| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Laboratory of Molecular Endocrinology, McGill University Health Centre, Royal Victoria Hospital, Montréal, Québec, Canada H3A 1A1
Address all correspondence and requests for reprints to: Hans Zingg, Laboratory of Molecular Endocrinology, Royal Victoria Hospital, Room H7.64, 687 Pine Avenue West, Montréal, Québec, Canada H3A 1A1. E-mail: hans.zingg{at}mcgill.ca.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
B (RANK) and its ligand, as well as the involvement of novel signaling molecules such as estrogen-responsive gene 1, decay-accelerating factor 1, and ebnerin. The present results provide the basis for further studies that will enlarge our knowledge of the mechanisms underlying labor and parturition under physiological and pathophysiological conditions. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
The process of parturition has been divided into four different phases (1). During most of the course of pregnancy, the uterus is kept in a state of quiescence due to the repression of uterotonic signals and the activation of uterolytic signals as well as due to the relative unresponsiveness of the uterus to uterotonic stimulation. This stage is referred to as phase 0. As term approaches, the uterus enters phase 1 (activation phase), which is characterized by an increase in its capacity to respond efficiently to uterotonic signals. In phase 2, (stimulation phase), uterotonic signals induce coordinated uterine contractions that lead to birth. The ensuing expulsion of the placenta and the uterine involution process form phase 3 of parturition.
A limited number of signals that play a role in the control of parturition have so far been identified. Inducers of myometrial relaxation include progesterone, relaxin, prostacyclin PGI2, nitric oxide, PTH-related peptide, CRH, and LH (1, 3). Only two endogenous physiological activators have so far been identified with certainty, namely prostaglandins and oxytocin. In addition, uterine activation is characterized by the up-regulation of a cassette of contraction-associated proteins, which include connexin-43, the oxytocin receptor, certain prostaglandin receptors (foremost the prostaglandin F receptor), and certain ion channels (1).
We hypothesized that the different phases of uterine activation, stimulation, and involution are accompanied each by a specific pattern of gene expression, and that the systematic large scale elucidation of uterine gene expression patterns during parturition will provide the basis for a broadening of our understanding of this process at a molecular level. As a long-term aim, this approach will provide potential targets for the design of novel diagnostic and predictive tests as well as for effective pharmacotherapies for the treatment of preterm labor. We present here our first result from the implementation of this approach using the rat as an animal model and Affymetrix rat genome U34A DNA microarrays in conjunction with conventional Northern blotting and real-time RT-PCR.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
RNA extraction
Uteri were opened by longitudinal incision on the mesometrial side, and fetuses and placentas were removed. To avoid any contamination with fetal or placental tissue components, the placental insertion sites were also removed from the uteri. No attempts were made to mechanically separate myometrium from endometrium. For each time point, a single pool containing the uteri from at least four animals was established. The pooled tissues were frozen in liquid nitrogen pending further analysis. Two independent experiments A and B were performed. RNA was extracted and purified using the TRIzol reagent (Invitrogen-Life Technologies, Inc., Gaithersburg, MD) according to the manufacturer’s instructions. The quality of the total RNA was assessed by determining the 260/280 nm absorbance ratios and by gel electrophoresis in agarose/formaldehyde gels. RT, second-strand synthesis, and cRNA labeling were all accomplished by the standard Affymetrix protocol (Affymetrix, Santa Clara, CA), and biotinylated cRNA was hybridized to rat Genome U34A GeneChips (Affymetrix) on an Affymetrix Fluidics Station at the McGill Genome Centre. Fluorescence labeling, confocal scanning and data generation for analysis was performed by the same facility. Two independent experiments A and B were performed.
Elimination of genes expressed below detection limits
In the Affymetrix chip used, each gene is represented by an array of sixteen different oligonucleotides, paralleled by an array of negative control oligonucleotides that differ from the probes by one point mutation. Based on a proprietary algorithm developed by Affymetrix, analysis of the hybridization signal intensity over the 32 pixels results for each gene Gi in an expression intensity value gi as well as a call of either present, marginal, or absent. Genes that had an absent call at all time points in both experiments were eliminated from further analysis.
