Susceptibility to genital herpes as a biomarker predictive of increased HIV risk: expansion of a murine model of microbicide safety

Abstract

Background: A crucial gap in the development of microbicides for HIV prevention is the absence of models predictive of safety. Previous studies have demonstrated an increased susceptibility to genital herpes in mice following repeated applications of nonoxynol-9 (N-9). This study was designed to explore the underlying mechanisms, focusing on the effects that N-9 has on genital tract epithelium and to apply this expanded model to evaluate the safety of microbicides that have been advanced to clinical trials.

Methods: Mice were treated intravaginally with formulated 3.5% N-9, 1% tenofovir, 0.5% or 2% PRO 2000, hydroxyethylcellulose (HEC) placebo or no treatment and the effect on herpes simplex virus 2 (HSV-2) susceptibility, epithelial cell architecture, junctional proteins and inflammation were assessed.

Results: Mice treated with seven daily doses of N-9, but not tenofovir, PRO 2000 or HEC, were significantly more susceptible to challenge with low doses of HSV-2; confocal microscopy demonstrated increased numbers of viral particles deep within the genital tract. N-9 disrupted the epithelium with loss of tight and adherens junctional proteins. By contrast, the epithelium was relatively preserved following tenofovir, PRO 2000 and HEC exposure. Additionally, N-9, but not the other microbicides, triggered a significant inflammatory response relative to untreated mice.

Conclusions: These findings indicate that disruption of the epithelium contributes to increased HSV-2 susceptibility and might provide a biomarker predictive of increased risk for HIV acquisition. The results are consistent with the safety outcomes of the recently completed Phase IIb clinical trial with 0.5% PRO 2000 gel, and predict that tenofovir gel will not adversely affect the genital tract.

Figure 1. N-9, but not tenofovir or PRO 2000, increases the susceptibility to HSV-2 infection

Mice were challenged with 10³, 10⁴ or 10⁵ plaque-forming units (PFU)/mouse of herpes simplex virus 2 strain G (HSV-2[G]) (A) at 12 h after receiving the seventh daily dose of nonoxynol-9 (N-9), tenofovir, 0.5% PRO 2000, 2% PRO 2000 or hydroxyethylcellulose (HEC) or (B) after being left untreated and were observed daily for 15 days for signs and symptoms of disease. Symptoms were scored on a 0–5 point scale: 0, no apparent infection; 1, slight redness of the vagina; 2, moderate redness and swelling of the vagina and surrounding tissue; 3, severe redness and swelling; 4, genital ulceration or hair loss of genital and surrounding tissue; and 5, evidence of hind-limb paralysis. Mice reaching stage 4 or 5 disease were euthanized. Results show the percentage of survival pooled from at least two independent experiments (n=20 mice/group). ^aP<0.05. To assess whether changes in susceptibility were associated with disruption of the epithelial barrier, confocal images of extracted lower genital tract tissue were examined. Mice were infected with HSV-2(G) at 12 h after the seventh daily gel application and were then sacrificed 4 h after infection. Tissues were stained for viral capsids (green), nectin-1 (red) and nuclei (blue). (C) Three-dimensional reconstruction representative of images taken from at least four animals per treatment group. At least six independent randomly selected images were acquired per mouse and each grid mark represents 10 μm.

Figure 2. N-9 causes disruption of the epithelium with loss of junctional proteins

Mice were treated daily for 7 days with microbicides and at 12 h after the final application, the mice were sacrificed and the entire vaginal canal up to the uterine bifurcation was excised. (A & B) The tissue was then fixed and stained with EZ-link sulfo-NHS-biotin (EZ-link; Pierce, Rockford, IL, USA) to detect the apical surface (magenta; [A] only) 4′,6′-diamidino-2-phenylindole nucleic acid stain to detect nuclei (blue), anti-zonula occludens protein 1 (ZO-1; green) and desmoglein (red) and viewed by confocal microscopy. Representative three-dimensional (upper panel) images are shown. Images were taken from at least six animals per treatment group and at least six independent randomly selected images were acquired per animal. Pixel intensity results were obtained for successive μm thick axial sections from the apical to the basal membrane for Z0-1 and desmoglein-1 and are shown in the graphs below each panel set. (C) After seven daily drug applications, gene expression of E-cadherin, ZO-1, desmoglein and nectin-1 were measured from RNA extracted from genital tract tissue by quantitative competitive real-time PCR. Results represent the β-actin normalized values compared with untreated samples (n= at least five mice/group in at least two independent experiments) and are presented as log₁₀ change in expression (mean ± SE). ^aP<0.05. No change in E-cadherin expression was observed in response to hydroxyethylcellulose (HEC) or tenofovir. N-9, nonoxynol-9.

