Abstract
Estrogens act through nuclear and membrane-initiated signaling. Estrogen receptor alpha (ERα) is critical for reproduction, but the relative contribution of its nuclear and membrane signaling to the central regulation of reproduction is unclear. To address this question, two complementary approaches were used: estetrol (E4) a natural estrogen acting as an agonist of nuclear ERs, but as an antagonist of their membrane fraction, and the C451A-ERα mouse lacking mERα. E4 dose- dependently blocks ovulation in female rats, but the central mechanism underlying this effect is unknown. To determine whether E4 acts centrally to control ovulation, its effect was tested on the positive feedback of estradiol (E2) on neural circuits underlying luteinizing hormone (LH) secretion. In ovariectomized females chronically exposed to a low dose of E2, estradiol benzoate (EB) alone or combined with progesterone (P) induced an increase in the number of kisspeptin (Kp) and gonadotropin-releasing hormone (GnRH) neurons coexpressing Fos, a marker of neuronal activation. E4 blocked these effects of EB, but not when combined to P. These results indicate that E4 blocked the central induction of the positive feedback in the absence of P, suggesting an antagonistic effect of E4 on mERα in the brain as shown in peripheral tissues. In parallel, as opposed to wild-type females, C451A-ERα females did not show the activation of Kp and GnRH neurons in response to EB unless they are treated with P. Together these effects support a role for membrane-initiated estrogen signaling in the activation of the circuit mediating the LH surge.
Significance Statement
Estrogen receptor alpha (ERα) is critical for the activation of the neural circuits underlying ovulation. However, the relative contribution of its nuclear and membrane signaling to this neuroendocrine phenomenon is unclear. Using two complementary approaches to block membrane ERα signaling, the present study reveals that membrane ERα signaling is required for the activation by estrogens of gonadotropin-releasing hormone (GnRH) and kisspeptin (Kp) neurons, two key neuronal populations underlying the surge of luteinizing hormone which triggers ovulation. Interestingly, the absence of activation of Kp and GnRH neurons is alleviated in both models by progesterone (P). Collectively the results of these two approaches converge to provide evidence that membrane estrogen signaling contributes to this key event for the central regulation of reproduction.
Introduction
Gonadotropin-releasing hormone (GnRH) neurons stand at the top of the hypothalamus–pituitary–gonadal (HPG) axis that governs reproduction. Their activity drives the pulsatile release of gonadotropins to govern ovarian steroidogenesis and folliculogenesis. During most of the cycle, estrogens exert a negative feedback on GnRH and gonadotropin secretion. At mid-cycle, estrogens switch from negative to positive feedback to generate a continuous surge of GnRH and subsequently a luteinizing hormone (LH) surge which triggers ovulation (Herbison, 1998, 2020; Wang and Moenter, 2020). The mechanisms underlying the action of estrogens leading to the initiation of the preovulatory LH surge remain however unclear.
The nuclear estrogen receptor alpha (nERα) is the primary estrogen receptor (ER) involved in the central control of reproduction (Hamilton et al., 2014). As GnRH neurons do not express ERα (Herbison and Pape, 2001), the positive feedback is mediated by ERα-expressing afferents to GnRH neurons mainly originating from the anteroventral periventricular nucleus (AVPv; Wintermantel et al., 2006; Campbell and Herbison, 2007). In particular, kisspeptin (Kp) neurons exert a pivotal role in translating changes in circulating estrogens into changes in the activity of GnRH neurons and LH surge generation (Wang et al., 2016, 2018; Porteous and Herbison, 2019). Although other neuronal populations likely contribute to the estrogenic regulation of GnRH neurons, the current view posits Kp neurons located in the AVPv as key elements of the core surge generator (Goodman et al., 2022).
Estrogens act through nuclear and membrane-initiated signaling. Nuclear signaling regulates the transcription of target genes through direct interaction of the liganded receptor with an estrogen response element (ERE; classical genomic action) on the DNA or via protein–protein interaction with another transcription factor (tethered genomic action; McDevitt et al., 2008). Upon palmitoylation, ERs are translocated to the membrane where they can signal to activate intracellular signaling cascades (Arnal et al., 2017; Acconcia et al., 2021). Additionally, estrogens also act on membrane-specific G-protein-coupled receptors such as GPER1 (Kelly and Rønnekleiv, 2015). While nuclear actions lead to relatively slow and long-lasting effects, membrane-initiated actions occur within seconds to minutes (Kelly and Rønnekleiv, 2015; Balthazart, 2021).
Whether the central regulation of LH surge involves nuclear- or membrane estrogen-initiated signaling or a combination of both is currently unclear. Early evidence indicated that a prolonged exposure to high circulating estrogens is required to elicit an LH surge (Legan et al., 1975; Evans et al., 1997), suggesting that classical estrogen signaling is involved. This is supported by reports indicating that ERE-independent ERα activity alone is not sufficient to restore E2-induced changes in the firing rate of GnRH neurons or the LH surge (Glidewell-Kenney et al., 2007; Christian et al., 2008). However, that transcriptional signaling is required does not preclude a role of membrane-initiated signaling. Moreover, membrane-initiated estrogen signaling also influences GnRH neurons in vitro (Herbison, 2009; Moenter and Chu, 2012; Terasawa and Kenealy, 2012). While ERβ (Abraham et al., 2003; Chu et al., 2009) or membrane-specific estrogen receptors, such as the STX-activated receptor (Zhang et al., 2010) or GPER1 (Sun et al., 2010), appear to mediate a direct action of estrogens on the activity of GnRH neurons, ERα would mediate indirect estrogenic actions by modulating inputs to GnRH neurons (Romano et al., 2008; Chu et al., 2009; Romanò and Herbison, 2012). In particular, membrane ERα (mERα) stimulates neuronal activity and contributes to the regulation of Kp expression in immortalized Kp neurons with features of AVPv Kp neurons (Mittelman-Smith et al., 2015). Evidence obtained in vitro also indicates that the activation of mERα mediates the synthesis of neuroprogesterone by rostral hypothalamic astrocytes (Micevych et al., 2007; Kuo et al., 2010; Mohr et al., 2022), whose action on Kp neurons is necessary for both Kp release (Mittelman-Smith et al., 2018; Mohr et al., 2021) and LH surge induction (Micevych et al., 2003; Mohr et al., 2019; Chuon et al., 2022). Thus, mERα signaling appears to be able to modulate the activity of AVPv Kp neurons both directly and indirectly in vitro. Yet, to our knowledge, whether direct or indirect, a role for membrane estrogen signaling on the activation of Kp neurons and the subsequent activation of GnRH neurons has never been demonstrated in vivo.
