Viral Suppression and Immune Restoration in the Gastrointestinal Mucosa of Human Immunodeficiency Virus Type 1-Infected Patients Initiating Therapy during Primary or Chronic Infection

Human subjects and sample collection.

HIV-infected patients (n = 10) participating in the study had initiated HAART during either the primary (group PHI, n = 3) or the chronic (group CHI, n = 7) stage of HIV infection. Longitudinal jejunal biopsy samples and peripheral blood samples were obtained prior to and following the initiation of HAART over a period of 36 months and assessed for the restoration of T-cell subsets and their function, viral loads, HIV-specific CD8⁺ T-cell responses, and gene expression profiles (Table 1). Four HIV seronegative healthy individuals were enrolled in the study to provide baseline values as negative controls. Jejunal biopsy samples were collected by upper endoscopy under sedation as previously described (13, 34). Studies were performed under informed consent and an Institutional Review Board (IRB #200311088-7)-approved protocol.

Measurement of HIV loads.

HIV copy numbers in plasma samples were measured by branched DNA assay (Bayer Diagnostics, Emeryville, California) and in jejunal biopsy samples by real-time PCR as previously described (34).

Isolation of gut lymphocytes.

Mononuclear cells were obtained from longitudinal jejunal biopsy samples according to previously published protocols (13, 24, 41). Briefly, jejunal biopsy samples were placed in lymphocyte isolation media containing RPMI 1640 (Invitrogen/Gibco BRL, Grand Island, New York), 5% fetal calf serum (Gibco BRL), 100 U/ml of penicillin (Gibco BRL), 100 U/ml of streptomycin (Gibco BRL), and 100 U/100 ml of collagenase type II (Sigma Chemical Co., St. Louis, Missouri) and subjected to shaking, at 37°C with 5% CO₂, for three periods of 30 min. Cells were then washed with RPMI media and left to rest overnight at 37°C with 5% CO₂. Cell viability (>95%) was determined by trypan blue exclusion assay.

Immunophenotypic analysis.

Surface analysis of immunophenotypic markers on lymphocytes from peripheral blood and GALT was performed as previously described (13, 24, 41). Briefly, peripheral blood mononuclear cells and isolated lymphocytes from jejunal biopsy samples were incubated with fluorescent mouse anti-human monoclonal antibodies for 30 min at 4°C. Cells were washed, fixed with 1% paraformaldehyde, and analyzed by four-color flow cytometry (FACScalibur; Becton Dickinson). A minimum of 100,000 events were collected for each sample, and data were analyzed using Flow Jo (Tree Star, Inc., San Carlos, California).

Monoclonal antibodies.

Antibodies used for the analysis of T cells in this study included CD3, CD4, CD28 pure, CD103, CD69, and gamma interferon (IFN-γ) (Becton Dickinson, Mountain View, California); CD8 and CD11α (Caltag Laboratories, South San Francisco, California); and CCR5, CXCR4, and Ki67 (Coulter Immunotech, Miami, Florida).

HIV antigen-specific CD8⁺ T-cell responses.

Isolated T lymphocytes from GALT and peripheral blood were placed in 96-well V-bottom plates (Costar/Corning Incorporated, Corning, New York) and stimulated with either HIV Gag peptides (BD), cytomegalovirus peptides (BD), staphylococcal enterotoxin B (Sigma), or media alone at 37°C with 5% CO₂, as previously described (16, 38). Anti-CD28 pure monoclonal antibody (clone L293) was added to facilitate costimulation, followed by incubation with brefeldin A for 5 h to prevent Golgi transport of cytokines. Cells were washed, fixed in 4% paraformaldehyde (Sigma), and placed in phosphate-buffered saline-1% bovine serum albumin overnight at 4°C. Cells were washed, permeabilized with Caltag permeabilization solution B (Caltag, Calif.), and stained with monoclonal antibodies specific for CD3, CD8, CD69, and IFN-γ antibodies for 30 min at 4°C. Samples were washed, fixed with 1% paraformaldehyde, and analyzed by flow cytometry (FACScalibur; Becton Dickinson). A minimum of 100,000 events were collected for each sample whenever possible and analyzed using Flow Jo software (Tree Star, Inc., San Carlos, California).

Immunohistochemistry.

