The proteinaceous biofilm of Gardnerella vaginalis enables a novel enzymatic therapy for bacterial vaginosis

Abstract

Since it was first reported in the 1950s, bacterial vaginosis (BV) has become a globally vital concern among women of childbearing age. Gardnerella vaginalis is widely recognized as the primary causative agent responsible for BV development. G. vaginalis has a strong tendency to form biofilms which have been linked to widespread antimicrobial tolerance and recurrent or persistent BV episodes. Our study demonstrated that proteins constitute more than 50% of the G. vaginalis biofilm matrix, which significantly protects the bacterium from degradation by lysozyme, and that LasA, an elastase derived from Pseudomonas aeruginosa, effectively disrupts G. vaginalis biofilms and subsequently lyses the bacterial cell wall, leading to cell death. Four candidate biofilm-associated proteins of G. vaginalis were identified using cross-linking mass spectrometry (XL-MS) and subsequently confirmed as LasA substrates through purification and LasA digestion. Peptidoglycan debris was observed after treating the G. vaginalis cell wall extract with LasA. We also found that LasA showed a minimal adverse effect on lactobacilli strains when used in vitro. In vivo studies utilizing murine models artificially infected with G. vaginalis further demonstrated that a single dose of LasA effectively reduces G. vaginalis colonization while exerting a negligible adverse effect on lactobacilli populations. The safety of LasA was further supported by the fact that no negative effects were observed on the treated mice's vaginal tissue sections during the post-treatment administration period.

1. Introduction

Bacterial vaginosis (BV) is a complex polymicrobial vaginal infection occurring in women of reproductive age. Patients with BV may experience abnormal vaginal discharge, odour and irritation [[1], [2], [3], [4]]. Approximately 50% of patients with BV are symptomatic and experience itching, whereas the others are asymptomatic [4,5]. Furthermore, BV is characterised by an increase in opportunistic bacteria, such as Gardnerella vaginalis, and a decrease in lactobacilli [2,3,5,6]. Studies have shown that the recurrence rate of BV is high and that it can increase the risk of sexually transmitted diseases (STDs), such as those caused by human immunodeficiency virus (HIV) and human papillomavirus (HPV) [1,4,7]. In addition, BV is associated with other gynaecological complications, such as preterm labour, abortion, intrauterine infection, and pelvic inflammatory disease [1,4,5,8].

G. vaginalis (GV) is the major causative agent of BV and the leading member of a specific group of BV-associated anaerobes that can form biofilms on the surface of vaginal epithelial cells (ECs) [[9], [10], [11], [12]]. In a previous study, GV accounted for 90% of the bacteria observed in biofilms, whereas lactobacilli were predominant in healthy women but did not form biofilms [13]. GV has been detected not only in women with BV but also in those without BV [14,15]. Among BV-related pathogens, GV is particularly characterized by three virulence traits: cytotoxicity, adhesion to vaginal ECs, and biofilm formation [15,16]. Biofilm formation is crucial for the survival of GV in the vaginal environment [16]. Other anaerobes, including Fannyhessea vaginae, Prevotella bivia and Sneathia amnii, are also embeded within the biofilm [12], where they contribute to inflammation and disrupt the epithelial barriar homeostasis [17].

BV biofilms are surface-associated microbial communities formed on vaginal epithelial cells that produce an extracellular polymeric substance matrix, which acts as a protective barrier against the influence of vaginal factors [12,18,19]. This self-produced matrix mainly consists of polysaccharides, proteins, nucleic acids and few lipids [20]. GV can form biofilms and then adhere to the surface of vaginal ECs [2]. Biofilms may help reduce the penetration of antimicrobial agents, thereby facilitating the persistence of antibiotic-susceptible microbes and creating a dormant state to increase the tolerance of GV to antibiotics [21]. The bacteria in biofilms can exist in a dormant state and revive once the antibiotics treatment is stopped, resulting in repeated attacks of BV.

Recurrent BV is defined as a confirmed diagnosis of BV three or more times in the same year [22]. Globally, although the therapeutic options have increased, the recurrence rate of BV remains high even after extended treatment. One of the current treatment approaches for BV is antimicrobial therapy involving metronidazole (MTZ) and clindamycin [3,5,23]. Our previous study revealed that the most prevalent GV strains in local China were resistant to MTZ, with the minimum inhibitory concentration (MIC) exceeding 128 μg/mL [24]. Moreover, a recent study reported that MTZ treatment failure could enhance HIV susceptibility [25]. Under normal conditions, lactobacilli are present in high concentrations in the vagina; they can inhibit the growth of pathogenic organisms and maintain vaginal pH by producing hydrogen peroxide and lactic acid [22,26]. Treating BV with antibiotics not only kills GV but also destroys microecological balance, contributing to the recurrence of BV. Studies have suggested that the recurrence rate of BV is as high as approximately 80% within 12 months of effective treatment because of antibiotic resistance of the pathogenic bacteria and their biofilms [1,5]. Although probiotic therapy has exhibited promising results, it cannot fully control the recurrence rate [27,28]. Consequently, the management of this issue has turned into a major challenge [2,4]. The persistence of certain BV-associated bacteria and biofilms may be a determinant of recurrence because biofilms can persist after putatively successful antimicrobial therapy in women. The function of biofilms may well explain the failure of effective antimicrobial therapy [21]. Therefore, for treating BV, there is an urgent need to explore new strategies that can disrupt biofilms with high efficiency. The optimal efficacy of antimicrobial therapies could be achieved by combining antimicrobial agents with biofilm-disrupting agents, particularly in individuals at risk of recurrence.

Despite the growing clinical burden of stubborn BV associated with GV biofilms, the composition and resilience of these biofilms remain poorly understood. In this study, we aimed to characterize the composition and structural features of GV biofilms and evaluate potential strategies for their disruption. Our approach combined direct compositional analysis with immunofluorescence confirmation, revealing that the GV biofilm is predominantly proteinaceous. This finding indicated that protein-dominated pattern underlined the biofilm's stability and its resistance to treatment. Notably, LasA from Pseudomonas aeruginosa effectively degrades the GV biofilm and lyses GV cells, highlighting a potential therapeutic strategy for overcoming BV recalcitrance.

