Variations in the heme oxygenase-1 microsatellite polymorphism are associated with plasma CD14 and viral load in HIV-infected African Americans

Abstract

Heme oxygenase-1 (HO-1) is an anti-inflammatory enzyme that maintains homeostasis during cellular stress. Given previous findings that shorter length variants of a HO-1 promoter-region GT_n microsatellite polymorphism are associated with increased HO-1 expression in cell lines, we hypothesized that shorter variants would also be associated with increased levels of HO-1 expression, less inflammation, and lower levels of inflammation-associated viral replication in HIV-infected subjects. Healthy donors (n=20) with shorter GT_n repeats had higher HO-1 mRNA transcript in peripheral blood mononuclear cells stimulated with lipopolysaccharide (LPS) (r= −0.38, p=0.05). The presence of fewer GT_n repeats in subjects with untreated HIV disease was associated with higher HO-1 mRNA levels in peripheral blood (r= −0.41, p=0.02); similar observations were made in CD14⁺ monocytes from antiretroviral-treated subjects (r= −0.36, p=0.04). In African-Americans, but not Caucasians, greater GT_n repeats were correlated with higher soluble CD14 (sCD14) levels during highly active antiretroviral therapy (HAART) (r= 0.38, p=0.007) as well as higher mean viral load off-therapy (r= 0.24, p=0.04). These data demonstrate that the HO-1 GT_n microsatellite polymorphism is associated with higher levels of HO-1 expression and that this pathway may have important effects on the association between inflammation and HIV replication.

INTRODUCTION

Although host immune activation is critical for the eradication of infectious agents, the inflammatory response can also enhance pathogenesis. An example of this delicate balance is revealed by human immunodeficiency virus (HIV) disease: progressive and pathogenic infection is associated with chronic immune activation, including heightened destruction and diminished production of T lymphocytes (1), an increased frequency of T cells with an activated phenotype (2), increased T cell induction of the pro-inflammatory transcription factors, NFκB and NF-AT (3), and increased serum levels of pro-inflammatory cytokines (e.g., TNFα, IL-1 and IL-6) (4,5).

The impact of chronic inflammation on HIV (and simian immunodeficiency virus; SIV) disease progression has been well studied in both humans and non-human primates. Experiments in non-human primate models of acute SIV infection have demonstrated that increased bystander immune activation heightens viral cellular transmission by increasing the T cell target to T cell effector ratio (6). T lymphocyte activation as measured by cell surface markers has also been associated with viral load “set-point” in untreated HIV disease (7) and is associated with disease progression in both untreated and treated HIV infection (8,9).

Multiple mechanisms likely contribute to HIV-associated inflammation, including the direct effects of the virus. Recent studies have shown that circulating lipopolysaccharide (LPS) resulting from microbial translocation from the gut is a cause of systemic immune activation in chronic HIV infection (10), and is present even in HAART-suppressed subjects with undetectable viremia (11). An extension of this finding was the discovery that soluble CD14, the portion of the LPS-receptor that gets cleaved upon binding to its ligand, is elevated in HIV disease and strongly associated with enhanced mortality in chronically HIV-infected subjects (12), and is a major force driving chronic inflammation in HAART-treated patients (13).

Given the impact of immune activation/inflammation on the pace of HIV disease progression, we hypothesized that host factors controlling inflammation might be protective. In this study, we focused on heme oxygenase-1 (HO-1), the rate-limiting enzyme that initiates heme degradation and maintains cellular homeostasis during stress through its depletion of pro-oxidant heme, through generation of cytoprotective carbon monoxide and biliverdin, and through induction of ferritin by Fe²⁺ release. HO-1 is induced by a myriad of stress signals including oxidative stress, ultraviolet (UVA) radiation, and the pro-inflammatory cytokines IL-1, IL-6, TNFα (14). Furthermore, we have recently demonstrated that the HO-1 inhibitor, tin mesoporphyrin IX (SnMP), induces activation, proliferation, and maturation of naïve human T cells via interactions with CD14⁺ monocytes (15).

Given the important role that HO-1 has as an anti-inflammatory mediator, numerous studies have linked proximal promoter HO-1 polymorphisms to disease states that are driven by inflammation (e.g., graft versus host disease, ischemic stroke, coronary artery disease, etc.) [reviewed in (16)]. The GT_n microsatellite polymorphism is the best-characterized HO-1 genetic locus to date. Previous studies using promoter-luciferase assays in cell lines have shown an association between GT_n microsatellites with a small number of repeats and enhanced transcription of the HO-1 gene (17–20). Likewise, a recent study using primary endothelial cells isolated from newborns showed that HO-1 induction was more robust in the context of fewer (GT_n<23) than greater (GT_≥29) repeats (21). Although many candidate gene studies have detected associations between the short HO-1 GT_n microsatellite variant and control of inflammatory disease states, no study has examined its functional role in regulating gene expression in primary immune cells that directly influence pathogenic outcomes in HIV-infected subjects.

Here, we have studied the influence of the promoter GT_n microsatellite repeat number on HO-1 transcript expression in peripheral blood mononuclear cells (PBMCs) of healthy adults after in vitro stimulation and in PBMCs isolated from HIV-infected subjects on and off antiretroviral therapy. Next, we conducted a candidate genotyping study for two biomarkers of HIV disease progression: mean viral load and soluble CD14. Lastly, we determined the association between HO-1 expression and CD14 expression on primary blood CD14⁺ monocytes from healthy donors stimulated in vitro with LPS. In aggregate, our results suggest that the HO-1 GT_n microsatellite polymorphism is a functional determinant of HO-1 expression in primary immune cells and is associated with important biomarkers of HIV disease outcome.

