Bioinspired gelated cell sheet–supported lactobacillus biofilm for aerobic vaginitis diagnosis and treatment

Abstract

Aerobic vaginitis (AV) is a long-standing inflammatory disease that affects female patients. The use of antibiotics is a common means for AV treatment, but it will indiscriminately kill both pathogenic bacteria and beneficial strains, which easily causes vaginal dysbacteriosis and infection recurrence. Herein, we describe a bioinspired strategy for fabricating gelated cell sheet–supported lactobacillus biofilms (GCS-LBs) for AV treatment. Compared with common planktonic probiotic formulations, probiotic biofilms forming on a robust GCS exhibit enhanced stress tolerance and better colonization capacity in the mouse vagina. Moreover, DNA nanodevices are decorated on the GCS and dynamically report the microenvironment change of biofilms for timely evaluating bacterium activity, both in vitro and in vivo. Consequently, GCS-LBs are used for treating AV in an Escherichia coli–infected mouse model, which shows enhanced therapeutic efficacy compared with conventional antibiotic or lactobacillus monotherapy. Overall, the GCS-LB shows promise as a potent multifunctional tool to combat bacterial infection.

A strategy to fabricate gelated cell sheet–supported lactobacillus biofilm was developed for vaginitis treatment.

INTRODUCTION

Aerobic vaginitis (AV) is a vaginal infection characterized by mucosal congestion, pain, inflammation, an increase in aerobic bacteria, and a decrease in lactobacilli, which can even increase the risk of sexually transmitted diseases (e.g., HIV) and preterm labor (1, 2). It is one of the most common vaginal infections, with a prevalence ranging from 12 to 23.7% among nonpregnant women and 4 to 8% during pregnancy (3). In clinical routine, empirical use of antibiotics is the main treatment method for AV; however, antibiotics indiscriminately kill pathogenic bacteria and beneficial strains in the vagina (4), leading to an imbalance in the vaginal flora and a high recurrence rate. Moreover, the overuse of antibiotics may cause the development of multidrug-resistant pathogenic bacteria, further compromising therapeutic effects.

Lactobacillus plays critical roles in the maintenance of vaginal homeostasis and health through several mechanisms, including competitive exclusion of pathogenic bacteria, secretion of antibacterial substances (e.g., lactic acid, H₂O₂, and antimicrobial peptides, etc.), and modulation of the immune system of the host (5). Thus, the delivery of living lactobacilli into the pathological vagina has been an acceptable means to fight against bacterial vaginosis and modulate the vaginal microenvironment (6). Despite its usefulness, the therapeutic efficiency of probiotic monotherapy remains low (7, 8) as most of the administered probiotics are rapidly cleared away or lose activity before successful colonization and survival in an exclusive vaginal environment. To achieve durable effects, multiple doses of lactobacilli must be administered to improve cell colonization, but this substantially increases treatment costs and reduces patient compliance. Recently, some attractive strategies for engineering probiotics have been developed for protecting them from hostile conditions and enhancing their colonization ability (9–12), which shows great promise in biomedical applications. However, most of these methods are designed for treating enteric diseases (13–16), and few studies on maintaining female reproductive health have been reported. In addition, it remains a big challenge to dynamically evaluate probiotic activity in vivo, resulting in unpredictable treatment outcomes in real applications. Therefore, there is an urgent need to develop new therapeutic probiotic formulations for the diagnosis and treatment of vaginitis.

Biofilm, a thick layer of bacteria that aggregate to form a colony, is a vital adaptation and survival strategy commonly shared by bacteria (17). Bacteria in the biofilm produce various soluble microbial products and extracellular polymeric substances (EPSs) to protect them from hostile environments (18). For example, biofilm cells have a very high tolerance to antibiotics (10 to 1000 times more than planktonic cells) (19). In terms of the excellent stress tolerance of biofilms, we therefore hypothesized that the application of lactobacillus biofilms would be an alternative to develop next-generation probiotic therapy with enhanced efficiency for bacterial diagnosis and therapy.

Naturally, lactobacilli adhere to the vaginal epithelial cells, which serve as an ideal substrate for probiotic adhesion and colonization and provide adequate nutrition (i.e., glycogen) for maintaining bacterial activity and proliferation (5). Inspired by these arguments, we report a strategy to fabricate gelated cell sheet–supported lactobacillus biofilms (denoted as GCS-LBs) for diagnosing and treating AV. As shown in Fig. 1A, vaginal epithelial cells are cultured until an intact cell sheet forms on the plate, followed by intracellular gelation treatment (20, 21). This process not only retains the main characteristics of epithelial cells but also greatly improves cell stability because of the formation of intracellular hydrogel cores, making them suitable for both in vitro and in vivo applications. Then, lactobacilli were seeded on the GCSs, and excess bacteria were removed by thorough washing. After culturing in de Man–Rogosa–Sharpe (MRS) broth media for 48 hours, a thick probiotic biofilm forms on the cell sheets, which can be then encapsulated and collected for subsequent use.

Fig. 1. Construction and application of GCS-LB for monitoring of the biofilm microenvironment and AV treatment.

(A) Fabrication of the GCS-LB. (B) Application of GCS-LB in monitoring of the biofilm microenvironment. (C) Application of GCS-LB in AV treatment.

Because of the GCSs, lactobacilli with relatively poor biofilm formation capacity can still form a thick biofilm and protect them from different adverse conditions. Compared with planktonic lactobacilli (PLA), GCS-LBs have several attractive features: (i) enhanced resistance capacity against different stresses, (ii) longer vagina reservation time in the mouse vagina, (iii) flexible decoration of a DNA nanodevice for dynamically monitoring probiotic activity (Fig. 1B), and (iv) better therapeutic efficiency for bacterial vaginosis, as demonstrated in a mouse model (Fig. 1C). Collectively, we anticipate the formation of probiotic biofilms on a bioactive substrate to be a universal approach for the preparation of cell-bacteria architectures with improved functionality and believe that probiotic biofilms represent a useful tool for biomedical applications with enhanced therapeutic efficacy.

RESULTS

Preparation and characterization of GCS-LBs

We chose human vaginal epithelial cells (VK2/E6E7) as the cell brick and Lactobacillus acidophilus (L. acidophilus) as the probiotic strain because it has already been used to modulate the vaginal microenvironment and combat microbial infection with high safety (22). Also of note, L. acidophilus is a commonly used strain for treating human vaginitis, which has excellent antibacterial and antifungal effects and can be found in some commercially available products (23). After culturing VK2 cells in a 6-well plate, an intact two-dimensional (2D) cell sheet was conveniently harvested. The cell sheets were then reinforced using an intracellular gelation technique according to previous reports. No morphology changes in the cell sheets after intracellular gelation were observed (Fig. 2A), suggesting that the main cellular characteristics were well preserved. Successful gelation was confirmed by fluorescence microscopy (fig. S1) and enhanced stability (vide infra). This gelation step was essential because only GCSs could endure a long-term bacteria culture process in the MRS medium. When challenged with pure water, lactobacilli culture medium, and long-term storage (30 days), native cell sheets rapidly ruptured and detached from the plate surface, whereas GCSs retained intact structures without obvious changes (Fig. 2B), which offered a stable platform for bacterial adhesion, proliferation, and biofilm formation.

