Prevention of SIV Rectal Transmission and Priming of T Cell Responses in Macaques after Local Pre-exposure Application of Tenofovir Gel

Background The rectum is particularly vulnerable to HIV transmission having only a single protective layer of columnar epithelium overlying tissue rich in activated lymphoid cells; thus, unprotected anal intercourse in both women and men carries a higher risk of infection than other sexual routes. In the absence of effective prophylactic vaccines, increasing attention is being given to the use of microbicides and preventative antiretroviral (ARV) drugs. To prevent mucosal transmission of HIV, a microbicide/ARV should ideally act locally at and near the virus portal of entry. As part of an integrated rectal microbicide development programme, we have evaluated rectal application of the nucleotide reverse transcriptase (RT) inhibitor tenofovir (PMPA, 9-[(R)-2-(phosphonomethoxy) propyl] adenine monohydrate), a drug licensed for therapeutic use, for protective efficacy against rectal challenge with simian immunodeficiency virus (SIV) in a well-established and standardised macaque model. Methods and Findings A total of 20 purpose-bred Indian rhesus macaques were used to evaluate the protective efficacy of topical tenofovir. Nine animals received 1% tenofovir gel per rectum up to 2 h prior to virus challenge, four macaques received placebo gel, and four macaques remained untreated. In addition, three macaques were given tenofovir gel 2 h after virus challenge. Following intrarectal instillation of 20 median rectal infectious doses (MID50) of a noncloned, virulent stock of SIVmac251/32H, all animals were analysed for virus infection, by virus isolation from peripheral blood mononuclear cells (PBMC), quantitative proviral DNA load in PBMC, plasma viral RNA (vRNA) load by sensitive quantitative competitive (qc) RT-PCR, and presence of SIV-specific serum antibodies by ELISA. We report here a significant protective effect (p = 0.003; Fisher exact probability test) wherein eight of nine macaques given tenofovir per rectum up to 2 h prior to virus challenge were protected from infection (n = 6) or had modified virus outcomes (n = 2), while all untreated macaques and three of four macaques given placebo gel were infected, as were two of three animals receiving tenofovir gel after challenge. Moreover, analysis of lymphoid tissues post mortem failed to reveal sequestration of SIV in the protected animals. We found a strong positive association between the concentration of tenofovir in the plasma 15 min after rectal application of gel and the degree of protection in the six animals challenged with virus at this time point. Moreover, colorectal explants from non-SIV challenged tenofovir-treated macaques were resistant to infection ex vivo, whereas no inhibition was seen in explants from the small intestine. Tissue-specific inhibition of infection was associated with the intracellular detection of tenofovir. Intriguingly, in the absence of seroconversion, Gag-specific gamma interferon (IFN-γ)-secreting T cells were detected in the blood of four of seven protected animals tested, with frequencies ranging from 144 spot forming cells (SFC)/106 PBMC to 261 spot forming cells (SFC)/106 PBMC. Conclusions These results indicate that colorectal pretreatment with ARV drugs, such as tenofovir, has potential as a clinically relevant strategy for the prevention of HIV transmission. We conclude that plasma tenofovir concentration measured 15 min after rectal administration may serve as a surrogate indicator of protective efficacy. This may prove to be useful in the design of clinical studies. Furthermore, in vitro intestinal explants served as a model for drug distribution in vivo and susceptibility to virus infection. The finding of T cell priming following exposure to virus in the absence of overt infection is provocative. Further studies would reveal if a combined modality microbicide and vaccination strategy is feasible by determining the full extent of local immune responses induced and their protective potential.


A B S T R A C T Background
The rectum is particularly vulnerable to HIV transmission having only a single protective layer of columnar epithelium overlying tissue rich in activated lymphoid cells; thus, unprotected anal intercourse in both women and men carries a higher risk of infection than other sexual routes. In the absence of effective prophylactic vaccines, increasing attention is being given to the use of microbicides and preventative antiretroviral (ARV) drugs. To prevent mucosal transmission of HIV, a microbicide/ARV should ideally act locally at and near the virus portal of entry. As part of an integrated rectal microbicide development programme, we have evaluated rectal application of the nucleotide reverse transcriptase (RT) inhibitor tenofovir (PMPA, 9-[(R)-2-(phosphonomethoxy) propyl] adenine monohydrate), a drug licensed for therapeutic use, for protective efficacy against rectal challenge with simian immunodeficiency virus (SIV) in a well-established and standardised macaque model.

