Antibody and Antiretroviral Preexposure Prophylaxis Prevent Cervicovaginal HIV-1 Infection in a Transgenic Mouse Model

Abstract

The development of an effective vaccine preventing HIV-1 infection remains elusive. Thus, the development of novel approaches capable of preventing HIV-1 transmission is of paramount importance. However, this is partly hindered by the lack of an easily accessible small-animal model to rapidly measure viral entry. Here, we report the generation of a human CD4- and human CCR5-expressing transgenic luciferase reporter mouse that facilitates measurement of peritoneal and genitomucosal HIV-1 pseudovirus entry in vivo. We show that antibodies and antiretrovirals mediate preexposure protection in this mouse model and that the serum antibody concentration required for protection from cervicovaginal infection is comparable to that required to protect macaques. Our results suggest that this system represents a model for the preclinical evaluation of prophylactic or vaccine candidates. It further supports the idea that broadly neutralizing antibodies should be evaluated for use as preexposure prophylaxis in clinical trials.

INTRODUCTION

Thirty years after first reports of patients presenting with symptoms of human immunodeficiency virus (HIV) infection (1), AIDS remains one of the leading global health problems (2). While the RV144 trial demonstrated the modest efficacy of the RV144 vaccine in a community risk vaccination setting (3, 4), a highly protective vaccine remains elusive.

Recent studies have shown that antiretroviral therapy (ART) of HIV type 1 (HIV-1)-infected patients can significantly reduce the rate of HIV-1 transmission (5) and decrease the risk of infection when used in seronegative probands (6–8). Other trials, however, failed to demonstrate protective effects for antiretroviral medication, which may be attributable to low drug adherence (9, 10), indicating the need for further research on optimized administration strategies.

Broadly neutralizing antibodies (bnAbs) directed against the envelope protein (Env) of HIV-1 may present an additional option for HIV-1 preexposure prophylaxis. Passive administration of neutralizing antibodies to macaques can provide sterilizing immunity to simian-human immunodeficiency virus (SHIV) infection (11–15) and protect from HIV-1 infection in humanized rodent models (16–20). Many of these studies were, however, performed with earlier-generation neutralizing antibodies. Recently introduced methods for isolation and cloning of monoclonal antibodies have led to the production of human antibodies with increased neutralizing breadth and potency (21–27), which could prove beneficial for clinical use, including therapy (28).

A variety of small-animal models have been developed to study different aspects of HIV-1 infection in vivo, including models with several strains of immunodeficient mice reconstituted with components of the human immune system (reviewed in references 29 to 31). These mice are susceptible to HIV-1 infection, and they can recapitulate the viral life cycle and show long-term viremia. However, infections in individual mice vary, depending on the amount of engraftment, and the mice lack a fully functional immune system, showing only sporadic humoral immune responses to HIV-1 (32–34). The use of transgenic (35, 36) or adenovirus-mediated (19) expression of HIV-1 entry factors represents an alternative approach to rendering rodents susceptible to viral entry but not to efficient replication (37, 38). Mice and rats expressing human CD4 (hCD4) and human CCR5 (hCCR5) were infected with HIV-1 after intravenous injection with HIV-1; however, this route accounts for only a minority of infections in humans (39).

Here, we describe a transgenic reporter mouse expressing hCD4 and hCCR5 under the control of the ubiquitin promoter, allowing in vivo detection of HIV-1 pseudovirus infection after intraperitoneal (i.p.) and intravaginal (i.vag.) application by means of bioluminescence.

MATERIALS AND METHODS

Mice.

Gt(ROSA)26Sor^tm1(Luc)Kael/J (ROSA-Stop-Luc) mice and Gt(ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J (ROSA-Stop-tdTomato) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). hCD4/hCCR5 transgenic mice were generated by cloning the hCCR5-2A-hCD4 open reading frame (19) downstream of the human ubiquitin C promoter (40) and injecting a linearized 4.5-kb fragment containing the transgene into fertilized female C57BL/6J mouse pronuclei. A founder line was chosen on the basis of surface expression of hCD4 and hCCR5 on peripheral blood mononuclear cells (PBMCs), as determined by flow cytometry (data not shown). HIV-Luc_TG mice were produced by breeding hCD4/hCCR5 transgenic mice to ROSA-Stop-Luc mice and backcrossing to ROSA-Stop-Luc mice to obtain mice homozygous for firefly luciferase and were screened by PCR genotyping or were produced by breeding mice homozygous for the transgene and luciferase to ROSA-Stop-Luc mice. Mice homozygous for hCD4 and hCCR5 were obtained by breeding HIV-Luc_TG mice and selected on the basis of surface expression levels, as determined by flow cytometry. Transgene-negative littermates or ROSA-Stop-Luc mice served as negative controls, as indicated. Mice were bred and maintained at the Comparative Bioscience Center at The Rockefeller University according to guidelines established by the Institutional Animal Committee. Experiments were performed with authorization from the Institutional Review Board (IRB) and the IACUC at The Rockefeller University.

Pseudovirus production.

