Absence of Nuclear Receptors for Oxysterols Liver X Receptor Induces Ovarian Hyperstimulation Syndrome in Mice

Animals

LXRα and LXRβ knockout mice (LXR^−/−) and their wild-type controls were maintained on a mixed-strain background (C57BL/6:129Sv), housed in a temperature-controlled room with a 12-h light, 12-h dark cycle and fed ad libitum with water and Global-diet 2016S (Harlan, Gannat, France). These mice have been extensively detailed in previous articles (18,19,20,21). All experiments were performed on age-matched female mice. Except when indicated, mice were 6–12 months old. For all experiments, except for estradiol synchronization, animals were treated with a superovulation protocol as previously described (21). Briefly, mice received an ip injection of 7.5 IU pregnant mare’s serum gonadotropin on d 1 at 1400 h and 5 IU human chorionic gonadotropin on d 3 at 0800 h, and then killed on d 5 at the end of metaestrus, starting at 1400 h. For real-time PCR (qPCR) (1) experiments, mice were gavaged with 45 mg/kg T0901317 (T1317) (Cayman Chemical, Montigny le Bretonneux, France) or vehicle (methyl-cellulose) (20). In estradiol synchronization protocol, estrus was induced with a single injection of 10 μg estradiol benzoate (Sigma-Aldrich, L’Isle D'Abeau, France) 18 h before the animals were killed, i.e. at 2000 h (21). Before oocyte counting, mice were killed on d 4, a short time after ovulation. The oocyte cumulus masses were surgically harvested from the oviducts and cultured in DMEM (Invitrogen, Cergy Pontoise, France) supplemented with 20% fetal calf serum and antibiotics. After treating with 0.3% hyaluronidase (Sigma-Aldrich) for 10 min, oocytes were observed and counted. To minimize the effect of stress, the elapsed time between the capture of a mouse and when it was killed was less than 30 sec. All aspects of animal care were approved by the Regional Ethics Committee (authorization CE1-04).

Anatomy and pathology analyses

Ovaries were collected, fixed, and embedded in paraffin, and 5-μm-thick sections were prepared and stained with hematoxylin/eosin/saffron and analyzed with Axiovision 4.2 software (Carl Zeiss Vision GmbH, Le Pecq, France) as previously described (21).

Laser microdissection (LMD)

Serial cryosections (25 μm) of fresh-frozen ovaries were cut at −16 C, thaw mounted onto PEN slides (Leica, Québec, Canada), stained with toluidine blue, and dehydrated in 70 and 100% ethanol followed by incubation for 1 h at 37 C. The granulosa cells were excised by LMD (Leica LMD6000) at ×20 magnification directly into either RNA-lysis (RNeasy microkit; QIAGEN, Courtaboeuf, France) as previously described (22).

Real-time PCR

Total RNA was isolated from whole ovary using the Trizol method (Invitrogen) according to the manufacturer’s instructions. cDNA was synthesized with Moloney murine leukemia virus reverse transcriptase (Promega, Charbonnières, France) and random hexamer primers (Promega) according to the manufacturer’s recommendations. The real-time PCR was performed on an iCycler (Bio-Rad, Marnes-la-Coquette, France) as described before (21). Primers are given in Table 1.

Evaluation of Evans blue leakage

Before the animals were killed, mice from both groups were anesthetized with a single ip injection of chloral (3.75 mg/10 g). Mice received then an intracardiac injection with 100 μl of 25 mg/ml of Evans blue. Ovaries were collected and incubated in 1 ml of formamide for 24 h at 60 C. To evaluate ovarian capillary permeability, Evans blue concentration in the formamide extract was measured by light absorption at 620 nm using a spectrophotometer as described (23). The ovarian Evans blue content was expressed as ng/mg of tissue.

Steroid measurements

After decapitation of superovulated animals, sera were collected and steroid hormone concentrations were assessed by electrochimioluminescence (Elecsys Progesterone II and Estradiol II; Roche Diagnostics, Meylan, France).

Statistical analysis

Student’s t test was performed to determine whether there were significant differences between the groups. P = 0.05 was considered significant.