The present study took advantage of genetic and pharmacological complementary approaches to explore the role of mERα on the central regulation of LH surge. First, a knock-in mouse model with a point mutation of the palmitoylation site Cys451 into an alanine leads to a selective loss of function of ERα membrane signaling, allowing to dissociate the two modes of action of estrogens on ERα (Adlanmerini et al., 2014; Pedram et al., 2014). Second, estetrol (E4) is an estrogen exclusively synthesized in human fetal liver which selectively binds ERα and ERβ with a lower affinity than E2 (Holinka et al., 2008). E4 presents unique properties allowing to distinguish nuclear and membrane estrogen signaling in rodents notably, as it mimics estrogenic actions induced via the activation of nuclear ERα but antagonizes membrane ERα in different tissues (Gérard et al., 2022). E4 inhibits ovulation when administered alone in rats (Coelingh Bennink et al., 2008) and when combined with progesterone (P) in humans (Duijkers et al., 2015; Apter et al., 2016). E4 is now included in an oral contraceptive formulation (Klipping et al., 2021). However, its central mechanism of action on the HPG axis remains unknown.
Materials and Methods
Animals and general procedures
All wild-type (WT-ERα) and C451A-ERα mice of the CD1 strain, obtained by backcrossing the original C451A-ERα mice (C57Bl/6) into the CD1 background (Adlanmerini et al., 2014), were housed and bred in the animal facility of the University of Liège. Mice were genotyped by PCR analysis of DNA collected from the tail as described previously (Adlanmerini et al., 2014). Mice were weaned at 3–4 weeks of age and housed in same-sex cages. All animals had ad libitum access to food and water. The room temperature was maintained at 24 ± 2°C. Animals were housed under a reversed 12 h light/dark cycle (lights on at 1 A.M.) when tested for positive feedback (Exp. 1 and 2). All experimental procedures were in accordance with laws on the “Protection and Welfare of Animals” and on the “Protection of Experimental Animals” and were approved by the Ethics Committee of the University of Liège.
General procedures
Surgery
Between 2 and 3 months of age, females were bilaterally ovariectomized (OVX) under general anesthesia using a mixture of Domitor (Domitor, Pfizer, 1 mg/kg) and medetomidine (Ketamine, 80 mg/kg) administered subcutaneously (s.c.). In some experiments, animals were implanted at the time of ovariectomy with a subcutaneous Silastic capsule filled with E2. At the end of surgery, medetomidine-induced effects were antagonized by atipamezole (Antisedan, Pfizer, 4 mg/kg, s.c.) to accelerate recovery.
Hormones
17β-Estradiol (E2, E8875), β-estradiol-3-benzoate (EB, E8515), and progesterone (P, P0130) were purchased from Sigma-Aldrich and dissolved in sesame oil, used as vehicle, unless stated otherwise. EB (1 µg, s.c.) and P (500 µg, s.c.) were injected subcutaneously, while E2 (1 µg diluted in 7.35 µl of sesame oil/20 g of body weight) was provided through subcutaneous Silastic capsules (inner diameter, 1.02 mm; outer diameter, 2.16 mm; Dow Corning) which yield physiological circulating E2 concentration (Bronson, 1981). Estetrol (E4) was provided by Mithra Pharmaceuticals and dissolved in sesame oil with 5% ethanol (0.2 mg, 50 µl, s.c.). Unless stated otherwise, treatments were counterbalanced across housing cages, such that each/every cage contained animals with different treatments.
Blood collection
Depending on the question and the method used for blood analysis, blood drops or trunk blood were collected. For repeated sampling of blood drops on a same day or assay with ultrasensitive immune-enzyme assays [EIA; Exp. 1-part1 (1.1)], blood was collected using the repetitive tail-tip blood sampling (Czieselsky et al., 2016). Briefly, mice were habituated to handling for a few minutes while massaging the tail every day during 2 or 3 weeks. For blood drop collection, a single excision of the tail tip was made with a razor blade. When females were OVX (regardless of whether they were treated with EB and/or P), one blood sample (5.2 µl) was collected with a pipette and immediately diluted in 98.8 µl phosphate-buffered saline with 0.05% of Tween 20 (PBST), quickly frozen in dry ice, and stored at −80°C until further use. In Exp. 1.1, blood drops were collected every 30 min for 4 h. For Exp. 1.2, mice were placed under a red lamp to allow dilation of blood vessels and were briefly restrained in the immobilizing cage where a single excision of the tail with a razor blade was made. Blood (200 µl) was collected in heparinized microhematocrit capillary tubes filled by capillarity. The tail was massaged to facilitate blood dripping. Blood was stored in a 1.5 ml microfuge tube containing a drop of heparin (Leo, 012866-08, 5,000 U.E/ml). Blood was centrifuged 10 min at 1,500 × g at 4°C, the plasma was collected and stored at −80°C until quantification by radioimmunoassay (RIA). At the end of experiments (Exp. 1.2 and Exp. 2), trunk blood was also collected in 1.5 ml microfuge tubes containing a drop of heparin. Plasma was collected as previously and stored at −80°C until further use.
LH assay
Two methods were used to assay LH: an ultrasensitive sandwich ELISA and a classical RIA. The ultrasensitive sandwich ELISA was used for blood drops [Exp. 1.1 and 1.2 (day 39)], while the RIA was used for all the other types of blood samples (Exp. 1.2 and Exp. 2).
We used the sensitive sandwich ELISA previously described and validated (Steyn et al., 2013) with few modifications. Briefly, 96-well high-affinity binding microplates (9018, Corning) were coated with 50 µl of a monoclonal antibody directed against bovine LH beta subunit (1:1,000; 518B7; RRID: AB_2665514, University of California, UC Davis) and incubated overnight at 4°C. Unspecific binding was blocked by incubating each well with 200 µl of blocking buffer for 24 h at 4°C. Samples (50 µl) and LH standards [50 µl; generated by serial twofold dilution of mouse LH starting at 400 pg/well until 0,19 pg/well, AFP-5306A, National Institute of Diabetes and Digestive and Kidney Diseases – National Hormone and Pituitary Program (NIDDK-NHPP)] were incubated for 2 h before adding 50 µl of detection antibody (1:10,000; polyclonal antibody, rabbit LH antiserum, AFP240580Rb; RRID:AB_2665533, NIDDK-NHPP) for 1.5 h at room temperature (RT). A horseradish peroxidase-conjugated polyclonal Goat Anti-Rabbit antibody (50 µl, 1:2,000; P0448, Dako; RRID:AB_2617138) was added in each well for 1.5 h at RT. Then, the substrate of the peroxidase (100 µl, 3,3′,5,5′-tetramentylbenzidine solution; 1-Step Ultra TMB-ELISA, 34029, Thermo Fisher Scientific) was added in each well for 10 to 25 min at RT and in darkness. The reaction was stopped by 3 M HCl (50 µl). The absorbance of each well was read at a wavelength of 450 nm and at a wavelength of 650 nm (background). The optical density (OD) obtained at 650 nm was subtracted from this obtained at 450 nm. The amount of LH present in each well was determined by interpolating the resulting OD of unknown samples against a nonlinear regression of the OD of the LH standard curve (GraphPad Prism 8). Standards were run in duplicate and yielded a nonlinear curve fitting with a R2 > 0.95. The sensitivity of the assay was 0.03 ng/ml. All samples from a same mouse were assayed on the same plate, and genotypes and treatments were counterbalanced within plates. The intra- and inter-assay coefficients of variation were <10 and 15%, respectively.
