Cervical dilation and normal progress of labor are preceded by complex tissular alterations that transform the dense and nonpliable, “unripe” uterine cervix into a soft and distractible organ. Therefore, cervical softening is an essential prerequisite for fetal expulsion and delivery. This timely and highly controlled process is a key element of normal parturition, and cervical dysfunction can lead to serious conditions occurring during pregnancy. Preterm birth is often caused by premature cervical effacement and is the leading causative factor for neonatal morbidity.1,2 A delay or lack of timely cervical softening is often linked with postterm pregnancy and is likely to result in prolonged and dysfunctional labor. Traditionally, cervical ripening and dilation have been viewed as an active and dynamic process involving inflammatory and neurogenic components.3,4 The cervix, which is mainly composed of collagen types I and III (70–80%) and smooth muscle fibers (10–20%), thereby undergoes extensive extracellular matrix (ECM) remodeling and degradation of connective tissue at the end of gestation.3,5,6
Several studies have examined a number of factors that are thought to contribute to cervical effacement, and the increase in brain nitric oxide synthase and inducible nitric oxide synthase in term and postpartum cervical tissue has led investigators to believe that the local release of the nitric oxide, a powerful myometrial relaxant, is a key process in cervical ripening.7–9 In addition, interleukin (IL)-1, IL-6, IL-8, and granulocyte colony stimulating factor (G-CSF) have also been shown to take part in ECM remodeling of cervical tissue, which fits well the hypothesis that cervical effacement can be seen as some sort of inflammatory process.3,10–13 In addition, Sugano et al14 have demonstrated that the proinflammatory cytokine, platelet-activating factor (PAF), accelerates collagenolysis via induction of monocyte chemoattractant protein 1 and RANTES (regulated upon activation, normal T cell expressed and secreted) in softened cervical tissue samples. Finally, a number of neuropeptides, such as substance P, capsaicin, neurokinin A, calcitonin-gene–related peptide, and secretneurin, complete the list of substances that have been found to favor cervical softening and to participate in a neuroinflammatory reaction leading to cervical ECM degradation.4 However, although an impressive number of individual molecules have already been identified as factors being potentially involved in cervical ripening, it has not yet been possible to fit all possible components into a uniform hypothesis that could explain the principal mechanisms leading to cervical effacement.
Complementary DNA (cDNA) arrays provide a powerful tool for overcoming this obstacle and allow for monitoring the simultaneous expression of hundreds of possible genetic/molecular factors in tissues or body fluids.15 The present study was therefore undertaken to investigate the gene expression pattern in pooled samples from 10 ripe and 10 unripe cervical tissue biopsies to identify alterations in gene expression and possible novel markers that are associated with physiological ripening of the human uterine cervix. Real-time polymerase chain reaction (PCR) was then performed for selected genes to validate the cDNA array results and confirm the gene expression pattern in each of the 20 samples separately.
MATERIALS AND METHODS
The study was approved by the Institutional Review Board of the University of Vienna, and only women who had signed an informed consent were enrolled. The study was performed at the Department of Obstetrics and Gynecology of the Medical University of Vienna from February through March 2004. Women with a history of cervical intraepithelial neoplasia, pelvic inflammatory disease, bleeding disorders, or surgery of the lower genital tract were excluded from the study. None of the women had received prostaglandin or oxytocin infusions during pregnancy. Ten women with completely dilated cervices after spontaneous ripening, who were undergoing normal, noninstrumental vaginal delivery, were included in group A. Each of the 10 women underwent cervical biopsies immediately after fetal expulsion, but before delivery of the placenta, during routine cervical inspection. Ten women who underwent primary cesarean delivery for non–cervix-related reasons, such as breech position or prior cesarean delivery, were included in group B. In these women, cervical biopsies were obtained from competent cervices (Bishop score < 4) during cesarean delivery. Patient characteristics are presented in Table 1. The Mann-Whitney U test was used to identify significant differences in maternal age, gestational age, and length of labor, and the Fisher exact test was used to identify differences in parity and race. P < .05 was considered statistically significant.