2. Results

2.1. GV has a protein-rich extracellular matrix

The bacterial extracellular matrix assumes an important role in maintaining the structure of biofilms. Tiny granules were distributed around the cells as revealed by atomic force microscopy (AFM, Fig. 1a–d). These granular structures, which represent the major component of the extracellular matrix, are unlikely to be polysaccharides or extracellular DNA [29,30], as these polymers typically form fibrous or thread-like architectures. Direct compositional analysis of the extracted biofilms indicated that extracellular DNA (eDNA) and polysaccharides remained stable at 10.6% and 22.4%, respectively, over three extraction cycles. Protein concentrations varied depending on the assay employed, yielding 54.7% by the Bradford method and 60.3% by the BCA assay. Following removal of DNA and proteins during polysaccharide purification, samples were freeze-dried and reweighed to quantify DNA and protein. Given eDNA is 10.6%, the residual protein accounts for ∼68.5% of the biofilm mass. The remaining 0.7–9.1% was attributed to lipids or other minor constituents (Fig. 1e). Confocal laser scanning microscopy (CLSM) further demonstrated that surface proteins in GV biofilms, crosslinked with miniGFP, were widely distributed throughout the biofilm matrix. (Fig. 1f–i).

Fig. 1.

AFM images of G. vaginalis JNFY14 cells (a-d; bar, 1 μm), biofilm components distribution (e) and CLSM images of G. vaginalis JNFY14 biofilms after being crosslinked with miniGFP-spyTag. (f-i; bar, 10 μm). Cells of GV JNFY14 were cultured in BHIs broth for 48 h and washed with ddH2O for three times, and then were laid on mica sheet (a-d; arrows for extracellular substances). Panels a and c show peak force error; b and d show corresponding height with a scale from −3 nm to 3 nm. Pink areas in the height images represent higher structures that were beyond the height range for display. Distribution of biofilm matrix components, including proteins, polysaccharides and extracellular DNA (eDNA), quantified by protein assay (Bradford/BCA), phenol-sulfuric acid method and spectrophotometry, respectively (e). GV JNFY14 biofilms were grown in 24-well plates for 48 h, treated with 1–2% (w/v) glutaraldehyde and miniGFP-SpyTag (2 mg), and incubated at 37 °C for 3 h quenching with glycine. Biofilms were washed with 50 mM PBS (pH7.0) prior to imaging. Biofilms without glutaraldehyde (f, g); with glutaraldehyde (h, i); bright-field images (f, h); EGFP channel (g, i).

2.2. LasA from P. aeruginosa disintegrates GV biofilms when supplied exogenously

LasA is an elastase from P. aeruginosa that specifically recognises a Gly-Gly-X sequence, where X is preferably glycine or alanine, cleaving the peptide bond after the second glycine [31]. LasA can also hydrolyse the glycine-containing elastin pentapeptide [32]. Therefore, we tentatively treated GV biofilms with LasA to assess whether it would have a disruptive effect.

Pre-LasA was purified via Ni-NTA (Nickel-Nitrilotriacetic acid) chromatography, and mature LasA was obtained through trypsin digestion, then further purified to homogeneity using gel filtration (Fig. S1a and b). Enzyme activity assays employing a synthetic fluorogenic peptide substrate revealed minimal differences between ambient temperatures (25 °C) and 37 °C, with modest attenuation observed at supra-physiological temperatures (45–50 °C) (Fig. S1c). The pH optimum for LasA activity occurred in the slightly alkaline range (pH 8.0–9.0). Notably, approximately 70% of maximal activity was retained under acidic conditions (pH 5.0), demonstrating its potential application in the acidic vaginal microenvironment (Fig. S1d).

To evaluate the activity of LasA on GV biofilms, we applied increasing gradient of LasA solutions to the corresponding wells containing established GV biofilms and incubated the mixture for 6 h at 37 °C. The thick biofilms formed in the presence of sheep blood were extensively disintegrated (Fig. 2a).

Fig. 2.

Degradation of GV biofilms by LasA at gradient concentrations and comparation with erythromycin (ERM) and lysozyme (LZM).

GV biofilms were treated with serial dilutions of LasA (0, 320, 160, 80, 40, 20, 10 and 5 μg/mL) and incubated at 37 °C for 6 h (a); Comparative treatments with 1 mg/mL lysozyme, 1 mg/mL erythromycin and 20 μg/mL LasA were performed at 37 °C for 2 h (b); GV biofilms were cultivated on mica sheet for 48 h prior to treatment with 1 mg/mL lysozyme (d), 1 mg/mL erythromycin (e) and 20 μg/mL LasA (f) at 37 °C for 2 h, respectively. c, control. Bar, 2 μm. Statistical analysis was carried out using GraphPad Prism 9.5.1. ∗∗∗, p < 0.01.

The abilities of LasA, lysozyme and ERM to disrupt GV biofilms were also compared in order to assess whether LasA is a more effective therapeutic option. Crystal violet staining was used to view the results clearly. Treatment with 20 μg/mL of LasA significantly disrupted GV biofilms. In contrast, lysozyme showed very weak activity, even at a concentration of 1 mg/mL. Moreover, 1 mg/mL ERM did not exhibit any obvious effect on the integrity of GV biofilms (Fig. 2b).

AFM was used to observe the morphological changes in biofilms and bacterial cells after treatment with lysozyme, ERM and LasA (Fig. 2c–f). The untreated biofilm cells were structurally intact and tightly packed together on the mica sheets (Fig. 2c). In contrast, the lysozyme-treated biofilm cells were aggregated and covered with a layer which appeared to prevent further biofilm disruption, and the cells still adhered to the mica sheets. These findings indicate that lysozyme cannot eliminate GV biofilms (Fig. 2d). The cells of the ERM-treated biofilm appeared structurally intact, although a few showed a decrease in thickness. However, overall, the cells were still tightly packed together and adhered to the mica sheets. A previous study revealed that this GV strain is resistant to many antibiotics but sensitive to ERM [33]. Although most bacteria with active metabolism inside the biofilm were inhibited by ERM, the biofilm still had an intact structure and adhered to the contact carrier, potentially allowing persistent cells to survive and resume growth later (Fig. 2e) [34]. LasA treatment had a significant effect on GV biofilms. In particular, most cells were removed from the mica sheet, and the residual bacterial cells lost their intact structure, resulting in cytoplasmic leakage. Thus, LasA can not only disrupt GV biofilms by digesting the protein components but also lyse the cell wall of GV to achieve bactericidal purpose (Fig. 2f), which was further verified in the following experiments.