The RIA consisted of a double antibody method with reagents provided by the National Institutes of Health [Dr. A. F. Parlow, National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Hormone and Peptide Program, Torrance, CA]. LH was detected by a rat LH-I-10 (AFP-11536B) labeled with 125I and precipitated with a Rabbit anti-mouse LH (AFP-240580; RRID: AB_2784499). Mouse LH reference preparation (AFP-5306A) was used to prepare the standard curve. The intra- and inter-assay coefficients were <10 and 7%, respectively, and the sensitivity of the method was set at 4 pg/100 ml based on the lowest detectable point of the standard curve.
The values of LH concentrations obtained for each animal on the day of LH induction were compared with the average values measured in all samples within each genotype collected on the morning of the day preceding the LH surge induction. This average plus two times the standard deviation was considered as the threshold for considering an LH surge (Dror et al., 2013). The percentage of animals that presented a surge was then calculated for each group.
Euthanasia
Animals were humanely anesthetized with isoflurane and decapitated 30 min after lights off (Exp. 1 and 2). Their brain was then removed from the skull and immersed in a solution of 0.5% of acrolein in 0.01 M PBS for 2 h at RT. For this type of fixation, brains were rinsed thrice for 30 min in PBS before being transferred in 30% sucrose overnight. Brains were then frozen on dry ice and stored at −80°C until further use. All brains were cryosectioned in four series of 30 µm thick coronal slices from the corpus callosum level to the end of the hypothalamus. Sections were stored in antifreeze solution and kept at −20°C.
Histology and immunostaining
Brains were double labeled for Fos and Kp or GnRH. Briefly, brain sections were first rinsed three times for 5 min in 0.05 M Tris-buffered saline (TBS), pH 7.6, at RT. Unless mentioned otherwise, all following incubations were carried out at RT and followed by similar rinses. Sections were first incubated in 0.1% sodium borohydrate for 15 min. They were then incubated in hydrogen peroxide (H2O2, 1% for 20 min) to block endogenous peroxidase activity. Sections were blocked and permeabilized for 1 h in normal goat serum (NGS) in TBS with 0.1% Triton X-100 (TBST) and immediately incubated at 4°C in the primary antibody against the N terminus of human Fos [overnight, 1:2,000; Rabbit polyclonal, ABE457, Millipore; RRID: AB_2631318 (Alvisi et al., 2016; Exp. 1.2); overnight, 1:2,000; monoclonal antibody, sc-166940, Santa Cruz Biotechnology; RRID: AB_10609634 (Exp. 2)] in NGS and TBST. Sections were then incubated for 2 h in a goat anti-rabbit biotinylated antibody (111-065-003; RRID: AB_2337959; Jackson ImmunoResearch) followed by 1 h in the AB complex solution (PK-6100; Vector Laboratories) diluted at 1:400 or 1:800 (for Fos when followed by GnRH labeling or for Fos when followed by Kp labeling, respectively). The immunoproduct was visualized with 0.05% diaminobenzidine with 0.012% H2O2 in TBS.
The first visualization was followed by a blockade of avidins and biotins using avidin-biotin blocking kit (SP-2001; Vector Laboratories) for 15 min prior to an additional blocking and permeabilization step. Sections were immediately incubated overnight in a polyclonal rabbit antibody directed against GnRH-I [1:400, polyclonal, #20075, Immunostar; RRID: AB_572248 (Memi et al., 2013)] or twice overnight in a rabbit antibody directed against mouse Kp [1:10,000; rabbit polyclonal, Ac566 kindly provided by Isabelle Franceschini and Massimiliano Beltramo, INRA, Nouzilly, Tours, France; RRID: AB_2296529 (Clarkson and Herbison, 2006)] in NGS and TBST. Sections were then incubated in a goat anti-rabbit biotinylated secondary antibody (111-065-003; Jackson ImmunoResearch). Finally, the immunoproduct was visualized by a last incubation in the substrate of the Vector SG Peroxidase Substrate Kit (SK-4700; Vector Laboratories). After final rinses, sections were mounted on microscope slides and coverslipped with Eukitt (Sigma-Aldrich).
Image analysis
The number of single-labeled Kp-immunoreactive (IR) neurons or the number of Kp-IR and GnRH-IR neurons colabeled with Fos was analyzed by direct observation at 40× magnification using Leica DMRB microscope. The number of Kp-IR cell bodies was investigated bilaterally in 10 consecutive brain sections (each separated by a distance of 90 µm) encompassing the AVPv and the rostral periventricular nucleus (PeN) continuum [corresponding to plates 29–35 of the Paxinos Mouse Atlas (Franklin and Paxinos, 2001)]. The number of GnRH-IR cell bodies was analyzed bilaterally in 10 consecutive brain sections (each separated by a distance of 90 µm) corresponding to plates 21–31 of the Paxinos Mouse Atlas (Franklin and Paxinos, 2001). Kp and GnRH immunolabeling is cytoplasmic, while Fos immunolabeling is detectable only in the nucleus. All Kp or GnRH neurons detected in this region were counted and analyzed for the presence of nuclear immunostaining for Fos. The values obtained for each side of the 10 sections were summed to provide a total number of Kp or GnRH expressing neurons and the percentage of Kp or GnRH neurons coexpressing the protein Fos.
Experimental designs
Experiment 1—positive feedback
The role of mERα in the induction of LH surge was repeatedly assessed in two cohorts of 2-month-old WT-ERα (Cohort 1: n = 24; Cohort 2: n = 17) and C451A-ERα (Cohort 1: n = 18; Cohort 2: n = 19) females. The two cohorts were subjected to the exact same protocol except that females from the second cohort were housed based on their treatment. In each cohort, females were tested twice following a paradigm of LH surge induction, i.e., by implantation of a subcutaneous capsule delivering low levels of E2 mimicking diestrus levels and administration of EB 7–8 d after OVX (Fig. 1A). The first test was designed to examine the time-response profile of the EB-induced LH surge following blood sampling every 30 min for 4 h [Part 1 (Exp. 1.1), Days 0–8], while the second investigated the central activation of the circuit underlying the LH surge [Part 2 (Exp. 1.2), Days 30–39].
Briefly, females were OVX and implanted with a subcutaneous capsule containing E2 (1 µg). A first blood sample was collected on Day 6 post-OVX between 08.20 A.M. and 09.00 A.M. (3 µl immediately diluted in 57 µl of PBST for EIA). Females of each genotype were subdivided in three groups of equal size subjected to three different hormonal treatments (s.c.): veh + veh, EB + veh, and EB + P. On Day 7 (10 A.M.), they were injected with EB or its vehicle (veh). On Day 8 (10 A.M.), they were injected with P or veh 3 h before lights off, while females that had received veh on Day 6 received veh again. Blood sampling was then carried out every 30 min for 4 h starting 60 min before lights off. All samples were assayed in duplicate. Three to 7 d later, their implant was removed, and they were treated every 3 or 4 d with EB until the beginning of the second part.
