Parturition failure in mice lacking Mamld1

Abstract

In mice, the onset of parturition is triggered by a rapid decline in circulating progesterone. Progesterone withdrawal occurs as a result of functional luteolysis, which is characterized by an increase in the enzymatic activity of 20α-hydroxysteroid dehydrogenase (20α-HSD) in the corpus luteum and is mediated by the prostaglandin F2α (PGF_2α) signaling. Here, we report that the genetic knockout (KO) of Mamld1, which encodes a putative non-DNA-binding regulator of testicular steroidogenesis, caused defective functional luteolysis and subsequent parturition failure and neonatal deaths. Progesterone receptor inhibition induced the onset of parturition in pregnant KO mice and MAMLD1 regulated the expression of Akr1c18, the gene encoding 20α-HSD, in cultured cells. Ovaries of KO mice at late gestation were morphologically unremarkable; however, Akr1c18 expression was reduced and expression of its suppressor Stat5b was markedly increased. Several other genes including Prlr, Cyp19a1, Oxtr and Lgals3 were also dysregulated in the KO ovaries, whereas PGF_2α signaling genes remained unaffected. These results highlight the role of MAMLD1 in labour initiation. MAMLD1 likely participates in functional luteolysis by regulating Stat5b and other genes, independent of the PGF_2α signaling pathway.

Introduction

In most mammals including mice, uterine quiescence during pregnancy is maintained by circulating progesterone synthesized primarily in the ovarian luteal cells^1,2. Progesterone binds to its receptor in the uterus and suppresses the expression of genes involved in myometrial contraction^3,4. Previous studies have shown that signal transducer and activator of transcription 5b (STAT5B) is essential to sustain blood progesterone levels in pregnant mice^5,6,7. STAT5B inhibits ovarian expression of Akr1c18, the gene for 20α-hydroxysteroid dehydrogenase (20α-HSD) that converts progesterone into an inactive metabolite 20α-hydroxyprogesterone (20α-OHP)⁵. From 18 days post coitum (dpc), i.e., 24–36 hours before term, progesterone secretion from the ovary progressively declines through processes referred to as functional and structural luteolysis^1,5. Functional luteolysis is an enzymatic shift characterized by an increase in 20α-HSD activity^1,5. This process is followed by structural luteolysis, in which the corpus luteum undergoes morphological changes and cellular apoptosis^1,8. To date, multiple molecules have been implicated in functional luteolysis¹. Of these, prostaglandin F2α (PGF_2α) upregulates Akr1c18 via a signaling pathway consisting of PGF_2α, PGF_2α receptor (FP), JUND and nuclear receptor subfamily 4 group A member 1 (NR4A1, also known as NUR77)^1,9,10. The Gq/11 protein family also serves as a component of the PGF_2α signaling pathway¹¹. Genetic knockout (KO) of Akr1c18, Fp, or Gα_q/11 leads to persistent progesterone production and subsequent parturition failure^5,11,12,13. In addition, Fp KO perturbs expression of several steroidogenic genes in the corpus luteum, which may also be relevant to delayed parturition¹⁴. Other factors, including NOTCH 1 and 4, oxytocin receptor (OXTR) and galectin 3, also participate in luteolytic processes and/or regulation of Akr1c18^1,15,16,17; however, it remains unknown whether STAT5B plays a role in functional luteolysis.

MAMLD1 on the human X chromosome (NM_001177465) is a causative gene for disorders of sex development in 46,XY individuals¹⁸. Loss-of-function mutations in MAMLD1 have been identified in male patients with hypospadias^18,19,20. Murine Mamld1 (NM_001081354) also resides on the X chromosome and is strongly expressed in the Leydig and Sertoli cells of the fetal testis¹⁸. In vitro knockdown assays using mouse Leydig tumour cells (MLTC1) and in vivo analysis of male Mamld1 KO mice indicated that MAMLD1 transactivates several Leydig cell-specific genes including Star, Cyp11a1, Cyp17a1, Hsd3b1 and Insl3 without exerting a demonstrable DNA-binding capacity^21,22,23. While male Mamld1 KO mice showed no hypospadias, the phenotypic difference between human patients and KO mice was explicable by species differences in the process of male sex development²³. To date, the function of MAMLD1 in females has not been investigated, although previous analyses detected strong expression of Mamld1 in the ovaries of adult mice¹⁸. In the present study, we analyzed phenotypic and molecular characteristics of female Mamld1 KO mice.