HIV-1 pseudoviruses were prepared as described previously (19). Briefly, pTripCre, HIV gag-pol, and pSVIIIenv_YU-2 (41) or pSVIIIenv_JR-FL (42) plasmids were cotransfected in HEK293T cells using X-tremeGENE 9 reagent (Roche) according to the manufacturer's instructions at a 1.42:1.00:1.68 ratio. Supernatants were harvested, cleared of debris by centrifugation at 300 × g, filtered using a 0.45-μm-pore-size filter (Thermo Scientific), concentrated by stirred-cell ultrafiltration using a 300,000-nominal-molecular-weight-limit membrane (Millipore), and stored at −80°C. Viruses pseudotyped with the envelope protein G of the vesicular stomatitis virus (VSVg) were prepared by cotransfection of pTripCre, HIV gag-pol, and pVSVg (43) plasmids at a 3.18:2.24:1.00 ratio either in HEK293T cells using X-tremeGENE 9 and adjusted to a final concentration of 4 μg/ml Polybrene (Sigma) and 50 mM HEPES (Gibco) after harvesting or in HEK293-6E cells using polyethyleneimine (PEI) at a 1.5:1.0 ratio of PEI/DNA and adding sodium butyrate to a final concentration of 5 mM after 12 h. Supernatants were harvested, cleared of debris by centrifugation at 300 × g, filtered using a 0.45-μm-pore-size filter, and stored at −80°C. The p24 content was determined by enzyme-linked immunosorbent assay (ELISA; PerkinElmer) according to the manufacturer's instructions. The 50% tissue culture infective dose (TCID₅₀) was determined using the mouse embryonic fibroblast (MEF)-based assay described below and calculated according to the Reed-Muench method (44). The mean luminescence of replicates of uninfected cells × 1.5 was used as the negative cutoff.

Antibody production.

Anti-VSVg antibodies were purified from supernatants of the I1 hybridoma cell line (ATCC), maintained as per the manufacturer's instructions, using protein G-Sepharose 4 Fast Flow (GE Healthcare). A mouse IgG2a antibody (clone UPC-10, Sigma) dialyzed against phosphate-buffered saline (PBS) was used as an isotype control. Anti-HIV-1 antibodies (24, 45) and the human IgG1 isotype control mGO53 (46) were expressed in HEK293T cells and purified using protein G-Sepharose 4 Fast Flow as described previously (19, 47). Purified antibodies were dialyzed against PBS and sterile filtered through a 0.22-μm-pore-size filter unit.

Cells.

MEFs from HIV-Luc_TG and ROSA-Stop-Luc mice were produced by spontaneous immortalization. The placenta, head, and internal organs were thoroughly removed from individual fetuses placed in ice-cold PBS. Fetuses were minced, incubated in 0.025% trypsin-EDTA (Gibco) at 37°C, homogenized by thorough pipetting, filtered using a cell strainer, and washed twice, and cells were maintained in MEF medium (RPMI 1640 supplemented with 1× antibiotic-antimycotic, 2 mM l-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 55 μM β-mercaptoethanol [all from Gibco], 10% fetal bovine serum [FBS; HyClone]). hCD4 and hCCR5 expression was confirmed by flow cytometry. HEK293T cells were maintained in 293T medium (Dulbecco's modified Eagle medium [DMEM; Gibco] supplemented with 1 mM sodium pyruvate, 1× antibiotic-antimycotic, and 10% FBS) and cultured in DMEM supplemented with 1 mM sodium pyruvate and 3% FBS after transfection for pseudovirus production. HEK293-6E cells were maintained in FreeStyle expression medium supplemented with 0.2× penicillin-streptomycin (both from Gibco) on an orbital shaker and cultured in FreeStyle medium without penicillin-streptomycin after transfection. Human healthy control PBMCs were used in accordance with the IRB protocols of The Rockefeller University.

Antiretrovirals.

For use in in vivo experiments, clinical formulations of maraviroc (Pfizer) and efavirenz (Bristol-Myers Squibb) were thoroughly ground using a mortar and pestle, resuspended in H₂O, and stored at −20°C. For use in in vitro experiments, maraviroc and efavirenz were acquired through the NIH AIDS Research and Reference Reagent Program, resuspended in dimethyl sulfoxide (DMSO), and stored at −20°C. A clinical formulation of enfuvirtide (Roche) was resuspended in sterile H₂O, stored at −20°C, and used for in vivo and in vitro experiments.

Flow cytometry.

Whole blood, healthy human PBMC controls, and MEFs were used to assess expression of hCD4 and hCCR5. MEFs were harvested using enzyme-free dissociation buffer (Gibco). The following antibodies were used for stainings: anti-human CCR5-fluorescein isothiocyanate (FITC) and CD4-allophycocyanin (APC) (BD Pharmingen). Stainings were performed in the presence of mouse FcBlock (2.4G2; Bio X Cell). To determine cells expressing tdTomato, peritoneal cells were acquired by lavage with ice-cold PBS and stained with anti-mouse F4/80-APC and B220-FITC (eBioscience) in the presence of FcBlock. Cells were acquired on a BD LSRFortessa cell analyzer (BD Biosciences) and analyzed with FlowJo software (TreeStar).