Part 2 started 30 d after Part 1. Females were reimplanted with a new subcutaneous E2 implant. Two blood samples were collected on Day 38 between 8 A.M. and 9 A.M.: 5.2 µl immediately diluted in PBST for EIA and 200 µl for plasma collection and RIA. Females were then treated with veh or EB at ∼10 A.M. The next day (Day 39), veh or P was injected 3 h before lights off. Mice were anaesthetized with isoflurane 30 min after lights went off and killed by rapid decapitation. Trunk blood was collected, extracted for plasma as described above, and assayed by EIA. Brains were fixed in 0.5% acrolein (Fig. 2A).
Experiment 2—E4 and positive feedback
This experiment investigated the effect of E4 on the induction of LH surge in WT-ERα females (n = 45) subjected to a classical paradigm of induction of the LH surge by administration of EB with or without P in OVX females chronically exposed to low estrogen levels mimicking diestrus levels (Fig. 3A). Briefly, females were OVX and implanted with a subcutaneous E2 capsule. Prior to treatment, one blood sample (200 µl) was collected on Day 8 after OVX. Females were subdivided into five groups and subjected to five different hormonal treatments: veh + P, EB + veh, EB + P, EB + E4, and EB + E4 + P. On Day 8 (10 A.M.), they were injected with veh, EB, or EB + E4. On Day 9 (10 A.M.), they were injected with P or veh 4 h before lights off. Females were killed by rapid decapitation within 1 h after lights off, trunk blood was collected, and the brain was dissected out of the skull and fixed in 0.5% acrolein.
Statistical analysis
All statistical analyses were performed using Prism 8 (version 8.0.0, GraphPad Software). Continuous data were analyzed by parametric unpaired Student’s t tests and two-way ANOVAs or by nonparametric Mann–Whitney and Kruskal–Wallis tests when the normality and homoscedasticity assumptions were violated. Significant parametric and nonparametric ANOVAs were followed by Tukey' and Dunn’s post hoc tests, respectively. Contingency data were analyzed by Fisher’s exact tests. Bonferroni’s correction was applied when multiple Mann–Whitney tests were applied to a data set. The resulting p value is then called adjusted p value (padj). Due to technical issues such as the loss or the degradation of sections during processing, the final sample size may differ from the initial number of samples collected, thus explaining the variability in the degrees of freedom between analyses of samples originating from the same experiments. Effects sizes from ANOVA (partial eta squares, ηp2) were calculated based on the sums of squares provided by the ANOVAs or using calculators available at https://www.psychometrica.de/effect_size.html for Kruskal–Wallis analyses. Effect sizes for Student’s t or Mann–Whitney test (Cohen’s d) were obtained using calculators available at https://wwhttps://www.psychometrica.de/effect_size.html. Results were considered significant when p < 0.05. All results are represented as means ± SEM unless mentioned otherwise.
Results
Are C451A-ERα mice able to show an LH surge in response to EB and is P necessary?
Although the paradigm of rising E2 levels can induce an LH surge in the absence of P, the combination of E2 and P yields changes of higher amplitude (Bronson and Vom Saal, 1979; Waring and Turgeon, 1992). Therefore, the first experiment investigated the role of mERα on the LH surge profile induced by EB combined or not with P. OVX females were implanted with a capsule delivering low E2 amounts mimicking circulating E2 levels at diestrus (Dror et al., 2013), and blood was collected by tail-tip blood sampling every 30 min for 4 h starting 1 h prior to lights off (Exp. 1). This experiment was conducted in two cohorts of mice, subjected to the exact same protocol, whose data were pooled. First, looking at baseline LH levels (6 d after OVX and implantation of a subcutaneous capsule delivering low levels of E2), C451A-ERα females showed significantly higher LH levels than their WT-ERα littermates (WT-ERα, median = 1.4 ng/ml, n = 42; C451A-ERα, median = 21.6 ng/ml, n = 36; U = 47, p < 0.0001, d = 2.474; Fig. 1B), indicating that C451A-ERα females may present some impairment of the negative feedback.
Profiles of LH changes induced by estradiol benzoate (EB) alone or in combination with progesterone (P) in ovariectomized WT-ERα (white) or C451A-ERα (gray mice). A, Protocol used to induce a positive feedback: females were ovariectomized (OVX), chronically treated with estradiol (E2) from day 0 to day 8, injected with estradiol benzoate on day 7, and injected with progesterone or its vehicle (sesame oil) on day 8. B, On day 6, C451A-ERα females (n = 32) showed higher baseline LH levels than WT-ERα females (n = 31; Mann–Whitney test). C, D, Profiles of LH levels measured every 30 min starting 1 h before lights off following treatment on day 8 in WT-ERα and C451A-ERα females, respectively. C, LH profiles obtained in WT-ERα mice (OVX + E2 + veh + veh n = 12, OVX + E2 + EB + veh n = 11, OVX + E2 + EB + P n = 14). D, LH profiles obtained in C451A-ERα mice (OVX + E2 + veh + Veh n = 10, OVX + E2 + EB + veh n = 11 OVX + E2 + EB + P n = 12). E, Regardless of treatment, WT-ERα females showed an increased LH concentration at one time point (peak) between 0 and 2.5 h after lights off compared with prior (day 6, pre) and during 3 h after lights off (post; two-way ANOVA; Tukey’s post hoc test following significant time effect: ++p < 0.01 vs “pre”; †p < 0.05 vs “post”). F, EB + P induced an increased LH concentration in C451A-ERα females within 0 and 2.5 h after lights off (peak) compared with prior (day 6, pre) and during 3 h after lights off (post; two-way ANOVA; Tukey’s post hoc test, following significant interaction: +++p = 0.001 vs “pre” within same treatment; †††p = 0.001 vs “post“ within same treatment; ∇p < 0.05 EB + P “pre” vs “post” within same treatment. Symbols in the statistical boxes: *, **, ***, p < 0.05, 0.01, 0.001; N.S., nonsignificant).
The qualitative analysis of the average profiles of LH concentration measured every 30 min on Day 8 indicates that treatment with EB + P resulted in an increased LH concentration, while no surge was induced neither in the control condition (veh + veh) nor following EB alone in both WT-ERα and C451A-ERα mice (Fig. 1C,D). Of note, in WT-ERα, LH began to rise before lights off, peaked at lights off, and slightly decreased afterward while remaining elevated for the next 3 h (Fig. 1C), while in C451A-ERα mice, the LH surge began with a slight delay compared with WT-ERα, peaked 30 min after lights off and remained elevated for the next 2.5 h (Fig. 1D). Interestingly, in C451A-ERα females treated with EB + P and EB + veh, the first time point (−1 h) shows a clear decrease in LH concentration compared with the measure taken 48 h earlier (day 6) potentially reflecting a negative feedback exerted by EB. In both genotypes, there is a large variability around the mean for most time points which is explained by the variability in individual profiles. Additional work is warranted to confirm the existence of a delay in the response of C451A-ERα females.
