Durable Protection from Vaginal Simian-Human Immunodeficiency Virus Infection in Macaques by Tenofovir Gel and Its Relationship to Drug Levels in Tissue

Gel formulation.

Tenofovir [(R)-9-(2-phosphonylmethoxypropyl)adenine] (TFV) was kindly provided by Gilead Sciences; 1% (wt/wt) TFV was formulated in 2% hydroxyethyl cellulose (HEC) gel as previously described, with minor modifications (16). Gels were formulated at pH 6.5 to mimic the average vaginal pH of pigtailed macaques (12). A placebo 2% HEC gel was formulated similarly and used as a control. Products were stored at room temperature, and antiviral activity was confirmed throughout the course of the study using a TZM-bl reporter assay to measure drug activity (23).

Virus stock.

SHIV_SF162P3 (simian immunodeficiency virus SIVmac239 backbone with an HIV-1 subtype B, CCR5-tropic envelope) (6, 7) was obtained from the National Institutes of Health (NIH) AIDS Research and Reference Reagent Program and was propagated in PBMCs from pigtailed macaques as previously described (20, 29). The virus stock titer was calculated in pigtailed macaque PBMCs, and virus was diluted to the challenge dose of 10 50% tissue culture infective doses (TCID₅₀) (1.5× 10⁶ RNA copies) and stored separately in 1-ml aliquots in liquid nitrogen. Individual vials were thawed on ice prior to challenge.

Efficacy study using the repeat-exposure macaque model.

We previously reported the efficacy of 1% TFV gel administered 30 min before vaginal SHIV exposures in female pigtailed macaques (21). Macaques were challenged twice weekly for 10 weeks (20 challenges over ∼2 menstrual cycles), and all remained protected. Here we assessed the window of protection conferred by 1% TFV gel using a similar study design. Macaques received a single vaginal application of placebo (n = 1) or 1% TFV (n = 6) gel but were exposed to two SHIV_SF162P3 challenges at 30 min and 3 days after the gel application (Fig. 1A). The once-weekly gel application followed by two virus exposures was repeated for 10 weeks or up to 20 challenges, occurring over two menstrual cycles. All animals were anesthetized and remained recumbent for 30 min after each gel application and 15 min after virus inoculation. Blood was collected after each challenge throughout the study period and twice per week, for 10 weeks, after the last virus challenge. Virus exposures were stopped when a macaque became SHIV RNA positive in plasma. Infected macaques continued to receive weekly gel dosing for a median of 13 weeks (range, 12 to 16 weeks) after infection. All experiments were done under highly controlled conditions by the same personnel using the same virus stock, inoculum dose, and procedures as described in previous studies (21). These studies adhered to the Guide for the Care and Use of Laboratory Animals (18a); all procedures were approved by the Institutional Animal Care and Use Committees (IACUC) of both the Centers for Disease Control and Prevention (CDC) and the Yerkes National Primate Research Center (Emory University).

Menstrual cycle monitoring.

Menstrual cycles in macaques were monitored by plasma progesterone levels. Plasma samples collected twice weekly were shipped to the University of Wisconsin National Primate Research Center and analyzed for progesterone levels using an enzyme immunoassay (EIA) (25).

Infection monitoring by PCR and serology.

SHIV infection was monitored twice weekly by the detection of SHIV RNA in plasma using a reverse transcriptase PCR (RT-PCR) method with a sensitivity of 50 RNA copies per milliliter as previously described (4, 20, 21). Proviral DNA was monitored in peripheral blood mononuclear cells (PBMCs) using primers and probes specific for SIVmac239 pol (20, 21). Serologic testing was done using a synthetic-peptide EIA (Genetic Systems HIV-1/HIV-2 plus O; Bio-Rad). The estimated time of infection was defined as 7 days prior to the first positive RNA RT-PCR in plasma to account for the eclipse period between virus inoculation and detection of SHIV RNA in infected macaques (13). Animals were considered protected if they tested negative for SHIV plasma RNA and SHIV DNA in PBMCs and remained seronegative during the course of the study and the following 10 weeks of washout in the absence of challenge and gel application.