2.3. Bactericidal effect of LasA and lysozyme on planktonic GV

LasA, an M23 metallopeptidase with highly restricted specificity, has been shown to lyse the cell wall of Staphylococcus aureus [35]. Based on our experimental results, we investigated whether it can similarly degrade the cell wall of GV. As shown in Fig. 3, the lytic activity of LasA against GV increased with the concentration of the applied enzyme. At 10 μg/mL, LasA lysed 44%, 63% and 79% cells at 30, 60 and 180 min, respectively. Increasing the concentration to 20 μg/mL resulted in lysis rates of 57%, 72% and 82% at the same time points. When the concentration exceeded 40 μg/mL, the lysis rate approached 80% within 3 h, comparable to lower concentrations. These findings suggest that the minimum effective concentration of LasA is approximately 10–20 μg/mL (Fig. 3a and b). Additionally, to evaluate the lytic effects of LasA on planktonic GV, suspensions of JNFY14 cells were treated with 10 μg/mL ERM and 1 mg/mL lysozyme. LasA at up to 80 μg/mL exhibited activity comparable to ERM, whereas lysozyme did not produce any observable damage to the GV cell wall, suggesting that even minimal amounts of extracellular matrix in the planktonic state may impede lysozyme access (Fig. 3b).

Fig. 3.

Degradation of GV cell walls by LasA at gradient concentrations and comparation with lysozyme (LZM) treatment. Suspensions of GV JNFY14 were incubated with LasA of 0, 5, 10, 20, 40, 80 μg/mL at 37 °C for 3 h, respectively (a); Suspensions of GV JNFY14 were treated with LasA (0, 10, 20, 40, 80 μg/mL), erythromycin (ERM, 10 μg/mL) or lysozyme (LZM, 1 mg/mL) at 37 °C for 3 h, serially diluted 105–107-fold, and plated on BHIs agar. Plates were incubated at 37 °C with 5% CO2 (v/v) for 24 h and colony numbers were counted to determine survival rates (b); Cell wall extracts (2 mg) were incubated with LasA (20 μg/mL) at 37 °C for 3 h (c); Cell wall extracts (2 mg) and LasA-digested debris (from c) were each incubated with lysozyme (1 mg/mL) at 37 °C for 3 h (d). Statistical analysis was carried out using GraphPad Prism 9.5.1. ∗∗∗, p < 0.01; ∗∗∗∗, p < 0.005.

To elucidate whether LasA and lysozyme cooperate with and degrade the GV cell wall, peptidoglycan was extracted from GV cells and incubated separately with 20 μg/mL LasA and 1 mg/mL lysozyme. Peptidoglycan debris was observed after LasA digestion (Fig. 3c). In contrast, treatment with lysozyme alone yielded no detectable degradation products; sequential treatment of peptidoglycan with LasA followed by lysozyme generated only two distinct muropeptide species (Fig. 3d).

2.4. Identification of potential substrate proteins of LasA in GV biofilm

Nine proteins containing signal peptides identified by cross-linking mass spectrum (XL-MS) were considered as potential substrates of LasA (Table 1). Among them, cholesterol-dependent cytolysin vaginolysin (WP_004574832.1), ABC transporter substrate-binding protein (WP_014554287.1), FTR1 family iron permease (WP_004573679.1) and CHAP domain-containing protein (WP_248898768.1) were successfully expressed in E. coli and the gradual diminution of target protein bands following LasA treatment supports their susceptibility to LasA-mediated degradation (Fig. S2). These results strongly suggest that these four proteins are candidate components of GV biofilm and substrates of LasA.

Table 1.

Candidate proteins on the surface of GV biofilms identified by XL-MS.

Function	MW (kDa)	Number of Gly-Gly-X motifs	Comment	Accession No.
Peptidase	278.8	5	Competitive exclusion	WP_248898599.1
Phosphate ABC transporter substrate-binding protein	41.1	1	Ingestion	WP_248898628.1
Cholesterol-dependent cytolysin vaginolysina	56.9	2	Cytotoxin	WP_004574832.1
ABC transporter substrate-binding proteina	46.1	3	Ingestion	WP_014554287.1
FTR1 family iron permeasea	62.6	3	Iron acquisition	WP_004573679.1
Rib/alpha-like domain-containing protein	338.7	7	Immune response evasion and mitigation	WP_248898518.1
GA module	213.1	6	Immune response evasion and mitigation	WP_248898505.1
CHAP domain-containing proteina	49.7	3	Competitive exclusion	WP_248898768.1
Tat pathway signal sequence domain protein	159.3	6	Protein transport	WP_248898776.1

^aSuccessfully expressed in E.coli and degraded by LasA.

2.5. LasA has no adverse effects on the vaginal beneficial bacterium L. crispatus

L. crispatus is a common beneficial bacterium in the female genital tract in Asia and plays important roles in maintaining the health of the vaginal microecology. We treated an L. crispatus CXD strain isolated from a patient's vagina with various concentrations of LasA and assessed its effect using the colony counting method (Fig. S3). Although L. crispatus is also a Gram-positive bacterium like S. aureus, it was not harmed by LasA, which might be attributed to the fact that it lacks Gly-Gly-X motif in the cell wall peptidoglycan [36]. This means that LasA can selectively eliminate pathogenic Gardnerella species in vagina while leaving beneficial bacterium L. crispatus unaffected, thus facilitating the recovery of the microecological balance. These data suggest that LasA could be a good therapeutic option for BV.

