Streptococcus agalactiae Induces Placental Macrophages To Release Extracellular Traps Loaded with Tissue Remodeling Enzymes via an Oxidative Burst-Dependent Mechanism

Streptococcus agalactiae, also known as group B Streptococcus (GBS), is a common pathogen during pregnancy where infection can result in chorioamnionitis, preterm premature rupture of membranes (PPROM), preterm labor, stillbirth, and neonatal sepsis. Mechanisms by which GBS infection results in adverse pregnancy outcomes are still incompletely understood. This study evaluated interactions between GBS and placental macrophages. The data demonstrate that in response to infection, placental macrophages release extracellular traps capable of killing GBS. Additionally, this work establishes that proteins associated with extracellular trap fibers include several matrix metalloproteinases that have been associated with chorioamnionitis. In the context of pregnancy, placental macrophage responses to bacterial infection might have beneficial and adverse consequences, including protective effects against bacterial invasion, but they may also release important mediators of membrane breakdown that could contribute to membrane rupture or preterm labor.

IMPORTANCE Streptococcus agalactiae, also known as group B Streptococcus (GBS), is a common pathogen during pregnancy where infection can result in chorioamnionitis, preterm premature rupture of membranes (PPROM), preterm labor, stillbirth, and neonatal sepsis. Mechanisms by which GBS infection results in adverse pregnancy outcomes are still incompletely understood. This study evaluated interactions between GBS and placental macrophages. The data demonstrate that in response to infection, placental macrophages release extracellular traps capable of killing GBS. Additionally, this work establishes that proteins associated with extracellular trap fibers include several matrix metalloproteinases that have been associated with chorioamnionitis. In the context of pregnancy, placental macrophage responses to bacterial infection might have beneficial and adverse consequences, including protective effects against bacterial invasion, but they may also release important mediators of membrane breakdown that could contribute to membrane rupture or preterm labor. Placental macrophages were infected for 1 h with GBS cells at an MOI of 20:1. Scanning electron micrographs (bottom row) demonstrate extracellular structures released from macrophages (white arrows), which are not seen after DNase I treatment. PMs were also stained with SYTOX Green, a double-stranded DNA dye, and evaluated by scanning laser confocal microscopy, which demonstrates extracellular structures composed of DNA (white arrowheads). Bars represent 100 m. (B) Placental macrophage extracellular traps were stained with Hoechst 33342 (blue), a condensed chromatin/nuclear stain, SYTOX Green (green), and specific antibodies for either histone H3, myeloperoxidase, or neutrophil elastase as listed (red). Histone, myeloperoxidase, and neutrophil elastase staining colocalizes with extracellular DNA staining, suggesting that MET structures contain these proteins. Bars represent 100 m. (C) PMs releasing METs were quantified by counting MET-producing cells seen in SEM images and expressed as the number of macrophages releasing METs per field. GBS-infected PMs release significantly more METs than uninfected cells, and DNase I treatment degraded these structures. Data represent samples from 6 to (Continued on next page) Doster et al. against GBS were specific to PMs or might represent a broader macrophage response. The immortalized monocyte-like cell line, THP-1 cells, was evaluated after differentiation into macrophage-like cells with phorbol 12-myristate 13-acetate (PMA) for 24 h. THP-1 macrophage-like cells infected with GBS released significantly more METs than uninfected cells, and DNase I treatment degraded the MET structures (Fig. S5). The THP-1 MET response required contact with bacterial cells, as treatment of the macrophage-like cells with sterile filtered bacterial culture supernatant did not stimulate MET release compared to uninfected cells.
Actin polymerization is required for GBS-induced MET release. Actin polymerization has been shown to be important for MET release (36). A similar role of cytoskeletal changes on GBS-induced MET release in THP-1 macrophage-like cells was examined. Treatment prior to infection with the actin polymerization inhibitor, cytochalasin D, but not nocodazole, which inhibits microtubule polymerization, inhibited MET release compared to GBS-infected, untreated cells (Fig. S5). As noted below (and shown in Fig. 2B), cytochalasin D also inhibited MET release by human PMs infected with GBS.
