|Home | About | Journals | Submit | Contact Us | Français|
There are currently more than 15 million preterm births each year. We propose that gene-environment interaction is a major contributor to preterm birth. To address this experimentally, we generated a mouse model with uterine deletion of Trp53, which exhibits approximately 50% incidence of spontaneous preterm birth due to premature decidual senescence with increased mTORC1 activity and COX2 signaling. Here we provide evidence that this predisposition provoked preterm birth in 100% of females exposed to a mild inflammatory insult with LPS, revealing the high significance of gene-environment interactions in preterm birth. More intriguingly, preterm birth was rescued in LPS-treated Trp53-deficient mice when they were treated with a combination of rapamycin (mTORC1 inhibitor) and progesterone (P4), without adverse effects on maternal or fetal health. These results provide evidence for the cooperative contributions of two sites of action (decidua and ovary) toward preterm birth. Moreover, a similar signature of decidual senescence with increased mTORC1 and COX2 signaling was observed in women undergoing preterm birth. Collectively, our findings show that superimposition of inflammation on genetic predisposition results in high incidence of preterm birth and suggest that combined treatment with low doses of rapamycin and P4 may help reduce the incidence of preterm birth in high-risk women.
Preterm birth in humans, defined as parturition occurring prior to 37 weeks of gestation, is still a global problem. One in 10 babies is born preterm worldwide, accounting annually for more than 15 million premature births (1–4). With more than 1 million deaths attributed to complications arising from prematurity each year, preterm birth is a leading cause of neonatal death (1, 5, 6). In addition, premature babies who survive are at increased risk for a number of health challenges, including respiratory distress, underdeveloped organ systems, and cerebral palsy with learning and developmental disabilities (1, 2, 7). Several risk factors, including genetic predisposition, infection/inflammation, oxidative stress, short cervix, progesterone (P4) resistance, increased maternal age, and stretch signaling originating from multiple pregnancy, contribute to this multifactorial disorder (1, 8). Due to its complex nature, mechanisms underpinning preterm birth are not clearly understood.
While more than 60% of preterm births occur in developing countries in Africa and South Asia, a recent report from the WHO identified the United States as being among the ten countries with the highest numbers of preterm birth (1), underscoring the global nature of this problem. Etiologies behind high preterm birth rates in developing and developed countries may be disparate — infection/inflammation affects developing countries more, while assisted reproductive technology (ART) compounded by increased maternal age at conception as well as increased prevalence of diabetes and high blood pressure are additional risk factors in developed countries (1). These observations suggest that preterm birth is the end result of many different causative factors. Therefore, multiple approaches and model systems are warranted in addressing this problem.
Mechanistic preclinical studies utilizing mouse models of preterm birth should help in addressing this problem. Inflammatory mediators such as endotoxin (LPS) and inflammatory cytokines (such as IL-1β, IL-6, and TNF-α) are known to induce preterm labor coincident with ovarian luteolysis with a decrease in serum P4 levels (9). Along the same lines, administration of RU-486 (mifepristone), a progesterone receptor (PR) antagonist, leads to similar effects (8, 10). However, these mouse models may not adequately define the mechanism of parturition timing, since human preterm birth is considered to occur without a drop in serum P4 levels (11), although further studies are required to assess P4 levels accounting for disparate etiologies of preterm birth.
Mouse and human studies have shown that aberrations in early pregnancy can be propagated during the subsequent course of pregnancy and lead to compromised pregnancy outcomes, including preterm birth (12). We have generated a mouse model harboring a conditional uterine deletion of Trp53, encoding tumor suppressor protein p53 (Trp53loxP/loxPPgrCre/+). These mice exhibit premature decidual senescence associated with heightened mTORC1 signaling early in pregnancy. Strikingly, they show genetic predisposition to preterm birth; approximately 50% incidence of spontaneous preterm delivery with fetal death and dystocia is observed in these females without a drop in serum P4 levels (13). Progressive decidual senescence was coincident with aberrantly higher levels of COX2 and prostaglandin F synthase (PGFS), reflected in increased uterine levels of PGF2α, triggering myometrial contractility in a paracrine manner to induce preterm birth (13). These results led us to speculate that targeting COX2 and/or mTORC1 signaling could prevent preterm birth in this model. Indeed, we were successful in preventing preterm birth in Trp53loxP/loxPPgrCre/+ females by oral gavage of celecoxib, a selective inhibitor of COX2, on day 16 of pregnancy or by inhibition of mTORC1 signaling by a low dose of rapamycin, which attenuated the progression of premature decidual senescence (13, 14).
Preterm birth is the result of many contributing factors, including genetic predisposition and environmental insults. Although gene-environment interactions are thought to contribute to the disease process, this concept has not been experimentally tested in a predisposed model. Our current studies provide evidence-based proof of principle that when genetic predisposition is superimposed with inflammation, the rate of preterm birth is profoundly enhanced. We show that Trp53loxP/loxPPgrCre/+ females exposed to even a mild inflammatory insult are exquisitely susceptible to preterm birth, with a decrease in P4 levels. Intriguingly, combination therapy with rapamycin and P4 targeting both the decidual senescence imposed by Trp53 deletion and the inflammatory insult can effectively rescue preterm birth. These findings point to gene-environment interactions in preterm birth through cooperative contributions of at least two sites of action (decidua and ovary) and an effective preventative strategy.
These studies in mice prompted us to assess whether similar signaling pathways are operative in human preterm delivery. We present evidence that a signature of decidual senescence with increased mTORC1 signaling and COX2 expression is active in human preterm birth. Superficial decidual cells adherent to the term placentae of women were isolated for culture and showed increased levels of the inflammatory mediators IL-6 and IL-8 in the presence of TLR4-specific LPS, but their levels were reduced by addition of P4 and/or rapamycin. Collectively, the results suggest a close relationship between genetic predisposition and environmental insults in exacerbating preterm delivery.
