The Molecular Mechanisms of Cervical Ripening Differ between Term and Preterm Birth

Premature cervical ripening can occur by more than one mechanism, and premature ripening associated with preterm birth is not simply an acceleration of term ripening.

Abstract

In the current study, the mechanisms of premature cervical ripening in murine models of preterm birth resulting from infection or early progesterone withdrawal were compared with the process of term cervical ripening. Tissue morphology, weight, gene expression, and collagen content along with immune cell populations were evaluated. Premature ripening induced by the progesterone receptor antagonist mifepristone results from an acceleration of processes in place during term ripening as well as partial activation of proinflammatory and immunosuppressive processes observed during postpartum repair. In contrast to term or mifepristone-induced preterm ripening, premature ripening induced in an infection model occurs by a distinct mechanism which is dominated by an influx of neutrophils into the cervix, a robust proinflammatory response and increased expression of prostaglandin-cyclooxygenase-endoperoxide synthase 2, important in prostaglandin biosynthesis. Key findings from this study confirm that cervical ripening can be initiated by more than one mechanism and is not necessarily an acceleration of the physiologic process at term. These results will influence current strategies for identifying specific etiologies of preterm birth and developing subsequent therapies.

Preterm birth (PTB) accounts for 12.7% of all births in the United States (1). Preterm infants suffer morbidities including respiratory distress syndrome, intraventricular hemorrhage, necrotizing enterocolitis, periventricular leukomalacia, and cerebral palsy (2). These outcomes can have a life-long health impact, and costs associated with care of these babies exceed $26 billion per year. Therapies to prevent PTB and development of improved tools for detection are hindered by the fact that PTB has multiple etiologies (2). While the cause of PTB remains elusive in roughly 50% of cases, identified risk factors include smoking, low maternal body-mass index, alcohol consumption, advanced maternal age, stress, genetic factors, previous PTB, threatened abortion, multifetal gestation, cervical insufficiency, cervical shortening and infection.

Irrespective of etiology, PTB results in premature cervical ripening, premature uterine contraction, and delivery of an infant unprepared for extrauterine life. Areas of intense investigation include both the hormonal regulation and the molecular mechanisms involved in initiating uterine contractions and remodeling of the cervix from a closed rigid structure to one that can open sufficiently for passage of a term fetus. With respect to the cervix, the slow progressive phase of remodeling termed cervical softening begins early in pregnancy when progesterone levels are high and estrogen levels are relatively low (3). This early phase then overlaps with the accelerated remodeling phase toward the end of pregnancy termed cervical ripening when progesterone action is blunted and estrogen levels are high. While the mechanism for loss of progesterone function required for the initiation of parturition can vary between species, many of the downstream processes that bring about cervical ripening are well conserved between human and animal models (4). Animal models of PTB resulting from infection or premature progesterone withdrawal have relevance to human cervical biology. Well-studied PTB models include administration of the progesterone receptor (PR) antagonist, mifepristone (RU486) or the endotoxin lipopolysaccharide (LPS) to mice resulting in premature cervical ripening and PTB (5). The first model serves as a hormonal withdrawal or noninfection model, while the latter is a model for infection-induced PTB. Microarray and gene expression studies comparing expression profiles in term cervical ripening vs. infection and noninfection-induced PTB report distinct differences in gene pathways between term and infection-induced cervical ripening (6, 7). Additionally, our laboratory has observed differences in collagen structure during cervical ripening induced by premature progesterone withdrawal compared with normal remodeling (8). These studies have led to the insight that the mechanisms of preterm cervical remodeling can occur via distinct mechanisms from the physiologic process and secondly that activation of immune pathways are sufficient but not necessary for ripening (6, 9, 10).

These novel findings highlight the necessity of understanding the similarities and differences in the process of cervical ripening between term and PTBs with a defined etiology to identify effective therapies in the prevention of PTBs. In the current study, we sought to elucidate the mechanism of ripening in infection and noninfection-mediated mouse PTB models compared with physiologic term ripening through analysis of tissue morphology, weight, gene expression, collagen content, and immune cell populations.

Materials and Methods

Mice

Animals were housed under a 12L:12D photoperiod (lights-on, 0600–1800) at 22 C. Mice used in the present studies were from Black 6/129SvEv and NIH Swiss strains. The Black 6/129SvEv mice were generated and maintained as a breeder colony at the University of Texas Southwestern Medical Center (Dallas, TX). Female mice were housed overnight with males and checked in the morning for vaginal plugs to obtain accurately timed pregnant mice. The day of plug formation was counted as d 0, and birth occurred in the early morning hours of d 19. Samples were collected at midday unless otherwise specified. Cervical samples indicated as d 18.75 were collected in the evening of d 18, generally between 1800 and 2000. The NIH Swiss timed-pregnant were purchased from Harlan Laboratories (Houston, TX). All mice in these studies were 3–6 months old and nulliparous. All studies were conducted on approval by the University of Texas Southwestern Medical Center Institutional Animal Care and Research Advisory Committee.

