Discussion

Membrane vesicles are bilayered structures, found to be secreted almost ubiquitously by gram negative bacteria and a large number of gram positive bacteria [24,25,28,29,31–34]. In the current study, we demonstrate for the first time that GBS (a gram positive bacteria) is capable of secreting MVs in a strain independent manner. Employing different microscopic techniques we could not only detect the presence of vesicles in GBS culture supernatant, but also demonstrate the budding of vesicular structures from GBS surface. Interestingly, similar budding structures from GBS surface resembling MVs have been reported in GBS colonized in the murine genital tract, implying that GBS could also produce MVs in vivo [35]. We isolated and characterized these MVs both physically and biochemically and examined their pattern of interaction with host cells in order to explore their potential contribution in GBS pathogenesis. Intriguingly, we could extract DNA and amplify the cfb gene, an important virulence factor of GBS, using the MV DNA as template. Occasionally, DNA has been found to be associated with MVs from other microorganisms such as P. aeruginosa [36], S. vesicuolsa [37] and C.perfringens [26]. Though DNA is detected in GBS MVs, we believe this DNA is fragmented as we could only detect the cfb gene, but genes representing other virulence factors (cylE, pepB,zooA) or housekeeping gene (gapN) could not be amplified. This implies that the DNA packaged in MVs is perhaps not a random phenomenon but may involve a regulated mechanism. While the functional significance of the DNA packaging in MVs is yet unknown, since the MVs fuse with the host or other bacterial cells, these might act as carriers of DNA from one cell to another [38]. It will be of interest to determine if the MV DNA specifically enters the cells to induce a pathological activation of host nuclease machinery.

Proteomic analysis of the GBS MV proteins revealed presence of numerous ECM degrading enzymes. The preferential packaging of ECM degrading enzymes compared to other abundant membrane proteins implies presence of a selective sorting mechanism. Such a sorting mechanism have been identified in P. gingivalis which not only facilitates preferential packaging of important virulence factors but also enables it to exclude other abundant outer membrane proteins from the cargo [39]. Similar enrichment of acidic glycosidases and proteases were observed in Bacteroides species suggesting presence of a species specific machinery devoted to selectively pack proteins into the vesicles to do specific jobs, in this case securing nutrients for the benefit of the whole bacterial community present in the microbiota [40].

The presence of various ECM degrading enzymes in GBS MVs therefore points towards its significant role in GBS pathogenesis. Indeed, we observed that the MVs could not only interact with HeLa cells extracellularly but also internalize and cause cell death. These observations prompted us to hypothesize that the presence of various virulence factors and ECM degrading proteins in GBS MVs might lead to tissue degradation and cell death that may contribute to PPROM and preterm births. We next performed a series of experiments to investigate this hypothesis.

The fetal membranes (amnion and chorion) rest upon a collagenous basement membrane of type II and IV collagen and beneath this layer lays a fibrous layer that contains collagen types I, III, V, and VI. Collagen, therefore provides major structural strength for the membranes and degradation of it leads to loss of membrane integrity [41–43]. The enrichment of ECM degrading proteases, coupled with the presence of gelatinolytic activity in GBS MVs suggested that these might cause loss of fetal membrane integrity by ECM degradation. Indeed, ex vivo treatment of chorio-decidual membrane with GBS MVs led to collagen fragmentation. This collagen fragmentation had profound effect on fetal membrane integrity as the stiffness of the membranes challenged with MVs were significantly reduced. It is likely that during pregnancy, GBS via its MVs lead to collagen degradation and reduction in tissue stiffness which together would resist further expansion of the amniotic sac, a prerequisite to accommodate the growing fetus. Thus our findings suggest that GBS MVs lead to loss of ECM and weaken the amniotic membrane making it susceptible to rupture upon pressure from the growing fetus.

Epidemiological data suggest that colonization of vagina and cervix with GBS increases the probability of chorio-amnionitis. We suspected that the MVs produced by GBS while colonizing can move upwards in the female reproductive tract and affect cells and tissue at distant sites. Budding structures resembling MVs have been observed in GBS colonized in the murine genital tract [35]. Extending this data, herein we show that GBS MVs when vaginally instilled could traverse anterogradely upto anterior most segment of the uterus. Recently, it has been shown that GBS mutant for hyaluronidase is not capable of ascending infection in a mouse model [44]. Since, hyaluronidase is also enriched in our GBS MV preparations; it is possible that it could aid ascend of MVs into the female tract. While it needs to be demonstrated that like GBS, MVs can also anterogradely transport into the gravid uterus; it is tempting to hypothesize that GBS colonization at a distant site such as vagina can affect the feto-maternal tissues in the uterus by the virtue of secreting MVs.

