Viral Infection Sensitizes Human Fetal Membranes to Bacterial Lipopolysaccharide by MERTK Inhibition and Inflammasome Activation

Chorioamnionitis, preterm premature rupture of membranes (PPROM) and preterm birth resulting from infection are thought to be initiated by bacteria ascending from the lower genital tract, gaining access to the fetal membranes (FMs), and activating innate immune responses (1). The proinflammatory cytokine IL-1β is an important mediator of PPROM and preterm birth (2–5). Normal human FMs express a range of innate immune pattern recognition receptors, such as TLRs, Nod-like receptors, and inflammasome family members, and can generate inflammatory responses following their activation by infectious components (6–8). Although clinical and experimental studies have correlated bacterial infection and inflammation at the maternal–fetal interface with prematurity (9–16), no single bacterium has been attributed to preterm birth (17), and identifiable bacteria associated with chorioamnionitis, PPROM, and preterm birth are often common to the genital tract and the placenta (18). Moreover, although the FMs are likely the first tissue colonized by the normal flora of the lower genital tract or by an ascending pathogen (19), most FMs from normal deliveries also have bacteria present (20). Thus, bacterial stimulation of the FMs may, in and of itself, be insufficient to trigger chorioamnionitis and preterm birth.

A number of diseases are caused by polymicrobial infections, including disorders of the urogenital tract, like vaginosis (21). Thus, one potential risk factor that could contribute to bacterial-associated preterm birth may be another type of infection, such as a virus. Although not all women with a viral infection during pregnancy will have complications, some viruses that are detected in the amniotic fluid or gestational tissues have been linked to an increased risk for chorioamnionitis and preterm birth. These include adenovirus and herpes viruses, such as CMV, EBV, and HSV (22–31). If a virus, which can infect the placenta and FMs increases a woman’s risk for preterm birth by altering local responses to bacterial components, then the mechanisms likely involve modulation of innate immune receptors and their regulators. TLR-driven immune responses are tightly controlled by inhibitors, including the TAM tyrosine kinase receptors (32, 33). Three TAM receptors, TYRO3, AXL, and MERTK, are activated by two endogenous ligands: growth arrest specific 6 (GAS6) and protein S1 (PROS1) (33). GAS6 activates all three TAM receptors, whereas PROS1 activates TYRO3 and MERTK (33). Upon ligand binding, TAM receptors trigger STAT1 phosphorylation, inducing SOCS1 and SOCS3, which broadly inhibit TLR signaling (33, 34). In this study, we investigated how a polymicrobial infection could impact human FM innate immune responses and, thus, pregnancy outcome. Using an ex vivo human FM explant system and an in vivo mouse model of pregnancy, we examined the effect that a herpes virus infection had on FM responses to low levels of bacterial LPS, as well as the role of the regulatory TAM receptors.

Human FMs were collected from planned uncomplicated term (37–41-wk) cesarean deliveries without labor or known infection/inflammation, as previously described (7, 8). Tissue collection was approved by Yale University’s Human Research Protection Program. After washing the FMs with sterile PBS supplemented with penicillin (100 U/ml) and streptomycin (100 μg/ml) (Life Technologies, Grand Island, NY), adherent blood clots were removed, and explants in which the chorion and amnion were intact were obtained using a 6-mm biopsy punch. The FM explants were placed in 0.4-μm pore cell culture inserts (BD Falcon, Franklin Lakes, NJ) with 500 μl of DMEM (Life Technologies) supplemented with 10% FBS (HyClone, Logan, UT), and these were placed in a 24-well plate containing 500 μl of the same DMEM medium for 24 h, as previously described (7, 8, 35). The next day, the medium was removed and replaced with serum-free Opti-MEM (Life Technologies). After 3 h, treatments were initiated in serum-free Opti-MEM.