To generate mice with uterine-specific deletion of Trp53, we mated Trp53loxp/loxP females with males expressing Cre recombinase driven by the Pgr promoter (PgrCre/+), as previously described (13, 14). Trp53loxP/loxPPgrCre/+ females show approximately 50% incidence of spontaneous preterm delivery with dystocia and fetal death compared with floxed littermates showing normal pregnancy outcome; preterm delivery is defined as birth occurring before day 19 of pregnancy (14). To compare the sensitivity of these females to inflammation with respect to preterm birth, we injected TLR4-specific LPS i.p. into Trp53loxp/loxPPgr+/+ and Trp53loxP/loxPPgrCre/+ females. Ultrapure LPS was used to avoid contamination by other TLR agonists normally found in commercial preparations that are often used in preterm birth studies. Trp53loxp/loxPPgr+/+ females showed 100% incidence of preterm birth and/or fetal resorption and death when injected with 75 μg TLR4-specific LPS on day 16 of pregnancy. While a dose of even 50 μg was quite effective in inducing preterm birth (71%), lower doses of LPS (10 or 37 μg) were ineffective in inducing preterm birth in floxed females (Table (Table1).1). Remarkably, an injection of 10 μg LPS on day 16 induced preterm birth with stillbirth in all Trp53loxP/loxPPgrCre/+ littermates examined (n = 20). These results clearly demonstrate that these females are exquisitely sensitive to preterm delivery.
Prostaglandins (PGs) are normally generated by the COX system, which exists in two isoforms, COX1 and COX2. While constitutive COX1 is considered to maintain basal levels of PGs, COX2-induced PGs are normally generated by inflammatory stimuli and are known to participate in parturition (13, 15, 16). Among various PGs, PGF2α is implicated in parturition timing by synchronizing myometrial contractility. We have previously shown that levels of uterine COX1 remains unaltered in Trp53loxP/loxPPgr+/+ and Trp53loxP/loxPPgrCre/+ females, while uterine COX2 levels are upregulated in Trp53loxP/loxPPgrCre/+ females, with increased levels of PGFS and PGF2α (13). We also found that upregulation of COX2 primarily occurs in the decidua, suggesting that decidua-derived PGF2α causes myometrial contractions in a paracrine manner.
In the present study, we found that COX2 immunostaining was more intense in deciduae and endothelial cells of decidual vasculature of Trp53loxP/loxPPgrCre/+ females as compared with Trp53loxP/loxPPgr+/+ females after 12-hour treatment with a low dose of LPS (10 μg) given on day 16 of pregnancy (Figure (Figure1A).1A). Increases in decidual COX2 levels in Trp53loxP/loxPPgrCre/+ females after exposure to LPS were also reflected in higher levels of PGF2α without alterations in PGE2 levels, as determined by mass spectrometric analysis (Figure (Figure1B).1B). Notably, these increased levels were similar to those in untreated Trp53loxP/loxPPgrCre/+ females, but much higher than in Trp53loxP/loxPPgr+/+ mice treated with LPS. Increased levels of PGF2α after LPS administration were attenuated by oral administration of celecoxib. These results raised the question of how even a small insult results in such profound effects on preterm birth. Notably, acute exposure to LPS did not increase the intensity of senescence-associated β-gal staining (SA-β-gal), an established marker of cellular senescence, which was already higher in Trp53loxP/loxPPgrCre/+ deciduae (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI70098DS1).
P4 is an absolute requirement for pregnancy success, and withdrawal of P4 signaling is critical for parturition (12). Therefore, we examined whether the ovarian output of P4 is altered in the Trp53loxP/loxPPgrCre/+ females receiving 10 μg LPS. Large doses of LPS ranging from 50 to 250 μg have been shown to induce ovarian luteolysis (9), the functional and structural disintegration of corpora lutea (CL); this process triggers a rapid decrease in serum P4 levels, leading to preterm birth or pregnancy termination in rodents (9, 17), although other systemic responses cannot be ruled out. Normally, Trp53loxP/loxPPgrCre/+ females do not show ovarian luteolysis or a decrease in serum P4 levels but still exhibit approximately 50% spontaneous preterm birth (13). In the present investigation, we found that an injection of even 10 μg LPS in Trp53loxP/loxPPgrCre/+ females on day 16 of pregnancy triggered a drop in serum P4 levels 12 hours after injection (Figure (Figure1C).1C). This was reflected in higher expression of ovarian Akr1c18, encoding mouse 20α-hydroxysteroid dehydrogenase (20αHSD), which metabolizes P4 to an inactive form (Figure (Figure1,1, D and E). No significant drop in P4 levels nor increased 20αHSD expression at the mRNA or protein levels was noted in Trp53loxP/loxPPgr+/+ females after such treatment. As previously shown by others (18, 19), ovarian insufficiency in LPS-treated Trp53loxP/loxPPgrCre/+ females was also reflected in upregulation of Socs1 and Socs3 expression (Supplemental Figure 2, A and B). These results suggest that ovaries in Trp53loxP/loxPPgrCre/+ mice are more sensitive to luteolysis even with a mild environmental insult, raising the question regarding the mechanism for the heightened sensitivity of Trp53loxP/loxPPgrCre/+ females to ovarian luteolysis with inflammatory stimulus, leading to preterm birth.