2.6. Evaluation of the therapeutic effect of LasA and antibiotics using murine vaginal infection models

To investigate the in vivo effect of LasA, we established a murine model of BV induced by GV and evaluated its colonization fate (Fig. S4). However, we observed a rapid decline in GV biomass without any pretreatment of vaginas (Fig. S5a and c). Therefore, pretreatments with alkaline buffer, bactericide, and dexamethasone (DXMS) were performed to promote GV colonization (Figs. S5 and S6). Both alkaline buffer and cetylpyridinium chloride lavage are believed to disrupt the microecological balance, thereby fostering GV occupancy. In contrast, lactobacilli biomass decreased after both alkaline buffer and bactericide treatments three days after inoculation (Fig. S5b and d). This suggests that both treatments harm endogenous lactobacilli. Additionally, the immune system naturally protects against pathogenic invasion. To enhance GV colonization without damaging beneficial bacteria, DXMS was used to suppress the mice's immune response. Following this intervention, both GV strains achieved successful colonization, with a slight decrease observed in biomass of JNFY14 (Fig. S6a) and a threefold increase observed in that of ATCC 14019 (Fig. S6c). Regardless of the infecting strain, the relative biomass of lactobacilli increased six-fold and two-fold after DXMS treatment, exhibiting a consistent pattern with the control group, which suggested that DXMS merely suppressed the immune system without posing any detrimental effects on lactobacilli viability (Fig. S6b and d). Therefore, this model would facilitate further research regarding the interaction between GV and lactobacilli during the growth process.

The morphology of normal ECs changed after infection. In particular, the cells exhibit a rough-looking surface with irregular edges, spots and granules, and a large number of short bacilli are attached to the surface. These cells are known as clue cells and are a sign of BV. When many clue cells are present in the vaginal discharge, the patient may be initially suspected of having BV. When mice were infected with the clinical isolate GV JNFY14, ECs covered with a large number of small bacilli from the vaginal sample were observed (Fig. 4b). These cells exhibited the characteristics of clue cells, which were absent in the control group but accounted for approximately 65.7% of the total ECs (Fig. S7). After LasA treatment, the percentage of clue cells decreased to 22.8%. Although a small number of ECs covered with bacilli remained (Fig. 4c), the number of bacteria attached to the surface of ECs decreased significantly (Fig. S7). In contrast, following ERM treatment, clue cells, which accounted for 46.7% of the total ECs, were still present in the vaginal secretions of the mice (Fig. 4d & Fig. S7). These results strongly suggest that LasA may be a more potential option than antibiotics for treating BV.

Fig. 4.

Vaginal epithelial cell morphology (a–d) and changes in 16S rRNA gene copy number of G. vaginalis and lactobacilli in the vaginas of GV-infected mice following LasA or antibiotic treatment (e,f). Vaginal secretion smears from GV JNFY14-infected mice showing epithelial cells and adherent GV cells. Insets highlight zoomed-in regions (scale bar, 10 μm; inset bar, 2 μm). (a) Uninfected epithelial cells; (b) GV-infected vaginal epithelial cells without treatment; (c) GV-infected vaginal epithelial cells after LasA treatment; (d) GV-infected vaginal epithelial cells after ERM treatment. Arrows indicate GV cells adhered to epithelial cell surface. Biomass changes of G. vaginalis and lactobacilli populations after treatment in the vagina of mice infected with GV JNFY14 (e) or GV ATCC 14019 (f). Successfully infected mice were divided into 3 groups and treated once a day for 6 days: Group 1 was treated with 20 μg/mL LasA, group 2 was treated with the same concentration of antibiotics (ERM for strain JNFY14 and MTZ for strain ATCC 14019, respectively); and the control group was treated with an equal volume of 50 mM PBS buffer (pH 7.0). Statistical analysis was carried out using GraphPad Prism 9.5.1. ∗∗, p < 0.05; ∗∗∗, p < 0.01; ∗∗∗∗, p < 0.005.

LasA significantly reduces the number of GV cells in vivo without negatively affacting lactobacilli.

The effects of treatment with LasA and antibiotics in the BV murine models were also quantitatively evaluated using qPCR. The relative abundance of G. vaginalis, lactobacilli, and total bacteria was quantified by 16S rRNA gene copy numbers, which were calculated from Ct value standard curves generated for each primer pair (Fig. S8). In both models (infected with the clinical isolate JNFY14 and the type strain ATCC 14019), LasA reduced the number of GV cells much more effectively than antibiotics. The number of GV cells in the control group of the ATCC 14019-infected murine model decreased to a greater extent than that in the JNFY14 group (Fig. 4e and f). This observation may be attributed to the JNFY14 strain's better adaptation to the vaginal environment of mice compared to the ATCC 14019 strain. Notably, both strains were more effectively eliminated by LasA than by antibiotics. In addition, opposite trends in total bacterial biomass were observed following the two treatments in both the JNFY14- and ATCC14019-infected murine models: LasA treatment resulted in an increase in total bacterial biomass, whereas antibiotic treatment led to a decrease (Fig. S8).

LasA has been shown to have no bactericidal effect on a clinically isolated L. crispatus strain (Fig. S3). Our results indicated that antibiotics were significantly more harmful to lactobacilli than LasA. Notably, biomass of lactobacilli remained stable in JNFY14-infected mice, while they decreased by approximately 50% in ATCC 14019-infected mice. This discrepancy may be attributed to a greater incompatibility between ATCC 14019 and lactobacilli. Notably, the lactobacilli-specific primers, designed based on human-derived Lactobacillus strains, is also applicable to most mouse-derived Lactobacillus species [37,38]. In particular, lactobacilli detected in vaginal samples is likely derived from the murine intestinal microbiota, consistent with the neutral pH of the mouse vagina. Moreover, LasA showed minimal adverse effects on lactobacilli viability in vivo, as indicated by sustained bacterial colonization (Fig. 4e and f).

Impact of LasA on the vaginal epithelial tissue is non-injurious.

