that differential HO-1 transcriptional regulation directly modulates CNS inflammation independent of viral replication and systemic disease progression. Our data thus strongly implicate for the first time a significant CNS effect of HO-1 on HIV-driven neuroinflammation and neurodegeneration, which varies with an individual’s HO-1 promoter genotype, and which itself might be a predictor of risk for development of HAND in HIV infection.
The mechanisms by which HO-1 may modulate HIV neuropathogenesis are not fully apparent, but they likely include effects on cellular oxidation state, glial cell neurotoxin production, and modulation of neuroinflammation. Beyond the direct ability of HO-1 to detoxify free heme, a potent pro-oxidant product of cellular metabolism as well as hemoglobin degradation, HO-1 has been shown to limit cellular and tissue injury in response to non-heme-driven oxidative stress. Adult HO-1-deficient mice express elevated markers of oxidative stress, including oxidized proteins and lipid peroxidation [
38,
39] and multiple primary cell lineages (e.g., fibroblasts, vascular smooth muscle, astrocytes) derived from HO-1-deficient mice are highly susceptible to oxidant-induced injury [
40‐
42]. Moreover, in cell culture systems, induction of HO-1 has been shown to protect cells of different lineages (including astrocytes [
43,
44] and neurons [
45,
46]) from oxidative stress insults [
47‐
50].
We recently demonstrated another mechanism by which HO-1 can specifically modulate neuronal injury, with particular relevance to HIV infection. We showed that loss of HO-1 expression by HIV replication in human macrophages increases their release of glutamate at levels sufficient to induce excitotoxic injury in neurons [
2,
5]. Furthermore, this loss of HO-1 protein in HIV-infected macrophages is selective for HO-1 among a number of antioxidant response genes examined [
2,
4]. Reduction of HO-1 protein expression in HIV-infected macrophages occurs even with viral strains with limited replication, and ART exposure at CNS-relevant concentrations fails to prevent HO-1 loss and enhanced glutamate release once infection is established [
5]. Notably, siRNA knockdown of HO-1 expression or pharmacologic inhibition of residual HO-1 enzymatic activity in HIV-infected macrophages augments glutamate release and associated neurotoxicity, while HO-1 induction decreases glutamate release and associated neurotoxicity [
2]. These studies thus suggest that individuals with shorter HO-1 promoter region (GT)n repeat length alleles might be less prone to robust CNS glutamate release and associated neurotoxicity during HIV infection and that CNS induction of HO-1 expression might augment neuroprotective effects of viral suppression by ART.
Additional neuroprotective and cytoprotective effects of HO-1 are likely. In addition to its ability to limit oxidative stress and glutamate production from HIV-infected macrophages, HO-1 can modulate immune activation and inflammation. HO-1 deficient mice exhibit chronic inflammation, as demonstrated by high peripheral white blood cell counts, high CD4+:CD8+ T-lymphocyte ratios, elevated activated CD4+ T-lymphocyte levels, increased monocyte vascular adhesion, increased baseline serum IgM, inflammatory cell tissue infiltrates (particularly in the liver), and elevated splenocyte secretion of pro-inflammatory cytokines in response to endotoxin or anti-CD3/anti-CD28 stimulation [
39,
51]. Consistent with an anti-inflammatory role for HO-1, the first identified human patient with HO-1 deficiency demonstrated enhanced systemic inflammation in addition to asplenia, intravascular hemolysis, and systemic vascular endothelial dysfunction, among other symptoms [
52,
53]. The anti-inflammatory and immunomodulatory functions of HO-1 have been studied in various immune cells, including those of the myeloid and lymphocyte lineages. In macrophages induction of HO-1 limits expression of pro-inflammatory factors including IL-6, TNFα, inducible nitric oxide synthase (iNOS), and cyclooxygenase-2 (COX-2). In monocytes induction of HO-1 suppresses expression of human leukocyte antigen-DR, CD36, and CD11B [
51,
54]. HO-1 also regulates maturation and proper functioning of dendritic cells [
55]. In T lymphocytes, HO-1 modulates T lymphocyte-mediated immunity [
56], particularly with respect to the suppressive capacity of regulatory T lymphocytes [
57]. These data highlight the important roles for HO-1 in modulating immune activation and inflammatory responses and also highlight potential mechanisms by which increased expression of HO-1, through inherent promoter polymorphism differences or exogenous induction, may reduce immune activation and inflammation within the peripheral and CNS compartments.
The decreased neuroimmune activation and neuroinflammation, as indicated by reduced type I interferon responses and T-cell activation markers in the brains of HIV-infected individuals with shorter HO-1 (GT)n alleles, may account for the decreased prevalence of HIVE in these individuals. Subjects with shorter HO-1 (GT)n alleles may have higher basal HO-1 expression and increased induction in response to oxidative stress, inflammation, and other stimuli over the course of their HIV infection, effects that could reduce neuroinflammation, immune activation, and a potential associated risk for developing HIVE. Although HIVE is rarely seen in autopsies from individuals on suppressive ART, the clinical diagnosis of HIV-associated neurocognitive disorders (HAND) remains prevalent in virally suppressed, HIV-infected individuals (~ 15% are functionally impaired) [
30,
31]. Moreover, virally suppressed ART-treated individuals express elevated markers of peripheral and CNS inflammation and immune activation, albeit to a lesser extent than untreated HIV individuals [
32‐
34]. Persistent and/or recurrent low-level inflammation and immune activation in ART-treated HIV-infected individuals could contribute to the persistence of HAND [
58‐
61]. A recent retrospective neuropathology study of simian immunodeficiency virus (SIV)-infected pigtailed macaques on ART demonstrated significant prevalence of lymphocyte-dominant inflammation in the CNS [
62], suggesting that adaptive immune responses may persist in the CNS in HIV-infected ART-treated individuals and may contribute to neuronal dysfunction and subsequent neurocognitive decline. Moreover, human studies have demonstrated that increased T-lymphocyte activation [
63,
64] and type I interferon responses [
24,
65,
66] associate with more pronounced neurocognitive dysfunction in HIV-infected individuals.
We hypothesize that shorter HO-1 (GT)n promoter repeat lengths associated with decreased neuroinflammation and neuroimmune activation might decrease the risk of HAND and progression of neurocognitive impairment in HIV-infected individuals on suppressive ART. However, our autopsy study is not ideally suited to assessing the potential contribution of the HO-1 (GT)n promoter repeat length to the development of HAND, given the potential confounding factors of variable ART use and the agonal state of individuals prior to death in autopsy-based cohort studies. However, the data presented in this study build upon our previous data to further support the hypothesis that HO-1 induction would have significant therapeutic benefit in well-suppressed HIV-infected individuals on ART who continue to have elevated markers of immune activation and inflammation and associated disease sequelae. This is additionally built upon previous studies of potentially therapeutic HO-1 inducers and their ability to limit inflammation, oxidative stress, and disease pathology in different disease states [
67‐
69]. To these ends, we are initiating a study to determine whether the presence of short HO-1 (GT)n repeat length alleles associates with HAND, neurocognitive changes over time, and/or with peripheral and CNS biomarkers in a clinically tracked, virally suppressed ART-treated, living HIV patient cohort. Future clinical studies should directly address the potential therapeutic use of HO-1 inducers for neuroprotection against HIV in ART-treated individuals.