The Animal Care and Use Committee (AICUC) of the Centers for Disease Control and Prevention (CDC) approved this study (protocol CDC-AICUC 1414OTTMONC-A5). All animal procedures were done at the animal facilities at CDC. Results reported and analyzed adhere to the predetermined study design and objectives.

The study design is shown in Figure 1. Three drug treatments with different antiviral activity were each given once daily to a group of six rhesus macaques. Group 1 was treated subcutaneously with a human equivalent dose of FTC (20 mg/kg; see next paragraph). Group 2 received oral FTC and TDF (20 mg/kg and 22 mg/kg, respectively) at a dose equivalent to Truvada in humans. Group 3 received subcutaneous FTC (20 mg/kg) and a higher dose of tenofovir (22 mg/kg). Based on approximately double the molecular weight and a 25%–50% tenofovir bioavailability in macaques following oral TDF administration, this dose corresponds to about 66–88 mg/kg of TDF given orally, or 3- to 4-fold the human-equivalent dosing. Intermittent PrEP was given to a fourth group of macaques (group 4) who received subcutaneously the FTC and high-dose tenofovir used in group 3 only 2 h before and 24 h after each weekly virus challenge. A total of 18 control macaques did not receive any drug treatment. Of these macaques, nine were part of this study and nine were historical controls from earlier studies done under identical conditions in Rhesus macaques with the same virus stock, inoculum size, and inoculation protocol.

Oral administration of FTC and TDF was performed by mixing the drug powders with fruit (apples or bananas and infrequently oranges). Stability analysis for up to 24 h by high-pressure liquid chromatography (HPLC) demonstrated that TDF was stable in fresh apple, banana, or orange juice (96% in fruit compared to 98% in water; unpublished data). Macaques were observed to ensure they had taken the drugs. Overall daily drug ingestion was very high; animals missed only one to seven doses during the whole study period. Infected animal AG-81 (see below) missed five doses; one dose 14 d before and four sporadic doses after viral RNA was detected in plasma. Infected animal AI-54 only missed one dose 10 d before the first detectable RNA. Stock solutions of tenofovir and FTC were prepared in distilled water or phosphate-buffered saline, respectively, and injected subcutaneously [21,22].

Tenofovir disoproxil fumarate (9-[(R)-2[[bis[[isopropoxycarbonyl)oxy]-methoxy] phosphinyl]methoxy]propyl]adenine fumarate; TDF), tenofovir ((R)-9-(2-phosphonylmethoxypropyl) adenine; PMPA), and emtricitabine (5-fluoro-1-(2R,5S)-[2-(hydroxymethyl)-1,3-oxathiolan-5-yl]cytosine; FTC) were kindly provided by Gilead Sciences through a material transfer agreement.

The efficacy of different PrEP modalities was evaluated using a repeated exposure macaque model of rectal transmission [12,19]. Rhesus macaques were exposed rectally once weekly for up to 14 wk to a SHIV_SF162P3 chimeric virus that contains the tat, rev, and env coding regions of HIV-1_SF162 in a background of SIVmac239 (National Institutes of Health [NIH] AIDS Research and Reference Reagent Program [23,24]). The SHIV162p3 challenge dose was 10 TCID₅₀ or 7.6 × 10⁵ RNA copies, which is within the range of HIV-1 RNA levels in semen (10³–10⁶ copies/ml) during acute infection in humans and higher than the levels (10²–10⁴ copies/ml) seen after primary viremia [20,25]. Virus exposures were done 2 h after drug treatment by a non-traumatic inoculation of 1 ml of SHIV_SF162P3 into the rectal vault via a sterile gastric feeding tube of adjusted length [19]. Macaques were anesthetized using standard doses of ketamine hydrochloride. Anesthetized macaques remained recumbent for at least 15 min after each intrarectal inoculation. Virus exposures were stopped when a macaque became SHIV RNA positive. All experiments were done under highly controlled conditions by the same personnel using the same virus stock, inoculum dose, and inoculation method.