Cervical biopsies were obtained by clipping a 3 × 3 × 3 mm piece of tissue from the anterior margin of the cervix, between 10 and 2 o'clock, and these were immediately snap-frozen in liquid nitrogen and stored at −80°C until used. For total RNA isolation, the tissues were pulverized and dissolved in 1,000 μL of homogenization buffer containing 0.3 mol/L sodium acetate, pH 6, 0.02 mol/L ethylenediaminetetraacetic acid (EDTA), and 1:50 (w/w) vanadyl ribonucleoside complex (Gibco BRL, Gaithersburg, MD). After addition of 1 vol of SS-phenol and 1 vol chloroform/isoamyl alcohol (24:1), the tubes were heated to 55°C for 5 minutes before being placed into ice for another 5 minutes. Phases were separated by a 2-minute spin at 5,000g, and the upper phase was removed to a new tube. RNA was precipitated in the presence of 2.5 vol of absolute ethanol for 10 minutes on dry ice and microcentrifuged for 10 minutes at 4°C. RNA pellets were then washed with 80% ice-cold ethanol and dissolved in RNase-free water. Total RNA concentrations were assessed by ultraviolet spectrophotometry (Pharmacia, Uppsala, Sweden).
We chose to pool 10 cervical samples in each of the 2 groups to minimize unbalanced cDNA expression profiles resulting from inaccurate pipetting, while at the same time yielding an acceptable “background-to-noise ratio” (the ratio would have increased by pooling a larger number of patient RNAs per group).
32P-labeled cDNAs were synthesized with the use of pooled total RNA from cells collected from either normal cervices or invasive cervical cancers by reverse transcription in the presence of [α-32]deoxycytidine triphosphate ([α-32]dCTP). Briefly, total RNAs (20 μg each) were denatured at 75°C for 10 minutes in the presence of 8 pmol of dT15VN (V = A, G, and C; N = A, G, C, and T) mix. After the denaturation step, cDNAs were synthesized by incubation at 37°C for 1 hour in a master mix of 40 μL total volume, containing 3 μL of deoxyribonucleoside triphosphate (500 μmol/L, without dCTP), 5 μL of [α-32]dCTP (3,000 Ci/mmol; Amersham Life Science, Cleveland, OH), and 1,600 units of moloney murine leukemia virus–reverse transcriptase (Promega, Madison, WI) in 1 × reverse transcriptase buffer (as supplied by the manufacturer). The reaction was terminated by heating for 10 minutes at 75°C, and unincorporated nucleotides were removed by spin column purification. (Chroma Spin-2000; Clontech Laboratories, Palo Alto, CA). For each reaction, approximately 2 × 107 cpm were incorporated in the final product.
After purification, labeled cDNAs were denatured by boiling for 5 minutes and then hybridized to an Atlas human cancer cDNA array (Clontech Laboratories) in 2 × 106 cpm/mL ExpressHyb hybridization solution (Clontech). Membranes were prehybridized at 68°C for at least 2 hours before probe addition. Hybridization was performed at 68°C in a rolling bottle overnight. After the first 2 washes with 2 × standard saline citrate and 0.1% sodium dodecyl sulfate (SDS) at 68°C for 20 minutes, the membranes were subjected to a stringent wash with 0.1 × standard saline citrate, 0.5% SDS, and 0.1 mmol/L EDTA at 68°C. Membranes were then exposed to X-ray film (Hyperfilm, Amersham) for 1–5 days at −70°C. Signal densities were determined by using the optical densitometer DESKTop Plus Scanner (Pharmacia). Further analysis of the digitalized images was done on a SPARC-Station 10 (Sun Microsystems, Mountain View, CA) using PDQUEST 5.0 software (PDI Inc, Huntington Station, NY).
Differences in gene expression are depicted as ratios of signal intensity in the array containing RNA of dilated cervices to signal intensity in the array containing competent cervices. Because the cDNA array becomes less reliable in distinguishing differential gene expression as expression ratios approach 1, a ratio of more than 2.5 for up-regulated genes and of less than 0.5 for down-regulated genes was chosen to distinguish genes with a pronounced differential gene expression from unchanged genes.
Total RNA (0.5–1 μg) of each of the cervix samples was incubated with random hexamers (Promega) and adjusted to 5 μL with diethyl pyrocarbonate–treated, double-distilled H2O before being heat-denatured at 70°C for 5 minutes and chilled on ice. The samples were then added to a reaction mix consisting of 4 μL of 5 × reverse transcriptase buffer (250 mmol/L Tris-HCl, pH 8.3, 375 mmol/L KCl, 15 mmol/L MgCl2), 2 μL deoxyribonucleoside triphosphate mixed stock solution (10 mmol/L, each from Pharmacia), 1 μL RNase inhibitor (Applied Biosystems, Vienna, Austria), 1 μL dithiothreitol, and 1 μL moloney murine leukemia virus–reverse transcriptase (200 U/μL, Amersham Bioscience Ltd). The reaction mix was vortexed and centrifuged briefly before being incubated at 37°C for 1 hour. The reaction was then stopped by heating to 80°C for 10 minutes, and tubes were chilled on ice before they were briefly centrifuged and stored at −20°C.