Figure 2. N-9 causes disruption of the epithelium with loss of junctional proteins

Mice were treated daily for 7 days with microbicides and at 12 h after the final application, the mice were sacrificed and the entire vaginal canal up to the uterine bifurcation was excised. (A & B) The tissue was then fixed and stained with EZ-link sulfo-NHS-biotin (EZ-link; Pierce, Rockford, IL, USA) to detect the apical surface (magenta; [A] only) 4′,6′-diamidino-2-phenylindole nucleic acid stain to detect nuclei (blue), anti-zonula occludens protein 1 (ZO-1; green) and desmoglein (red) and viewed by confocal microscopy. Representative three-dimensional (upper panel) images are shown. Images were taken from at least six animals per treatment group and at least six independent randomly selected images were acquired per animal. Pixel intensity results were obtained for successive μm thick axial sections from the apical to the basal membrane for Z0-1 and desmoglein-1 and are shown in the graphs below each panel set. (C) After seven daily drug applications, gene expression of E-cadherin, ZO-1, desmoglein and nectin-1 were measured from RNA extracted from genital tract tissue by quantitative competitive real-time PCR. Results represent the β-actin normalized values compared with untreated samples (n= at least five mice/group in at least two independent experiments) and are presented as log₁₀ change in expression (mean ± SE). ^aP<0.05. No change in E-cadherin expression was observed in response to hydroxyethylcellulose (HEC) or tenofovir. N-9, nonoxynol-9.

Figure 3. Effect of microbicides on SLPI and defensin gene expression

After (A) three or (B) seven daily applications of each formulated microbicide, gene expression of β-defensin (BD)1, BD2 and secretory leukocyte protease inhibitor (SLPI) were measured from vaginal tissue by quantitative real-time PCR. Results represent the β-actin normalized values compared with untreated mice (n= at least five mice/group in at least two independent experiments) and are presented as log₁₀ change in gene expression (mean ± SE). ^aP<0.05. There was no change in gene expression observed for BD1 or BD2 in response to hydroxyethylcellulose (HEC) or tenofovir. N-9, nonoxynol-9.

Figure 4. Changes in gene expression and protein levels of cytokines and chemokines in response to microbicides

Levels of messenger RNA were quantified from tissue on (A) day 3 and (B) day 7 by quantitative real-time PCR. Results represent β-actin-normalized values relative to untreated samples (n= at least five mice/group in at least two independent experiments) and are presented as the log₁₀ change in gene expression (mean ± SE). Vaginal washes pooled from at least five mice were collected before microbicide application in normal saline on days 3, 7, 14 and 21 from mice treated daily for 14 days with each microbicide formulation or untreated control mice. The levels of chemokines and cytokines were measured in the vaginal lavage by BioLuminex (Austin, TX, USA). The sensitivities for each analyte (in pg/ml) were as follows: (C) monocyte chemotactic protein (MCP)-1=0.95, (D) tumour necrosis factor (TNF)-α=0.42, (E) macrophage inflammatory protein (MIP)-2=2.2 and (F) interleukin (IL)-1β=3.3. In addition, the sensitivity for IL-6 was 0.71. Results for the detectable mediators (n = at least five mice/group in at least two independent experiments) are presented as mean ± SE. No protein was detected in the lavage samples for which no bar is observed. ^aP<0.05. CCL5, C–C motif ligand 5; CXCL11, C-X-C motif ligand 11; HEC, hydroxyethylcellulose; IP-10, interferon-inducible protein 10; N-9, nonoxynol-9; RANTES, regulated upon activation, normal T-cell expressed and secreted protein.

Figure 5. N-9, but not tenofovir, activates NF-κB

Nuclear levels of nuclear factor (NF)-κB (p65) were determined by ELISA performed on nuclear extracts prepared from harvested genital tract tissue. Results are shown as mean ± SE values using tissue from at least five mice/group from at least two replicate experiments. ^aP<0.05 relative to untreated control mice. HEC, hydroxyethylcellulose; N-9, nonoxynol-9; RLU, relative luciferase units.