Fig. 2. Preparation and characterization of GCS-LBs.

(A) Optical images of the VK2 cell sheet before and after intracellular gelation. White scale bar, 50 μm. Black scale bar, 10 μm. (B) Nongelated or gelated cell sheets (NCS and GCS, respectively) after different treatments, including pure water (PW), lactobacilli culture medium (LCM), and long-time storage (LTS) in PBS (30 days). Scale bars, 10 μm. (C) Optical images showing the formation of the lactobacilli biofilm on the GCS at different time points. Arrows show the margin of the bacterial colony. (D) Absorbance of lactobacilli (LA), GCS, and GCS-LBs after staining with crystal violet. Insets show the corresponding staining images. (E) Confocal images of the mature process of the lactobacilli biofilm on the GCS. (F) SEM images of the GCS after incubating PLA for 2 hours and the lactobacilli biofilm on the GCS after 48 hours of culture. Blue arrows show the adsorbed lactobacilli on the GCS. Red arrows show the network-structured EPSs in the biofilm. (G) Glycogen and (H) protein content of GCSs determined using commercial kits. Error bars represent the SD (n = 3). Statistical significance was analyzed using unpaired two-tailed Student’s t test, giving P values, ***P < 0.001, and ****P < 0.0001; NS, not significant.

Next, we seeded Lactobacillus on the GCSs and investigated the formation of the biofilm. Optical microscopy was used to monitor the formation of biofilms. Compared with the initial GCSs, numerous bacteria were clearly observed on the cells after 2 hours of incubation (Fig. 2C and fig. S2), indicating successful lactobacilli colonization on the GCSs. After culturing for 12 hours, small bacterial films could be clearly observed on the GCSs, indicating that lactobacilli could successfully proliferate on the GCSs. Then, these bacterial colonies became larger and thicker and lastly covered most of the areas after 48 hours. Crystal violet staining results revealed the enhanced biofilm formation ability of Lactobacillus on the GCS surface compared to a common polystyrene surface (Fig. 2D), which was probably due to the highly biocompatible cell surface for lactobacilli adhesion and survival. As shown in Fig. 2E, the confocal laser scanning microscopy (CLSM) results demonstrated that the thickness of biofilms gradually reached a plateau until 48 hours (~30 μm; Fig. 2E). Of note, GCSs offered a favorable environment for biofilm formation, as evidenced by the incomplete biofilms with a thickness of only ~10 μm on a polystyrene plate after 48 hours of culture (fig. S3). Scanning electron microscopy (SEM) confirmed the presence of numerous lactobacilli on the GCSs after 48 hours (Fig. 2F), and abundant EPSs were observed in the biofilm, further verifying the successful formation of the lactobacillus biofilm. Compared with traditional cell fixing methods, the cellular gelation process did not involve toxic reagents (e.g., aldehydes), which offered a highly biocompatible surface for lactobacilli adhesion. To test this, we also prepared 4% paraformaldehyde and 2.5% glutaraldehyde-fixed cell sheets. Unexpectedly, few biofilms were observed on these fixed cell sheets even after 48 hours (fig. S4), which was possibly due to the toxic effect of chemicals on cell sheets. GCSs, as a native material, were rich in bacteria-required nutrition, with glycogen and protein contents of 13.7 and 0.575 g g⁻¹, respectively (Fig. 2, G and H). This property was different from that of other man-made polymer materials, so we tested if lactobacilli could absorb these nutrients. When GCS-LBs were cultured without carbon and nitrogen sources, gelated cells were gradually digested by lactobacilli (fig. S5) and the bacteria remained active (vide infra), consistent with our speculation.

In vitro resistance of GCS-LBs to environmental assaults

Probiotics frequently endure many challenges when transplanted into the pathological vagina. To determine the capacity of GCS-LBs against survival stresses, we evaluated their tolerance to antibacterial agents (i.e., antibiotics and copper ions), nutritional deficiencies, and high temperatures. Of note, we used PLA as control because it was widely used in currently available Lactobacillus preparations. The production of lactic acid is the main characteristic of lactobacilli; therefore, we tested the pH value of culture media after adding PLA. As shown in fig. S6, the pH value of the alkalized MRS media containing PLA rapidly decreased from 6.5 to 4.2 after 14 hours, suggesting the production of lactic acid by PLA. However, when PLA was pretreated with levofloxacin (Lev) (a broad-spectrum antibiotic) for 1 hour and then cultured in alkalized culture media, the media could not be acidized, which was mainly ascribed to the loss of bacterial activity. Consequently, we can conveniently use pH change to rapidly assess Lactobacillus activity because a conventional plate culture assay often requires at least 36 hours. As shown in Fig. 3A, we conducted Live/Dead experiments to characterize the probiotics before and after incubating Lev. Numerous PLA showed red fluorescence after drug treatment (Fig. 3A), suggesting that these bacteria were killed by antibiotics. In sharp contrast, the overwhelming majority of bacteria survived in the biofilm, as evidenced by strong calcein fluorescence and negligible prodium iodide (PI) fluorescence. Then, the probiotic activity was also evaluated by pH change, and the results are shown in Fig. 3B. Consistent with the results in Fig. 3A, PLA was highly sensitive to Lev, whose ~75% activity decreased even at the lowest antibiotic concentration. However, GCS-LBs retained ~58% activity even after treating the highest concentration of Lev. Notably, this pH decrease was not due to GCSs (fig. S7), thus confirming that GCS-LBs could efficiently protect bacteria from antibiotic attack. A standard plate culture assay was performed, and the same trends were observed (fig. S8), further verifying the antibiotic resistance of GCS-LBs. The enhanced resistance capacity of GCS-LBs was also observed after exposing the bacteria to another antibacterial agent, i.e., CuSO₄ (Fig. 3C). For curing vaginitis, women frequently use antimicrobial agents; nevertheless, these drugs indiscriminately kill both pathogenic and probiotic bacteria, making it difficult to restore the normal vaginal ecosystem. Because of the antibiotic-resistant GCS-LBs, it may offer a probability to cooperatively apply antibiotics and probiotics for vaginitis treatment.

Fig. 3. In vitro resistance of GCS-LBs to environmental assaults.

(A) Live/Dead images of PLA and GCS-LBs treated with Lev (12.5 μg ml⁻¹) for 1 hour. Green fluorescence indicates living bacteria, and red fluorescence indicates dead bacteria. Scale bars, 10 μm. Activity of PLA and GCS-LBs after treatment with (B) Lev, (C) copper ions, (D) high temperature, and (E) nutrition deficiency. (F) Stress-associated gene expression profiles of GCS-LBs. Antimicrobial capacity of GCS-LBs against (G) E. coli, (H) S. aureus, and (I) C. albicans. Error bars represent the SD (n = 3). Statistical significance was analyzed using unpaired two-tailed Student’s t test, giving P values, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; NS, not significant.