Placental macrophage MET responses require ROS production. Neutrophil release of NETs occurs in a ROS-dependent manner (40). It was hypothesized that MET release from PMs may require production of ROS. Treatment of PMs prior to infection with the NADPH oxidase inhibitor diphenyleneiodonium (DPI) inhibited release of METs, whereas treatment of uninfected macrophages with PMA resulted in similar levels of MET release to GBS-infected cells ( Fig. 2A and B). As with the THP-1 macrophage-like cells, treatment of PMs with cytochalasin D prior to infection inhibited MET release. To further define that ROS production was associated with MET release, a fluorescent ROS dye was used to evaluate PMs for intracellular ROS production. Treatment with DPI inhibited ROS production, and GBS infection as well as PMA treatment of uninfected PMs resulted in significantly more ROS production than uninfected cells ( Fig. 2C and D). Interestingly, pretreatment with cytochalasin D decreased levels of intracellular ROS production similar to those of DPI, suggesting that pretreatment with the actin cytoskeletal inhibitor may actually be preventing MET release by impeding ROS production. Additionally, ROS production in these experiments mirrored the degree of MET release under similar conditions (Fig. 2B), suggesting that ROS production is necessary for MET release from these macrophages.
Placental macrophage METs contain MMPs. During pregnancy, PMs support gestational tissue remodeling through release of MMPs. Because macrophage release of MMPs has been implicated in the pathogenesis of fetal membrane rupture (41), we hypothesized that these proteases may also be released in METs. Five MMPs that have been implicated in development and pathological pregnancies were evaluated. Immunofluorescent staining of METs was significant for the colocalization of MMP-1, -7, -8, -9, and -12 with extracellular DNA structures (Fig. 3A). As MMPs are present within METs and GBS infection induced MET release, metalloprotease concentrations within coculture supernatants were examined to determine whether GBS infection would result in an increase in metalloprotease release. MMP-8 and MMP-9 have been investigated as potential biomarkers for intrauterine infection (42)(43)(44), and concentrations of both were significantly elevated in supernatants of GBS-infected cells compared to uninfected controls ( Fig. 3B and C). Global MMP activity of coculture supernatants was then assessed to determine whether the MMPs released were active by using a MMP activity assay, which uses fluorescence resonance energy transfer peptides that, when cleaved by MMPs, are fluorescent. Supernatants taken from placental macrophages cocultured with GBS demonstrated significantly more MMP activity compared to those of uninfected controls (Fig. 3D). Together, these data suggest that PMs express several MMPs and that these MMPs are released during bacterial infection within METs and into the extracellular spaces, where they might contribute to breakdown of gestational tissue extracellular matrix. To determine whether METs were present within gestational tissues in response to infection, fetal membrane tissues from healthy, term, nonlaboring caesarian sections were obtained, excised, and organized into Transwell structures, creating two chambers separated by the fetal membranes. GBS cells were added to the choriodecidual surface, and infection was allowed to progress for 48 h prior to fixing tissues for immunohistochemistry and immunofluorescence analysis. CD163-positive cells were found localized to an area of GBS microcolonies within the membranes that demonstrated histone staining extending beyond the nucleus and into the extracellular space, suggesting the release of a MET-like structure (Fig. 4A). Immunofluorescence staining demonstrated CD163-positive cells associated with extracellular material that stained positive for histones and MMP-9 (Fig. 4B). Additional staining of fetal membrane tissues using neutrophil elastase as was previously described for identification of NETs in tissues (45), identified cells within the choriodecidua with long extensions that stained strongly for neutrophil elastase that colocalized to histone H3 and DNA staining (Fig. S6). Cells releasing MET-like structures could be compared to cells with intact nuclei and neutrophil elastase staining limited to granule structures, suggesting that cells releasing MET-like structures had undergone cellular changes consistent with etosis.