Methods and Findings
A total of 20 purpose-bred Indian rhesus macaques were used to evaluate the protective efficacy of topical tenofovir. Nine animals received 1% tenofovir gel per rectum up to 2 h prior to virus challenge, four macaques received placebo gel, and four macaques remained untreated. In addition, three macaques were given tenofovir gel 2 h after virus challenge. Following intrarectal instillation of 20 median rectal infectious doses (MID 50 ) of a noncloned, virulent stock of SIV mac251/32H , all animals were analysed for virus infection, by virus isolation from peripheral blood mononuclear cells (PBMC), quantitative proviral DNA load in PBMC, plasma viral RNA (vRNA) load by sensitive quantitative competitive (qc) RT-PCR, and presence of SIV-specific serum antibodies by ELISA. We report here a significant protective effect (p ¼ 0.003; Fisher exact probability test) wherein eight of nine macaques given tenofovir per rectum up to 2 h prior to virus challenge were protected from infection (n¼6) or had modified virus outcomes (n ¼ 2), while all untreated macaques and three of four macaques given placebo gel were infected, as were two of three animals receiving tenofovir gel after challenge. Moreover, analysis of lymphoid tissues post mortem failed to reveal sequestration of SIV in the protected animals. We found a strong positive association between the concentration of tenofovir in the plasma 15 min after rectal application of gel and the degree of protection in the six animals challenged with virus at this time point. Moreover, colorectal explants from non-SIV challenged tenofovir-treated macaques were resistant to infection ex vivo, whereas no inhibition was seen in explants from the small intestine. Tissue-specific inhibition of infection was associated with the intracellular detection of tenofovir. Intriguingly, in the absence of seroconversion, Gag-specific gamma interferon (IFN-c)secreting T cells were detected in the blood of four of seven protected animals tested, with frequencies ranging from 144 spot forming cells (SFC)/10 6 PBMC to 261 spot forming cells (SFC)/10 6 PBMC.

Conclusions
These results indicate that colorectal pretreatment with ARV drugs, such as tenofovir, has potential as a clinically relevant strategy for the prevention of HIV transmission. We conclude that plasma tenofovir concentration measured 15 min after rectal administration may serve as a surrogate indicator of protective efficacy. This may prove to be useful in the design of clinical studies. Furthermore, in vitro intestinal explants served as a model for drug distribution in vivo and susceptibility to virus infection. The finding of T cell priming following exposure to virus in the absence of overt infection is provocative. Further studies would reveal if a combined modality microbicide and vaccination strategy is feasible by determining the full extent of local immune responses induced and their protective potential.

Introduction
The development of an effective vaccine against HIV is still thought to be a long-term endeavour; meanwhile the HIV pandemic continues relentlessly, fuelled primarily by sexual transmission. The relative ease by which HIV is transmitted rectally [1][2][3][4] makes this a particularly important, although relatively neglected, route to target with prevention strategies. Unlike the vagina, the rectal canal has only a single layer of columnar epithelium overlying tissue rich in activated lymphoid cells [5,6] and therefore for reasons of both anatomy and immunological status presents a particular challenge for preventative modalities. It is difficult to estimate the prevalence of anal intercourse (AI) in the heterosexual population, but recent studies have indicated that it may be far more common than had been thought. In a population-based study of 2,547 Northern Californian women aged 18-29 y, AI was reported in 21.7% of sexually active individuals [7]. Higher rates of AI have been described in women at particular risk of HIV infection through drug use, forced intercourse, and prostitution [8][9][10][11]. Data from Africa are very limited, but levels in excess of 40% incidence have been described in commercial sex workers in South Africa [12,13]. These figures alone demonstrate the need for the development of rectal microbicides, a focus catalyzed by the more visible need of the population of men who have sex with men, which still accounts for the majority of new infections in North America, South America, and Europe. Moreover, many of these men who have sex with men are having unprotected AI irrespective of their HIV status [14]. In the absence of vaccines, increasing effort is being applied to the parallel approach of preventative microbicides/pre-exposure use of antiretrovirals (ARVs). To date preclinical studies of tenofovir in the macaque rectal challenge model using either simian immunodeficiency virus (SIV) or recombinant chimeric simian HIV (SHIV) have focussed on the use of the orally bio-available prodrug tenofovir disoproxil fumarate (TDF) [15,16]. Although partial protection was observed against multiple low dose mucosal virus challenge, it appeared that the dose of available drug may have been suboptimal. We have taken a different approach, reasoning that the optimal route to protect cells of the rectum and possibly beyond may be through local application of tenofovir given, in effect, as a rectal microbicide. As well as testing this hypothesis, we sought to determine if protective efficacy correlated with drug uptake and whether exposure to virus in the absence of overt infection stimulated virus-specific T cell responses.

