|Home | About | Journals | Submit | Contact Us | Français|
Post hoc analysis of the phase 2b Step study evaluating a recombinant adenovirus serotype 5 (rAd5)-based HIV-1 vaccine candidate suggested a potential increased risk of HIV-1 acquisition in subjects who were baseline Ad5 seropositive and uncircumcised. These concerns had a profound impact on the HIV-1 vaccine development field, although the mechanism underlying this observation remains unknown. It has been hypothesized that rAd5 vaccination of baseline Ad5-seropositive individuals may have resulted in anamnestic, vector-specific CD4+ T lymphocytes that could have trafficked to mucosal sites and served as increased targets for HIV-1 infection. Here we show that Ad5-specific CD4+ T lymphocyte responses at mucosal sites following rAd5-Gag/Pol/Nef vaccination were comparable in rhesus monkeys with and without baseline Ad5 immunity. Moreover, the total cellular inflammatory infiltrates and the CD3+, CD4+, HLA-DR+, Ki67+, and langerin+ cellular subpopulations in colorectal and foreskin mucosa were similar in both groups. Thus, no greater trafficking of Ad5-specific CD4+ T lymphocytes to mucosal target sites was observed following rAd5 vaccination of rhesus monkeys with baseline Ad5 immunity. These findings from this nonhuman primate model provide evidence against the hypothesis that recruitment of vector-specific target cells to mucosal sites led to increased HIV-1 acquisition in Ad5-seropositive, uncircumcised vaccinees in the Step study.
The Step study revealed a potential increased risk of HIV-1 acquisition among adenovirus serotype 5 (Ad5)-seropositive, uncircumcised subjects who received the Merck recombinant Ad5 (rAd5)-Gag/Pol/Nef vaccine candidate (2, 6). It has been hypothesized that rAd5 vaccination of Ad5-seropositive individuals may have resulted in robust expansion and activation of vector-specific CD4+ T lymphocytes that could have trafficked to mucosal sites and served as increased targets for HIV-1 infection. Our laboratory and others have recently demonstrated that total and vector-specific CD4+ T lymphocytes in peripheral blood in Ad5-seropositive volunteers were comparable to or lower than the levels in Ad5-seronegative volunteers following rAd5-Gag vaccination in the Merck phase 1 studies (4, 8). However, mucosal biopsy specimens were not obtained in these clinical trials, and thus the extent of inflammatory infiltrates and vector-specific CD4+ T lymphocytes in colorectal and foreskin mucosa could not be evaluated in these prior studies.
It has also recently been reported that vector-specific CD4+ T lymphocytes may upregulate mucosal homing integrin expression following exposure to Ad5 in short-term in vitro cultures (1). These findings highlight the importance of directly investigating the extent and nature of vector-specific CD4+ T lymphocytes at mucosal sites following rAd5 vaccination. Given the lack of mucosal biopsy samples from human subjects in the Step study, we developed a nonhuman primate model of preexisting adenovirus immunity to evaluate the extent and nature of inflammatory cell populations at mucosal sites following rAd5 vaccination.
Adult Indian-origin rhesus monkeys (n = 12) were housed in the biosafety level 3 containment facility at the New England Primate Research Center (NEPRC), and all studies were approved by the Harvard Medical School Institutional Animal Care and Use Committee (IACUC). Baseline endoscopic biopsy specimens of the colorectum and snip biopsy specimens of inner penile foreskin (IPF) were collected prior to inoculation. Animals were inoculated twice at an interval of 4 weeks by the intranasal route with either 1011 viral particles of replication-competent Ad5 containing host range mutations (n = 6) or saline (n = 6). After 4 weeks, animals were vaccinated by the intramuscular route with 1011 viral particles of replication-incompetent, E1-/E3-deleted rAd5-Gag/Pol/Nef vectors in the quadriceps muscles. Of note, replication-incompetent, E1-deleted rAd5-Gag/Pol/Nef vectors were utilized in the Step study. Animals were sacrificed 2 weeks after vaccination, and a full necropsy was performed.