Results

Mamld1 KO causes parturition failure in female mice

Prior to this study, we generated a mouse strain in which the genomic structure of Mamld1 was disrupted by substituting a PGK-neo cassette for Mamld1 exon 3 that corresponds to approximately two-thirds of the coding region²³. We have reported that male Mamld1 KO mice retained normal external genitalia and fertility, despite having mildly impaired expression of Leydig cell-specific genes in the fetal testis²³.

In this study, we analyzed the phenotype of female Mamld1 KO mice. The mice were healthy and exhibited no discernible anomalies. Furthermore, the mice were fertile when mated with male wildtype (WT) or Mamld1 KO mice. However, female KO mice frequently showed delayed parturition (Table 1). More than 50% of KO mice gave birth to their first pups at 20.5 dpc or later, while approximately 80% of WT animals gave birth at 19.5 dpc. The frequency of delayed parturition (≥20.5 dpc) in WT animals was comparable between this study and previous studies^3,24. The genotype of the mated male mice (WT or KO) had no influence on the parturition timing of the female WT or KO mice.

Table 1 Frequency of delayed parturition.

Pups born to Mamld1 KO mothers have a high neonatal mortality rate and can be rescued by caesarean operation

We examined the number of pups born to WT and Mamld1 KO mothers. Although the average number of pups at birth was comparable between the two groups, the average number of pups alive at postnatal day 1 was significantly lower in KO mothers (Fig. 1a,b). Approximately half of the pups born to Mamld1 KO mothers died within the first 24 hours after birth, while >80% of pups born to WT mothers survived beyond this period. The dead pups of KO mothers exhibited no apparent malformations (Fig. 1a). Most pups survived beyond postnatal day 1 remained alive until adulthood. The newborn mortality rates of WT and KO mothers were not affected by paternal genotype (WT or KO). The sex ratio of the dead pups was almost 1:1. Thus, the neonatal deaths were more likely the result of an aberrant maternal condition rather than inborn defects in the pups.

Figure 1

Phenotypes of pups born to Mamld1 knockout (KO) mothers.

(a) Pups born to KO mothers at 20.5 dpc. Pups showed a high neonatal mortality, although they had no congenital anomalies. The photographs were taken by M.M. and M.F. at the National Research Institute for Child Health and Development. (b) Average number of births (white bars) and that of live pups at postnatal day 1 (gray bars). Pups born to KO mothers showed a significantly higher newborn mortality rate than those born to wildtype (WT) mothers (asterisks). Paternal genotype had no effect on the number of pups. Frequent newborn deaths were eliminated by caesarean operation. The results are expressed as the mean ± SEM.

It is known that parturition failure in female Fp KO mice results in frequent fetal death^12,13. To clarify whether the high mortality rate of pups born to Mamld1 KO mothers was due to delayed parturition, we performed caesarean operations on the day of the expected term (19.5 dpc). The operations significantly improved the survival rate of pups; at postnatal day 1, the average number of live pups born to the operated KO mothers was comparable to that born to non-operated WT mothers (Fig. 1b).

Progesterone withdrawal is impaired in pregnant Mamld1 KO mice

Previous studies have shown that parturition failure is caused by defects in functional luteolysis that lead to persistent progesterone production^{5,12,13,15,24}; however, it can also be caused by uterine lesions such as defective myometrial contraction or delayed cervical ripening^25,26. To determine whether progesterone withdrawal is impaired in pregnant Mamld1 KO mice, we measured serum levels of progesterone and other steroids. In this study, we utilized liquid chromatography tandem mass spectrometry (LC-MS/MS), which is more sensitive and accurate than conventional immunoassays²⁷. Serum samples were collected from pregnant WT and KO mice at 18.5 dpc, a stage at which circulating progesterone usually declines in WT mice²⁵. Serum progesterone was significantly higher in KO mice than in WT animals (Table 2). In contrast, serum levels of 20α-OHP, the inactive metabolite of progesterone, remained low in KO mice. Altered serum levels of progesterone and 20α-OHP were also observed in KO mice at 20.5 dpc (Table 2). Blood levels of testosterone and estradiol were comparable between WT and KO mice.

Table 2 Serum steroid hormone levels in WT and Mamld1 KO mice.