When probiotics are transplanted into the pathological vagina, a competitive environment and limited nutrition conditions may cause their death or low activity. To test this, active PLA was transferred into a nutrient-deficient solution. As shown in Fig. 3D, as the culture time was prolonged, the activity of PLA markedly decreased. Impressively, the activity of PLA declined by more than 95% over 60 hours, which could not be restored even after adding fresh media. As a comparison, the activity of the biofilm only fell ~50% at the same incubation time, which was largely because bacteria could directly acquire nutrition from the gelated cells, which was critical for their long-term survival. Live probiotics always require appropriate external conditions (e.g., cold-chain storage), and a high temperature rapidly causes bacteria death. To test this, both PLA and GCS-LBs were exposed to high temperatures ranging from 37° to 65°C for 1 hour. Although the activity of the two groups gradually decreased (Fig. 3E), the activity of GCS-LBs was consistently higher than that of PLA at all tested temperature points. Impressively, almost all PLA lost their activity after culturing at 65°C for 1 hour, whereas GCS-LBs retained 20% activity.

To investigate the underlying mechanism of GCS-LB resistance to adverse conditions, the expression of critical genes (24) was analyzed. Concordantly, the high levels of mRNA expression of stress proteins such as heat shock protein (hsp), molecular chaperone DnaK (dnak), molecular chaperone GroEL (groel), 10-kDa chaperonin GroES (groes), adenosine 5′-triphosphate–dependent Clp protease proteolytic subunit (clpp), and catalase (cat) in GCS-LBs may explain their high survival rate under external stress conditions (Fig. 3F).

Lactobacilli can produce abundant antimicrobial chemicals, so we further examined the antibacterial and antifungal capacity of the fabricated GCS-LBs using three representative pathogens, i.e., Escherichia coli (E. coli), Staphylococcus aureus (S. aureus), and Candida albicans (C. albicans). As expected, the biofilm could efficiently kill high concentrations of E. coli and S. aureus in the solution (Fig. 3, G and H). Although the biofilm did not completely eradicate C. albicans, it significantly inhibited the growth and hyphal formation of C. albicans (Fig. 3I and fig. S9). Thus, the lactobacillus biofilm had a broad-spectrum antimicrobial capacity.

Colonization of the GCS-LB and its effect on modulating the vaginal microenvironment

Encouraged by the enhanced survival of GCS-LBs under adverse conditions, we next examined if GCSs could enhance the colonization of lactobacilli in the vagina. Because the surface of the vagina mucosa is covered by a layer of glycoproteins (25), we first tested the adhesion of gelated VK2 cells (GVKs) on a mucosa-mimicking surface. As depicted in Fig. 4A, suspended GVKs were barely adsorbed on a bare cell plate without a mucin coating, implying their weak binding capacity on the polystyrene surface. However, once the plates were covered with mucins, the adsorbed number of GVKs on the surface significantly increased, indicating that GVKs could strongly interact with mucins via noncovalent interactions. The large-sized GCS-LBs also firmly bound to the mucin-coated surface (Fig. 4B), which could not be removed even by repeated washing with phosphate-buffered saline (PBS). Also of note, GCS-LBs were easily subjected to physical operations such as scraping and pipetting during collection and use, which caused biofilm damage or even collapse. To address this, a thin alginate hydrogel was used to coat the biofilms to protect them from the external environment. As expected, because of the encapsulation of the alginate hydrogel, the morphology of GCS-LBs was well preserved without obvious collapse after repeated scraping and pipetting (fig. S10), which proved the good mechanical stability of probiotic biofilms. Also of note, the coating of a thin alginate hydrogel did not affect the mass exchange between the biofilm and external environment because the porous structure of the hydrogel layer still allowed free diffusion of antibacterial substances and other functional metabolites produced by lactobacilli, as evidenced by the excellent antibacterial capacity of alginate-coated GCS-LBs (vide supra). Thus, alginate-coated GCS-LBs were used for both in vitro and vivo.

Fig. 4. Colonization of the GCS-LB and its effect on modulating the vaginal microenvironment.

(A) Binding of GVKs on the surface coated with different concentrations of mucins. Inset shows optical images of cells adsorbed on mucin-coated surfaces. Scale bars, 20 μm. (B) GCS-LB number on the mucin-coated surfaces before and after three PBS rinses. Insets show corresponding confocal images of GCS-LBs on the surface. Scale bars, 50 μm. (C) In vivo imaging system measurements of mice after vaginal administration of PLA (1 × 10⁷ CFUs) or GCS-LBs (1 × 10⁶ CFUs) at different time points. Cells are labeled with Cy7. (D) Retention rate of PLA and GCS-LBs in mouse vaginas at different time points. (E) Plate images and (F) lactobacilli colonies in the vaginal secretions collected from mice after injecting PLA (1 × 10⁷ CFUs) and GCS-LBs (1 × 10⁶ CFUs) for 48 hours. Mice without any treatment were used as negative controls. Scale bars, 100 μm. (G) Lactobacillus colonies in vaginal secretions collected from mice treated with PLA (1 × 10⁷ CFUs) and GCS-LBs (1 × 10⁶ CFUs) for 12 hours in the absence or presence of Lev. Error bars represent the SD (n = 3). Statistical significance was analyzed using unpaired two-tailed Student’s t test; NS, not significant.

After verifying the strong binding between GCS-LBs and mucin-coated surfaces, we used an in vivo imaging system (IVIS) system to evaluate lactobacillus colonization in the mouse vagina using Cy7-labeled bacteria. As shown in Fig. 4C, a significantly weaker fluorescent signal was observed after the mice were treated with PLA at 4 hours after injection. These results indicated that common PLA could be excreted quickly from the body, and the signal completely vanished at 48 hours, implying that a single inoculation was not sufficient for successful lactobacillus colonization. In contrast, the retention of GCS-LBs in the vagina was markedly prolonged, which could still be seen even at 96 hours after injection. Also of note, the lactobacilli fluorescence even increased in the initial 8 hours (Fig. 4D), which was probably due to rapid bacterial proliferation.

Meanwhile, vaginal tissue samples were collected at 48 hours after injection, and histofluorescence experiments were conducted. As shown in Fig. 4E, negligible lactobacillus fluorescence was observed in the PLA-treated mice, suggesting the limited colonization capacity of common lactobacilli, consistent with the IVIS images. In contrast, strong fluorescence signals could still be seen in the mouse vagina treated with GCS-LBs, thus confirming the enhanced colonization capacity of GCS-LBs compared with PLA. To further test if the injected lactobacilli were still alive, a plate assay was conducted by culturing the vaginal secretions collected from the mice at 24 hours after injection. Consistent with the IVIS results, the colony number of the PLA-treated mice was much less than that of mice treated with GCS-LBs (Fig. 4F), indicating that most lactobacilli in the biofilm survived in the mouse vagina despite the marked changes in the living environment. The injected probiotic biofilms retained their drug resistance ability (Fig. 4G), which could endure a high concentration of antibiotics compared with convention PLA, implying the possibility of coadministration of probiotics and antibiotics for combating complex bacterial infection. Collectively, the formation of the biofilm not only improved the colonization capacity of lactobacilli in vivo but also contributed to its survival under hostile conditions.

