Local production of neurostradiol affects gonadotropin‐releasing hormone (GnRH) secretion at mid‐gestation in Lagostomus maximus (Rodentia, Caviomorpha)

Abstract Females of the South American plains vizcacha, Lagostomus maximus, show peculiar reproductive features such as massive polyovulation up to 800 oocytes per estrous cycle and an ovulatory process around mid‐gestation arising from the reactivation of the hypothalamic–hypophyseal–ovary (H.H.O.) axis. Estradiol (E2) regulates gonadotropin‐releasing hormone (GnRH) expression. Biosynthesis of estrogens results from the aromatization of androgens by aromatase, which mainly occurs in the gonads, but has also been described in the hypothalamus. The recently described correlation between GnRH and ER α expression patterns in the hypothalamus of the vizcacha during pregnancy, with coexpression in the same neurons of the medial preoptic area, suggests that hypothalamic synthesis of E2 may affect GnRH neurons and contribute with systemic E2 to modulate GnRH delivery during the gestation. To elucidate this hypothesis, hypothalamic expression and the action of aromatase on GnRH release were evaluated in female vizcachas throughout pregnancy. Aromatase and GnRH expression was increased significantly in mid‐pregnant and term‐pregnant vizcachas compared to early‐pregnant and nonpregnant females. In addition, aromatase and GnRH were colocalized in neurons of the medial preoptic area of the hypothalamus throughout gestation. The blockage of the negative feedback of E2 induced by the inhibition of aromatase resulted in a significant increment of GnRH‐secreted mass by hypothalamic explants. E2 produced in the same neurons as GnRH may drive intracellular E2 to higher levels than those obtained from systemic circulation alone. This may trigger for a prompt GnRH availability enabling H.H.O. activity at mid‐gestation with ovulation and formation of accessory corpora lutea with steroidogenic activity that produce the necessary progesterone to maintain gestation to term and guarantee the reproductive success.


Introduction
The South American plains vizcacha, Lagostomus maximus, a hystricognath fossorial rodent that inhabits the Pampean region in Argentina (Jackson et al. 1996), has been reported to show peculiar reproductive features such as preovulatory follicle formation during the 155-day long-lasting pregnancy with an ovulatory process around mid-gestation that produces a considerable number of secondary corpora lutea with oocyte retention, and the consequent increment in circulating progesterone (P4) levels (Jensen et al. 2008;Dorfman et al. 2013). In contrast to most mammals, female vizcacha shows massive polyovulation of up to 800 oocytes per estrous cycle, the highest ovulatory rate so far recorded for a mammal (Weir 1971a,b;Jensen et al. 2006). Massive ovulation results from an unusual constitutive suppression of apoptosis that precludes intraovarian oocyte dismissal through follicular atresia (Jensen et al. 2006(Jensen et al. , 2008Leopardo et al. 2011;Inserra et al. 2014).
Gonadotropin-releasing hormone (GnRH), the central regulator of fertility in mammals, is involved in the modulation of the hypothalamic-hypophyseal-ovary (H.H.O.) axis. We have previously shown that GnRH distribution in the hypothalamic areas of the vizcacha, as medial preoptic area (mPOA), suprachiasmatic area, supraoptic area (SON), arcuate nucleus, and medial eminence, is comparable to a variety of other mammalian species (Dorfman et al. 2011;Inserra et al. 2017). Estradiol (E 2 ) regulates GnRH expression through its binding to specific receptors. Two nuclear estrogen receptors isoforms, alpha (ERa) and beta (ERb), have been described to bind E 2 with similar affinity (Kuiper et al. 1996). Both receptors can bind to estrogen response elements (ERE) localized in the GnRH gene promoter activating GnRH synthesis (Radovick et al. 1991). Estrogen has a bimodal effect on the hypothalamus, with either an inhibitory or stimulatory action on GnRH delivery. A rapid E 2 increase at the end of follicular phase triggers a stimulatory effect on GnRH surge resulting in the secretion of gonadotropins, follicle-stimulating hormone (FSH) and luteinizing hormone (LH) from the pituitary gland, which are essential for follicular maturation and ovulation, respectively (Abraham et al. 1972;Simerly 2002). In contrast, during the luteal phase, E 2 produces a negative feedback over GnRH secretion (March et al. 1979). In human pregnancy, maternal serum E 2 level increases considerably above preovulatory levels remaining elevated until the end of gestation, inhibiting menstrual cycle (Simpson and McDonald 1981). The ability of estrogen to induce GnRH surge declines with age both in rodent and human females (Shaw et al. 2011). Lack of estrogen in postmenopausal women contributes to the increase in GnRH mRNA levels (Morrison et al. 2006;Shaw et al., 2010).
Biosynthesis of estrogens is due to aromatization of androgens through aromatase (CYP19A1), an enzyme located in the endoplasmic reticulum and mitochondria (De Montellano 2005). Testosterone (TT) and androstenedione (A4) are converted into E 2 and estrone (E 1 ) by aromatase, respectively. In addition, E 1 is converted into E 2 through 17b-hydroxysteroid dehydrogenase type I and A4 is converted into TT by a 17b-hydroxysteroid dehydrogenase type III (Luu-The et al. 1995;Conley and Hinshelwood 2001). E 2 is mainly produced by ovaries and it crosses the hematoencephalic barrier; however, the synthesis of E 2 in the hypothalamus has been also described, suggesting that neurosteroidogenesis is needed for LH surge to induce ovulation and luteinization (Naftolin et al. 1972, Roselli et al. 2009). Brain neuroendocrine areas involved in the control of reproduction have shown aromatase expression and activity (Selmanoff et al. 1977;Resko 1987, 1993) suggesting the importance of local synthesis of E 2 , called neuroestradiol, on reproduction.
The peculiar reproductive cycle of the South American plains vizcacha suggests a particular modulation of its neuroendocrine activity to enable ovulation at mid-gestation. We have recently described a correlation between GnRH and ERa expression in the hypothalamus of the vizcacha during pregnancy and the expression of ERa in the same neurons that synthetize GnRH, and that these observations occur together at mid-gestation with preovulatory follicle development and secondary corpora lutea formation, suggesting E 2 feedback effects on GnRH surge during gestation (Inserra et al. 2017). These observations prompted us to examine the possible contribution of hypothalamic synthetized E 2 on GnRH expression and delivery, and its involvement on the ovulation during pregnancy observed in the vizcacha.