In vitro MEF assay.

MEFs were harvested using enzyme-free dissociation buffer (Gibco) and plated in a 24-well plate at a density of 5 × 10⁴ cells/well in MEF medium. Serial dilutions of pseudoviruses were added after 8 to 12 h and incubated for 1 day at 37°C. Pseudoviral supernatants were replaced by fresh MEF medium, and cells were incubated at 37°C for another day. Medium was removed, cells were lysed in 100 μl ONE-Glo luciferase assay reagent (Promega) and 100 μl MEF medium for 5 min, and the lysates were thoroughly homogenized by pipetting. A volume of 150 μl of each sample was transferred to a white 96-well plate, and luminescence was measured using a FLUOstar Omega reader (BMG Labtech). Uninfected cells served as background controls. In some experiments, serial dilutions of antiretroviral drugs were added at the step of pseudovirus infection and DMSO carrier controls were performed (data not shown).

In vivo infections.

For vaginal challenge, mice were pretreated by subcutaneous (s.c.) injection of 3 mg of depot medroxyprogesterone acetate (DMPA; Depo-Provera; Pfizer) 7 days prior to intravaginal application of 40 μl of a commercially available 4.0% nonoxynol-9 (N-9) gel formulation (Conceptrol) 6 h prior to the first viral challenge. A volume of 40 μl of pseudovirus was carefully pipetted into the vagina of isoflurane-anesthetized mice, which were held in an inverted position for 5 min after application to allow adsorption of virus. Pseudovirus application was repeated after 2 and 4 h. For peritoneal challenge, mice were injected in both lower abdominal quadrants. Monoclonal antibodies were injected s.c. 1 day prior to virus challenge. For ART experiments, efavirenz and maraviroc were applied orally twice by gavaging 0.5 mg or 1.5 mg, respectively, 24 h prior to and at the time of viral challenge; a dose of 40 μg of enfuvirtide was applied s.c. every 12 h, starting 24 h prior to vaginal challenge and continuing to up to 12 h after viral challenge. For anti-VSVg experiments, virus was preincubated on ice for 30 min with anti-VSVg antibody I1 or an isotype control at a final concentration of 1 μg/ml each prior to injection. Control medium for VSVg-Cre was DMEM supplemented with 4 μg/ml Polybrene, 50 mM HEPES, and 3% FBS.

Serum antibody ELISA.

The determination of the antibody concentrations in serum was performed as described previously (19). ELISA plates (Corning) were coated with goat anti-human IgG (Jackson ImmunoResearch Laboratories) at 2.5 μg/ml overnight. Plates were blocked with 2 mM EDTA and 0.05% Tween 20 in PBS for 1 h at room temperature. Serial dilutions of mouse serum in PBS were incubated and detected with a horseradish peroxidase-conjugated goat anti-human IgG (Jackson ImmunoResearch Laboratories) at a 1:1,000 dilution in blocking buffer. Samples were subsequently developed with ABTS [2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid); Invitrogen]. Purified human IgG was included to generate the standard curve.

Animal luciferase assay.

Mice were anesthetized using isoflurane. In mice that received pseudovirus i.p., 150 μl of a 30-mg/ml solution of d-luciferin potassium salt (Regis Technologies) was injected i.p., and peak luminescence was acquired by serial measurements; in mice that received pseudovirus i.vag., 20 μl of the d-luciferin reagent was carefully applied i.vag., and luminescence was acquired immediately thereafter. Luminescence was acquired using an IVIS Lumina II apparatus with an exposure time of 60 s, and regions of interest (ROIs) were analyzed using Living Image software (both from Caliper Life Sciences).

Ex vivo analysis.

For analysis of omental tissues, mice were injected i.p. with 4.5 mg d-luciferin reagent as described above. Mice were sacrificed after 10 min, omenta were dissected and placed into a black 96-well plate, and luminescence was acquired using the IVIS Lumina II apparatus with an exposure time of 120 s.

Statistical analysis.

Statistical analysis was performed using Prism (version 5) software for the Mac OS X operating system (GraphPad Software). Percent infection was calculated using the arithmetic means of non-log-transformed absolute photon counts per second. A two-tailed Mann-Whitney U test was used to compare groups, and tests were not adjusted for multiple comparisons. Data are presented as means ± standard errors of the means (SEMs).