Dynamic monitoring of pH changes of GCS-LBs in vitro and in vivo

Abnormal pH levels in vaginal fluids are a hallmark of vaginitis (26), and pH test strips are frequently used to measure this parameter using vaginal secretions. However, this method suffers from low sensitivity and inconvenience because it cannot dynamically report probiotic activity or microenvironment change in vivo during treatment, especially in a small-animal model. Currently, various vaginal sensors have been developed to analyze physical [e.g., vaginal electromyography (27), temperature (28), and pressure (29)] or chemical clues [e.g., pH (30, 31)], proteins, and other metabolites (32). Nevertheless, the development of nanobiosensors for dynamically monitoring vaginal pH remains unexplored (33). Thus, we explored the use of a pH-sensing DNA nanodevice in our system for dynamically monitoring pH changes around the biofilm, which could be further used to evaluate the change of the vaginal microenvironment.

To achieve this goal, we designed an acid-sensitive DNA duplex, which comprised an i-motif strand (IS) and a signal strand (SS). Specifically, SS had a cholesterol moiety at the 3′ end, a Cy5 fluorophore (E_x/E_m = 650/670 nm) at the 5′ end, and a Cy7 fluorophore (E_x/E_m = 750/780 nm) at the middle nucleotide. IS was labeled with a cholesterol moiety at the 5′ end and a quencher [Black Hole Quencher 3 (BHQ3)] at the 3′ end (Fig. 5A and fig. S11). Initially, SS could hybridize with IS and form a duplex structure, and the close proximity between Cy5 and BHQ3 caused efficient fluorescence quenching because of the fluorescence resonance energy transfer effect. However, as the pH value decreased, IS rapidly folded into a compact i-motif structure (34), leading to duplex dehybridization and the recovery of Cy5 fluorescence. In this system, the “always-on” Cy7 fluorophore that was not sensitive to pH change generated a stable fluorescence signal, which could directly locate the biofilms in vivo. Moreover, this ratiometric fluorescent probe with two independent fluorescence peaks can markedly reduce the limitations of single-signal response probes, such as bleaching and background fluorescence in biosensing (35).

Fig. 5. Dynamic monitoring of pH changes of GCS-LBs in vitro and in vivo.

(A) Schematic illustration of the working principle of the pH-responsive DNA nanodevice. (B) Flow cytometry analysis of DNA-GVKs responding to PBS solutions with different pH values. (C) pH-dependent fluorescence dynamic ranges of the DNA nanodevice. (D) Confocal images of GCS, SS-modified GCS, and SS/IS-decorated GCSs. Scale bars, 20 μm. (E) Confocal images and (F) corresponding fluorescence intensities of SS/IS-decorated GCS-LBs initially cultured in a neutral MRS buffer (pH 7.0) for different time points. The top panels in (E) show the merged images of the Cy5 and Cy7 channels, and the bottom panels show the corresponding ratiometric (Cy5/Cy7) images. Scale bar, 10 μm. (G) Confocal images and (H) signal ratio changes demonstrating the reversibility of the DNA-GCS-LB response to external pH changes. (I) In vivo imaging system measurement of mice after vaginally administrating PBS, GCS-LBs (1 × 10⁶ CFUs), DNA-GCS-LBs (1 × 10⁶ CFUs), and DNA-GCS-iLB (1 × 10⁶ CFUs) at different time points. Error bars represent the SD (n = 3).

To test the pH sensing capacity of the DNA nanodevice, we first modified it on GVKs and then used flow cytometry to determine the fluorescence changes under different pH conditions. As shown in Fig. 5B, between pH 4.5 and 7.0, the Cy5 signal decreased by 22.3-fold, whereas the Cy7 signal remained constant. The fluorescence ratio between Cy5 and Cy7 (I_cy5/I_Cy7) exhibited a sharp increase in the pH range from 6.0 to 5.0, which was consistent with the acidification range induced by lactobacillus fermentation (Fig. 5C). The signal response was highly specific to pH change and could not be disturbed by other substances such as DNA, proteins, monosaccharides, antibiotics, or cells (fig. S12).

Then, the DNA nanodevice was introduced into our GCS system, which was used to monitor bacterial activity in a real-time manner. As evidenced by CLSM analysis (Fig. 5D), strong Cy5 and Cy7 fluorescence signals were observed on the cell sheets after incubation with SSs, thus confirming successful DNA modification. After adding ISs, Cy5 fluorescence was significantly suppressed, verifying the successful fabrication of DNA nanodevices on GCSs. The number of DNA molecules on each cell was estimated to be 7.6 × 10⁶ strands cell⁻¹. Then, the fluorescence change of the resulting products (denoted as DNA-GCS-LBs) was dynamically recorded after incubating the biofilm with fresh culture media (pH 7.0). The Cy5 fluorescence of DNA-GCS-LBs rapidly increased within 0.5 hours and then reached a plateau at 1 hour (Figs. 4F and 5E). In comparison to the slow pH change of the bulk solution (more than 10 hours), these results indicated rapid acidification of the biofilm, which could be explained by the high local concentration of lactic acids in the biofilm and the close distance between DNA nanodevices and bacteria. Moreover, this nanodevice was highly reversible because DNA molecules could freely move on the cell membranes, as evidenced by the cyclical fluorescence changes driven by a pH change (Figs. 4H and 5G). Furthermore, DNA nanodevices on the GCS-LB were rather stable, which still reported pH change after continuous culture for at least 5 days (fig. S13). Collectively, these interesting properties of the DNA nanodevice were vital for timely evaluation of probiotic activity.