To confirm that impaired progesterone withdrawal is the major cause of parturition failure in Mamld1 KO mice, we treated pregnant mice with the progesterone receptor antagonist RU486. Administration of 150 μg RU486 at 17.5 or 18.5 dpc invariably induced vaginal bleeding (the signs of labour initiation) and/or delivery of a pup(s) within 24 hours in both WT and KO mice (Supplementary Table S1).

We also examined whether Mamld1 KO affects ovarian structures. The size and appearance of the ovaries were comparable between pregnant WT and KO mice at 18.5 dpc (Fig. 2a,b). No apparent histological changes were observed in the ovaries of KO mice (Fig. 2c–f). Furthermore, the average number of corpora lutea in the ovary and that of implants in the uterus were similar between WT and KO mice (Fig. 2g). The position of uterine implantation was also normal in KO mice. These data indicate that Mamld1 KO exerts a deleterious effect on functional luteolysis, but not on ovary development, ovulation, luteinization or implantation.

Figure 2

Morphological analysis.

(a–f) Morphological findings of the ovaries obtained from pregnant WT and Mamld1 KO mice at 18.5 dpc. Scale bars: 1 mm (a,b), 500 μm (c,d) and 100 μm (e,f). CL, corpus luteum. (g) Average number of corpora lutea in the ovary (upper panel) and that of implants in the uterus (lower panel) at 18.5 dpc. The results are expressed as the mean ± SEM.

In ovaries of WT mice during late gestation, Mamld1 is continuously expressed, while expression levels of Akr1c18, Nr4a1, and Stat5b drastically change after 17.5 dpc

We examined Mamld1 expression in the ovaries of WT mice at late gestation. Real-time PCR detected continuous expression in the ovaries, with the highest expression at 17.5 dpc (Fig. 3a). In situ hybridization of the murine ovary at 18.5 dpc showed clear signals for Mamld1 mRNA in the corpora lutea as well as in the primary, secondary, vesicular and mature follicles (Fig. 3b–d).

Figure 3

Mamld1 expression in pregnant WT mice.

(a) Mamld1 expression in whole ovaries from pregnant WT mice at 16.5 (n = 3), 17.5 (n = 5) and 18.5 dpc (n = 8). mRNA levels relative to that of Gapdh are shown. The results are expressed as the mean ± SEM. The average of mRNA levels at 16.5 dpc was defined as 1.0. Asterisks indicate statistical significance. (b–d) Mamld1 expression in corpora lutea and follicles. Arrowheads indicate Mamld1 signals in corpus luteum. Scale bars: 200 μm (b), 100 μm (c) and 20 μm (d). No specific expression of the negative control (a sense probe). CL, corpus luteum. (e) Known factors involved in functional luteolysis. Arrow and bar headed lines indicate stimulatory and inhibitory effects, respectively. Double lines indicate protein-receptor bindings. FP, prostaglandin F2α receptor; PRL, prolactin; PRLR, PRL receptor. (f) Gene expression in the whole ovaries in pregnant WT mice at 17.5 and 18.5 dpc (n = 5 and 8, respectively). The average of mRNA levels at 17.5 dpc was defined as 1.0. (g) Mamld1 expression in the whole ovaries (Ovary, n = 8) and uteri (Uterus, n = 4) from pregnant WT mice at 18.5 dpc. The average of mRNA levels in the whole ovaries was defined as 1.0.

We also analyzed mRNA levels of Akr1c18, Nr4a1 and Stat5b in ovaries of pregnant WT mice at 17.5 and 18.5 dpc. These genes showed drastic changes in expression between 17.5 and 18.5 dpc, as reported previously¹⁰. Akr1c18 and Nr4a1 expression was significantly higher at 18.5 dpc than at 17.5 dpc, while Stat5b expression was markedly decreased at 18.5 dpc (Fig. 3f).

In addition, we analyzed Mamld1 expression in the uteri of pregnant WT mice at 18.5 dpc. A relatively weak expression was detected in the uteri, as compared to that in the ovaries (Fig. 3g).

MAMLD1 regulates Akr1c18 expression in vivo and in vitro

We examined the expression of Akr1c18/20α-HSD in pregnant WT and Mamld1 KO mice at 18.5 dpc. Real-time PCR analysis showed significantly decreased Akr1c18 expression in the whole ovaries and corpora lutea of KO mice (Fig. 4a) and Western blot analysis confirmed the reduction of 20α-HSD protein expression in the ovaries of KO mice (Supplementary Fig. S1). In contrast, mRNA levels of Akr1c18 in the uteri were comparable between WT and KO mice at 18.5 dpc (Fig. 4b). Expression of Srd5a1 for steroid 5α reductase, which mediates local progesterone metabolism in the uterus, remained unaffected in KO mice (Fig. 4b). Akr1c18 expression remained low in the KO mice at 20.5 dpc (Supplementary Fig. S2).