DISCUSSION
In the initial description of NETs in 2004, several potential immune functions were described, including the trapping and killing of microorganisms and degradation of bacterial products (37). Release of cellular DNA and proteins within extracellular traps has also been associated with autoimmune pathology in systemic lupus erythematosus and antineutrophil cytoplasmic autoantibodies (ANCA)-associated vasculitis, as well as in diseases of aseptic inflammation (46)(47)(48)(49)(50). Other leukocytes, including mast cells, eosinophils, basophils, and macrophages, have now been shown to release extracellular trap structures (36,(51)(52)(53)(54).
In this report, PMs are added to a growing list of monocytes and tissue differentiated macrophages capable of releasing METs, which includes human alveolar macrophages, glomerular macrophages, peripheral blood monocytes, and macrophages from other mammalian and nonmammalian species (36). These data demonstrate that human PMs and PMA-differentiated THP-1 macrophage-like cells release METs in response to bacterial infection and after treatment with the protein kinase C agonist, PMA. The present data correlate with previous reports that neutrophils and murine macrophages release traps in response to GBS infection and that METs are capable of killing GBS cells (55,56).
These data also demonstrate that PM METs contain many proteins previously identified in NETs, including histones, neutrophil elastase, and myeloperoxidase (37). These results mirror other MET investigations demonstrating that diverse macrophages produce and release proteins, including neutrophil elastase within MET structures. For example, human glomerular macrophages releasing METs containing myeloperoxidase have been demonstrated in cases of ANCA-associated glomerulonephritis, and human alveolar macrophages have been shown to release METs containing histones and   (57,58). Human blood monocytes release METs containing H3 histones, myeloperoxidase, lactoferrin, and neutrophil elastase in response to Candida albicans cells, and similar MET contents have been demonstrated in THP-1 macrophage-like cells infected with Mycobacterium massiliense (59,60).
Similar to neutrophil etosis, these data suggest that ROS generation is necessary for PM MET release, as this response was inhibited by treatment with the NADPH oxidase inhibitor DPI. These results mirror reports that inhibition of ROS production via chemical inhibitors resulted in diminished MET release in bovine, caprine, murine, and human macrophages (61)(62)(63)(64)(65). In neutrophils, ROS act to break down intracellular membranes and activate neutrophil elastase, which translocates to the nucleus where it degrades histones and promotes chromatin decondensation (39). Myeloperoxidase is thought to contribute chromatin decondensation by an enzyme-independent mechanism (39). It is unclear at this time whether neutrophil elastase and myeloperoxidase perform similar roles in macrophages.
Our data indicate an increase in LDH release during infection, which is consistent with reports that MET release results in cell death (66). Etosis has been noted to be distinct from other cellular death pathways, including pyroptosis and apoptosis. After 1 h of infection when MET responses were identified, there was no significant difference in TUNEL staining or IL-1␤ release, suggesting that PM METs occur by a distinct pathway. This is notable, as previous reports have shown that the GBS toxin ␤-hemolysin is capable of inducing pyroptosis of macrophages, though in this study infection was allowed to progress for 4 h, longer than that required for the PM MET response (67). Previous studies have also demonstrated that GBS is capable of inducing macrophage apoptosis, but again this occurred over longer periods of infection than the 1 h that was capable of inducing MET responses (68).