FM explants were treated or not for 24 h with murine γ herpes virus 68 (MHV-68; 1.5 × 10⁴ PFU/ml) (36), HSV-2 (6.4 × 10² PFU/ml), or the viral dsRNA mimic polyinosinic-polycytidylic acid [Poly(I:C)] (high m.w., 20 μg/ml; InvivoGen, San Diego, CA). FMs were then treated or not with LPS isolated from Escherichia coli 0111:B4 (Sigma-Aldrich, St. Louis, MO) at 1 or 100 ng/ml. For some experiments during the LPS treatment, FMs were also treated or not with the caspase-1 inhibitor Z-WEHD-FMK (1 μM; R&D Systems, Minneapolis, MN) (7, 8), the NLRP3 inflammasome inhibitor 3,4-methylenedioxy-β-nitrostyrene (MNS; 10 μM; Cayman Chemical, Ann Arbor, MI) (37), or recombinant (r) human GAS6 (50 ng/ml; R&D Systems). Twenty-four hours later, culture supernatants and FM tissues were collected, snap-frozen, and stored at −80°C until further analysis. In separate experiments, FMs were pretreated for 30 min with blocking Abs (0.5 μg/ml) to human TYRO3 (mouse mAb; catalog number MAB859), human AXL (goat polyclonal; catalog number AF154), and human MERTK (goat polyclonal; catalog number AF891; all from R&D Systems). FMs were also pretreated with isotype-control Abs mouse IgG1 (catalog number MAB002) and goat IgG (catalog number AB-108-C) at the same concentrations (both from R&D Systems). FMs were then treated or not with LPS (1 ng/ml), and culture supernatants and FM tissues were collected 24 h later and stored.

All mouse studies were approved by Yale University’s Institutional Animal Care and Use Committee. Pregnant wild-type C57BL/6 or pregnant AXL^−/−MERTK^−/− mice (38) were injected i.p. with PBS or low-dose LPS (20 μg/kg) on embryonic day (E)15.5 (36, 39). Pregnant wild-type C57BL/6 mice were also injected i.p. with PBS or MHV-68 (1 × 10⁶ PFU) on E8.5, followed by PBS or LPS (20 μg/kg) injected i.p. on E15.5. After 6 h, mice were sacrificed. FMs were collected, pooled, snap-frozen, and stored at −80°C until further analysis.

Supernatants were measured for IL-1β by ELISA (R&D Systems), and the following cytokines/chemokines were measured by multiplex analysis (Bio-Rad, Hercules, CA): IL-6, IL-8, IL-10, IL-12, IL-17, G-CSF, GM-CSF, IFN-γ, MCP-1/CCL2, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, TNF-α, vascular endothelial growth factor (VEGF), growth regulated oncogene-α (GRO-α), and IFN-γ–induced protein 10 (IP-10/CXCL10) (7, 8). Soluble (s)MERTK was measured by ELISA (R&D Systems). Synergistic responses induced by combined treatments were defined as greater than the additive value of the two treatments alone.

Human and mouse FMs were homogenized, and total RNA was extracted as previously described (6, 40). Quantitative real-time RT-PCR was performed using the KAPA SYBR FAST qPCR Kit (Kapa Biosystems, Woburn, MA), and PCR amplification was performed on the Bio-Rad CFX Connect Real-time System. Data were normalized to the housekeeping gene GAPDH, analyzed using the ΔΔ threshold cycle method, and presented as relative abundance. The primers used are shown in Table I.

Human FM explants were homogenized, and proteins were extracted and quantified as previously described (8, 35). TYRO3, AXL, and MERTK levels were evaluated by Western blot, as previously described (8), using the anti-human primary Abs against TYRO3 (catalog number MAB859), AXL (catalog number AF154), and MERTK (catalog number AF891; all from R&D Systems). IL-1β levels were evaluated using the anti-human primary Abs against pro–IL-1β (product number 12703) and the active form (product number 2022; both from Cell Signaling Technology). β-Actin was used as a loading control (Sigma-Aldrich). Images were recorded, and semiquantitative densitometry was performed using the Gel Logic 100 system and Carestream software (Carestream Molecular Imaging, Woodbridge, CT). ELISA was used to measure the tissue levels of GAS6 (R&D Systems) and total PROS1 (Innovative Research, Novi, MI).

Each in vitro FM treatment experiment was performed in triplicate and repeated at least three times. For in vivo studies, all FMs from each pregnant animal were pooled. All data are reported as mean ± SEM of pooled experiments. The number of independent experiments or individual mice from which data were pooled are indicated in the figures or figure legends as n. Statistical significance (p < 0.05) was determined by one-way ANOVA or a nonparametric test for multiple comparisons or a t test or a Wilcoxon matched-pairs signed-rank test for the comparison of two groups, using Prism software (GraphPad, La Jolla, CA).

