Identification and characterization of novel endolysins targeting Gardnerella vaginalis biofilms to treat bacterial vaginosis

Abstract

Bacterial vaginosis (BV) is a recurrent dysbiosis that is frequently associated with preterm birth, increased risk for acquisition of human immunodeficiency virus (HIV) and other sexually transmitted infections (STIs). The overgrowth of a key pathobiont, Gardnerella vaginalis, as a recalcitrant biofilm is central to the development of this dysbiosis. Overgrowth of vaginal biofilms, seeded by initial G. vaginalis colonization, leads to recurrent symptomatic BV which is poorly resolved by classically used antibiotics. In this light, the use of bacteriophages and/or their proteins, represents a promising alternative. Here we identify 84 diverse anti-Gardnerella endolysins across 7 protein families. A subset of 36 endolysin candidates were refactored and overexpressed in an E. coli BL21 (DE3) system and 5 biochemically and structurally diverse endolysins were fully characterized. Each candidate endolysin showed good lytic activity against planktonic G. vaginalis ATCC14018, as well as G. vaginalis clinical isolates. These endolysin candidates were assayed in biofilm prevention and disruption assays, with biofilm disruption at low microgram concentrations (5 μg/ml) observed. In addition to clonal G. vaginalis biofilms, endolysin candidates could also successfully disrupt polyspecies biofilms. Importantly, none of our candidates showed lytic activity against commensal lactobacilli present in the vaginal microbiota such as L. crispatus, L. jensenii, L. gasseri, and L. iners or against Atopobium vaginae (currently classified as Fannyhessa vaginae). The potency and selectivity of these novel endolysins constitute a promising alternative treatment to combat BV, avoiding problems associated with antibiotic resistance, while retaining beneficial commensal bacteria in the vaginal flora. The diverse library of candidates reported here represents a strong repository of endolysins for further preclinical development.

Fig. 1. Phylogenetic analysis of anti-Gardnerella endolysin candidate library.

Individual families are denoted by coloured clades. Selected bootstrap values of >95% are shown for clarity. CCB2 homologues are described previously.

Fig. 2. Endolysin biofilm prevention activity: Biofilm prevention assay using G. vaginalis ATCC14018, G. vaginalis UG860107, and G. vaginalis BUL001 cultures.

Final concentrations of A CCB2M94_8, B CCB7.1, C CCB8.1, D CCB4.1 or E CCB2.2, from 0.39 to 200 μg/ml were used to prevent biofilms. OD₆₀₀ readings are the average of biological triplicates. Controls: MES corresponds to bacteria with MES buffer and control medium is the no bacterial control.

Fig. 3. Antibiotic biofilm prevention activity: Biofilm prevention assays using G. vaginalis ATCC14018, G. vaginalis UG860107 and G. vaginalis BUL001 cultures.

Final concentrations of A metronidazole or B clindamycin from 0.39 to 200 μg/ml, were used to disrupt biofilms. OD₆₀₀ readings are the average of biological triplicates. Controls: MES corresponds to bacteria with MES buffer and control medium is the no bacterial control.

Fig. 4. Endolysin biofilm disruption activity: Biofilm disruption assay using G. vaginalis ATCC14018, G. vaginalis UG860107 and G. vaginalis BUL001 biofilms.

Final concentrations of A CCB2M94_8, B CCB7.1, C CCB8.1, D CCB2.2 or E CCB4.1, from 5 to 200 μg/ml were used to disrupt biofilms. OD₆₀₀ readings are the average of biological triplicates plus/minus their standard deviation. The asterisks in figures indicate levels of significance. p-values < 0.05 are represented by *, p-values < 0.01 are represented by **, p-values < 0.001 are represented by ***, and p-values < 0.0001 are represented by ****. No statistical differences are represented by ns. Controls: MES corresponds to bacteria with MES buffer and control medium is the no bacterial control. p-values for ANOVA and Tukey tests.

Fig. 5. Antibiotic biofilm disruption activity: Pre- formed biofilm disruption assays using G. vaginalis ATCC14018, G. vaginalis UG860107 and G. vaginalis BUL001 biofilms.

Final concentrations of A metronidazole or B clindamycin from 5 to 200 μg/ml, were used to disrupt biofilms. OD₆₀₀ readings are the average of biological triplicates plus/minus their standard deviation. No statistical differences are represented by ns. Controls: MES corresponds to bacteria with MES buffer and control medium is the no bacterial control. p-values for ANOVA and Tukey tests.

Fig. 6. Synergistic activity: Biofilm disruption assay using G. vaginalis ATCC14018, G. vaginalis UG860107 and G. vaginalis BUL001 pre-established biofilms.

Final concentrations of A CCB2M94_8, B CCB7.1, C CCB8.1, D CCB2.2 or E CCB4.1 ranged from 2.5 to 50 μg/ml, were used to disrupt biofilms. OD₆₀₀ readings are the average of biological triplicates plus/minus their standard deviation. The asterisks in figures indicate levels of significance. p-values < 0.05 are represented by *, p-values < 0.01 are represented by **, p-values < 0.001 are represented by *** and p-values < 0.0001 are represented by ****. No statistical differences are represented by ns. Controls: MES corresponds to bacteria with MES buffer and control medium is the no bacterial control. p-values for ANOVA and Tukey tests.

Fig. 7. Polymicrobial biofilm disruption.

A Biofilm disruption assay using G. vaginalis ATCC14018 and A. vaginae UG71161 polymicrobial biofilms. The different endolysin candidates, metronidazole and clindamycin were used in a final concentration of 200 μg/ml. B Biofilm disruption assay using A. vaginae UG71161 biofilms. The different endolysin candidates, metronidazole and clindamycin were used in a final concentration of 200 μg/ml. OD₆₀₀ readings are the average of biological triplicates plus/minus their standard deviation. The asterisks in figures indicate levels of significance. p-values < 0.05 are represented by *, p-values < 0.01 are represented by **, p-values < 0.001 are represented by *** and p-values < 0.0001 are represented by ****. No statistical differences are represented by ns. p-values for ANOVA and Tukey tests.

Fig. 8. Proposed combined endolysin antibiotic therapeutic: The presence of recalcitrant biofilms typically mean that front line antibiotics are infective when treating G. vaginalis associated BV (right).

Here we suggest that the lytic activity of phage endolysins leads to biofilm disruption and increased activity of front-line antibiotics (left) potentially leading to greater effective clearance and lower recurrence rates. Figure created using BioRender.