Colorectal biopsy specimens in RPMI 1640 medium containing 10% fetal bovine serum supplemented with 300 U/ml of type IV collagenase (Sigma-Aldrich) and 30 U/ml of DNase I (Sigma) were incubated at 37°C with rocking for 30 min. The digested biopsy tissues were then homogenized, and the solution was strained with a 70-μm cell strainer (BD Falcon). The cell suspension was then centrifuged at 1,800 rpm for 25 min on a 35% Percoll gradient (Sigma). Pellets containing the lymphocytes were collected and processed for intracellular cytokine staining. Colorectal tissue specimens obtained at necropsy were cut into pieces of similar size to the biopsy specimens and were then processed similarly.
Ad5-specific gamma interferon (IFN-γ) enzyme-linked immunospot (ELISPOT) assays were conducted essentially as described previously (4, 8). Peripheral blood mononuclear cells (PBMC) were stimulated by replication-incompetent Ad5 virus (104 multiplicity of infection) or by pooled overlapping peptides covering the Ad5 E2A or Ad5 hexon protein (2 μg ml−1 of each peptide).
The magnitudes and phenotypes of Ad5-specific cellular immune responses were assessed by multiparameter intracellular cytokine staining (ICS) assays essentially as described previously (4, 8). PBMC (1.5 × 106) or mucosal lymphocytes were incubated for 6 h at 37°C with the medium as the negative control, 10 pg ml−1 phorbol myristate acetate (PMA), and 1 μg ml−1 ionomycin (Sigma-Aldrich) as the positive control, or 5 ng ml−1 Ad5 hexon peptide pools. The cultures contained monensin (GolgiStop; BD Biosciences) and 1 μg ml−1 of a monoclonal antibody (MAb) against CD49d. All MAbs were purchased from BD Biosciences unless otherwise indicated. The cells were then stained with predetermined titers of MAbs against CD3 (SP34.2; allophycocyanin [APC] Cy7), CD4 (L200; AmCyan), CD8 (7pt3F9; Ab was conjugated with Dot605 in the laboratory), CD28 (L293; peridinium chlorophyll protein [PerCP]-Cy5.5), CD69 (TP1.55.3, phycoerythrin-Texas red), CD95 (DX2; phycoerythrin [PE]), Ki67 (B56; fluorescein isothiocyanate [FITC]), or α4β7 (APC; kindly provided by Keith Reimann, Beth Israel Deaconess Medical Center) and were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences). The cells were then stained intracellularly with MAbs against IFN-γ (B27; PE-Cy7), tumor necrosis factor alpha (TNF-α) (Mab11; Pacific blue; eBiosciences), and interleukin-2 (IL-2) (MQ1-17H12; Alexa 700; Biolegend) and were fixed with 1% paraformaldehyde. Samples were evaluated with an LSR II (BD Biosciences) and analyzed using FlowJo software (TreeStar). Approximately 300,000 to 500,000 events were collected per sample. The medium control background levels were typically <0.02% of the gated CD8+ or CD4+ T lymphocyte levels.
Tissue specimens were fixed in 10% neutral buffered formalin (NBF) and embedded in paraffin. After fixation, the colorectum and jejunum were serially sectioned circumferentially at 0.5-cm intervals to yield a minimum of 5 sections per animal. For foreskin, the penis was fully extended to reveal the glans, inner penile foreskin (IPF), and inner foreskin (IF) and allowed to fix in NBF. Cross-sections at 0.5-cm intervals were cut from the glans, IPF, and IF, and serial 5-μm tissue sections were either stained with hematoxylin and eosin (H&E) for histopathologic examination or subjected to immunohistochemical staining.
Briefly, tissue sections were heated and rehydrated, incubated in 1% hydrogen peroxide in methanol (CD4) or 3% hydrogen peroxidase in phosphate-buffered saline (PBS) (HLA-DR, Ki67, and CD3), and subjected to antigen retrieval procedures, including microwaving in 0.01 M citrate buffer (Dako, Carpinteria, CA) (Ki67 and CD3) or heating under pressure in 1 mM EDTA-0.05% Tween (CD4), as previously described (11). Nonspecific staining was blocked with Dako protein block, and sections were incubated overnight at 4°C with primary antibodies to Ki67, HLA-DR, CD3 (Dako), or CD4 (Vector Laboratories, Burlingame, CA) or appropriate isotype-matched irrelevant MAb controls, followed by incubation with biotinylated secondary antibody (Dako). Slides were then treated with Vectastain ABC elite, followed by development with chromogen (Dako). Rehydrated slides to be stained for langerin were heated under pressure in Trilogy solution (Cell Marque Corporation, Rocklin, CA), and endogenous hydrogen peroxide activity and nonspecific staining were inhibited by serial incubation in Envision dual endogenous enzyme block (Dako), followed by Dako protein block. Langerin primary antibody (AbCam, Cambridge, MA) was applied for 60 min at room temperature, followed by serial incubation with Envision labeled polymer (Dako) and Envision diaminobenzidine (DAB) chromogen solution (Dako). All slides were counterstained with Mayer's hematoxylin (Newcomer Supply).