Tissue collection

Cervices were isolated by dissection at the uterocervical junction (caudal to the uterine bifurcation), and all vaginal tissue was removed. Cervices were weighed with an analytical balance (Mettler Toledo, Columbus, OH) before flash freezing in liquid nitrogen. To estimate tissue hydration, cervices were lyophilized overnight, and dry weight was measured. Water content was determined by [(wet weight − dry weight)/wet weight] and expressed as a percentage. Cervices for histological and immunohistochemical analysis were fixed in 4% paraformaldehyde (Sigma Aldrich, St. Louis, MO) and paraffin embedded. Longitudinal sections were stained with Masson trichrome as described previously (11). Trichrome stains fibril collagen blue, keratin and muscle fibers red, and cytoplasm light red or pink.

Infection-induced PTB model

Labor was induced in gestation d 15 mice as previously described (12). Briefly, the mice were anesthetized with avertin (0.2cc/g body weight). The skin was shaved and subsequently sterilized with iodine and ethanol. Small abdominal incisions were made, and a 150 μg LPS (O55:B5 Sigma, St. Louis, MO) dose in 30 μl sterile water was injected between two gestational sacs in the left uterine horn being careful to avoid intraamniotic injection. The peritoneal fascia was aseptically stitched using sterile surgical silk (Ethicon, Bridgewater, NJ), and the skin was closed with sterile surgical wound clips (Autoclips by Becton Dickinson, Sparks, MD). Animals were placed on a heating pad to maintain body temperature during anesthetic recovery. Buprenorphine (0.1 μg/g) was administered during recovery for pain control. The animals were killed after 6 h for cervical tissue collection. Sterile water alone was injected similarly for the 6-h sham surgical controls. Performance of surgery under sterile conditions as described was found to be critical to prevent global activation of proinflammatory responses in the sham surgery controls. Under the conditions and dose described, PTB occurred 7–8 h after LPS injection but not sham surgery. To ensure initiation of PTB at a similar time point in gestation as the RU486 model, LPS was administered early on gestation d 15 (approx 7–8 am) and cervices collected at midday on d 15.

Noninfection PTB model

RU486 (M 8046, Sigma), a PR antagonist, was used to simulate premature progesterone withdrawal. RU486 was solubilized in ethanol and brought up in glyceryl trioleate (Sigma). A dose of 0.5 mg/200 μl was injected subcutaneously late on the night of gestation d 14 (approximately 2100–2200), and cervical tissue was collected 12 h later on gestation d 15. Fifty microliters of ethanol in 150 μl of glyceryl trioleate was administered subcutaneously as a 12-h vehicle control. The dose used will cause delivery 13–16 h after injection (13).

RNA isolation and quantitative real-time PCR (QRTPCR)

Total RNA was extracted from frozen mouse cervices using RNA Stat 60 (Tel-Test Inc, Friendswood, TX) and was subsequently treated with DNase I (DNA-free; Ambion Inc, Austin, TX) to remove genomic DNA. cDNA synthesis was performed using 2 μg total RNA in a 100-μl volume (TaqMan cDNA Synthesis Kit; Applied Biosystems, Foster City, CA). QRTPCR was performed using SYBR Green and a PRISM7900HT Sequence Detection System (Applied Biosystems). Aliquots (20 ng) of cDNA were used for each quantitative PCR reaction, and each reaction was run in triplicate. Relative gene expression between experimental groups was determined using the ddCt method, as described in User Bulletin No. 2 (Applied Biosystems). The housekeeping gene cyclophilin B was used as the internal normalizer. Data were further normalized to the average of the untreated d 15 cervices.

Flow cytometry

Cervical cells were dispersed and stained using methods described previously (10). Peripheral blood was collected by submandibular bleed. One hundred microliters of blood were used for each animal tested, and the cells were stained and fixed as described previously (10). The Fc receptors were blocked using mAb 24G2 (BD Biosciences, San Jose, CA). Fluorescently conjugated mAbs used included neutrophil (Neu) 7/4-PE and Neu 7/4-biotin (Serotec, Raliegh, NC); F4/80-allophycocyanin, CD45-PE-Cy7, CD11b-eFluor 450, and Gr-1 (Ly6C/Ly6G)-allophycocyanin-Cy7 (eBiosciences, San Diego, CA), and Siglec F-PE (BD Biosciences). Stained cells were run on a BD Biosciences LSRII flow cytometer using BD FACSDiva (BD Biosciences) software and analyzed with FloJo 7.1 analysis software (Tree Star Inc., Ashland, OR).