Immunohistochemical analysis was performed for the detection and localization of CD4⁺ lymphocytes in jejunal biopsy samples. Tissue samples were embedded in tissue freezing medium (Triangle Biomedical Sciences, Durham, North Carolina) at the time of sample collection and flash frozen. Tissue sections were fixed in STRECK (Streck laboratories, La Vista, Nebraska) and incubated with AR10 (Biogenex, San Ramon, California) for 5 min. Nonspecific binding was blocked by reacting the slides with 10% goat serum (Jackson ImmunoResearch, West Grove, Pennsylvania) and 5% bovine serum albumin for 1 h. Tissue sections were incubated with anti-CD4 (BD clone L200; Becton Dickinson, Mountain View, California) and polyclonal anti-CD3 (Dako, Carpinteria, California) antibodies for 12 h at 4°C, followed by an incubation with anti-mouse Cy3 (Jackson ImmunoResearch) and anti-rabbit Alexa 488 (Invitrogen, Carlsbad, California) for 1 h. The sections were mounted using Slow Fade with 4′,6′-diamidino-2-phenylindole (DAPI; Invitrogen). Images were captured by confocal laser microscopy using an LSM 5 and PASCAL software (Zeiss, New York), and cells positive for both CD3 and CD4 markers were localized in the tissue samples.

Gene expression analysis.

Intestinal mucosal gene expression was investigated by oligonucleotide microarray analysis as previously described (11, 34). Briefly, total RNA from jejunal biopsy samples was extracted using an RNeasy kit (QIAGEN, Valencia, California), and cDNA was synthesized (Superscript Choice system; Gibco Life Technologies, Rockville, Maryland) utilizing an oligo(dT) primer. Biotinylated cRNA was synthesized using a Bioarray high-yield RNA transcription labeling kit (ENZO* Diagnostics, Farmingdale, New York) and purified through RNeasy nucleic acid columns. The quality of the cRNA was evaluated by hybridization to Test3 GeneChips (Affymetrix, Santa Clara, California), and only samples whose 3′/5′ ratios were less than 3 were utilized for subsequent hybridization to HuGene U95-AV2 GeneChips (Affymetrix). The GeneChips were scanned using a laser scanner (Agilent Technologies, Palo Alto, California), and fluorescence data were processed initially through GeneChip Operating System, version 1.1 (Affymetrix). Background correction, normalization, generation of expression values, and statistical analysis of differential gene expression were performed using dChip analysis software (DNA-Chip Analyzer [dChip], version 1.3; Harvard University). Genes differentially expressed (≥1.5-fold above or below mean baseline from healthy, uninfected controls; P value of ≤0.05) in GALT were hierarchically clustered to identify patterns of up (red) and down (green) regulation in response to HIV infection and HAART. The prevalence of enriched biological functions within up and down regulated genes was analyzed (EASE, version 1.0) to determine which processes were statistically overrepresented in the microarray data set.

Statistical analysis.

Longitudinal patterns of change in CD4⁺ T-cell levels in HIV-infected patients following HAART were assessed using linear mixed-effects models. These models are useful in the analysis of longitudinal data in a small number of sample groups. In each model, the outcome variable represents percentage points of increase, determined as absolute CD4⁺ T-cell counts or the CD4⁺ T-cell percentage of total CD3⁺ T cells, with time as the predictor variable, measured by months from the initiation of HAART, and group PHI or CHI (early versus late initiation of HAART) as the covariate. These models provide, for each patient and for each group, an overall level and a rate of change or increase in counts or percentage points per month, as well as average values and average rates of change for both groups. Two-sample t tests were performed to assess differences in the rate of CD4⁺ T-cell restoration between the groups initiating HAART during PHI and during CHI.

Clinical characteristics of HIV-infected patients.

HAART regimens and clinical data from all patients in the study are presented in Table 1. Patients 116, 134, and 140 were identified during primary HIV infection within 4 to 6 weeks following HIV exposure. Patients were diagnosed for HIV infection based on the viral RNA burden in plasma. All three patients (PHI) chose to initiate HAART shortly after diagnosis. At the time of enrollment, all three patients had viral loads of >500,000 copies/ml and >500 cells/mm³ absolute CD4⁺ T-cell numbers in the peripheral blood (Fig. 1A).