Polymerase chain reaction was performed by adding 20 μL of reaction mix to 2.5 μL of 10 × PCR buffer, 2 μL deoxyribonucleoside triphosphate mix (10 mmol/L each), 0.25 μL primer (100 μmol/L), and 5 μL Taq polymerase (5 U/mL) to the respective primers (vascular endothelial growth factor [VEGF] precursor: CGA AGT GGT GAA GTT CAT GGA TGT and TTC TGT ATC AGT TTC CTG GTG; intercellular adhesion molecule-1 [ICAM-1] precursor: GTT GTG ACA GCT GCT GTT CCC ATC and GAT GAA GGG GTT CCT GAA TGC CTC; 5-hydroxytryptamine 1A receptor [5HT1A]: CGA GGC GAA TCT TCG CGC TGC and CAG GGA TCC TGT AGC CTC GAC; cyclic adenosine monophosphate [cAMP]-response element binding protein [CREB2]: GCC ACT GTG ATA GAG GCT GA and TGA GAA ATG TTG ACC ACA CAC T). A total volume of 45 μL was reached by adding diethyl pyrocarbonate–treated, double-distilled H2O. Cycling conditions were as follows: depending on the primers, 25–35 cycles were carried out at 94°C for 1 minute, 68°C for 2 minutes, and 72°C for 2 minutes, with an extension of 5 seconds with each subsequent cycle. Placental total RNA was used in reverse transcriptase–PCR experiments for each of the primers, and double-distilled H2O was used instead of total RNA for negative controls.
Agarose gel electrophoresis was carried out by adding 20 μL of each of the PCR products, which were subjected to 1.2% Nusieve agarose gel electrophoresis in 1 × tris-borate-EDTA buffer and separated by applying a constant voltage at 80 V for 1–2 hours. DNA bands were then visualized by ethidium bromide. Band size was determined by a coloaded DNA size marker.
Figure 1 shows cDNA array hybridization signals on an Atlas human cancer cDNA array displaying pronounced differences in gene expression profiles of pooled postpartum cervical tissue compared with that of preterm, ie, unripe cervical specimens. The intensity of each spot was measured by densitometry. Expression levels of individual genes in postpartum cervical cells are shown here in relation to corresponding signal intensities in unripe, nondilated cervical cells. We have decided to consider differences in gene expression “up-regulated” if at least a 2.5-fold difference in mRNA was detected. This resulted in 40 genes, which were up- or down-regulated during transition from nonsoftened, preterm cervical epithelial cells to cervical carcinoma cells (Tables 2 and 3).
The following gene products acting as chemokines, cytokines, and growth factors were found to be altered in postpartum specimens: interleukin-8 precursor (IL-8), ratio (R) 999.0; VEGF, R 999.0; monocyte chemotactic protein 1 precursor (MCP1), R 999.0; connective tissue growth factor precursor (CTGF), R 999.0; tumor necrosis factor receptor 1 (TNFR1), R 999.9; insulin-like growth factor IA precursor (IGF1A), R 999.0; glutaredoxin, R 999.0; macrophage inflammatory protein 2 α (MIP2-α), R 19.37; autocrine motility factor receptor (AMF receptor; AMFR), R 3.04; interleukin-13 precursor (IL-13), R 2.83; and endothelin 2 (ET2), R 2.64.
Basic transcription factors found to be elevated in ripe cervical tissue included basic transcription element-binding protein 2 (BTEB2), R 999.0; transcriptional enhancer factor (TEF1), R 999.0; mitotic growth and transcription activator, R 999.0; activator of RNA decay (ARD-1), R 59.54; cAMP-response element binding protein, R 42.03; transcription factor TFIIB, R 10.04; Ini, R 5.57; transcription factor ETR101, R 3.61; transcription factor TFIIIB 90 kDa subunit (HTFIIIB90), R 2.58; and transcription factor DP2 (Humdp2), R 2.51.