Characterization of experimental scatter
As illustrated in Fig. 1A
on the example of the time point d 0, analysis of scatter plots of two independent but methodologically identical experiments, A and B, revealed that the relative scatter of experimentally determined gene expression intensity values in two different experiments is the function of the expression intensity value itself and increases with decreasing levels of gene expression. It is therefore inappropriate to judge the significance of any observed change in gene expression solely on the level of the observed-fold difference. The shape of the scatter was characterized by a pair of lines that enclosed an area containing 95% of all points on the scatter plot diagram. These lines were defined by the equations 
and 
for the upper and lower borders, respectively. In double logarithmic representation, as in Fig. 1, the boundaries take the shape of an oblique inverted funnel; in linear representation (not shown), they represent straight lines, with a and 1/a representing the slopes of the upper and lower boundaries, respectively, and b the intercepts of the upper and lower boundary lines with the y- and x-axes, respectively. For the determination, of parameters a and b, the totality of all 43,700 data points in experiment A were plotted against those of the identical but independent experiment B. Parameters were determined such that the following two conditions were satisfied: 1) the area defined by the boundary lines encompassed 95% of all points; and 2) outliers were distributed at similar densities along the boundary lines. The parameter values that satisfied both these criteria were a = 1.28 and b = 135.
|
, B and C. It follows from the above that, if a given gene Gi was represented by a scatter point outside the above-defined 95th percentile interval, the chances that the difference in the expression values for gene Gi between the time points t1 and t2 was solely due to experimental error was less than 5%, and there was a more than 95% probability that it represented a gene with a true difference in its expression value between the two given time points. If a gene Gi exhibited a scatter point outside the 95th percentile interval in both independent experiments A and B, the probability that the difference was due to experimental error was even lower, i.e. less than 5% of 5%, corresponding to less than 0.25%. Genes exhibiting a difference in the expression levels that satisfied the aforementioned criteria in both experiments A and B between D0 and at least one other time point were pooled for further analysis. Implementation of the above algorithms and production of scatter plots was performed using Excel Microsoft (Redmond, WA). Data were further analyzed by cluster analysis using Gene-Cluster (4). Following log transformation, normalization and mean centering, data were analyzed by K Means Clustering (9 nodes, maximum 100 iterations) and self-organizing maps (SOM), with the parameters: x-dimension = 1, y-dimension = 10, number of iterations = 100,000. Gene patterns were visualized using the program TreeView (4).
Northern blot
Ten micrograms of total RNA were electrophoretically separated on a 1.2% agarose gel containing 1.5 M formaldehyde and transferred to nylon membrane (Hybond N, Amersham Pharmacia Biotech, Arlington Heights, IL). Blots were prehybridized with QuickHyb solution (Stratagene, La Jolla, CA) for 1 h at 65 C and hybridized at 65 C for 1 h in the same solution containing 2 x 106 cpm/ml of a 32P-labeled specific PCR fragment corresponding to the gene of interest. Nonspecific binding was eliminated by one wash at room temperature for 30 min in 2x standard sodium citrate containing 0.1% sodium dodecyl sulfate, followed by two washes at 65 C in 0.1x standard sodium citrate/0.1% sodium dodecyl sulfate. The membranes were exposed to a phosphorimager (Typhoon, Amersham Pharmacia Biotech) and the specific signals quantitated using ImageQuant for Mac version 1.2.
Two step kinetic RT-PCR
Total RNA extracted from uteri at time points D0, D20, NL, L, and PP was treated with ribonuclease-free deoxyribonuclease (Ambion, Inc., Austin, TX). Eight micrograms of each sample was reverse transcribed using random hexanucleotides (Amersham Pharmacia Biotech) and Moloney murine leukemia virus reverse transcriptase (Invitrogen). Control reactions without reverse transcriptase were set up in parallel. The samples of the two independent experiments A and B were processed separately. For real-time quantitative PCR, equal amounts of cDNA were added to 20 µl of FastStart SYBR green reaction mix (Roche Molecular Biochemicals, Indianapolis, IN) containing 0.4 µM of gene specific oligonucleotide primers (for oligonucleotide sequences, see the supplemental data published on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org). For each sample, a parallel reaction was set up with 18S RNA specific primers. 18S RNA was chosen as endogenous control because of its invariant expression across tissues. Standard curves were generated simultaneously for both the target and the endogenous control gene using serial dilutions of one of the cDNA samples. The reactions were run in a Light Cycler (Roche Molecular Biochemicals) with cycling profile as follows: denaturation/activation at 95 C for 10 min, followed by 35–50 cycles of denaturation at 95 C for 10 sec, annealing at 57–60 C for 5 sec, elongation at 72 C for 10 sec. The presence of a specific and unique PCR product for each pair of gene-specific primers was verified by a Light Cycler-generated melting curve profile and by conventional agarose gel electrophoresis. Relative quantification of initial amounts (in arbitrary units) of target and endogenous control were extrapolated from the respective standard curves using the Light Cycler second derivative maximum algorithm. For each time point, the target gene expression values were normalized to the respective 18S values.
