Neither Microbial Translocation Nor TLR Responsiveness Are Likely Explanations for Preexisting Immune Activation in Women Who Subsequently Acquired HIV in CAPRISA 004

Abstract

Innate immune activation was a strong predictor of HIV acquisition in women at risk for HIV in CAPRISA 004. Identifying the cause(s) of activation could enable targeted prevention interventions. In this study, plasma concentrations of lipopolysaccharide, soluble CD14, and intestinal fatty acid–binding protein did not differ between subjects who did or did not subsequently acquire HIV nor were these levels correlated with plasma cytokines or natural killer cell activation. There was no difference between HIV acquirers and non-acquirers in the chemokine and cytokine responses of peripheral blood mononuclear cells stimulated with TLR2, 4, or 7/8 agonists. Further studies are required.

INTRODUCTION

Nonspecific activation of the immune system is a pathological feature of HIV infection.¹ The degree of activation of CD8⁺ T cells is associated with the rate of HIV disease progression.² But our recent studies³ validate previous studies^4–6 in suggesting that innate immune activation is also a pathological determinant of HIV acquisition. In CAPRISA 004, a randomized placebo-controlled trial of 1% Tenofovir gel, women who acquired HIV had significantly higher plasma levels of interleukin (IL)-2, IL-7, IL12p70, and tumor necrosis factor alpha (TNF-α), and more frequent activation and degranulation of natural killer (NK) cells in the blood than the women who remained uninfected.³ Even after adjusting for potential confounders, innate immune activation was associated with HIV acquisition. The identification of determinants of activation would allow for the design and testing of targeted interventions to reduce activation as a method to reduce HIV acquisition.

During established HIV infection, increased microbial translocation¹ and chronic cytomegalovirus infection⁷ are associated with greater immune activation. There has been some speculation that a higher pathogen burden in settings with poorer sanitation may result in higher levels of microbial translocation, even among HIV-uninfected individuals.⁸ Others have suggested that differences in toll-like receptor (TLR) responsiveness exist between women who are exposed to HIV but remain uninfected, and HIV-uninfected controls.⁹ We previously reported that the prevalence of cytomegalovirus infection in this cohort is almost universal³ so this is not likely to drive the differences. Although HSV-2 was associated with HIV acquisition and suppressed NK cell antiviral response,¹⁰ it did not weaken the association between innate immune activation and HIV acquisition.³ Here, we report on further studies that aimed to determine whether differences in microbial translocation or TLR responsiveness could explain differences in innate immune activation in women who were exposed to HIV in CAPRISA 004.

METHODS

Study Population and Sampling

For this study, specimens from women who acquired HIV and women who remained uninfected were used. Peripheral blood mononuclear cells (PBMCs) and plasma were collected from participants of CAPRISA 004, a randomized controlled trial of 1% Tenofovir gel conducted in eThekweni and Vulindlela (KwaZulu-Natal, South Africa), as previously described.¹¹ The participants in this study matched those used in previous studies of immune activation in this trial.³ Briefly, samples from HIV acquirers were from the last trial visit before HIV acquisition. Samples from HIV non-acquirers were from the trial visit at which sexual activity in the preceding month was each participant's personal maximum. HIV non-acquirers were eligible for selection if they reported an average of more than 2 sex acts per week throughout the trial, returned a corresponding number of used gel applicators (as a surrogate measure of actual sexual activity), had PBMC samples available, and had given informed consent for future research. HIV-uninfected status was verified by polymerase chain reactions, Western blot, and rapid antibody assays. Due to limited cell numbers, a different number of samples were used for each TLR agonist studied (16 acquirers and 13 non-acquirers for aldrithiol-inactivated HIV [AT-2]; 14 and 9, respectively, for heat-killed Listeria monocytogenes [HKLM]; and 10 and 6 for lipopolysaccharide [LPS]). The baseline demographic characteristics, sexual history, and study procedures have been previously described.³ Receipt of Tenofovir gel did not affect any of the measures presented^3,12 so is not further described in this report.

This study was approved by the University of KwaZulu-Natal Biomedical Research Ethics Committee (BE073/010). Participants gave informed consent for their samples to be used for these studies.

Measurement of LPS, Soluble CD14, and Intestinal Fatty Acid–Binding Protein

LPS, soluble CD14 (sCD14; a marker of monocyte response to LPS), and intestinal fatty acid–binding protein (I-FABP; a marker of enterocyte damage) were measured in duplicate in cryopreserved plasma from participants. Plasma had not been thawed more than once previously. LPS levels were measured using the Limulus Amebocyte Assay (Lonza, Walkersville, MD), as previously described.¹³ A commercially available enzyme-linked immunosorbent assay was used for sCD14 (R&D Systems, Minneapolis, MN) on plasma diluted to 0.5% according to the manufacturer's instructions. I-FABP was measured using an adaptation of FABP2 kit on plasma diluted to 10% in 50% fetal calf serum in phosphate buffered saline (R&D Systems).