The four study groups were staggered for logistic feasibility given the long follow-up (6.5 mo) in this model and to avoid unnecessary use of macaques. Group 3, which received the highest dose of FTC/tenofovir, was started before the FTC-only animals in group 1 because if this combination was not protective, testing FTC only would have not been pursued. Similarly, the intermittent FTC/tenofovir arm was done after data were available from the daily interventions. Four protected animals (three from group 3, one from group 1) were used again after a washout period of 8–12 mo; two were used in group 2 and two were used in group 4. Details on all the individual macaques, the regimens received, the sequence and outcome of each series, and the re-use of four animals are shown in Text S1 and Table S1. Blinding of the animal handlers was not done.

Small mammals usually eliminate drugs faster than larger mammals [26,27]. We therefore performed single-dose pharmacokinetic studies in Rhesus macaques to determine the FTC and TDF doses that achieve plasma and intracellular levels comparable to those seen clinically in humans. Six macaques were each given 48, 30, 20, 16, 15, and 12 mg/kg FTC by oral gavage; plasma and intracellular drug levels in peripheral blood mononuclear cells (PBMCs) were determined as described below at 1, 2, 4, 8, and 24 h. The area under the plasma concentration time curve (AUC) values over a 24 h interval increased linearly with the FTC dosing (unpublished data). The dose of 20 mg/kg FTC resulted in an AUC value comparable to that achieved in humans receiving 200 mg FTC (13.2 μg × h/ml and 10 ± 3.1 μg × h/ml, respectively) [17]. Intracellular FTC-triphosphate (FTC-TP) levels in this animal were essentially similar to those in humans [17]. FTC-TP levels were highest at 1 h (1,419 fmol/10⁶ cells), between 1,173–966 fmol/10⁶ cells from 2–8 h, and 273 fmol/10⁶ cells at 24 h. To assess oral bioavailability of FTC in macaques, two additional macaques received a dose of 11 or 17 mg/kg of FTC subcutaneously. Plasma FTC levels (AUC values of 6.1 and 12.2 μg × h/ml, respectively) were comparable to those achieved in the two macaques that received 12 or 16 mg/kg of FTC orally (7.9 and 11.2 μg × h/ml, respectively) indicating approximately 100% absorption. Therefore, we considered a once-daily oral or subcutaneous dose of 20 mg/kg of FTC as comparable to the dosing currently used in humans.

In humans, a wide range of tenofovir exposure is generally seen with once daily dosing of 300 mg of TDF (plasma AUC values of 2.3 ± 0.69 μg × h/ml; range, 2.1–3.2 μg × h/ml) [28]. We performed single-dose pharmacokinetic studies in five macaques that were each given by oral gavage a dose of 15, 20, 24, 37, and 46 mg/kg. Consistent with previous studies, we found a linear relationship between the dose of TDF and plasma AUC values. We also found that a dose of 20 or 24 mg/kg of TDF resulted in plasma tenofovir levels (AUC values of 3.2 and 3.4, respectively) that are within the range of those achieved in humans, while the other doses were outside this range (1.3, 4.1, and 5.8 μg × h/ml for 15, 37, and 46 mg/kg, respectively) [12,13,27,28]. Mean intracellular tenofovir-diphosphate (tenofovir-DP) levels measured between 1–4 h in these two macaques were 107 fmol/10⁶ cells (range, 60–169 fmol/10⁶ cells) and 248 fmol/10⁶ cells (range, 175–335 fmol/10⁶ cells), respectively, also comparable to the levels observed in humans 1–4 h after oral administration of 300 mg TDF (129–373 fmol/10⁶ cells) [16]. We therefore considered a once-daily dose of 22 mg/kg of TDF to be equivalent to the human dose.

Plasma SHIV RNA was quantified using a real-time RT-PCR assay previously described [12]. This assay format has a sensitivity of 50 RNA copies/ml. RNA was extracted from virus pellets obtained by ultracentrifugation of 1 ml plasma. Prior to RNA extraction and to control for the efficiency of extraction, a known amount of virus particles (3 × 10⁵) from an HIV-1 CM240 virus stock was added to each plasma sample. Reverse transcription and PCR amplification of SHIV (gag) and HIV-1 (env) sequences was done using primers specific for SIVmac239 and HIV-1 CM240, respectively [12]. HIV-1 CM240 was obtained from the NIH AIDS Research and Reference Reagent Program. Amplification of proviral DNA from PBMCs was done as previously described using primers and probes specific for SIVmac239 pol [19].