It is known that decidua-derived factors serve as luteotrophins to extend the lifespan of the CL and maintain luteal P4 secretion in rodents (20–22). In fact, experimentally induced decidualization in pseudopregnant rats extends the CL lifespan with continued P4 secretion, as opposed to shortened CL lifespan in pseudopregnant rats without decidualization (23). Our recent proteomics analysis found downregulation of a decidua-enriched prolactin-like hormone, prolactin-like protein J (PLP-J), in Trp53loxP/loxPPgrCre/+ deciduae (24). PLP-J was identified in rat decidua and is implicated in maintaining decidual cell survival (25–28). Our in situ hybridization results show that Prl3c1, encoding PLP-J, was highly expressed in the decidua surrounding the developing embryo on day 8 in Trp53loxP/loxPPgr+/+ females but was markedly downregulated in Trp53loxP/loxPPgrCre/+ deciduae, consistent with our previous proteomics results (Figure (Figure2A).2A). We also found that expression of this gene was downregulated at the decidual-placental interface in Trp53loxP/loxPPgrCre/+ females on day 16 (Figure (Figure2B).2B). Immunolocalization of CDX2 demarcated trophoblast cell invasion into the decidual bed on this day of pregnancy (Supplemental Figure 3). Prolactin receptor (Prlr), which is also expressed in the decidua (29, 30), was downregulated in this tissue on day 16 (Figure (Figure2B).2B). These results suggest that decidual health is inferior in Trp53loxP/loxPPgrCre/+ females, which compromises its luteotrophic task, making the CL more vulnerable to even a small insult.
As reported previously (13, 14), administration of either celecoxib or rapamycin rescues spontaneous preterm birth in Trp53loxP/loxPPgrCre/+ females without any apparent adverse effects on the dam or fetuses. These results led us to test whether this treatment would effectively reverse inflammation-exaggerated preterm birth in Trp53loxP/loxPPgrCre/+ females. First, we used rapamycin or celecoxib singly and found them to be insufficient in preventing LPS-induced preterm birth (Supplemental Figure 4 and Supplemental Table 1). We also found that although P4 (2 mg) supplementation prior to and after LPS injection extended the parturition timing in Trp53loxP/loxPPgrCre/+ females, higher rates of fetal death and/or stillbirth were frequently encountered under this condition (Supplemental Table 1). Moreover, a combination treatment of celecoxib plus rapamycin or of celecoxib and P4 did not rescue preterm birth in Trp53loxP/loxPPgrCre/+ females exposed to LPS (Supplemental Table 1). We next asked whether combinatory treatments with celecoxib, P4, and/or rapamycin would rescue preterm birth with neonatal survival. Both floxed and deleted mice received an oral gavage of rapamycin (0.25 mg/kg BW) on days 8, 12, and 16 of pregnancy, followed by an oral gavage of celecoxib (10 mg/kg BW) twice on day 16, once 3 hours prior to and 4 hours after LPS (10 μg) injection. In addition, P4 was given twice on day 16 at around the same time points as celecoxib. This combination treatment rescued preterm birth in Trp53loxP/loxPPgrCre/+ females exposed to LPS, with survival of a full complement of pups (Figure (Figure3,3, A–C, and Supplemental Table 2); maternal weight gain due to fetal growth from day 16 to delivery and neonatal pup growth over a period of 10 days were comparable to those of untreated Trp53loxP/loxPPgr+/+ females with term delivery (Supplemental Figure 5, A and B). However, this treatment schedule adversely affected fetal viability, with high incidence of resorption in littermate Trp53loxP/loxPPgr+/+ females (Figure (Figure3C).3C). These results were surprising and led us to reevaluate our approach to treating LPS-induced preterm birth in Trp53loxP/loxPPgrCre/+ females without incurring adverse effects on fetal survival in control floxed littermates. We found that a combination of rapamycin and P4 was not only sufficient to rescue preterm birth in Trp53loxP/loxPPgrCre/+ females, but also did not significantly alter pregnancy outcome in Trp53loxP/loxPPgr+/+ females (Figure (Figure3,3, A–C, and Supplemental Table 2). Again, this treatment did not interfere with maternal weight gain due to fetal growth during pregnancy or neonatal growth over a period of 10 days in either group (Supplemental Figure 5, A and B). An alternative schedule of rapamycin treatment on days 8, 10, and 12 of pregnancy with P4 on day 16 was also effective in rescuing preterm birth in Trp53loxP/loxPPgrCre/+ females and did not result in adverse pregnancy outcome in Trp53loxP/loxPPgr+/+ females (Supplemental Table 3). The combined treatment of rapamycin and P4 prevented an increase in decidual COX2 (Figure (Figure3D3D and Supplemental Figure 6), and this was reflected in reduced levels of PGF2α compared with those in Trp53loxP/loxPPgrCre/+ females treated with LPS alone (Figure (Figure3E).3E). Rapamycin and P4 treatment also reduced the expression level of ovarian Akr1c18 in LPS-treated Trp53loxP/loxPPgrCre/+ females (Figure (Figure1D),1D), along with downregulation of ovarian Socs1 and Socs3 expression (Supplemental Figure 2). Collectively, these results point toward a potential therapeutic application for this combination treatment in preterm birth.
The next objective was to see whether our findings in mice were applicable in the case of human preterm birth. Placentae to which the decidua basalis remained adherent were collected from women following vaginal delivery at term (37–41 weeks of gestation) and preterm (25–36 weeks); the etiology of preterm birth ranged from unknown to diagnosed infection (Supplemental Table 4). When placental-decidual sections were processed for SA-β-gal staining, we found clear evidence of positive β-gal staining in preterm deciduae, with little or no staining in term deciduae (Figure (Figure4A,4A, Supplemental Figure 7A, and Supplemental Figure 8). We also found increased intensity of nuclear immunolocalization of γH2AX, another marker of senescence associated with DNA damage response, in preterm deciduae (Figure (Figure4B,4B, Supplemental Figure 7B, Supplemental Figure 8, and ref. 31). Increased expression of these senescence markers correlated with increased immunostaining of COX2 and phosphorylated ribosomal protein S6 (pS6), a signature of heightened mTORC1 signaling (Figure (Figure4,4, C and D, Supplemental Figure 7, C and D, and Supplemental Figure 8). To determine whether this decidual signature is specific to preterm delivery, we performed similar analyses in decidua-placental sections from women undergoing non-laboring cesarean deliveries at 31–36 weeks gestation due to various pathologies, including preeclampsia, placenta previa, placental abruption, and fetal anomalies/distress. The signature was not observed in these deciduae (Supplemental Figure 9 and Supplemental Table 5). Collectively, the results provide evidence that decidual senescence is associated with higher mTORC1 and COX2 signaling in deciduae from preterm deliveries, corroborating the results in Trp53loxP/loxPPgrCre/+ mice (13, 14).