RESULTS

Heme oxygenase-1 (HO-1) promoter description across HIV-infected subjects of different ethnicities

The HO-1 promoter region analyzed extends from the nucleotide position 1876 base pairs upstream and +75 base pairs downstream of the transcription start site at position +1 (Fig. 1a). This promoter region contains the GT_n dinucleotide repeat and two common SNPs (-413AT/rs2071746 and -1195AG/rs3761439) that have been analyzed in previous studies of the HO-1 gene (19,22). Two common insertion/deletion variants (dbSNP ID rs72441698 and rs58433947) with GT₇ repeats may explain the large frequency of GT₂₃ and GT₃₀ variants seen across all populations of HIV-infected subjects (n=717, median age ± interquartile range = 46.2 ± 10.2, 85.5% male, 28.6% African American, 55.2% Caucasian) (Suppl. Fig. 1). The GT_n repeats in the African-Americans follow a tri-modal distribution in contrast to the bi-modal distribution seen in Caucasians (Box III in Fig. 1b). The addition of this GT₇ insertion results in greater repeat numbers in African Americans (mean 31.5 ± 0.27, n=205) than in Caucasians (mean 28.1 ± 0.14, n=396) (p<0.001) (Fig. 1b) and in other ethnic groups (Suppl. Fig. 1a). Allele frequencies for these SNPs were also calculated across these ethnic groups (Suppl. Fig. 1b).

Figure 1. Heme oxygenase-1 (HO-1) promoter description and variation across HIV-infected subjects of different ethnicities.

(a) HO-1 promoter region (−1876 to +75) encompassing subcloning region, GT_n repeat, sequencing primers, and transcription start site (TSS) at position +1. The GT_n dinucleotide repeat is shown in the boxed area (extending from 260 base pairs to 200 base pairs upstream of the TSS) and two common SNPs (-413AT/rs2071746 and -1195AG/rs3761439) are shown. Two common insertion deletion variations (dbSNP ID rs72441698 and rs58433947) with GT₇ repeats are shown in bold font within the GT_n repeat. Microsatellite sequencing primers for capillary electrophoresis are denoted by “MS-Primer1” and “MS-Primer2.” Subcloning primers for the promoter-reporter gene expression assay are denoted by “S-primer1” and “S-primer2.” (b) GT_n allele frequencies within HIV-infected patients reported as (mean ± s.e.m.): African Americans (31.5 ± 0.27, n=205) and Caucasians (28.1 ± 0.13, n=396) (difference in mean length between ethnic groups, p<0.0001, Student’s unpaired T-test). Boxes I, II, and III represent the distribution of the additive GT_n repeats with peaks at GT₂₃, GT₃₀, and GT₃₇. (c) The GT_n microsatellite repeats that are most represented in all populations (n=23 and 30) are not in linkage disequilibrium (LD) with -413AT/rs2071746 and -1195AG/rs3761439 in either Caucasians (n=396) or African Americans (n=205). The top two heat maps represent the pair-wise LD results for African American patients with GT₂₃ (left) and GT₃₀ (right), with the values in the legend corresponding to the r² values for each pair-wise comparison between the SNPs -413AT/rs2071746, -1195AG/rs3761439, and the GT_n microsatellite repeat. The bottom two heat maps represent the same for Caucasian patients.

Prior studies of the HO-1 promoter region examined the association between the -413AT/rs2071746 and the -1195AG/rs3761439 SNPs with coronary artery disease outcomes and cardiac function during exercise (19,22). These SNPs were not in significant linkage disequilibrium with the most common GT₂₃ and GT₃₀ repeats in both African American and Caucasian populations (Fig. 1c), with r² values below 0.4. These results suggest limited linkage disequilibrium present at this locus.

Heme oxygenase-1 additive GT_n repeats negatively correlate with gene expression in PBMCs and CD14⁺ monocytes from healthy donors

To extend these studies to primary cells that might be involved in the immune response to HIV, fresh peripheral blood mononuclear cells (PBMCs) (n=20) or enriched CD14⁺ monocytes from healthy donors (n=6) (subsets of donors from those listed in Table 1) were analyzed for HO-1 relative transcript expression after stimulation with the HO-1 chemical inducers cobalt protoporphyrin IX (CoPP), hydrogen peroxide (H₂O₂), and lipopolysaccharide (LPS). Compared to the phosphate buffered saline (PBS) control, stimulation with CoPP led to a 46-fold increase in relative transcript abundance of HO-1 (p<0.0001, paired t-test) whereas H₂O₂ stimulation led to a 1.3-fold increase (p=0.005) and LPS to a 1.3-fold decrease (p=0.007) (Fig. 2a). Since prior studies have reported an important role of HO-1 in the function of CD14⁺ monocytes (15,23), enriched CD14⁺ monocytes were analyzed after in vitro stimulation with CoPP, H₂O₂, or LPS, which led to a 30-fold increase (p<0.0001), a 1.2-fold increase (p=0.02), and a 1.4-fold decrease (p=0.02), respectively (paired t-test, Fig. 2b).

Table 1.

Healthy donors

Characteristics	Healthy donors for HO-1 phenotyping studies (n=22)
Age (median ± IQR)	33 ± 13
Gender (% Male)	45.4%
Ethnicity
Caucasian	77.3%
African	4.5%
Asian	18.2%

Figure 2. Heme oxygenase-1 additive GT_n repeats negatively correlate with relative gene expression in PBMCs and CD14⁺ monocytes from healthy donors.