To assess the interassay coefficient of variation, we used luciferase cRNA as an internal control to spike two RNA samples (D0 and L). We produced three separate cDNA reactions from each RNA sample and performed real-time PCR on each cDNA using luciferase-specific oligonucleotides. The amount of luciferase RNA added was comparable to the levels of endogenous 18S RNA. The results obtained from four independent experiments were compared. The maximum interassay coefficient of variation measured was 2%, well below the threshold of expression differences we set to measure (50%).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Point-to-point comparisons
We next analyzed differences in uterine gene expression pair wise between different time points using the approach described in Materials and Methods. Gene expression levels were compared between d 0 and each time point during pregnancy as well as between each consecutive time points. For each comparison, only genes that corresponded to scatter plot points outside the 95th percentile interval in both experiments A and B were considered for further analysis. Representative examples of scatter plots are shown in Fig. 1. Two scatter plots each comparing two different time points from experiment A are shown, including a comparison between d 0 and 20 (Fig. 1B) and a comparison between labor and nonlabor at d 23 (Fig. 1C). The scatter plot shown in Fig. 1A illustrates, on the other hand, the scatter resulting from technical and biological variability by comparing the expression values obtained for the same time point (d 0) in two independent experiments A and B.
Of the 8740 genes represented on the microarray, 622 genes or 7.1% passed our criteria for a significant and consistent expression change at least one of the time points under study. The actual number of genes undergoing biologically relevant changes is in all likelihood larger but certainly not smaller. Due to existing detection limits of the DNA microarray hybridization procedure as well as due to inherent technical and biological margins of error, we had adopted stringent criteria for the definition of a significant and consistent change in expression levels, thus limiting the number of genes considered for analysis. For example, the expression values of genes encoding the oxytocin receptor, connexin-43, relaxin, PTHrP, and cyclooxygenases 1 and 2, all genes known to be up-regulated at term, did all reach a maximum at d 23 (NL or L), but due to the low overall expression values, these differences did not meet the criteria for statistical significance established here.
Results from the quantitative analysis are depicted in Fig. 2. As expected, the uterine expression of an appreciable number of genes (249) were increased during pregnancy with a somewhat lesser number (112) being significantly decreased (Figs. 1B and 2). Progression from d 20 (D20) to d 23 (NL) was accompanied by the recruitment of a limited number (54) of additional genes. These included mainly genes encoding extracellular matrix and cytoskeletal products (together 42% of the known genes in this group). However, as the uterus progressed at term from NL to L, we found that the extent of gene suppression markedly outweighed the extent of gene activation, as evident from inspection of Fig. 1C and confirmed by quantitative analysis in Fig. 2. This implies that, for the induction of active labor, uterine gene suppression is equally if not more important than the more extensively studied mechanisms of gene activation.
|
; a full listing is given in the supplemental data, and an analysis of the possible biological implication and significance of expression changes of selected genes is presented in the Discussion.
|
). Whereas K means clustering results in the generation of a predetermined number of distinct clusters, the ordering by the generation of a SOM results in a continuous array. As depicted in Fig. 3, A and B, the two different analyses yielded very comparable results. The results of the clustering procedures give a vivid graphical picture of the dynamic and diverse expression changes that the genome undergoes in the uterus during the periparturient period.
|
lists for each gene its membership to a given cluster, as defined in Fig. 3B. A complete list of all genes belonging to each cluster is given in the supplemental data. Clusters 1–3 represent genes that are overall up-regulated during gestation, cluster 4 contains genes that undergo little change during gestation but are suppressed PP, and cluster 5 contains genes that are globally suppressed during gestation. Within the first group, cluster 1 contains the genes that are further increased at labor, whereas cluster 3 contains a group of genes that are specifically suppressed with the onset of labor. Clusters 2 and 5 can be further subdivided with respect to regulation PP, where genes are either suppressed (clusters 2b and 5a) or up-regulated or kept constant (clusters 2a and 5b).
Genes in each cluster were classified according to proposed or demonstrated functions (Fig. 4). In cluster 5, we observed a striking abundance of ribosomal protein genes (30%, with 29 ribosomal genes clustered in close succession). These genes were highly clustered in the area of cluster 5a. The significance of this phenomenon is unclear, but it is known that certain ribosomal genes subserve additional functions. For example, ribosomal proteins L29, P2, and S25 have been proposed to play a role as cell surface molecules and have each been shown to be specifically expressed in the uterus (5, 6). Cluster 1 represents the gene pool activated specifically at labor, containing the highest abundance of immune defense and inflammation-related genes (12%), consistent with the idea that the parturition process involves activation of pathways related to inflammation and immune defense. On the other hand, cluster 3, representing genes suppressed at labor, contained a high abundance of extracellular matrix-related genes (20%), supporting the idea that changes in extra cellular matrix configuration and cell-cell interactions may represent a so far understudied component associated with the mechanisms of parturition. Cluster 2, representing genes that are up-regulated throughout pregnancy, contained the highest abundance of genes linked to metabolism (10%) and intracellular trafficking (11%), again in accordance with the increase overall activity of the uterus during the activation and stimulation phases of parturition. Finally, in cluster 4, genes related to cytoskeleton and cellular motility were most abundant (29%), perhaps reflecting the changes in cytoskeletal architecture concomitant with PP involution.