As previously reported (8), low levels of bacterial LPS significantly upregulated normal human FM explant secretion of IL-1β compared with the not treated (NT) control (Fig. 1A–C). Infection with the herpes virus MHV-68 had no significant effect on FM IL-1β secretion compared with the NT control (Fig. 1A). However, similarly to human placental trophoblast cells (36, 39), MHV-68 efficiently infected human FM tissues. At 24 h postinfection, the FM viral load was 7.76 × 10⁵/500 ng of DNA, as measured by quantitative PCR for the MHV-68 early-late lytic gene ORF-53 (36, 41) (data not shown). In combination with LPS, pretreatment with MHV-68 significantly and synergistically augmented IL-1β secretion, as detected by ELISA, by 3.4 ± 1.4–fold compared with LPS alone and by 6.0 ± 1.1–fold compared with MHV-68 alone (Fig. 1A). Western blot analysis of the culture supernatants confirmed that only the mature active form of IL-1β was released from the FM tissue; no precursor was detected in the culture media (data not shown). When FMs were treated with LPS, followed by MHV-68 infection, a similar synergistic 5.2 ± 2.9–fold augmentation of IL-1β secretion was seen (data not shown). However, because we sought to build on previous studies that treated with MHV-68 prior to LPS exposure (36, 39), we continued our studies using this model.

To validate the findings for a human viral infection, human FMs were infected with HSV-2 prior to LPS exposure. HSV-2 alone had no effect on FM IL-1β secretion compared with NT control. However, HSV-2 infection significantly and synergistically augmented IL-1β secretion by 1.9 ± 0.4–fold compared with LPS alone (Fig. 1B). Similarly, the viral dsRNA mimic Poly(I:C) alone did not induce an FM IL-1β response, as previously reported (7). However, in combination with LPS, pretreatment with Poly(I:C) also significantly and synergistically augmented IL-1β secretion by 1.8 ± 0.2–fold compared with LPS alone and by 28.8 ± 4.5–fold compared with Poly(I:C) alone (Fig. 1C). Of note, although Poly(I:C) and HSV-2 had similar efficacies, MHV-68 was more efficient (1.7-fold) at augmenting LPS-induced IL-1β secretion by FMs.

To validate our in vitro findings in vivo, pregnant wild-type mice were injected with PBS or MHV-68 at E8.5, followed by PBS or low-dose LPS at E15.5, as previously described (36, 39). Exposure to LPS alone or MHV-68 alone had no significant effect on IL-1B mRNA levels in mouse FMs compared with the PBS control. However, the combination of MHV-68 and LPS induced a significantly synergistic increase in FM IL-1B mRNA expression that was 3.1 ± 0.7–fold higher compared with LPS alone and 4.0 ± 0.9–fold higher compared with MHV-68 alone (Fig. 1D).

Having established in a number of systems that a viral infection or viral dsRNA sensitizes FMs to bacterial LPS by synergistically augmenting IL-1β production, we investigated the mechanism by which this response was mediated. Using the model of human FMs infected with MHV-68, the pro- and active forms of IL-1β were measured. Under NT conditions, FM tissues did not express detectable levels of either form of IL-1β (Fig. 2A). Treatment with LPS alone significantly induced expression of pro–IL-1β and processing into its active form. Although treatment with MHV-68 alone induced some active and pro–IL-1β expression by FMs, the levels were not significantly different from the NT control (Fig. 2A). MHV-68 and LPS in combination significantly induced pro–IL-1β expression to levels similar to LPS alone. Furthermore, MHV-68 and LPS in combination significantly and synergistically induced the processing of 7.9 ± 2.3–fold more IL-1β into its active form compared with LPS alone (Fig. 2A). Inhibition of the common mediator of IL-1β processing, caspase-1 (42), significantly reduced FM secretion of IL-1β in response to combined MHV-68 and LPS by 53.3 ± 8.7% (Fig. 2B). Inhibition of NLRP3 inflammasome activity using the inhibitor MNS (37) significantly reduced FM secretion of IL-1β in response to combined MHV-68 and LPS by 43.3 ± 11.3% (Fig. 2C).