Next, we injected DNA-GCS-LBs into the mouse vagina to evaluate lactobacilli activity in vivo. As shown in Fig. 5I and fig. S14, DNA-GCS-LBs in the healthy vagina could be easily distinguished using Cy7 fluorescence, and the Cy5 signal reached a maximum at 4 hours after injection, suggesting the production of lactic acids in the mouse vagina by lactobacilli. The activation of DNA nanodevices in vivo was obviously slower than that in culture media, which was probably ascribed to the longer adaptive time of lactobacilli in the mouse vagina. Of note, the overall pH value of the mouse vagina remained slightly acidic (fig. S15), which, in principle, could not activate DNA nanodevices. Thus, the fluorescence response was largely owing to the microenvironment change in the biofilm. In contrast, when the heat-inactivated DNA-GCS-LB (denoted as GCS-iLB) was injected into vagina, no Cy5 signal was observed during the experiment, indicating that this signal resulted from probiotic activity. Furthermore, the influence of antibiotics (e.g., Lev) on lactobacilli in the mouse vagina was assessed. DNA nanodevice–modified GCS-supported lactobacilli (DNA-GCS-LA) were used as a control. After injecting DNA-GCS-LA into the mouse vagina, the Cy5 signal could also be detected, indicating that the adsorbed bacteria were active. However, when the mice were treated with Lev, the Cy5 signal quickly disappeared within 4 hours (fig. S16), which was largely attributed to the loss of Lactobacillus activity. In contrast, the fluorescence signal was not profoundly affected by antibiotics for the DNA-GCS-LB–treated mice, thus revealing that the probiotic activity in the biofilm was not disturbed by antibiotic administration. Thus, we confirmed that the developed DNA nanodevice could dynamically and remotely monitor probiotic activity in vivo without tedious secretion collection and pH testing steps.

Also of note, pH is a key parameter for vaginitis diagnosis or prognosis evaluation because patients frequently show a remarkable pH increase, which can be also analyzed by our DNA nanodevice. However, rodents maintain near neutral vaginal pH (~6.8 to 7.0), which is in contrast to the acidic vaginal pH (4.0 to 4.5) in women. Humans are one of the few mammalian species that harbor an acidic vaginal pH, so we hope to recruit volunteers to verify the unique analytical performance of DNA-modified probiotic biofilms in the future.

Treatment of E. coli–induced vaginitis using GCS-LBs

The excellent antiresistance and colonization capacity of GCS-LBs inspired us to further investigate its potential in preventing bacterial vaginitis. To this end, female BALB/c mice were first treated with β-estradiol for 1 week to disturb the vaginal microenvironment and then administered E. coli for 5 days [1 × 10⁸ colony-forming units (CFUs) per mouse per day] to cause bacterial vaginitis. Subsequently, these mice were randomly divided into four groups and subjected to different treatments, i.e., (i) PBS, (ii) Lev, (iii) PLA, and (iv) GCS-LB. The treatment strategies are shown in Fig. 6A. All mice were euthanized, and key parameters were assessed for evaluating the therapeutic effect after 8 days. We used E. coli as the pathogenic strain because it is one of the most common bacteria that cause AV and severe local inflammation compared with other anaerobic bacteria (36).

Fig. 6. Treatment of E. coli–induced vaginitis using GCS-LBs.

(A) Overview of the four treatment schedules for bacterial vaginitis. The infected mice treated daily with PBS solution were used as a control. (B) Colonies of E. coli and (C) lactobacilli in the vaginal secretions after different treatments. Levels of (D) H₂O₂, (E) LE, and (F) BAs in the vaginal secretions of E. coli–infected mice after different treatments. (G) Representative photos of E. coli–infected mouse introitus before and after treatment. (H) Representative H&E staining images of vaginal tissues after different treatments. Scale bars, 50 μm. Black boxes and arrows show inflammatory cell infiltration. Red arrows indicate the ulcerative epithelia. The levels of (I to K) proinflammatory and (L) anti-inflammatory cytokines in vaginal tissues from treated mice were measured using ELISA kits. Error bars represent the SD (n = 6). Statistical significance was analyzed using unpaired two-tailed Student’s t test, giving P values, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; NS, not significant.

We first tracked CFUs in vaginal secretions before and after treatment. For the PBS group, the number of E. coli in the vaginal secretions was high (10⁵ to 10⁶ CFUs ml⁻¹), indicating the successful establishment of bacterial vaginitis models. After treatment, PLA monotherapy led to variable treatment outcomes, whereas GCS-LBs had a stable inhibitory effect on pathogenic bacteria, killing 99.99% of E. coli. Of note, this outstanding antibacterial effect was comparable to antibiotic therapy. Of note, the GCS itself did not have any antibacterial activity (fig. S17). The survival of lactobacilli was also investigated. As shown in Fig. 6C, although the mice were injected with PLA each day, most lactobacilli could not survive in some mice because of the hostile vaginal microenvironment, as evidenced by the limited colony numbers. In contrast, a large number of probiotic colonies could be stably detected in the GCS-LB group, which was 72.5-fold higher than that in mice treated with PLA. This enhanced cell vitality in the biofilm may explain its outstanding antibacterial capacity. Also of note, the use of antibiotics could not improve the content of lactobacilli.

We determined three key factors [i.e., H₂O₂, leukocyte esterase (LE), and biogenic amines (BAs)] in the vaginal secretions to evaluate the microenvironment changes in the vagina after treatment. Among them, H₂O₂ is an antibacterial agent produced by lactobacilli (37), which can reflect probiotic activity. LE is an enzyme produced by leukocytes, whose abnormal rise indicates infection and inflammation (38). BAs are massively produced during bacteria breeding, which indicates the bacterial biomass (39). As shown in Fig. 6D, the H₂O₂ level in the infected mice was significantly lower than that in healthy mice, indicating a vaginal microenvironment disorder. Although antibiotics could efficiently kill pathogenic bacteria, however, this monotherapy could not restore H₂O₂ levels, implying that the sole use of antibiotics could not improve lactobacilli quantity or activity. In contrast, after supplementation with lactobacilli, the H₂O₂ level of the infected mouse was markedly improved. Of note, mice treated with a probiotic biofilm plus antibiotic showed comparable H₂O₂ levels to GCS-LB–treated mice, further verifying the feasibility of coadministration of biofilms and antibiotics. In addition, the LE level of mice treated with both biofilms and antibiotics almost reduced to the baseline level (Fig. 6E), which was the lowest among all treated groups, suggesting that local inflammation was successfully suppressed. Last, the BA level of all treated mice significantly decreased compared with that of the infected mice (Fig. 6F), indicating that most E. coli was eradicated, consistent with the plate assay.

After treatment, inflammation in most mice seemed to be relieved, as evidenced by hyperemia and edema dissipation of introitus (Fig. 6G). To better examine the therapeutic effect, hematoxylin and eosin (H&E) staining was used to further assess the quality of the vaginal tissues in different groups (Fig. 6H). As expected, obvious damage in vaginal epithelia and noticeable inflammatory cell infiltration were observed in the PBS group, indicating severe local inflammation. The use of GCS-LBs notably alleviated these symptoms, with continuous epithelia, regular histological structure, and low inflammatory cell infiltration. In contrast, obvious inflammatory cell infiltration was still observed in other treated groups, and even the ulcerative epithelium was still observed in the PLA group. Thus, although antibiotic monotherapy efficiently killed pathogenic bacteria in the vagina, its effectiveness in suppressing inflammation was still limited. Furthermore, the main tissues of the GCS-LB–treated mice were also analyzed using an H&E assay, and no toxicity was observed, suggesting the high biocompatibility of the probiotic biofilm (fig. S18). In addition, we analyzed the cytokine levels of vaginal tissues from different groups using an enzyme-linked immunosorbent assay (ELISA). The high levels of proinflammatory cytokines [tumor necrosis factor–α (TNF-α), interleukin-1β (IL-1β), and interferon-γ (IFN-γ)] in vagina tissues was efficiently down-regulated by GCS-LBs (Fig. 6, I to K), and the low level of anti-inflammatory transforming growth factor–β (TGF-β) was up-regulated (Fig. 6L), further confirming the excellent anti-inflammation effect of GCS-LBs. In comparison, antibiotics or PLA moderately inhibited inflammatory cell infiltration and cytokine expression. Overall, these results revealed that probiotic biofilms not only efficiently eliminated pathogenic bacteria but also successfully inhibited local inflammation in AV mice.

