Many complementary or competing signalling pathways bear an influence on the myometrium at any one time, and because the retinoic acid signalling pathway influences differentiation of a wide array of human tissues, this may be one of the determinants of myometrial differentiation during pregnancy. We have explored the novel hypothesis that the retinoids may act as important regulators in controlling the differentiated state of the human myometrium during pregnancy by characterizing the expression profiles for cellular retinoid-binding proteins CRBPI, CRABPI and CRABPII in non-pregnant, pregnant (non-labouring) and labouring human myometrium taken from the functionally distinct upper and lower uterine segments. In addition, we have investigated the effect of all-trans retinoic acid (ATRA) on the expression of several retinoic acid response genes including cyclooxygenase-2 (COX-2) and connexin-43 (Cx-43). Different spatial and temporal patterns of expression were observed for CRBPI, CRABPI and CRABPII within the upper and lower uterine segments through the three trimesters of pregnancy and in labour. Furthermore, the expression of COX-2, Cx-43, CRABPI, the transcription factor c-Jun and the retinoic acid receptor RARβ altered in response to different concentrations of ATRA, suggesting that the differential expression of cellular retinoid-binding proteins may lead to different levels of retinoic acid being delivered to its nuclear targets, leading to the differential expression of specific target genes within the myometrium during pregnancy.
Human myometrium differentiates during pregnancy to define functionally distinct upper and lower uterine segments: the contractile capability of upper segment myometrium is enhanced by the hypertrophy and hyperplasia of muscle fibres, whereas lower segment myometrium is modified, so that it can accommodate expulsion of the fetus during parturition. Strong, co-ordinated myometrial contractions must be avoided however until 37–42 weeks of gestation while the uterus supports the maturing fetus. If the uterus fails to maintain its state of quiescence until this time, preterm labour ensues leading to the birth of a baby ill equipped to deal with the physiological challenges of survival ex utero. As a consequence, although only 7–10% of pregnancies in the developed world culminate in preterm birth, there is a great excess of mortality and serious long-term morbidity in preterm babies compared with their counterparts born at full term.
The differential expression of specific sets of genes in the upper and lower segments of the uterus is crucial to the control of myometrial differentiation and to the maintenance of uterine quiescence in human pregnancy (Fuchs et al., 1984; Brodt-Eppley and Myatt, 1999; Sparey et al., 1999; MacDougall et al., 2003). Many complementary or competing signalling pathways bear an influence on the myometrium at any one time, and because the retinoic acid signalling pathway influences the differentiation of a wide array of human tissues (Ross et al., 2000), this may be one of the determinants of myometrial differentiation. The ability of retinoids to influence the differentiated state of a tissue is dependent on the presence of cellular retinoid-binding proteins and also nuclear receptors. In retinoid target cells, circulating retinol enters the cytoplasm where it is bound by cellular retinol-binding protein I (CRBPI; Posch et al., 1991). Through a series of reduction reactions, retinol is then converted to the active ligand, all-trans retinoic acid (ATRA), which is in turn bound by one of two cellular retinoic acid-binding proteins, CRABPI and CRABPII (Astrom et al., 1991; Posch et al., 1992; Napoli, 1996). These proteins determine the fate of ATRA by (i) encouraging sequestration of active ligand from the cell, (ii) presenting the ligand to cytoplasmic enzymes for inactivation or (iii) transporting the ligand to its biological site of action, the nucleus (Dong et al., 1997; Delva et al., 1999; Noy, 2000; Budhu and Noy, 2002). There is evidence to suggest that in some tissues, binding by CRABPI preferentially leads to CYP26-mediated inactivation of ATRA (Boylan and Gudas, 1991; Won et al., 2004). In contrast, CRABPII appears to facilitate the transport of bound ATRA to the nucleus, and possibly its transfer into the nucleus, because it has been detected in both the cytoplasmic and the nuclear fractions of many cultured cell lines and fresh-tissue preparations (Dong et al., 1997; Gaub et al., 1998).