A broader range of cytokines and chemokines secreted by human FMs in response to MHV-68, HSV-2, and Poly(I:C), either alone or in combination with LPS, was examined. Data from these studies have been summarized in Table II. Treatment of FMs with LPS alone significantly increased the secretion of IL-6, IL-8, IL-10, IL-12, IL-17, G-CSF, GM-CSF, IFN-γ, MCP-1, MIP-1β, RANTES, TNF-α, VEGF, GRO-α, and IP-10 compared with the NT control, whereas it had no significant effect on MIP-1α production (Figs. 3–5).

As shown in Fig. 3, infection of FMs with MHV-68 alone significantly increased the secretion of IL-6, IL-8, IL-10, IL-12, IL-17, G-CSF, IFN-γ, and GRO-α compared with the NT control. MHV-68 infection alone significantly reduced basal FM secretion of MCP-1 and had no significant effect on FM GM-CSF, MIP-1α, MIP-1β, RANTES, TNF-α, VEGF, or IP-10 secretion (Fig. 3). When FMs were treated with MHV-68 and then exposed to LPS, secretion of IL-6, IL-8, G-CSF, and GRO-α was further significantly increased compared with LPS alone and, with the exception of IL-8, compared with MHV-68 alone, all in an additive manner. In contrast, MHV-68 infection of FMs, followed by exposure to LPS, significantly inhibited the LPS-induced secretion of MCP-1, TNF-α, and IP-10 by 84.7 ± 5.2, 68.3 ± 8.8, and 52.9 ± 10.0%, respectively. The secretion of IL-10, IL-12, IL-17, GM-CSF, IFN-γ, MIP-1α, MIP-1β, RANTES, and VEGF was not significantly altered by the combination of MHV-68 and LPS compared with LPS alone or, with the exception of RANTES, compared with MHV-68 alone (Fig. 3, Table II).

As shown in Fig. 4, infection of FMs with HSV-2 alone had no significant effect on the FM secretion of any of the factors tested. When FMs were treated with HSV-2 and then exposed to LPS, FM secretion of G-CSF, MIP-1α, and GRO-α was significantly and synergistically augmented by 1.3 ± 0.1–fold, 1.2 ± 0.1–fold, and 1.2 ± 0.1–fold, respectively, compared with LPS alone. Similarly to infection with MHV-68, HSV-2 significantly reduced FM secretion of MCP-1 in response to LPS by 16.0 ± 7.3%. The secretion of IL-6, IL-8, IL-10, IL-12, IL-17, GM-CSF, IFN-γ, MIP-1β, RANTES, TNF-α, VEGF, or IP-10 was not significantly altered by the combination of HSV-2 and LPS compared with LPS alone (Fig. 4, Table II).

As shown in Fig. 5, treatment of FMs with Poly(I:C) alone significantly increased the secretion of IL-6, IL-17, G-CSF, GM-CSF, IFN-γ, MCP-1, MIP-1β, RANTES, TNF-α, VEGF, GRO-α, and IP-10 compared with the NT control. Similar to infection with MHV-68, pretreatment of FMs with Poly(I:C) significantly augmented the LPS-induced secretion of IL-6, G-CSF, and GRO-α compared with LPS or Poly(I:C) alone, in an additive manner. However, additional factors were also augmented in a similar manner. Poly(I:C) significantly augmented LPS-induced FM secretion of GM-CSF, VEGF, and IP-10 in an additive manner compared with LPS alone or, with the exception of IP-10, compared with Poly(I:C) alone. Also, similarly to MHV-68, pretreatment with Poly(I:C) significantly inhibited the LPS-induced FM secretion of TNF-α by 36.6 ± 1.3%. Similar to infection with HSV-2, the combination of Poly(I:C) and LPS significantly and synergistically augmented FM secretion of MIP-1α by 206.6 ± 55.5–fold compared with LPS alone and by 2563.9 ± 179.1–fold compared with Poly(I:C) alone. The combination of Poly(I:C) and LPS also significantly and synergistically augmented FM secretion of RANTES by 1.6 ± 0.1–fold compared with LPS alone and by 4.7 ± 0.2–fold compared with Poly(I:C) alone. The secretion of IL-8, IL-10, IL-12, IL-17, IFN-γ, MCP-1, and MIP-1β was not significantly altered by the combination of Poly(I:C) and LPS compared with LPS alone or, with the exception of IL-10 and MIP-1β, compared with Poly(I:C) alone (Fig. 5, Table II).