Patients 120, 123, 127, 133, and 136 were chronically HIV-infected individuals (CHI) who had viral loads of >50,000 copies/ml of plasma and <300 CD4⁺ cells/mm³ in peripheral blood prior to the initiation of HAART and were known to be HIV positive for >1 year (Fig. 1B). Patients 144 and 145 (CHI) were enrolled in the study after receiving therapy for a period of 5 to 6 years and served as long-term HAART controls. Therefore, longitudinal samples prior to and immediately after HAART were not available from these patients for analysis. These patients had undetectable viral loads, >500 cells/mm³ CD4⁺ T cells in peripheral blood (data not shown) and normal levels of CD4⁺ T cells in jejunal mucosal biopsy samples (43% and 65%, respectively).

Rapid suppression of HIV replication and maintenance of CD4⁺ T cells in peripheral blood of patients initiating HAART during primary infection.

To determine whether the timing of the initiation of HAART influenced the level of viral suppression and CD4⁺ T-cell restoration in peripheral blood, we compared viral loads and CD4⁺ T-cell numbers in PHI and CHI patients. Prior to the initiation of HAART, both PHI and CHI patients displayed high plasma viral loads of >50,000 RNA copies/ml (Fig. 1A). PHI patients 116 and 140 had undetectable viral loads following 4 months of therapy, while patient 134 had <400 RNA copies/ml. In contrast, four of the five CHI patients still had detectable viral loads (347 to 6,682 RNA copies/ml) at 3 to 4 months post-HAART (Fig. 1B). Thus, the initiation of HAART during primary HIV infection resulted in a more rapid and sustained suppression of viral burden in the peripheral blood than the initiation of therapy during chronic HIV infection.

PHI patients did not experience marked CD4⁺ T-cell depletion in peripheral blood prior to initiation of HAART and maintained relatively high CD4⁺ T-cell numbers (510 to 1,146 cells/mm³) during HAART (Fig. 1A). In contrast, a gradual but steady increase in CD4⁺ T-cell numbers was seen in most CHI patients during HAART, which correlated with the suppression of viral loads. Patient 127 experienced a rebound in viral loads after 6 months of therapy and did not show progressive increase in the peripheral CD4⁺ T-cell numbers that correlated with lack of effective suppression of plasma viral burden (Fig. 1B). To determine differences in the rate of CD4⁺ T-cell restoration in peripheral blood of PHI and CHI patients, we evaluated the longitudinal patterns of change in CD4⁺ T-cell levels, measured as an average change in percentage points per month (Fig. 1C). Our results demonstrated a significantly lower (P = 0.00082) rate of CD4⁺ T-cell restoration in PHI patients during HAART. These results were expected, since PHI patients did not display substantial CD4⁺ T-cell depletion in peripheral blood prior to therapy.

Incomplete viral suppression and delayed CD4⁺ T-cell restoration in GALT of HIV-infected patients during HAART.

To determine the effects of initiating HAART during primary or chronic HIV infection on viral suppression and restoration of CD4⁺ T cells in GALT, we evaluated HIV burden and distribution and function of CD4⁺ T-cell subsets in longitudinal jejunal biopsy samples from PHI and CHI patients. While both PHI and CHI patients had detectable HIV RNA levels in GALT prior to the initiation of therapy, the level of viral suppression was different between PHI and CHI patients during HAART (Fig. 2A). Viral loads were undetectable in GALT of PHI patients at 6 months post-HAART and remained undetectable thereafter. In contrast, suppression of viral replication was incomplete in GALT of CHI patients, suggesting that initiation of HAART during primary HIV infection may result in a more rapid and sustained suppression of viral replication in GALT than initiation during chronic HIV infection.

Both CHI and PHI patients displayed severe depletion of CD4⁺ T cells in GALT prior to HAART that was accompanied by an increase in the percentage of mucosal CD8⁺ T cells (Fig. 2B and C). During the first year of HAART, restoration of gut mucosal CD4⁺ T cells was slow in both groups of patients and CD8⁺ T-cell percentages remained elevated in most patients, demonstrating that T-cell homeostasis in GALT was not normalized. Despite a modest level of mucosal CD4⁺ T-cell restoration during the initial 8 months of HAART, patient 116 (PHI group) experienced a strong and steady increase in gut mucosal CD4⁺ T-cell percentages (from 16% to 55%) during the course of therapy (Fig. 2B). PHI patients 134 and 140 were monitored for 8 and 14 months post-HAART, respectively. Both of these patients opted to discontinue HAART thereafter. CD4⁺ T-cell depletion in GALT of these patients (28% and 23%, respectively) was not as severe as that observed in patient 116 (7%) prior to therapy, and neither patient displayed a significant change in CD4⁺ T-cell levels in GALT during therapy (<14 months). We hypothesize that viral replication was not fully suppressed in these patients and may have contributed to the lack of CD4⁺ T-cell restoration during the initial 1 year of therapy.