Previous studies have described the variable expression of CRBPI, CRABPI and CRABPII within the rat uterus (Bucco et al., 1996; Zheng and Ong, 1998) and also shown that their expression levels alter in response to treatment with estradiol (Li and Ong, 2003). The expression of nuclear retinoic acid receptors (RARs) and retinoid-X receptors (RXRs) has been reported in endometrial epithelial cells (Siddiqui et al., 1994), non-pregnant human myometrial cells and leiomyomata (Tsibris et al., 1999, Vienonen et al., 2004); however, the expression of cellular retinoid-binding proteins has not been defined in human myometrium to date. The aim of the present study was to seek evidence in support of a role for the retinoic acid signalling pathway in determining the differentiated state of human myometrium by characterizing the expression profiles of CRBPI, CRABPI and CRABPII protein and mRNA in non-pregnant, pregnant (non-labouring) and labouring human myometrium taken from the upper and lower uterine segments. In addition, the effect of ATRA on the expression of known retinoic acid response genes was observed in cultured human myometrial cells.
Materials and methods
Myometrial tissue samples
Myometrial biopsies were obtained from non-pregnant premenopausal women (age range 32–44 years) undergoing hysterectomy for benign gynaecological disorders such as menorrhagia. Patients with uterine leiomyoma were excluded from the study. In addition, biopsies were taken from women having a termination during the 12th or 14th week of pregnancy, women with a full-term pregnancy undergoing elective Caesarean section before labour and women labouring spontaneously at term who were undergoing emergency Caesarean section. Note, in this study, non-pregnant patients were age matched with the pregnant patients. Non-pregnant samples were taken from the middle of the uterine corpus. In pregnancy, upper segment samples were taken from an area close to the fundus avoiding the placental bed, whereas lower segment samples were taken a short distance above the internal cervical os. Tissue was snap-frozen in cooled isopentane and liquid nitrogen and stored at −70°C unless being used to establish primary cells in culture. Prior written consent for tissue biopsy was obtained in each case under terms agreed with the Newcastle and North Tyneside Health Authority Ethics Committee.
Myometrial cells were isolated and established in primary culture, as previously described (Bailey et al., 2005), with complete d-valine Dulbecco’s modified Eagle’s medium (DMEM), which is known to inhibit fibroblast growth, supplemented with 10% fetal calf serum, penicillin (1 IU/ml) and streptomycin (1 ng/ml) at 37°C with 5°C CO2. Myometrial cells were treated with 0, 1, 5 or 10 µM ATRA at 80–90% confluency in antibiotic-free DMEM medium as required, at 37°C for 24 h. All experiments were undertaken on subculture passages 2–3 and performed in triplicate. Cells were then harvested by scraping, rinsed in phosphate-buffered saline (PBS) and stored at −70°C. Note, in experiments undertaken to determine the effect of ATRA on the transcriptional activation of COX-2, myometrial cells were also treated with lipopolysaccharide (LPS) at 1 µg/ml to induce COX-2 expression.
RNA extraction and RT–PCR
Total RNA was prepared from non-pregnant (NP), 12 and 14 weeks of gestation, term pregnant (P) and term labouring (L) myometrium using the SV Total RNA Isolation System, as recommended by the manufacturer (Promega). Eluted and quantified RNA was stored at −70°C. First-strand cDNA was synthesized at 50°C using 1 µg of each RNA sample with 20 U of Superscript III reverse transcriptase (Invitrogen Life Technologies), 10 U of RNase inhibitor and 100 ng of oligo(dT) as primer. PCR amplification was performed with 1–3 µl of cDNA template with CRBPI primers (sense: GTCGACTTCACTGGGTACTGGA; anti-sense TTGAATACTTGCTTGCAGACCACA), CRABPI primers (sense: CGGCACCTGGAAGATGCGCA; anti-sense: CCACGTCATCGGCGCCAAAC) or CRABPII primers (sense: CCCAACTTCTCTGGCAACTGGA; anti-sense CTCTCGGACGTAGACCCTGGT) using 5 U of Platinum Taq Polymerase (Invitrogen Life Technologies). PCR was carried out with an initial hot start at 94°C (4 min), 55°C (30 s) and 72°C (1 min), followed by 20–25 cycles at 94°C (30 s), 55°C (30 s) and 72°C (1 min). Reactions were terminated in the exponential phase to allow semi-quantitative analyses. Glyceraldehyde-3-phosphatedehydrogenase (GAPDH)-specific primers (spanning a short 104-bp intron) were also designed as control primers. DNA sequences for the GAPDH primers were sense CTGCCGTCTAGAAAAACC and anti-sense CCACCTTCGTTGTCATACC.