Myeloperoxidase (MPO) assay

Tissue-associated MPO activity was determined by modification of the methods of Krawisz et al. as described previously (9, 14). For each experiment, untreated nonpregnant (NP) metestrus, gestation d 15, late d 18, and 2–4 h postpartum (2–4 PP), as well as d 15 treated with LPS, RU486, or vehicle controls as described above were tested. MPO activity was normalized to mg total protein and expressed relative to activity of the NP cervix in each experiment. The data are reported as a percentage activity compared with that observed in NP cervix.

Immunohistochemistry

MMP8 staining was carried out using antimouse MMP8 rabbit monoclonal antibody (Epitomics, Burlingame, CA). Longitudinal sections were deparaffinized and hydrated in xylene and a series of graded ethanol solutions followed by three PBS washes. Slides were subjected to citrate buffer (10 mm Citric Acid, 0.05% Tween20) antigen retrieval for 30 min at 95 C. Endogenous peroxidases were quenched using 0.5% H₂O₂ in methanol for 10 min at room temperature. Nonspecific binding was blocked with 1.5% normal donkey serum (Jackson Laboratories, Westgrove, PA). Slides were incubated with anti-MMP8 (1:250 in PBS) for 1 h at room temperature. Biotinylated donkey antirabbit (Jackson Laboratories) and peroxidase conjugated avidin-biotin complex (Vector Laboratories, Burlingame, CA) were applied in sequence followed by 5-min incubation with DAB (Invitrogen) at room temperature. Tissues were counterstained in hematoxylin (Surgipath, Richmond, IL) for 10 sec.

Collagen content

Cervices were lyophilized, and dry weight was measured. Dried cervices were digested with 0.5 mg/ml proteinase K (Roche) in 100 mm ammonium acetate pH 7.0 for 4 h at 60 C. An aliquot of the digested product equivalent to 1 mg of tissue dry weight was hydrolyzed in 6 m HCL at 110 C overnight. The hydrolysate was dried at 110 C and dissolved in ddH₂O. The hydroxyproline assay was carried out as previously described (15, 16). The amount of total collagen was determined by using a mass ratio of collagen to hydroxyproline of 7.46:1 (17). The collagen weight was normalized to tissue dry weight.

Statistics

Statistics for QRTPCR were performed with logarithmic transformation of relative gene expression with one-way ANOVA and adjusted with Dunnett's pairwise comparisons to gestation d 15. Further pairwise multiple comparisons were done with Student-Newman-Keuls testing. Wet weight and flow cytometry cervix data were analyzed using Kruskal-Wallis one-way ANOVA on Ranks, and pairwise comparisons were performed using Dunn's Method. Peripheral blood, water content, and collagen content data were analyzed with one-way ANOVA followed by Dunnett's pairwise comparison test. The choice as to whether ANOVA or Kruskal-Wallis test was used is based upon the results of the Shapiro-Wilk test for statistical normality. The rank methods, Kruskal-Wallis test, were used when the data and the transformed data both reject the hypothesis of normality.

Results

Histology and cervical wet weight

Morphological assessment of Masson's Trichrome stained sections was performed in the two PTB models and compared with morphological features before (d 15), during (d 18.75), and after (PP) cervical ripening. Compared with gestation d 15, the intensity of the collagen fibers, stained blue, appeared reduced during physiologic cervical ripening on gestation d 18 (Fig. 1, A vs. B). In addition, fiber collagen packing was disorganized with greater spacing between fibers. This is proposed to be due to a change in extracellular matrix (ECM) composition and increased tissue hydration (4). Shortly PP (Fig. 1C) the blue staining of collagen fibers was weak, and spacing between fibers appeared further increased. In addition, tissue wet weight and hydration was higher than before birth (Fig. 1F). In the LPS-treated mice (Fig. 1D), the collagen matrix appeared similar to term ripening with increased spacing between collagen fibers though perhaps to a lesser extent than observed on d 18. Matrix disorganization was more pronounced in RU486-treated cervices (Fig. 1E) compared with term cervical ripening and was most similar to the cervix shortly PP (Fig. 1C). Most notable were the increased number of spaces and reduction of blue staining. LPS sham and RU486 vehicle controls were morphologically indistinct from untreated gestation d 15 (data not shown).