With these results in hand, we questioned whether an inflammatory insult will provoke inflammatory cytokine production in decidual cells. Human decidual cells adherent to term placentae were isolated and cultured as previously described (32). Vimentin and cytokeratin immunostaining (markers of stroma-derived decidual cells and of ectoderm-derived trophoblast and amnion cells, respectively) confirmed that isolated decidual cells were approximately 99% pure and negative for CD45 (pan-immune cell marker) (Supplemental Figure 10A). Notably, TLR4 was expressed in decidual cells free of immune cells (Supplemental Figure 10B). Decidual cells were cultured in the presence or absence of TLR4-specific LPS. We found that decidual cells exposed to LPS secreted higher levels of IL-6 and IL-8 (Figure (Figure5,5, A and B), with increased decidual COX2 at the protein and mRNA levels after LPS exposure (Figure (Figure5,5, C and D). Intriguingly, higher expression levels of decidual AKR1C1, encoding human 20αHSD, were also observed (Figure (Figure5E),5E), suggesting that the decidua can be a site for P4 metabolism.
Since P4 administration reduces the incidence of preterm birth in a specific population with higher risk of preterm birth and since P4 is considered immunosuppressive and antiinflammatory (33), LPS-treated cells were cultured in the presence or absence of P4 (100 nM). We found that P4 reduced the levels of IL-6 and IL-8 in the spent media (Figure (Figure5,5, A and B). We then asked whether mTORC1 inhibitor would also suppress the levels of these cytokines in the presence of LPS. Indeed, rapamycin (1 μM) reduced the levels of IL-6 and IL-8 in the spent media (Figure (Figure5,5, A and B), with further reduction by a combination of P4 and rapamycin (Figure (Figure5,5, A and B). While higher levels of decidual PTGS2 after LPS exposure were suppressed by rapamycin, but not P4, higher levels of decidual AKR1C1 after LPS exposure were suppressed by rapamycin or P4 treatment (Supplemental Figure 11, A and B). Collectively, these results suggest that LPS can increase the release of inflammatory cytokines/mediators and the primary P4-metabolizing enzyme in decidual cells without the participation of immune cells, and that P4 and/or rapamycin can dampen these responses.
The highlights of the present study are that: (a) genetically predisposed females with uterine deletion of Trp53 are more susceptible to preterm birth if exposed to a mild inflammatory stimulus; (b) under these conditions, preterm birth appears to involve both premature decidual senescence and ovarian luteolysis with a drop in P4 levels; (c) targeting premature decidual senescence by inhibiting mTORC1 signaling and compensating the drop in P4 levels by exogenous supplementation rescue preterm birth; (d) decidual senescence with increased mTORC1 and COX2 signaling is also evident in women undergoing preterm birth (Figure (Figure6);6); and (e) cultured human term decidual cells respond to LPS, P4, and rapamycin, as predicted from mouse experiments.
Paradoxically, the role of p53 in aging is dependent upon its cellular and physiological context (34–37). It has been shown that inactivation of p53 in mouse embryonic fibroblasts results in heightened cellular senescence with increased mTORC1 signaling, which is attenuated by rapamycin (38). This is consistent with our present and previous in vivo findings (14). The finding that spontaneous preterm birth occurred in approximately 50% of Trp53loxP/loxPPgrCre/+ females with increased decidual senescence motivated us to ask whether another site of action could be targeted to reproducibly increase the incidence of preterm birth to 100% in Trp53loxP/loxPPgrCre/+ females. Indeed, our present results showing preterm birth in all Trp53loxP/loxPPgrCre/+ females treated with a low dose of LPS as opposed to none in floxed littermate females suggest that superimposition of even a mild inflammation on genetic predisposition can profoundly aggravate this phenotype. This raises the question as to the site of action of this second insult. While it is possible that Trp53loxP/loxPPgrCre/+ deciduae are more responsive to inflammation than floxed deciduae, our results showing signs of ovarian luteolysis with a drop in serum P4 levels after an exposure to 10 μg LPS in Trp53loxP/loxPPgrCre/+ females, but not in Trp53loxP/loxPPgr+/+ dams, suggest that the ovary is also a potential target of this inflammatory insult in Trp53loxP/loxPPgrCre/+ females. Although other systemic effects of LPS cannot be ruled out, a previous study showed antigen-induced inflammation during early pregnancy induces luteolysis leading to pregnancy failure, which could be rescued by P4 supplementation (19); this is consistent with our present findings. There is evidence that higher doses of LPS (50–250 μg) can trigger preterm birth in rodents with ovarian luteolysis and a decrease in P4 levels (9, 39).