Western blot immunodetection
Samples of myometrial tissue were prepared for western blot immunodetection, as previously described (Pollard et al., 2000). Myometrial nuclear and cytoplasmic fractions were isolated from cultures of pregnant myometrial cells using an Active-Motif nuclear extraction kit following the manufacturer’s specified protocol. Briefly, cells were suspended in 1× hypotonic buffer and vortexed vigorously after the addition of 25 µl of detergent and centrifuged at 14 000 × g for 30 s. The supernatant (cytoplasmic fraction) was removed and the nuclear pellet resuspended in 50-µl ice-cold complete lysis buffer; the suspension was incubated for 30 min on ice and then vortexed vigorously and centrifuged for 10 min at 14 000 × g. The supernatant (nuclear fraction) was then stored at −80°C. The concentration of cytoplasmic fractions was increased by freeze-drying. Protein concentration was determined using a DC protein assay kit (Bio-Rad). A total of 50 or 100 µg of myometrial homogenate sample was denatured under reducing conditions in 0.5 mol/l of Tris, 5 mol/l of urea, 2.5% sodium dodecyl sulphate (SDS) and 3.5% β-mercaptoethanol and resolved on 15 or 18% polyacrylamide gels. Proteins were transferred to Hybond polyvinylidene difluoride (PVDF) membrane by voltage transfer. Membranes were then blocked with 5% non-fat milk and specific proteins detected with the following antibodies: anti-CRBPI (1:200, provided by Alexis Desmoulière, Centre Médicale Universitaire, Switzerland), anti-CRABPI and CRABPII (1:100, Santa Cruz) or a loading control antibody, the G-protein subunit, Gβ (1:2000, Santa Cruz) overnight at 4°C. ATRA-treated cells were separated on 10 or 15% polyacrylamide gels, transferred to PVDF membranes as previously described and membranes probed with the CRABPI primary antibody or specific antibodies to COX-2 (1:1000) (Upstate Biotechnology), connexin-43, Cx-43 (1:300), RARβ (1:200) and c-Jun (1:300) (Santa Cruz). Detection was carried out using an enhanced chemiluminescence (ECL) assay system (Amersham Pharmacia Biotechnologies).
ECL signals were quantified by scanning densitometry using a UMAX PS 2400 scanner at 700 dpi coupled to Intelligent Quantifier software (BioImage, Ann Arbor, MI, USA). The data presented are the mean ± SEM (n = 12 for all non-pregnant, pregnant and labouring paired samples). Quantification was performed using Prism 2.01 software (GraphPad Software), and data were then analysed using one-way analysis of variance (ANOVA) with Bonferroni’s multiple comparison test.
Expression profile for CRBPI in the myometrium during pregnancy
Western blot analyses and RT–PCR were undertaken to determine the relative levels of CRBPI, CRABPI and CRABPII within the myometrium. A 15-kDa band representing CRBPI protein was detected in all non-pregnant myometrium samples (Figure 1A and B). In upper segment tissue, in 12 and 14 weeks of gestation, the expression of CRBPI protein levels fell (Figure 1A), but they rose transiently to pre-pregnancy levels at term before falling once again, although not statistically significant, in labour (Figure 1B). Conversely in lower segment tissue, CRBPI protein expression was significantly low (P < 0.001) throughout all stages of pregnancy as well as in labour when compared with non-pregnant samples (Figure 1A and B). RT–PCR using RNA extracted from non-pregnant and term pregnant myometrium largely mirrored the results obtained by western blotting in terms of the spatial pattern for CRBPI within the upper and lower segment myometrium at term and in labour, although the elevated levels of CRBPI mRNA observed in the upper segment at term were maintained in labour (Figure 1D) and low levels of CRBPI mRNA expression were observed consistently in pregnant and labouring lower segment tissue (Figure 1D). No spatial variation was observed for CRBPI mRNA within the upper and lower segment myometrium at 12 or 14 weeks of gestation (Figure 1C).