Gene expressions involved in mechanisms of mitosis, cell cycle regulation, and cell metabolism included helix-loop-helix protein, R 999.0; DNA topoisomerase I (TOP1), R 999.0; 23-kDa highly basic protein, R 25.40; dual specificity mitogen-activated protein kinase 3 (MAP kinase kinase 3), R 10.05; liver glyceraldehyde 3-phosphate dehydrogenase (GAPDH), R 5.11; and CDC25B, R 5.00.
Cell-cell interaction and extracellular communication-related gene products included neural cell adhesion molecule phosphatidylinositol-linked isoform precursor (NCAM120), R 999.0; integrin-linked kinase (ILK), R 999.0; intercellular adhesion molecule 1 (ICAM-1) precursor, R 44.65; epithelial discoidin domain receptor 1 precursor (EDDR1; DDR1), R 3.03; neuronal acetylcholine receptor protein α-3 subunit precursor (NACHRA3), R 999.0; glutamate receptor 1 precursor (GLUR-1), R 999.0; 5-hydroxytryptamine 1A (5HT1A) receptor, R 999.0; and 5-hydroxytryptamine 3 receptor precursor (5-HT-3), R 3.48. Apoptosis-related genes included tumor necrosis factor (TNF)-related apoptosis inducing ligand (TRAIL), R 999.0; caspase-4 precursor (CASP4), R 999.0; BCL-2-related protein A1 (BCL2A1), R 999.0; apoptosis regulator bclw, R 999.0; caspase-9 precursor (CASP9), R 3.90; and nucleobindin precursor (NUC), R 2.68.
Genes that showed decreased expression in cells deriving from postpartal cervices included inducible nitric oxide synthase (INOS), R 0.45; fasL receptor; apoptosis-mediating surface antigen fas, R 0.37; tight junction protein zonula occludens (ZO-1), R 0.22; mitochondrial matrix protein P1 precursor, R 0.20; fibronectin receptor α subunit (FNRA), R 0.14; and integrin α 6 precursor (ITGA6), R 0.05 (Table 2). Five hundred and fifty-six genes that were differentially up-regulated less than 2.5-fold or that were down-regulated less than 0.5 fold in completely dilated cervices were excluded from further analysis.
To validate differential gene expression levels and to confirm that mRNA levels of genes were also differentially expressed in each of the individual unripened versus completely dilated tissues, reverse transcriptase PCR analysis was performed in selected, highly up-regulated genes. Semiquantitative mRNA expression of VEGF precursor, ICAM-1 precursor, 5-hydroxytryptamine 1A receptor, and cAMP-response element binding protein (CREB2) was determined for each of the 10 patients in both groups. As expected, we observed a consistent and strong signal in the samples from completely dilated cervices when compared with unripe cervices (Fig. 2).
The softening and dilation of cervical tissue requires the rapid reorganization of the ECM and involves several intra- and intercellular regulation mechanisms, finally resulting in a state that has been titled neuro-inflammation.3,4,12 To gain more insight into the complex transformational changes in gene expression, which lead to cervical effacement, we have applied cDNA array technology to unripe, nonsoftened cervical tissues and compared them with completely dilated samples obtained during spontaneous vaginal delivery in women matched for age, race, parity, and week of gestation (Table 1). While we found no difference in the 2 groups in terms of maternal age and parity, women undergoing cesarean delivery had a mean gestational age that was slightly higher (284 versus 268 days). Although this difference was statistically significant, we do not believe it to be a clinically relevant confounder in gene expression in our patient cohort, because none of the 5 genes investigated in our validation study showed variations along the gestational age (Fig. 2).