FTC and tenofovir resistance was monitored by standard sequence analysis of SIV RT (551 bp; amino acids 52 to 234) and by sensitive allele-specific real-time PCR methods for the K65R and M184V mutations. Sequence analysis was done from plasma viruses using an RT-PCR procedure as previously described [12]. The Vector NTI program (version 7, 2001) was used to analyze the data and to determine deduced amino-acid sequences. Detection of low-frequency K65R and M184V mutants in plasma by real-time PCR was done as previously described [29]. A similar testing approach is used to assess minority drug-resistant HIV subpopulations in humans [30]. The sensitive SIV assays can detect 0.4% of K65R and 0.6% of M184V cloned sequences in a background of wild-type plasmid.

Virus-specific serologic responses (IgG and IgM) were measured using a synthetic-peptide EIA (Genetic Systems HIV-1/HIV-2; BioRad) assay.

Tenofovir and FTC were extracted from 100 μl plasma by protein precipitation with 400 μl methanol containing 250 nM 3TC as an internal standard. Following precipitation, the solution was maintained at 4 °C for 10 min followed by vortexing for another 10 min. The supernatant was then evaporated to dryness in a heated vacuum centrifuge. Dry samples were stored at −20 °C until liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. Drug levels were analyzed by using HPLC (LC Packings Ultimate 3000 modular LC System; Dionex) and ultra triple quadrupole mass spectrometer (Thermo Electron). Calibration curves were generated from standards of tenofovir and FTC by serial dilutions in human plasma over the range from 3.33 to 100,000 ng/ml. The lower limit of quantification was 3.33 ng/ml. All calibration curves had r² values greater than 0.99.

Tenofovir-DP and FTC-TP were extracted from macaque PBMCs using 60% methanol. Briefly, PBMCs were washed twice with phosphate-buffered saline and then resuspended in 1 ml of ice-cold methanol containing 4 nM 3TC-TP as an internal standard for both tenofovir-DP and FTC-TP. Samples were extracted overnight at −80 °C, centrifuged, and dried under a stream of air. Dried extracts were maintained at −80 °C until analysis. Prior to analysis, each sample was reconstituted in 100 μl 50 mM ammonium acetate buffer (pH 7.0) and centrifuged at 17,000g to remove insoluble particulates. Calibration curves were generated from standards of tenofovir-DP and FTC-TP by serial dilutions in blank human PBM extract (10⁷ cells/100 μl 50 mM ammonium acetate buffer) over the range from 0.25 to 10 nM for tenofovir-DP and 2.5 to 100 nM for FTC-TP. The lower limit of quantification was 0.25 nM for tenofovir-DP and 2.5 nM for FTC-TP. All calibration curves had r² values greater than 0.99. Drug levels were analyzed by HPLC (Agilent 1100 LC system; Agilent Technologies) with mass spectrometry detection (TSQ Quantum; Thermo Electron).

The exact log-rank test was used to ascertain statistical difference between groups of macaques (treatment and control or two treatment groups). The Cox proportional hazards model was used to estimate the relative hazard ratio (HR), for a discrete-time survival analysis of the treatment and control groups, with use of the number of inoculations as the time variable. Though the sample sizes were small, model diagnostics supported the use of the Cox proportional hazards model. All statistical analyses for calculation of the efficacy of the different interventions were performed using SAS software (version 9.1; SAS Institute) and StatXact software (version 6.3; Cytel). For infected macaques, a linear mixed effects model was fit to the log₁₀ of viral load with the macaques as a random effect and group (treatment or control) and time (weeks after peak viral load) as fixed effects.

Figure 2 shows the rate of infection in animals that received PrEP and in 18 untreated control macaques based on the first detectable viral RNA in plasma. Untreated macaques were infected after a median of two rectal exposures (mean, four rectal exposures). This suggests that an animal receiving PrEP and remaining uninfected after 14 virus challenges would have been protected against a median of seven transmission events. The majority of control animals (13/18 or 72%) were infected during the first four challenges; four (22%) were infected between exposures 8 and 12, and only one macaque (6%) remained uninfected after 14 exposures. Proviral DNA was usually detected at the time of plasma RNA detection or within 1–2 wk thereafter. Virus-specific antibody responses were observed within 7 to 35 d (median, 21 d) after the first detectable RNA in plasma. Three control macaques had persistently high viral loads in plasma but remained seronegative. Animals in the PrEP arms were considered protected from systemic SHIV infection if they were seronegative and negative for SHIV plasma RNA and SHIV DNA in PBMCs during PrEP and the following 70 d of washout in the absence of any drug treatment [31].