Expression profile for CRABPI in the myometrium during pregnancy
The expression levels for CRABPI were next determined. Immunoblotting detected a single 16-kDa protein band in all non-pregnant samples (Figure 2B). There appeared to be a significant decrease in the expression of CRABPI levels in myometrial samples from 12 and 14 weeks of gestation in both the upper and the lower segments, with virtually no protein being detected in any samples tested when compared with non-pregnant samples. The levels of CRABPI protein remained depressed in upper and lower uterine samples from term and labouring (Figure 2B). CRABPI mRNA was detected in all myometrial samples, with slightly higher levels of expression being found in non-pregnant myometrium tissue compared with 12 and 14 weeks of gestation, and uterine samples from term and labouring samples (Figure 2C and D). In addition, CRBPI mRNA expression appeared to decrease in the lower uterine tissue at 12 and 14 weeks of gestation when compared with the upper uterine tissue (Figure 3C).
Expression profile for CRABPII in the myometrium during pregnancy
A 16-kDa protein band representing CRABPII was detected in all non-pregnant samples (Figure 3A and B). CRABPII protein levels increased in pregnancy, although not to a statistically significant extent in both the upper and the lower segments of the uterus at 12 and 14 weeks of gestation, with the expression in the upper segment being highest (Figure 3A). A further rise in the levels of CRABPII was noted in upper (P < 0.01) and lower segments (P < 0.01) at term, and these levels were maintained in labour (Figure 3B) when compared with those in non-pregnant myometrium. Western immunoblotting also indicated that CRABPII is present in both the nuclei and the cytoplasm of the myometrium (Figure 3E). A 411-bp product representing CRABPII mRNA was detected by RT–PCR at relatively low levels in all non-pregnant tissue samples (Figure 3C and D). The levels of mRNA expression rose in the 12th and 14th weeks of pregnancy, with the greatest rise being evident in the upper myometrial segment (Figure 3C). Interestingly, the levels of CRABPII mRNA increased in the lower uterine regions when comparing all term and labouring samples with non-pregnant samples, which also reflects the pattern of expression for CRABPII at the protein level.
The effect of ATRA on retinoic acid target genes
Myometrial cell cultures established from non-pregnant tissues were treated with increasing concentrations of ATRA (0, 1, 5 and 10 µM) for 24 h to determine the ligand’s effect on the expression of specific genes known to be retinoic acid response genes, including CRABPI, RARβ, Cx-43 and COX-2. The results shown in Figure 4A indicate that increasing concentrations of ATRA influence the levels of expression for all four target proteins in human myometrial cell cultures in both a gene-specific and a dose-dependent manner. The levels of CRABPI protein increased at 10 µM (Figure 4A; i), whereas the stimulatory effect of ATRA on RARβ and Cx-43 protein levels was observed at the lower concentration of 5 µM (Figure 4A; ii and iii). Two protein bands were observed for RARβ (50 and 55 kDa), but only the 50-kDa band increased sequentially as the concentration of ATRA increased. It is feasible that this doublet either is the result of alternative splicing (Nagpal et al., 1992) or represents variations in phosphorylation status. In contrast, treatment with ATRA in the presence of LPS resulted in a substantial reduction in the levels of COX-2 protein in myometrial cells reflected by the fall in the intensity of the 72-kDa band representing COX-2 protein (Figure 4A; iv). Quantification demonstrated that the levels of COX-2 decreased sequentially as the concentration of ATRA increased from 1 to 5 to 10 µM. In an additional set of experiments, we determined the effect of ATRA, together with the histone deacetylase (HDAC) inhibitor trichostatin A (TSA) on COX-2 protein expression. We observed that using ATRA (1 µM) and TSA (300 nM) in combination further decreased COX-2 protein levels when compared with the application of ATRA on its own (Figure 4B). We also demonstrate that the treatment with 1, 5 and 10 µM ATRA resulted in a decrease in protein levels of the transcription factor, c-Jun (Figure 4C, lanes 2–4) when compared with no treatment of ATRA (Figure 4C, lane 1).