To better understand the mechanism by which viral infection of human FMs synergistically augmented the LPS-induced production of IL-1β, expression of the TAM receptor family was examined in these tissues. Under control conditions, human FM explants expressed the TAM receptors TYRO3, AXL, and MERTK, as well as their ligands GAS6 and PROS1, at the mRNA level, although TYRO3 expression was very low (Fig. 6A). This was reflected at the protein level, because, under NT conditions, FMs expressed AXL (Fig. 6B, 6D) and MERTK (Fig. 6C, 6D), whereas expression of TYRO3 was undetectable (data not shown).

Treatment of human FMs with MHV-68 or LPS, alone or in combination, had no significant effect on AXL protein expression levels (Fig. 6B). MERTK protein expression by FMs treated with MHV-68 plus LPS was reduced significantly (by 42.2 ± 5.3%) compared with the NT control and compared with LPS alone (by 34.3 ± 9.7%) (Fig. 6C). Similarly to FMs exposed to MHV-68 plus LPS, Poly(I:C) plus LPS significantly reduced FM MERTK protein expression by 38.7 ± 9.2% compared with the NT control (Fig. 6D). However, Poly(I:C) plus LPS significantly increased FM AXL protein expression by 2.1 ± 0.3–fold compared with the NT control (Fig. 6D). To assess whether the reduction in MERTK expression correlated with decoy receptor release (43), sMERTK levels were measured. Treatment of FM explants with LPS alone or MHV-68 alone significantly reduced FM sMERTK levels by 36.4 ± 19.4 and 44.8 ± 19.2%, respectively, compared with the NT control (Fig. 6E). MHV-68 in combination with LPS significantly augmented sMERTK by 1.7 ± 0.6–fold compared with MHV-68 alone, to near baseline (NT) levels (Fig. 6E).

The presence of the common TAM receptor agonist rGAS6 significantly increased AXL expression in FMs exposed to MHV-68 plus LPS by 1.6 ± 0.2–fold (Fig. 6B), and it significantly increased MERTK expression in FMs exposed to MHV-68 alone by 1.3 ± 0.1–fold (Fig. 6C). Although significance was not reached, rGAS6 increased MERTK expression in FMs exposed to MHV-68 in combination with LPS by 2.4 ± 0.7–fold (Fig. 6C). FM expression of GAS6 and total PROS1 protein was also evaluated. As shown in Fig. 6F, treatment of FMs with MHV-68, either alone or in combination with LPS, significantly increased GAS6 levels compared with the NT control. In contrast, PROS1 levels did not change significantly with any of the treatments (Fig. 6G); however, under NT, LPS, and MHV-68 plus LPS conditions, the presence of rGAS6 significantly elevated PROS1 levels (Fig. 6G).

Having established that MHV-68 plus LPS inhibited FM MERTK expression and promoted sMERTK levels, and that this was reversed by rGAS6, we next sought to determine whether this modulation of the TAM receptor pathway played a direct role in the IL-1β response. Human FM IL-1β secretion in response to MHV-68 plus LPS was significantly inhibited by 41.2 ± 10.4% in the presence of rGAS6 (Fig. 7Ai). To further explore the mechanism by which GAS6 regulated FM IL-1β secretion in response to MHV-68 plus LPS, Western blot of tissue lysates for active and pro–IL-1β was performed. rGAS6 had no effect on the levels of pro–IL-1β under all conditions tested (Fig. 7Aii). However, rGAS6 significantly inhibited FM expression of active IL-1β by 52.3 ± 5.8% when treated with MHV-68 plus LPS (Fig. 7Aiii).