|
Verification of microarray data
To validate further the expression, patterns detected by the microarray approach, expression levels of selected genes representing distinct expression patterns were assessed with another hybridization-based method, Northern blotting, and with an amplification-based method, two-step real-time RT-PCR (Fig. 5). The genes chosen for the validation were: decay-accelerating factor 1 (DAF1; AF039583, cluster 1), IGF binding protein (IGFBP) 2 (M91595, cluster 2), osteopontin (OPN; M14656, cluster 3), frizzled-related protein (FrpAP; AF012891, cluster 4), osteoprotegerin (OPG; U94330, cluster 4), estrogen-responsive gene 1 (ERG1; AF022147, cluster 5), ebnerin (U32681, cluster 5), IGFBP-5 (AI029920, cluster 5). For some genes (FrpAP, IGFBP-5, ERG1) both Northern blotting and real-time PCR were used in the validation. IGFBP-2 expression was validated only by Northern analysis and OPN, OPG, ebnerin, and DAF1 were validated only by real-time RT-PCR. In all cases, excellent agreement was found in the expression trends identified by all three methods (Fig. 5).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Genes up-regulated at d 20
These genes belong predominantly to clusters 2 and 3 and thus represent either genes that are expressed constantly throughout pregnancy and parturition (cluster 2) or genes that are highly expressed up to d 23 and then fall abruptly with the initiation of labor (cluster 3). As expected, this group includes genes that are well known to be expressed in the uterus close to term. These included oxytocin (7), fibronectin (8), IL-1, OPN (9), P-glycoprotein (10), and IGFBP-2 (11).
Among the novel genes, we identified uroguanylin as the gene with the highest induction rate (>70-fold; Table 1A). Uroguanylin (also called guanyl cyclase activator IIB) has not been previously known to be expressed in the uterus. It is a typical representative of a cluster 3 type gene: following its dramatic induction during pregnancy, it undergoes an equally drastic 20-fold decrease before parturition (Table 1C). This relatively recently discovered 15-residue peptide increases intracellular cyclic GMP (cGMP) via binding to, and activation of, particulate guanyline cyclase as well as via a separate pathway involving a pertussis toxin-sensitive G protein (12). cGMP relaxes smooth muscle cells by inhibiting calcium entry. Interestingly, myometrial cGMP rises several hundred-fold during pregnancy, only to decline abruptly before the onset of labor (13). It is therefore tempting to speculate that uterine uroguanyline is part of an intrauterine system that maintains uterine quiescence during pregnancy, and that the acute decrease in its expression contributes to the events of successful labor and parturition.
As shown in Table 1A, the gene encoding the extracellular matrix protein OPN underwent the second highest induction of all the genes assessed between d 0 and 20. The pattern of expression also conformed to cluster 3, with a marked decrease at labor. This expression pattern was verified independently by real-time RT-PCR (Fig. 5E). OPN has been known to be expressed in early pregnancy in the uterine endometrium and, as a ligand of several integrins, it is thought to play a role in blastocyst adhesion and implantation (14, 15). We now extend these findings and show here that OPN continues to be highly expressed in late pregnancy but then decreases rapidly at labor, a feature that may be compatible with a role of OPN in attachment of the placenta during gestation and its release at labor.
OPN is a representative of a series of additional cell adhesion molecules that undergo marked changes in their expression pattern. These include fibronectin-1 (28-fold), thrombospondin-4 (24-fold), connexin-26, fibronectin-3, laminin 
-1, tropoelastin, and biglycan, with the first three exhibiting the cluster-3 expression pattern. Concomitantly, the extracellular matrix-degrading enzymes matrilysin [matrix metalloproteinase (MMP)-7] and stromelysin (MMP-3) are strongly repressed. In fact, matrilysin is the gene that undergoes the highest level of suppression between d 0 and 20 (Table 1A) only to be stimulated again 3-fold at labor (Table 1C). It equally fits this concept that the metalloprotease inhibitors, tissue inhibitor of metalloproteinase 2 (TIMP-2) and 
2-macroglobulin, are concomitantly up-regulated. Concurrently, enzymes involved in extracellular matrix molecule biosynthesis such as calnexin, farnesyl diphosphate synthase, and calreticulin are also up-regulated (Table 1A). These data further support the established concept that the coordinated temporal expression pattern of MMP genes and their tissue inhibitors are important for extracellular matrix remodeling in uterus, much as in many other tissues undergoing dynamic architectural alterations (16).