Pretreatment of THP-1 macrophage-like cells and PMs with the actin cytoskeletal inhibitor cytochalasin D inhibited MET release, but not the microtubule inhibitor nocodazole. Conflicting reports exist regarding the role of actin polymerization in etosis. Studies evaluating bovine macrophages and THP-1 cells demonstrated a decrease in MET release after cytochalasin treatment, but similar treatment of murine J744A.1 macrophage-like cells, RAW macrophage-like cells, and bovine blood monocytes did not have a significant effect on MET release (60,61,64,69). This collection of conflicting reports mirrors the NET literature. In the original NET description, cytochalasin D prevented cell phagocytosis but not NET release (37). Others have documented NET inhibition with nocodazole or cytochalasin D in response to LPS or enrofloxacin (70,71). Because of the differential responses, some authors have postulated that phagocytosis may be an important first step toward cell stimulation and ROS generation, and cytoskeletal inhibition may block the initial steps toward MET release. Another possibility is that the pretreatment with cytochalasin D may interrupt trafficking of the NADPH oxidase complex, thus impairing ROS production. NADPH oxidase is a complex of six components, and the cytosolic proteins p40 phox and p47 phox are known to interact with F-actin; treatment with cytochalasins have been shown to interrupt NADPH complex formation and lead to impaired ROS formation (72,73). Timing of the cytochalasin treatment is important, as treatment of cells after prestimulation with molecules such as LPS, which stimulates NADPH oxidase assembly, may actually increase generation of ROS in these cells (74). In our study, macrophages were pretreated with cytochalasin D and were not stimulated prior to infection. It remains unclear whether the conflicting literature with regard to the impact of cytoskeletal inhibition on extracellular traps may be explained by the timing of cytoskeletal inhibition and subsequent effects on ROS production.
PMs were found to produce and release several MMPs within MET structures. During chorioamnionitis, inflammatory mediators lead to the production and release of several metalloproteinases, including MMP-1, MMP-7, MMP-8, and MMP-9 (75,76). MMP-9 is considered to be the major MMP responsible for collagenase activity within the membranes, but many other MMPs are thought to contribute to the processes of membrane weakening (75,77). This study reinforces and expands previous reports that identified placental leukocytes as being able to secrete MMPs, including MMP-1, -7, and -9 (78). Several MMPs have been implicated in preterm birth and pathological pregnancies. MMP-1 and MMP-9 were found to be elevated in placental tissues of women with preterm births compared to women delivering at full term (79). MMP-1 and neutrophil elastase have been shown to stimulate uterine contractions (80). Interestingly, proteomic comparisons of amniotic fluid samples from women with premature preterm rupture of membranes demonstrated increases in histones (H3, H4, and H2B), myeloperoxidase, neutrophil elastase, and MMP-9 in women with histologic chorioamnionitis and proven intrauterine infection, which likely represents the influx of inflammatory cells into these tissues and potentially the release of extracellular traps (81). MMP-12, or macrophage metalloelastase, is a key mediator of the breakdown of elastase and has been shown to be important for spiral artery remodeling during parturition, but to date, there are no studies demonstrating changes in MMP-12 release during cases of pathological pregnancies (82). MMP-12 is better studied in conditions of lung pathology, including emphysema, and alveolar macrophages are known to release MMP-12 in METs during infection, suggesting that protease release from leukocytes may contribute to this disease process (65). Analogous to a controlled burn, we speculate that tethering MMPs to MET structures allows the host to control the release of these potent enzymes, thereby limiting their capacity to broadly weaken membrane structure in response to infection.
MET release appears to occur within fetal membrane tissue, as demonstrated by our immunohistochemistry and immunofluorescence data. This report adds to the growing relevance of these structures in cases of disease pathology. NETs have previously been identified in placenta tissue samples from women with pregnancies complicated by systemic lupus erythematosus and preeclampsia (83,84). NETs were also found in fetal membrane samples from women with spontaneous preterm labor due to acute chorioamnionitis (85). Interestingly, in this report, antibody staining with histone H3 and neutrophil elastase was used to denote NET structures, but given our data, this staining pattern would not have differentiated METs from NETs. Additionally, our group and others have demonstrated that in animal models of vaginal colonization and perinatal infection with GBS, neutrophils traffic to GBS-infected gestational tissues and release NETs containing antimicrobial peptides, including lactoferrin, as a means to control bacterial growth and invasion (86)(87)(88).