Cells from healthy donors (n=20) were stimulated with CoPP (25 μM), H₂O₂ (100 uM), or LPS (100 ng/mL), and then harvested 48 hours later for HO-1 transcript analysis. Paired T-test analyses yield statistically significant increases in HO-1 relative transcript abundance normalized to HPRT in (a) PBMCs and (b) CD14⁺ monocytes upon stimulation with CoPP or H₂O₂ and decreases upon stimulation with LPS. (c) PBMCs from healthy donors were stimulated with CoPP (25 uM), H₂O₂ (100 uM) or LPS (100 ng/mL) and genotyped for their HO-1 GT_n repeat polymorphism. Multiple regression and correlation analyses with ethnicity as a covariate showed a significant decline in normalized HO-1 transcript levels within LPS-stimulated cells as the additive GT_n repeats increased (r= −0.38, β= −0.024, p=0.05), and trends in PBMCs stimulated with CoPP (r= −0.30, β= −0.98, p=0.06) and H₂O₂ (r= −0.36, β= −0.076, p=0.08). The solid line represents a fitted linear regression line and dashed lines represent the 95% confidence interval band.

Using an HO-1 promoter-reporter gene assay in HEK293T cells, the additive sum of an individual’s GT_n repeat number (GT_n^{locus 1} + GT_n^{locus 2}) was found to be an informative variable for subsequent functional analyses of this polymorphism (i.e., the greater the repeat number the lower the level of HO-1 expression) (Suppl. Fig. 2). Healthy donors were genotyped for the HO-1 GT_n polymorphism and transcript levels in PBMCs were measured. Multiple regression analysis (controlling for self-identified ethnicity) of normalized HO-1 transcript level and additive GT_n repeats showed statistically significant negative correlations in PBMCs from healthy donors stimulated with LPS (r= −0.38,β= −0.024, p=0.05) (Fig. 2c). The same analyses in PBMCs stimulated with CoPP (r= −0.30, β= −0.98, p=0.06) and H₂O₂ (r= −0.36, β= −0.076, p=0.08) showed trends towards negative correlations (Fig. 2c). Consistent with previous findings, the HO-1 transcript levels in unstimulated PBMCs did not show a correlation with an individual’s additive GT_n repeat number (20) (Suppl. Fig. 3). These results are consistent with our gene expression results in the HEK293T cell lines as well as prior reports showing that greater GT_n repeats lead to decreased gene expression (17–19).

Finally, given prior evidence that the -413AT/rs2071746 may be associated with HO-1 expression in cell lines (19), we tested whether this SNP was associated with HO-1 induction in primary PBMCs exposed to CoPP, LPS, or H₂O₂ (Suppl. Fig. 4a–c, respectively). ANOVA analysis showed no significant association between HO-1 gene expression with this SNP or another common SNP (-1195AG/rs3761439). In summary, these results show that the GT_n microsatellite polymorphism is associated with the regulation of HO-1 gene transcription in primary immune cells.

Heme oxygenase-1 additive GT_n repeats negatively correlate with gene expression in PBMCs and CD14⁺ monocytes from HIV-infected subjects

Prior studies have shown that HIV infection leads to up-regulation of HO-1 mRNA levels in peripheral blood cells (24). To determine whether HIV-infected subjects with a greater number of additive GT_n repeats displayed lower HO-1 expression, we harvested thawed PBMCs and measured relative HO-1 transcript abundance. Multiple regression analysis with ethnicity as a covariate was performed to determine the relationship between the number of additive GT_n repeats and HO-1 relative transcript levels in “viral non-controllers” (prototypic HIV-infected adults with plasma HIV RNA levels above 10 000 copies RNA/mL, n=34) (Table 2a) and “viral controllers” (HIV-infected adults able to fully or partially control HIV replication with plasma HIV RNA level below 1 000 copies RNA/mL in the absence of therapy, n=30) (Table 2a). Among non-controllers, there was a significant correlation between additive GT_n repeats and the HO-1 relative transcript level (Fig. 3a) (r= −0.41, β= −3.08, p=0.02). There was no association between these factors in the controllers (r=0.03, β= −0.36, p=0.4).

Table 2a.

Clinical characteristics of HIV-infected subjects off HAART

Characteristics (median ± IQR)	Viral controllers (n=30)	Viral non-controllers (n=34)
Age	45 ± 9	44 ± 8
Gender (% male)	68.0%	91.2%
Plasma viral load (copies/mL)	75 ± 98	84 142 ± 112 100
CD4 (cells/uL)	745 ± 433	317 ± 264
Ethnicity (%African American)	40%	29.4%
HLA-DR+CD38+% of CD4+ T cells	3.12 ± 1.5	11 ± 11.4
HLA-DR+CD38+% of CD8+ T cells	8.4 ± 10.8	23.5 ± 11.4

Figure 3. Heme oxygenase-1 additive GT_n repeats negatively correlate with gene expression in PBMCs and CD14⁺ monocytes from HIV-infected subjects.

PBMCs from HIV-infected subjects were analyzed for HO-1 transcript analysis. (a) Total mRNA was harvested from thawed PBMCs of HIV viral non-controllers and multiple regression analysis with ethnicity as a covariate yields a significant decline in normalized HO-1 transcript levels as the additive GT_n repeats increase (n=34, r= −0.41, β= −3.08, p=0.02). Total mRNA was harvested from thawed PBMCs of HIV viral controllers and multiple regression analysis with ethnicity as a covariate does not lead to a significant decline in normalized HO-1 transcript levels as the additive GT_n repeats increase (n=30, r=0.03, β= −0.36, p=0.4). (b) Thawed PBMCs from HAART-suppressed subjects were analyzed by multiparameter flow cytometry for HO-1 expression within CD14⁺ monocytes. Multiple regression analysis with ethnicity as a covariate yields a significant decline in normalized HO-1 gMFI as the additive GT_n repeats increase within CD14⁺ monocytes (n=25, r= −0.36, β= −30.45, p=0.04). The solid line represents a fitted linear regression line and dashed lines represent the 95% confidence interval band.