11ß-Hydroxysteroid dehydrogenase1 [11ß-HSD1; represented on the array as an expressed sequence tag (EST)] is found in fourth position on Table 1A. As the three preceding genes on the list, its expression profile conforms to cluster 3, with a strong increase during gestation up to d 23, followed by a decrease at labor (Table 1C). 11ß-HSD1 modulates cortisol availability by converting biologically active cortisol into inactive cortisone (oxidase activity) or vice versa (reductase activity). The enzyme is known to be expressed in human fetal membranes (17) and in rat myometrium at term before labor (18). The present data indicate for the first time that labor is associated with a rapid fall in 11ß-HSD1 expression (Table 1C). The precise role of the observed pattern of uterine 11ß-HSD1 expression and its potential significance for the triggering of labor now requests further investigation.
Several enzymes involved in vesicle transport were also up-regulated, such as caveolin (15-fold), rab-7 (7-fold), rab GDP dissociation inhibitor (GDI) ß, clathrin-associated protein 17, and ADP ribosylation factor-1 (ARF-1) and -6, reflecting an important but perhaps overlooked role of cellular trafficking mechanisms in the final stages of pregnancy. Caveolin expression in the uterus has recently been shown to be hormonally regulated (19), and caveolin has a role in the regulation of the activity of cell surface signaling molecules, such as the oxytocin receptor (20). Rab7 controls trafficking through late endosomes to lysosomes and its role, along with the other rab proteins, in the uterus is yet unexplored.
IGFBP-2 was up-regulated 8-fold and stayed high throughout pregnancy. This is in stark contrast to IGFBP-5, which was suppressed throughout pregnancy and underwent a dramatic rebound PP. As shown in Fig. 5, B and D, these two differing expression profiles have been independently confirmed using both real-time RT-PCR and Northern blot analysis. IGFBP-3 was only expressed at low levels and IGFBP-6 expression remained constant throughout late pregnancy and decreased PP. The IGF system has been studied extensively for its role in endometrial activity (21), uterine differentiation (22), implantation (23), and early pregnancy (24), but the role, if any, of the six IGFBPs at term remains uncharacterized. It is likely that growth changes affecting the uterus before and after labor are the result of differential local actions of IGF and the modulating influences of IGFBPs. Previous studies in the rat have shown that IGFBP-2 and IGFBP-5 are localized in distinct regions of the uterus (25, 26, 27, 28). In the human uterus, IGFBP-2 has been shown to increase at active labor (29), in agreement with our data. IGFBP-5 expression was shown to increase during uterine involution mediated by IGF and estrogen (30), again consistent with our finding of elevated expression of IGFBP-5 only in the PP period. Thus, the fact that the IGFBP profile we obtained is in overall agreement with previous reports constitutes a further validation of our analysis.
The last two genes that we wish to discuss in this series are FrpAP and OPG. These are both called decoy receptors, i.e. soluble forms of receptors that inhibit ligand action by sequestering ligand and their expression in the uterus has so far not been described. As shown in Fig. 5, A and F, their expression pattern has been independently verified by real-time RT-PCR (FrpAP and OPG) as well as by Northern blotting (FrpAP). Although they were both up- regulated during pregnancy, they reached a maximum either before labor (FrpAP) or at labor (OPG). At d 20, FrpAP gene expression was increased 6-fold over d 0, and it stayed elevated for the rest of pregnancy and parturition. FrpAPs are secreted proteins that bind Wnt, the ligand of frizzled, and thus antagonize Wnt signaling. Using real-time RT-PCR, we observed that frizzled receptor-1 was down-regulated at late pregnancy and labor (data not shown). Together with the elevated levels of FrpAP, the data indicate that the Wnt signaling pathway is down-regulated at term. Secreted FrpAPs are thought to be modulators of apoptosis in certain systems (31, 32) by modulating the Wnt pathway (33). The Wnt pathway is also of known importance for uterine development (34), and the uterine expression of certain modulators of Wnt signaling have recently been shown to be highly regulated at the time of implantation (15). Our findings suggest that this pathway subserves further roles later during pregnancy and that down-regulation of this pathway at term by FrpAP is of physiological importance.
OPG is a soluble member of the TNF- receptor family that acts as a negative regulator for receptor activator of nuclear factor-B (RANK) by neutralizing the effects of RANK ligand (RANKL). The RANK/RANKL pathway is one of the main regulators of bone resorption (35). OPG expression in transgenic mice blocks the last stages of osteoclast differentiation, inducing bone osteopetrosis, whereas OPG knockout mice develop osteoporosis (36). Although the RANK/RANKL system has been found to influence mammary gland development and differentiation at pregnancy (37), its role in the uterus has never been investigated. We report here elevated OPG expression at late pregnancy with a peak in labor. These findings suggest a role for the RANKL/RANK system in the uterus at parturition and should prompt further investigations into the role of the RANK/RANKL system and its negative regulator OPG in the uterus.