Drug resistance testing.

Drug resistance emergence was monitored in both plasma and cervicovaginal lavage (CVL) fluid from infected animals using standard sequence analysis of the RT-coding region (551 bp; amino acids 52 to 234) and by sensitive allele-specific real-time PCR methods for the K65R mutation as previously described (9, 29). CVL specimens were collected by instilling and recollecting approximately 5 ml of sterile phosphate-buffered saline in the vaginal cavity. CVL fluids were centrifuged for 15 min at 400 × g to pellet cells and debris. Virion-associated RNA was extracted from CVL supernatant (1 ml) using a Qiagen viral RNA kit.

Measurement of drug concentrations in plasma, tissue lymphocytes, and PBMCs.

TFV concentrations in plasma were measured in macaques at 30 min after vaginal administration of TFV gel, resulting in the analysis of at least 10 samples from each macaque. Briefly, TFV was extracted from 100 μl of plasma by protein precipitation with 350 μl of methanol containing 200 ng of ¹³C-labeled TFV as an internal standard. Supernatant containing the drug from precipitation was evaporated to near dryness under vacuum and then resuspended in high-pressure liquid chromatography (HPLC) buffer containing 9.9 mM acetic acid, 5.9 mM ammonium hydroxide, and 9.4 mM formic acid (pH ∼3). Drug levels were analyzed by using liquid chromatography-mass spectrometry (LC-MS) (21, 36). The assay had a lower limit of quantification (LLOQ) of 5 ng/ml and standard curve r² values of greater than 0.99.

For tissue lymphocyte drug levels, SHIV-infected female macaques (n = 10) were sacrificed at 4 h (n = 3), 24 h (n = 2), 2 days (n = 2), and 3 days (n = 3) after receiving a single dose of 1% TFV gel (30 mg). All animals were within the luteal (progesterone-dominant) phase of their menstrual cycle at the time of dosing. Vaginal tissue collected at time of necropsy were dissociated using an enzyme cocktail containing collagenase type II (62.5 U/ml), elastase (0.07 U/ml), hyaluronidase (0.8 U/ml), and DNase I (0.083 U/ml) following procedures kindly provided by Craig Hendrix with minor modifications. Crude cell preparations isolated from tissues were separated using lymphocyte separation medium (LSM) (MP Biomedicals, Aurora, OH) to enrich for mononuclear cells. Total cell populations were gated and counted for lymphocytes using a Guava cell counter with CytoSoft data acquisition and analysis software version 6.0.2 (Millipore, Billerica, MA).

To compare TFV-DP concentrations observed in vivo with those achieved in vitro, macaque PBMCs were incubated with various concentrations of TFV. Intracellular TFV-DP and extracellular TFV concentrations were then plotted in a standard curve. Briefly, PBMCs (5 × 10⁶) were incubated for 24 h in RPMI medium containing TFV concentrations within approximately the range of the 100% inhibitory concentration (IC₁₀₀) to the IC₀ (300, 60, 12, 2.4, 0.48, and 0.096 μM) (28, 31). Additional PBMCs were incubated with 32.8 mM TFV, which corresponds to the TFV concentration in the gel. Cells were washed twice with saline buffer solution and counted using a Guava cell counter.

In addition, we measured intracellular TFV-DP levels in PBMCs following oral administration of tenofovir disoproxil-fumarate (TDF) as previously described (4). Briefly, six macaques each received a single human equivalent oral TDF dose (22 mg/kg), and TFV-DP levels were measured in PBMCs from blood collected at 2, 5, 24, 48, 120, and 168 h.