Further phenotyping analyses in PBMCs from HIV-infected subjects on highly active antiretroviral therapy (HAART) showed that HO-1 was expressed to the highest extent in CD14⁺ monocytes. Among individuals with a favorable response to HAART (as defined by having a plasma viral loads with <75 copies RNA/mL) (Table 2b), there was a negative correlation in CD14⁺ monocytes (r= −0.36, β= −30.45, p=0.04) (Fig. 3b). These results suggest that the GT_n microsatellite polymorphism is an important correlate of HO-1 expression within peripheral immune cells during the course of HIV disease, especially in those subjects with progressive disease and high viremia.

Table 2b.

Clinical characteristics of HIV-infected subjects on HAART

Characteristics (median ± IQR)	HAART patients (n=25)
Age	45 ± 7
Gender (% male)	84.0%
Months on HAART	20.1 ± 6.7
Plasma viral load (copies/mL)	75 ± 25
CD4 (cells/uL)	300 ± 151
Ethnicity (% African American)	24.0%
HLA-DR+CD38+% of CD4+ T cells	9.7 ± 7.7
HLA-DR+CD38+% of CD8+ T cells	41.2 ± 26.4

HO-1 additive GT_n repeats are correlated with levels of soluble CD14 and viremia in African American HIV-infected subjects

Given recent studies linking high levels of soluble CD14 (sCD14) with enhanced mortality in chronically HIV-infected subjects on HAART (12) and the functional role that HO-1 plays in CD14⁺ monocyte biology, we measured levels of plasma sCD14 levels in a larger cohort of African American and Caucasian HIV-infected, HAART-suppressed subjects. Clinical and laboratory data describing these subjects is provided in Table 3a.

Table 3a.

Clinical characteristics of HIV-infected HAART suppressed subjects

Characteristics (median ± IQR)	African American (n=50)	Caucasian (n=123)
Age	44.0 ± 8.8	43.8 ± 10.0
Gender (% male)	62.0%	88.6%
Soluble CD14 (median ± IQR)	260 ± 72	249 ± 66
Months on HAART	21.2 ± 27.0	23.0 ± 32.8
Pre-HAART VL (copies/mL)	29 809 ± 64 571	64 649 ± 149 040
Post-HAART VL (copies/mL)	75 ± 28	75 ± 25
Pre-CD4 T cell count (cells/uL)	162 ± 182	192 ± 242
Post-CD4 T cell count (cells/uL)	310.5 ± 398	430 ± 374

Greater HO-1 additive GT_n repeats correlated with higher levels of sCD14 in African American HAART subjects (n=50, r= 0.38, p=0.007) (Table 3a) (Fig. 4a). Furthermore, each unit increase in additive GT_n repeats correlated with a sCD14 level increase of 2.3 ± 0.8 × 10⁴ pg/mL (regression coefficient β= 2.3). Greater additive GT_n repeats also correlated with higher mean plasma HIV RNA levels in African American subjects that were during the chronic phase of infection (n=74, r=0.24, p=0.04) (Fig. 4b). Furthermore, each unit increase in additive GT_n repeats correlated with a mean viral load increase of 3 805.13 ± 1 803.7 HIV RNA copies/mL (regression coefficient β= 3 805.13). Neither the soluble CD14 levels during suppressive HAART (n=123, r= 0.0016, β=0.80, p=1.0) nor the mean viral load level in the absence of therapy (n=177, r= −0.02, β=3 119.26, p=0.2) were statistically different in Caucasian subjects with greater HO-1 additive GT_n repeat numbers.

Figure 4. Increased levels of soluble CD14 during HAART and mean viral load are detected in African American HIV-infected subjects with higher HO-1 additive GT_n repeats.

Two important clinical parameters associated with accelerated HIV disease (high levels of circulating soluble CD14 and a mean viral load) were measured. (a) Plasma soluble CD14 levels were measured by ELISA in African American and Caucasian HAART-suppressed HIV-infected subjects. Greater HO-1 additive GT_n repeats correlated with higher levels of sCD14 in African American (n=50, r= 0.38, β= 2.3, p=0.007) but not Caucasian HIV-infected HAART subjects (n=123, r= 0.0016, β=0.80, p=1.0); this correlation remained after adjusting for gender as a covariate. (b) Mean viral loads were measured in African American and Caucasian subjects and in the chronic stage of infection before they were placed on HAART. Greater HO-1 additive GT_n repeats correlated with higher mean viral loads in African American but not Caucasian HIV-infected HAART subjects (n=74, r=0.24, β= 3 805.13, p=0.04), and stayed significantly associated after adjusting for gender as a covariate. The mean viral load level in the absence of therapy was not statistically associated with the additive GT_n repeat numbers in Caucasian HIV subjects (n=177, r= −0.02, β=3 119.26, p=0.2). Mean viral loads are shown in logarithmic base 10 scale. The solid line represents a fitted linear regression line, and dashed lines represent the 95% confidence interval band.

In summary, these data suggest that the genetically determined level of HO-1 expression in CD14⁺ monocytes may play a role in HIV disease outcome in African-Americans.