Genes suppressed at d 20
Next in the list is ebnerin, a gene strongly suppressed at d 20 with continuously low expression throughout term and labor. We have validated this expression pattern independently by real-time RT-PCR (Fig. 5H). Ebnerin (also called CRP-ductin) was cloned originally from the lingual salivary von Ebner’s gland and localized to secretory duct epithelial cells (38). Subsequently it was shown that ebnerin, hensin, and deleted in malignant brain tumor-1 are alternately spliced products from the same gene. Hensin has been implicated in the determination of epithelial cell polarity and differentiation (39). None of these splice products had so far been demonstrated in the uterus. Although the precise function of ebnerin remains to be defined, certain structural features of the molecule suggest that it may be a cell-surface molecules and posses multiligand binding capacity (40).
Another strongly suppressed gene is ERG1, a protein that exhibits amino acid similarity to ebnerin and bone morphogenic protein (41). In contrast to ebnerin, however, ERG1 expression rose again during pregnancy and reached its highest expression level at the time of labor. We have confirmed this expression profile by real-time RT-PCR as well as by Northern blotting (Fig. 5). ERG1 has been cloned independently by two different groups and is referred to as either uterus-ovary specific putative transmembrane protein (UO, unpublished, accession no. AF022147) or ERG1 (41). ERG1 has been identified as an estrogen-induced transmembrane protein of unknown function, highly expressed in the rat oviduct and the endometrial epithelium, but acutely down-regulated at the beginning of pregnancy (41). Here we show for the first time that ERG1 expression increases a dramatic 12-fold in the uterus a few hours before labor and increases further at labor. This suggests a previously unrecognized role of ERG1 during parturition and the active phase of labor in the rat. ERG1 also represents one of several examples of genes that are suppressed at the onset of pregnancy during the implantation phase but which strongly rise at term. Other examples identified here include OPN, apolipoprotein (Apo) E, matrilysin, and the FrpAPs. Thus it appears that, with respect to the expression of certain genes, the parturition phase represents a mirror image of the implantation phase.
Genes up-regulated at term
Considering the list of genes that undergo significant up-regulation between d 20 and 23 (Table 1B), we find at the top of the list -fetoprotein (AFP), followed by a hitherto unknown AFP-like gene, represented on the array as an EST. Elevated circulating levels of AFP represent a well-known marker for threatening preterm labor. As confirmed by a recent NIH study, elevated AFP levels are significantly associated with subsequent spontaneous preterm birth in otherwise asymptomatic pregnant women at 24 and 28 wk (42). At 24 wk, the odds ratio for spontaneous preterm birth at less than 32 wk was 8.3. We now show here that AFP is produced in the rat uterus, and its expression undergoes an 80-fold increase within the last 3 d of pregnancy, only to be acutely suppressed again at labor. Thus, uterine AFP expression appears to be part of the physiological progress toward normal birth, and this expression program may be accelerated in the context of preterm birth. The role of AFP is, however, unknown. It acts as regulator of hormonally induced growth. AFP binds estrogen and suppresses the uterotropic response to estrogen (43), and a peptide derived from AFP prevents the growth of estrogen-dependent human breast cancers (44). AFP is also produced by a number of malignant uterine tumors (45). Interestingly, AFP is strongly suppressed at the moment of active labor. In fact, as shown in Table 1C, in our study AFP is the gene that undergoes the strongest degree of suppression at labor (L vs. NL). It is possible to speculate that the acute suppression of AFP production at labor removes the estrogen-antagonistic influence and thus may contribute to a local increase in the estrogen/progesterone ratio.
ApoA1, represented under two different accession numbers) figures next on the list of genes up-regulated between d 20 and 23 (Table 1B). ApoA1 has been detected as an abundant-pregnancy associated protein in ovine uterine luminal fluid (46). The present study now shows that ApoA1 is expressed in the rat uterus and that its expression undergoes, like AFP, a dramatic more than 20-fold increase within the last 3 d of pregnancy, followed by an acute suppression at labor (Table 1C; cluster 3). The role of ApoA1 in the uterus is still unclear. Apos are known to be involved in cholesterol transport and thus perhaps indirectly in steroid biosynthesis. ApoA1 has been specifically identified as a factor that binds to elastin and mediates tethering of elastin to the extracellular matrix of vascular smooth muscle cells, thereby activating intracellular tyrosine kinase activity (47). Whether ApoA1 forms part of an intrauterine system mediating elastin/myometrium interactions represents a speculative but testable hypothesis.
Matrilysin (metalloproteinase-7) was suppressed at d 20 but up-regulated 14-fold between d 20 and 23. Whereas the suppression of matrilysin during gestation has been observed previously (15, 48), the acute up-regulation at term and the further 3-fold up-regulation at labor (Table 1C) described here is original and may be an important part of the physiological and possibly pathophysiological parturition mechanism. Indeed, increased matrilysin content in amniotic fluid has been reported in association with preterm labor (49).