For analysis purposes, the highest LH concentrations obtained in each animal between 0 and 2.5 h after lights off (Peak) were averaged across females and compared with the concentration measured 48 h before (Pre) and 3 h after (Post) lights off (Fig. 1E,F). Confirming the qualitative observations, no LH surge was observed following treatment with veh or EB alone in both genotypes. In WT-ERα, the analysis revealed no effect of treatment (F(2,33) = 1.488; p = 0.2405;
Three weeks later, the same mice were then subjected to the same protocol with minor changes. Their blood and brain were collected between 30 min and 1 h after lights off to evaluate the impact of the mutation on the neuronal circuits underlying the induction of a LH surge by estrogens. As before, C451A-ERα showed higher LH concentrations than WT-ERα prior to EB (WT-ERα: 0.95 ng/ml ±0.17, n = 16, C451A-ERα: 14.89 ng/ml ±1.80, n = 15; t(29) = 7.952, p < 0.001, d = 2.858). The analyses of blood samples collected at euthanasia identified an increase in LH in WT-ERα females treated with EB + P, but not with veh + EB compared with veh + veh (H = 10.02; p = 0.0067;
Effect of mERα absence on the positive feedback of estrogens on LH concentration and the activation of the associated neurocircuits. A, Protocol used to induce positive feedback: following a first round of injections to induce the positive feedback (Fig. 1), the E2 implant was replaced by a new one on Day 30, and females were treated again with veh + veh, EB + veh, or EB + P on Days 38 and 39. Blood and brains were collected 30–60 min after lights off. B, In WT-ERα females (white), EB + P, but not EB + veh, induced a significant rise in LH (Kruskal–Wallis test: **p < 0.01 vs veh + veh), while in C451A-ERα females (gray), EB + veh induced a significant reduction in LH (Kruskal–Wallis test: *p < 0.05 vs veh + veh; Mann–Whitney tests: ##, ### < 0.01, 0.001 vs WT-ERα within same treatment). C, WT-ERα females displayed more kisspeptin (Kp) neurons in RP3 V (AVPv + PeN) than C451A-ERα females (two-way ANOVA). D, A higher percentage of Kp neurons coexpressed Fos following EB and EB + P than veh + veh in WT-ERα, while only EB + P induced such activation in C451A-ERα (two-way ANOVA; * and ***, p < 0.05 and 0.001 vs veh + veh same genotype; $$$p < 0.0001 vs EB + veh same genotype; ###p < 0.001 vs same treatment in WT-ERα). E, GnRH neurons counted in POA were slightly more abundant in C451A-ERα females than in WT-ERα females (two-way ANOVA). F, A higher percentage of GnRH neurons coexpressed Fos following EB and EB + P than veh + veh in WT-ERα, while only EB + P induced such activation in C451A-ERα (two-way ANOVA; ***, p < 0.001 vs veh + veh same genotype; $$$p < 0.0001 vs EB + veh same genotype; ##p < 0.01 vs same treatment in WT-ERα). Sample size: B, C. WT-ERα: veh + veh, n = 14, veh + EB, n = 14, EB + P, n = 13, C451A-ERα: veh + veh, n = 11, veh + EB, n = 11, EB + P, n = 13. C–F. WT-ERα: veh + veh, n = 13, veh + EB, n = 14, EB + P, n = 13, C451A-ERα: veh + veh, n = 12, veh + EB, n = 11, EB + P, n = 13. Symbols in the statistical boxes: *, **, ***, p < 0.05, 0.01, 0.001; N.S., nonsignificant.
The brains of these females were then immunostained for Kp (Fig. 3) and GnRH (Fig. 4) along with Fos to determine the effect of the mutation on the activation of the hypothalamic circuits underlying the LH surge (Clarkson et al., 2008; Gonzalez-Martinez et al., 2008). This neuronal response is considered a more reliable index of surge initiation than LH itself (Clarkson et al., 2023). The analysis of the total number of Kp neurons in the AVPv-PeN continuum revealed a reduced number of Kp neurons in C451A-ERα females compared with their WT-ERα littermates (F(1,70) = 38.61; p < 0.0001;
Representative photomicrographs of Kp-IR neurons (in blue) and their coexpression of the neuronal activity marker Fos (in orange) as a function of the treatment and genotype. Black arrows point at double-labeled neurons, while white arrows point at single-labeled neurons.
Representative photomicrographs of GnRH-IR neurons (in blue) and their coexpression of the neuronal activity marker Fos (in orange) as a function of the treatment and genotype. Black arrows point at double-labeled neurons, while white arrows point at single-labeled neurons.
The analysis of the percentage of Kp and GnRH neurons colabeled with Fos revealed a very different pattern of response between genotypes (Figs. 2–4). In WT-ERα, EB administered alone or along with P activated a higher percentage of Kp neurons. In contrast, only EB + P elicited such an increase in C451A-ERα females. A two-way ANOVA indeed identified a trend toward a genotype effect (F(1,70) = 3.735; p = 0.0573;
Similarly, the percentage of GnRH neurons colabeled with Fos increased after EB alone and EB + P in WT-ERα females, while only EB + P resulted in such an increase in C451A-ERα females, which resulted in a genotype effect (F(1,69) = 8.481; p = 0.0048;
The percentages of activated Kp and GnRH neurons correlate with circulating LH concentrations in WT-ERα females treated with EB + P (Kp, R = 0.6314, p = 0.0206; GnRH, R = 0.8388, p = 0.0003), while it is not the case for the circulating LH in C451A-ERα females (C451A-ERα, Kp, R = 0.1253, p = 0.6835; GnRH; R = 0.1117, p = 0.7163; data not shown). This difference could be explained by the fact that brains and bloods were collected too early to detect the surge in most individuals. This interpretation goes along with the relatively low percentage of animals displaying a surge, regardless of treatment and genotype, but even less so in C451A-ERα females.
Does E4 block the LH surge induced by estradiol benzoate (EB)?
This lack of activation of Kp and GnRH neurons in C451A-ERα females treated with EB alone but not with EB + P suggested that mERα signaling is required for the activation of the neural circuitry underlying LH surge generation by EB but that P can bypass the effect of mERα. This latter effect could be interpreted as an indirect confirmation of the role of mERα for neuroprogesterone synthesis and its pivotal role for the activation of this circuit. As E4 was described as an antagonist of mERα (Gérard et al., 2022), we wondered whether E4 could block the LH surge induced by EB and whether this effect could be prevented by P.