RESULTS

Mice carrying a conditionally transcribed firefly luciferase gene in the ROSA26 locus, regulated by a loxP-flanked transcriptional stop element (ROSA-Stop-Luc) (48), can be used to detect Cre recombinase (Cre) activity by photon emission (48, 49). To test whether ROSA-Stop-Luc mice can be used to measure viral infection by HIV-based viruses, we produced Cre-encoding replication-deficient viruses pseudotyped with the envelope protein G of the vesicular stomatitis virus (VSVg-Cre). Viruses pseudotyped with VSVg are pantropic and can infect a broad range of mammalian target cells (50). Bioluminescence was readily detected in ROSA-Stop-Luc mice after i.p. injection of VSVg-Cre (75 ng p24), while control mice did not show any specific luciferase activity (Fig. 1A and B). The signal in the peritoneal cavity showed a major hot spot at the site of the omentum that peaked 6 days after injection and that had a signal of 1.5 orders of magnitude above that for the control (Fig. 1B; P = 0.0159 at day 6). In addition, bioluminescence could be detected in the mediastinal region, the site of the primary draining lymph nodes of the peritoneal cavity (51); however, the bioluminescence was at a much lower intensity and had higher levels of variability than in the omentum (data not shown). Cre activity results from viral entry mediated by VSVg, as preincubation of VSVg-Cre with the neutralizing anti-VSVg antibody I1 (52) results in nearly complete blockage of infection (P < 0.001), while preincubation with an isotype control did not reduce infection (P = 0.281) (Fig. 1C). In addition, these results were reflected by bioluminescence in ex vivo tissue samples of the omentum (Fig. 1C) and cells harvested by peritoneal lavage (data not shown). We conclude that HIV-based pseudoviruses encoding Cre can be used to detect viral infection in vivo and ex vivo.

Fig 1.

Intraperitoneal injection of VSVg-Cre (75 ng p24) into ROSA-Stop-Luc mice results in bioluminescence in the omentum. (A) Representative luminescence 6 days after i.p. injection of control medium or VSVg-Cre. The unmarked signal in lower left abdomen depicts one site of injection. (B) Time course of photon flux per second for omental ROI in mice i.p. injected with VSVg-Cre (n = 5) or control medium (n = 4). †, off-scale value (3.27). (C) (Left) Preincubation with anti-VSVg antibody I1 (total n = 7) blocks VSVg-Cre infection measured 4 days (d4) after i.p. injection by an omental ROI compared to the result for isotype control mice (total n = 8) and untreated mice (total n = 7), used to define 100% infection (I1 and isotype controls, respectively; final concentration, 1 μg/ml). Data were pooled from two experiments. For each experiment, 0% infection was defined by luminescence of uninfected mice (n = 1 to 2). Symbols depict different individual experiments. (Right) Photon flux per second for omenta dissected after i.p. injection of d-luciferin 6 days (d6) after VSVg-Cre injection (n = 4 per group). Dashed line, mean of naive mice; †, negative off-scale values (for I1 mice, −944.7; for naive mice; −6,075); p/s/cm²/sr, photon flux per second per square centimeter per steradian; Ab, antibody.

To overcome host restriction for HIV-1 entry, we produced transgenic mice that carry human CD4 (hCD4), the primary receptor for HIV-1 (53, 54), and human CCR5 (hCCR5), the most commonly used coreceptor in initial HIV-1 infection (55–57). Coexpression of hCCR5 and hCD4 was achieved by linking them on a single polyprotein transcript separated by a ribosomal skip 2A peptide sequence (19, 58) driven by the human ubiquitin C promoter (pUP-R plasmid [40]) (Fig. 2A). A founder line was chosen on the basis of cell surface expression of hCCR5 and hCD4 and bred to ROSA-Stop-Luc mice to obtain hCCR5⁺ hCD4⁺ ROSA26^Luc/Luc mice (termed HIV-Luc_TG mice). While expression was considerably lower than that on human control PBMCs, hCCR5 and hCD4 could be readily detected on HIV-Luc_TG bulk PBMCs (Fig. 2B) and on subsets of splenocytes, including T and B cells, dendritic cells, and monocytes/macrophages, as well as on subsets of peritoneal cells, including B cells and macrophages (data not shown).

Fig 2.

HIV-Luc_TG mice express hCCR5 and hCD4. (A) Construct used to produce transgenic mice expressing hCCR5 and hCD4 under the control of the human ubiquitin C (UbC) promoter. SV40, simian virus 40; BGH, bovine growth hormone. (B) Representative fluorescence-activated cell sorter analysis of hCCR5 and hCD4 expression on PBMCs of an HIV-Luc_TG mouse, a transgene-negative littermate, and a human healthy control.

Replication-deficient pseudoviruses are frequently used to assay viral infection and antibody neutralization in vitro. For example, the single-round-infection TZM-bl assay (59) measures luciferase or β-galactosidase activity driven by the viral Tat protein expressed after infection of target cells. To facilitate quantification of infectious particles of HIV-1-enveloped and Cre-encoding but tat-negative pseudoviruses (19), we produced MEFs expressing hCCR5 and hCD4 from HIV-Luc_TG mice (Fig. 3A). Infection with VSVg-Cre, as quantified by luciferase activity, was readily detected in MEFs generated from both ROSA-Stop-Luc and HIV-Luc_TG mice (Fig. 3B). In contrast, viruses pseudotyped with the HIV-1 envelope protein gp160 of YU2, a clade B and difficult-to-neutralize tier 2 HIV-1 strain (YU2-Cre) (60), were able to infect only HIV-Luc_TG MEFs (Fig. 3B). Antiretroviral drugs, which interfere with distinct steps of the viral life cycle, are the cornerstone of HIV therapy (61, 62). They include efavirenz, a nonnucleoside reverse transcriptase inhibitor (NNRTI) (63); maraviroc, an hCCR5 antagonist (64); and enfuvirtide, a peptide fusion inhibitor (65). As expected, infection with YU2-Cre was inhibited in the presence of all three antiretrovirals tested in our MEF-based in vitro assay, with 50% inhibitory concentrations (IC₅₀s) being in the nanomolar range (IC₅₀ of efavirenz, 1.101 nM; IC₅₀ of maraviroc, 0.662 nM; IC₅₀ of enfuvirtide, 5.004 nM), while VSVg-Cre infection was inhibited only by the NNRTI efavirenz (Fig. 3C). We conclude that HIV-Luc_TG MEFs resemble TZM-bl cells, in that they can be used to titrate viral preparations of Cre-encoding HIV-1 pseudoviruses in vitro, and that the HIV-Luc_TG MEF assay can be used to assess the effects of antiretroviral drugs on HIV-1 pseudovirus infection in vitro.