The expression of the adult skeletal muscle gene 15 is 11-fold increased in the d 20/23 time span, but followed by an acute 30-fold suppression at labor. Adult skeletal muscle gene 15 is the rat homolog of mouse H19, a tumor suppressor with a proposed role in muscle differentiation (50). This gene has been shown to be expressed in the mouse uterus and to be induced by estrogens (51). However, its pattern of acute expression at d 23 and suppression at labor as described here is novel and points to a possible unrecognized function in the uterine activation phase.
IGF-II follows exactly the same pattern of expression as H19, with an 11-fold up-regulation between d 20 and 23 and a 30-fold down-regulation at labor. It is an interesting coincidence that, in the human, these coordinately regulated genes are reciprocally imprinted and located in close proximity on chromosome 11p15.5, a region subject to loss of heterozygosity. Loss of heterozygosity or loss of imprinting of the H19 and IGF2 genes has been shown to occur in 58% of cervical cancers (52), suggesting a coordinate role of these genes in the control of growth and differentiation.
There are only a few selected genes that follow the very characteristic expression pattern of IGF-II and H19, i.e. a peak of high expression immediately before labor at d 23 (NL) and low expression at d 20 and at labor. These genes have been grouped in a small but distinct cluster by the K-tuple clustering algorithm (cluster I in Fig. 3, bottom left), and include, in addition to IGFII and H19, also ApoA1, AFP, and three ESTs: EST189329, EST220183, and EST197895. It remains to be determined, whether these genes share additional common control mechanisms and/or common functions.
Genes undergoing dynamic changes at parturition
The discussion now leads us to the small but interesting group of genes that remain suppressed until term and undergo significant up-regulation with the onset of labor (L vs. NL, Table 1C). This group includes several immediate early response genes, such as c-fos, the immediate-early serum-responsive JE gene, an EST closely related to the chemo-attractant macrophage inflammatory protein-2 (Ref.53 ; daf-1 or CD55; Fig. 5), as well as the transcription factors NGFI-A (also known as Egr1, Zif268, Krox 24, and Tis8; Ref.54) and NGFI-B (also called nurr-77 or TIS-1). Daf-1 or CD55 is an estrogen- and cytokine-induced TGFß type I receptor-like membrane glycoprotein that protects cells from complement-mediated attack and mediates so far incompletely understood signaling functions (55). The two transcription factors NGFI-A and -B are key regulators of growth and differentiation signals and are induced by a variety of stimuli, including stretch, hypoxia, and cytokines (54). They are both known to be induced transiently in the rat uterus 30–120 min after estrogen administration (56). Thus, their rapid induction at labor, as shown here, may be triggered by stretch as well as by the shift in the progesterone/estrogen ratio. NGFI-B or nurr77 is stimulated by PGF2 and is an activator of 20 hydroxysteroid dehydrogenase (20HSD), an enzyme that metabolizes progesterone. This chain of events operates in the mouse corpus luteum and forms part of the mechanism inducing the decrease in serum progesterone which, in turn, triggers parturition in the mouse (57). Whether a similar mechanisms operates locally in the uterus is suggested by the present data but remains to be determined. The very acute up-regulation of these two factors observed here establishes them as potential key regulators of the gene expression program that initiates labor.
Previous studies
Chan et al. (58) identified six novel up-regulated genes associated with human labor. We found that for five of them (elongation factor 1, -actin, manganese superoxide dismutase, cyclophilin, and nucleophosmin) the rat homolog was also moderately up-regulated (1.4- to 2.9-fold). Aguan et al. (29) identified 12 genes that increased at human labor. We found that for four of them the rat homolog was also moderately increased with angiotensin converting enzyme being induced the highest (1.4-fold). Among the nine genes listed as decreased at labor, we strongly confirmed the decrease of IGF-II, although we also noted that IGF-II undergoes a strong increase between d 20 and 23 just before labor.
|
Conclusion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Current address: National Jewish Hospital, Department of Pediatrics, Room K906, 1400 Jackson Street, Denver, Colorado 80206. 
Abbreviations: AFP, -Fetoprotein; Apo, apolipoprotein; DAF1, decay-accelerating factor 1; ERG1, estrogen-regulated gene-1; EST, expressed sequence tag; FrpAP, frizzled-related protein; 11ß-HSD1, 11ß-hydroxysteroid dehydrogenase 1; IGFBP, IGF binding protein; L, labor; MMP, matrix metalloproteinase; NL, nonlabor; OPG, osteoprotegerin; OPN, osteopontin; PP, post partum; RANK, receptor activator of nuclear factor-B; RANKL, RANK ligand; SOM, self-organizing maps.
Received December 26, 2002.