This experiment followed a similar design as the second part of the previous experiment, except that in this experiment five treatments (veh + P, EB, EB + P, EB + E4, EB + E4 + P) were compared in wild-type mice (Fig. 5A). As expected, LH levels assayed on samples collected before treatment (day 8) did not differ between groups (F(4,40) = 0.4620; p = 0.7631;
Effect of estetrol on the LH surge induced by estradiol and the neurocircuits underlying this response. A, Protocol used to induce a positive feedback. WT mice were ovariectomized (OVX) on day 0, treated with subcutaneous estradiol (E2) implant from day 0 to day 9, and injected on day 8 with estradiol benzoate (EB) alone or combined with estetrol (E4, 200 µg, s.c.) or their vehicle (sesame oil) and on day 9 with progesterone (P) or its vehicle (sesame oil). Blood samples were collected prior to treatment on day 8 and within 1 h of lights off on day 9, when brains were also collected for immunohistological analyses. B, LH levels did not differ between groups (n = 9) on day 8. C, Females treated with EB alone (n = 9) or EB + E4 (n = 9) did not show a LH surge compared with veh + veh (n = 9) unless they were treated with P (EB + P, n = 9, and EB + E4 + P, n = 9). D, F, The number of kisspeptin (Kp) neurons in RP3V (AVPv + PeN, D) or GnRH neurons in POA (F) did not differ across treatments (Kp: veh + veh, n = 7, EB, n = 9, EB + P, n = 7, EB + E4, n = 6, EB + E4 + P, n = 7; GnRH: veh + veh, n = 9, EB, n = 8, EB + P, n = 9, EB + E4, n = 8, EB + E4 + P, n = 8). E, G, The percentages of Kp (E) and GnRH (G) neurons coexpressing Fos were higher in females treated with EB, EB + P, and EB + E4 + P than females treated with veh + P and EB + E4 (same sample sizes as in D and F). All data were analyzed by one-way ANOVA followed by Tukey’s post hoc test when significant: *, **, and *** p < 0.05, 0.01, and 0.001 versus veh + P; #, p < 0.05 versus EB; Δ, ΔΔ, and ΔΔΔ, p < 0.05, 0.01, and 0.001 versus EB + E4.
As previously, the brains of these females were immunostained for Kp or GnRH along with Fos to determine the effect of E4 on the activation of the hypothalamic circuits underlying the LH surge. The total number of Kp neurons in the AVPv-PeN continuum (Fig. 5D) and preoptic GnRH neurons (Fig. 5F) did not differ between treatments (Kp: F(4,31) = 0.4461, p = 0.7744,
Together, these results indicate that, in the absence of exogenous P, E4 prevents the activation of the neural circuit underlying the induction of an LH surge.
Discussion
The present results indicate that, in the absence of P, a constitutive lack of mERα signaling as well as an acute treatment with E4 prevent the ability of E2 to activate Kp and GnRH neurons which are key neuronal populations for the LH surge generation. These pronounced effects (with
It should be noted however that the present data cannot extent this conclusion to the LH surge itself due to a lack of statistically significant LH surge in EB-treated WT females, despite numerous females showing higher LH than the average of the control group. cFos expression represents a transcriptional coupling to various types of stimuli, which reflects synaptic activation, accompanied or not by concurrent spike activity, mainly associated with an increased calcium influx and the activation of the MAPK pathway leading to the activation of the AP1 pathway and of late genes (Morgan and Curran, 1989; Luckman et al., 1994; Kovacs, 2008; Chung, 2015; Hudson, 2018). Increased cFos expression has long been used as a cell-specific marker of neuronal activity, notably in GnRH and Kp neurons in the context of LH surge induction (Hoffman et al., 1993; Clarkson et al., 2008, 2023; Dror et al., 2013). Transient cFos expression requires strong synaptic activation and is detected as a protein between 45 min and 3 h (peaking between 90 and 120 min; Kovacs, 2008). Such an extended time window of detection following stimulation leaves room for a mismatch between the measure of neuronal activation and the detection of a rise in LH. As previously shown, the amplitude of LH surge induced by EB alone is lower than this induced by the EB + P combination (Bronson and Vom Saal, 1979; Waring and Turgeon, 1992), which limits the detection of the surge. Moreover, the onset of LH surge is notoriously highly variable (Czieselsky et al., 2016). As blood samples were collected shortly after lights off, it is possible that LH surges of lower amplitude in this experimental group have been missed. Finally, C451A-ERα mice present elevated basal LH concentrations that could limit the detection of a surge in terminal blood. Therefore, given that the activation of Kp and GnRH neurons is considered as a reliable marker of the LH surge especially when only one blood collection is available (Clarkson et al., 2023) and the mechanisms underlying the positive and negative feedbacks operate independently from each other (Herbison, 2020; Goodman et al., 2022), the discussion of the present results will focus on the role of mERα in the activation of the neurocircuits underlying the induction of the LH surge rather than the surge itself.
C451A-ERα females show a distinct phenotype of LH secretion
The idea that the positive feedback of estrogens depends on nuclear estrogen signaling is mainly based on the observation that the induction of a LH surge requires a prolonged exposure to high estrogen levels (Legan et al., 1975; Evans et al., 1997). Moreover, restoring ERE-independent ERα signaling had failed to restore the capacity to mount an LH surge in response to estrogens in ERαKO mice indicating that nonclassical signaling alone is not sufficient for positive feedback (Glidewell-Kenney et al., 2007). However, previous evidence supports the existence of a cooperation between nuclear- and membrane-initiated estrogen signaling (Vasudevan et al., 2001; Seredynski et al., 2013). Therefore, it is likely that membrane estrogen signaling requires nuclear estrogen signaling to properly function even if classical signaling constitutes the prime requisite for LH induction. To test this possibility, we used two complementary approaches. With a point mutation at the site of palmitoylation of ERα, C451A-ERα mice allow the study of the impact of a lack of membrane signaling of ERα while preserving its nuclear activity (Adlanmerini et al., 2014). Although this mutation does not seem to alter the sexual differentiation of females (Khbouz et al., 2019), the constitutive absence of mERα is an obvious limitation of such a model that may result in developmental defects and/or compensations. On the other hand, the antagonistic action of E4 on the membrane estrogen signaling of the classical estrogen receptors ERα (and possibly ERβ) provides a mean to circumvent developmental deficits or compensation. Although the preference of E4 for ERα over ERβ, GPER1, or STX-activated receptor as well as its properties in the brain (notably whether it activates or inhibits them) remain poorly documented, the comparison of effects obtained with both approaches provides confidence that membrane signaling plays a role in the activation of the circuit underlying the positive feedback of estrogens on LH secretion.