Fig. 1.

Histomorphological and wet weight assessment of cervical ripening in PTB models. Sections were stained with Masson's trichrome. A, Fibrillar collagen (blue) in the untreated gestation d 15 was densely packed. B, Spaces between collagen increased while intensity of the blue staining decreased by late on d 18. C, Two to 4 h PP, spacing between collagen fibers was even further increased. D, LPS-treated cervix appeared similar to d 18.75 however, (E) the RU486-treated cervix was morphologically similar to PP cervix. Ep, epithelium; S, cervical stroma. Bar, 100 μm (A). F, Tissue wet weight was determined in gestation d 15, d 15-LPS–treated (LPS), d 15-RU486–treated (RU486), late d 18 (d18.75) and PP cervices. Data represent an average of 13–27 cervices ± sem. *, P ≤ 0.05 when compared with untreated d 15.

While cervical wet weight was increased during term cervical ripening, increases in tissue wet weight differed between the two preterm models (Fig. 1F). Compared with gestation d 15, tissue wet weight of RU486-treated cervices increased to levels similar to term cervical ripening late on d 18 (d18.75). In contrast, the cervices from LPS-treated mice failed to increase in size with similar weights to the untreated d 15 mice. Relative to d 15, water content increased significantly by 6% in the RU486-treated cervices but not in the LPS treated cervices or LPS/RU486 sham controls (Supplemental Fig. 1 published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org/). Taken together, both tissue weights and morphological assessment indicate increased matrix disorganization and edema in RU486-treated cervices compared with LPS-treated.

Gene expression by QRTPCR

Genes previously identified to be induced during term cervical ripening in mice were evaluated by QRTPCR (Fig. 2) (13, 18, 19). These included hyaluronan (HA) synthase 2 (Has2), connexin 26 (Gjb2), oxytocin receptor (Oxtr), and steroid 5 α reductase type I (Srd5a1). While Has2 gene expression increased significantly from gestation day 15 to late day 18 (d18.75) during normal gestation, the expression did not increase during premature ripening induced by LPS- or RU486-induced PTB. HAS2 synthesizes HA, a hydrophilic glycosaminoglycan that facilitates collagen matrix disorganization. The expression of Gjb2 and Oxtr increased significantly, and there was a trend for increased Srd5a1 expression with RU486 treatment. In the LPS-treated mice, only Oxtr expression increased significantly above the d 15 sham and untreated control but not to a level that was similar to d 18.75 term cervical ripening.

Fig. 2.

Genes up-regulated during term cervical ripening are similarly induced in premature ripening by RU486 but not LPS treatment. QRTPCR was performed to compare gene expression in PTB models with untreated d 15 and d 18.75 cervices. Gap junction β 2 (Gjb2, connexin 26), oxytocin receptor (Oxtr), steroid 5-α reductase type 1 (Srd5a1), and HA synthase 2 (Has2) expression were evaluated. Data represent an average ± sem. The study groups are d 15, RU486 vehicle control and treatment (−, +), LPS sham surgery and treatment (Sham, +), and d18.75. The n for each group include: d 15 = 4, RU486 control = 3–4, RU486 = 5–6, LPS Sham = 4, LPS = 4–6, and d 18.75 = 5. An asterisk indicates significance (P ≤ 0.05) from d 15 untreated cervices. A + indicates significance from d 18.75 (P ≤ 0.05).