Our results showing reduced expression of Prl3c1 and Prlr in deciduae of pregnant Trp53loxP/loxPPgrCre/+ mice suggest that increased sensitivity to ovarian luteolysis under mild inflammation could be due to decidual insufficiency. There is physiological and molecular evidence that decidual factors affect CL lifespan (20–23). Gibori’s group has also shown that decidual “luteotrophins” regulate ovarian adenylyl cyclase activity, luteinizing hormone receptor, and steroidogenesis (40). In addition, we and others have previously shown that implantation failure in Prlr mutant females is rescued by P4 administration, suggesting its effect at the ovarian level (29, 30). However, Prlr is also expressed in the decidua, and P4-treated Prlr mutant mice fail to give a full complement of pups and show an increased number of resorptions even with continued P4 administration, suggesting the significance of decidual Prlr in supporting the later course of pregnancy. The importance of decidual health is also reflected in our present findings of rescue of preterm birth timing in LPS-treated Trp53loxP/loxPPgrCre/+ females with P4 alone, but with a large number of resorptions and fetal deaths (~40%) (Supplemental Table 1). In contrast, in LPS-treated Trp53loxP/loxPPgrCre/+ females, treatment with rapamycin and P4 not only rescued preterm birth but maintained fetal survival (91% survival), which was comparable to that in floxed females under similar treatment conditions (Supplemental Table 2). It would be interesting to determine whether the decidua undergoing senescence influences CL function or whether both decidua and ovary are affected by inflammation in genetically predisposed females, such as Trp53loxP/loxPPgrCre/+ mice. Nonetheless, our results provide evidence that Trp53loxP/loxPPgrCre/+ deciduae and/or ovaries are more sensitive to exacerbation of preterm birth by inflammatory stimuli.
Our findings of decreased expression of AKR1C18 by rapamycin and P4 are consistent with a previous observation of attenuation of AKR1C18 promoter activity in mouse luteal cells by rapamycin (21). We have previously shown that rapamycin inhibits decidual COX2 levels in Trp53loxP/loxPPgrCre/+ females and in a cell line with increased mTORC1 activity (14). Moreover, our observations that attenuation of premature decidual senescence by rapamycin together with P4 supplementation prevents preterm birth in Trp53loxP/loxPPgrCre/+ females exposed to a small dose of LPS — without any observable effects on fetal viability and growth — suggest that targeting decidual senescence and ovarian luteolysis is a potential therapy for preventing preterm birth in the context of genetic predisposition and infection/inflammation.
The therapeutic approach using rapamycin and P4 seems more desirable, since floxed mice given this therapy along with LPS did not appreciably show adverse effects on pregnancy outcome. A combinatory treatment with rapamycin and P4 with celecoxib also rescued preterm birth in Trp53loxP/loxPPgrCre/+ females, but this regimen produced adverse effects on pregnancy outcome in floxed littermates. The reasons for the rescue of preterm birth with normal fetal viability and health in Trp53loxP/loxPPgrCre/+ mice versus widespread fetal death and resorption in floxed mice is not clearly understood at this time, although it is possible that under normal pregnancy conditions, drug-drug interactions between celecoxib and rapamycin may adversely affect fetal survival. In fact, co-administration of rapamycin and NSAIDs is not clinically recommended (41). It is also possible that other PGs inhibited by this combination may be harmful to the overall health of normal pregnancy. Notably, celecoxib or rapamycin given alone in Trp53loxP/loxPPgrCre/+ females completely rescued spontaneous preterm birth and had no apparent effects on fetal viability or growth in deleted and floxed females (13, 14).
With regard to the doses and schedule of rapamycin and P4 treatments, rapamycin is normally given at a loading dose of 6 mg followed by a daily oral maintenance dose of 2 mg in transplant patients (42). However, the dosage depends on the response of the patient, and the daily maintenance dose can be up to the recommended limit of 40 mg (43). In our mouse studies, we used only 3 intermittent doses of 0.25 mg/kg BW rapamycin; however, it is difficult to directly compare the doses in mice and in humans due to differential metabolic and clearance rates of drugs.
Regarding the dose of P4, the use of 2 mg P4 in mouse studies is common and widespread in implantation and pregnancy maintenance (12, 44, 45). Besides, we used only 2 doses of P4 on day 16 of pregnancy, which did not result in any adverse effects in floxed wild-type mice, delivering a full complement of healthy pups. In humans, there is a large range of P4 doses given via different routes and for variable lengths of treatment, and the American College of Obstetrics and Gynecology has not yet identified an appropriate dose, route, or formulation for P4 supplementation for the prevention of preterm birth (46). Two major clinical studies have shown that P4 supplementation can reduce the incidence of preterm delivery in select patient populations. One study used 250 mg 17α-hydroxyprogesterone caproate injections (i.m.) weekly from 16 to 20 weeks gestation until 37 weeks or delivery (47), while another study used 200 mg vaginal P4 each night from 24 to 34 weeks gestation until 34 weeks (48). These long-term studies apparently did not show adverse effects on the mother and babies. Again, the doses used in these studies cannot directly be compared with mouse studies due to differential metabolic and clearance rates between the two species. Nonetheless, we believe that the dose of P4 we have used is an appropriate treatment for our present mouse studies.
In this mouse model, we address two known effectors of preterm birth and its rescue. However, the multifactorial aspects of human preterm delivery must be recognized and further studied. Extension of parturition timing but with poor neonatal outcome in LPS-treated Trp53loxP/loxPPgrCre/+ females by P4 treatment alone suggests that it can address the ovarian insufficiency of P4 secretion but cannot overcome the adverse effects of premature decidual senescence. Indeed, there is evidence that P4 supplementation in humans can prevent preterm delivery only in certain patient populations with specific risk factors (47–51).
The role of p53 in pregnancy maintenance in relation to P4 levels and its responsiveness remain to be ascertained. There is evidence that certain TRP53 polymorphisms in women correlate with recurrent pregnancy failure (52); however, this issue remains unsettled (53). TRP53 polymorphisms have also been associated with aging and lifespan in humans (36, 54). Our recent proteomics study showed that deciduae in Trp53loxP/loxPPgrCre/+ females manifest an increased signature for oxidative stress, with downregulation of many antioxidant enzymes, including PRDX6 (24). PRDX6 plays a role during pregnancy in mice with deletion of Fkbp52, an immunophilin co-chaperone for nuclear PR, which show reduced uterine P4 responsiveness (55–57). Therefore, it is possible that oxidative stress makes Trp53loxP/loxPPgrCre/+ females more sensitive to preterm birth, since it is considered a contributing factor (1–3, 8). In-depth studies will be required to assess the definitive role of p53 at various stages of pregnancy.