Figure 4

The effect of MAMLD1 on Akr1c18 expression.

(a) Akr1c18 expression in the whole ovaries (Ovary) and corpora lutea (CL) of pregnant WT (n = 8, white bars) and Mamld1 KO (n = 9, black bars) mice at 18.5 dpc. mRNA levels relative to that of Gapdh are shown. The results are expressed as the mean ± SEM. The average of mRNA levels in the whole ovaries of WT mice was defined as 1.0. Significant differences between WT and KO animals are indicated by asterisks. (b) Akr1c18 and Srd5a1 expression in the uteri (Uterus) of pregnant WT (n = 4, white bars) and KO (n = 4, black bars) mice at 18.5 dpc. The average of mRNA levels in WT mice was defined as 1.0. (c) Mamld1 knockdown assays. N.C., negative control (non-targeting siRNA). The average of mRNA levels in N.C. was defined as 1.0. (d) Overexpression experiments of Mamld1. Vector, empty expression vector. The average of mRNA levels in Vector was defined as 1.0.

To confirm the effect of MAMLD1 on Akr1c18 expression, we performed in vitro assays. In these experiments, we used MLTC1, which has high endogenous expression of both Mamld1 and Akr1c18. First, we carried out knockdown assays using two siRNAs for Mamld1. When Mamld1 mRNA levels were suppressed to ~25% by the siRNAs, Akr1c18 mRNA levels were reduced to ~75% (Fig. 4c). Next, we performed Mamld1 overexpression experiments. Transient transfection with a Mamld1 expression vector resulted in a ~2-fold increase of Akr1c18 mRNA after a 24-hour cell culture (Fig. 4d).

Mamld1 KO dysregulates Stat5b and other genes in the ovaries of pregnant mice

We examined gene expression patterns in the whole ovaries and corpora lutea of pregnant WT and Mamld1 KO mice at 18.5 dpc (Fig. 5). The most remarkable changes in KO mice were the significantly increased mRNA levels of Stat5b, despite overexpression of Socs3, which encodes a putative inhibitor of Stat5. Prlr and Esr1 were also upregulated. In contrast, Fp, Jund and Nr4a1 were not affected, except for a slightly decreased expression of Jund in the whole ovaries. Increased levels of STAT5B protein and unaffected levels of NR4A1 protein in KO mice ovaries were confirmed by Western blot analysis (Supplementary Fig. S1). Markedly increased Stat5b mRNA expression was also observed in pregnant KO mice at 20.5 dpc (Supplementary Fig. S2).

Figure 5

Expression patterns of functional luteolysis-related genes in pregnant WT and Mamld1 KO mice.

Relative mRNA levels of genes in the whole ovaries (Ovary) and corpora lutea (CL) in pregnant WT (n = 6, white bars) and KO (n = 6, black bars) mice at 18.5 dpc are shown. The results are expressed as the mean ± SEM. The average of mRNA levels in the whole ovaries of WT mice was defined as 1.0. Asterisks indicate statistical significance.

We also analyzed mRNA levels of other genes involved in ovarian steroidogenesis and in the luteolytic processes (Figs 5 and 6). Gene expression patterns were grossly similar in the whole ovaries and corpora lutea. Among the steroidogenic genes, Cyp19a1 was significantly upregulated. Expression levels of Hsd17b3, Hsd17b1 and Hsd17b7 were mildly increased, while mRNA levels of Cyp11a1 and Cyp17a1 remained unchanged. Star expression was slightly decreased, but only in the whole ovaries. Of the genes involved in the luteolytic processes, Oxtr was upregulated, while Lgals3 encoding anti-apoptotic factor galectin 3 was downregulated. Notch 1 and 4 were unaffected.

Figure 6

Expression patterns of steroidogenic genes in pregnant WT and Mamld1 KO mice.