While immune modulating genes are not induced in the cervix until onset of term labor, many of these genes are reported to be induced in the cervix before infection-mediated PTB (6, 9, 10, 20, 21). Immune modulating genes in term labor or PP repair include both proinflammatory genes such as interleukin 6 (Il6), tumor necrosis factor-α (Tnf), interleukin 1 α (Il1a), and chemokine (C-X-C motif) ligand 2 (Cxcl2 formerly Mip2) and those involved in tissue repair (arginase 1, transforming growth factor β1) and immunosuppression [interleukin 1 receptor antagonist (Il1rn), interleukin 13] (10). The expression of genes associated with classically and alternatively activated immune cells were evaluated in the two PTB models and compared with gene expression before and during physiologic cervical ripening on d 15 and d 18, respectively (Fig. 3). The neutrophil chemoattractant Cxcl2 and Tnf had increased gene expression after LPS administration in contrast to the RU486 preterm, d 15 controls and untreated term cervix. While RU486 and the vehicle control were increased, they were found to be similar to each other (P > 0.05) in contrast to LPS compared with its sham (P ≤ 0.05). Other proinflammatory markers such as Il6 and Il1a were significantly increased in both preterm models compared with untreated gestation d 15 and vehicle/sham controls though the level of induction with LPS was many-fold increased compared with RU486. Interestingly, genes expressed by alternatively activated M2-macrophages such as chitinase 3-like 3 (Chi3l3 formerly Ym1) and Il1rn were up-regulated in both the noninfection RU486 preterm model and the infection preterm model. Chi3l3 is a marker of M2 macrophages though its function is unclear while Il1rn is a decoy receptor for Il1a that blunts Il1a function (22, 23). Chi3l3 increased to levels that were significantly higher than term in both preterm models, while the increase in Il1rn was more pronounced in the infection model. Prostaglandins are synthesized by prostaglandin-cyclooxygenase-endoperoxide synthase 1 and 2 (Ptgs1 and 2 formerly Cox1 and 2) and stimulate proinflammatory responses (24). A previous study reports expression of Ptgs1 in the term cervix, while Ptgs2 is low to absent during term cervical ripening (19). In contrast, premature cervical ripening induced by RU486 resulted in a significant increase in Ptgs1 expression. Ptgs2 expression was low but significantly increased compared with d 15 controls. Ptgs1 expression was low in the infection model yet Ptgs2 expression was markedly induced.

Fig. 3.

Immune modulating genes have differing expression patterns in term and preterm ripening. Comparison of relative mRNA expression with QRTPCR from d 15 untreated, RU486- or LPS-treated, and respective controls and late d 18 (d 18.75) cervices is shown. Proinflammatory markers Cxcl2, Il6, Tnf, Il1a, as well as genes expressed by alternatively activated M2 macrophages, Chi3l3 and Il1rn, were measured. Ptgs1 was up-regulated in the RU486 model while Ptgs2 was highly up-regulated in the LPS model. Data represent an average ± sem for 5–6 cervices per group. The study groups are d 15, RU486 vehicle control and treatment (-, +), LPS sham surgery and treatment (Sham, +), and d 18.75. An asterisk indicates significance (P ≤ 0.05) from d 15 untreated cervices. A + indicates significance from d 18.75 (P ≤ 0.05).

Matrix metalloproteinases (MMPs) and ADAMTS (a distintegrin-like and metalloproteinase with thrombospondin type 1 motif) are two groups of proteases that clip or breakdown ECM molecules such as collagen and proteoglycans, respectively. The mRNA expression of Adamts1, described to be up-regulated during cervical ripening (25), was evaluated along with Adamts4, Mmp2, and Mmp8 in cervices from preterm models (Fig. 4). Adamts1 and Adamts4 mRNA expression is increased 2- to 3-fold during physiologic cervical ripening and were increased 3- to 4-fold with RU486-induced premature ripening. Adamts1 and Adamts4 expression was increased approximately 3-fold and 7-fold, respectively, in LPS-treated cervices. While Mmp2 expression was abundant, levels were unchanged between normal d 15, d 18, and both preterm models. We have previously observed Mmp2 and Mmp9 transcripts in the cervix with little change in expression in term cervical ripening (data not shown). The lack of induction of Mmp2 in the LPS-treated mice in this study differs from previous reports (6). The expression of Mmp8 expressed primarily by neutrophils was significantly induced in both PTB models, but expression was absent in term ripening. The expression of Mmp8 was much greater in the LPS group compared with RU486.

Fig. 4.

Adamts1, Adamts4, and Mmp8 are up-regulated during premature cervical ripening induced by RU486 and LPS. Adamts1 and 4 show increased expression in both preterm models similar to d 18 term cervical ripening. Mmp8, expressed by neutrophils, was significantly increased in both the LPS and RU486 models. Mmp2 expression was unchanged between all groups. Data represent an average ± sem of 5–6 cervices per group. The study groups are d 15, RU486 vehicle control and treatment (−, +), LPS sham surgery and treatment (Sham, +), and d 18.75. An asterisk indicates significance (P ≤ 0.05) from d 15 untreated cervices. A + indicates significance from d 18.75 (P ≤ 0.05).