This is the first study to characterize the cellular retinoid-binding proteins CRBPI, CRABPI and CRABPII in human myometrium cells during pregnancy. Spatial variations have been observed, with differences being found in patterns of expression between the functionally distinct upper and lower uterine segments. Temporal variations have also been found, with patterns of expression evolving through the three trimesters of pregnancy and in labour. Our results also provide evidence to show that CRABPII localizes to the nucleus of cultured myometrial cells and that human myometrial cells undergo a transcriptional response to the presence of ATRA.
In non-pregnant myometrium, the ready availability of CRBPI raises the possibility that this tissue manufactures ATRA. If so, the presence of both CRABPI and CRABPII in non-pregnant myometrium suggests that a balance exists between the inactivation of the ligand and its transport in an active form to the nucleus. These pathways remain to be tested by functional experiments, but active ligand made available to the nucleus could be important for the control of myometrial proliferation in vivo as has been demonstrated previously in other tissues (Won et al., 2004).
The retinoids may also be important in influencing other aspects of the differentiated state of the myometrium such as its contractile capability. For example, Cx-43 is recognized as an important modulator in communicating signals that promote contraction of the myometrium (Lefebvre et al., 1995; Sparey et al., 1999). Moreover, the up-regulation of Cx-43 and changes in its phosphorylation in response to retinoic acid have been reported in other cell types including human endometrial stromal cells (Tanmahasamut and Sidell, 2005; Vine et al., 2005). In this context, we also show that ATRA up-regulates Cx-43 protein expression in myometrial cell cultures, although further studies will need to be undertaken to determine whether ATRA influences the phosphorylation of Cx-43 in the myometrium.
In human epithelial cells, ATRA inhibits the binding of activator protein 1 to the cyclic AMP response element in the COX-2 promoter (Subbaramaiah et al., 2002). This inhibits phorbol ester-mediated induction of COX-2 gene transcription, leading to a fall in intracellular COX-2 levels and hence reduced prostaglandin production (Mestre et al., 1997; Subbaramaiah et al., 2002). If this process is replicated in the human myometrium in pregnancy, ATRA could thereby contribute to the maintenance of uterine quiescence. This would be in keeping with the present demonstration that ATRA substantially inhibits COX-2 activation in LPS-treated myometrial cell cultures. In an initial attempt to examine the mechanism for the inhibitory effect of ATRA on COX-2 expression, we carried out additional experiments treating cultured cells with ATRA. The HDAC inhibitor TSA was also added to the cultures because previous studies have shown HDAC inhibitors to have a suppressive effect on COX-2 activation (Tong et al., 2005; Yamaguchi et al., 2005). Yamaguchi et al. (2005) also provided mechanistic evidence indicating that HDAC inhibitors suppress COX-2 transcription as a consequence of suppressing expression, and blocking the function, of the transcription factor, c-Jun. Interestingly, we also show that ATRA had an inhibitory effect on the expression of c-Jun protein in myometrial cell cultures and that the inhibitory effect of ATRA on COX-2 activation was enhanced in the presence of the HDAC inhibitor TSA. Although it remains unclear whether the observed inhibitory effect using ATRA in combination with TSA is simply additive, further studies should be explored to unravel the mechanism(s) of action for both ATRA and HDAC inhibitors in regulating transcription of myometrial retinoic acid response genes. In this context, recent studies in cancer research demonstrate that HDAC inhibitors can revert or enhance changes in gene expression in response to retinoid acid (Emionite et al., 2004; Trus et al., 2005; Wang et al., 2005).