Macaques, Tenofovir Gel, and Virus
A well-established macaque rectal challenge model using uncloned SIVmac 251/32H ; a virus that results in high cellassociated and plasma viral RNA (vRNA) loads shortly after a single application to naïve macaques [17][18][19][20] was used for this study. Purpose-bred male rhesus macaques (Macaca mulatta) of Indian origin aged 4-6 y were obtained from a UK breeding colony. All animals were housed according to the Code of Practice of the UK Home Office (1989) and were sedated with ketamine hydrochloride prior to inoculation with virus and/ or venepuncture. Animals were killed humanely by an overdose of anaesthetic. All procedures involving animals were approved by the Ethical Review Committee of the Centre for Emergency Preparedness and Response, Salisbury, UK. None of the animals had been used previously for experimental procedures and all animals were shown to be SIV negative (by virus isolation [VI], PCR, and serology) before entering the experiment. Tenofovir gel (1%) or control placebo gel, a proprietary formulation containing purified water with edetate disodium, citric acid, glycerin, methylparaben, propylparaben, and hydroxyethylcellulose (pH 4.5) supplied by Gilead Sciences was transferred to 5ml syringes and administered in 3-ml volumes via a 10 FG soft catheter introduced 8 cm into the rectum. The process was atraumatic with no obvious leakage. Assignment of macaques to different treatment groups is shown in Table 1.
All macaques were challenged with 20 median rectal infectious doses MID 50 of SIVmac 251/32H 11/88 stock equivalent to 18,974 50% tissue culture doses (TCID 50 ) as determined by previous titration [18]. Virus inoculum was diluted aseptically shortly before challenge and administered in 3-ml volumes with a 6 FG soft catheter 8 cm into the rectum. At 1, 2, 6, 12, 16, and 20 wk postchallenge macaques were bled to provide 15 ml of heparinised blood for VI and proviral DNA assay, 2 ml of EDTA blood for plasma vRNA load determination, and 2 ml of clotted blood for serum antibody analysis. Animals were observed for adverse effects immediately after injection of test substance, followed by daily observation for the duration of study. At each time of sedation the following clinical data were taken for each animal: body weight, temperature, axillary and inguinal lymph node scores, and haemoglobin concentration. Necropsies were performed on all the animals from which SIV had not been recovered and on two SIV-infected animals. At necropsy, the animals were assessed for gross pathology and the following specimens collected for analysis: heparinised blood, spleen, inguinal, mesenteric, and iliac lymph nodes, 20cm sections of ileum/jejunum, and colorectum tissue. The gut tissues were washed to remove faecal material and each sample placed into 20 ml of RPMI supplemented with Lglutamine (2 mM), penicillin (50 U/ml), streptomycin (50 lg/ ml), and gentamicin (80 lg/ml) before rapid transport to the laboratory (3-4 h) on water ice for explant culture. All other tissues, as well as small samples of the gut tissues, were frozen at À70 8C for later analysis by PCR.
Two other SIV-naïve rhesus macaques were used to obtain gut tissues for explant studies of tenofovir antiviral activity in vitro (see below). A further four SIV-naïve macaques were dosed with tenofovir gel as described above 3 h prior to necropsy, and gut tissues used for ex vivo analysis of intracellular drug concentration and viral inhibitory activity (Table 1).

Virus Isolation
At each time point of sampling, 2 3 10 6 peripheral blood mononuclear cells (PBMC) were cocultured with C8166 cells essentially as described previously [21]. Cultures were examined at the time of feeding every 3-4 d for cytopathic effect (cpe). Cultures where cpe was not seen were kept for at least 28 d. Virus growth was confirmed by immunofluorescent staining of acetone-methanol fixed cells for the detection of SIV antigens using a polyclonal antiserum from an SIV infected macaque. A similar procedure was used for back titration of the challenge stock virus, where 10-fold dilutions were added to four replicate flasks of C8166 cells for each dilution tested.

Plasma vRNA
Plasma vRNA concentrations were determined by quantitative competitive (qc) real time reverse transcriptase (RT)-PCR with a cut-off sensitivity of 40 RNA equivalents/ml [22].

Cell Associated Proviral DNA
Levels of proviral SIV gag DNA were determined using a highly sensitive PCR assay. Genomic DNA was extracted from whole blood or lymphoid samples by proteinase K digestion and phenol/chloroform using standard protocols, and the amount included in each PCR determined retrospectively by microfluorometry using a total DNA quantification kit on the basis of calf-thymus DNA standards (Sigma). One microlitre aliquots of DNA extract were assayed, in triplicate, using a Taqman Universal PCR Master mix (Applied Biosystems) with primers and probe located in conserved regions of gag. Tenfold serial dilutions of a p2-LTR plasmid [23], which encompasses the target gag sequences and diluted in 60 lg/ ml herring sperm DNA, were included in each run as controls. Primer sequences were 59-AGTGCCAACAGGCTCAGAAAA-39 (forward, 1253-1273) and 59-TGCGTGAATGCACCA-GATG-39 (reverse, 1304-1322) numbering based on SIV mac239 , at an optimal concentration of 900 nM for each primer. T a q m a n hy d r o l y s i s p r o b e s e q u e n c e 59-( 69-F A M -TTAAAAAGCCTTTATAATACTGTCTGCG-BHQ1)-39 (forward, 1275-1301) was used at a concentration of 225 nM. Amplifications were performed on an Mx3000P genetic analyser (Stratagene) with a thermoprofile of initial HotStart of 95 8C/10 min, followed by 95 8C/30 s, and 60 8C, 1 min/45 cycles. Quantitative SIV DNA levels were calculated using the Mx3000P software, expressed as copies of SIV DNA per 10 5 PBMC/mononuclear cells (MNC). The absolute detection limit of the assay, as determined by Poisson statistics, was a single SIV DNA copy. Similar data were generated on DNA extracted from low-copy number SIV-infected cell preparations where the actual copy number per PCR reaction was determined to be 1.48 copies of SIV gag DNA. Each assay was controlled internally with a spiked DNA preparation and validated by detection of ,three copies of SIV DNA.