(a) Enzymes involved in the steroidogenic pathway. 17-OHP, 17-hydroxyprogesterone; 20α-OHP, 20α-hydroxyprogesterone; E₁, estrone; Δ4A, androstenedione; T, testosterone; E₂, estradiol. (b) Gene expression in the whole ovaries (Ovary) and corpora lutea (CL) in pregnant WT (n = 6, white bars) and KO (n = 6, black bars) mice at 18.5 dpc. mRNA levels relative to that of Gapdh are shown. The results are expressed as the mean ± SEM. The average of mRNA levels in the whole ovaries of WT mice was defined as 1.0. Significant differences between WT and KO animals are indicated by asterisks. Hsd3b2 was undetectable in both WT and KO mice.

Discussion

Targeted deletion of Mamld1 in female mice caused parturition failure and frequent neonatal deaths without affecting ovarian morphology. This phenotype likely results from attenuated functional luteolysis, because expression of Akr1c18 mRNA and 20α-HSD protein was markedly decreased in the ovaries of pregnant Mamld1 KO mice at 18.5 dpc. Consistent with this, ratios of 20α-OHP to progesterone in blood samples were lower in KO mice than in WT animals. Although the serum levels of progesterone and 20α-OHP in our mice differed from those in previous reports^5,10, this can be ascribed to the difference in the methods (LC-MS/MS vs. conventional immunoassays) and sampling points (the day when a vaginal plug was observed was designated as 0.5 dpc in this study and as 1.0 dpc in previous studies). Attenuated functional luteolysis seemed to persist in KO mice after the day of the expected term. We found that inhibition of progesterone signaling by RU486 induced vaginal bleeding (the signs of labour initiation) and/or delivery of a pup(s) in KO mice. In vitro assays indicated that MAMLD1 upregulates Akr1c18 in MLTC1, although these results need to be confirmed in further studies using cells of ovarian origin. While previous studies have shown that local progesterone metabolism in the uterus can also affect parturition timing^25,26, mRNA levels of Akr1c18 and Srd5a1 in the uteri remained unaffected in Mamld1 KO mice. Furthermore, Mamld1 was continuously expressed in the ovaries during late gestation and only weakly expressed in the uteri. Collectively, the results suggest that MAMLD1 is involved in upregulation of Akr1c18 in ovaries of pregnant mice at late gestation.

The phenotype of pregnant Mamld1 KO mice overlaps with that of Fp KO mice^12,13; however, expression of the PGF_2α signaling pathway genes, Fp, Jund and Nr4a1, was not significantly altered in the ovaries of Mamld1 KO mice at 18.5 dpc. Likewise, protein expression of NR4A1, the most downstream component of the PGF_2α signaling pathway that directly binds to the Akr1c18 promoter, remained unaffected in KO mice ovaries. Thus, the function of MAMLD1 appears to be independent of the PGF_2α signaling pathway, although mRNA expression of the Gq/11 protein family, a recently identified component of this pathway¹¹, was not analyzed in the present study. In contrast, Stat5b and Prlr were markedly upregulated in KO mice ovaries. Increased Prlr expression can be ascribed to high STAT5B activity, which transactivates Prlr²⁸. Likewise, Esr1, the potential target of STAT5B in rats²⁹, was also upregulated in Mamld1 KO mice. To date, STAT5B has not been implicated in functional luteolysis, although it suppresses Akr1c18 during mid-gestation⁵. We confirmed that Stat5b expression significantly declined in pregnant WT mice ovaries after 17.5 dpc. Our data imply that Stat5b suppression mediated by MAMLD1 is critical for functional luteolysis. Since MAMLD1 protein transactivates various genes in the fetal testis without demonstrable DNA binding capacity^21,23, MAMLD1 may regulate Stat5b expression as a non-DNA-binding co-activator. In this regard, it is noteworthy that the phenotypic severity of pregnant Mamld1 KO mice was milder than that of Fp KO mice. While Mamld1 KO permits a term delivery in approximately half of pregnant mice, Fp KO leads to parturition failure and loss of pups in all mice^12,13. Likewise, the increase in blood progesterone levels at the end of pregnancy was less significant in Mamld1 KO mice than in Fp KO mice. These results are consistent with the findings that Akr1c18 mRNA levels in the ovaries were decreased by 70–80% in pregnant Mamld1 KO mice and by 100% in Fp KO mice¹⁰. This suggests that although MAMLD1 and PGF_2α signaling are essential for the luteolytic process, the role of MAMLD1 is relatively minor compared to that of PGF_2α signaling.

