Background
To start a pregnancy, the endometrium must be in a receptive state for embryo implantation. In women, the endometrium undergoes extensive vascular growth, remodeling, and regression on an approximately monthly basis [
1]. During the proliferative phase of the cycle, the endometrium and its associated vasculature undergo a period of rapid growth, with subsequent maturation occurring in the secretory phase [
2]. Failure of the endometrium to appropriately remodel and vascularize can result in miscarriage and other pregnancy complications such as fetal growth restriction [
3,
4].
Growth, maturation and remodeling of the endometrium and its blood vessels are primarily regulated by the ovarian steroid hormones estrogen and progesterone [
5,
6]. In mice, endometrial endothelial cell proliferation increases prior to embryo implantation in response to increasing plasma progesterone concentrations. Progesterone treatment in ovariectomized mice stimulates angiogenesis; estrogen priming moderates this progesterone-induced endothelial cell proliferation [
6]. The peptide hormone relaxin also contributes to remodeling in early pregnancy in primates [
7]. Although primarily ovarian in origin, relaxin is also produced by other tissues during pregnancy, including the endometrium, where it is hypothesized to exert paracrine and autocrine uterotrophic effects [
8,
9]. Relaxin treatment of ovariectomized rats and rhesus monkeys primed with estrogen increases uterine weight relative to controls, and also causes increased arteriole number per unit area, dilated blood vessels on the endometrial luminal surface and pronounced endothelial cell proliferation in arterioles and capillaries of the endometrium [
9‐
12]. Implantation bleeding and increased endometrial thickness are observed in relaxin-treated macaque and marmoset monkeys [
13,
14], with slightly higher implantation rates following in vitro fertilization and embryo transfer in the macaques [
14]. In the context of human physiology, women in a clinical trial for the treatment of scleroderma reported heavier or irregular menstrual bleeding if they were treated with relaxin, possibly due to increased endometrial vascularization [
15,
16].
Few studies have investigated the direct effects of relaxin on endometrial remodeling and angiogenesis, although relaxin receptors are expressed in the human endometrium and relaxin is known to modify the expression of numerous key regulatory factors [
17‐
20]. For instance, in rhesus macaques relaxin treatment inhibits endometrial estrogen receptor α (ERα), progesterone receptor A (PRA) and progesterone receptor B (PRB) protein levels [
9]. Relaxin also stimulates expression and secretion of the pro-angiogenic factor vascular endothelial growth factor (VEGF) in normal human endometrial stromal and glandular epithelial cells in a dose-dependent manner in vitro [
15,
21]. This activation occurs through ERα [
22]. Relaxin treatment in macaques has negative impacts on matrix metalloproteinase (MMP) expression causing decreased pro-MMP and an increase in the MMP inhibitor tissue inhibitor of metalloproteinase 1 (TIMP1) [
9]. Although these studies provide evidence to support a role for relaxin in uterine vascularization and remodeling in the pre-implantation period, they are limited in their analysis of angiogenesis-related genes. Therefore, the first aim was to undertake a more comprehensive microarray approach using human endometrial stromal (HES) cells treated with relaxin
in vitro to identify candidate angiogenesis-related genes for further investigation.
The relaxin gene knockout (
Rln
-/-
) mouse provides a model to investigate endogenous relaxin physiology in the pre-implantation period
in vivo; fetal weights in late pregnancy and pup weight at birth are both significantly reduced, suggesting that mice born to
Rln
-/-
mice are growth restricted
in utero [
23,
24]. Based on the relaxin treatment effects in rodents and rhesus monkeys [
9‐
14], we predicted that this growth restriction could be due to reduced uterine angiogenesis and remodeling in the uterus of
Rln
-/-
mice. Therefore, the second aim of this study was to investigate whether or not relaxin deficiency affected uterine expression of representative genes associated with angiogenesis and uterine remodeling, and also blood vessel proliferation in the pre-implantation endometrium (days 1–4 of pregnancy in mice). This time frame was selected so we could focus on the endometrial changes that occur prior to implantation and independently of any influence from the implanting embryo. As there are no data on paracrine relaxin signaling (relaxin and relaxin receptor,
Rxfp1, expression) in the mouse uterus, a third aim was to demonstrate that relaxin receptors are present in the endometrium and that there is a uterine source of relaxin in early pregnancy.