None of the patients in the CHI group were able to restore mucosal CD4⁺ T cells to normal levels during HAART (Fig. 2C). After 24 months of HAART, patients 120 and 123 experienced a minimal increase in CD4⁺ T-cell levels. Patient 133 exhibited modest restoration of mucosal CD4⁺ T cells initially but then experienced a substantial loss of CD4⁺ T cells after 12 months of HAART (Fig. 2C). Patient 127 experienced fluctuations in the levels of mucosal CD4⁺ T cells, possibly due to a documented lack of adherence to the HAART regimen. Patient 136 maintained mucosal CD4⁺ T cells but showed a decline in the CD4⁺ T-cell levels at 12 months post-HAART. HIV RNA was detected in jejunal biopsy samples from these patients at various time points during therapy, suggesting that the incomplete suppression of viral replication in GALT might have contributed to poor restoration of gut mucosal CD4⁺ T cells.

Repopulating CD4⁺ T cells were localized to mucosal effector sites.

We performed fluorescence immunohistochemistry on jejunal biopsy samples to detect and localize the CD4⁺ T cells repopulating GALT during HAART (Fig. 3). While the CD3⁺ CD4⁺ T cells (yellow) were uniformly scattered throughout the villi and in the crypts of uninfected individuals, severe depletion of CD4⁺ T cells throughout the villi and crypts was observed in HIV-infected patients, with only a few CD4⁺ T cells in a nonspecific pattern near the base of villus. Major aggregates of lymphocytes, likely to be CD8⁺ T cells (CD3⁺ CD4⁻) (green), were detected near the epithelial layer surrounding the villus. At 6 months post-HAART, most of the repopulating CD4⁺ T cells were detected in the crypt region at the base of the villus, with a few cells in the villus. Although patients on long-term therapy (>2 years) showed uniformly scattered CD4⁺ cells in the crypts and villi, the numbers of cells were still lower than those detected in uninfected healthy controls. These data demonstrated that CD4⁺ T cells were depleted from the villus effector sites during HIV infection and that these sites were not adequately replenished even during long-term therapy.

Immunophenotypic characteristics of CD4⁺ T cells repopulating GALT during HAART.

Immunophenotypic and functional characteristics of CD4⁺ T cells repopulating GALT during HAART were evaluated by determining the expression of CD45RA and CD45RO (naive/memory markers), CCR5 and CXCR4 (HIV coreceptors), LFA-1 (CD11a), and α_Eβ₇ (CD103) (trafficking markers) in comparison to healthy HIV-negative controls. The majority of CD4⁺ T cells in GALT of uninfected healthy individuals were of the memory phenotype expressing CD45RO⁺ (>90%) (data not shown). While the naive phenotype (CD45RA⁺) predominated in the CD4⁺ and CD8⁺ T-cell subsets (>45% and >70%, respectively) in peripheral blood prior to and following HAART, the majority of the CD4⁺ T cells in GALT (>90%) during HAART were of a memory phenotype (Fig. 4A).

The expression of CCR5 and CXCR4 was detected on the CD4⁺ T cells repopulating GALT during HAART in HIV-infected patients (Fig. 4B). However, in both CHI and PHI patients, the percentages of CCR5⁺ (16% and 8%, respectively) and CXCR4⁺ CD4⁺ T cells (10% and 10%, respectively) were significantly lower than in healthy uninfected controls (38% and 51.0%, respectively). The expression of HIV coreceptors on repopulating CD4⁺ T cells in GALT highlights their susceptibility to infection by both R5- and X4-tropic viruses and provides further evidence that ongoing viral replication may impair the efficacy of the restoration.