We next asked if GBS MVs have any pathogenic effects in vivo. Corroborating our ex-vivoobservations; we observed that instillation of GBS MVs intra-amniotically led to extensive disruption of the chorio-decidual structure. This disruption is also associated with collagen fragmentation which could be due to the proteases present in the MVs. Beyond collagen, extensive alterations in the expression of genes encoding for extracellular matrix proteins has been reported in chorio-decidua of rhesus monkeys infected with GBS [16]. While it will be of interest to see the alterations in other ECM molecules in response to MVs, we believe that the changes in extracellular matrix and collagen degradation would contribute to the mechanical weakening of the tissue resulting in the loss of membrane stiffness. Indeed we did see a reduction on stiffness of the chorio-decidua tissue challenged with MVs ex vivo. It will be of interest to test the effects of MVs derived from GBS mutants for various ECM degrading enzymes and determine the mechanistic basis of this phenomenon. Along with loss of collagen, we also detected extensive apoptosis in the chorio-decidua derived from sacs injected with MVs. These observations suggested that coupled with mechanical weakening, cell death might further contribute to loosening of the membrane. Such tissue damage by MVs could make the membrane prone to rupture leading to PPROM.

A leading feature of PPROM and preterm births due to inflammation is chorio-amnionitis. Chorio-amnionitis is acute inflammation of the fetal membrane and chorion which is typically caused due to ascending microbial infection especially in case of infection with genital mycoplasmas [45]. Histologically, chorio-amnionitis is characterized by leukocyte infiltration in the chorion/amnion and the decidua. In the context of GBS, in vivo administration of the live bacteria in pregnant mice led to histopathologic characteristics resembling chorio-amnionitis [16]. Herein for the first time we demonstrated that chorio-amnionitis can be caused by MVs even in absence of live bacteria. We observed that intra-amniotic administration of GBS MVs caused extensive leukocytic infiltration in the subchorion layer and in decidua resembling chorio-amnionitis. Beyond leukocytic infiltration, the decidua of mice injected with GBS MVs also had macrophage infiltration. Moreover, these vesicles also induced inflammatory cytokines such as, Kc, Il-1β, Il-6, and Tnf-α in the decidua. Levels of these pro-inflammatory cytokines are also known to be elevated in the amniotic fluid of patients in independent contexts of GBS infections and PPROM [16,46]. Since HeLa cells also showed similar response upon treatment with MVs (S3B Fig), we believe that the elevation in levels of these cytokines in the decidua may not be exclusively due to higher numbers of immune cells present in these tissues, but because of the ability of MVs to generate an inflammatory response in the decidual cells. We therefore conclude that GBS by the virtue of production of MVs activates the host immune system which might trigger the immune cell homing and activation leading to chorio-amnionitis. These observations are novel as for the first time we have shown that even in absence of active infection in the chorio-decidua, features resembling clinical chorio-amnionitis could be mimicked by MVs. Clinically this observation is highly relevant as 50–80% women with chorio-amnionitis do not have bacteria in their amniotic fluid or the decidual tissue [47–50]. Based on our findings we can hypothesize that MVs secreted by the pathogens residing in lower genital tract may be responsible for cases with unexplained chorio-amnionitis. It would be imperative to study the presence of MVs in amniotic fluid and chorio-decidual tissues of woman with culture negative chorio-amnionitis.

Since, GBS MVs lead to tissue damage, inflammation and chorio-amnionitis, resulting in mechanical weakening of the fetal membrane, we finally asked if GBS MVs can lead to premature birth. Indeed, we observed that intra-amniotic injection of GBS MVs lead to preterm delivery. More than 50% of mouse pups were delivered preterm (almost 2 days prior to their expected day of delivery) when challenged with MVs. Finally, we also observed that like live GBS [17], the MVs are also highly pathogenic to the fetus. Almost 30–40% of fetuses died in utero amounting to resorption and some fetuses were still born. The fetuses that were stillborn or recovered from uterine sacs after GBS MV challenge, were smaller and had major damage to its organs. These results together imply that GBS MVs can cause IUFD or preterm births. To our knowledge this is the first report describing the direct role of membrane vesicles produced by any pathogenic bacteria in disease pathogenesis. In the light of the fact that intrauterine infections can lead to autism like changes in the brain [51], it is possible that GBS MVs might have a similar effect. It will be of interest to study the effects of sub-lethal dose of GBS MVs on fetal development and physiology.