The safety of LasA was assessed by examining vaginal tissue sections from treated mice. It is evident that even after 3–6 days of treatment with either 20 μg/mL or 200 μg/mL of LasA (equivalent to 1–2 treatments), the thickness of the vaginal epithelial tissue remained at 100 μm, comparable to that observed in the control group (Fig. 5 & S9). Proteomic profiling of the PBS- and 200 μg/mL LasA-treated groups over 6 days revealed the stable expression of most proteins involved in pathological pathways. Strikingly, key immune regulators—including CARD9 signaling mediators, proteins related to inflammatory responses, peroxisome proliferator-activated receptor (PPAR), and T cell receptor activation markers—were absent from both treatment groups (Table S1).

Fig. 5.

Vaginal epithelial thickness of mice vaginas remains unchanged after LasA treatment. Vaginas of healthy mice were washed with 20 μg/mL (b, e) or 200 μg/mL (c, f) once per day for 3 days (a–c) or 6 days (d–f). Control mice were washed with 50 mM PBS buffer (pH 7.0, a, d). Vaginal tissues were fixed in 10% neutral buffered formalin at room temperature, embedded in paraffin, sectioned and stained with hematoxylin and eosin (H&E). Stained sections were imaged using an Olympus IX71 microscope and epithelial thickness was measured using ImageJ 1.54i software as the mean of 6 measurements per section. Bar, 50 μm.

3. Discussion

The high recurrence rate of BV has long been a major problem affecting patients. In those accepting conventional therapy, the recurrence rate could be as high as 80% [5,39]. These medications have been used for many years but have not provided a lasting cure. Effective, long-lasting treatments are urgently needed, as recurrent BV has a devastating impact on patients’ health [37]. The unacceptably high recurrence rates of BV highlight the urgent need for major advances in developing effective treatments that can disrupt biofilms and be used in combination with antimicrobial agents [21].

AFM imaging initially indicated that the GV biofilm matrix differs from typical extracellular polysaccharide (EPS)-dominated biofilms. While quantitative analyses of biofilm components showed some variation across three measurements, the protein fraction consistently exceeded 50% of the total biomass. This observation was further supported by the widespread distribution of miniGFP-tagged proteins throughout the biofilm matrix, as visualized by CLSM. Therefore, it was speculated that the GV biofilm matrix contains a diverse array of protein components. This protein-rich nature of the biofilm matrix is also consistent with the fact that proteinase K can disrupt the GV biofilm [40]. However, proteinase K is a broad-spectrum protease that can digest a wide variety of proteins; therefore, its use for treatment purposes may cause unpredictable damage to the epithelia. For this reason, a protease with a narrow substrate spectrum would be a more suitable option for the treatment of BV.

AFM also revealed that an elastase, LasA, could efficiently disrupt GV biofilms, while lysozyme and antibiotics had no effect on the biofilm structure. This result was further confirmed by subsequent experiments. LasA has been reported to specifically cleave the peptide bond after the second glycine in the Gly-Gly-X motif [32,33]. Sequence analysis of candidate proteins identified in the GV biofilm matrix revealed the presence of one or more Gly-Gly-X motifs, suggesting that these proteins could serve as potential substrates for LasA-mediated cleavage. In addition to its activity on peptidoglycan, which is thinner in GV than in S. aureus, LasA may also target protein components of the biofilm matrix. This dual activity could contribute to both biofilm destabilization and bacterial cells lysis.

The nine identified potential components of the biofilm proteinaceous matrix are primarily extracellular proteins that may serve as integral parts of the biofilm and fulfil various functions, including competition (WP_248898599.1, WP_248898768.1) [41,42], cytotoxicity (WP_004574832.1, WP_004573679.1) [40,41], predation (WP_248898628.1, WP_014554287.1), and immune evasion (WP_248898518.1, WP_248898505.1) [43,44]. Additionally, WP_248898776.1 is predicted to be a Tat pathway signal sequence domain protein that may be closely associated with the proteinaceous biofilm formation [45]. With the exception of WP_248898599.1, WP_248898518.1, WP_248898505.1 and WP_248898776.1, which could not be successfully expressed probably due to their large molecular weights, and the transmembrane protein WP_248898628.1, which exhibited low expression levels, WP_004574832.1, WP_014554287.1, WP_004573679.1 and WP_248898768.1 were successfully expressed in E. coli. All of the purified proteins were cleaved by LasA, further validating the target sites through which LasA disrupts the GV biofilm. After breaching the initial layer of the biofilm barrier, LasA continues to target the peptide cross-link structures within the peptidoglycan of GV cell walls, ultimately resulting in the lysis of GV cells. From the immunofluorescence labeling results, Vly appears to be a prominent protein within the GV biofilm matrix and may contribute to virulence rather than serving as a dedicated structural scaffold. Its ability to associate with vaginal epithelial cells and induce host cell damage could support bacterial persistence and biofilm development on epithelial surfaces. More broadly, protein-host and protein-matrix interactions are likely to play important roles in GV biofilm biology. Further characterization of the proteinaceous fraction of the biofilm will therefore be an important focus of future studies and may reveal additional therapeutic targets for disrupting GV biofilms in the context of bacterial vaginosis.

Surprisingly, lysozyme has little effect on GV in both the biofilm and planktonic forms. This phenomenon may be attributed to the presence of an extracellular matrix in both states, which acts as a barrier preventing lysozyme from accessing the cell wall. Additionally, the absence of teichoic acid may lead to a deficiency of recognition sites for lysozyme, in contrast to the cell walls of typical Gram-positive bacteria [46]. Notably, further treatment of LasA-digested cell wall debris with lysozyme results in the production of only a limited number of muropeptides, suggesting that LasA digestion may expose additional binding sites for lysozyme by cleaving peptide bridges.

This study further explained why antibiotics are often ineffective in treating BV. Except for the widespread drug-resistant strains of GV, biofilms are the decisive factor for resistance. Although antibiotics may eliminate most of the active bacteria in biofilms, they cannot disrupt the biofilm structure, which shelters persistent cells that eventually resume growth, resulting in recurrence. Moreover, antibiotic treatment leads to the dysregulation of the vaginal microflora. This not only decreases the bacterial population but also causes certain beneficial bacteria to develop drug resistance, altering their interactions with pathogens and impairing their regulatory functions. This can result in long-lasting and recurrent infection.