To further confirm a role for the TAM receptors in modulating human FM responses to LPS, their function was inhibited using blocking anti-TAM Abs instead of virus. Combination anti-TAM Abs significantly augmented LPS-induced human FM IL-1β secretion by 1.6 ± 0.05–fold compared with the levels secreted in response to LPS in the presence of isotype-Ab controls (Fig. 7B). To validate the role of the TAM receptors in regulating human FM responses to LPS, an in vivo mouse model was used. Similar to human FMs, FMs collected from pregnant wild-type mice expressed the TAM receptors TYRO3, AXL, and MERTK, as well as their ligands GAS6 and PROS1, at the mRNA level, although, again, TYRO3 expression was very low (Fig. 7C). Because mouse FMs predominantly expressed AXL and MERTK, we used double-knockout mice as a surrogate for a viral infection (38). Thus, wild-type or AXL^−/−MERTK^−/− mice were exposed to PBS or LPS at a dose that does not induce inflammation or preterm birth in wild-type mice (36, 39). FMs collected from pregnant wild-type mice exposed to low levels of LPS expressed levels of IL-1B that were similar to those in pregnant wild-type mice exposed to PBS. However AXL^−/−MERTK^−/− mice exposed to low levels of LPS expressed significantly higher levels of IL-1B compared with PBS-treated AXL^−/−MERTK^−/− mice (5.0 ± 2.4–fold increase) and LPS-treated wild-type mice (3.6 ± 1.7–fold increase) (Fig. 7D).

Although chorioamnionitis, PPROM, and subsequent preterm birth are associated with infection and inflammation, the underlying mechanisms involved are not fully understood. Furthermore, although local bacterial infections may be easily identified, other more difficult-to-detect infections may still play a role in the pathogenesis of preterm birth. Indeed, viral infections are becoming increasingly linked to pregnancy complications and, thus, represent an important clinical problem. It is established that viruses able to cross the placenta and infect the fetus, such as Zika virus, CMV, and rubella, often cause congenital defects and adverse pregnancy outcomes (44–46). Much less is known about viruses that may not readily transmit vertically during pregnancy, like HSV-2 and influenza H1N1. However, such viruses may still impact maternal and fetal morbidity and mortality by infecting gestational tissues and modulating local innate immune responses (47–52). The results presented in this article indicate that a herpes viral infection that infects, but does not cross, the placenta (36, 39) sensitizes human FM tissue to low levels of bacterial LPS, giving rise to an exaggerated inflammatory IL-1β response. Using an ex vivo human FM explant system and an in vivo mouse model of pregnancy, we have found that the mechanism by which this aggravated FM inflammation arises is through disabling of the TAM receptor pathway and subsequent activation of the NLRP3 inflammasome.

In chorioamnionitis, PPROM, and preterm birth, the most commonly isolated microbes are bacteria that normally colonize the lower genital tract, including E. coli, Streptococcus, and Ureaplasma (53–58). Mixed bacterial infections within the amniotic cavity have been demonstrated in preterm birth (53, 56) and chorioamnionitis (59), suggesting that the normal (or pathologic) bacteria of the genital tract might be an underlying cause. A recent study tested this by exposing human FM explants to a mixed bacterial culture of E. coli and Streptococcus agalactiae; however, there was no difference in the inflammatory responses generated after mixed or single infection (60). Thus, this normal (or abnormal) bacterial flora alone may not be sufficient to trigger an inflammatory response at the maternal–fetal interface that can initiate preterm birth, suggesting a role for polymicrobial infections of different types. Increasingly, infectious and inflammatory diseases are being attributed to combinations of different infectious kingdoms (21), including pregnancy complications like preterm birth (61). Furthermore, the impact that viruses may have on pregnancy outcomes is beginning to be appreciated (31, 62, 63). The current study set out to test the impact that a polymicrobial infection could have on human FM innate immune responses and the mechanisms involved.