In summary, the results of the present study have shown that GBS MVs cause host cell death, membrane weakening and inflammation of the feto-maternal interface which causes preterm birth and IUFD. Coupling this with the fact that pathogen derived vesicles can function as vehicles for long distance delivery of virulence factors [52], our results imply that the production of extracellular membrane vesicles serves not only as a tool for secretion but also arms GBS with an additional weapon by which while colonizing it can orchestrate events at distant sites including the fetal membrane. Our findings provide a plausible explanation for the occurrence of PPROM and premature delivery in woman with chorio-amnionitis without detectable bacteria in their amniotic fluid or the decidual tissue. We conclude that GBS utilizes MVs as a surrogate to spread its virulence factors in the host which are responsible for the clinical features of GBS infection during pregnancy. Prevention of vesicle biogenesis may therefore be a viable therapeutic option to prevent GBS mediated preterm birth.

Materials and Methods

Ethics statement

All the experimental work on animals was done as per the guidelines of the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), India. The study protocol has been reviewed and approved by the Institutional Animal Ethics Committee of National Institute for Research in Reproductive Health (NIRRH) under the project number 10/15.

Bacterial strains and isolation of MVs

S. agalactiae strains A909 (serotype IA), COH1 (serotype III), NEM316 (serotype III) and 2603V/R (serotype V) (kindly provided by Dr. Lakshmi Rajagopal, Department of Pediatric Infectious Diseases, University of Washington School of Medicine, Univ. of Washington, Seattle, USA) were cultured in Todd Hewitt broth (THB) at 37^○C. Unless otherwise stated MVs from GBS strain A909 was used for further experiments. MVs were purified from growing GBS cultures as per the standard protocol used for many other bacteria [25,27,32,53,54]. Briefly, GBS cells grown upto optical density (OD_600nm) 1.2, were harvested by centrifugation (12,000 x g, 30 min, 4^○C) and the supernatant was passed through 0.22μm filter. The filtrate was concentrated using Amicon ultrafiltration system (10 kDa) and ultracentrifuged (150000 x g, 3 h, 4^○C) to pellet down MVs. The MV pellet was resuspended in PBS (pH 7.2) and protein content was determined by Bradford assay following lysis of MVs by 0.05% Triton X-100.

To quantitate the number of MVs produced from GBS culture, MVs were labeled with 20 μM of Vybrant DiO cell labeling solution (Molecular Probes) and quantified by flow cytometry (FACS Aria II, BD Biosciences, USA) using fluorescent counting beads (CountBright Absolute counting beads; Invitrogen) as standards as described earlier [25]. The varying diameters and size distribution of MV preparations were measured using a Goniometer (Brookhaven Instrument Co., USA) as described earlier [55].

Electron microscopy and atomic force microscopy for GBS MVs

For Scanning Electron Microscopy (SEM), GBS cells were air dried, desiccated overnight and visualized with Field Emission Gun Scanning Electron Microscope (JEOL, USA) at an accelerating voltage of 5 kV. For Transmission Electron Microscopy (TEM), MV samples were applied to Formvar/Carbon film coated 200-mesh copper grids (Pacific Grid-Tech) and negatively stained with 2.5% uranyl acetate followed by visualization under Transmission Electron Microscope (FEI Technai, USA) (120 kV). For Atomic Force Microscope (AFM), bacterial suspension was loaded onto poly-L-lysine coated coverslips and visualized under an atomic force microscope (Asylum Research, USA) under contact mode at a scanning rate of 1 Hz using silicon nitride cantilevers.

Fatty acid analysis

Lipids were extracted from intact GBS cells and MVs using protocol described earlier [25]. Following extraction, lipids were silylated using 1:1 ratio of N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA) and pyridine at 75^○C for 30 min. Silylated lipids were then analyzed using a GC-MS (Agilent, USA) fitted with a HP-5MS fused silica capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness) with Helium as carrier gas (flow rate 1ml/min). Identification of compounds was based on mass spectra including comparison to a library (NIST).

Protein analysis

150 μg of MV proteins were separated on 12% SDS-PAGE. After staining with Coomassie Brilliant Blue R-250 (CBB), bands were excised and processed for in-gel trypsin digestion [56]. The eluted oligopeptides were co-crystallized with CHCA (5 mg/ml) and spectra were acquired using MALDI-ToF mass spectrometer (UltraFlex III, Bruker Daltonics). Mascot (Version 2.2.04, Matrix Science) searches were conducted using the NCBI non-redundant database with the following settings: 1 missed cleavage; Carbamidomethyl on cysteine as fixed modifications, methionine oxidation as variable modification and 100 ppm error (150 ppm error for band No. 4). A match with S. agalactiae protein with the best score in each Mascot search was accepted as successful identification (p< 0.05). Subcellular locations were predicted by LocateP database and confirmed by pSORT and TMHMM algorithms.