Our findings indicate that LasA has more prospect for treating BV. Upon administeration, LasA first degrades the protein-rich matrix of GV biofilms, exposing the bacterial cell wall to outside attack. LasA continues to hydrolyse the peptide bonds in peptidoglycan, lysing the bacterial cells. As LasA targets the major component of the biofilm or cell wall, which is generally highly conserved, it is difficult for GV to develop drug resistance. The fact that LasA does not negatively affect lactobacilli is an unexpected benefit. This feature has the potential to help patients recover faster and reduce the likelihood of relapse. Moreover, safety evaluations of LasA demonstrated that treatment with both low and high concentrations did not cause any observable damage to the vaginal epithelial tissue in mice. This finding is also consistent with a previous report that LasA is not a corneal virulence factor [47]. In summary, LasA is an effective and safe therapeutic option for BV. Its early clinical application will help relieve pain of a greater number of patients.

4. Materials and methods

4.1. Strains, culture conditions and biofilm formation

GV ATCC14019 was purchased from the American typical strain collection center. GV JNFY14 and L. crispatus CXD were isolated from clinical samples of patients provided by Jinan Maternal and Child Health Hospital [24]. BHIs broth medium contained 3.85% brain-heart infusion broth (Haibo, Qingdao, China), 2% (w/v) yeast extract, 2% (w/v) gelatin, 0.1% (w/v) glucose and 0.1% (w/v) soluble starch. 0.1% (w/v) glucose was replaced by 0.3% (w/v) maltose and 5% sheep blood was added into the Biofilm medium. For planktonic culture, GV cells were inoculated in BHIs broth medium, then incubated at 37 °C within 5% CO₂ for 24 h. For biofilm formation, 1% (v/v) seed culture (OD₆₀₀ = 1.5) was inoculated into the biofilm medium in a 24- or 96-well microplate and the plates were incubated under the same condition for 48 h. L. crispatus CXD was cultured in MRS medium (Haibo, Qingdao, China) at 37 °C within 5% CO₂ for 24 h.

4.2. Plasmid construction, protein expression and purification

Sequences encoding full-length LasA (Signal peptide is not included) was amplified and cloned into pGLO1, a modified vector from pET15b with a PreScission Protease (PPase) cleavage site for the removal of His-tag. The resulting expression plasmid was transformed into E. coli BL21(DE3), then cultured at 37 °C till OD₆₀₀ reached 0.8, and induced overnight with 0.1 mM isopropyl β-d-thiogalactopyranoside at 18 °C.

For the purification of protein, bacterial cells were harvested by centrifugation for 20 min at 5000×g then resuspended in binding buffer (25 mM Tris-HCl, pH8.0, 500 mM NaCl). Afterwards, the cells were lysed by sonication and then centrifuged for 50 min at 25200×g to remove the precipitate. The supernatant containing LasA was loaded onto a Ni-NTA column (GE Healthcare), then washed for three times with binding buffer (supplemented with 20 mM imidazole) to remove the miscellaneous proteins. The protein bound to the resin was resuspended in 3 mL elution buffer mixed with Trypsin (a final concentration of 100 μg/mL) and incubated for 30 min to produce a mature active protease LasA. Then the protein sample was eluted with binding buffer. The elute was further purified by size-exclusion chromatography (Superdex200 10/3000 GL, GE Healthcare) in 10 mM Tris-HCl buffer with 500 mM NaCl at pH 8.0.

4.3. Temperature and pH tolerance assay of LasA on synthetic peptide, whole cells of GV & Lactobacillus crispatus and the cell wall of GV

The activity of LasA was assessed following the methodology outlined in a previous study [48], using the synthetic internally quenched fluorogenic peptide Dabsyl-Leu-Gly-Gly-Gly-Ala-Edans as the substrate. A concentration of 20 μg/mL of LasA was incubated with 50 μM of the substrate dissolved in 50 mM phosphate buffer solution (PBS, pH 8.0) at temperatures of 25, 37, 45, and 50 °C for durations of 10, 30, 60, 120, and 180 min, respectively, to evaluate the enzyme's thermostability. The pH tolerance range of LasA was determined by following the same procedure at pH values of 5.0, 6.0, 7.0, 8.0, and 9.0, with all reactions maintained at 37 °C.

After culturing GV JNFY14 in BHIs broth for 24 h, the planktonic cells were harvested and resuspended in Tris buffer (25 mM Tris-HCl, pH8.0, 500 mM NaCl). Then a series of LasA (0, 10, 20, 40 and 80 μg/mL) were added into the suspension and detected the cell density changes over time detected using microplate reader. Suspensions of GV JNFY14 and L. crispatus CXD cells were treated with LasA at multiple concentrations (0, 10, 20, 40, and 80 μg/mL). Two additional control groups, treated with 10 μg/mL ERM or 1 mg/mL lysozyme, were included using the same GV suspensions to assess the sterilizing effect of LasA. Following a 3 h incubation at 37 °C, samples were serially diluted to 10⁵–10⁷-fold and plated on BHIs for GV and on MRS for L. crispatus. Plates were incubated at 37 °C within 5% CO₂ for 24 h (GV) and overnight (L. crispatus), respectively, after which colonies were counted to determine survival rates.

To elucidate how the GV cell wall was affected by LasA and lysozyme treatments, 2 mg cell wall peptidoglycan of strain JNFY14 was extracted following previously described procedures and subsequently degraded by incubation with 20 μg/mL LasA at 37 °C for 3 h [49,50]. The reactions were terminated by heating, and the resulting lysate was then treated with 1 mg/mL lysozyme at 37 °C for 3 h. Finally, the supernatants were analyzed by HPLC (Shimadzu, Japan) using a C18 column (150 mm × 4.6 mm, 5 μm; Agela, China). The mobile phase comprised a gradient elution from 50 mM PBS (pH 4.31) to 75 mM PBS (pH 4.89) with 15% (v/v) methanol; separation was performed at 45 °C with a flow rate of 0.5 mL/min over 140 min. The eluent was monitored using a UV detector set at 205 nm. To eliminate background interference from PBS, enzymes, and substrates, the experiment was conducted in four independent trials, each performed in triplicate.