Gonadally intact C451A-ERα females exhibit elevated LH levels and fewer corpus lutea than wild-type females (Adlanmerini et al., 2014), suggesting a potential role of mERα signaling in both negative and positive feedbacks. The C451A mutation does not alter brain expression of ERα and Kp (Khbouz et al., 2019), suggesting the preserved transcriptional activity of the nuclear fraction of the receptor. The present study reveals however that C451A-ERα mice exhibit a distinct profile of LH secretion in response to estrogens compared with other ERαKO mouse models. Indeed, contrasting with ubiquitous ERαKO, neuron-specific ERαKO, and ARC specific ERαKO mice, which exhibit altered LH responses to ovariectomy and/or E2 (Wersinger et al., 1999; Cheong et al., 2014; Yeo and Herbison, 2014), C451A-ERα females respond to both ovariectomy (Fig. 1B) and provision of exogenous estrogens in the context of negative feedback (Faure et al., Submitted). However, C451A-ERα females are also unable to respond to increasing E2 levels by the activation of Kp and GnRH neurons (Fig. 2) which is typical following a surge induction protocol including P (Gonzalez-Martinez et al., 2008; Szymanski and Bakker, 2012). This is congruent with previous observations in other ERαKO or knockdown models which showed that the ubiquitous or neuron-/site-specific lack of ERα leads to impaired LH surge (Wintermantel et al., 2006; Cheong et al., 2015; Dubois et al., 2015; Porteous and Herbison, 2019; Wang et al., 2019). Interestingly, when treated with P, C451A-ERα females exhibit the typical activation of Kp and GnRH neurons (Fig. 2). Although one might wonder how the activation of the LH surge generation circuit is possible in mice showing such elevated circulating LH concentrations, this idea is compatible with the “two component model” of control of GnRH secretion which poses that the positive and negative feedbacks of gonadal steroids on LH regulation are regulated by two anatomically distinct and independent mechanisms (Herbison 2020). This idea is also supported by a recent study showing that a surge can be elicited over high LH levels (Chuon et al., 2022).
The inability of ovariectomized C451A-ERα females to show the characteristic activation of Kp and GnRH neurons by estrogens converges with the observation of reduced numbers of corpus lutea in gonadally intact C451A-ERα females (Adlanmerini et al., 2014) and suggests that this mutation leads to impaired ovulation and infertility. This was initially supported by the absence of pups in the nest of C451A-ERα females (Adlanmerini et al., 2014) and NOER females, another model generated following the same mutation of the palmitoylation site into an alanine (Pedram et al., 2014). However, a more careful investigation revealed that C451A-ERα females do get pregnant but lose their fetuses during the course of pregnancy and delivery, due to placental dysfunction and delayed labor induction, respectively (Rusidzé et al., 2022). Moreover, the ovulation rate of females mated overnight with a male did not differ between genotypes. This result is very surprising when compared with the present data as they indicate that ovary-intact C451A-ERα females are able to ovulate. Yet, it is important to consider that, when housed with females only, C451A-ERα females present irregular cycles with very rare estrous (they are essentially blocked in diestrus) suggesting very rare natural ovulations, matching the reduced number of corpora lutea. To assess ovulation rate, females were housed overnight with a male and mating was assessed by the presence of a vaginal plug, considered as an indication of estrous. Surprisingly, both WT and C451A-ERα females presented the same percentage of females with a plug on the next day (Rusidzé et al., 2022). A potential explanation is that exposure to male cues has overridden the blockade of the axis caused by the absence of mERα. Male cues are known to stimulate the activation of GnRH neurons (Taziaux and Bakker, 2015). In immature females, male cues induce estrus cycling and accelerate the cycle in adult group housed females (Whitten, 1956, 1958). However, this effect was reported to occur within 48 h, not overnight. This being said, in OVX mice chronically treated with a low dose of estrogens and hence presenting a high level of LH (similar to ovary-intact C451A-ERα females), the exposure to a male induces a rise in LH within 24 h (Bronson, 1976). Finally, overnight housing of acyclic aged female rats with a sexually active male led to a surge of LH secretion and ovulation whether they were allowed to copulate or not (Matt et al., 1987; Day et al., 1988). These effects are likely mediated by olfactory cues emitted by the males since exposure to male urine can restore ovulation in young females in persistent estrous (Johns et al. 1978). Interestingly, ovulation in aged females in persistent estrous cannot be mimicked by treatment with estrogens (Matt et al., 1987) and the reflexive LH surge elicited by male cues is associated with a rise of circulating progesterone concentration (Day et al., 1988). Together, these observations suggest the intriguing possibility that the absence of mERα from conception onwards may hamper spontaneous ovulation but somehow permit reflex ovulation in the presence of a mate. The mechanism underlying such an effect remains to be tested but would likely depend on the activation of GnRH and a rise of LH secretion following mating as in induced ovulators (Bakker and Baum, 2000).
E4 acts as a mERα antagonist in the brain
E4 mimics the nuclear actions mediated by E2 on ERα in several tissues including the uterus (Abot et al., 2014), vagina (Benoit et al., 2017), the mammary gland (Gérard et al., 2015a), and cardiovascular system (Guivarc’h et al., 2018). Although antagonistic actions of E4 have been reported in several tissues including the brain (Pluchino et al., 2014; Gérard et al., 2015a,b; Pluchino et al., 2015), whether these effects are mediated by transcriptional or membrane ERα signaling is not known, with the exception of the membrane-mediated action identified in endothelial and breast cancer cells (Abot et al., 2014; Giretti et al., 2014) and in the brain (de Bournonville et al., 2023). The blockade of E2-induced activation of Kp and GnRH neurons by E4 in parallel with the absence of such response in mice lacking mERα signaling therefore provide converging evidence of the antagonist action of E4 on mERα in the brain and most probably within the preoptic area. However, the timing of these effects (estrogens being administered >24 h prior to sample collection) does not allow to determine whether they reflect direct membrane actions or membrane-initiated transcriptional effects. Further studies will be needed to identify the mechanism underlying these effects.
An alternative interpretation of these results is that E4 would act as an agonist of ER rather than an antagonist, thus exerting its effect through a negative feedback mechanism. However, it must be noted that E4 has a very short half-life in mice (2 h), contrasting with the situation in women (28 h; Gallez et al., 2023), making it unlikely that the injection received 34 h prior to sample collection could have resulted in a negative feedback effect that would have prevented the surge. Moreover, in a parallel study, a single dose of E4 induced a very moderate reduction of LH measured 3 h later, such that it does not reach statistical significance for several doses, including the one used in the present study, while chronic treatment with much lower doses resulted in a massive reduction in LH secretion (Faure et al., Submitted). Finally, it is important to note that E4 does not block the activation of Kp and GnRH neurons in the presence of the combination of EB and P, while P alone (Veh + P condition) does not activate these neuronal populations. As high estrogen levels are known to prevent LH induction by P (Bronson and Vom Saal 1979), this observation further argues in favor of an antagonist action of E4 and membrane ERα. In conclusion, although it cannot be ruled out that the absence of activation of Kp and GnRH neurons in mice treated with EB and E4 results from a negative feedback effect, this possibility seems unlikely. Future work targeting specific brain regions and neuronal populations is however necessary to confirm this hypothesis.