All PrEP regimens offered some degree of protection. Of the six macaques in group 1 receiving FTC only, two were protected after 14 exposures, whereas four had detectable viral RNA at exposures 5 (AG-80), 10 (AG-46), 12 (AH-04), and 13 (AG-07); infection in these four macaques was delayed compared to untreated controls (p = 0.004, exact log-rank test). These data demonstrate the prophylactic activity of FTC. Infection was also confirmed by DNA-PCR for provirus in PBMC and seroconversion observed 3, 1, 2, and 6 wk after the first detectable RNA, respectively. To compare the temporal probability of risk for SHIV infection in controls relative to treated animals, Cox proportional HRs were calculated. Compared to controls, animals receiving FTC had a 3.8-fold lower risk of infection (Cox HR, 3.8, p = 0.02).

More protection was seen in group 2 (oral FTC and TDF); four of the six macaques were protected and only two (AG-81 and AI-54) were infected, at exposures 9 and 12. Both infections were significantly delayed from controls (p = 0.0004, exact log-rank test). These two macaques were seropositive 2 wk after the first detectable viral RNA in plasma and both were proviral DNA positive at weeks 10 and 12, respectively. Risk of infection in group 2 macaques receiving FTC and TDF was significantly reduced by 7.8-fold compared to controls (Cox HR, 7.8, p = 0.008).

When PrEP fails to completely prevent transmission, the antiretroviral regimen may select for viruses with resistance to the drugs [32]. We therefore continued drug treatment in the six animals that failed PrEP in groups 1 and 2 and followed virus load dynamics and drug resistance emergence. Differences were first noted in acute viremia between breakthrough and control infections. Figure 3 compares virus loads in the breakthrough infections and 10 untreated macaques with sufficient longitudinal data. The mean peak viremia in the six treated macaques was 4.9 ± 0.5 log₁₀ RNA copies/ml, 2.0 log₁₀ lower than in untreated controls (6.9 ± 0.3 log₁₀ RNA; p = 0.007, Mann-Whitney test; Figure 3). This difference was maintained up to week 11 as virus loads declined at similar rates in both groups (−0.24 ± 0.02 log₁₀/wk and −0.24 ± 0.03 log₁₀/wk respectively; p = 0.99). Figure 4 shows that three FTC and one FTC/TDF failures had the first undetectable virus loads as early as weeks 3 (AG-07), 4 (AH-04), 7 (AG-81), and 11 (AG-80) after the peak viremia. All four animals maintained low or undetectable virus loads for up to 20 wk. In contrast, all 10 untreated macaques had detectable virus loads during a median follow-up period of 7 wk (range, 5–40 wk; Figure 3).

Sequence analysis of SIV RT showed that all six breakthrough infections were initiated with wild-type virus. Four animals had no evidence of drug resistance by both sensitive and conventional testing despite extended (median, 23 wk; range, 13–29 wk) treatment. Two animals had M184V (AG-46; FTC-treated) or M184I (AI-54; FTC/TDF-treated) mutations associated with FTC resistance 10 and 3 wk after the first detectable RNA, respectively (Figure 4). The tenofovir-associated K65R mutation was not detected in the two animals receiving FTC/TDF.

In an effort to completely block SHIV transmission, we administered subcutaneously FTC (20 mg/kg) and a higher dose of tenofovir (22 mg/kg) to a third group of six macaques (group 3). All of the six macaques that received this daily PrEP regimen remained uninfected after 14 exposures (p = 0.00005 compared to untreated controls, exact log-rank test; Figure 2). We also assessed if PrEP given intermittently around each virus exposure is equally protective to daily PrEP. A group of six macaques was injected subcutaneously with this highly effective regimen only 2 h before and 24 h after each weekly rectal challenge. All six animals received 14 weekly challenges, and all were found to be protected from infection (Figure 2). None of the animals were seropositive or had detectable viral sequences either in plasma or PBMCs at any of the weekly time points during the 14-wk intermittent PrEP period and up to 70 d thereafter.