The expression of the lymphocyte trafficking molecules, LFA-1 and integrin α_Eβ₇, was also analyzed in CD4⁺ T cells repopulating GALT during therapy (Fig. 4B). Although levels of α_Eβ₇ expressing CD4⁺ T cells were similar in PHI and CHI patients, increased levels of LFA-1-expressing mucosal CD4⁺ T cells were observed in PHI (56%) compared to CHI patients (27%). No significant change was observed in the levels of CD4⁺ T cells expressing Ki67, a cell proliferation marker (<3%) (data not shown). These findings suggest that increased mucosal T-cell trafficking and not local proliferation may be the major mechanism of CD4⁺ T-cell repopulation in GALT during early stages of HAART.

HIV-specific CD8⁺ T-cell responses in GALT and peripheral blood of HIV-infected patients during HAART.

To examine the correlation between the HIV-specific antiviral cellular responses in GALT and viral suppression during therapy, HIV Gag-specific CD8⁺ T-cell responses were measured from both PHI and CHI patients at various time points post-HAART. Gag-specific CD8⁺ T-cell responses were detected in GALT of both PHI and CHI patients with detectable viral loads in GALT at the time of the sample collection (Fig. 5). One patient (123) in particular displayed strong HIV-specific responses and had significantly higher viral loads in GALT at the time of sample collection. These data suggested that incomplete suppression of viral replication in GALT led to increased levels of HIV Gag-specific CD8⁺ T-cell responses.

Mucosal gene expression profiles in GALT may be predictive of response to HAART.

To determine the molecular mechanisms involved in the restoration of the gut mucosal immune system during HAART, we compared changes in the mucosal gene expression profiles of PHI and CHI patients. We hypothesized that changes in gene expression patterns in the gut microenvironment may precede detectable changes at the cellular and tissue levels. Therefore, we compared the gut mucosal gene expression profiles prior to HAART and at 3 months posttherapy in two patients who showed efficient mucosal CD4⁺ T-cell restoration (116 and 133) with two patients who displayed poor mucosal CD4⁺ T-cell restoration following HAART (120 and 123). Hierarchical clustering of differentially expressed genes in these patients revealed a distinct pattern of up regulated gene expression in those patients with poor CD4⁺ T-cell reconstitution (120 and 123) after 3 months of HAART. This included increased expression of genes involved in innate and acquired immune responses, apoptosis, and cell injury (Fig. 6). Genes associated with stress and inflammation were also up regulated in patients with poor CD4⁺ T-cell repopulation, including Toll-like receptors 1 and 5, myxovirus resistance proteins, and numerous interferon-induced factors (Fig. 7A). In contrast, patients 116 and 133 showed a marked decrease in the level of immune and inflammatory response-related genes at 3 months posttherapy. Thus, molecular processes leading to immune activation and tissue damage dominated the gut microenvironment in patients who failed to restore mucosal CD4⁺ T cells.

In agreement with recent reports (34), we found decreased expression of genes involved in digestive and absorptive functions, lipid metabolism, and xenobiotic metabolism in all HIV-infected patients (Fig. 7B). Prior to therapy, the degree of suppression appeared more severe in patients 120 and 123. This functional category included genes involved in lipid, carbohydrate, and steroid metabolism as well as large- and small-molecule transporter proteins and catabolic factors. Importantly, several cytochrome P450 family members involved in metabolism of HAART compounds were substantially down regulated, indicating a potential to impact the efficacy of therapy.

Viral loads were readily detectable in GALT of patient 123 after 6 months of therapy (Fig. 8), coinciding with increased expression of genes associated with inflammatory response and immune activation in GALT (RANTES, NF-κB, multiple interferon response factors, and Mx-2). In contrast, patient 116 maintained a relatively normal to minimally decreased expression of these genes in GALT at 6 months post-HAART. Patient 116 also displayed increased expression of genes involved in mucosal trafficking (alpha integrins 4 and E, ICAM, and VCAM) and growth (Ki67, platelet and transforming growth factors, and trefoil factor 3). In summary, mucosal gene expression profiles prior to therapy differentiated patients who progressed to restore gut mucosal CD4⁺ T cells poorly or efficiently during the first few months of therapy. These data suggest that the extent of inflammation, cell activation, and apoptosis may be indicative of the degree of disruption in the mucosal microenvironment prior to therapy and may directly impact the efficacy of viral suppression and CD4⁺ T-cell restoration during therapy.