In this article, we report that infection of human FM with herpes virus MHV-68 synergistically augmented the processing and secretion of mature IL-1β in response to low levels of bacterial LPS; this was mediated by activation of the inflammasome, most likely NLRP3, which is expressed by human FMs (8). That we only detected mature IL-1β in the culture supernatants indicates that combined MHV-68 and LPS treatment of FMs promoted classical pyroptosis-associated inflammasome activation. The reduction in FM IL-1β secretion by MNS, which blocks the assembly of NLRP3 inflammasome (37), and by the caspase-1 inhibitor, a major inflammasome component, indicates that inflammasome activation, most likely NLRP3, mediated this response. However, there is the possibility that other non-NLRP3 inflammasomes might be involved, because MNS can also inhibit Syk kinase signaling (37). Our finding is in keeping with recent clinical observations showing that FMs from patients with preterm birth or acute histologic chorioamnionitis have elevated IL-1β, and involvement of the inflammasome is indicated (64–66). Our in vivo observations of augmented FM IL-1β production with the combination of MHV-68 and LPS also complement previous reports that MHV-68 infection during pregnancy, in combination with bacterial LPS, triggered elevated uterine and placental inflammation and induced preterm birth (36, 39). Similarly, intrauterine delivery of viral dsRNA, which is produced by many viruses, including herpes virus (67), in combination with bacterial peptidoglycan, amplifies inflammation and preterm delivery (68). That we observed a similar synergistic augmentation of IL-1β secretion whether FMs were infected prior to or after LPS demonstrates that this response is due to the polymicrobial exposure rather than the timing of virus and LPS.

Although treatment of FMs with LPS induced strong pro–IL-1β expression, low-dose LPS alone was not sufficient to trigger strong IL-1β processing and secretion of active IL-1β. Furthermore, unlike the secretion of IL-1β in response to combination MHV-68 and LPS that was inhibited by MNS, IL-1β secretion by LPS alone appeared to be NLRP3 independent, unlike higher concentrations (69). Similarly, although MHV-68 alone induced some FM pro–IL-1β expression, there was not significant IL-1β processing. Instead, it was the combination of LPS and MHV-68 that significantly augmented FM IL-1β processing and subsequent secretion and secretion of mature IL-1β. This suggests that LPS, by activating TLR4, serves as a classical signal 1 and that MHV-68 may provide signal 2 by putatively activating NLRP3. NLRP3 can be activated by viruses (70, 71), including the herpes virus, HSV-1, although how is unclear (72). How MHV-68 may be contributing to NLRP3 inflammasome activation is an area for future study.

HSV-2 infection and viral dsRNA [Poly(I:C)] had a similar impact on human FM IL-1β secretion in response to LPS, and they may also activate the NLRP3 inflammasome (73).

Thus, the observed augmented IL-1β by virally infected human FMs in response to low levels of bacterial LPS might be common to viruses able to produce dsRNA (67) and activate the TLR3 pathway (74). However, we did note a difference in the magnitude of the response. Although Poly(I:C) and HSV-2 had similar efficacies, MHV-68 was more efficient at augmenting LPS-induced IL-1β secretion by FMs. This may be due to additional mechanism(s) used by MHV-68. Indeed, this possibility is highlighted by the additional cytokine/chemokine data in which we observed overlapping and differential responses for MHV-68, HSV-2, and Poly(I:C). HSV-2 and Poly(I:C) synergistically augmented LPS-induced MIP-1α, whereas MHV-68 and Poly(I:C) additively augmented LPS-induced IL-6, G-CSF, and GRO-α secretion by FMs, again indicating a potential role for TLR3. MHV-68 and HSV-2 suppressed LPS-induced FM MCP-1, whereas MHV-68 and Poly(I:C) reduced LPS-induced TNF-α secretion. MHV-68 also reduced LPS-induced IP-10, whereas HSV-2 synergistically augmented LPS-induced GRO-α. The inhibition of TNF-α by MHV-68 has been reported in CD8⁺ T cells stimulated by the viral protein encoded by ORF-61 (75). Thus, similar mechanisms may be involved in MHV-68–infected FMs for the suppression of LPS-induced TNF-α, MCP-1, and IP-10. Alternatively, or in combination, the MHV-68– and LPS-induced production of IL-10 may be involved in the suppression of TNF-α and IP-10 (76). HSV immediate-early protein ICP0 has been shown to inhibit TLR4-mediated inflammatory responses to HSV and, thus, may account for HSV-2 suppression of the FM MCP-1 responses (77). Where the FM cytokine/chemokine profiles diverge in response to MHV-68, HSV-2, or Poly(I:C) combined with LPS likely reflects the ability of different live viruses to simultaneously regulate a number of host innate immune pathways, as well as produce their own immunomodulatory factors.