Isolation and analysis of nucleic acid from MVs

MVs were initially treated with 50 μg/ml DNaseI in presence of 10 mM MgCl₂ at 37°C for 1 h to remove any surface bound DNA. Following heat inactivation (10 min at 80°C) and lysis of MVs using Triton X-100 (0.05%) at 37⁰C for 30 min, DNA was extracted by phenol-chloroform-isoamyl alcohol, precipitated using ammonium acetate as described earlier [26]. The purified DNA was quantified and used for PCR using primers specific for cylE, cfb, pepB, zooA andgapN genes (S1 Table). Genomic DNA isolated from GBS strain A909 served as control template.

Cell culture, internalization of MVs and cytotoxicity

HeLa cells (procured from National Center for Cell Science, Pune, India) were grown in DMEM media (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) at 37^○C in 5% CO₂. HeLa cells were incubated with increasing protein concentrations of MVs (10 μg/ml to 300 μg/ml) for 24 h and the viability was assessed using MTT assay kit (HiMedia, India).

To study internalization, MVs were labeled with FITC (Sigma) in 0.1 M sodium carbonate buffer (pH 9.0) for 1 h at room temperature and washed to remove unbound FITC [57]. FITC labeled MVs (30 μg MV protein) were then allowed to interact with HeLa cells for 6 h. After washing to remove unbound MVs, cells were fixed and stained with mouse polyclonal anti-FITC antibody (Invitrogen) followed by an anti-mouse secondary antibody conjugated to AlexaFluor 555 (Invitrogen). Images were acquired with an oil immersion Plan-Apochromat 63X/1.4 NA objective using a confocal laser scanning microscope (Zeiss).

Gelatin zymography

10 μg of MV proteins were separated (150 V, 5 h, 4°C) on 12% SDS-PAGE with 0.1% co-polymerized gelatin. After treatment in 1% Triton X-100, gelatinolysis was promoted by further incubation for 16 h at 37°C in developing buffer (50 mM Tris-HCl, 50 mM Tris base, 200 mM NaCl, 5 mM CaCl₂, 0.02% Brij 35, pH 7.6). The gel was then stained with Coomassie Brilliant Blue R-250 and destained to intensify the digestion halos.

Preparation of liposome

BSA entrapped liposomes were prepared by thin film hydration method [58]. Briefly, phosphatidylcholine (10 mg) was dissolved in a mixture of chloroform and methanol (2:1) and the solvent was evaporated under vacuum using a rotary evaporator to form a thin layer of lipid film in a round bottom flask. The dried lipid film was hydrated using 1 ml of PBS containing BSA (10 mg/ml) at 45°C for 1 h to form multilamellar vesicles (MLV). The liposome suspension was then sonicated using a probe sonicator at 40% amplitude to obtain unilamellar vesicles.

Atomic force microscopy for tissues

Chorio-decidual membrane was collected from pregnant mice on E14.5. Immediately after dissection, the membranes were incubated in DMEM (Invitrogen) containing 10% FBS (Invitrogen) at 37°C and treated for 24 h with either PBS, or BSA entrapped liposomes (100 μg/ml) or MVs (100 μg/ml) in presence or absence of protease inhibitor cocktail (Sigma). Collagenase (10 μg/ml) was used as positive control. Following incubation, the membranes were washed, carefully spread on a slide layered with double adhesive tape, allowed to air dry for 5 min. Subsequently, the membranes were hydrated with PBS, and their mechanical properties were probed with an Atomic Force Microscope (AFM). A 5 μm diameter spherical probe with a nominal spring constant of 32.66 pN/nm was used. Using a custom-written code, force-indentation curves were fitted with Hertz model to estimate the Young’s modulus of elasticity of each of the membranes. Each sample was probed randomly multiple number of times (greater than 50) at multiple different positions to estimate average stiffness of the membranes.

Anterograde transport of MVs

The vagina of mice in estrus phase were flushed thrice with 40 μl of 0.2% Triton X-100 in 0.9% saline followed by 40 μl 0.9% saline. MVs were labeled with FITC as described earlier [57] and 100 μg of FITC labeled GBS MVs in 100 μl PBS were vaginally instilled in 25–30 μl aliquots using a micropipette. Control mice received PBS only. After 6 h mice were euthanized and the reproductive tract (cervix to uterus) was collected. Tissues were briefly fixed in 4% paraformaldehyde and mounted on slides with Vectashield containing DAPI (Vector Laboratories). Images of different parts of the tissue such as utero-cervical junction, distal uterus and proximal uterus were captured with an oil immersion Plan-Apochromat 40X/1.3 NA objective of a confocal laser scanning microscope.