In consistence with previous studies that reported the altered expression of inflammatory cytokines and chemoattractants in ripened cervical tissue, we detected elevated expression levels of glutaredoxin, IL-8, VEGF, monocyte chemotactic protein 1, and insulin growth factor-1A precursor in postpartum cervical tissue.5,12,14,16,17 Interestingly, we also observed altered expression of the proinflammatory autocrine motility factor receptor, IL-13 precursor, endothelin 2, and macrophage inflammatory protein 2 α, which have not been associated with cervical effacement before. Endothelin 2 is a well-known chemoattractant18 that has been shown to stimulate the invasion of breast tumor cells19 and synergistically increases the expression of IL-13 in eosinophils.20 Similarly, autocrine motility factor receptor and macrophage inflammatory protein 2 α have been demonstrated to exert proinflammatory properties and appear to facilitate cell motility and ECM breakdown,21,22 thereby supporting the pivotal roles of inflammatory compounds in the ripening process. Within this, the dissolution of collagen fibers and cell-cell junctions appears to be an essential precondition for ECM-lysis. This is reflected by our observations of the decreased expression of the cell-cell interaction-related genes such as tight junction protein zonula occludens (ZO-1) and integrin α 6 precursor. Both have been shown to have functional roles in intercellular attachment, cell migration, and cytoskeletal anchorage23,24 and might also contribute to the stabilization of ECM components of competent cervical tissue. In contrast, expression levels of integrin-linked kinase, which has been demonstrated to induce expression of ECM-degrading matrix metalloproteinase 9 in brain tumor cell lines25 was found to be elevated in postpartum specimens.
Attention has also been drawn to studies indicating that primary afferent nerve fibers, corresponding receptors, and neuropeptides participate in mechanisms that induce cervical remodeling toward parturition. In this context, Stjernholm and coworkers6 provided supporting evidence for the abundant presence of neuronal marker protein S-100 in term cervices and suggested the release of the neuropeptides vasoactive intestinal peptide (VIP), calcitonin-gene–related peptide, and human peptide histidine isoleucine amide (PHM-27) to positively influence cervical remodeling and dilation. The present study identified the neurogenic receptor proteins 5-hydroxytryptamine 1A and -3 receptor, glutamate receptor 1 precursor (GLUR-1), and neuronal acetylcholine receptor protein α3 subunit precursor (NACHRA3) to be slightly overexpressed in ripened cervical tissue. Interestingly, 5-hydroxytryptamine receptors have been linked to relaxation of uterine smooth muscle cells and might therefore also have a role in the relaxation of cervical smooth muscle fibers at the time of effacement and dilation.
Finally, we also observed several apoptosis-related genes to be differentially expressed in unripe versus postpartal cervical tissue. For example, proapoptotic caspase-4 precursor (CASP4), caspase-9 precursor (CASP9), BCL-2-related protein A1 (BCL2A1), and TNF-related apoptosis-inducing ligand (TRAIL) were found to be elevated in ripe cervices. This may reflect the physiological breakdown of cellular components that build up the integrity of premature cervical tissue and their loss in cervical softening or dysfunction.
Taken together, we have investigated tissue-specific changes in the gene expression pattern of competent versus completely effaced uterine cervices by using cDNA array technology. cDNA arrays provide a valuable tool to monitor the simultaneous expression of several differentially expressed genes. However, their use primarily serves as a survey and is often limited by the relatively small number of analyzed samples and the high cost of the procedure. Although these obstacles also limit the present study, we have identified several angiogenetic, proinflammatory, proapoptotic, and ECM-associated genes that have not been associated with the ripening process before. The differential expression of these compounds hints at novel factors and mechanisms that influence the ripening and dilation of the uterine cervix. A thorough understanding of the molecular events that accompany physiological effacement is essential to better understand pathological conditions such as premature delivery and postterm pregnancy.
1. Copper RL, Goldenberg RL, Creasy RK, DuBard MB, Davis RO, Entman SS, et al. A multicenter study of preterm birth weight and gestational age–specific neonatal mortality. Am J Obstet Gynecol 1993;168:78–84.
2. Robertson PA, Sniderman SH, Laros RK Jr, Cowan R, Heilbron D, Goldenberg RL, et al. Neonatal morbidity according to gestational age and birth weight from five tertiary care centers in the United States, 1983 through 1986. Am J Obstet Gynecol 1992;166:1629–41; discussion 1641–5.
3. Winkler M. Role of cytokines and other inflammatory mediators. BJOG 2003;110(suppl):118–23.
4. Collins JJ, Usip S, McCarson KE, Papka RE. Sensory nerves and neuropeptides in uterine cervical ripening. Peptides 2002;23:167–83.
5. Ekman-Ordeberg G, Stjernholm Y, Wang H, Stygar D, Sahlin L. Endocrine regulation of cervical ripening in humans: potential roles for gonadal steroids and insulin-like growth factor-I. Steroids 2003;68:837–47.
6. Stjernholm Y, Sennstrom M, Granstrom L, Ekman G, Liang Y, Johansson O. Neurochemical and cellular markers in human cervix of late pregnant, postpartal, and nonpregnant women. Acta Obstet Gynecol Scand 2000;79:528–37.