In conclusion, we demonstrate that placental macrophages as well as PMAdifferentiated THP-1 cells respond to bacterial infection by releasing METs. These MET structures contain proteins similar to those of NETs, including histones, myeloperoxidase, and neutrophil elastase. MET release from these macrophages can be stimulated in the absence of bacterial cells with PMA and is inhibited by pathways that impair ROS production. Placental macrophage METs contain several MMPs that have been implicated in pathological pregnancies, including premature rupture of membranes. MET structures were identified in human fetal membrane tissue infected ex vivo. Together, these results suggest that placental macrophages, which are thought to help maintain maternal fetal tolerance and aid in extracellular matrix remodeling, are capable of responding to GBS infection in a way that may trap and kill GBS cells but may also release important mediators of fetal membrane extracellular matrix digestion that could potentially contribute to infection-related pathologies, including preterm rupture of membrane and preterm birth.

MATERIALS AND METHODS
Placental macrophage isolation and culture. Human placental macrophages (PMs) and fetal membrane tissues were isolated from placental tissue samples from women who delivered healthy infants at full term by cesarean section (without labor). Deidentified tissue samples were provided by the Cooperative Human Tissue Network, which is funded by the National Cancer Institute. All tissues were collected in accordance with the guidelines of the Vanderbilt University Institutional Review Board (approval 131607). Macrophage isolation occurred as previously described (89); briefly, placental villous tissue samples were minced followed by digestion with DNase, collagenase, and hyaluronidase (all from Sigma-Aldrich, St. Louis, MO). Cells were filtered and centrifuged, and CD14 ϩ cells were isolated using the magnetic MACS Cell Separation system with CD14 microbeads (Miltenyi Biotec, Auburn, CA). Cells were Doster et al. incubated in RPMI 1640 medium (ThermoFisher, Waltham, MA) with 10% charcoal stripped fetal bovine serum (ThermoFisher) and 1% antibiotic/antimycotic solution (ThermoFisher) overnight at 37°C in 5% carbon dioxide. The following day, PMs were suspended in RPMI 1640 medium without antibiotic/ antimycotic and distributed into polystyrene plates. Cells were incubated for at least 1 h prior to infection to allow for cell adherence to the plate or to poly-L-lysine-coated glass coverslips (Corning, Bedford, MA) for microscopy assays. THP-1 cell culture. THP-1 cells (ATCC, Manassas VA) were cultured in RPMI 1640 medium with 10% charcoal-treated FBS and 1% antibiotic/antimycotic medium at 37°C in 5% carbon dioxide. Twenty-four to 48 h prior to coculture experiments, cells were treated with 100 nM phorbol 12-myristate 13-acetate (PMA) (Sigma-Aldrich) to induce differentiation to macrophage-like cells. Prior to coculture experiments, cells were suspended in RPMI 1640 medium without antibiotic/antimycotic and distributed into polystyrene plates containing poly-L-lysine-coated glass coverslips and allowed to rest for at least 1 h prior to infection to promote cell adherence.
Bacterial culture. Streptococcus agalactiae strain GB590 is a capsular type III, ST-17 strain isolated from a woman with asymptomatic colonization (90), and GB037 is a capsular type V strain obtained from a case of neonatal sepsis (91,92). Escherichia coli serotype 075:H5:K1 is a clinical isolate obtained from a fatal case of neonatal meningitis (93). Bacterial cells were cultured on tryptic soy agar plates supplemented with 5% sheep blood (blood agar plates) at 37°C in ambient air overnight. Bacteria were subcultured from blood agar plates into Todd-Hewitt broth or Luria broth and incubated under aerobic shaking conditions at 37°C in ambient air to stationary phase. Bacterial supernatant was collected and sterile filtered using a 0.1-m filter (Millipore Sigma, Burlington, MA) and incubated with THP-1 cells at a concentration of 10% volume. Bacterial cells were washed and suspended in phosphate-buffered saline (PBS) (pH 7.4), and bacterial density was measured spectrophotometrically at an optical density of 600 nm (OD 600 ), and bacterial numbers were determined with a coefficient of 1 OD 600 ϭ 10 9 CFU/ml. Bacterium-macrophage cocultures. PMs or PMA-differentiated macrophage-like cells in RPMI 1640 medium without antibiotics were infected with GBS or E. coli cells at a multiplicity of infection (MOI) of 20:1 unless otherwise noted. Cocultured cells were incubated at 37°C in air supplemented with 5% carbon dioxide for 1 h. As stated, some cells were pretreated with 10 g/ml cytochalasin D (Thermo-Fisher), 10 nM nocodazole, 100 U/ml DNase I, 500 nM PMA, or 10 M diphenyleneiodonium chloride (all from Sigma-Aldrich) for at least 20 min prior to infection. At 1 h, supernatants were collected, and cells were fixed with 2.0% paraformaldehyde and 2.5% glutaraldehyde in 0.05 M sodium cacodylate buffer (Electron Microscopy Sciences, Hatfield, PA) for at least 12 h prior to processing for microscopy.