Tenofovir Quantification
Plasma samples were assayed for tenofovir concentration by the Clinical Pharmacology and Analytical Chemistry Core of the University of North Carolina Center for AIDS Research.
Drug concentrations in plasma were determined by a validated high pressure, liquid chromatography (HPLC) method with ultraviolet detection [24]. This method utilized a dynamic range of 10-10,000 ng/ml, with intra-and interday variability of ,10% across this range. Total tenofovir concentrations were assayed in tissues using a fully validated HPLC method with mass spectrometry detection (S. Choi, et al., unpublished data). Briefly, internal standard (tolbutamide) was added to tissue prior to complete homogenization using a Fast prep-24 Homogenizer and Lysing Matrix A (MP Biomedicals). Samples were subjected to solid phase extraction on a Varian Bond Elut-C 18 column. Chromatographic separation of TNF and the internal standard (tolbutamide) was achieved with a Varian Polaris 3C 18 -A, 150 mm 3 2 mm, reversed phase analytical column, using a gradient method of 0.1% formic acid in water and 0.1% formic acid in acetonitrile. Detection of TNF and tolbutamide was achieved by electrospray ionization MS in the positive mode using 288.05 and 271.00 m/z, respectively. Linear tenofovir calibration curves were obtained between 1.0 and 1,000.0 ng/ml, with a correlation coefficient (r 2 ) greater than 0.999. Intra-and interday accuracy for tenofovir ranged from 89.7% and 109.4% and from 97.3% and 104.9%, and precision ranged from 1.3% and 10.9% and 2.6% and 9.0%, respectively. Assays were performed with and without incubating tissue with phosphatase, to enable comparison between total parent tenofovir concentrations, and indirect measures of the combination of mono-and diphosphate (active compound) concentrations. Both approaches were validated according to Food and Drug Administration (FDA) standards (http://www. fda.gov/cder/guidance/) with respect to calibration curve performance, recovery, intra-and interday accuracy, and precision of the method, stability of the analytes at various test conditions (determined by TNF and TNF-DP QC performance), evaluation of specificity and matrix effect, and comparisons of differences between fresh and frozen tissue.

In Vitro Assay of Infectivity Inhibition in Indicator Cells and Explant Cultures
Full-length, replication and infection competent proviral HIV-1 clones pYU2 [27,28] and pNL4-3 [29] were provided by the NIH AIDS Research and Reference Reagent Program (http://www.aidsreagent.org/). TZM-bl cells, a CXCR4-positive HeLa cell clone engineered to express CD4 and CCR5 and to contain integrated reporter genes for firefly luciferase and Escherichia coli b-galactosidase under the control of an HIV long terminal repeat sequence, were obtained from the NIH AIDS Research and Reference Reagent Program and were grown at 37 8C in Dulbecco's Minimal Essential Medium containing 10% foetal calf serum, 2 mM L-glutamine, and antibiotics (100 U of penicillin/ml, 100 lg of streptomycin/ml). For quantification of virus infectivity cells were seeded at 3 3 10 3 cells/well 24 h before addition of virus. After incubation with virus for 2 d, the cells were washed with PBS and treated with 100 ll of luciferase cell culture lysis reagent. As previously described [30], 50 ll of each lysate was transferred to a white, opaque assay plate for luciferase quantification in a Synergy HT Multi-Detection Microplate Reader (BioTek Instruments) using 50 ll of luciferase assay reagent (Promega). The extent of luciferase expression was recorded in relative light units (r.l.u.). Virus infectivity inhibition assays were performed using a standardized amount of virus culture supernatants normalized for infectivity. Cells were incubated with serial dilutions of drugs for 1 h at 37 8C, and then virus was added to cells and left for 48 h. Fifty percent inhibitory concentrations (IC 50 ) were calculated from the sigmoidal dose response curves.
Colorectum and ileum-jejenum explants were prepared and cultivated at an air-media interface as described previously for human tissue [31]. Briefly, after two washes and removal of the muscle from the resected tissue, 2-3 mm 3 explants were cut. Explants from nontreated animals were incubated with or without PMPA at 100 lg/ml for 1 h and then incubated with SIVmac 251/32H for 2 h without removal of the drug. Explants from tenofovir-treated animals were incubated directly with virus. After four washes in PBS, the explants were cultivated on gelfoam (Pharmacia & Upjohn) for 15 d with Dulbecco's Minimal Essential Medium (Sigma) containing 10% foetal calf serum, 2 mM L-glutamine, 100 U/ ml penicillin, 100 lg/ml streptomycin, and 80 lg/ml gentamicin. Assays were performed with four replicate wells per sample each well containing three explants. Supernatants were harvested at 3, 7, 11, and 15 d and Gag p27 concentrations were measured by ELISA (p27 antigen ELISA, Zeptometrix Corporation) following the manufacturer's instructions.

Statistical Analysis
The significance of the protection data was determined using the two-tailed Fisher exact probability test by manual calculation. Drug activity titration curves were compared by cross-sectional time series analysis using Stata v10 (Stata Corp). p-values of ,0.05 were considered to be significant.

Comparison of Tenofovir-Mediated Inhibition of HIV-1 and SIV Replication In Vitro
Having shown that tenofovir had no evidence of toxicity on TZM-bl cells, as assessed by MTT viability assay at the highest dose tested (unpublished data), we initially compared the activity of tenofovir, formulated as a solution from PMPA powder or formulated as a gel, as supplied by Gilead Sciences, against a panel of HIV-1 isolates with alternative secondary receptor usage (CXCR4-using RF, IIIB, and NL4.3; CCR5using BaL, YU.2, and R8BaL) and against SIVmac 251/32H ; the latter being used for the macaque challenge experiments. There was no evidence of any difference in the mean average level of inhibition of SIV mac251/32H replication with either formulation at any dose (p ¼ 0.89). This result was determined allowing for the inherent correlation across doses within the same intervention group using cross-sectional time series analysis. Furthermore, tenofovir IC 50 values for SIV mac251/32H were of the same order of magnitude as those measured against both R5 and X4 restricted isolates of HIV (Figure 1).