Fig 3.

MEFs produced from HIV-Luc_TG mice can be used for in vitro analysis of HIV-1 pseudovirus infection. (A) hCCR5 and hCD4 expression on MEFs from an HIV-Luc_TG mouse (solid line) and a ROSA-Stop-Luc mouse (tinted region) determined by flow cytometry. (B) Luminescence after infection of HIV-Luc_TG and ROSA-Stop-Luc MEFs with serial dilutions of preparations of YU2-Cre or VSVg-Cre. Infection of HIV-Luc_TG MEFs was performed in quadruplicate; infection of ROSA-Stop-Luc MEFs was performed in duplicate. Results of a representative experiment for different viral preparations are presented. (C) Luminescence after infection of HIV-Luc_TG MEFs with YU2-Cre (13 TCID₅₀s; red) or VSVg-Cre (35 TCID₅₀s; blue) in the presence of different antiretrovirals. Dashed lines, luminescence in the absence of antiretrovirals (red, YU2-Cre; blue, VSVg-Cre) and for uninfected MEFs (black, YU2-Cre experiments; gray, VSVg-Cre experiments). Infections were performed in duplicate. IC₅₀s were calculated using a nonlinear regression curve. n.a., not applicable.

Next, we sought to determine whether HIV-Luc_TG mice can be used to detect HIV-1 pseudovirus infection in vivo after i.p. injection. Omental luciferase activity after i.p. injection of 585 TCID₅₀s of YU2-Cre increased longitudinally in HIV-Luc_TG mice, peaking at between days 5 and 8 after injection with a mean signal of ≈0.7 to 0.8 order of magnitude (arithmetic and geometric mean) above that for the hCCR5-negative (hCCR5⁻)/hCD4-negative (hCD4⁻) littermates, which remained unaffected (Fig. 4A and B). To determine which cells become infected after i.p. injection of HIV-1 pseudoviruses, we bred HIV-Luc_TG mice to mice harboring a loxP-flanked tdTomato gene in the ROSA26 locus (66). Peritoneal lavage performed 6 days after injection of YU2-Cre revealed infection to be relatively inefficient, with nearly all of the infected cells being of a macrophage-like phenotype (F4/80-positive B220-negative), as determined by flow cytometry (Fig. 4C). To test whether HIV-Luc_TG mice can be used as a model for preexposure prophylaxis, mice were pretreated with antiretroviral drugs or the broadly neutralizing anti-CD4-binding-site antibody 3BNC117 (24), and dissected omenta were analyzed 8 (YU2-Cre) or 6 (VSVg-Cre) days after i.p. injection of the respective virus. Pretreatment with efavirenz (0.5 mg twice per os [p.o.]), maraviroc (1.5 mg twice p.o.), or enfuvirtide (40 μg four times s.c.), starting 1 day prior to viral challenge, significantly reduced infection with YU2-Cre (585 TCID₅₀s; P = 0.0159), while only the NNRTI efavirenz showed a significant effect on VSVg-Cre infection (2,190 TCID₅₀s; P = 0.0286) (Fig. 4D). In addition, while s.c. injection of 20 μg 3BNC117 1 day prior to challenge with YU2-Cre failed to show protection (P = 0.6905), injection of 200 μg 3BNC117, while not reaching a level of statistical significance, showed clear protective effects in three of five treated mice compared to the effect of the isotype control mGO53 (46) (P = 0.0952) (Fig. 4E). We conclude that HIV-Luc_TG mice can serve as a tool to rapidly monitor HIV-1 entry after i.p. injection, primarily by infection of macrophage-like cells, and that infection can be blocked by antiretrovirals and neutralizing antibodies.

Fig 4.