Our results are clinically relevant because some aspects of the molecular signature observed in mouse studies are consistent with those observed in deciduae of patients undergoing preterm birth. As presented here, it is remarkable that decidual senescence indicated by SA-β-gal and γH2AX staining (58, 59), along with higher mTORC1 and COX2 signaling, are also characteristics of human preterm deciduae. Interestingly, this signature was observed in deciduae irrespective of the etiology of preterm birth, ranging from unknown to diagnosed infection (e.g., chorioamnionitis). These results suggest that disparate signaling pathways converge toward mTORC1-induced decidual senescence and COX2 signaling. However, these studies must be repeated with a larger cohort of patients undergoing preterm birth. Nonetheless, the finding that P4 and/or rapamycin inhibited the inflammatory cytokine release from cultured human term decidual cells in response to LPS suggests that maintaining decidual health will help to prevent preterm birth. It is interesting that TLR4 is expressed in human decidual cells free of leukocytes, suggesting a direct effect of TLR4-mediated effects in the decidua in addition to the effects exerted by immune cells. Whether results from cultured decidual cells correctly reflect the effects of inflammation/infection in vivo remains to be determined.
The placenta is a major source of P4 in human pregnancy after 10 weeks of gestation, as opposed to the situation in rodents, in which ovaries are the major source of P4 throughout the course of pregnancy (60). Although a decrease in P4 levels in rodent models of preterm birth is well established, peripheral P4 levels in women undergoing term and preterm delivery needs to be carefully assessed. A recent report from the United Kingdom shows decreases in salivary P4 levels in women undergoing preterm birth before 34 weeks of gestation; this study suggested that P4 levels are different in early preterm and late preterm birth (61). However, an earlier, U.S. study failed to observe such decline in salivary P4 levels (62). Therefore, P4 levels during human pregnancy in the context of the etiology of preterm birth and parturition timing remain unsettled. A recent report shows that microRNA-200a via STAT5b increases local metabolism of P4 by increasing the expression of AKR1C1 in immortalized human myometrial cells in culture (63). Another report shows AKR1C1 expression in human deciduae (64). Our results showing increased AKR1C1 expression levels in human term decidual cells in culture exposed to LPS, which could be attenuated by rapamycin or P4 treatment, suggest that decidua is also a site for P4 metabolism. It is interesting that the decidual PTGS2 levels are downregulated by rapamycin, which is consistent with our previous and present findings (14). Collectively, human studies showing different aspects of P4 signaling in parturition timing and multiple sites regulating P4 levels indicate that further investigation is warranted.
P4 executes its functions via two PR isoforms, PR-A and PR-B (65, 66). Analysis of promoter activity in cell culture systems suggests that while PR-A functions as a repressor, PR-B serves to increase P4 signaling (67). Notably, the placenta does not express PR. Therefore, P4 should exert its effects via decidual or myometrial PR; which site of P4 signaling is more important in parturition remains to be ascertained. Functional withdrawal of P4 signaling in the myometrium has been proposed to trigger labor in humans (67). There could be several reasons for withdrawal: reduced P4 levels, local metabolism of P4 in the myometrium and/or decidua, an altered ratio of PR isoforms (PR-A/PR-B), or reduced transactivation or heightened transrepression due to recruitment of coactivators or corepressors (68). There is also evidence that inflammation via NF-κB can reduce P4 effectiveness and PGF2α increases PR-A expression without affecting PR-B expression (69, 70). In addition, several studies reported that human labor is associated with reduced decidual expression of PR (71–73). Taken together, the evidence indicates that P4 signaling in the context of myometrial contractility in human parturition requires further investigation.
Chronological aging is a contributing factor to cellular senescence (74). Therefore, it is possible that uterine senescence due to maternal aging compounded by environmental stressors, such as infection/inflammation, can increase the risk of preterm birth. Epidemiologic evidence suggests that advanced maternal age is associated with human preterm birth (75–77). Furthermore, women of advanced maternal age undergoing ART procedures show higher incidence of preterm birth, even when receiving oocytes from young donors (78, 79), suggesting that uterine factors can contribute to this disorder.
Although gene-environment interactions are assumed to be major contributors to preterm birth, this concept has not been experimentally interrogated. Our studies in mice provide evidence that when a genetic predisposition is superimposed by mild inflammation, the rate of preterm birth is profoundly exaggerated. More importantly, success in reversing preterm birth in our mouse model by rapamycin and P4 provides clues to the cooperative contributions of at least two sites of action (decidua and ovary) toward preterm birth. Our present findings in mouse and human studies point toward decidual senescence as a contributor to preterm birth, a concept not previously entertained. These findings provide new insights and should encourage further investigation in the field. Future studies integrating findings from multiple models of preterm delivery will help to define the mechanism behind parturition timing and allow for the design of strategies to prevent preterm birth.
Trp53loxP/loxPPgrCre/+ mice were generated as described previously (13). Briefly, Trp53loxP/loxP mice (FVB/129) were crossed with PgrCre/+ mice (C57BL6/129) to generate mice with uterine deletion of Trp53. Trp53loxP/loxP mice were obtained from the Mouse Models of Human Cancers Consortium, while PgrCre/+ mice were initially provided by J.B. Lydon and F.J. DeMayo (Baylor College of Medicine, Houston, Texas, USA). For experiments, littermate Trp53loxP/loxPPgr+/+ and Trp53loxP/loxPPgrCre/+ mice were used. Mice were provided with autoclaved rodent LabDiet 5010 (Purina) and UV light–sterilized RO/DI constant circulation water ad libitum and were housed under a constant 12-hour light/12-hour dark cycle.