Dosing and Outcome of Virus Challenge
We used a 1% tenofovir gel formulation made and supplied by Gilead Sciences as part of a programme on the development of a vaginal microbicide. Although this formulation had not been optimised for rectal use, we considered that it would provide a suitable starting point for preclinical evaluation. No adverse effects were seen following rectal gel administration. Virological analysis over a 20-wk period after rectal challenge with 20 MID 50 showed that four of six animals in group A (tenofovir 15 min before challenge) and two of three animals in group D (tenofovir 2 h before challenge) were protected from systemic infection on the basis of failure to (a) recover virus from PBMC, (b) to detect proviral DNA in PBMC, (c) to detect vRNA in plasma, and (d) to detect SIVspecific antibodies in serum. All the naïve/not-dosed animals and three of four animals receiving placebo gel were infected and had high levels of circulating vRNA, and proviral DNA and virus was recoverable from the PBMC at all times of testing ( Figure 2). This protective effect just reached statistical significance at p ¼ 0.05 (Table 2). Furthermore, in comparison with the controls, two of three animals given tenofovir rectally prechallenge that became infected had modified virological outcomes. Virus was recovered from the PBMCs of animal D43 only at weeks two and six corresponding to the only times that vRNA was detected in plasma. Interestingly, the proviral DNA load remained below 50 copies/10 5 PBMC and by week 20, the last point of sampling, had declined to just above the limit of detection. In animal D56 the earliest time of virus detection by any of the methods used was 12 wk after challenge; a delay not seen in other animals in this study or in control animals rectally infected with the same virus stock in previous studies. Taken together, these data showed that tenofovir gel administered at least up to 2 h prior to virus challenge had a significant protective effect (p ¼ 0.003) ( Table 3). Inclusion of another 17 naïve controls that had been challenged intrarectally with the same virus challenge stock [17][18][19][20] revealed a highly significant level of protection from detectable virus infection (p , 0.001) ( Table 4). One of three animals given similar intrarectal dosing 2 h after virus challenge (group E) was protected from infection. Back titration of the virus challenge stock in C8166 cells confirmed infectivity titre had been maintained upon storage.
Of the eight animals that remained VI/PCR negative, seven were clinically normal throughout the study, and one, D68, which had received placebo gel, whilst clinically normal at necropsy had enlarged axillary and inguinal nodes at week 12.
In contrast, all of the animals in which virus was recovered at high frequency had some clinical signs and/or necropsy findings consistent with SIV infection, such as enlarged iliac, axillary and inguinal lymph nodes, and splenomegaly. One animal, D83, had evidence of progression to AIDS including loss of weight, tenting of skin, lung abnormalities including, grey discolouration, numerous petechial haemorrhages on the surface, and adhesions to the thoracic wall. Animal D43, which had been given tenofovir 15 min prior to virus challenge and showed only a weak and transient viraemia, remained clinically normal throughout the study.
To determine if virus was sequestered in tissue associated with the virus challenge, quantitative proviral PCR was used to examine MNC isolated from rectum and ileum, as well as iliac, inguinal, and mesenteric lymph nodes. In the apparently protected animals, there was no evidence of infection whereas in animal E81, a naïve challenged control, proviral loads of 26, 5, 130, 2,100, and 105 proviral DNA copies/10 5 MNC equivalents were detected in each tissue respectively.

T Cell Responses in Protected Animals
To determine if mucosal exposure to virus in the absence of overt infection had stimulated T cell immunity, SIVspecific IFN-c ELISpot analysis was performed on PBMC from protected animals and one infected animal (E81) taken 20 wk after challenge. Four of seven protected animals had IFN-c-secreting Gag-specific T cells at frequencies ranging from 144 to 261 SFU/10 6 cells, whereas the infected animal E81 had both Gag and Tat-specific circulating T cells ( Figure  3A). In contrast, no significant reactivity was seen in 12 naïve macaques tested (,40 SFU/10 6 cells). The finding of T cell responses in the virus exposed macaques was found to be statistically significant; p ¼ 0.009 (Fisher exact test). Interestingly, the presence of Gag-specific IFN-c secreting T cells in animal D68 (pre-exposed to placebo gel) confirmed that challenge virus gained access to antigen presenting cells. Interpretation of T cell reactivity in MNC from colorectal tissue was precluded by a very high spontaneous background; however, MNC from the small intestine of three of four protected animals contained SIV-specific IFN-c-secreting T cells with broader antigen specificity (including specificity for nonstructural virus-encoded antigens) and in two of these cases no responses in PBMC ( Figure 3B). Thus the local mucosal immune system may have been primed by SIV antigens produced de novo during drug-modulated abortive or limited infection. Despite evidence of T cell priming, none of the protected animals had detectable SIVspecific serum antibodies ( Figure 3C).