HIV-Luc_TG mice become infected with HIV-1 pseudovirus after i.p. injection. (A) Representative luminescence 8 days after i.p. injection. (B) Time course of changes in photon flux per second for an omental ROI for HIV-Luc_TG mice and transgene-negative littermates injected i.p. with YU2-Cre (585 TCID₅₀s) compared to the luminescence of uninjected mice (left) and photon flux per second 8 days after injection (right). Pooled data from two experiments depicted by individual symbols are presented (total n starting on day 3, 7 to 8 [one mouse died during the course of the experiment] for HIV-Luc_TG mice, 8 for negative mice, and 2 for uninjected mice). (C) Detection of tdTomato-positive (tdTomato⁺) cells in peritoneal lavage fluid obtained 6 days after i.p. injection of YU2-Cre (585 TCID₅₀s) in hCCR5⁺ hCD4⁺ ROSA26^tdTomato/Luc mice or transgene-negative littermates. Representative tdTomato expression found in different cellular subsets of an hCCR5⁺/hCD4⁺ mouse is shown. (Left) Gating strategy for cells pregated on the basis of forward scatter and sideward scatter (SSC); (right) mean frequency (n = 3 per group) of tdTomato-positive cells in cellular subsets of hCCR5⁺/hCD4⁺ mice and negative littermates. (D) Ex vivo luminescence analysis of dissected omenta after i.p. d-luciferin injection. Analysis was performed 8 days after i.p. YU2-Cre injection (585 TCID₅₀s) in HIV-Luc_TG mice (left) and 6 days after i.p. VSVg-Cre injection (2,190 TCID₅₀s) in ROSA-Stop-Luc mice (right). Mice were pretreated with antiretroviral compounds starting 24 h prior to viral challenge (efavirenz, 0.5 mg twice p.o.; maraviroc, 1.5 mg twice p.o.; enfuvirtide, 40 μg four times s.c.) or untreated. The log differences in the arithmetic means of photon flux per second defining 0% and 100% infection are 1.2 (YU2) and 3.0 (VSVg). (E) Ex vivo luminescence analysis of dissected omenta after i.p. d-luciferin injection. Analysis was performed 8 days after i.p. YU2-Cre (585 TCID₅₀s) injection in HIV-Luc_TG mice. Mice were s.c. pretreated with 3BNC117 or 200 μg mGO53 24 h prior to viral challenge. The log difference in the arithmetic mean of photon flux per second defining 0% and 100% infection is 0.74. For panels D and E, untreated or isotype-treated mice were used to define 100% infection, respectively, and naive mice were used to define 0% infection (n ≥ 4 for virus-injected groups, n = 2 to 3 for naive mice).

Cervicovaginal exposure accounts for the majority of global HIV-1 infections (39). To establish whether HIV-Luc_TG mice can be used to monitor genital mucosal infection, we adapted protocols used in macaques and mice and pretreated mice with 3 mg of DMPA (13, 14, 67–70). Seven days later, a volume of 40 μl of a 4% N-9 gel formulation (71, 72) was applied to the vagina, 6 h prior to three vaginal challenges with 160 TCID₅₀s of YU2-Cre every 2 h. In vivo imaging revealed a luminescent signal in HIV-Luc_TG mice that was not seen in controls, peaking at an ≈0.7 order of magnitude (arithmetic and geometric means) above that for the controls at between 3 and 5 days after application, and that was not affected by pretreatment with mGO53 24 h prior to viral challenge (Fig. 5A and B). Similar results were obtained when mice homozygous for hCCR5 and hCD4 were i.vag. subjected to viruses pseudotyped with the JR-FL gp160 (MEF titer, >1:270), an additional clade B tier 2 isolate (73), highlighting the flexibility of the system for infection by different HIV-1 Env pseudotypes (Fig. 5C). To determine whether HIV-Luc_TG mice can be used to examine antiretrovirals or anti-HIV-1 antibodies for prophylaxis, we administered maraviroc, 3BNC117, or the broadly neutralizing antibody 3BC176, which targets a not yet precisely defined conformational Env epitope (45), before viral challenge. Pretreatment with maraviroc (1.5 mg twice p.o.) nearly completely blocked infection with YU2-Cre (P = 0.0286), while infection with VSVg-Cre (three challenges with 175 TCID₅₀s) was not significantly reduced (P = 0.8182) (Fig. 5D). Subcutaneous injection of 3BNC117 1 day prior to virus application led to a dose-dependent reduction of the level of YU2-Cre infection compared to that after injection of mGO53 (Fig. 5E). While an injected dose of 2 μg of 3BNC117 failed to prevent YU2-Cre infection (P = 0.3706), a dose of 20 μg reduced infection by a mean of 49% (P = 0.025) or 75%, when one unprotected outlier is excluded, and a dose of 200 μg led to complete protection (≥95% reduction of infection) (P = 0.0002). Compared to the infection in PBS-treated mice, we found a reduction of YU2-Cre infection by a mean of 71% after s.c. pretreatment with as little as 2 μg of 3BC176 1 day prior to viral challenge (P = 0.0286), corresponding to a serum concentration of 0.5 μg/ml at the time of viral challenge, and complete protection (≥96% reduction of infection) after pretreatment with 200 μg (P = 0.0286) (Fig. 5F). We conclude that HIV-Luc_TG mice can be used to assay cervicovaginal infection of HIV-1 pseudoviruses in vivo and to measure prophylaxis mediated by antiretroviral drugs and neutralizing antibodies.

Fig 5.