Discrepancies between the two approaches
Although the two approaches employed in this study lead to similar conclusions, differences were observed. E4 altered EB-induced activation of Kp and GnRH neurons (Fig. 3) but had no impact on the number of Kp and GnRH neurons. In contrast, OVX and E2-treated C451A-ERα females exhibited elevated LH concentrations along with fewer Kp neurons and more GnRH neurons than their wild-type counterparts (Fig. 2). The presence of Kp in AVPv neurons of C451A-ERα mice confirms the preserved transcriptional activity of ERα, contrasting with the absent or greatly reduced Kp expression in the complete absence of ERα in Kp neurons (Smith et al., 2005; Gottsch et al., 2009; Dubois et al., 2015). However, the lower number of Kp neurons observed in the present experiment could be explained by a developmental effect of the constitutive mERα absence or by an effect of the mutation on Kp transcription. Although developmental defects cannot be ruled out, several observations points to the latter. First, the early programming of AVPv Kp neurons is affected by estrogen exposure in two critical periods. Perinatal exposure to estrogens leads to few detectable Kp neurons that are typical of males (Gonzalez-Martinez et al., 2008). In females, prepubertal exposure to estrogens is required to observe normal adult Kp neuronal numbers (Clarkson et al., 2009; Szymanski and Bakker, 2012; Brock and Bakker, 2013). Accordingly, C451A-ERα females exhibit expected numbers of Kp in the AVPv supporting an absence of programming defect in this cell population in females (Khbouz et al., 2019). Second, the number of Kp neurons in the AVPv of C451A-ERα females appears to be influenced by the dose of estrogens. Comparable numbers of Kp neurons were counted in the AVPv of wild-type and C451A-ERα females injected daily with EB (1 µg) for 2 weeks (Khbouz et al., 2019). Moreover, a parallel study found a difference between genotypes in females implanted with a Silastic capsule filled with 1 µg of E2 but not with a capsule containing 5 µg of E2 (Faure et al., Submitted). ERα may be less sensitive to estrogens in the mutant mice, thus requiring higher circulating concentrations of E2 to stimulate normal Kp expression as was recently shown to be the case in other tissues (Jiang et al., 2023). Finally, ERE-independent pathways are not sufficient to stimulate Kp expression in the AVPv of ERαKO mice (Gottsch et al., 2009). The lower number of Kp neurons in C451A-ERα mice thus seems attributable to a lower expression of Kp in the presence of low circulating estrogens. One report mentions, however, a stimulatory role for mERα in the expression of Kp in mHypo51A cells, an immortalized line derived from adult mouse hypothalamic neurons presumed to be AVPv Kp neurons (Mittelman-Smith et al., 2015).
Role of progesterone signaling
Genetic or pharmacological blockade of mERα signaling prevented key neuronal populations for the induction of a LH surge by EB. In both cases, neuronal activation was restored by the administration of P 3–4 h before lights off. The potentiating effect of P on EB-induced surge has long been known (Bronson and Vom Saal, 1979). Its importance is underlined by studies focusing on progesterone receptors (PR), whose expression is stimulated by estrogens through an ERE-dependent genomic action (Moffatt et al., 1998). Knockout PR mice (PRKO) and mice lacking PR exclusively in Kp neurons (KissPRKO) are unable to mount an EB-induced surge (Chappell et al., 1999; Stephens et al., 2015; Gal et al., 2016). However, the reintroduction of PR expression specifically in Kp neurons of the AVPv of KissPRKO mice restores the LH surge, demonstrating the essential role of P action on this neuronal population for the induction of the LH surge (Mohr et al., 2021). Our results could thus suggest that the absence or blockade of mERα impedes PR expression. This hypothesis seems however unlikely given that C451A-ERα mice respond well to exogenous P in terms of Kp and GnRH activation. Moreover, E4 mimics the action of E2 on PR expression and E2 induces PR expression in the brains of C451A-ERα females, albeit to a lesser extent than in wild-type mice (Faure et al., Submitted). Membrane estrogen signaling could also interfere with another aspect of P signaling, such as its membrane-initiated or ligand-independent signaling (Tetel and Lange, 2009).
Alternatively, the present results support the notion that mERα modulates local P synthesis to contribute to LH surge induction (Micevych et al., 2015). Remarkably, all the studies supporting the necessity of PR to induce an LH surge, in particular within Kp neurons, did not provide exogenous P, suggesting that an endogenous source of P exists in OVX females which may be necessary to elicit the surge. This idea is supported by an absence of correlation between circadian fluctuations of brain and plasma P concentration in ovary-intact mice (Corpechot et al., 1997). Moreover, the work of Paul Micevych and his collaborators indicates that (1) a rise in neuroprogesterone produced by hypothalamic astrocytes is a prerequisite for the LH surge (Micevych et al., 2003; Micevych and Sinchak, 2008; Mohr et al., 2019; Chuon et al., 2022), (2) this rise depends on mERα activation (Micevych et al., 2007; Kuo et al., 2010; Mohr et al., 2022), and (3) neuroprogesterone’s action on LH is mediated by its action on Kp neurons (Mittelman-Smith et al., 2018). Therefore, it is possible that the lack of activation of the central pathway leading to LH surge in mice lacking mERα or following E4 treatment is explained by a blockade of hypothalamic P synthesis which is necessary for LH induction. In this model, mERα activation would thus stimulate neuroprogesterone synthesis by hypothalamic astrocytes and indirectly activate Kp neurons and in turn GnRH neurons.
Conclusions
The present results contradict the idea that the central induction of a LH surge by rising concentrations of circulating estrogens is mediated by genomic effects only. Although it has long been known that the LH surge requires a prolonged exposure to high estrogen concentrations, it is also recognized that estrogens do not have to be present the whole time for the surge to occur (Legan et al., 1975; Evans et al., 1997). Moreover, membrane estrogen signaling through modulation of intracellular signaling cascades can potentiate the slower transcriptional actions of estrogens (Vasudevan et al., 2001; Kow and Pfaff, 2004). A role for membrane-initiated signaling in the induction of LH surge by estrogens is supported by the acute actions of E2 reported on the activity of GnRH neurons (Romano et al., 2008; Chu et al., 2009; Romanò and Herbison, 2012). It should also be pointed out that membrane-initiated signaling does not necessarily imply rapid actions, as indirect genomic signaling is also possible (Vasudevan and Pfaff, 2007). The present results cannot discriminate between these possibilities, nor can they determine the contribution of mERα located in the AVPv and ARC. Although it cannot be ruled out that the impaired positive feedback observed in mutant mice is an indirect result of the expected dysregulation of the negative feedback, this would not explain why Kp and GnRH neurons are still able to respond normally when provided with P along with EB. Moreover, the similarity of the responses of C451A-ERα mice to wild-type females treated with E4 supports a role for membrane-initiated estrogen signaling in the central induction of LH surge, probably through the activation of neuroprogesterone synthesis by hypothalamic astrocytes (Micevych et al., 2015). Further work will be necessary to identify where this contribution occurs.
Footnotes
The authors declare no competing financial interests.
We thank Laura Vandries for her help with the immunostaining and Arlette Gérard for carrying out the RIA assay.
This work was supported by grants from the Fonds National pour la Recherche Scientifique (F.R.S.-FNRS PDR T.0042.15) and the Special Funds for Research from the University of Liège (FSR-S-SS-19/40), a research project (E4Liberty) with Mithra Pharmaceuticals and the Walloon Region of Belgium. C.A.C. is a Research Director of the F.R.S.-FNRS.
R.C.’s present address: Departamento de Neurobiología Celular y Molecular, Instituto de Neurobiología, UNAM, Querétaro, México.
M.C.F.’s present address: Center for Integrative Genomics, University of Lausanne, Lausanne, Switzerland.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.