Plasma Tenofovir Concentration and Protection Status
Analysis of plasma tenofovir concentration at the time of virus challenge, 15 min after gel administration, revealed a strong association with protective efficacy. The lowest concentration of plasma tenofovir associated with protection as defined by failure to isolate and/or detect virus in PBMC and plasma and lack of seroconversion, was 119.9 ng/ml ( Table 5). Taking into account estimated plasma volume, protection was associated with as little as 0.11% of the total tenofovir applied. Moreover, an effect upon plasma viraemia was observed with as little as 0.06% of applied tenofovir detected in plasma at 15 min. In animals given tenofovir 2 h prior virus challenge plasma tenofovir, at the time of challenge ranged between below the 10 ng/ml limit of detection to 23.3 ng/ml. These results suggested therefore that drug concentration peaked rapidly after rectal dosing.

Ex Vivo Modelling of Tenofovir Activity
To further address the possible mechanism of the protection observed we utilised our recently described in vitro colorectal explant model [31] and adapted it to macaque tissue. Using tissues from another group of SIV-naïve macaques (group F), first, we demonstrated that SIVmac 251/ 32H replicated in this system and that replication was sensitive to addition of tenofovir in vitro ( Figure 4A). We also found similar replication in ileum-jejunum explants (unpublished data). Interestingly, replication in macaque explants peaked earlier (day seven) and at a lower level than that seen for HIV in human explants as determined by Gag p27/24 production (C. Herrera, unpublished data and [31]), regardless of virus dose. Next, we investigated the replication kinetics in intestinal tissue explants taken from another four SIV naïve macaques that were given tenofovir gel in vivo 3 h prior to necropsy (group G). In colorectal explants from three of four animals, complete or nearly complete inhibition of virus replication was seen and in the other animal a high level of variability between replicate samples resulted in lower mean inhibition ( Figure 4B). In contrast, inhibition of virus replication was not seen in explants from the small intestine (unpublished data) suggesting that tenofovir was, at least in part, acting on cells at the virus portal of entry. Analysis of intestinal tissue samples collected at necropsy showed that all tenofovir-dosed animals had measurable concentrations of drug in lysates of colorectal tissue at concentrations between 20.8 and 54.2 lg/g protein but no drug was detected in lysates of homogenates from the small intestine (Table 6). Tissues from untreated animals (group F) acted as negative controls. To indirectly estimate the amount of intracellular phosphorylated tenofovir in tissues, samples were analysed with (to measure the combination of tenofovir þ tenofovir monophosphate þ tenofovir diphosphate) and without (to measure tenofovir only) phosphatase hydrolysis. Subtracting the concentration of tenofovir obtained from tissue samples without phosphatase, from the concentration of tenofovir obtained from tissue samples with phosphatase, demonstrated that between 46%-75% of total tenofovir in tissues was present as the intracellular mono-and diphosphate forms. On the basis of intracellular data describing tenofovir monophosphate:diphosphate ratios [32][33][34], it was estimated that approximately 30%-60% of total tenofovir in tissues was present as the intracellular diphosphate form.

Summary of Results
The preclinical study reported here using the SIV-macaque challenge model, showing a statistically significant protective effect, strongly supports the notion that local application of an ARV agent can efficiently protect against subsequent intrarectal challenge with virus. In addition, the study revealed a potential metric of protective efficacy wherein, above a critical threshold, the concentration of tenofovir detectable in the plasma 15 min after rectal application was positively associated with protection. Topical use of tenofovir prior to virus exposure also facilitated the priming of SIVspecific T cell responses in a proportion of macaques, opening up the possibility that this type of prophylaxis may be able to prime and/or boost immune responses elicited with experimental vaccines.

Protection from Infection
It has become increasingly evident that current experimental approaches to the development of an HIV vaccine, largely based upon the generation of T cell immunity, are, at best, likely only to reduce the level of viraemia following exposure to virus. Although this may have partial benefit, in the longer term, as shown in the SIV model, virus frequently escapes from immune control following the selection of variants mutated in T cell epitopes [35,36]. The development of mucosal targeted vaccines, that may be able to completely prevent infection, is a major roadblock to progress and has refocused the field on prevention of infection at the viral portal of entry. The concept of using ARVs topically to prevent infection with HIV is gaining momentum and several clinical trials are currently in progress [37]. The findings reported here showing protection from overt infection in six of nine macaques against rectal transmission of SIV after topical application of tenofovir gel, suggest that this approach has high potential for clinical translation. Interestingly, even in two of three unprotected animals a delay in appearance of viraemia (D56) or a low, transient viraemia (D43) was seen. Although it is not possible to rule out that these effects were due to natural animal-to-animal variability, the VI profiles were highly atypical compared to those observed in our previous studies with this stock of virus [17,18]. We suggest that there may have been low level seeding of virus into a sparse population of cells containing suboptimal concentrations of drug and that the majority of target cells were blocked by tenofovir providing only limited potential for local replication of virus.
The long intracellular half-life of tenofovir diphosphate, the pharmacologically active metabolite of tenofovir [32,33], potentially makes tenofovir an ideal compound for topical use. Indirect analysis to measure total intracellular phosphorylated tenofovir (mono-and diphosphate combined) concentration in rectal tissue, taken from animals dosed in vivo, indicated that up to approximately 60% of the drug had been converted to the active form. It is known that the metabolism of tenofovir varies in different cell types, at least  in vitro [33,34]. Of particular relevance to the present study, it has been reported that in dendritic and Langerhans cells the mono-to diphosphate ratio is approximately 1:3 dependent upon time of sampling [34]. It was encouraging that two of three animals given tenofovir 2 h prior to virus challenge were protected against overt infection. Further experiments are now required to fully determine the longevity of protection and its relationship of tenofovir metabolism in rectal tissue. Another critical consideration is whether protective efficacy can be increased to 100%. Of particular interest in the present study was the observation of a positive association between the concentration of tenofovir measured in the circulation 15 min after topical application and the outcome (C) The group mean þ/À standard deviation profile of anti-SIV Gag p27 binding antibody titres (measured by ELISA) from animals infected with SIV () and the individual profile for an SIV-infected macaque E81 ( * ), in which T cell ELISpot was analysed was compared with animals from which no virus was detected following challenge (m). doi:10.1371/journal.pmed.0050157.g003 of virus challenge. Our working hypothesis is that this metric provides a surrogate measure of local uptake of tenofovir; thus only in animals where the local tissue concentration is beyond a critical threshold is protection achieved. Using the newly developed assay for intracellular tenofovir it should now be possible to test this hypothesis directly. If indeed the rate limiting step to 100% protective efficacy is local uptake of tenofovir, several possibilities require further investigation, including: is efficacy increased (a) at higher concentrations of administered tenofovir gel; (b) with larger volumes of administered gel; (c) with optimised gel formulations; (d) with combinations of ARTs having differing pharmacokinetics; (e) with prior washing to remove residual faecal matter? Thus, the quantification of plasma tenofovir following rectal administration could be used to inform the optimisation of rectal dosing in terms of concentration, volume, formulation, and timing to accelerate the progression of this approach to clinical trials.