Higher levels of HO-1 expression in primary human CD14^hi monocytes correlate with decreased loss of cell surface CD14 after LPS stimulation

CD14 is expressed on the cell surface of monocytes and is the receptor for LPS, a cell wall component of gram-negative bacteria. Upon binding to LPS, a cleaved form of CD14 is shed into the circulation (25,26). Given the finding that greater numbers of HO-1 additive GT_n repeats are positively correlated with higher levels of sCD14 (Fig. 4a), we tested the possibility that HO-1 expression in CD14+ monocytes predicts their ability to retain CD14 cell surface expression upon LPS stimulation. HO-1 protein levels were quantified in PBMCs from 22 healthy donors (Table 1) by multiparameter flow cytometry analysis (Fig. 5a). As previously demonstrated (15,23), HO-1 was most highly expressed at baseline in PBMCs within the CD14⁺ blood monocyte population. Upon exposure to CoPP, HO-1 was up-regulated to a greater extent within monocytes that are CD14^hi than in “non-classical” CD14^dimCD16⁺ monocytes (55 226 vs. 13 448 in geometric mean fluorescence intensity (gMFI); p<1×10⁻⁸). By contrast, CD3⁺ T cells displayed very low HO-1 expression, even after induction with 25 uM CoPP (Fig. 5a).

Figure 5. A higher level of HO-1 expression in primary human CD14^hi monocytes correlates with decreased loss of cell-surface CD14 after LPS stimulation.

(a) Fresh PBMCs from healthy donors were stimulated with either 25 uM CoPP or PBS for 48 hours. Analysis was performed by sequentially gating on live cells, singlets (FSC-A/FSC-H), non-lymphocyte (SSC-A high/FSC-A high), and CD3⁻ populations. Monocyte populations were further gated on CD14 and CD16. HO-1 was induced to a greater extent within CD14^hi monocytes upon stimulation with 25uM CoPP as compared to CD14^dimCD16^pos monocytes (55 226 3 ± 916.9 vs. 13 448 ± 2 458.4 in gMFI; unpaired Student’s T-test p<1×10⁻⁸). Histograms are shown depicting the intensity of HO-1 staining in CD3+ T cells, CD14^dimCD16⁺ monocytes, or CD14^hi monocytes (25 uM CoPP or PBS) as well as with the secondary antibody (goat anti-rabbit-FITC). (b) PBMCs were stimulated with either LPS (100 ng/mL) or PBS for 48 hours. HO-1 gMFI within CD14^hi monocytes is reduced upon LPS stimulation (5 068 ± 256.5 vs. 4 481 ± 196.2 gMFI, p=0.04) paired Student’s T-test). (c) Multiple linear regression with ethnicity, age, and gender as covariates showed that normalized delta HO-1 gMFI in CD14^hi monocytes correlates with the normalized delta CD14 gMFI upon LPS stimulation (n=22, r=0.54, β=0.60, p=0.05). The solid line represents a fitted linear regression line and dashed lines represent the 95% confidence interval band. Boxed data points refer to individuals with GT_>60 repeats.

We measured HO-1 expression after LPS stimulation and saw a significant decrease in expression of HO-1 gMFI within CD14^hi monocytes (paired t-test, p=0.04) (Fig. 5b). Since HO-1 is generally considered to be anti-inflammatory, we hypothesized that induction of HO-1 in CD14^hi monocytes after LPS stimulation might positively correlate with the ability of these cells to retain cell surface CD14. To test this hypothesis, the relationship between the change in staining intensities for HO-1 and for CD14 (normalized ΔCD14 gMFI and ΔHO1 gMFI) were determined in CD14^hi blood monocytes of healthy donors upon in vitro LPS stimulation (Fig. 5c). Multiple regression analysis (covariates included ethnicity, age, and gender) showed an association between the expression of HO-1 and retention of cell surface CD14 in LPS-stimulated CD14^hi monocytes. (n=22, r=0.54, β=0.60, p=0.05). Notably, CD14^hi monocytes from individuals with very high GT repeat numbers (>60) generally had less induction of HO-1 and greater loss of CD14 when stimulated with LPS (boxed data points, Fig. 5c).

These results suggest that the ability to induce HO-1 in CD14^hi monocytes upon exposure to LPS is protective against loss of CD14 from the cell-surface, which itself is a favorable predictor of outcome in HIV disease.

Discussion

Since high levels of immune activation and inflammation predict a more rapid pace of HIV disease progression (7,9,12), host immunoregulatory factors that blunt immune activation and inflammation may contribute to delayed disease progression. Amongst such factors, HO-1 is an important anti-inflammatory enzyme that has been implicated in multiple disorders in the past. In these studies, we have performed an assessment of HO-1 promoter polymorphism genetics, HO-1 expression, and two biomarkers associated with HIV disease outcome in a large cohort of well-characterized HIV-infected and -uninfected adults. Our data show that HO-1 GT_n microsatellite genotypes predict HO-1 expression in primary cells, and that there is a consistent association between HO-1 promoter genotype and HO-1 expression. We also show that greater additive GT_n repeats correlate with higher viral loads and higher levels of the inflammatory sCD14 biomarker in HIV-infected African Americans. Finally, we show that higher levels of HO-1 induction in monocytes correlate with retention of CD14 on the cell surface after LPS stimulation. In light of these data, we speculate that induction of the anti-inflammatory enzyme HO-1 may play a role in limiting the injurious effects of immune activation in chronic HIV disease, e.g., that caused upon microbial translocation and circulation of LPS (10,12). This is in line with evidence that HO-1 induces tolerogenic properties in monocyte-derived antigen presenting cells (15,23). By influencing monocyte maturation and activation, HO-1 may accordingly represent an anti-inflammatory control mechanism for monocyte-mediated inflammation induced by LPS.