7. Bao S, Rai J, Schreiber J. Brain nitric oxide synthase expression is enhanced in the human cervix in labor. J Soc Gynecol Investig 2001;8:158–64.
8. Ali M, Buhimschi I, Chwalisz K, Garfield RE. Changes in expression of the nitric oxide synthase isoforms in rat uterus and cervix during pregnancy and parturition. Mol Hum Reprod 1997;3:995–1003.
9. Buhimschi I, Ali M, Jain V, Chwalisz K, Garfield RE. Differential regulation of nitric oxide in the rat uterus and cervix during pregnancy and labour. Hum Reprod 1996;11:1755–66.
10. Schmitz T, Leroy MJ, Dallot E, Breuiller-Fouche M, Ferre F, Cabrol D. Interleukin-1β induces glycosaminoglycan synthesis via the prostaglandin E2 pathway in cultured human cervical fibroblasts. Mol Hum Reprod 2003;9:1–8.
11. Winkler M, Rath W. Changes in the cervical extracellular matrix during pregnancy and parturition. J Perinat Med 1999;27:45–60.
12. Sennstrom MK, Brauner A, Lu Y, Granstrom LM, Malmstrom AL, Ekman GE. Interleukin-8 is a mediator of the final cervical ripening in humans. Eur J Obstet Gynecol Reprod Biol 1997;74:89–92.
13. Liggins GC. The role of the hypothalamic-pituitary-adrenal axis in preparing the fetus for birth. Am J Obstet Gynecol 2000;182:475–7.
14. Sugano T, Narahara H, Nasu K, Arima K, Fujisawa K, Miyakawa I. Effects of platelet-activating factor on cytokine production by human uterine cervical fibroblasts. Mol Hum Reprod 2001;7:475–81.
15. Ramaswamy S, Golub TR. DNA microarrays in clinical oncology. J Clin Oncol 2002;20:1932–41.
16. Marvin KW, Keelan JA, Eykholt RL, Sato TA, Mitchell MD. Expression of angiogenic and neurotrophic factors in the human amnion and choriodecidua. Am J Obstet Gynecol 2002;187:728–34.
17. Sahlin L, Wang H, Stjernholm Y, Lundberg M, Ekman G, Holmgren A, et al. The expression of glutaredoxin is increased in the human cervix in term pregnancy and immediately post-partum, particularly after prostaglandin-induced delivery. Mol Hum Reprod 2000;6:1147–53.
18. Grimshaw MJ, Wilson JL, Balkwill FR. Endothelin-2 is a macrophage chemoattractant: implications for macrophage distribution in tumors. Eur J Immunol 2002;32:2393–400.
19. Grimshaw MJ, Hagemann T, Ayhan A, Gillett CE, Binder C, Balkwill FR. A role for endothelin-2 and its receptors in breast tumor cell invasion. Cancer Res 2004;64:2461–8.
20. Cui P, Sharmin S, Okumura Y, Yamada H, Yano M, Mizuno D, et al. Endothelin-1 peptides and IL-5 synergistically increase the expression of IL-13 in eosinophils [published erratum appears in Biochem Biophys Res Commun 2004;318:802]. Biochem Biophys Res Commun 2004;315:782–7.
21. Yanagawa T, Watanabe H, Takeuchi T, Fujimoto S, Kurihara H, Takagishi K. Overexpression of autocrine motility factor in metastatic tumor cells: possible association with augmented expression of KIF3A and GDI-β. Lab Invest 2004;84:513–22.
22. Call DR, Nemzek JA, Ebong SJ, Bolgos GR, Newcomb DE, Wollenberg GK, et al. Differential local and systemic regulation of the murine chemokines KC and MIP2. Shock 2001;15:278–84.
23. Gabler C, Chapman DA, Killian GJ. Expression and presence of osteopontin and integrins in the bovine oviduct during the oestrous cycle. Reproduction 2003;126:721–9.
24. Giepmans BN. Gap junctions and connexin-interacting proteins. Cardiovasc Res 2004;62:233–45.
25. Troussard AA, Costello P, Yoganathan TN, Kumagai S, Roskelley CD, Dedhar S. The integrin linked kinase (ILK) induces an invasive phenotype via AP-1 transcription factor–dependent upregulation of matrix metalloproteinase 9 (MMP-9). Oncogene 2000;19:5444–52.