Utility of the Ex Vivo Explant System
Our study has also further validated the use of the ex-vivo rectal explant system by, for the first time, extending its use to macaques. Not only is this system useful for the evaluation of antiviral compounds and combinations in vitro; as we have shown here, it may also be used to measure antiviral activity following application of drug in vivo. Detailed pharmacokinetic studies are required to determine when the peak concentration of tenofovir appears in the plasma, but our results from this preliminary study suggest that the peak is early. Interestingly ileum/jejunum tissue taken from dosed macaques remained susceptible to infection. Indeed, the lack of drug in the small intestine was confirmed by analysis of intracellular tenofovir concentration, suggesting that secondary adsorption at this site is insignificant. The relatively low rectal dose of tenofovir applied, equating to an average of 10 lg/kg, of which a maximum of 0.19% was detected in plasma 15 min later, was far below the dose used in oral pre-exposure prophylaxis [16]. Rectal administration of tenofovir gel may be less likely to select for RT escape mutants even when inadvertently used by HIV-infected individuals; however, this possibility requires further investigations to determine the cumulative concentration of drug entering the circulation. Even though, following optimisation, rectal use of tenofovir alone may prove to be sufficient to confer protection, use in combination with other topical ARVs/microbicides may be advantageous. Experiments are in progress in vitro to model synergistic or additive effects of drug combinations in the explant system. Furthermore, we have recently shown that drug combinations can protect colorectal explants against infection with rarely transmitted RT escape mutants (C. Herrera, unpublished data).

Possible Mechanisms of Distal Protection
At the doses used in the present study it is predicted that the protective effects will be confined to the rectal portal of entry; however, the high degree of protection against rectal challenge seen in the present study, including the absence of SIV detection by highly sensitive qPCR in regional lymph nodes was perhaps surprising given that virus may translocate rapidly in dendritic cells [38]. An interesting possibility that requires further investigation is that tenofovir may be reaching local nodes through lymphatic drainage. The interior iliac lymph nodes are known to be a site of early virus replication and have common drainage of the female genital tract and rectum. Indeed, targeting an SIV vaccine preparation to the vicinity of these nodes has conferred significant protection from rectal challenge [39].
Although our study was not powered to allow statistical analysis of rectal postexposure efficacy, in the group of animals receiving gel 2 h after virus exposure only one of three animals was protected. Given that the virus replication cycle is approximately 24 h, this preliminary observation is in accord with the hypothesis that virus is translocated rapidly from the portal of entry and is sequestered either in a cell in which tenofovir is metabolised only slowly and/or that the virus may be held in trans in a nonreplicating state, as has been proposed for dendritic cell-mediated infection [40,41]. Thus, topical application of ARVs such as tenofovir following local viral challenge could be used to further investigate early virus trafficking in vivo, an area of study in which there is still a critical paucity of information.