It has long been recognized that individuals with HIV have a higher proportion of circulating CD14^dimCD16⁺ inflammatory monocytes than do uninfected controls (27–29). Reciprocally, we have observed that initiation of suppressive HAART therapy in chronically HIV-infected subjects is associated with a decrease in the percentage of circulating CD14^dimCD16⁺ inflammatory monocytes and an increase in the percentage of CD14^hi ‘conventional’ monocytes (Suppl. Fig. 6). Shedding of CD14 from the surface of non-inflammatory CD14^hi monocytes upon exposure to an activating stimulus (e.g., LPS) may represent an important part of their transition into pro-inflammatory CD14^dimCD16⁺ monocytes, and further studies exploring this possibility are underway.

We also demonstrate that higher copy numbers within the HO-1 GT_n repeat promoter polymorphism correlate with higher mean viral load during chronic HIV infection in African American subjects, suggesting that HO-1 activity can modulate viral load. For this analysis, we focused on individuals with high mean viral loads due to the fact that HO-1 expression is not induced to as high an extent in individuals that are able to control viral loads (Figs. 3a and 3b). HO-1 induction may limit bystander immune activation during the acute stage of HIV infection, and thereby reduce the harmful T cell target to T cell effector ratio that leads to heightened viral transmission leading into the chronic stage of infection (6).

Our data show that having fewer GT_n repeats within the microsatellite polymorphism of the HO-1 promoter region is associated with higher levels of HO-1 gene expression in primary immune cells. The alternating purine-pyrimidine repeats within the GT_n microsatellite form a left-handed helical Z-DNA conformation and may affect gene transcription through the inhibition of RNA polymerase-mediated transcription (30). The HO-1 GT_n microsatellite locus is unique in being able to influence gene transcription under multiple cellular stress conditions. Our study shows that additive GT_n repeats influence HO-1 expression upon CoPP, H₂O₂, and LPS stimulation, confirming and extending the findings of a recent study analyzing this polymorphism in primary human newborn endothelial cells (21). The central mediator implicated in this pathway is the Nrf2 transcription factor during HO-1 induction with CoPP and H₂O₂ (31,32), and HO-1 down-regulation with LPS stimulation (33).

In this study, we report an association between greater additive GT_n repeats and plasma sCD14 in HAART-suppressed subjects and mean viral load in chronically HIV-infected, untreated African American subjects. Interestingly, this association was not observed in Caucasians. Though individuals of African ancestry have a greater average number of HO-1 GT_n repeats within this microsatellite polymorphism than do other ethnic groups (Suppl. Fig. 1), it does appear that the positive correlation between sCD14 remains consistent even in the range of 40–70 repeats, which overlaps with the number of repeats seen in Caucasians. We cannot rule out that there may be epistatic factors, whether environmental or genetic, that differentially influence HO-1 gene function in African Americans vs. Caucasians. For example, ethnicity-specific differences in the interaction of Nrf2 or Bach1 with the promoter region containing the GT_n satellite polymorphism may also influence HO-1 gene expression. Also, we cannot rule out the possibility that interactions with Toll-like receptors, such as TLR4, may influence the extent of CD14 expression on monocytes upon LPS stimulation (Figure 5c). To our knowledge, this type of regulation has not been explored in the past (25, 26), but future studies along these lines may reveal new insights about monocyte biology. To better understand the basis of the associations observed in African American subjects, it will be necessary to carry out larger cross-population comparisons of the HO-1 genetic locus as well as the up-stream mediators influencing its expression.

The prevalence of higher GT_n repeat numbers in individuals with African ancestry may reflect an interplay between HO-1 activity and the various hemolytic conditions that are prevalent in this population (e.g., about 75% of the global incidence of hemoglobinopathy is in sub-Saharan Africa) (34). Indeed, there is evidence showing positive selection for the genomic region specific to African ancestral populations, with this group showing a positive Cross Population Expression Haplotype Homozygosity (XP-EHH) coefficient of 1.8 as opposed to other populations showing negative XP-EHH coefficients (Human Genome Diversity Project: http://hgdp.uchicago.edu/) (Suppl. Fig. 5) (35).

The results of this study suggest that pharmacological augmentation of HO-1 may represent a strategy for the treatment of diseases such as HIV, in which unchecked immune activation results in deleterious clinical outcomes. Given the increased burden of HIV disease in African and African American subjects, such a strategy might prove to be particularly useful in these populations. Our study demonstrates that there are population-specific genetic variations of a GT_n dinucleotide polymorphism involved in the control of HO-1 gene expression in immune cells that have an important role in HIV disease outcome. A recent clinical study reported that the synthetic metalloporphyrin Hematin (Fe²⁺ protoporphyrin IX) (Panhematin ®, Lundbeck Inc.) can induce a 15-fold increase in HO-1 activity (36). Further investigation of the HO-1 GT_n microsatellite genetic locus may help us better understand the clinical pharmacogenetics of this type of intervention.

In summary, our data show that the HO-1 GT_n microsatellite promoter polymorphism predicts expression of HO-1 in primary immune cells that play an important role in host interactions with HIV. We show that that greater additive GT_n repeats correlate with higher levels of the inflammatory sCD14 biomarker in African Americans on HAART and of the mean viral load in subjects off therapy. Finally, we show that the genetically-determined level of HO-1 expression in monocytes correlates with the extent of LPS-induced loss of cell-surface CD14 expression within healthy donors. This study highlights the potential role of the HO-1 GT_n genetic locus as a predictive biomarker for disease outcome in HIV-infected subjects on and off antiretroviral therapy.