Accepted for publication February 6, 2003.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
- fetoprotein, plasma corticotropin-releasing hormone, and other serum markers for the prediction of spontaneous preterm birth. Am J Obstet Gynecol 186:990–996[CrossRef][Medline]
-fetoprotein suppresses the uterotropic response to estrogens. Proc Natl Acad Sci USA 80:2733–2737-fetoprotein prevents the growth of estrogen-dependent human breast cancers sensitive and resistant to tamoxifen. Proc Natl Acad Sci USA 99:2211–2215-Fetoprotein production by a malignant mixed Mullerian tumour of the uterus. J Clin Pathol 49:349–351 chemoattractant for human neutrophils and dendritic cells. J Immunol 165:2588–2595-induced expression of 20-hydroxysteroid dehydrogenase involves the transcription factor NUR77. J Biol Chem 275:37202–37211This article has been cited by other articles:
 |
|
 |
|
  G. Helguera, M. Eghbali, D. Sforza, T. Y. Minosyan, L. Toro, and E. Stefani Changes in global gene expression in rat myometrium in transition from late pregnancy to parturition Physiol Genomics, January 8, 2009; 36(2): 89 - 97. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. T. Chiquet, S. H. Blanton, A. Burt, D. Ma, S. Stal, J. B. Mulliken, and J. T. Hecht Variation in WNT genes is associated with non-syndromic cleft lip with or without cleft palate Hum. Mol. Genet., July 15, 2008; 17(14): 2212 - 2218. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Sitras, R.H. Paulssen, H. Gronaas, A. Vartun, and G. Acharya Gene expression profile in labouring and non-labouring human placenta near term Mol. Hum. Reprod., January 1, 2008; 14(1): 61 - 65. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Chevillard, A. Derjuga, D. Devost, H. H Zingg, and V. Blank Identification of interleukin-1{beta} regulated genes in uterine smooth muscle cells Reproduction, December 1, 2007; 134(6): 811 - 822. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B E Cross, H M O'Dea, and D J MacPhee Expression of small heat shock-related protein 20 (HSP20) in rat myometrium is markedly decreased during late pregnancy and labour Reproduction, April 1, 2007; 133(4): 807 - 817. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Bdolah, A. Segal, P. Tanksale, S. A. Karumanchi, and S. H. Lecker Atrophy-related ubiquitin ligases atrogin-1 and MuRF-1 are associated with uterine smooth muscle involution in the postpartum period Am J Physiol Regulatory Integrative Comp Physiol, February 1, 2007; 292(2): R971 - R976. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Breuiller-Fouche and G. Germain Gene and protein expression in the myometrium in pregnancy and labor. Reproduction, May 1, 2006; 131(5): 837 - 850. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. J. White, J. W. Ross, M. M. Joyce, R. D. Geisert, R. C. Burghardt, and G. A. Johnson Steroid Regulation of Cell Specific Secreted Phosphoprotein 1 (Osteopontin) Expression in the Pregnant Porcine Uterus Biol Reprod, December 1, 2005; 73(6): 1294 - 1301. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Kwekel, L. D. Burgoon, J. W. Burt, J. R. Harkema, and T. R. Zacharewski A cross-species analysis of the rodent uterotrophic program: elucidation of conserved responses and targets of estrogen signaling Physiol Genomics, November 17, 2005; 23(3): 327 - 342. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Hirsch and H. Wang The Molecular Pathophysiology of Bacterially Induced Preterm Labor: Insights From the Murine Model Reproductive Sciences, April 1, 2005; 12(3): 145 - 155. [Abstract] [PDF]
|
|
|
|
|
|
J. C. Havelock, P. Keller, N. Muleba, B. A. Mayhew, B. M. Casey, W. E. Rainey, and R. A. Word Human Myometrial Gene Expression Before and During Parturition Biol Reprod, March 1, 2005; 72(3): 707 - 719. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Dogan, D. A. Deshpande, M. S. Kannan, and T. F. Walseth Changes in CD38 Expression and ADP-Ribosyl Cyclase Activity in Rat Myometrium During Pregnancy: Influence of Sex Steroid Hormones Biol Reprod, July 1, 2004; 71(1): 97 - 103. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. L. O. Buxton Regulation of Uterine Function: a Biochemical Conundrum in the Regulation of Smooth Muscle Relaxation Mol. Pharmacol., May 1, 2004; 65(5): 1051 - 1059. [Abstract] [Full Text]
|
|
|
|
|
|
G. A. Johnson, R. C. Burghardt, F. W. Bazer, and T. E. Spencer Osteopontin: Roles in Implantation and Placentation Biol Reprod, November 1, 2003; 69(5): 1458 - 1471. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Bazer The Endocrinology and Physiology of Parturition: Understanding the Process through Mining Genome Databases Endocrinology, June 1, 2003; 144(6): 2253 - 2253. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