4.4. Image acquisition and identification of biofilm surface proteins

After culturing GV JNFY14 on mica for 48 h, the culture medium was removed, and the biofilm was gently washed twice with deionized water. The mature biofilm was treated with 1 mg/mL lysozyme, 1 mg/mL erythromycin and 20 μg/mL LasA at 37 °C for 2 h, respectively. Then the biofilm was gently washed with deionized water, dried and observed by AFM. AFM imaging was carried out using a Multimode VIII atomic force microscopy equipped with a Nanoscope V controller (Bruker, USA). A cantilever (XSC11/AI BS, MikroMasch) with a nominal spring constant of 2.7 N/m was used for imaging. Scans were conducted in ScanAsyst mode under ambient air condition.

Biofilms were grown for 48 h in five 24-well plates, with a total of approximately 250 mL of GV JNFY14 culture, using 2 mL per well. After incubation, the liquid medium containing planktonic cells were carefully discarded to remove non-adherent cells. The remaining attached biofilm were then resuspended by adding 1 mL of 50 mM PBS (pH 7.0) to each well. The resuspended biofilms were collected, centrifuged, and washed three times with 25 mL of PBS. Subsequently, 1–2% (w/v) glutaraldehyde and 2 mg of miniGFP-SpyTag were added to the resuspended mixture [51,52]. The mixture was incubated at 37 °C for 3 h, after which glycine was added to terminate the reaction. The biofilms were washed with PBS and observed under the cofocal laser scanning microscopy (CLSM, LSM900, Zeiss, Germany) to assess the efficiency of cross-linking. The miniGFP cross-linked biofilms were then lysed using high-pressure homogenization. The lysate was mixed with a final concentration of 1% n-dodecyl-β-d-maltopyranoside and centrifuged at 12,000×g for 40 min, the supernatant was loaded to a gravity column containing MBP beads conjugated with SpyCatcher [51]. After PBS wash, 20 mM maltose was added to elute the crosslinked surface proteins, which were subsequently analyzed by using liquid chromatography-mass spectrometry (ThermoFisher Scientific, USA). The detailed experimental workflow is shown in Fig. S10.

4.5. Quantitative analysis of biofilm matrix components

Biofilms of GV JNFY14 were collected as described above and sonicated in an ultrasonic cleaner for 30 min at 70% power. The supernatant was collected by centrifugation at 2000×g for 20 min, and the pellet was discarded. The supernatant was desalted using a disposable PD-10 column (GE Healthcare, USA), lyophilized, and the dry biofilm mass was weighed. The dry biofilm was then dissolved in 50 mM PBS (pH 7.0), and DNA concentration was measured using a Nanodrop spectrophotometer (GE Healthcare, USA). Due to the difficulty of accurately determining protein concentrations, protein levels were quantified using the Bradford assay kit (Beyotime, China) and the BCA assay kit (SparkJade, China). Subsequently, 20 μg/mL DNase I (Sigma-Aldrich, USA) was added and incubated at 37 °C for 30 min, followed by the addition of 1 mg/mL proteinase K (Beyotime, China) and incubation at 65 °C for 30 min. Next, one-third volume of Sevage reagent (chloroform: n-butanol = 4:1, v:v) was added and the mixture was allowed to stand at room temperature for 30 min. The top aqueous phase was carefully collected, and the extraction was repeated five times to remove residual proteins. Organic solvents were then removed by rotary evaporation at 35 °C. The sample was centrifuged at 8000×g for 10 min, and the resulting supernatant was freeze-dried to determine the protein removal efficiency before polysaccharide purification by ion-exchange chromatography. A DEAE-cellulose DE-52 column (Borui Saccharide Biotech, China) was pretreated by soaked in 0.5 M hydrochloric acid for 1 h, then washed to remove impurities, and equilibrated with distilled water. After adjusting the flow rate to 5 mL min⁻¹ and equilibrating for 2 h, the sample was loaded onto the column, and elution was performed with distilled water at 15 mL min⁻¹. Polysaccharide content was monitored by the phenol-sulfuric acid method at 490 nm using a spectrophotometer (Tecan Sunrise, Switzerland).

5. Biofilm disassembly assay

After culturing biofilm of GV JNFY14 in 96-well plate for 48 h, medium was removed, and the biofilm was gently washed with deionized water twice. A series of LasA with concentrations of 0, 5, 10, 20, 40, 80, 160 and 320 μg/mL were added into each well of the plate. After incubation at 37 °C for 6 h, the supernatant was removed, and each well was washed twice with deionized water gently. For quantitative analysis, 0.05% (w/v) crystal violet dye were added and then poured out the dye and washed twice after 10 min. Decolorization was carried out with 33% (v/v) acetic acid for 15 min, the absorbance at 570 nm was measured with Microplate Reader (Tecan Spark, Switzerland). Each treatment was performed in triplicate, including three biological replicates and three technical replicates.

5.1. Heterogeneous expression, purification of the identified proteins and LasA degradation test

All identified proteins listed in Table 1 were heterogeneously expressed and purified using Ni-NTA affinity chromatography, followed by anion-exchange chromatography, and gel filtration. The purified substrate proteins—vaginolysin, ABC transporter substrate-binding protein, FTR1 family iron permease, and CHAP-domain-containing protein—were adjusted to a final concentration of 250 μg/mL. Various concentrations of LasA (0, 10, 20, 50, 100, and 200 μg/mL) were then incubated with the substrate proteins at 37 °C for 1 h. The reaction mixtures were subsequently analyzed by SDS-PAGE to visualize the protein degradation profiles.