NK Cell Phenotypic Characterization

NK cell activation was measured in cryopreserved PBMCs in batch analyses using optimized procedures and flow cytometry methods as previously described.³

Characterization of Cytokine and Chemokine Response of PBMC to TLR Agonists

Cryopreserved PBMCs were thawed in warm R10 culture media: RPMI 1640 (Gibco; Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum (Invitrogen), 2 mmol/L of L-glutamine and 1% penicillin/streptomycin. After 2 hours of incubation at 37°C, cells were stained with Viacount (Millipore, Billerica, MA) and counted with a Guava PCA (Millipore). PBMC were resuspended at 1 × 10⁶/mL in R10 culture media. TLR agonists or R10 media alone were added to 1 × 10⁶ PBMC in 4 mL FACS Tubes (Becton Dickinson, Franklin Lakes, NJ) at the following concentrations 0.7 μg/mL AT-2 HIV, 2 μL/mL HKLM, or 0.1 μg/mL LPS. Cells were cultured overnight at 37°C in a sterile incubator maintained at 5% CO₂. Supernatant was harvested by centrifugation and immediately cryopreserved at −80°C.

The concentrations of 13 cytokines: GMCSF, interferon (IFN)-γ, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12 (p70), IL-13, and TNF-α were assessed using a high-sensitivity human cytokine premixed 13-plex kit (Millipore) as per the manufacturer's instructions. Samples were assayed in duplicate after a single thaw, without dilution. Samples from HIV acquirers and non-acquirers were equally spread across assay plates to minimize plate-to-plate variability as a confounder. Cytokine measures beneath the detection limit of the assay were given a value of the midpoint between 0 and the lower detection limit of the assay and were included in the analysis.

Statistical Analysis

Assays were conducted blinded to whether the sample was from an HIV acquirer or non-acquirer. The study had 80% power to detect a difference of approximately 4 pg/mL in LPS, 0.17 × 10⁶ pg/mL in sCD14, or 208 pg/mL in I-FABP levels between the 2 groups.

Comparisons were made using a nonparametric Mann–Whitney test in GraphPad Prism v5 (GraphPad Software). To account for multiple comparisons, the unadjusted P value is reported and the relevant Bonferroni adjusted P value is given in the text or figure caption.

RESULTS

HIV-exposed women who acquired HIV while enrolled in CAPRISA 004 were previously reported to have higher levels of circulating IL-2, IL-7, IL12p70, and TNF-α in the blood, and higher levels of activated NK cells relative to women who did not acquire HIV.³ To explore possible explanations for the differences in activation, samples from the same 44 women who acquired HIV and 37 who remained uninfected and were previously studied, were evaluated for evidence of microbial translocation or differential response to microbial products. The baseline characteristics of this study population have been previously described.³

No Appreciable Difference in Microbial Translocation between HIV Acquirers and Non-acquirers

In primary or chronic HIV infection, plasma levels of LPS are a marker of gut permeability and microbial translocation.¹ Both LPS and its ligand sCD14 are associated with immune activation in HIV infection. We measured LPS, sCD14, and I-FABP—a marker of enterocyte damage.¹⁴ There was no significant difference in plasma levels of LPS, sCD14, or I-FABP in HIV acquirers and non-acquirers, although surprisingly, some of the clinically well participants had high levels of LPS or I-FABP consistent with significant translocation or intestinal epithelial cell necrosis (Fig. 1).

FIGURE 1: Plasma levels of LPS (A), sCD14 (B), and I-FABP (C) do not differ between women who acquired HIV samples preinfection (gray squares) and women who remained HIV uninfected (black triangles). Lines demarcate the median and interquartile range. P value from Mann–Whitney test.

Levels of LPS, sCD14, and I-FABP were not associated with NK cell activation, regardless of which marker of NK cell activation was used (HLA-DR, CD38, or CD69), as shown in Figure S1 (see Supplemental Digital Content, https://links.lww.com/QAI/A414). Similarly, levels of LPS, sCD14, or I-FABP were not significantly associated with any of the plasma cytokines measured in previous studies after correcting for multiple comparisons (the Bonferroni corrected alpha for significance was P < 0.0013).

TLR Responsiveness

Some previous studies have suggested that differences in the ability to respond to microbial products may lead to differential levels of TLR-mediated activation.⁹ To determine whether there were any differences in TLR responsiveness between HIV acquirers (n = 21) and non-acquirers (n = 18), cytokine and chemokine release was quantified after overnight in vitro culture of PBMC with agonists for TLR2 (HKLM), TLR4 (LPS), or TLR7/8 (AT-2 HIV).