Ovulation induction by gonadotropins leads to ovarian enlargement and hemorrhagic corpora lutea

Morphological examination of superovulated ovaries showed an increased size of the ovaries in the LXR^−/− mice compared with the wild-type mice (Fig. 1A) and the presence of multiple hemorrhagic cysts (data not shown). Somatic indexes of LXR^−/− mice ovaries were altered compared with wild-type mice: even though a higher (yet not significant, P = 0.07) increase of the somatic index was pointed for the animals from 2 and 3 months of age, a significant increase of this index was clearly observed for all the tested ages, suggesting the effect of the superovulation was not age dependent. In contrast, estradiol treatment alone to synchronize female cycles did not result in ovary size differences in LXR-deficient mice, showing that only the superovulation protocol led to ovarian enlargement. These data suggest that LXR^−/− mice have an increased ovarian response to gonadotropins. In rare cases it could be observed a severe phenotype (Fig. 1B) with massive ovarian hemorrhages of the corpus luteum (Fig. 1C), as suggested by the presence of luteal cells surrounding the hemorrhage (Fig. 1D). This fact has already been described in the human pathology (2,3). The reasons that OHSS range from slight to dramatic situations are still unknown and probably involve the sensitivity to some paracrine factors. In the other cases, gross examination showed that hemorrhagic lesions were more numerous and larger in LXR-deficient ovaries (Fig. 2A). It can also be noticed the presence of the follicles at all stages.

One of the main causes of OHSS symptoms is the presence of third space fluid shift. Ovarian vascular permeability was thus measured in ovaries using Evans blue, as previously described in rat (23). Dye content in the ovaries from LXR-deficient mice (0.21 ± 0.06 ng/mg tissue) was significantly higher (1.9-fold, P < 0.05) than in the control group (0.11 ± 0.05 ng/mg tissue; Fig. 2B) showing that LXRs control vascular permeability. Expression of lxrα and lxrβ was measured during superovulation protocol in granulosa/luteal cells purified by LMD from wild-type ovaries collected at specific time points (Fig. 2C). Both isoforms were expressed in all tested follicular stages. Interestingly, lxrα clearly showed a developmental regulation. In addition, a significant peak of its expression was seen at the time when corpus luteum function is maximal. These data strongly suggested that the lack of LXR could be, at least in part, associated to an abnormal response of the corpus luteum.

LXR-null mice have increased ovarian response to gonadotropins

Because OHSS is caused by an abnormal ovarian response to gonadotropins, we measured the ovary response after follicle retrieval following an in vitro fertilization protocol. Expulsed follicle number was greater in LXR^−/− females compared with wild-type mice (Fig. 3A; 2.2-fold, P < 0.02), pointing out the crucial role of LXRs in the control of the ovarian response. Among the expelled oocytes, 45% were dead or presented empty zonae pellucidae (Fig. 3B).

Ovarian hyperstimulation is linked to increased estradiol production in LXR-deficient mice

Because estradiol concentration is one of the major factors involved in the progression of the syndrome (5), ovarian steroids were measured in the sera of superovulated mice. Whereas no difference in progesterone concentration was observed among all groups, estradiol (E2) was shown to be increased in LXR-deficient mice compared with wild-type females. Conversely, gavage of wild-type mice with the synthetic LXR agonist T1317 induced an increase of E2 production (Fig. 4A). Consistent with increased E2-production, Star (steroidogenic acute regulatory protein) and cyp11a1 (cytochrome P450 side chain cleavage) expression was increased in wild-type animals gavaged with T1317 (Fig. 4B; 1.8- and 1.5-fold induction for star and cyp11a1, respectively; P < 0.05), as previously described in other steroidogenic organs (19,20). Basal level of cyp11a1 was increased in LXR^−/− mice (1.4-fold, P < 0.05) compared with wild-type mice. This could be partially responsible for the higher basal level of E2 in this group. No difference in 3βhsd (3-β-hydroxysteroid dehydrogenase I) expression was observed. Aromatase/cyp19A1 expression was also measured in the ovaries. Unexpectedly, its levels were basally lower than the wild-type animals. T1317 treatment had no effect on the various levels.

Ovarian hyperstimulation is associated with a deregulation of the inflammatory status in LXR-null mice

Measurement of the blood sedimentation speeds, a widely used clinical marker of general inflammatory status, at 1 and 2 h showed an increase in LXR^−/− mice (5.3- and 6.0-fold, respectively, P < 0.05) compared with wild-type mice (Fig. 5A). Real-time qPCR analysis of immunity markers and candidate genes involved in the progression of the symptoms (Fig. 5B) was performed in whole ovaries of T1317-gavaged wild-type and LXR^−/− animals.