Intracellular TFV-DP concentrations in lymphocytes were measured with an automated online weak anion-exchange solid-phase extraction method coupled with ion-pair chromatography-MS/MS (4, 11). Briefly, TFV-DP was extracted from lymphocytes using 80% methanol. Lymphocytes were washed with 0.9% NaCl buffer and then resuspended in 1 ml of ice-cold 80% methanol containing ¹³C-labeled TFV-DP as an internal standard. Samples were extracted overnight at −80°C, centrifuged, and dried to completion. Samples were reconstituted in 100 μl of 50 mM ammonium acetate buffer (pH 7.0) and centrifuged at 17,000 × g to remove insoluble particulates. Calibration curves were generated from standards of tenofovir-DP by serial dilutions in blank human PBMC extract over the range from 0.25 to 10 nM. The lower limit of quantification for tenofovir-DP was 10 ng/ml. All calibration curves had r² values of greater than 0.99.

Statistical methods.

Fisher's exact test was used to compare the number of infections per number of months at risk for new infection by study group. Risk for infection per month, or approximately per menstrual cycle, with biweekly challenge doses throughout each month, considers that cycling pigtailed macaques are more susceptible to infection during the luteal phase of the menstrual cycle, as we previously demonstrated in this model (34). Intervention efficacy was calculated as 1-(p₁/p₀), where p₁ and p₀ denote the proportions of months with incident infection for intervention and control animals, respectively (15).

Linear mixed-effects regression models (17) were used to estimate the TFV-DP half-life and 95% confidence limits (CL) based upon observations at and after peak drug absorption and to test for differences in rate of drug elimination between vaginal lymphocytes and systemic PBMCs following topical TFV gel and oral TDF dosing, respectively. A delta approximation method was used to calculate variance for drug half-life (19). The use of this methodology to estimate drug half-life assumes first-order drug PKs. SAS version 9.2.1 was used for all statistical analyses (17).

TFV gel maintains protection at 3 days against challenge with SHIV-162P3.

Figure 1B shows the infection survival rates in animals administered TFV gel at 3 days before SHIV exposure. The infection outcome was compared to that seen in 10 untreated controls receiving a placebo 2% HEC gel application (1 real-time control and 9 historic controls). The historic controls were from a study completed 2 months prior to the start of this work in which control animals received the same placebo gel and were challenged with the same virus stock under the same experimental conditions (21). Figure 1B shows that 9 of 10 controls became infected after 10 weeks, 7 animals (including the real-time control) became infected during the first month of challenges, and 2 additional animals became infected in the following weeks. In contrast, 4 of the 6 macaques receiving 1% TFV gel remained uninfected after 10 weeks of twice-weekly SHIV exposures; 2 treated animals became infected, 1 during the first month and 1 in the following weeks. All infected macaques had detectable proviral DNA and seroconverted at 1 to 3 weeks after first detectable plasma SHIV RNA was observed (data not shown). Risk for infection differed by study group (P = 0.02 by Fisher's exact test). TFV gel efficacy was estimated at 74%, based upon 9/13 infections per number of months at risk among controls and 2/11 among treated animals.

Intracellular half-life of TFV-DP in vaginal tissue lymphocytes.

To better understand the relationship between protection and tissue drug levels, we assessed the decay of intracellular TFV-DP over time in vaginal lymphocytes by administration of a single vaginal gel application to 10 macaques followed by necropsy at specified time points and collection of vaginal tissues. Gel was applied at 4 h (n = 3), 1 day (n = 2), 2 days (n = 2), or 3 days (n = 3) prior to tissue collection. Figure 2 shows TFV-DP concentrations detected in tissue lymphocytes from all 10 macaques. The TFV-DP levels detected in vaginal tissues were highest at 4 h (median, 1,810 fmol/10⁶ cells; range, 1,312 to 2,684) and remained above 1,000 fmol/10⁶ cells within 1 to 2 days (median, 1,107.5 fmol/10⁶ cells; range, 771 to 2,043), demonstrating rapid accumulation and persistence of TFV-DP. However, the median TFV-DP concentrations at 3 days precipitously dropped to 252 fmol/10⁶ cells (range, 196 to 295). We compared the half-life of TFV-DP in vaginal lymphocytes after vaginal gel dosing with that in circulating PBMCs in six pigtail macaques after oral TDF dosing. In lymphocytes extracted from vaginal tissues, the estimated half-life was 25 h (95% CL, 16 to 34). In contrast, TFV-DP persisted much longer in PBMCs (P = 0.02) following oral dosing, with a half-life of 49 h (95% CL, 39 to 60).