In the present study, we report that CRBPI is down-regulated in both the upper and the lower uterine segments during the first and second trimesters of pregnancy. This mitigates the production of large amounts of ATRA in the cytosol at these times and may therefore be the rate-limiting step in determining the ligand’s availability to the nucleus. If so, the change may be permissive for myometrial cellular proliferation, as ATRA has an anti-proliferative effect in myometrial culture systems. Conversely, it is possible that some ATRA gains access to the nucleus at this time because CRABPII is expressed in both the upper and the lower segments throughout the first two trimesters of pregnancy. If so, its role could be to contribute to the maintenance of uterine quiescence, for example, by inhibiting COX-2 expression as previously outlined. The fall in CRABPI levels in the first and second trimesters may be a consequence of falling levels of ATRA reaching the nuclei of myometrial cells because the expression of CRABPI has been shown in this study to be proportionate to ATRA exposure in cultured myometrium.
By the end of the third trimester, CRBPI is up-regulated in upper segment myometrium. Together with a rise in CRABPII expression at this time and the nuclear localization of the latter, it may be inferred that ATRA is readily available to its myometrial nuclear-binding proteins by the end of pregnancy. The effect may be to prevent further cellular proliferation primarily, possibly in addition to contributing to the inhibition of uterine contractions. The maintained reduction in CRABPI protein levels in upper segment tissue in the third trimester may be important in reducing inactivation of the ligand before its transport to the nucleus. Because labour spans a relatively short space of time, this being hours rather than weeks, the potential anti-proliferative effect of the retinoids is unlikely to be clinically significant at this time. Despite this, clearly defined alterations may be found in cytoplasmic binding protein expression during labour. These may come about either immediately before or during labour, but in either case, a role for the retinoids in the switch from myometrial quiescence to increased contractility is implied. Because CRBPI is down-regulated in labouring myometrium, it is likely that less ATRA is being made at this time in the cells. Furthermore, the proliferation of myometrial cells is inhibited by ATRA in both rat and human culture systems. ATRA may therefore be a pro-quiescent factor whose withdrawal is important either for the initiation or for the maintenance of labour.
Although the level of retinoid-binding proteins found in myometrial tissues is in general mirrored by variations in the expression of its matching mRNA, some exceptions were found. For example, RT–PCR demonstrated spatial variation in the expression of CRABPI mRNA in the first two trimesters of pregnancy despite an apparent absence of protein on western blotting. There are many factors that could account for this disparity, including changes in the stability of mRNA or in an altered efficiency of mRNA translation, either of which could be influenced by a changing myometrial cellular environment as pregnancy progresses. It is also possible that protein levels were too low for detection by the assay systems employed in the present experiments but that more sensitive assay systems could identify spatial and temporal variations in the expression of very low levels of CRABPI protein throughout pregnancy and labour.
In myometrial culture experiments, CRABPI was up-regulated in proportionate response to increasing levels of ATRA. This may prove to be part of a negative feedback system controlling the influence of the retinoid signalling pathway on myometrial differentiation because high levels of CRABPI could promote the catabolism of ATRA. Rising levels of ATRA also increased the expression of the 50-kDa retinoic acid receptor, RARβ, providing the opportunity for the ligand to exert control over the transcription of its target genes.
In summary, these data provide evidence to suggest that the retinoids may be important players in the modulation of myometrial cell function. Because the action of ATRA is mediated through its nuclear receptors (Kastner et al., 1997; Bastien and Rochette-Egly, 2004), the expression of RARs and RXRs will now be characterized in both the upper and the lower uterine segments throughout pregnancy and labour. In additional experiments, the ability of myometrial cells to synthesis ATRA from retinol and then to metabolize ATRA to its polar metabolites will be investigated. The interaction between the retinoid signalling pathway and other determinants of myometrial differentiation will also be explored because, for example, estrogens are known to promote the synthesis and secretion of ATRA in other tissue types. Finally, future studies will be employed to unravel the molecular mechanisms by which ATRA elicits both its stimulatory and inhibitory effects on retinoic acid response genes associated with uterine activity.
1School of Surgical and Reproductive Sciences, The Medical School, University of Newcastle upon Tyne and 2Women’s Services, Newcastle upon Tyne Hospitals NHS Trust, Royal Victoria Infirmary, Newcastle upon Tyne, UK