Parturition events were monitored from day 16 through day 21 by observing mice daily, morning (0600h–0700h), noon (1200h), and evening (1800h–2000h). Birth timing was defined by the observation of the first born pup. Preterm birth was defined as birth occurring earlier than day 19 of pregnancy, with the day the vaginal plug was found designated day 1 of pregnancy. Dystocia was defined as difficult delivery lasting more than 12 hours. Resorption sites and placental scars were identified in dams showing preterm or difficult deliveries by examining the uterus after delivery. The number of pups/masses delivered were compared with the number resorption sites and placental scars identified.
Ultrapure TLR4-specific LPS (10, 37, 50, or 75 μg/mouse, i.p.; Invivogen) was administered on day 16 of pregnancy at 1200h. The selective COX2 inhibitor celecoxib was suspended in 5% PEG400 and 5% Tween-80 dissolved in water by constant stirring and was given by oral gavage as indicated (10 mg/kg BW/dose). The mTORC1 inhibitor rapamycin (0.25 mg/kg BW/d) was suspended in the same vehicle and given as a single oral gavage as indicated. The control group received vehicle alone. Progesterone was dissolved in sesame oil and administered subcutaneously (2 mg/0.1 ml/dose). Treatment schedules of various combinations of drugs are given in Supplemental Figure 4.
Implantation sites from which fetuses and placentae had been removed were collected on day 16 of pregnancy. These tissues were flash frozen and stored at –80°C until used for extractions. Methanolic extracts of tissues were partially purified using C18 solid-phase extraction columns (Agilent), and PGs were quantified by HPLC–tandem mass spectrometry as previously described (13).
In situ hybridization was performed as described (13). Whole implantation sites were collected and flash frozen. Frozen tissue sections (12 μm) were mounted onto baked poly-l-lysine–coated slides, fixed in cold 4% paraformaldehyde, acetylated, and hybridized at 45°C for 4 hours in formamide hybridization buffer containing 35S-labeled Plr3c1 (a gift from Michael Soares, University of Kansas Medical Center, Kansas City, Kansas, USA) and Prlr (long isoform) cRNA probes. RNase A–resistant hybrids were detected by autoradiography after 3- to 10-day exposure by using Kodak NTB-2 liquid emulsion. To compare mRNA localization in Trp53loxP/loxPPgr+/+ and Trp53loxP/loxPPgrCre/+ tissues, we placed sections of tissues of both genotypes under similar experimental conditions onto the same slide and processed them for hybridization.
Immunostaining was performed in formalin-fixed, paraffin-embedded sections using specific antibodies to 20αHSD (a gift from Geula Gibori, University of Illinois at Chicago, Chicago, Illinois, USA), COX2 (mouse, laboratory-generated; human, Santa Cruz Biotechnology Inc.), pS6 (Cell Signaling Technology), CDX2 (BioGenex), γH2AX (Millipore), vimentin (Dako), pan cytokeratin (Dako), and CD45 (Dako) as described previously (13, 14). Tissue sections from Trp53loxP/loxPPgr+/+ and Trp53loxP/loxPPgrCre/+ females on each day of pregnancy were processed onto the same slide.
Staining of SA-β-gal activity was performed as described previously (13, 14). In brief, frozen sections were fixed in 0.5% glutaraldehyde in PBS and stained for 6 hours in PBS (human tissues, pH 6.0; mouse tissues, pH 5.5) containing 1 mM MgCl2, 1 mg/ml X-gal, and 5 mM each of potassium ferricyanide and potassium ferrocyanide. Sections were counterstained with eosin.
The images of SA-β-gal staining and immunostaining were analyzed using inForm Image analysis software (PerkinElmer), which can detect the average signal intensity in the scanned area.
RNA was prepared from homogenized tissues using TRIzol reagent (Invitrogen). RNA extraction was performed as described previously (13, 14). Quantitative PCR (qPCR) was performed using StepOnePlus Real-Time PCR System (Applied Biosciences). PCR was performed using the following primers: 5′-CTCTGAAGCCAGGGAATGAG-3′ and 5′-ATGGCATTCTACCTGGTTGC-3′ for mouse Akr1c18 (encoding 20αHSD) (product size, 221 bp); 5′-TGTGCCGCAGCATTAAGTG-3′ and 5′-GGCATCTCACCCTCCACAAC-3′ for mouse Socs1 (product size, 125 bp); 5′-TGCTGGCCAAAGAAATAACCA-3′ and 5′-GGTCACCCCTTGCCACTCT-3′ for mouse Socs3 (product size, 88 bp); 5′-TCCATGACAACTTTGGCATTG-3′ and 5′-CAGTCTTCTGGGTGGCAGTGA-3′ for mouse Gapdh (product size, 72 bp); 5′-GATTGCCCGACTCCCTTGG-3′ and 5′-GTCTAGCCAGAGTTTCACCGT-3′ for human PTGS2 (encoding COX2) (product size, 250 bp); 5′-TGTGCGATATTTGACCCTTGA-3′ and 5′-TGCTGTAGCTTGCTGAAATCAC-3′ for human AKR1C1 (product size, 204 bp); 5′-CAACAAAGGTGGGAATGCTT-3′ and 5′-TGCCATTGAAAGCAACTCTG-3′ for human TLR4 (product size, 317 bp); and 5′-CACACTGTGCCCATCTACGA-3′ and 5′-CTCCTTAATGTCACGCACGA-3′ for human ACTB (product size, 162 bp). Gapdh and ACTB served as housekeeping genes for mouse tissues and human cells, respectively.
Mouse blood samples were collected on day 16 of pregnancy at the prescribed time after treatments. Serum levels of P4 were measured by EIA kits (Cayman Chemical).