Objective histopathologic scoring criteria were developed by evaluating serial sections of stained colorectum and penis/foreskin from control young adult male rhesus macaques (n = 6), as well as from animals in the present study (n = 12). At each site, 10 or more randomly selected regions of lamina propria, submucosa, and/or epithelium were examined at 200× magnification with an Olympus BX41 microscope (Olympus America, Inc.) and assigned scores ranging from 0 to 2 (colorectum) or 0 to 3 (penis/foreskin), corresponding to cell counts as detailed in the footnotes of Tables Tables11 and and2.2. Quantitative enumeration of stained cells in H&E-treated and immunohistochemically stained samples was performed at 400× magnification using an Olympus BX41 microscope equipped with a DP25 camera (Olympus). Nucleated or stained cells within the lamina propria or epithelium were enumerated, and the area evaluated was measured using DP2-BSW software (Olympus) to yield a series of measurements of the number of cells/area. Ten or more fields for each animal at each site were evaluated. Keratin thickness (μm) was measured at 200× magnification in a minimum of 10 fields at each anatomic site in the foreskin. Image acquisition was at 200× magnification with the above-listed microscope and camera.
All statistical analyses were conducted using GraphPad Prism, version 5. Statistical comparisons of keratin thickness and langerin+ cell counts between foreskin sites were conducted using the 2-tailed Wilcoxon matched-pair test, and P values of <0.05 were considered significant.
We induced baseline Ad5 immunity in rhesus monkeys by respiratory infection with replication-competent Ad5 virus, which contained the Ad5 E1 gene and host range mutations that allowed efficient virus replication in monkey cell lines in vitro. We infected adult male rhesus monkeys twice by the intranasal route with 1011 viral particles of replication-competent Ad5 (n = 6) or sham saline (n = 6). Replication-competent Ad5 infection resulted in robust respiratory virus shedding for 2 weeks as measured by virus-specific quantitative PCR assays (data not shown). Monkeys infected with replication-competent Ad5 also developed Ad5-specific neutralizing antibodies (NAbs; titers of 190 to 648) (Fig. (Fig.1a)1a) and Ad5-specific T lymphocyte responses that were detected by IFN-γ ELISPOT assays (Fig. (Fig.1b)1b) and multiparameter intracellular cytokine staining (ICS) assays (Fig. (Fig.1c)1c) at magnitudes and phenotypes comparable to those observed using these assays in humans (4, 8).
We next vaccinated all 12 monkeys by the intramuscular route with 1011 viral particles of E1/E3-deleted, replication-incompetent rAd5-Gag/Pol/Nef vectors 4 weeks after the second intranasal replication-competent Ad5 infection. Ad5-specific CD4+ T lymphocyte responses in peripheral blood were observed in all monkeys following vaccination. These responses were slightly lower in monkeys with baseline Ad5 immunity than in monkeys without baseline Ad5 immunity (Fig. (Fig.2a),2a), consistent with data obtained from human subjects in the Merck phase 1 clinical vaccine trials (4, 8). These immunologic data effectively bridge observations obtained from clinical trials with this preclinical model. At week 1 following vaccination, the majority of Ad5-specific CD4+ T lymphocytes in both groups exhibited an activated Ki67+ phenotype, although this cellular activation was transient and had largely disappeared by week 2 (Fig. (Fig.2a2a).
At week 2 following vaccination, all monkeys were sacrificed to facilitate a comprehensive evaluation of total and vector-specific CD4+ T lymphocytes at mucosal sites by flow cytometry, histopathology, and immunohistochemistry. The magnitude and phenotype of colorectal Ad5-specific CD4+ T lymphocyte responses proved comparable in baseline Ad5-seropositive and baseline Ad5-seronegative monkeys (Fig. (Fig.2b).2b). Moreover, colorectal Ad5-specific CD4+ T lymphocyte responses were similar in magnitude to those observed in peripheral blood at week 2 (Fig. 2a and b). These data show that there was no greater trafficking of Ad5-specific CD4+ T lymphocytes to the colorectal mucosa in monkeys with baseline Ad5 immunity than in monkeys without baseline Ad5 immunity.