Evaluation of myeloid-derived immune cells in cervix of PTB models

The differential expression of immunomodulating genes in the PTB models may suggest a different recruitment or activation of inflammatory cells in premature cervical ripening. Flow cytometry was used to compare the inflammatory cell populations in the PTB models vs. normal cervical ripening (Fig. 5). Dot plots in panel A and B represent the pan leukocyte marker CD45-positive populations. As described previously, neutrophils are defined as Neutrophil (Neu) 7/4⁺, GR1²⁺; monocytes as Neu 7/4²⁺, GR1^{intermediate/}; eosinophils as Neu 7/4^low, F4/80^low, SiglecF⁺; and macrophages as Neu 7/4^low, F4/80²⁺ (10). Similar to term cervical ripening, both tissue monocytes (Fig. 5, A and B) and eosinophils (data not shown) significantly increased during premature cervical ripening in mice treated with RU486. In contrast, tissue monocytes and eosinophils were not increased during premature cervical ripening induced by LPS treatment. Neutrophils, however, significantly increased in the cervix of the LPS but not the RU486-treated mice. No significant change in macrophages was observed with any of the treatments.

Fig. 5.

Recruitment of specific myeloid cell populations in the cervix is dependent on the etiology of the PTB. Panels A and B are representative dot plots of the white blood cell (CD45+) populations showing how monocytes (Mo) and neutrophils (Neus) (A) and macrophages (MØ) and eosinophils (Eos) (B) were identified. C, Monocytes and eosinophils significantly increased when the mice were treated with RU486 but not with LPS. Conversely, the neutrophil populations significantly increased in LPS-treated mice but not RU486. Data represent mean ± sem of 5–7. *, P ≤ 0.05 compared with d 15 untreated.

Peripheral blood from the treated and untreated mice was also stained with the above antibody panel (Supplemental Fig. 2). A significant increase in neutrophils and decrease in eosinophils were measured in the peripheral blood of LPS sham surgery controls but not with the LPS or RU486 treatment. Monocyte numbers in peripheral blood did not change with either the LPS or RU486 treatments. These data suggest that the abdominal surgery results in an increase in circulating neutrophils despite careful aseptic technique during surgery and absence of the LPS injection. However, the granulocytes were not increased in the tissue of the sham surgeries as they did with the LPS injection. Thus, the overall neutrophil numbers in the peripheral blood of LPS-injected animals remained low, indicating that changes in immune cell populations within the cervical microenvironment were not reflected in peripheral blood.

Characterization of neutrophil phenotype

MPO is secreted by activated neutrophils, and MPO activity can be used to quantify the number of neutrophils in tissue (26). In previous studies we have reported the presence of neutrophils in mice during term cervical ripening yet little to no MPO activity until 2–4 h after birth (9). In the current study, MPO activity was measured to assess the phenotype of neutrophils whose numbers increase during premature cervical ripening with LPS treatment. MPO activity was very low, and no significant changes were observed between untreated d 15 cervix and both PTB models (Fig. 6). A significant increase in MPO activity was seen in 2–4 h PP cervices and NP cervix compared with gestation d 15 cervix as reported previously (9). Immunohistochemical localization of MMP8 was performed to confirm its expression in neutrophils and validate QRTPCR findings. The few MMP8-positive neutrophils present in the d 15 cervix were primarily localized to the epithelia or subepithelial region of the tissue (Fig. 7A). Few to no MMP8-positive cells were detected during term cervical ripening (Fig. 7B) however, by 4 h PP (Fig. 7C), numerous MMP8-positive cells were distributed throughout the cervical stroma. Consistent with increased Mmp8 mRNA expression, there were an increased number of MMP8-expressing cells in the LPS treated cervix (Fig. 7D). Similar to d 15, most of the stained cells were in the subepithelial region. While MMP8-positive cells were present in the RU486-treated cervix, staining was primarily localized to cervical mucus in the cervical lumen as well as to neutrophils still present in blood vessels (Fig. 7E). A few positive cells were detected in the cervical stroma.

Fig. 6.

Similar to term ripening, neutrophil myeloperoxidae (MPO) activity does not increase with premature ripening induced by RU486 or LPS. MPO assays were performed on cervical tissue extracts from pregnant d 15 (untreated, LPS, RU486), late d 18 (d 18.75), along with 2–4 h PP, and NP animals in metestrus. Three to five animals were tested for each time point and treatment. Results were normalized to total protein of each cervix. MPO activity for each experiment was normalized to the NP cervix of that experiment. Data are reported as a percentage activity compared with NP cervix. Data represent mean ± sem. *, P < 0.05 compared with d15 untreated cervices.

Fig. 7.