Influence of different treatment strategies on the vagina microbiome

Next, we investigated the potential of GCS-LBs to modulate the composition of the vaginal microbiome in an E. coli–infected mouse model. The vaginal secretions of mice analyzed by 16S ribosomal RNA (rRNA) gene sequencing revealed that the abundance [operational taxonomic unit (OTU) richness] and α-diversity (Chao and Shannon indices) of bacterial communities in mice treated with GCS-LBs were significantly higher than those of mice subjected to antibiotics and PLA (Fig. 7, A to C), revealing that probiotic biofilms were more conducive to the remodeling of disturbed microflora. Bacterial taxonomic profiling at the phylum level of the vagina microbiota is presented in Fig. 7D. The abundance of Proteobacteria, which is closely associated with vaginal inflammatory diseases, was successfully suppressed in mice treated with GCS-LBs. Specifically, the overgrowth of Escherichia-Shigella, a highly virulent pathogen in Proteobacteria, could be seen in the PBS group, as evidenced by the higher relative abundance. The use of GCS-LBs effectively lowered the abundance of Escherichia-Shigella, whose effectiveness was comparable to that of the Lev groups (Fig. 7, E and F). Meanwhile, the abundance of Lactobacillus in the PBS and Lev groups significantly decreased (Fig. 7G), implying that infection or Lev treatment inhibited lactobacilli survival. In particular, the proportion of Lactobacillus in the GCS-LB group was significantly improved compared with that in the other groups (Fig. 7G), consistent with the plate culturing assay. Also of note, Lactobacillus species are the dominant bacteria in the vagina of healthy women (40), but their abundance in BALB/c mice was significantly low, consistent with previous reports (41, 42). The phylogenetic composition of murine vaginal microbiota is quite different from humans (43); however, Lactobacillus still plays an important role in maintaining vagina health in a mouse model (23, 44). Furthermore, principal components analysis (PCA) showed that GCS-LB–treated mice had a distinct vagina microbiota profile compared with other treatment groups (Fig. 7H). Noticeably, the GCS-LB and healthy groups exhibited significant overlap between their 95% confidence ellipses, suggesting that the microbiota composition in the GCS-LB group was the closest to that in the healthy group. Overall, these results confirmed that the use of GCS-LBs not only inhibited the proliferation of harmful bacteria but also successfully modulated vaginal microbiota.

Fig. 7. Influence of different treatment strategies on the vagina microbiome.

After in situ administration, vaginal secretions were collected and analyzed for gut microbial composition by high-throughput 16S ribosome RNA sequencing. (A) OTU richness and (B and C) α-diversity (Chao and Shannon indices) showing microbial communities (n = 4). Bacterial taxonomic profiling at (D) phylum and (E) genus levels of the vaginal microbiota from E. coli–infected mice after treatment with PBS, Lev, PLA, and GCS-LBs. Relative abundance of (F) Escherichia-Shegella and (G) Lactobacillus from the infected mice after treatment with PBS, Lev, PLA, and GCS-LBs. (H) PCA plot of the microbiota composition of mice after different treatments. Error bars represent the SD (n = 4). Statistical significance was analyzed using unpaired two-tailed Student’s t test, giving P values, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; NS, not significant.

DISCUSSION

In summary, we propose a therapy strategy to treat AV using a bioinspired probiotic biofilm. The probiotic biofilm supported by a bioactive cell sheet can be easily prepared using a standard culture method. Compared with common planktonic probiotics, the GCS-LB exhibits (i) an enhanced stress tolerance against different hostile conditions, (ii) a longer residence time in the mouse vagina, (iii) a general platform allowing the decoration of DNA sensing elements for evaluating probiotic activity, and (iv) a better therapeutic effect because of its excellent antimicrobial and anti-inflammatory abilities. Furthermore, the probiotic biofilm is highly biocompatible, which effectively restores the vaginal microbiota and reshapes a healthier vaginal microenvironment compared with traditional antibiotic therapy. Considering these unique characteristics, we anticipate the broad application of this tool in probiotic-based diagnosis and therapy applications.

MATERIALS AND METHODS

Materials

All used oligonucleotides (IS: 5′-cholesteryl-TTTTTTTCCCTAACCCTAACCCTAACCC-BHQ3-3′;SS: 5′-Cy5-GGGTTAGGGTTAGGGTT/Cy7/TTTTT-cholesteryl-3′), glycogen content assay kit, H₂O₂ content assay kit, and PBS purchased from Sangon Biotechnology Co. Ltd. (Shanghai, China). Dulbecco’s modified Eagle’s medium (DMEM), RPMI 1640 medium, and fetal bovine serum (FBS) were obtained from Gibco (Grand Island, United States). The MRS broth medium was obtained from Park Hope Bio-Technology Co. Ltd. (Qindao, China). 2-Hydroxy-4′-(2-hydroxyethoxy)-2-methylpropiophenone, sodium alginate, poly(ethylene glycol) diacrylate (Mn = 700) (PEG700-DA), and fluorescein O,O′-diacrylate were ordered from Aladdin (Shanghai, China). Bicinchoninic acid (BCA) kit, CCK-8 kit, Calcein-AM, RNA extraction kit, and PI were purchased from Beyotime Biotechnology (Shanghai, China). LE determination kit was bought from Hongye Antibody Biotechnology Co. Ltd. (Guangzhou, China). BA determination kit was purchased from Hezhong Biotechnology Co. Ltd. (Sanming, China). ELISA kits of IL-6, TNF-α, IFN-γ, and TGF-β were obtained from Amyjet Scientific Co. Ltd. (Wuhan, China). L. acidophilus (ATCC4356), E. coli, S. aureus, and C. albicans were ordered from the China General Microbiological Culture Collection Center (CGMCC, China). The VK2/E6E7 cell was obtained from the American Type Culture Collection (ATCC) and cultured in DMEM (Sigma-Aldrich, United States) supplemented with 10% (v/v) FBS and 1% (v/v) antibiotic/antimycotic solution (Sigma-Aldrich, United States) in a 37°C incubator with 5% CO₂. All other reagents were purchased from domestic suppliers and used as received. Deionized and ribonuclease-free water (resistance > 18 megohm·cm) was used throughout the experiments.