HIV-Luc_TG mice become infected with HIV-1 pseudoviruses after intravaginal application. (A) Representative luminescence in HIV-Luc_TG mice and transgene-negative littermates 5 days after i.vag. YU2-Cre application (three times with 160 TCID₅₀s). (B) Timeline for experiments (left; d, day; Tx, treatment) and time course of photon flux per second for vaginal ROI after i.vag. application of YU2-Cre (three times with 160 TCID₅₀s each time) (middle) in PBS-treated HIV-Luc_TG (n = 6) mice and transgene-negative littermates (n = 4). Dashed line, luminescence 5 days after application in ROSA-Stop-Luc mice (n = 3). (Right) Pooled luminescence 5 days (d5) after infection for PBS-treated (total n = 17) or mGO53-treated (200 μg s.c., total n = 10) HIV-Luc_TG mice and transgene-negative littermates (total n = 6). Symbols depict individual experiments (n = 7). (C) Photon flux per second 5 days after i.vag. application of JR-FL-Cre (MEF titer > 1:270; three times with 40 μl each time) and YU2-Cre (three times with 160 TCID₅₀s each time) in mice homozygous for hCCR5 and hCD4 compared to transgene-negative littermates. Pooled data from two experiments are presented (total n for hCCR5⁺/hCD4⁺ mice, ≥5; total n for negative mice, 2). (D) Infection in mice pretreated with maraviroc (1.5 mg twice p.o.) or controls 5 days after i.vag. application of pseudovirus. (Left) YU2-Cre infection (three times with 160 TCID₅₀s each time) in HIV-Luc_TG mice (n = 4). mGO53-pretreated mice (200 μg s.c., n = 4) (red bar) and ROSA-Stop-Luc mice (n = 3; see panel B) from the same experiment were used to define 100% and 0% infection, respectively. (Right) Total photon flux per second in ROSA-Stop-Luc mice after VSVg-Cre application (three times with 175 TCID₅₀s each time). Pooled data from two individual experiments are presented (total n = 6 per group). (E) Infection 5 days after intravaginal application of YU2-Cre (three times with 160 TCID₅₀s each time) in HIV-Luc_TG mice s.c. pretreated with 3BNC117 (for 2 μg, total n = 3; for 20 μg, total n = 7; for 200 μg, total n = 6) or 200 μg mGO53 (total n = 10; see panel B) 24 h prior to viral challenge. Pooled mGO53-treated mice and pooled transgene-negative littermates (total n = 6; see panel B) were used to define 100% and 0% infection, respectively. Pooled data from a total of five experiments depicted by individual symbols are presented. (F) Infection 5 days after intravaginal application of YU2-Cre (three times with 160 TCID₅₀s each time) in HIV-Luc_TG mice s.c. pretreated with 3BC176 or PBS (n = 4 per group) 24 h prior to viral challenge. PBS-treated mice and pooled transgene-negative littermates (total n = 6; see panel B) were used to define 100% and 0% infection, respectively. Numbers show human IgG serum concentration (n = 5 per group) in μg/ml at the time of viral challenge, and SEMs are in parentheses.

DISCUSSION

Although significant progress in HIV-1 prevention has been made, further research is required to establish an efficient vaccine or optimized pharmaceutical prevention strategies. While studies in nonhuman primates are the current “gold standard” for late-stage preclinical studies, the limited number of animals available and the cost of these experiments make them unsuitable for large-scale early preclinical studies.

Several small-animal models have been developed to allow assessment of HIV-1 infection. Humanized mice partially reconstituted with cellular components of the human immune system allow HIV-1 infection, including mucosal transmission, viral replication, and long-term viremia resembling the clinical situation (29–31). However, these mice fail to mount a robust immune response to HIV-1, rendering them unsuitable for vaccination studies. In addition, the need for reconstitution can lead to differences between individual mice. Immunocompetent transgenic rodent models have been established, but to date their use has been restricted to extramucosal routes of infection (35, 36, 74), which account for a minority of human HIV-1 infections (39).

HIV-Luc_TG mice and cells obtained from these mice allow rapid in vivo and in vitro detection of infection with HIV-1 pseudoviruses. These mice differ from humanized mice, in that they should be immunologically intact and could potentially be used for vaccine studies. They also differ from previously reported transgenic models, in that they can be used to assay infection by mucosal exposure. We have shown that HIV-Luc_TG mice can be used to detect infection and prophylaxis mediated by antiretrovirals and anti-HIV-1 antibodies after both intraperitoneal and cervicovaginal application of HIV-1 pseudoviruses.