Field-emission gun scanning electron microscopy. After the macrophages were treated and infected as described above, they were incubated in 2.0% paraformaldehyde and 2.5% glutaraldehyde in 0.05 M sodium cacodylate buffer for at least 12 h prior to sequential dehydration with increasing concentrations of ethanol. Samples were dried at the critical point, using a CO 2 drier (Tousimis, Rockville, MD), mounted onto an aluminum stub, and sputter coated with 80/20 gold-palladium. A thin strip of colloidal silver was painted at the sample edge to dissipate sample charging. Samples were imaged with an FEI Quanta 250 field-emission gun scanning electron microscope. Quantification of macrophages producing extracellular traps was determined by evaluating scanning electron micrograph images at a magnification of ϫ750 and counting total macrophages and those macrophages releasing extracellular traps. Extracellular traps were defined as previously described with typical appearing fibers extending from the cell body into the extracellular space (36).
Bacterial killing by macrophages releasing extracellular traps. PMs were infected with GBS cells at an MOI of 20:1 as described above. As indicated, some PMs were incubated with 100 U/ml DNase I during infection to degrade extracellular trap structures as has been described previously (37). At the end of 1 h, DNase I was added to previously untreated wells for 10 min to release trapped bacterial cells. Supernatants were collected, and PMs were permeabilized with 0.05% Tween 20 in sterile ice-cold water to release intracellular bacteria. Samples were vortexed vigorously, serially diluted, and plated on blood agar plates for enumeration of the bacterial cells. Untreated PMs were compared to DNase I-treated cells, and data are expressed as the percentage of colony-forming units (CFU) recovered compared to that of untreated cells. To further evaluate bacterial killing, PMs were seeded onto coverslips and infected as described above. Following infection, cells were stained using the Live/Dead BacLight bacterial viability kit (Invitrogen) prior to confocal laser scanning microscopy.
LDH cytotoxicity assay. Placental macrophages were incubated in RPMI 1640 medium without antibiotics or serum and infected as described above. Supernatants were collected and centrifuged to pellet cellular debris. Supernatants were analyzed using the Cytotoxicity Detection kit (Sigma-Aldrich) per the manufacturer's instructions. Results are expressed as percent toxicity using medium without cells as the low control and cells treated with 2% Triton X-100 as the high control. Percent cytotoxicity was calculated using the following equation: cytotoxicity (as a percentage) ϭ (experimental value Ϫ low control)/(high control Ϫ low control) ϫ 100.
Apoptosis assay. Placental macrophages were incubated in RPMI 1640 medium without antibiotics and infected as described above. Following infection, supernatants were removed, and cells were fixed with 2.0% paraformaldehyde and 2.5% glutaraldehyde in 0.05 M sodium cacodylate buffer for at least 15 min. Click-iT Plus TUNEL assay with Alexa Fluor 594 dye (ThermoFisher) was used to identify cells undergoing apoptosis, and staining was conducted per the manufacturer's instructions with additional staining that included Hoechst 33342 to visualize nuclei prior to confocal laser scanning microscopy.