Several other genes were dysregulated in pregnant Mamld1 KO mice ovaries. First, Cyp19a1, Hsd17b3, Hsd17b1 and Hsd17b7 involved in ovarian steroidogenesis were upregulated. These molecular alterations did not affect blood sex hormone levels. However, perturbed steroidogenesis may play a role in parturition failure of Mamld1 KO mice, because previous studies suggested that the androgen:estrogen synthesis ratio in the ovaries affects the luteolytic process¹⁴. Second, Oxtr expression was increased in the KO mice ovaries. It has been shown that administration of low-dose oxytocin results in persistent progesterone production and subsequent parturition failure, whereas high-dose oxytocin causes uterine contraction and early labour¹². Since downregulation of Oxtr in the ovaries and its upregulation in the uteri were proposed to induce the onset of parturition^2,30, elevated expression of Oxtr in the ovaries of Mamld1 KO mice may be associated with delayed parturition. Third, expression of Lgals3 was decreased in the whole ovaries and corpora lutea of KO mice. Lgals3 is co-expressed with Akr1c18 in the corpora lutea and galectin 3 encoded by Lgals3 contributes to the elimination of luteal cells⁸. Thus, decreased Lgals3 expression in the ovaries of Mamld1 KO mice may also be relevant to impaired luteolysis. Lastly, expression of Notch 1 and 4 remained intact in KO mice. Thus, although MAMLD1 has sequence similarity with a Notch co-factor Mastermind-like 2²¹, the function of MAMLD1 in the ovaries is unlikely to be associated with Notch signals.

In summary, our results indicate that MAMLD1-mediated Stat5b suppression is essential for term delivery in mice. MAMLD1 appears to participate in a complex molecular network in the ovaries and regulate functional luteolysis, without affecting expression of PGF_2α signaling genes. This study provides novel insights into molecular mechanisms of mammalian reproduction.

Methods

Treatment of animals

Animal experiments in this study were approved by the Animal Care Committee at the National Research Institute for Child Health and Development (project number: A2008-001). All experiments were performed in accordance with the institutional guidelines of the care and use of laboratory animals. All mice were housed under specific pathogen-free controlled conditions with a 12-hour light-dark cycle. Food and water were available ad libitum.

Mamld1 KO mice

Male Mamld1 KO mice were generated by targeting deletion of exon 3²³. The mice were backcrossed with the C57BL/6N strain (Sankyo Labo Service Corp. Inc., Tokyo, Japan).

Cross-mating and caesarean operation

Cross-mating was performed between female Mamld1 KO mice and male WT or KO mice and between female WT mice and male WT or KO mice. Female mice from 7 to 25 weeks of age and male mice from 8 to 40 weeks of age were used for mating. The noon of the day when a vaginal plug was observed was designated as 0.5 dpc. Vaginal bleeding (the signs of labour initiation) or delivery of the first pup was defined as the onset of parturition. Caesarean operation was performed for Mamld1 KO mice at 19.5 dpc. After birth, the pups were nursed by lactating WT animals.

Measurement of serum steroid metabolites

Blood samples were collected from the right ventricle of the heart of euthanized pregnant WT and KO mice at 18.5 dpc, pregnant KO mice at 20.5 dpc and WT mice at 0 or 1 day postpartum. The serum was separated by centrifugation and stored at −80 °C until hormone measurements were performed. Serum steroid metabolites were measured by LC-MS/MS (ASKA Pharma Medical, Kanagawa, Japan).

Parturition induction by progesterone receptor antagonist

The progesterone receptor antagonist RU486 (mifepristone; Sigma-Aldrich, St. Louis, MO) was administered to pregnant mice at 17.5 or 18.5 dpc. One ml of solution containing 150 μg RU486 in 6% ethanol was subcutaneously injected in the bilateral hind legs.

Morphological and quantitative analyses of corpora lutea and uterine implants

We analyzed the morphology of ovaries obtained from pregnant WT and KO mice at 18.5 dpc. Tissue samples were fixed with 4% paraformaldehyde, dehydrated and embedded in paraffin. Serial 6 μm sections were mounted on microscope slides. The samples were stained with hematoxylin-eosin and the number of corpora lutea in the ovary and implants in the uterus were counted under a stereoscope.

