Preclinical validation of electrospun fibers to achieve vaginal colonization by Lactobacillus crispatus

Abstract

Communities of bacteria collectively known as the vaginal microbiota reside in the human vagina. Bacterial vaginosis (BV) describes an imbalance of this microbiota, affecting more than 25% of women worldwide, and is linked to health problems such as infertility, cervical cancer, and preterm birth. Following antibiotic treatment, BV becomes recurrent in many individuals. Lactobacillus crispatus is widely believed to contribute to a healthy vaginal microbiome, and its therapeutic application has shown promise in early clinical trials investigating adjunct therapies for lasting treatment of conditions such as BV. There is a pressing need for therapeutic platforms that apply biologically active agents such as probiotic bacteria, to the vagina with little user effort but lasting effect. Here, we use a mouse model to investigate the functional utility and potential harms of soft, slow-dissolving fibers made by electrospinning polyethylene oxide (PEO) and poly(lactic-co-glycolic acid) (PLGA). Blank electrospun fibers that passed quality control checkpoints were administered vaginally in a murine model and compared to animals receiving mock procedures. Fiber administration had no significant effects on mucus glycan markers, vaginal epithelial exfoliation, keratinization, tissue edema or neutrophil infiltration. L. crispatus-loaded fibers enabled L. crispatus colonization in most animals for more than one week. Mice receiving L. crispatus-loaded fibers had significantly higher measured concentrations of lactate in vaginal washes at 48 hrs compared to pre-colonization washes. These data provide pre-clinical proof of concept that vaginal administration of electrospun fibers can achieve viable delivery and vaginal colonization by metabolically active L. crispatus, without eliciting inflammation or injury.

Keywords: Bacterial vaginosis; Lactobacillus crispatus; electrospun fibers; histopathology; lactate; vaginal colonization.

Figure 1.. Histopathology of vaginal and cervical tissues, 24 h after insertion of electrospun fibers made of PEO-PLGA in β-estradiol-valerate-treated C57Bl/6J mice, compared to a group that received the mock procedure.

A) Experimental schematic of the initial in vivo experiment. B) Heat map depicting average histological scores for each group of mice. Keratin and edema were scored on a scale of 0 to 2, where polymorphonuclear cells (PMNs) were scored on a scale of 0 to 4. C) Scoring method used to score histological samples. Observers were blinded to the experimental groups. These experiments were only powered to observe phenotypes with a large effect size.

Figure 2.. Epithelial exfoliation and mucus sialic acid content were indistinguishable in mice receiving vaginal electrospun fibers.

A) Experimental schematic for data in Figs 2–4. B) Vaginal epithelial cells were enumerated in 72 h vaginal washes using phase contrast microscopy of mouse vaginal washes. An average of 5 fields was used for each data point. C) Representative images of two mouse vaginal washes under phase contrast. D) Sialic acid (N-acetylneuraminic acid) concentrations, a proxy for intact mucosal secretions in mouse vaginal washes, were determined by high performance liquid chromatography as described in the methods. Sialic acid concentrations are measured in picomoles per 10 μL of vaginal wash. The Mann Whitney U-test did not detect statistically significant differences in epithelial exfoliation or mucus sialic acid content between animals receiving mock-administrations versus electrospun fibers.

Figure 3.. Scoring of histological images of the vaginal epithelium..

Representative images of scores are shown for each histological metric as defined in Figure 1C. Keratin and edema scale bars represent 200 μm, while neutrophil scale bars represent 100 μm. Black arrows indicate presence of keratin, green arrows indicate presence of edema, and red arrows indicate presence of polymorphonuclear cells (PMNs).

Figure 4.. Vaginal and cervical histopathology in mock- and electrospun fiber-treated mice.

A) Images of vaginal tissue from mice from the mock group and electrospun fiber group. Both show similar levels of keratinization, edema, and neutrophil infiltration. B-E) Blinded scoring of histology metrics was performed based on the rubric described in Figure 1C. The Mann Whitney U-test showedno significant differences between experimental groups for the vaginal or cervical tissues. Less keratinization of the cervix was more likely in the electrospun fiber group, but not statistically significant(p=0.0534 using nonparametric Mann-Whitney analysis). These data represent a total of n=40 mice, with n=10 per group in two independent experiments.

Figure 5.. L. crispatus-loaded PEO:PLGA fibers initiate colonization of the mouse vagina.

A) Vaginal wash titers of streptomycin-resistant L. crispatus strain MV-1A-US over the course of 7 days from mice treated with L. crispatus-loaded PEO:PLGA fibers (10⁵-10⁶ CFU/fiber). Statistical analysis assumed nonparametric data and matched data (Friedman’s test) reflecting the same subjects over time with Dunns post-hoc test to enable multiple comparisons. B) Lactic acid concentration in mouse vaginal washes before and after treatment with PEO:PLGA fibers. The Friedman test was used to compare each timepoint to a control timepoint (T0), comparing them as matched pairs from each individual animal C) Categorical analysis comparing the proportion of animals at each time point with undetectable versus measurable levels of lactate in vaginal washes, according to data in panel B. Results of the Fisher’s Exact test are shown. D) Correlation between lactate concentration and Lactobacillus crispatus titers in vaginal washes for all data points collected (n=100 paired datapoints, including T0). There was a weak correlation between L. crispatus titers and lactate concentration (p=0.0103, r=.2693) using Spearman correlation coefficient. DPI: Days Post Inoculation *P<0.05 **P<0.01***P<0.001