Macrophage viability assay. To determine whether DNase I treatment resulted in alterations in PM viability or function, TNF-␣ release was used a functional measure. Cells were left untreated or treated with 100 U/ml DNase I for 60 min. All cells were then washed, and fresh medium was added prior to stimulation with 150:1 heat-killed GBS cells (incubated at 42°C for 2 h) for 24 h. Supernatants were collected and centrifuged to pellet cellular debris before TNF-␣ release was determined using a DuoSet TNF-␣ ELISA (R&D Systems) per the manufacturer's instructions.
Measurement of intracellular ROS production. Measurements of intracellular ROS production were made by staining cells with CellRox Deep Red reagent (ThermoFisher) which measures oxidative stress by producing fluorescence upon oxidation by ROS. PMs were isolated and treated as described above. At the time of infection, a cellular stain mixture containing CellROX Deep Red (5 M final concentration), SYTOX Green, and Hoechst 33342 was added to cocultures. After 1 h of infection, cells were washed three times with PBS before a 15-min fixation with 3.7% formaldehyde to preserve CellRox Reagent signal. Coverslips were then mounted onto glass slides and visualized with a Zeiss LSM 710 confocal microscope as described above. Images obtained were analyzed using Fiji version 1.0 (94). In order to quantify ROS production, a cellular ROS production index was calculated using the following equation: (total image intensity Ϫ (mean background fluorescence ϫ image area))/(total macrophages counted ϫ (number of macrophages with ROS production/total macrophages counted)). Images capturing only ROS staining (without other stains/channels) were measured to determine the total corrected fluorescence for the total image area. Mean background fluorescence was determined by at least three different measurements in areas of the image lacking cellular contents (95). Data are presented as the mean Ϯ SE ROS cellular production index of 10 images per sample.
Metalloproteinase ELISA. Supernatants from macrophage-GBS cocultures were collected and centrifuged as described above to remove cellular debris. Supernatants were then evaluated for the concentration of human MMP-8 and MMP-9 using DuoSet ELISA kits (R&D Systems, Minneapolis, MN) per the manufacturer's protocol, and protein levels were calculated from a standard curve.
Matrix metalloproteinase activity. MMP activity of coculture supernatants was measured using the MMP Activity Assay kit (Abcam). Supernatants were incubated with assay buffer for 30 min, and fluorescence signal was measured with a fluorescence microplate reader at an Ex/Em ratio of 490/ 525 nm. Sample values were normalized to the values for uninfected cells from the same placental sample to calculate percent change for each placental sample assayed.
Human fetal membrane infections. Fetal membrane tissue was obtained and cultured as previously described (96). Briefly, fetal membranes were excised from placental tissues. Fetal membrane tissue sections were suspended over a 12-mm Transwell Permeable Support without membrane (Corning) and immobilized using a 1/4-inch intraoral elastic band (Ormco, Orange, CA) so that the choriodecidua was oriented facing up. Both Transwell chambers were incubated with Dulbecco's modified Eagle's medium (DMEM), high-glucose, HEPES, cell culture medium with no phenol red (Gibco, Carlsbad, California) supplemented with 1% fetal bovine serum and PEN-STREP antibiotic/antimycotic mixture (Gibco). Transwells were incubated overnight at 37°C in ambient air containing 5% CO 2 before the medium was replaced with DMEM, high-glucose, HEPES, cell culture medium with no phenol red (lacking the PEN-STREP antibiotic/antimycotic mixture). Bacterial cells were added to the choriodecidual surface of the gestational membranes at a multiplicity of infection of 1 ϫ 10 6 cells per Transwell. Cocultures were incubated at 37°C in ambient air containing 5% CO 2 for 48 h at which time membrane tissues were fixed in 10% neutral buffered formalin prior to paraffin embedding.