Growth of bacteria

Lactobacilli were grown at 37°C overnight in the MRS medium, and E. coli, S. aureus, and C. albicans were grown at 37°C in sterile LB broth media for 12 hours at 200 rpm at 37°C. Bacteria were collected by centrifugation at 4200g for 10 min and resuspended in ice-cold PBS. The concentration of bacteria was detected by measuring the optical density (OD) at 600 nm. Bacterial counts were determined by making dilutions of the bacterial suspension, culturing them on agar plates at 37°C for 24 hours and counting the CFUs.

Animals

Female BALB/c mice from 6 to 8 weeks old (~20 g) were obtained from the Animal Experiments Center of the Shanghai Institute of Materia Medica (Shanghai, China). The study was approved by the ethics committee of Tongren Hospital, Shanghai Jiao Tong University School of Medicine (no. A2023-073-01).

Fabrication and characterization of GCS-LBs

VK2 cells were seeded in a 6-well plate until an intact cell sheet formed. Then, the cell sheet was kept in 500 μl of phenol red–free DMEM containing a 1× protease inhibitor. The intracellular hydrogelation of VK2 cells was according to a previous report (20). Briefly, 1 ml of a gelation buffer was prepared by mixing 100 μl of 2-hydroxy-4′-(2-hydroxyethoxy)-2-methylpropiophenone (1.5 g ml⁻¹) and 900 μl of PEG700-DA. After that, the prepared gelation buffer was added to the cell plate to reach a 10 wt % PEG700-DA concentration. For fluorescence labeling of the gelated cells, fluorescein O,O′-diacrylate was incorporated in the gelation buffer. After 10-min incubation, the cell plate was thoroughly washed by DMEM and then subjected to 365-nm irradiation for 10 min using an ultraviolet (UV) oven (UVP Crosslinker, CL3000, United States). Of note, a 365-nm UV lamp was used to excite the photoinitiator as the wavelength reduces protein denaturation. Next, 1 ml of the PLA (OD_{600 nm} = 1.0) solution was added into the GCS plate and was incubated for 2 hours at room temperature. After removing unbound bacteria, fresh MRS media were added into the cell wells and cultured for 48 hours. Then, the cell plates were carefully washed to remove free bacteria, and 20 μl of CaCl₂ (20 mM) was dropped on the biofilm for 10 min, and then 200 μl of the alginate solution was used to encapsulate the biofilm surface followed by removing excess alginates. The final product (i.e., GCS-LBs) was collected using a cell scraper and redispersed in the PBS solution for subsequent use.

The structure of lactobacillus biofilms was examined by SEM and confocal analysis. Initially, cell sheets or biofilms were fixed in glutaraldehyde (2.5%; Sigma-Aldrich) for 24 hours at 4°C under dark conditions. Cells were then washed and treated with ethanol gradient dehydration (50, 70, 90, and 100%, twice), before being dried using a critical point dryer and coated with platinum sputter. Last, SEM images were obtained on a Hitachi S-4800 FE-SEM at a working voltage of 15.0 kV and a working current of 10 μA under a magnification of 40 K. For measuring the thickness of the biofilm, the GCS-LBs were incubated with Calcein-AM (1 μM) for 20 min. After thoroughly washing, the samples were subjected to confocal microscopy analysis (Carl Zeiss LSM980, Germany) for collecting 3D structures of biofilms. A typical crystal violet staining experiment was performed, and the results were collected by a microplate reader (TECAN Infinite M200, Switzerland). For nutrient detection, the fabricated GCSs were thoroughly homogenized and sonicated, and then the resulting solutions were subjected to glycogen and protein analysis using commercial kits according to the manufacturer’s protocols.

Stability study

Native VK2 cell sheets and GCSs were challenged with pure water for 1 hour, Lactobacillus culture medium (i.e., MRS broth medium) for 1 hour, and PBS buffer for 30 days. After that, the residual cell number was counted using an optical microscope. For each sample, initial cell sheets were used as the positive control. Cell retention (%) = Cell number_treatment/Cell number_initial × 100%.

Protein and glycogen analysis

The prepared GCSs were first lysed through radioimmunoprecipitation assay lysis buffer and sonication treatment to collect protein glycogen. The protein and glycogen concentrations were quantified using a BCA protein assay and glycogen determination kit according to the manufacturer’s protocol. After that, the concentrations of protein and glycogen were calculated according to the standard curve.

Stress tolerance experiments

Equal amounts of PLA and GCS-LBs were added into 1 ml of a medium containing different concentrations of Lev and CuSO₄ for 1 hour. Also, PLA and GCS-LBs were incubated in a heater for 1 hour at different temperatures or were cultured in neutral PBS for different time points. After that, bacterial cells were centrifuged at 5000 rpm for 10 min and resuspended in 1 ml of fresh media overnight to determine the bacterial activity. Activity = (pH_initial − pH_assay)/(pH_initial − pH_end) × 100%, where pH_initial represents the initial pH value of the fresh medium, pH_end represents the pH value of the final media after culturing fully active bacteria, and pH_assay represents the pH value of the final media after culturing PLA or GCS-LBs challenged with different stresses.

Live/Dead cell staining of Lactobacillus

The PLA or GCS-LBs treated with Lev (12.5 μg ml⁻¹) for 1 hour were stained with Calcein-AM and PI for 30 min and washed three times with a sterile 0.9% NaCl solution for visualization via a confocal fluorescence microscope.

Quantitative reverse transcription polymerase chain reaction analysis

Using the TRIzol reagent and adhering to the manufacturer’s instructions, the total RNA of PLA and GCS-LBs was extracted. A NanoDrop system (Thermo Fisher Scientific, United States) was used to assess the RNA’s purity and concentration. Of note, as recommended in the Promega DNase treatment kit, the DNase treatment was provided. cDNA was obtained by reverse transcribing the total RNA using a BeyoRT First Strand cDNA Synthesis Kit performed on an ABI PRISM 7900HT System (Applied Biosystems) using the SYBR Green Master Mix (Thermo Fisher Scientific). Prime sequences were shown in table S1. The 2^−ΔΔCt method was used to calculate the relative gene expression, and 16S rRNA was used as an internal control gene to normalize the expression of target genes.

Antimicrobial test

After forming GCS-LBs, three pathogenic microbes (OD_{600 nm} = 0.1), including E. coli, S. aureus, and C. albicans, were added into GCS-LB wells and incubated for 0, 30, 60, and 90 min. Next, the supernatants containing pathogenic microbes were collected and centrifuged at 5000 rpm for 10 min. After dispersing in the LB culture media, these microbes were incubated at 37°C with gentle shaking. The OD value of cultures was recorded at 600 nm at various time points by a NanoDrop spectrophotometer (Thermo Fisher Scientific, United States).