Neither the absolute increase in concentration nor the fold change increase in concentration relative to media alone of any of the measured cytokines differed between PBMC from HIV acquirers and non-acquirers stimulated with a TLR2, TLR4, or TLR7/8 agonist (Fig. 2). IFN-γ concentrations were higher in the supernatant of PBMC from HIV acquirers cultured with AT-2 HIV (P = 0.05), but after adjusting for the 39 comparisons made, the difference was not significant.

FIGURE 2: PBMC from HIV acquirers (black) and non-acquirers (gray) exhibit a similar fold change (log₁₀) in the secretion of chemokines and cytokines after stimulation with a TLR7/8 agonist (AT-2 HIV), TLR2 agonist HKLM, or a TLR4 agonist LPS relative to media alone. PBMC were stimulated overnight and secreted cytokines/chemokines measured by Luminex in duplicate. Data are shown as box and whiskers to 5th and 95th percentile. Adjusted P values are not shown as none achieved significance, although the nominal P value for the comparison of IFN secretion after AT-2 stimulation in HIV acquirers and non-acquirers was P = 0.05 (Bonferroni adjusted P < 0.0013 for significance).

Chronic stimulation by LPS is thought to reduce further responsiveness to stimulation of TLR4 by LPS (“LPS tolerance”).¹⁵ In this study, there was no evidence that circulating LPS levels were associated with in vitro cytokine or chemokine responses of PBMC stimulated with LPS (see Table S1, Supplemental Digital Content, https://links.lww.com/QAI/A414).

DISCUSSION

Immune activation was a potent predictor of HIV acquisition in CAPRISA 004.³ In chronic HIV infection, immune activation is associated with circulating levels of microbial products,¹ which predict disease course.^1,13 To our knowledge, there have not been any previous studies of microbial translocation in women who are exposed to HIV. Here, we found that microbial translocation was similar among HIV acquirers and non-acquirers. Microbial translocation was, furthermore, not associated with NK cell activation or any of the cytokines measured. We speculate that the levels of microbial translocation measured in HIV acquirers and non-acquirers may represent homeostatic levels and an alternative more potent mediator of activation may remain undiscovered among HIV acquirers. We tested an alternative hypothesis that TLR responsiveness may differ among HIV acquirers and non-acquirers. In this study, there was no significant difference in PBMC response to TLR agonists, whereas a previous report found greater TLR responsiveness in HIV-exposed uninfected women compared with healthy blood donors.⁹ The difference may be due to the comparison group in our study being women who actually acquired infection.

We observed that PBMC from HIV acquirers produced more of the antiviral cytokine, IFN-γ, after stimulation with AT-2 HIV than HIV non-acquirers. Although this finding is likely due to chance, as it did not reach statistical significance after controlling for the multiple tests performed, the finding points to the possibility that women who later acquired HIV may have had sustained exposure to HIV and consequently developed an HIV-specific immune response in the absence of infection. As reported elsewhere,¹⁶ we did not detect sustained or robust HIV-specific T-cell responses by Enzyme-Linked Immunosorbent SPOT in these participants. Therefore HIV-specific responses, if present in these women, are weak. Assessments of HIV-specific immune responses in future studies of HIV acquisition continue to be warranted.

Our findings are limited by the sample size; we are unable to rule out a difference smaller than what the study was designed to detect. Moreover, we studied bulk PBMC and plasma. Studies in established HIV infection have demonstrated that TLR responsiveness is cell type, disease stage, and agonist specific.¹⁷ Moreover, measurement of these parameters in the periphery may be inadequate to reveal what actually happens in the female genital tract, the site of infection, and therefore a study of the mucosa would perhaps be appropriate in future work.¹⁸

These data collectively suggest that neither microbial translocation nor TLR responsiveness of PBMC are sufficient to explain the higher levels of innate immune activation observed in women who subsequently acquired HIV in CAPRISA 004. Further studies to explore host genetic factors or differences in the enteric or genital tract microbiome may be required to identify the upstream causes of preexisting immune activation in women who subsequently acquired HIV in CAPRISA 004.

ACKNOWLEDGMENTS

The authors thank the participants; women whose dedication and commitment to improving their and their peers' health and kindly donating samples during the conduct of the trial make this research possible. The authors acknowledge the work of the study staff of CAPRISA 004 and the TRAPS team: Sengeziwe Sibeko, Leila Mansoor, Lise Werner, Shabashini Sidhoo, and Natasha Arulappan. The authors gratefully acknowledge Rachel Simmons for advice on cytokine assays, Marianne Mureithi for advice on designing TLR responsiveness assays, and Raveshni Durgiah for technical assistance. Tenofovir was provided by Gilead Sciences and the gel was manufactured and supplied for the CAPRISA 004 trial by CONRAD.