Because IL-1β is an important mediator of PPROM and preterm birth (2–5), and its production by human FMs was synergistically increased in our model of a polymicrobial infection for MHV-68, HSV-2, and Poly(I:C), as well as in mouse FMs, subsequent mechanistic studies focused on this inflammatory cytokine. Activation of the TAM receptors (TYRO3, AXL, MERTK) by their ligands (GAS6, PROS1) restrains TLR signaling, keeping the constitutive chemokine/cytokine expression regulated (33). Because FMs constitutively express high levels of AXL, MERTK, GAS6, and PROS1, we hypothesized that MHV-68 infection removed this brake, allowing heightened TLR4-mediated IL-1β production in response to LPS. Indeed, MHV-68 and Poly(I:C), in combination with LPS, reduced FM MERTK expression. Compared with treatment alone, LPS and MHV-68 also augmented sMERTK levels, which act as a decoy receptor for GAS6 (43), indicating reduced FM TAM receptor function. The increased GAS6 production under combined LPS and MHV-68 conditions may indicate a compensatory mechanism that was insufficient to restore receptor expression and function. Moreover, in the absence of viral stimulation, blocking human FM TAM receptor function also sensitized the tissue to LPS, augmenting IL-1β production. This is in contrast to studies using other enveloped viruses that can bind GAS6 and activate TAM receptors through their surface expression of phosphatidylserine (78). To further validate this finding, we turned back to an in vivo model of pregnancy using wild-type mice, which are hyporesponsive to low-dose LPS in terms of placental inflammation and pregnancy outcome (36, 39, 79, 80), and AXL^−/−MERTK^−/− mice, which generate hyperresponsive immune reactions to low-dose LPS (33, 81). In concert with our ex vivo human FM studies, in which TAM receptor function was inhibited instead of treating with virus, FMs from AXL^−/−MERTK^−/− mice generated an augmented IL-1β response 6 h following exposure to low-dose LPS compared with FMs from wild-type mice. When mice are not sacrificed for tissues, the same dose of LPS induced preterm birth in AXL^−/−MERTK^−/− mice but not in wild-type animals (G. Mor, manuscript in preparation). Conversely, the addition of exogenous GAS6 to human FMs treated with MHV-68 and LPS not only inhibited the augmented IL-1β response, it increased FM expression of AXL, MERTK, and PROS1. Thus, GAS6 may activate, as well as self-regulate, expression of its own signaling pathway to further enhance TAM receptor inhibition of TLR-induced inflammation. Furthermore, although rGAS6 was able to inhibit the processing and secretion of active IL-1β in response to MHV-68 plus LPS, it was unable to alter pro–IL-1β levels, indicating that GAS6 might be suppressing virus-induced inflammasome activation. Indeed GAS6–AXL signaling may prevent NLRP3 inflammasome activation in murine macrophages (82).

One strength of this study was the use of intact FM explants; we chose to work with an FM system in which the compartments were maintained as in vivo, because contact between the chorion and amnion likely influences each other’s response (8, 83). However, this also limits our knowledge about which cell types within the tissue (amniotic epithelial, chorionic decidual, or trophoblast; resident leukocytes) are the major targets for MHV-68, HSV-2, or Poly(I:C) or are the major producer of IL-1β and other cytokines/chemokines. Thus, in future studies, we intend to dissect out the relative contribution of the chorion and amnion and the specific cell types involved in the polymicrobial response.

In summary, FM inflammatory IL-1β responses to LPS become unrestrained postinfection with herpes virus by reduced TAM receptor MERTK expression and enhanced inflammasome activation. GAS6 re-establishes the normal FM response to bacterial LPS by restoring and augmenting TAM receptor and ligand expression and function, preventing the exacerbated IL-1β response. These findings suggest a novel mechanism by which viruses alter FM responses to intrauterine bacteria, giving rise to chorioamnionitis and preterm birth.

We thank the staff of Yale–New Haven Hospital’s Labor and Birth and the Yale University Reproductive Sciences Biobank for help with tissue collection.

The authors have no financial conflicts of interest.