Real-time RT-PCR analysis

Whole ovaries and corpora lutea were isolated from pregnant WT (n = 12–16) and Mamld1 KO (n = 12–18) mice at 18.5 dpc and uteri were isolated from four mice of each genotype at the same stage. Whole ovaries were also isolated from pregnant WT mice at 16.5 and 17.5 dpc (n = 3 and 5, respectively), pregnant KO mice at 20.5 dpc (n = 5) and WT mice at 0 or 1 day postpartum (n = 4). Tissues were immediately soaked in RNAlater solution (Life Technologies, Carlsbad, CA). Total RNA was extracted from homogenized samples by ISOGEN (Nippongene, Tokyo, Japan) and RNeasy Kit (QIAGEN, Valencia, CA). Contaminated genomic DNA was removed with a TURBO DNA-free kit (Life Technologies). cDNA was synthesized from 200 ng total RNA using a High Capacity cDNA Reverse Transcription kit (Life Technologies). We measured relative mRNA levels of genes implicated in the luteolytic process and/or regulation of Akr1c18. Gapdh was used as an internal control. The assays were performed using the ABI 7500 Fast real-time PCR system and TaqMan gene expression assay kit (Life Technologies). Primers and probes used in this study are listed in Supplementary Table S2.

In situ hybridization

We examined Mamld1 expression in the ovaries obtained from pregnant WT mice at 18.5 dpc. Paraffin sections were prepared as described above. In situ hybridization was performed using an antisense RNA probe for mouse Mamld1¹⁸ (Genostaff Inc., Tokyo, Japan). The probe was digoxigenin-labeled using DIG RNA Labeling Mix (Roche, Basel, Switzerland). A sense cRNA for mouse Mamld1 was used as a negative control. The colour of the probes was developed with NBT/BCIP solution (Sigma-Aldrich) and the sections were counterstained with Kernechtrot solution (Mutoh Chemical, Tokyo, Japan).

Western blot analysis

Tissue extracts were prepared from the ovaries of pregnant mice at 18.5 dpc and separated by standard SDS-PAGE (7.5% or 4–20% gradient gel; Bio-Rad, Hercules, CA). PVDF membranes were incubated in the solution containing the primary antibody. We used anti-20α-HSD antibodies (EB4002; KeraFAST Inc., Boston, MA), anti-NR4A1 antibodies (ab13851; Abcam, Cambridge, MA) and anti-STAT5B antibodies (ab178941; Abcam). Anti-ACTIN antibodies (A2066; Sigma-Aldrich) were used as an internal control. The signals were detected using Clarity Western ECL Substrate (Bio-Rad). All analyses were performed using three independent samples per group.

In vitro functional assays

MLTC1 (CRL-2065^TM; ATCC, Manassas, VA) were maintained in RPMI 1640 medium containing 10% fetal bovine serum. For Mamld1 knockdown assays, the cells were seeded in 6-well plates (1.0 × 10⁵ cells/well) and transiently transfected with two siRNAs, i.e., siRNA1 (sense: 5′-CAGGAAUCGGGAACCAGUAAGAGAA-3′; and anti-sense: 5′-UUCUCUUACUGGUUCCCGAUUCCUG-3′) and siRNA2 (sense: 5′-CAGAGAUGCAGAUGCCCACAUUAAA-3′; and anti-sense: 5′-UUUAAUGUGGGCAUCUGCAUCUCUG-3′), or with a non-targeting control RNA (4611G; Life Technologies) (20 nM final concentration), using Lipofectamine RNAiMAX (Life Technologies). For Mamld1 overexpression assays, the cells were seeded in 12-well plates (1.0 × 10⁵ cells/well) and transfected with 200  ng of the expression vector of Mamld1 or an empty expression vector (pCMV-Myc vector; Takara Bio, Otsu, Japan), using Lipofectamine 3000 (Life Technologies). The full-length Mamld1 cDNA, which contains 2,412 nucleotides corresponding to the coding region without both 5′- and 3′-untranslated regions, was amplified from mouse fetal testis-derived cDNA mixture (C57BL/6N; Sankyo Labo Service Corp. Inc.) and subcloned into a plasmid that was included in the TOPO TA cloning kit (Life Technologies). The cDNA that was missing the start codon was then subcloned into a pCMV-Myc vector to construct the Mamld1 expression vector.

The cells were harvested 24 hours after transfection. Total RNA were subjected to cDNA synthesis. Amounts of endogenous Mamld1 and Akr1c18 relative to that of Gapdh were analyzed by TaqMan real-time PCR in three independent experiments.

Statistical analysis

Data are expressed as the mean ± SEM. Statistical differences in mean values between two groups were examined by Student’s t-test or Mann-Whitney’s U-test and differences in frequencies were examined by χ² test. P values less than 0.05 were considered significant.