Intracellular TFV-DP in PBMCs following in vitro TFV dosing and comparison with in vivo concentrations.

We compared the TFV-DP concentrations observed in vivo with those achieved at 24 h after in vitro dosing. Figure 3 shows a standard curve relating extracellular TFV to intracellular TFV-DP concentrations. TFV-DP levels increased proportionally with extracellular TFV concentrations. Exposure of PBMCs to 2.4 and 60 μM TFV, the in vitro IC₅₀ and IC₉₅ (28, 31), resulted in intracellular TFV-DP concentrations of 109 and 1,373 fmol/10⁶ cells, respectively. Figure 3 also denotes the TFV-DP concentrations found in vaginal lymphocytes collected 4 h (median, 1,810 fmol/10⁶ cells) and 3 days (median, 252 fmol/10⁶ cells) after gel dosing in macaques, which were found to exceed the IC₉₅ and IC₅₀, respectively. Interestingly, the TFV-DP concentrations in PBMCs (79,244 fmol/10⁶ cells) after in vitro dosing at 32.8 mM, the concentration of TFV in 1% TFV gel, were 72-fold higher than those achieved in vivo in vaginal lymphocytes at 24 h after TFV gel dosing (1,107 fmol/10⁶ cells). These data demonstrate a lower in vivo permeability of TFV across the vaginal mucosa.

No evidence of blunted viremia in breakthrough infections.

We next explored the impact of continuous gel dosing on systemic virus replication by maintaining the once-weekly TFV gel applications in the two breakthrough animals. Animals identified as PBz-1 and PWw-1 received an additional 12 and 16 TFV gel applications postinfection, respectively. Figure 4A shows the kinetics of plasma viral replication compared to that in 9 untreated macaques. The peak viremia in the two treated macaques, 6.8 log₁₀ RNA copies/ml and 6.5 log₁₀ RNA copies/ml, was similar to the mean peak viremia of the 9 untreated controls (7.0 ± 0.96 log₁₀ RNA copies/ml), showing no difference in systemic virus levels during acute viremia.

No drug resistance emergence despite continued gel dosing.

Both breakthrough infections were initiated with wild-type viruses. As shown in Fig. 4B, sequence analysis of longitudinal specimens revealed no evidence of K65R emergence in viruses from plasma at the time of infection, peak viremia, or the following 8 to 12 weeks of the washout period. Likewise, despite weekly gel dosing, K65R was not detected by sequencing of SHIV RNA from CVL samples collected periodically up to 14 weeks from the time of initial infection. Additional sensitive testing by allele-specific PCR was also negative for all samples sequenced, thus indicating the absence of low-frequency K65R mutants in both plasma and CVL samples.

Systemic drug absorption following vaginal dosing with 1% TFV gel.

As previously reported, the median detectable TFV levels in plasma at 30 min after vaginal application were low (15.5 ng/ml) relative to median peak levels following oral dosing with TDF (80 ng/ml) (5, 21). Overall, the proportion of measurements with TFV detected in plasma of the 6 treated macaques was 60%, i.e., 36 of a total of 60 measurements (range, 20 to 90%), likely reflecting temporal changes in drug absorption associated with the menstrual cycle (Table 1). Both the infrequent (once-weekly) gel dosing and low systemic drug exposure may likely explain the lack of virus blunting and drug resistance emergence in the breakthrough infections.