Srebp1c, a known LXR-target gene, presented a 3.4-fold induction after treatment with T1317, demonstrating the efficiency of the treatment. Surprisingly, basal expression levels of the genes encoding LH receptor (LHR), FSH receptor (FSHR), and VEGF receptor 2 had significantly lower levels in the LXR^−/− group (from 25 to 55%). T1317 treatment induced a 60% decrease of fshr expression in wild-type females. IL-6 expression was increased by 1.75-fold in LXR-deficient mice compared with wild-type females. No difference in vegf, vegf receptor 1, cyclooxygenase 1 and 2, IL-1β, and tnfα expression levels were observed (data not shown). Similar results were obtained for claudin 5 and cadherin 5 encoding tight junction proteins whose deregulations are linked to OHSS (23,24) (data not shown). Thus, symptoms developed by LXR^−/− females are not underlined by all classical molecular mechanisms involved in OHSS but are associated with inflammatory response perturbations as in women developing the syndrome.

LXR deficiency leads to an acute ovarian response

Our work suggests for the first time an association between LXR signaling and OHSS. As for OHSS in women (5,25,26), LXR^−/− females show many signs of the abnormal ovarian response to an exogenous stimulation. Indeed a marked increase in expulsed follicles during ovulation followed by corpus luteum formation was observed as well as a higher production of estradiol. The poor quality of the oocytes retrieved in transgenic mice could result from the inability of these animals to perform a correct meiosis as described by Steffensen et al. (16).

It is now clear that OHSS progression in women is linked to high expression of cytokines and vascular permeability inducing factors. Among the many proinflammatory factors tested in this study, we identified IL-6 as an interesting candidate. Indeed, this factor is negatively regulated by LXRs in macrophages (7), and plasmatic and ascitic fluid IL-6 concentrations correlate with clinical features of OHSS (27). Our data suggest that this factor could be the molecular link between LXRs and the inflammatory status deregulation in these animals. Even though claudin 5 and cadherin 5 levels were unchanged, the increase in ovarian vascular permeability of LXR^−/− mice could be the result of a deregulation of another tight junction protein-encoding gene.

LXRs regulate ovarian endocrine function

We (19,20) and others (28) reported that activation of LXR could modulate the production of steroids in adrenal and testis. However, few studies have been available regarding the role of LXRs in the female reproductive physiology in vivo (16). Our work shows for the first time that activation of LXRs with a synthetic agonist increases estradiol levels in the serum, which seems to result from increased transcriptional regulation of star and cyp11a1. As already described for other LXR-target genes (19), the lack of LXRs in vivo is associated with a higher basal expression of cyp11a1.

Superovulated LXR^−/− female have high serum E2 levels despite a decreased concentration of gonadotropin receptor mRNAs. This paradoxical observation is explained by the fact that FSHR stimulation by its agonists leads to a decrease in the expression of its gene, both in vitro and in vivo (29). Because the LXR-deficient mice are more receptive to pregnant mare serum gonadotropin treatment, as pointed out by the oocyte counting, this could explain the highest decrease of fshr expression observed. In parallel, as shown by Tamura et al. (30), the high level of il-6 mRNA in the LXR^−/− females is associated with a decreased lhr expression, suggesting that il-6 deregulation is the primum movens of low lhr mRNA levels. A paradoxical point is that superovulated LXR^−/− mice have high E2 levels associated with normal progesterone levels. Unexpectedly, aromatase/cyp19A1 mRNA levels were lower in these animals. As suggested by Tamura et al. (31), the low aromatase expression could result from the high il-6 mRNA levels. It could therefore be hypothesized that estradiol increase in the LXR^−/− mice was also due to the up-regulation of liver Sult1e1 (32), which encodes estrogen sulfotransferase, an enzyme critical for metabolic estrogen deactivation.

It is important to note that circulating estradiol concentration in women undergoing controlled ovulation induction is often monitored because estrogens may be an index of ovarian response (4). If serum E2 is elevated, gonadotropin administration during treatment can be interrupted until E2 concentration reaches an acceptable range (33). Besides, whereas E2 is not the main factor influencing development of the OHSS in women, it is involved in the progression of the symptom.

Altogether, our data show that LXRs control oocyte production activity: LXR-deficient females have: 1) an increase of the inflammatory status possibly induced by an il-6 elevation and 2) high serum estradiol that could participate to the progression of the OHSS phenotype. LXRs are key regulators of the energetic metabolism. Thus, our data raise the question of a good control of the nutritional balance to improve the outcome of assisted reproductive technologies. LXR-deficient females are the first engineered mouse model linking OHSS and oxysterols signaling. They could help understanding the molecular mechanisms by which women with high body mass index are at risk of developing ovarian hyper stimulation syndrome.