Materials and Methods

Human subjects

Peripheral blood samples were collected from HIV-infected and -uninfected adults after written informed consent was obtained under protocols approved by the University of California at San Francisco Committee on Human Research (San Francisco, CA). Healthy adults (n=22) were recruited from the San Francisco Bay Area (Table 1). HIV-positive subjects were recruited from the San Francisco Bay Area into the UCSF-based Study of the Consequences of the Protease Inhibitor Era (SCOPE) Cohort, a clinic-based cohort of >1500 adults with chronic HIV infection (Table 2 and Table 3). Participants undergo clinical laboratory monitoring and have their biological specimens banked. HIV controllers are defined as HIV-infected subjects that are off antiretroviral therapy for at least the latest 12 month period for which follow-up is available, with at least three plasma HIV RNA levels taken during this time below 1 000 copies/mL. HIV viral non-controllers are defined as individuals with HIV RNA levels over 10 000 copies RNA/ml (by bDNA, Chiron Diagnostics, Emeryville, USA) or 20 000 copies RNA/mL (by Abbott RealTime HIV-1 PCR, Abbott Park, Illinois, USA) at some point in the past, regardless if during a period of treatment or no treatment.

Measurement of plasma CD14 in HAART-suppressed subjects and mean viral loads in HIV-infected patients off therapy

For sCD14 measurement in HAART-suppressed subjects (Table 3a), a commercially available enzyme-linked immunosorbent assay was used according to the manufacturer’s protocol (R&D Systems, Minneapolis, MN). The mean viral loads measured in HIV-infected patients (Table 3b) were defined as the mean of all viral load determinations known to have been measured in the absence of therapy. All experiments were performed in a blinded manner without prior knowledge of an individual’s genotype.

Table 3b.

Clinical characteristics of HIV-infected subjects off therapy

Characteristics (median ± IQR)	African American (n=74)	Caucasian (n=177)
Age	32.2 ± 9.0	46.8 ± 11.3
Gender (% male)	60.0%	89.8%
Mean VL (copies/mL)	63 935 ± 511 768	87 485 ± 161 380
CD4 T cell count (cells/uL)	330 ± 401.5	448 ± 344

Primary cell isolation and culture conditions

PBMCs were isolated from whole blood drawn into sodium heparin tubes by density centrifugation using Histopaque^®-1077 (Sigma Aldrich, Saint Louis, MO), and the MACS® Monocyte Isolation Kit II (Miltenyi, Auburn, CA) was used to enrich CD14⁺ monocytes from PBMCs (mean 95% purity, as assessed by multiparameter flow cytometry). PBMCs or CD14⁺ monocytes were cultured in RPMI 1640 (Life Technologies, Rockville, MD) supplemented with 10% FBS (Gemini Bio-Products, Woodland, CA), 1% penicillin/streptomycin (Mediatech, Washington, DC), and 2 mM L-glutamine (Mediatech) (hereafter referred to as R10 medium) on Upcell^™ 96F MicroWell plates (Thermo Scientific, Rochester, NY). Cobalt protoporphyrin IX (CoPP) was purchased in powdered form (Frontier Scientific, Park City, Utah), dissolved in 0.1 mM NaOH, and titrated to a pH of 7.6. Lipopolysaccharides (LPS) from Escherichia coli 055:B5 (Sigma Aldrich) were reconstituted in water to a final stock concentration of 1 mg/mL. 3% (w/w) hydrogen peroxide (H₂O₂) (Sigma Aldrich) was diluted in water for a final stock concentration of 2 mM. PBMCs or CD14⁺ monocytes were incubated at a final concentration of 25 uM of CoPP, 100 ng/mL LPS, or 100 uM H₂O₂ in RPMI-10 at 37° for 48 hours. Adherent cells were detached from the Upcell^™ plates by incubating the plates at 25°C for 20 minutes.

Cell preparation and antibody labeling

All cells were stained with a live/dead marker Amine-Aqua/Am-Cyan Live/Dead (Invitrogen) to exclude dead cells from analysis. The following fluorophore-conjugated monoclonal antibodies (mAbs) were used to detect cell surface markers: CD3-APC-Cy7 (SP34-2, BD Biosciences, Franklin Lakes, NJ), CD14-Qdot605 (Q10013, Invitrogen), and CD16-APC (3G8, Caltag-Invitrogen). The following antibodies were used for detection of intracellular antigens: HO-1 rabbit polyclonal (ab13243, unconjugated, Abcam, Cambridge, MA) and a secondary goat anti-rabbit IgG conjugate (554020, FITC, BD Biosciences). For flow cytometry analysis, cells were washed in staining buffer [PBS with 2% FBS and 2 mM ethylenediamine tetra-acetic acid – EDTA (Sigma)] and then incubated for 30 minutes at 4°C in the presence of directly-conjugated fluorescent mAbs. All cells were stained with a live/dead marker (Amine-Aqua/Am-Cyan Live/Dead; Invitrogen) so that dead cells could be excluded from analysis. The cells were washed in staining buffer, and then fixed and permeabilized in BD Cytofix/Cytoperm (BD Biosciences) according to the manufacturer’s protocol for intracellular staining. The cells were incubated for 1 hour at 4°C in the presence of a rabbit polyclonal antibody specific for HO-1 (ab13243, unconjugated, Abcam) and a secondary goat anti-rabbit IgG conjugate (554020, FITC, BD Biosciences). The cells were washed in staining buffer and then fixed in 1% paraformaldehyde (PFA). Data were acquired with an LSR-II flow cytometer (BD Biosciences) and analyzed with FlowJo software (Treestar, Ashland, OR).