In this study, we show that passive transfer of neutralizing anti-HIV-1 antibodies can protect HIV-Luc_TG mice from cervicovaginal infection with HIV-1 pseudoviruses. In previous experiments, we have determined an in vitro IC₅₀ of 0.054 μg/ml for the highly active agonistic anti-CD4-binding-site antibody (HAAD) 3BNC117 against YU2 (24) and measured the 3BNC117 serum concentration in ROSA-Stop-Luc mice 1 day after subcutaneous injection (2-μg dose, 0.4 ± 0.03 μg/ml; 20-μg dose, 5.0 ± 0.6 μg/ml; 200-μg dose, 24.3 ± 1.8 μg/ml [mean ± SEM, n = 3]) (19). Here, 3BNC117 reduces intravaginal infection by ≈50 to 75% after administration of 20 μg, corresponding to a serum concentration of ≈5 μg/ml at the time of viral challenge (19), which is roughly 100-fold higher than the in vitro IC₅₀ determined by the TZM-bl assay. These results are in good agreement with those of previous macaque studies of vaginal SHIV challenge, indicating that 50% protection in vivo requires serum antibody concentrations to be 100-fold or more of those needed for 50% neutralization in vitro (14, 69, 75). Pretreatment with 3BC176, targeting a conformational epitope found on cell surface-expressed gp160 (45), led to an ≈85% reduction of infection at a serum concentration of ≈6.8 μg/ml, only 23-fold higher than the IC₅₀ against YU2 determined in vitro (0.29 μg/ml) (45). Interestingly, a significantly reduced infection (≈70%) could also be detected in the presence of a serum concentration of as little as 0.53 μg/ml of 3BC176. These results are consistent with those of work in macaques showing that antibody-mediated protection from vaginal SHIV challenge can be achieved at relatively low titers (76, 77). However, it stands in contrast to the findings of antibody therapy experiments in YU2-infected humanized mice, where 3BC176 was ineffective at a serum concentration of up to more than 100 μg/ml (28). Thus, the requirements for prevention and therapy of HIV-1 infection by antibodies may differ significantly.

Establishing infection with HIV-1 pseudoviruses after vaginal application required pretreatment with both DMPA and N-9, which may alter the mucosal epithelia and milieu in a way that does not reflect the human cervicovaginal anatomy and its role in HIV-1 transmission. Progesterone is frequently used in vaginal infection models to synchronize estrous cycles and to increase susceptibility by thinning out the vaginal layer, including models of simian immunodeficiency virus/SHIV infection in nonhuman primates (13, 14, 67, 69, 70, 78) and mouse models of HIV-1 (68, 79), human papillomavirus (HPV) (80), and herpes simplex virus 2 (HSV-2) (81) infection. Increased susceptibility to viral infection after mucosal N-9 application has been described in models of HPV infection (80) and both vaginal and rectal HSV-2 infection (72, 82, 83). In addition, an increased frequency of HIV-1 infection has been found in female sex workers after application of high doses of N-9 (84). Lesions in cervical and vaginal epithelia described after application of N-9 are one likely contributor to these observations (71, 72), and induction of proinflammatory conditions with an influx of inflammatory cells may increase the number of potential target cells for HIV-1 infection (72, 85). While the hCCR5 antagonist maraviroc, which shows only minor inhibitory activity against murine CCR5 (86), may have reduced a murine chemokine-mediated influx of cells possibly targeted by pseudoviruses in our mouse model (87), in contrast to infection with YU2-Cre, infection with VSVg-Cre was not markedly reduced in the presence of maraviroc.

Expression of hCCR5 and hCD4 in HIV-Luc_TG mice is driven by the ubiquitin promoter, rendering virtually all murine cell types potential target cells for HIV-1 pseudoviruses. While both entry factors could be detected on a number of splenocyte and peritoneal cell subsets, different cell types may show different susceptibilities to both pseudovirus infection and protection mediated by antiretroviral compounds and antibodies. In addition, expression of entry factors in HIV-Luc_TG mice is markedly lower than that on human CD4⁺ and CCR5⁺ cells, likely leading to lower rates of entry factor-dependent infection in HIV-Luc_TG mice (88). Fluorescent analysis of individual cells revealed that murine macrophage-like cells are the main target of HIV-1 pseudoviruses after intraperitoneal challenge. While hCCR5-expressing macrophages are considered to be among the target cells for primary HIV-1 infection of R5-tropic HIV-1 in humans, it remains unclear how murine macrophages compare to their human counterpart. Further studies will be required to determine whether specific cells are targeted after vaginal application of HIV-1 pseudoviruses in HIV-Luc_TG mice and if they can serve as a model system for rectal transmission of HIV-1. With a signal approximately 0.7 order of magnitude above that for the controls for both i.p. and i.vag. infection, the dynamic range of our system is relatively narrow, and there is some considerable variation in the amount of infection between individual animals, which may result in limited sensitivity for the detection of minor differences between individual antibodies or doses. More importantly, while our study focused on the use of passively transferred, broadly neutralizing, and highly potent antibodies for antibody-mediated protection, vaccination strategies so far have not resulted in the generation of antibodies with comparable breadth and potency, and the dynamic range of our system may not be sufficient to detect the protective effects of antibodies with more limited potency.

While protection in macaque or humanized mouse studies is usually defined by the absence of viral RNA in peripheral blood, HIV-Luc_TG mice do not support efficient viral replication (35, 37), and therefore, this model is limited to studies involving the entry phase of the viral life cycle. Despite these caveats, HIV-Luc_TG mice can serve as a valuable tool for preclinical assessment of preexposure prophylaxis in HIV-1 infection. They should be suitable for vaccination studies, and, as shown in this study, they can be readily used to assess and compare the protective effects of both small-molecule antiretroviral drugs and potent anti-HIV-1 neutralizing antibodies.