GCS and Lactobacillus fluorescence labeling

The prepared GCSs (10⁷ cells ml⁻¹) were incubated with NHS-Cy7 (final concentration: 10 μM) in 1× PBS overnight at room temperature, and Cy7-labeled GCSs were collected via centrifugation. For labeling living Lactobacillus, a probiotic biofilm was formed in an MRS medium containing 1 mM 3-azido-d-alanine. After forming the biofilm, excess MRS media were removed and DBCO-PEG4-Cy7 was added and conjugation reaction was conducted for at least 4 hours. After thoroughly rinsing, a Cy7-labeled lactobacillus biofilm was obtained for in vivo colonization study.

Lactobacillus colonization and in vivo survival after vaginal injection

Different concentrations of mucin were used to coat plates overnight. After removing excess proteins, the GVKs (10⁶ cells) or GCS-LBs (1 × 10⁷ CFUs) were added into the mucin-coated plate and incubated for 0.5 hours. After thoroughly washing, the cell number or biofilm was counted using an optical microscope (Motic A2000, China) or a confocal microscope (Carl Zeiss LSM980, Germany).

After anesthetizing the mice with enflurane inhalation, PLA (1 × 10⁶ CFUs), GCS (1 × 10⁵ cells), or GCS-LBs (1 × 10⁵ CFUs) in 20 μl were injected into the mouse vagina. After 4, 8, 12, 24, 48, and 96 hours, the mice were euthanized and imaged using an IVIS system (Tannon ABL X6, China). In the meantime, at 48 hours, vagina tissues and secretions of some PLA-treated or GCS-LB–treated mice were collected for subsequent histofluorescence test and plate assay. Vaginal secretions were cultured using lactobacilli-selective media for 48 hours under anaerobic conditions. The Lactobacillus-selective (LBS) agar is a selective medium for culture of lactobacilli. The ammonium citrate and sodium acetate hydrate in LBS agar inhibit growth of streptococci, fungi, and bacteria other than lactobacilli. The low pH caused by the presence of acetic acid in the medium inhibits growth of bacteria other than lactobacilli. Furthermore, after injecting PLA or GCS-LBs into the mouse vagina, the mice were daily gavaged with 20 mg of Lev. After 48 hours, the vaginal secretions were collected and cultured for lactobacilli counting.

pH analysis in the probiotic biofilm

Equal amounts of SS and IS (final concentration: 2 μM) were added to the prepared GVKs or GCSs in the 24-well plate. After 30 min, excess DNA was removed and the resulting DNA nanodevice–modified GVKs (DNA-GVKs) or GCSs (DNA-GCSs) were obtained. Then, DNA-GCSs could be used for biofilm formation according to the abovementioned protocols, and the final products were named as DNA-GCS-LBs. To confirm the feasibility of the DNA nanodevice for pH analysis, 10 μl of DNA-GVKs (1 × 10⁵ cells) were incubated with 90 μl of different Tris buffers for 15 min followed by flow cytometry analysis (BD FACSCanto II, United States). The buffer used for pH adjustment contained 20 mM Tris and 140 mM NaCl. Acetic acid was added to the above solution until a pH of 8.5 was reached. Periodical pH switchings between 4.5 and 7.5 could then be done by careful additions of HCl and NaOH. For DNA-GCS-LB imaging, DNA-GCS-LB was added into an alkalized MRS buffer (pH 7.0) and then imaged at 0.25, 0.5, 1, and 2 hours using confocal analysis (Carl Zeiss LSM980, Germany) at sequential settings and ×20 magnification (UV settings for visualization of nuclear staining: λ_ex = 405 nm, λ_em = 420 to 452 nm; Cy5 settings for visualization of SS: λ_ex = 633 nm, λ_em = 650 to 700 nm; Cy7 settings for visualization of SS: λ_ex = 633 nm, λ_em 750 to 800 nm).

For in vivo analysis, mice were anesthetized with enflurane inhalation, and then 20 μl of PBS, GCS-LBs (1 × 10⁶ CFUs), DNA-GCS-LBs (1 × 10⁶ CFUs), or inactivated DNA-GCS-LB (DNA-GCS-iLB; 1 × 10⁶ CFUs) was injected into the mouse vagina. After 0, 1, 2, 4, and 8 hours, the mice were euthanized and imaged using an IVIS system at both Cy5 and Cy7 channels.

E. coli–infected vaginitis model and therapeutic process in vivo

The mice were intraperitoneally injected with 100 μl of β-estradiol (1 mg ml⁻¹) once every 2 days for 6 days. On the seventh day, 20 μl of E. coli (1 × 10⁹ CFUs ml⁻¹) was injected into the vagina once daily for 5 days, and then the mice were fed normally. Then, this day was designated as day 0. After successful modeling, the mice were randomly divided into five groups (six mice in each group): (i) health, (ii) PBS control, (iii) Lev, (iv) PLA, and (v) GCS-LB. E. coli–induced mice were intravaginally injected with 20 μl of 1× PBS, Lev, and Lactobacillus for six consecutive days (days 1, 2, 3, 4, 5, and 6). The GCS-LB was intravaginally injected once every 2 days for 6 days. The concentrations of PLA, Lev, and GCS-LB were 1 × 10⁷ CFUs, 1 × 10⁶ CFUs, and 0.5 mg kg⁻¹, respectively. On day 7, the vagina was washed repeatedly with sterile PBS (20 μl) by pipetting five times to obtain vaginal secretions, which were used for quantitative analysis of E. coli and Lactobacillus cell viability on agar plates and vaginal microbiome analysis. On day 8, all groups of mice were euthanized, and the vagina, heart, liver, spleen, lung, and kidney were collected and fixed with paraformaldehyde (4%). These tissues were dissected, analyzed by H&E staining, and imaged by an optical microscope.

Analysis of H₂O₂, LE, and BAs

The levels of H₂O₂, LE, and BAs in the vaginal secretions were detected using commercially available kits according to according to the manufacturer’s protocols. The absorbance of reaction solutions was measured using a UV2450 UV/vis spectrometer (Hitachi, Japan).

Analysis of cytokines

After treatment, the vaginal tissues were collected, homogenized, and lysed for the following cytokine detection. The protein from the shaved vaginal tissues (1 cm by 1 cm) was extracted with NP-40 (with protease-phosphatase inhibitor cocktail) and quantified by the BCA protein quantification method. The protein concentration of the final tissue extracts was determined and diluted to be 20 mg ml⁻¹. After that, the concentrations of IL-6, TNF-α, IFN-γ, and TGF-β were detected using ELISA kits.

16S sequencing and analysis

The 16S gene sequencing and analysis were performed at Biomarker Technologies. The total DNA was extracted from colonic content and sequenced by building a sequencing library on an Illumina HiSeq 2500. The results were stored in FASTQ (referred to as fq) format file, and data analysis was performed on BMKCloud. Alpha diversity that reflects the species richness of individual samples was conducted with Chao and Shannon indices. Beta diversity was analyzed with principal coordinates analysis.

Supplementary Materials

This PDF file includes:

Table S1

Figs. S1 to S18