Priming of Specific T Cells
The priming of SIV-specific T cells detectable in the circulation in the absence of seroconversion is reminiscent of similar responses seen in so-called highly exposed, persistently seronegative individuals (reviewed in [42]). The restriction of significant reactivity to Gag, a structural component of the virion, in the protected animals, suggested the virus inoculum itself, rather than viral proteins synthes- ised de novo, may have been processed by antigen presenting cells either for presentation to CD4 þ T cells via MHC class II or following processing through cross-presentation pathways, for presentation to CD8 þ T cells via MHC class I. It is possible that the tenofovir gel, which had not been formulated for rectal application, resulted in transient tissue damage, as reported for some sexual lubricants [43], resulting in seepage of virus inoculum into the submucosa where local DC could pick up antigen. Nonetheless, the majority of animals were protected from infection in the current study. Alternatively a low level of viral replication may have occurred locally with virus spread curtailed by tenofovir-blockade of secondary target cells. Indeed, transient early postexposure treatment of SIV-infected macaques with tenofovir has been associated with the stimulation of T cell responses, sometimes in the absence of SIV-specific serum binding antibodies [44]. Interestingly, in this instance, CD8 þ lymphocytes were also associated with resistance to rechallenge. If limited replication of virus did occur locally in the present experiment, this may explain the apparent broadening of T cell responsiveness in cells recovered from the intestine to include recognition of virally encoded nonstructural antigens. The latter observation however needs cautious interpretation because of the relatively high level of spontaneous background reactivity. Indeed in the large intestine, spontaneous IFN-c production was so high as to exclude analysis for antigen-specific cells, and although considerably lower in cells recovered from the small intestine, more work using cells recovered from SIVnaïve macaques is required to define a robust cut-off level for specific responses. We have reported previously that single low dose exposure to SIV by the rectal route can induce T and B cell priming in the absence of overt infection and a serum antibody response [25]. It is however, far from clear whether such responses have any antiviral significance or (B) Colorectal explants from four animals (group G: M1, M38, M5, M32) dosed in vivo with tenofovir per rectum 3 h before tissue removal were exposed to virus in vitro (as described above) and culture supernatants assayed for Gag p27. Mean percent inhibition of SIV Gag p27 production plus standard deviations are shown. doi:10.1371/journal.pmed.0050157.g004 whether they represent a ''footprint'' of abortive virus infection. Further detailed investigation is needed to determine if B cells are primed by virus exposure of tenofovirpredosed animals and whether B and T cell responses can be boosted and broadened in specificity upon subsequent exposure to antigen. The apparent longevity of the T cell response seen in the present study was interesting and further work is required to analyse the dynamics of T cell priming in this context. It is possible that the persistence of responsiveness was due to low level boosting by sequestered virus or antigen. Although proviral DNA was undetectable in tissues of protected animals it is not possible to absolutely rule out the possibility of a very low level occult infection as has been described in macaques following intrarectal exposure to low doses of virus (N. Polyanskaya et al., unpublished data) [44,45]. On-going and planned HIVpreventative clinical trials should provide further insight into the generation and nature of T cell responses in the absence of overt infection. This is of importance, as theoretically, although not supported by data on primed, repeatedly exposed macaques (N. Polyanskaya et al., unpublished data), activated CD4 þ T cells at the portal of entry could facilitate or enhance infection by providing a pool of target cells for virus replication. It is imperative that further rechallenge experiments are undertaken in appropriate macaque models before informed decisions can be made regarding advancement of this approach to clinical trial.

Limitations of the Study
The study reported here has several limitations. Firstly, the ARV-gel was delivered controllably under idealised conditions. Ensuring that this occurs in practice will require careful development. Secondly, finding an acceptable form of administration may require different strategies for different target populations. For some users application of a rectal gel may be acceptable, for others, formulation of the ARV into a suppository may be more acceptable. Compatibility with other products that may be used before AI is a further consideration. Thirdly, cell-free challenge virus was applied in a controlled, atraumatic manner in the absence of semen. Fourthly, the predictive value of the SIV-macaque model requires further investigation. This issue was highlighted recently by the much publicized failure of two large clinical trials in HIV prevention (the Merck STEP vaccine trial, and a trial of the vaginal microbicide cellulose sulphate), using modalities that had apparently shown promise in the macaque model. However, in the case of the recombinant replication deficient adenovirus vaccine (STEP trial), reduction in virus load was originally demonstrated following immunization with a construct expressing SIV Gag and challenge with simian HIV (SHIV)-89.6P [46]. This challenge virus has a highly atypical biological phenotype [47], and in a subsequent study with SIVmac239 a marginal effect was only seen in animals expressing the Mamu-A*01 MHC class I allele [48]. Although an enhanced effect was shown in a further study using adenovirus constructs expressing SIV Gag, Rev, Tat, and Nef, this approach was tested only in the more sensitive Mamu-A*01 macaques [49]. Likewise, in the evaluation of cellulose sulphate as a vaginal microbicide, the macaque protection data were not definitive with the majority of treated animals showing evidence of virus infection by PCR [50].

Conclusions
Given the caveats discussed above, our data nonetheless suggest that rectal mucosal dosing with tenofovir shows promise for protection against rectal transmission of an immunodeficiency virus. Moreover, given the functional IFNc T cell responses induced, there may be potential for synergy between topical ARV/microbicide use and vaccination as a two-pronged strategy for preventing infection with HIV. and recombinantly expressed p27 antigen; and to Jim Turpin (NIH) for useful discussion. We thank Alicja Rudnicka (SGUL) for statistical advice and analysis. Author contributions. MC, SS, and RS designed the experiments; SS and MD were responsible for the in vivo work; C. Herrera performed the explant experiments; AC performed the ELISpot experiments; NB, C. Ham, and JH performed the quantitative PCR analyses; NR and AK were responsible for the assay of tenofovir; PA coordinated the overall Microbicide Development Programme; IMcG had oversight of the preclinical testing project; and MC wrote the manuscript.