Discussion
The aims of this study were to confirm expression of relaxin and its receptor RXFP1 in the whole uterus of early pregnant mice prior to implantation and examine the effects of relaxin deficiency on endometrial angiogenesis and the expression of key angiogenesis-related and extracellular matrix remodeling genes in the pre-implantation period. Because fetal growth is compromised in Rln
-/- mice, and relaxin is thought to play a role in vascular growth in the endometrium in primates, our prediction was that pre-implantation angiogenesis in the endometrium would be compromised in Rln
-/-
mice. In fact, we observed that blood vessel proliferation was not different in Rln
-/-
mice compared to wild types before implantation. Unexpectedly, this was despite a general increase in the expression of many angiogenesis-related and other remodeling genes in the uteri of Rln
-/-
mice compared to wild type animals, particularly on day 1 of pregnancy. Relaxin deficiency was also associated with an increase in circulating progesterone levels and progesterone receptor mRNA expression. Overall, our data demonstrate phenotypic changes in the pre-implantation uterus of Rln
-/-
mice that include enhanced expression of steroid receptors, angiogenesis and matrix remodeling-related genes, but without the predicted impact on endometrial angiogenesis.
Circulating relaxin is undetectable in mice in early pregnancy [
35] but relaxin gene transcripts were detected in the ovaries and uterus from day 1 of pregnancy. The surprising finding was the high expression of relaxin on day 3 of pregnancy for which we have no explanation. Previous studies have shown relaxin gene expression in the uterus of early pregnant pigs [
36] and in HES cells in culture [
21]. These data provide good evidence that relaxin is locally produced and can act in a paracrine fashion in the uterus. It was also important to demonstrate
Rxfp1 expression in the uterus of mice in early pregnancy.
Rxfp1 increased between days 1 and 4 of pregnancy in both
Rln
+/+
and
Rln
-/- mice. In fact,
Rxfp1 was higher in
Rln
-/- mice at the start of pregnancy. Increased
Rxfp1 expression has also been shown in the cervix and vagina of
Rln
-/-
mice [
37] but not in the myometrium in late pregnancy [
38]. This suggests differential regulatory effects of the relaxin ligand on its own receptors dependent on tissue and stage of pregnancy [
39].
LacZ staining demonstrated that RXFP1 was localized predominantly in the inner circular layer of the myometrium, but also in the endometrial stroma similar to that shown by Kamat et al [
26]. However, this observation was made in non-pregnant mice, so it is possible that the pattern of RXFP1 staining may differ in early pregnant mice, likely intensified in the endometrium. Several studies have reported high expression of immunoreactive RXFP1 in endometrial stromal cells in uterine samples from women during the menstrual cycle [
17,
18] but this was not substantiated in later studies in which RXFP1 mRNA and relaxin binding sites were predominantly found in glandular and luminal epithelial cells [
19]. No studies to date have assessed RXFP1 in early pregnancy in relation to angiogenesis. In women, there was an increase in RXFP1 in endometrial biopsies collected during the secretory phase relative to those from the proliferative phase [
19]. It was not established if changes in RXFP1 occurred in stromal, epithelial or endothelial cells within the endometrium. However, Kohsaka et al [
40] identified relaxin binding sites in epithelial cells, smooth muscle cells, and blood vessels in the cervix, vagina and uterus in tissues obtained from women after hysterectomy. RXFP1 has been localized specifically to endothelial cells [
41,
42] in systemic and uterine blood vessels, but it remains to be demonstrated if relaxin receptors are expressed in endometrial endothelial cells in early pregnancy.
Because HES cells have previously been shown to express high affinity relaxin binding sites [
15] and to secrete cAMP [
27] and VEGF [
15] in response to relaxin, these cells were used to screen for other potentially relevant genes upregulated by relaxin. Our microarray analysis of gene transcripts altered by relaxin in HES cells revealed a number of genes whose expression is associated with stimulation of angiogenesis, including
Timp3,
Hgf, and
Hifiα. These and other genes known to modulate steroid responsiveness, angiogenesis and extracellular remodeling, including
Esr1,
VegfA, Vegfr2, Pgr,
Elgn1,
Mmp14 and
Ankrd37, were investigated in the mouse uterus by quantitative analysis. Genes including
Esr1,
Hgf,
Hif1α,
Timp3,
VegfA,
Vegfr2 and
Pgr were all upregulated by day 4 of pregnancy in the uterus of
Rln
+/+
mice relative to day 1. This correlates well with previously published data [
34,
43‐
46]. In addition, two other novel genes were upregulated in the uterus of early pregnancy,
Egln1 and
Mmp14. The latter remodels the ECM by activating other MMPs including MMP2 [
47].