MMP8-expressing cells are increased in cervices of LPS-treated mice yet collagen content is unaffected. Cervical sections from three animals per group or time point were stained with monoclonal antibody against mouse MMP8 and representative sections are indicated in A–E. Positive cells stained brown (indicated with an arrow). A, MMP8-positive cells were observed in untreated d 15 cervices and appeared to be localized in the epithelium/subepithelium. B, Few to no positive cells were observed in the late d 18 cervices. C, MMP8 expression was high in the PP cervices with the positive cells evenly spread throughout the stroma. D, With LPS treatment, numerous MMP8-positive cells were detected primarily in the subepithelial area and to a lesser extent in the stroma. E, The mucus secretions in the cervical lumen along with some cells in blood vessels expressed MMP8 with RU486 treatment though few to no positive cells in the cervical stroma. Ep, epithelium; S, cervical stroma. Bar, 100 μm (A). F, Collagen content was determined by hydroxyproline assay. Day 15 untreated (d 15) are compared with RU486, LPS, and normal term cervical ripening d 18 (d 18.75). Data represent an average μg collagen/mg of dry weight of tissue ± sem for 5–10 cervices per group or time point.

Collagen content

The increased mRNA expression of Mmp8, a neutrophil collagenase whose primary target is fibrillar collagen I, may lead to a change in collagen structure or loss of collagen in the cervical ECM. To test whether there is a loss of total collagen, the collagen content was measured via hydroxyproline assay comparing untreated d 15 to LPS- and RU486-treated tissues, as well as d 18 tissues. Collagen content levels did not change significantly with either treatment (Fig. 7F). Vehicle and surgical sham controls were also unchanged (data not shown). These results were similar to our previous observations that collagen content does not change during normal pregnancy from d 15 to d 18 (3).

Discussion

The development of clinical interventions that will reduce the rates of PTB is dependent on a clear understanding of the molecular mechanisms of normal parturition and the variety of causes and mechanisms by which PTB occurs. Previous studies report the potential for different mechanisms in initiation of preterm labor compared with term labor yet the processes are not defined (5, 7, 27). The current study identified similarities and differences between term and preterm cervical ripening in two mouse PTB models supports and builds upon the following findings: 1) cervical ripening can occur by more than one mechanism, and 2) premature ripening associated with PTB is not simply an acceleration of the physiologic term ripening process. These findings suggest the need to identify early markers of PTB risk that may be unique to the primary cause (infection, premature rupture of membranes, cervical insufficiency, etc.), as well as the need to devise therapies that are dependent on the etiology of PTB.

Both models of PTB used in this study have direct applicability to mechanisms of premature cervical remodeling in women. Administration of PR antagonists such as RU486 results in premature cervical ripening in women and animal models (12, 28, 29). Because loss of progesterone function occurs during normal term ripening, this model is predicted to mimic the physiologic process. Infection is a well-established cause of PTB and accounts for roughly 25% of PTBs in women (2). Ascending infection into the reproductive tract can elicit an inflammatory response which results in premature cervical ripening. Administration of intrauterine LPS, a component of the outer membrane of gram negative bacteria, or heat-killed Escherichia coli to mice will initiate an inflammatory response that mimics infection and results in PTB in the absence of a reduction in circulating progesterone (12, 30, 31). Once considered an essential component of the cervical ripening process at term, activation of proinflammatory responses has recently been shown to be sufficient but not required for ripening (6). Additionally, both proinflammatory and immunosuppressive responses are observed to be associated with PP repair of the cervix instead of ripening in mice and women (9, 10, 21, 32).

A sudden loss of PR function upon RU486 administration differs from the more gradual withdrawal of progesterone occurring during normal term birth in mice. This difference appears to result in a combination of premature activation of processes involved in term ripening in the mouse as well as a partial activation of resident neutrophils and macrophages similar to the PP repair phase of cervical remodeling. With respect to this noninfection model, the increased tissue size, changes in myeloid cell populations, and gene expression patterns were similar to term ripening (Figs. 1–5). Other features with RU486 treatment were similar to a cervix shortly PP, for example, morphological characteristics such as increased spacing between collagen bundles and induction of genes expressed by alternatively activated M2 macrophages (e.g., Ym1 and Il1rn), neutrophils (e.g., Mmp8) as well as increased expression of Ptgs1 (Cox1) and Il1a (Ref. 20 and data not shown). Notable dissimilarities in gene expression between the noninfection model and term were the small increases in Ptgs2, Il6, and the lack of induction of Has2 (Fig. 2). While Has2 expression did not increase in either PTB model, preliminary studies suggest compensation by increased expression of another HA synthase, Has1 (manuscript in preparation). The function of HA in matrix disorganization may thus have a similar function in premature ripening, and increased HA may in part account for the increase in cervical wet weight in RU486-treated mice in addition to increased hydration. Other contributing factors that may account for the notable increase in tissue wet weight include increased cell density attributable to increased number of immune cells, or the observed increase in cervical mucus. Further studies are required to identify the combination of processes that account for the dramatic increase in tissue wet weight with RU486 treatment.