We next characterized inflammatory cell populations in the colorectal and foreskin mucosa before and after vaccination. Multiple serial cross-sections of the colorectum were collected, and cell populations in lamina propria and mucosal epithelium were assessed by both objective scoring criteria and quantitative image analysis (Materials and Methods and Table Table1).1). The total cellular infiltrates in colorectal mucosa were determined by enumerating nucleated cells in serial sections of H&E-stained lamina propria and proved comparable at baseline and after rAd5 vaccination in monkeys with and without baseline Ad5 immunity (Fig. (Fig.3a3a).
We evaluated subpopulations of inflammatory cells in serial sections of colorectal mucosa by immunohistochemistry. The total CD3+ cellular infiltrates in both lamina propria and mucosal epithelium were similar at baseline and following rAd5 vaccination in monkeys with or without baseline Ad5 immunity (Fig. (Fig.3b).3b). Both objective scoring (data not shown) and quantitative image analysis (Fig. (Fig.3c)3c) also showed that the numbers and distribution of CD4+ lymphocytes in colorectal lamina propria were similar in baseline and vaccinated animals in both groups regardless of baseline Ad5 immunity. Moreover, the levels of proliferative activity and cellular activation in colorectal mucosa, as measured by Ki67 (Fig. (Fig.3d)3d) and HLA-DR (Fig. (Fig.3e)3e) expression, proved comparable between groups. Thus, no differences in the quantity or nature of colorectal cellular inflammatory infiltrates were detected in baseline Ad5-seropositive and baseline Ad5-seronegative monkeys following rAd5-Gag/Pol/Nef vaccination.
To determine the generalizability of these findings, we assessed the extent and nature of cellular inflammatory infiltrates in serial sections of small bowel mucosa. Similar to the results observed in colorectal mucosa, there were no differences in total (Fig. (Fig.4a),4a), CD4+ (Fig. (Fig.4b),4b), or Ki67+ (Fig. (Fig.4c)4c) cellular subsets within jejunal lamina propria in baseline Ad5-seropositive and baseline Ad5-seronegative monkeys following vaccination. These data indicate that baseline Ad5 immunity did not augment the trafficking of inflammatory cells to mucosal sites following vaccination with rAd5-Gag/Pol/Nef.
We next explored cellular inflammatory infiltrates in the foreskin (prepuce) and penis of these monkeys at week 2 following rAd5 vaccination, since the risk of acquisition of HIV-1 infection in the Step study appeared to be greatest in Ad5-seropositive, uncircumcised males (6). At necropsy, the penis and foreskin of all animals were fully extended to reveal the entire penile mucosa and were allowed to fix completely. Serial sections were then obtained from 3 anatomic locations: the glans, inner penile foreskin (IPF), and inner foreskin (IF) (Fig. (Fig.5a).5a). Baseline biopsy specimens prior to vaccination were obtained only from the IPF. Cellular infiltrates in these tissues were evaluated by histopathology and immunohistochemistry using both objective scoring criteria and quantitative image analysis. Importantly, histologic features of the rhesus penis and foreskin proved comparable to those of humans (10) and consisted of a moderately thick keratinized stratified squamous epithelium overlying a vascular submucosa containing a similar complement of resident immune cells (Fig. (Fig.5a).5a). Moreover, we measured keratin thickness at each anatomic site as a measure of barrier function and found that the variation in keratin thickness with anatomic location was comparable to that in humans (5) and that the keratin thickness proved to be significantly less in the IF than in the IPF or glans (P = 0.03 and P = 5 × 10−4, respectively) (Fig. (Fig.6a).6a). The keratin thickness was not affected by baseline Ad5 immunity at any anatomic site (Fig. (Fig.6b).6b). These data show that the rhesus monkey penis and foreskin share multiple key histologic features with the corresponding human tissues.