DNA isolation and genotyping

DNA was isolated from the cell pellets of both healthy donors and HIV-infected subjects using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA), according to the manufacturer’s instructions. HO-1 microsatellite genotyping was performed by the UCSF Genomics Core Facility. Primers were designed with the Primer3 algorithm http://fokker.wi.mit.edu/primer3/input.htm. The primer set was “MS-Primer1” 5′-FAM-CCAGCTTTCTGGAACCTTCTG and “MS-Primer2” 5′-GAAACAAAGTCTGGCCATAGGA. The samples were amplified on a touchdown PCR protocol. The resulting products were run on the 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA) and analyzed with GeneMapper (Applied Biosystems). The TaqMan® SNP Genotyping Assay was used to discriminate the allelic composition for the HO-1 SNPs at positions -1195AG/rs3761439 and -413AT/rs2071746. Unlabeled forward and reverse flanking PCR primers (900 nM final concentration) and two allele-specific probes labeled with either VIC or FAM reporter dye (200 mM final concentration) were added to 20 ng DNA in a 20-ml reaction containing TaqMan Universal PCR Mix. The sequence for the rs2071746 SNP probe was 5′-AGTTCCTGATGTTGCCCACCAGGCT[A/T]TTGCTCTGAGCAGCGCTGCCTCCCA (Assay ID: C__15869717_10, Applied Biosystems, Carlsbad, CA) and for the rs3761439 SNP probe was 5′-CATAGGGAGACCC[T/C]GTCT (Custom assay, Applied Biosystems). Samples were run on the StepOnePlus^™ Real-Time PCR system, and the results were analyzed using Step One v2.0 (Applied Biosystems)

Quantitative PCR

For quantitative PCR analysis, cells were harvested and RNA was isolated using TRIzol ® Reagent (Invitrogen, Carlsbad, CA). Total cellular RNA (0.2 μg) was used for cDNA synthesis with Oligo-dT primers and reverse transcriptase from Omniscript (Qiagen). Relative expression levels of HO-1 mRNA were measured by quantitative RT-PCR using validated Taqman® Gene Expression assay mixes for HO-1 (Assay ID: Hs00157965_m1, Applied Biosystems) and the reference gene, human hypoxanthine-guanine phosphoribosyl transferase (HPRT) (Assay ID: Hs99999909_m1, Applied Biosystems), according to the manufacturer’s protocol. The StepOnePlus^™ Real-Time PCR system (Applied Biosystems) was used for amplification and detection, and the efficiency corrected calculation (2^−ΔCT) of the threshold cycle C(t) was used to measure HO-1 gene expression relative to the HPRT gene.

Transient transfection assay for Dual-glo gene expression assay

Transient co-transfection with the HO-1-luciferase fusion plasmid and the pRL-ef1alpha plasmid (Promega, Madison, WI, USA) was performed using the calcium phosphate method. For transfection, HEK293T cells were split 24 h before transfection and seeded in Dulbecco’s Modified Eagle Medium (Gibco/Invitrogen, Carlsbad, CA) supplemented with 10% FBS (Gemini Bio-Products, Woodland, CA), 1% penicillin/streptomycin (Mediatech, Washington, DC), and 2 mM L-glutamine (Mediatech). The 10x co-transfection mix was made up with 1 ug of HMOX1-pGL4.20 and 0.1 ug of pRL-ef1alpha plasmid DNA in 200 uM CaCl₂ in Hepes buffered saline and added dropwise to the wells. Cells were harvested at 48 hours, and the transfection efficiency and cell viability was analyzed by flow cytometric analysis of %GFP+ cells (mean 80% efficiency) Luciferase activity was analyzed using the Dual-Glo ® luciferase reporter assay system (Promega, Madison, WI, USA). Renilla luciferase activity in the lysates was used to normalize the activity of HMOX1 promoter driven firefly luciferase activity on a Spectramax M2 luminometer. (Molecular Devices, Sunnyvale, CA)

Statistical and genetic analyses

Linkage disequilibrium analysis of the additive GT_n repeats and the SNPs -1195A/G (rs3761439) and -413A/T (rs2071746) was carried out using the “pwld” function (http://www.gene.cimr.cam.ac.uk/clayton). HO-1 GT_n repeat differences between African Americans and Caucasians (Fig. 1b), HO-1 relative qPCR levels before and after each stimulation condition using paired Student’s T-test analysis (Fig. 2ab), and HO-1 gMFI levels in CD14^hi monocytes before and after each stimulation condition using paired Student’s T-test analysis (Fig. 5b) were all performed using GraphPad Prism v5.0d (Graphpad Software, La Jolla, CA, USA). We performed multiple linear regression of healthy donor HO-1 additive GT_n repeats and the HO-1 relative qPCR levels upon different stimulation conditions (Fig. 2c) of HIV-infected subject additive GT_n repeats and the HO-1 relative qPCR levels (Figs. 3a and 3b), for additive GT_n repeats compared to mean viral load and post-HAART soluble CD14, and for comparing the normalized delta CD14 gMFI to normalized delta HO-1 gMFI in CD14^hi monocytes (Fig. 5c). For all correlation analyses, the Pearson rcorrelation and its p values are reported using GraphPad Prism v5.0d (Graphpad Software, La Jolla, CA, USA). For all multiple regression analyses, the regression coefficient β was calculated and adjusted for ethnicity, age, and gender when noted. All regression analyses were undertaken using STATA v11.2 (StataCorp LP, College Station, TX).