Fewer genes were analyzed in the
in vivo study in mice; surprisingly, there was a significant increase in most of the genes examined on day 1 of pregnancy in the
Rln
-/-
mice. Most notably, there was a large increase in
Pgr and
Esr1, but not
Esr2, as well as an increase in circulating progesterone on day 4. This contrasts with previous work that demonstrated increases in
Esr2, but not
Esr1, in the cervix and vagina of late pregnant
Rln
-/-
mice [
37]. These differences may simply be due to tissue-specific expression or because we investigated the uterus prior to implantation, a time when the steroid hormone profile is very different to that later in pregnancy. Treatment of ovariectomized rhesus monkeys with relaxin reduces endometrial progesterone receptor isoforms A and B, and ERα, without effecting ERβ [
9]. These data support our findings in the
Rln
-/- mice, and indicate a role for endogenous relaxin in the regulation of these receptors. Additionally, this accelerated increase in
Pgr in the
Rln
-/-
mice could indirectly mediate the increased expression of the angiogenesis and extracellular matrix remodeling related genes on day 1 of pregnancy; further, we suggest that increased plasma progesterone levels, with increased expression of progesterone receptors, may be sufficient to compensate for the lack of relaxin in the stimulation of angiogenesis.
Another surprising finding was the increased
VegfA expression in
Rln
-/-
mice on day 1 of pregnancy. Relaxin stimulates expression and secretion of VEGF in human endometrial stromal and glandular epithelial cells in a dose-dependent manner in vitro [
15,
21]. Therefore, we had predicted that in the absence of relaxin in early pregnancy,
VegfA would be compromised. However, progesterone mediates its angiogenic effects partly through VEGF so the enhanced progesterone/
Pgr expression on day 1 of pregnancy in
Rln
-/-
mice could explain the increase in
VegfA expression.
Similarly, given the rhRLX-induced increase in Hgf and Hif1α in the human microarray study, we anticipated that expression of these genes would be decreased in the uterus of the Rln
-/- mice. This was not the case. The significance of the increased expression of VegfA, Esr1, Egln1, Hif1α, Mmp14 and Ankrd37 in the Rln
-/-
mice on day 1 of pregnancy is unknown. One possibility is that in the absence of relaxin, the estrous cycle is modified, with a shorter follicular phase and more rapid onset of estrus. Although our mice were synchronized around estrus, they could mate any time between 4 pm and 8 am when they were first checked for mating plugs. If the Rln
-/- mice had shorter estrous cycles, they could mate up to 12 h before Rln
+/+ mice, and therefore post-coital stimuli that trigger changes in the uterine environment would be accelerated in Rln
-/- mice. Of interest, circulating corticosterone levels were significantly elevated in both genotypes on day 1 of pregnancy relative to day 4, suggesting a possible stress response to mating and/or human contact post-coitus.
There are two noteworthy limitations in our study. First, our gene analysis used the whole uterus because it was not possible to separate the endometrium from the myometrium in the mouse uterus in early pregnancy and yield sufficient quantities of high quality RNA. We acknowledge that the homogenized uteri contain both endometrium and myometrium, and that there are spatiotemporal differences in the genes analyzed. Thus it is possible that relaxin deficiency could affect spatiotemporal gene expression. This could only be assessed through
in situ hybridization. Second, we used HES cells in the microarray analysis to identify genes targeted by relaxin. Patterns of relaxin secretion vary in humans and rodents, with relaxin peaking in the first trimester in humans [
48] and increasing throughout pregnancy in rodents [
35,
49]. Furthermore, endometrial angiogenesis occurs in women before the arrival of the blastocyst and primarily occurs in mice after implantation. Therefore the role of relaxin in these two species could be quite different, and the pre-implantation uterus may not be a substantial relaxin target in mice. Importantly, we would argue that relaxin-deficient mice are not an appropriate animal model to further investigate the role of relaxin in the pre-implantation uterus in humans.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SM carried out the uterine gene analysis, immunohistochemistry, performed the statistical analysis and wrote the manuscript. LN completed the reverse transcriptase study. ENU carried out the gene array study and drafted the manuscript. LJP carried out the RXFP1 localization study. LJP and JEG conceived the study, participated in its design and coordination and the drafting of the manuscript. All authors approved the final manuscript.