The total numbers and distribution of CD3+ cells in foreskin proved comparable in baseline and vaccinated animals in both groups regardless of baseline Ad5 immunity (Fig. (Fig.5b).5b). Similar results were obtained for CD4+ (Fig. (Fig.5c),5c), Ki67+ (Fig. (Fig.5d),5d), and HLA-DR+ (Fig. (Fig.5e)5e) cells. We also assessed the numbers and distribution of Langerhans cells (LCs), which may be initial targets of viral infection at genital mucosal surfaces (3). LCs were present in the superficial epithelium (Fig. (Fig.5f)5f) and were most abundant in the IPF, followed by the IF and glans (Fig. (Fig.6c),6c), indicating an anatomic variation in cell density similar to that in humans (5). However, no differences in the density of langerin+ cells at any anatomic location were detected between baseline Ad5-seropositive and baseline Ad5-seronegative monkeys following vaccination (Fig. (Fig.5f5f).
Our data demonstrate that Ad5-specific CD4+ T lymphocytes at mucosal sites were comparable in rhesus monkeys with and without baseline Ad5 immunity following rAd5-Gag/Pol/Nef vaccination. Thus, we detected no evidence of increased trafficking of vector-specific CD4+ T lymphocytes to mucosal sites following rAd5 vaccination of baseline Ad5-seropositive monkeys compared to that in baseline Ad5-seronegative monkeys. Transient cellular activation was observed in peripheral blood at week 1 following rAd5 vaccination, consistent with recent in vitro findings (1), but this effect was no longer evident in the periphery or in the mucosa by week 2 following vaccination.
An important caveat of the present study is the use of the nonhuman primate model, which was required since mucosal biopsy specimens were not available from the Merck rAd5 clinical vaccine studies. We believe that this is a relevant model for studying Ad5 immunity, since multiple key features of this model correlate well with those in humans, including (i) the use of replication-competent Ad5 inoculated by the intranasal route, (ii) comparable duration of respiratory virus shedding, (iii) comparable induction of baseline Ad5-specific humoral and cellular immune responses (4, 8), (iv) comparable induction of Ad5-specific immune responses in the periphery following rAd5-Gag/Pol/Nef vaccination (4, 8), and (v) comparable histology of the colorectum and foreskin (5, 10). Moreover, male rhesus monkeys can be infected by inoculation of simian immunodeficiency virus to intact colorectal or penile mucosa, and LCs are implicated as initial targets for viral infection in genital mucosa in both rhesus monkeys and humans (7). One important difference between the nonhuman primate model and humans is the timing of the induction of Ad5 immunity, which was 4 and 8 weeks prior to vaccination in the present study compared with potentially much longer time frames in humans. We feel that this is a relatively minor difference, however, since the Ad5-specific humoral and cellular immune responses proved comparable. Thus, many key virologic and immunologic characteristics of Ad5 immunity are shared between the nonhuman primate model and humans.
Recent data based on short-term in vitro cultures have suggested that coculture of Ad5-specific CD4+ T lymphocytes with large quantities of rAd5 vectors can lead to cellular activation and induction of mucosal homing marker integrin expression (1), possibly as a result of immune complex formation and dendritic cell activation (9). Our data show that these in vitro observations do not translate into increased mucosal trafficking of Ad5-specific CD4+ T lymphocytes in vivo in this nonhuman primate model. In particular, there was no detectable increase in total or Ad5-specific CD4+ T lymphocytes at mucosal surfaces following rAd5-Gag/Pol/Nef vaccination of baseline Ad5-seropositive monkeys compared with baseline Ad5-seronegative monkeys. Although definitive data can only be derived from clinical vaccine studies of humans, these findings from this nonhuman primate model provide evidence against the hypothesis that recruitment of vector-specific target cells to mucosal sites led to increased HIV-1 acquisition in Ad5-seropositive, uncircumcised vaccinees in the Step study.
We thank A. Miller, L. Curran, N. Silva, N. Evangelous, K. Boisvert, H. Knight, D. Lynch, A. Riggs, F. Stephens, K. Reimann, M. Lifton, K. Furr, and J. Goudsmit for generous advice and assistance.
We acknowledge support from the Bill & Melinda Gates Foundation (38614) and the National Institutes of Health (AI066305, AI066924, AI078526, RR000168, and RR007000). The authors declare no financial conflicts of interest.
Published ahead of print on 4 August 2010.
†The authors have paid a fee to allow immediate free access to this article.