5.2. Construction of the vaginal infection murine model

After 7–11 weeks of age, C57BL/6 female mice were acclimatized for one week and injected intraperitoneally with 100 μL estradiol at a concentration of 2 mg/mL, which was repeated after 3 days, and then they could be injected once a week in the experiment to maintain their pseudoestrus status as reported in prior studies [53,54]. On the day after the second treatment with estradiol, mice were injected intraperitoneally with 200 μL dexamethasone sodium phosphate at a concentration of 1 mg/mL, once a day until the end of the experiment to suppress the immune system. The mice vagina was rinsed with an alkaline buffer of pH 8.0 and cetylpyridinium chloride as a bactericide to disrupt the microecological balance before GV inoculation, in order to compare the effect of GV colonization with dexamethasone injection. After pretreatment for 3 or 7 days, the cultured GV cells were adjusted to 10⁸ CFU/20 μL in sterilized 50 mM PBS (pH 7.0) bacterial suspension and inoculated into the vagina of mice at 0.5–1 cm with a pipette once a day for 6 days, and the blank group was injected with the same dose of PBS every day. The mice were rested for 3 min after each inoculation by lifting their tails. After serial inoculations were completed, 50 μL of PBS was pipetted into the mouse vagina and gently pipetted up and down to obtain mouse vaginal secretion washing solution. RNA was extracted from the secretion, reverse transcribed, and qPCR was used to detect the amount of GV and lactobacilli in the mouse secretion. The whole operation flow is shown in Fig. S4.

The successfully infected mice were divided into 3 groups for the follow-up treatment once a day for 6 days. Group 1 was treated with 20 μg/mL LasA, group 2 was treated with the same concentration of antibiotics, and the control group was treated with an equal volume of PBS buffer. The antibiotics for strain ATCC14019 and JNFY14 were metronidazole (MTZ) and erythromycin (ERM), respectively.

6. Ethics statement

All animal experiments were evaluated and approved by the Ethical Committee of Shandong University, with animal welfare prioritized throughout the experimental protocols. The approval number for this study is SYDWLL-2022-072.

7. Smear microscopy of vaginal secretions

Microscopic examination of vaginal secretion smear was performed with reference to the steps of Gram stain microscopy. The secretion was suspended in 0.9% NaCl (w/w) and the number of clue cells was summed by hemocyte counting plates. At least three smears were prepared for each sample.

7.1. Observation of vaginal tissue section

To evaluate the safety of LasA, vaginas of healthy mice were washed with 20 μg/mL and 200 μg/mL once per day, respectively. Mice in the control group were washed with 50 mM PBS buffer (pH 7.0). Mice were sacrificed at 3 d or 6 d to harvest vaginas and washed with PBS. The clean vaginal tissue from each mouse were fixed in 10% buffered formalin phosphate at room temperature followed by paraffin embedding. Histological slide preparation and H&E staining were performed according to previous study [55]. Then the stained sections were visualized on an Olympus IX71 microscope and epithelial thickness was measured using ImageJ 1.54i software, with averages calculated from six measurements per section.

7.2. Proteomics analysis of mice vaginal epithelial tissue

Tissues were first homogenized and then SDC buffer (5% SDS, 100 mM This-HCl, pH 8.5) was added. The lysates were further sonicated and boiled for 15 min. After being centrifuged at 25,200×g for 40 min, the supernatant was digested and the peptides from each sample were analyzed by OrbitrapTM AstralTM mass spectrometer connected to an Vanquish Neo system liquid chromatography (ThermoFisher Scientific, USA) in the data-independent acquisition (DIA) mode.

7.3. RNA extraction and biomass assay

Secretion RNA was extracted using magnetic bead method of tissue & cell RNA extraction kit (Accurate biology, Changsha, China). Then the RNA was reversely transcribed by the prime script Ⅱ 1st strand cDNA synthesis kit (TransGen, Beijing, China) and qPCR was performed on a CFX96TM thermal cycler (Jena, Germany) with the following cycling parameters: 3 min at 95 °C, followed by 40 cycles of 10 s at 95 °C, 10 s at 60 °C and 15 s at 72 °C. For relative cell quantification, target genes from the conserved and variable regions of 16S rDNA were cloned into the pEASY-T1 vector (TransGen, Beijing, China). The recombinant vectors were subjected to gradient dilution, followed by qPCR. Standard curves of gene copies and Ct values were plotted. Change Biomass were described as the delta gene copies after LasA or antibiotic treatments. The oligonucleotide primers were designed using SnapGene software (www.snapgene.com) (Table 2).

Table 2.

Primers used in qPCR.

Primers	Sequence (5′-3′)	Quantitation for:
universalF [56]	CTACGGGAGGCAGCAGT	Total bacteria
universalR	CCTACGTATTACCGCGGCTG
GV-PF [57]	GGGCGGGCTAGAGTGCA	G. vaginalis
GV-PR	GAACCCGTGGAATGGGCC
LactoF [58]	TGGAAACAGRTGCTAATACCG	lactobacilli
LactoR	GTCCATTGTGGAAGATTCCC

7.4. Statistical analysis

Summary statistics for the GV biofilm disassembly assay, GV cell walls degradation assay, vaginal infection murine model, and LasA treatment assay are presented as counts and percentages or means ± SDs. All statistical analysis were performed using GraphPad Prism 9.5.1.

CRediT authorship contribution statement

Kundi Zhang: Writing – original draft, Visualization, Validation, Supervision, Methodology, Funding acquisition, Data curation. Mengyao Lu: Writing – original draft, Visualization, Validation, Methodology, Data curation. Shangyi Fu: Visualization, Validation, Data curation. Chengcheng Jiang: Validation, Methodology. Xinyu Dong: Validation, Methodology. Tingting Liu: Validation, Methodology. Yifei Chen: Validation, Methodology. Xin Li: Visualization. Sujuan Xu: Supervision. Hainan Su: Visualization. Sifeng Jia: Validation, Methodology. Jian Zhang: Writing – review & editing, Supervision. Lichuan Gu: Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Kundi Zhang reports financial support was provided by National Key Research and Development Program of China. Kundi Zhang reports financial support was provided by Natural Science Foundation of Shandong Province. Lichuan Gu has patent #ZL 202210347294.X licensed to Stars Medical Co. Ltd. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Sequence data that support the findings of this study have been deposited in GenBank with the accession No. CP083171.1.