Cervical ripening resulting from infection occurs via a molecular process that is distinct from term ripening as well as premature ripening mediated by PR antagonism. Activation of proinflammatory responses that include Il6, Tnf, Il1a, and Cxcl2 predominate in infection-induced preterm ripening as did activation of prostaglandin synthesis via increased Ptgs2. Exposure to cytokines and growth factors induces Ptgs2 expression in macrophages and other immune cells (24). Another notable dissimilarity between the LPS model and term- or RU486-induced ripening was the alteration in immune cell populations, in particular increases in neutrophil numbers and the failed increase in tissue monocytes. While the neutrophil numbers are consistent with increased expression of the neutrophil chemoattractant, Cxcl2, the phenotype of the neutrophils is unique.

Unexpected findings in the preterm models are 1) the failure to induce MPO activity despite the increase in neutrophil numbers and neutrophil MMP8 expression in LPS-treated pregnant mice and 2) the increase in mRNA expression and immunohistochemical detection of MMP8 protein in neutrophils in the blood vessels and within mucus in the cervical lumen of RU486-treated cervices with no increase in neutrophil numbers or MPO activity (Fig. 7). This observation may in part be explained by earlier studies in women, which reported differing phenotypes in neutrophils from NP vs. pregnant women (33). Altered trafficking and cellular localization of the MPO enzyme in neutrophils from pregnant women compared with NP women resulted in a neutrophil that could not be fully activated. The neutrophil phenotype within the cervical microenvironment is unique in that certain functions are maintained such as MMP8 expression while MPO activity is suppressed. Further studies are required to confirm that MMP8 expression correlates with elevated MMP8 activity, to determine how the activation status of neutrophils is regulated by the tissue microenvironment, and which neutrophil functions are maintained during pregnancy and parturition.

Regardless of the mechanism, changes in the structure of the cervical ECM leading to loss of tensile strength must be common to preterm and term cervical ripening. While loss of total collagen in the cervical ECM does not appear to be a regulatory component of this process, the combined actions of Mmp8, Adamts1, and Adamts4 and perhaps additional unidentified proteases in PTB models likely result in modulation of the cervical ECM to facilitate loss of tensile strength and thus bypass processes in place during normal physiologic cervical ripening. Further studies are required to understand how MMP8 protease activity alters fibrillar collagen structure. In addition, studies to determine how Adamts1 and Adamts4 mediate the breakdown of versican and perhaps other proteoglycans leading to loss of tensile strength in the prematurely ripened cervix are needed. Protease expression could in part be regulated by prostaglandins resulting from increased synthesis of Ptgs1 in the RU486-treated mice and Ptgs2 in the LPS-treated mice. While neither Ptgs1 nor Ptgs2 is up-regulated in the mouse cervix during term ripening (Fig. 3), administration of prostaglandin E2 will induce cervical ripening in women and animal models and can affect mechanical properties of tissues through regulation of proteoglycan metabolism (34–38). A role of prostaglandins in PTB is supported by previous studies in mice and primates that report attenuation of infection-mediated PTB by administration of a PTGS2 inhibitor or a nonselective PTGS inhibitor (27, 39). Recently, transformation related protein 53 (Trp53), a tumor suppressor gene, has been identified as an upstream negative regulator of the Ptgs2 pathway in the uterus (40). Similar to the LPS infection model, loss of Trp53 expression in the murine female reproductive tract results in PTB through activation of the Ptgs2 pathway and can be reversed through administration of a PTGS2 inhibitor. Taken together, these observations suggest activation of PTGS2 in the female reproductive tract is an important mediator of inflammation induced PTB.

The clinical implications of the current findings are significant and have the potential to improve current strategies for development of therapies and clinical tools for detection of prematurity. First and foremost, these findings suggest that development and choice of therapies for prevention of prematurity will depend on 1) an understanding of molecular processes that regulate physiologic parturition and 2) a better understanding of the causes of PTB because 50% of cases have an unknown etiology. In addition, while the mechanisms for premature cervical ripening may differ between physiologic and premature ripening, induction of common gene markers were identified (e.g., Adamts1 and Adamts4). Future identification of molecular changes that occur very early in the progression of premature cervical ripening that are perhaps common to several models of PTB will be useful in identifying early detection biomarkers to assess risk of impending PTB and targets for therapy.