Intra-amniotic administration of MVs

For pregnancy-outcome experiments, C57BL6/J- were bred and maintained at the Experimental Animal Facility of NIRRH under constant temperature and 12 h light and dark cycles were used. Female mice in estrus were impregnated naturally by a male of proven fertility and mating was confirmed by the presence of a vaginal plug (E0.5). Intra-amniotic injections were performed on E14.5 of a 19–20 d gestation. Briefly, animals were anesthetized and a 1.5-cm midline incision was made in the lower abdomen. The mouse uterus is a bicornuate where the fetuses are arranged in a “beads-on-a-string” pattern. Both the horns were exposed and individual fetal sacs were injected with 100 μl of MVs (5 or 10 μg protein) or equivalent amount of PBS or BSA-liposomes. After injection the uterus was returned to the abdomen, muscle and skin layers were sutured and dams were returned to their cages and monitored on regular intervals. Surgical procedures lasted ∼10 min and post-operative care was taken as per standard protocols at NIRRH. Delivery of one or more pups in the cage or lower vagina within 48 h was considered preterm.

For collection of tissues, animals were euthanized 24 h after surgery. The inoculated horns were incised longitudinally along the anti-mesenteric border. Gestational tissues (full-thickness biopsies from the middle region) and fetal membranes were harvested and frozen in Trizol reagent (Invitrogen) and stored at −80°C for RNA extraction or fixed in 4% paraformaldehyde for histopathology. Fixed tissues were paraffin embedded and 5 μm thick paraffin sections were collected on poly-lysine coated glass slides and processed for routine hematoxylin and eosin staining.

RNA extraction, cDNA synthesis and qRT-PCR

RNA from HeLa cells or mouse tissue was isolated using Trizol reagent as detailed previously [59]. RNA was treated for 30 min with DNaseI to remove any DNA contamination and processed for reverse transcription. The details of qRT-PCR have been described previously [59]. Briefly, 1 μg of RNA was reverse-transcribed using SuperScriptTM First-Strand Synthesis System (Invitrogen). The cDNA was further used for quantitative RT-PCR (qRT-PCR) for various cytokines. β-actin (for HeLa cells) and 18S (for mouse tissues) were used as the house keeping genes. The sequences of the primers are as mentioned in S1 Table. Care was taken to design primers that spanned an intron to eliminate any amplification due to genomic DNA contamination. The specificity of amplicons was confirmed by performing dissociation / melt curve analysis. Only those primer pairs that resulted in a single sharp melt peak with a consistent melt temperature were included in the study. The relative changes in the expression of above genes was analyzed by 2^-ΔΔCt method [60].

Quantification of leukocytes

Hematoxylin and Eosin stained slides of the chorio-decidua were examined under 40X objective for neutrophils and lymphocytes. The cells were identified based on their morphology. The number of neutrophils and lymphocytes per field were counted in 8–10 random fields per section. Five random sections from each fetus were analyzed. The analysis was done in three biological replicates.

Immunohistochemistry

Immunohistochemistry on tissue sections was performed as detailed previously [61]. For detection of macrophages, paraffin embedded 5 μm sections of the chorio-decidual membrane (with or without MV treatment) were stained with rabbit polyclonal anti-F4/80 Ab (1:100; Santa Cruz Biotechnology) and detected using HRP conjugated goat anti-rabbit secondary antibody (1:100; Dako) and 3.3’ Diaminobenzidine with H₂O₂. Sections were counterstained with hematoxylin and mounted in DPX.

Collagen staining

Total collagen staining in chorio-decidual membrane sections were performed using Picrosirius red [62]. Paraffin sections were deparaffinized, rehydrated in graded methanol series and stained with hematoxylin followed by 0.5% Direct Red 80 (Sigma) prepared in picric acid. Following dehydration with 100% methanol and clearing in xylene, the slides were mounted with DPX.

TUNEL

Induction of apoptosis by GBS MVs was analyzed by In Situ Cell Death Determination Kit (Roche) according to manufacturer’s protocol. Briefly, following deparaffinization, tissue sections were rehydrated and digested using proteinase K. After permeabilization with 0.25% Triton X 100, the sections were incubated with terminal deoxynucleotidyl transferase and fluorescein-dUTPs for 1 h at 37°C. Following washing to remove unbound dUTPs, the sections were mounted with Vectashield containing DAPI (Vector Laboratories). Images were acquired with an oil immersion Plan-Apochromat 40X/1.3 NA objective of confocal laser scanning microscope (Zeiss).