Human fetal membrane immunohistochemistry staining. Tissues were cut into 5-m sections, and multiple sections were placed on each slide for analysis. For immunohistochemistry, slides were deparaffinized, and heat-induced antigen retrieval was performed on the Bond Max automated IHC stainer (Leica Biosystems, Buffalo Grove, IL) using their Epitope Retrieval 2 solution for 5 to 20 min. Slides were incubated with a rabbit polyclonal anti-GBS antibody (ab78846; Abcam), rabbit polyclonal antihistone H3 antibody (ab8580; Abcam), or a mouse monoclonal anti-CD163 antibody (MRQ-26; Cell Marque, Rocklin, CA) for 1 h. The Bond Polymer Refine detection system (Leica Biosystems) was used for visualization. Slides were dehydrated and cleared, and coverslips were added before light microscopy analysis was performed.
Human fetal membrane immunofluorescence staining. For immunofluorescence evaluation of METs within fetal membrane tissue, tissues were fixed and sectioned as described above. Sections were briefly incubated with xylene to deparaffinize. Tissues were blocked for more than 1 h with 10% bovine serum albumin (Sigma-Aldrich) before staining with 1/100 dilutions of mouse monoclonal anti-H3 antibodies conjugated with Alexa Fluor 647 (ab205729; Abcam), rabbit monoclonal anti-CD163 antibodies conjugated with Alexa Fluor 488 (ab218293; Abcam), and mouse monoclonal anti-MMP-9 antibodies conjugated with Alexa Fluor 405 (NBP-259699AF405; Novus Biological, Littleton, CO) overnight at room temperature. Additional tissue staining was conducted as previously described (45). Tissues were deparaffinized and then incubated in R Universal epitope recovery buffer (Electron Microscopy Sciences, Hatfield, PA) at 50°C for 90 min. Samples were then rinsed in deionized water three times followed by washing with Tris-buffered saline (TBS) (pH 7.4). Samples were permeabilized for 5 min with 0.5% Triton X-100 in TBS at room temperature, followed by three washes with TBS. Samples were then blocked with TBS with 10% BSA for 30 min prior to incubation with 1:50 dilutions of rabbit polyclonal anti-neutrophil elastase antibodies (catalog no. 481001; MilliporeSigma, Burlington, MA) and mouse monoclonal anti-H3 antibodies conjugated with Alexa Fluor 647 in blocking buffer at room temperature overnight. The following day, samples were washed in TBS, followed by repeat blocking with blocking buffer for 30 min at room temperature before incubation with 1/00 dilution of Alexa Fluor 488-conjugated donkey anti-rabbit IgG (Invitrogen) for 4 h at room temperature. Samples were then washed and incubated with 5 M Hoechst 33342 for 30 min to stain nuclei. After the final washes, slides were dried and coverslips were added to the slides. Tissues were visualized with a Zeiss LSM 710 META inverted laser scanning confocal microscope. Images shown are representative of four separate experiments using tissues from different placental samples.
Statistics. Statistical analysis of MET quantifications was performed using one-way ANOVA with either Tukey's or Dunnet's post hoc correction for multiple comparisons, and all reported P values are adjusted to account for multiple comparisons. MMP activity assays and bacterial killing assay were normalized to untreated or uninfected cells and analyzed with Student's t test or one-way ANOVA. P values of Յ0.05 were considered significant. All data analyzed in this work were derived from at least three biological replicates (representing different placental samples). Statistical analyses were performed using GraphPad Prism 6 for MAC OS X Software (version 6.0g; GraphPad Software Inc., La Jolla, CA).

ACKNOWLEDGMENTS
We thank Oscar Gomez-Duarte and Shannon Manning for providing the clinical bacterial isolates used in this study.
Core services, including use of the Cell Imaging Shared Resource, were performed through support from the Vanderbilt Institute for Clinical and Translational Research program supported by the National Center for Research Resources (grant UL1 RR024975-01) and the National Center for Advancing Translational Sciences (grant 2 UL1 TR000445-06). Deidentified, human fetal membrane tissue samples were provided by the Cooperative Human Tissue Network at Vanderbilt University, which is funded by the National Cancer Institute.
The funders of this study had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We have no conflicts of interest to disclose.