Term and preterm placentae were obtained from women with singleton vaginal term or preterm delivery. Placental samples from patients with hydramnios or newborns with any birth or chromosomal abnormalities were not included in the preterm study, although placental samples from patients with chorioamnionitis or preterm premature rupture of the membranes (PPROM) were included in the study. Detailed patient information for immunohistochemistry and SA-β-gal staining is shown in Supplemental Table 4. Placental samples from non-laboring patients undergoing cesarean section for preeclampsia, fetal distress/anomaly, placenta abruption, or previa were included as controls for the study (Supplemental Table 5). For decidual cell culture, term placentae were obtained from women undergoing elective cesarean section. Endometria for endometrial stromal cell culture were obtained from women undergoing hysterectomy due to benign gynecological diseases; none had hormone treatment 3 months prior to surgery. None of the women undergoing term vaginal delivery or term cesarean section showed any clinical or pathological signs of preterm delivery, infection, or other maternal or placental diseases. All women with singleton preterm vaginal delivery did not show clinical or pathological signs of other maternal or placental diseases apart from preterm delivery. Newborns did not have any apparent birth or chromosomal abnormalities.
Human term decidual cells were isolated and cultured according to previously described protocols with minor modifications (32, 80). Briefly, term deciduae from women with cesarean section were scraped from the maternal surface of the chorion, minced, and digested in Ham’s F-10 media containing 25 mg/ml collagenase and 6.25 U/ml DNase in a shaking water bath at 37°C for 30 minutes. Digested samples were passed through a 23-gauge needle to dissociate remaining cell clusters, centrifuged at 250 g for 5 minutes, and washed in the culture media. The cell pellet was resuspended in 20% Percoll, layered on a discontinuous (60%:50%:40%) Percoll gradient, and centrifuged at 540 g for 20 minutes. The upper cell layer was collected, washed, centrifuged, resuspended in 40% Percoll, layered on a discontinuous (55%:50%:45%) Percoll gradient, and centrifuged at 540 rpm for 20 min. The upper cell layer was washed and resuspended in DMEM/F-12 media containing 5% fetal bovine serum, 100 IU/ml penicillin, 0.1 mg/ml streptomycin, and 0.25 μg/ml amphotericin B and plated onto 100-mm dishes. Cells were cultured at 37°C in a humidified 5% CO2 chamber. When the cells attained confluence, they were dissociated with 0.25% trypsin-EDTA, harvested by centrifugation at 250 g for 5 minutes, and replated onto 100-mm dishes. Cells were passaged at least 3 times and plated in 12-well plates at 2 × 105 cells/well for experiments. The complete media were then removed and replaced with serum-free media containing antibiotics, and cells were cultured for an additional 12–24 hours before stimulation. Purity of the decidual cell population was determined by immunocytochemical staining of vimentin, pan-cytokeratin, and CD45, which served as markers for stromal cells, epithelial cells, and leukocytes, respectively. The purity of the passaged decidual cells was greater than 99%, as judged by positive staining for vimentin and negative staining for cytokeratin and CD45 (Supplemental Figure 10A).
To evaluate the effects of LPS on expression of PTGS2 and AKR1C1 in term decidual cells, wells were replenished with serum-free media with or without TLR4-specific LPS and cells were incubated for 6 hours (qPCR) or 24 hours (Western blotting). After termination of cultures, total RNA or protein was isolated and used for qPCR or Western blotting, respectively. To evaluate the effects of P4 and rapamycin on LPS-induced levels of PTGS2 and AKR1C1, cells were preincubated with P4 or rapamycin for 24 hours before addition of LPS and cultured for an additional 24 hours. To evaluate the effects of P4 and rapamycin on LPS-induced levels of IL-6 and IL-8, conditioned media were collected after termination of cultures, centrifuged, and stored at –80°C until assay. Concentrations of IL-6 and IL-8 were measured using specific ELISA kits according to the manufacturer’s protocol (R&D Systems). Absorbance was read at 450 nm with a DigiScan Microplate Reader.
Protein extraction and Western blotting were performed as previously described (13, 14). Antibodies to COX2 and actin (Santa Cruz Biotechnology Inc.) were used. Bands were visualized by using an ECL Prime Western blotting detection system (GE Healthcare). Actin served as a loading control.
Statistical analyses were performed using 2-tailed Student’s t test. P values less than 0.05 were considered statistically significant.
All mice used in this investigation were housed in the Cincinnati Children’s Hospital Medical Center Animal Care Facility according to NIH and institutional guidelines for the use of laboratory animals. All protocols of the present study were reviewed and approved by the Cincinnati Children’s Hospital Research Foundation Institutional Animal Care and Use Committee. Collection and processing of human samples were approved by the respective ethics committees at University of Tokyo and Yaizu City Hospital in Tokyo under the approved IRB protocol no. 3456, and all patients provided written informed consent. Tissue sample collections were based on the operating procedures of the University of Tokyo Tissue Procurement Resource, which strips samples of all patient identifiers before procurement (de-identified) and replaces these with new sample identifiers. This study was limited to female subjects because of the nature of the disease studied. Children were not included due to the rarity of preterm birth in the pediatric population.
We thank Tomoyuki Fujii, Yutaka Osuga, Kaori Koga (University of Tokyo, Tokyo, Japan), and Kazutoshi Naritaka (Yaizu City Hospital, Japan) for human sample collections and helpful discussions and Michael J. Soares (Kansas University Medical Center) for providing the Prl3c1 cDNA. This work was supported in part by grants from the NIH (HD12304 and DA06668), the March of Dimes (#21-FY12-127 and #22-FY13-543), and the Bill and Melinda Gates Foundation through the Grand Challenges Explorations Initiative (to S.K. Dey); by PRESTO, a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science, the Takeda Science Foundation, the Kowa Life Science Foundation, and the Yamaguchi Endocrine Research Foundation (to Y. Hirota); and by NIH/National Institute on Drug Abuse grant DA032150 (to H. Bradshaw). J. Cha is a predoctoral National Research Service Award fellow (NIA/NIH F30AG040858) of the University of Cincinnati Medical Scientist Training Program (T32GM063483).
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2013;123(9):4063–4075. doi:10.1172/JCI70098.