Animals
Adult female plains vizcachas [2.5-3.0 Kg body weight; 2-2.5 years old determined by the dry crystalline lens weight, according to Jackson (1986)] were captured from a resident natural population at the Estaci on de Cr ıa de Animales Silvestres (ECAS), Villa Elisa, Buenos Aires, Argentina. Animals were captured using live traps located at the entrance of burrows. All experimental protocols concerning animals were conducted in accordance with the guidelines published by the National Institutes of Health (NIH, USA) guide for the care and use of laboratory animals (NIH1985), and were reviewed and approved by the Institutional Committee on Use and Care of Experimental Animals (CICUAE) from Universidad Maim onides, Argentina. Handling and euthanasia of animals were performed in accordance with the NIH guidelines for the care and use of laboratory animals (NIH 1985, CCAC 2002, and CCAC 2003. In order to obtain females at different reproductive stages, captures were planned according to the natural reproductive cycle, previously described by Llanos and Crespo (1952), and on our own previous expertise in the field (Jensen et al. 2006(Jensen et al. , 2008Dorfman et al. 2011Dorfman et al. , 2013Leopardo et al. 2011;Halperin et al. 2013;Inserra et al. 2014Inserra et al. , 2017Charif et al. 2016). Pregnant vizcachas were captured during the breeding season that lasts from April to August. Gestational age was estimated on the basis of capture time and fetal development, including examination of fetal morphology, Theiler Stages of development, visible implantation sites in earlypregnant animals, and fetal crown-heel length for midpregnant animals (80-130 mm), according to Leopardo et al. (2011). Early-pregnant females (EP, N = 15) were captured in April, mid-pregnant females (MP, N = 15) in July, and term-pregnant females (TP, N = 15) in August; while nonpregnant females (NP, N = 15) were captured in mid-September after the end of the reproductive season. The ovulatory status was assessed by ovary inspection for the presence of ovulatory stigmata at the time of sacrifice and was correlated with the presence of follicles or corpora lutea in hematoxylin-eosin-stained ovary sections. Animals were housed under a 12:12 h low-light cycle to simulate their natural luminal exposure (low light of 12 watts followed by moon light), and 22 AE 2°C room temperature, with ad libitum access to food (alfalfa, potatoes, and apples) and tap water. Animals were housed for 1 week before euthanasia or before beginning with the hormonal treatment for estral cycle synchronization.

Tissue collection
Animals were anaesthetized by the intramuscular injection of 13.5 mg/kg body weight ketamine chlorhydrate (Holliday Scott S.A., Buenos Aires, Argentina) and 0.6 mg/kg body weight xylazine chlorhydrate (Richmond Laboratories, Veterinary Division, Buenos Aires, Argentina). Animals were sacrificed by trained technical staff by an intracardiac injection of 0.5 mL/kg body weight of Euthanyl TM (sodium pentobarbital, sodium diphenilhidanthoine, Brouwer S.A., Buenos Aires, Argentina) and brains were immediately removed. Isolated brains were either fixed in cold 4% neutral-buffered paraformaldehyde (PFA) (Sigma Aldrich Inc., St. Louis, Missouri) for immunohistochemical studies or the whole hypothalami were dissected out to a depth of approximately 4 mm with the following borders: the anterior edge of the optic chiasm, the anterior edge of the mammillary bodies, and the two hypothalamic sulci on either lateral side as Charif et al. (2016). Whole hypothalami were incubated in Krebs-Ringer buffer for pulsatile studies, or anterior hypothalami (AH) fragments surrounded by the anterior, posterior, and lateral borders of the optic chiasm (oc), including the preoptic area (POA), isolated, quickly frozen, and stored at À80°C for Western blot, RIA, or qPCR assays. Ovaries were also removed and fixed in cold 4% PFA for histological inspection of the ovulatory status and follicle development as previously described . Before performing euthanasia, blood samples were taken by cardiac puncture, centrifuged for 15 min at 3000 g and the serum fraction was aliquoted and stored at À20°C.

RNA isolation and quantitative polymerase chain reaction (qPCR)
AH were homogenized with TRIzol (Invitrogen, California, USA) according to the manufacturer's instructions to extract total RNA. RNA concentration was quantified by absorption at 260 nm (Genequant, Amersham Biosciences, England) and its integrity confirmed in a 1% agarose gel (Genbiotech, Argentina) in Tris (0.09 mol/L)-boric acid (0.045 mol/L)-EDTA (0.05 mol/L) (TBE) buffer (pH 8.3). RNA integrity was confirmed when the presence of S28 and S18 rRNA subunits was observed. Three lg of total RNA was treated with 1 lL DNaseI (Invitrogen, California, USA) in 1 lL 10X DNase Reaction Buffer (Invitrogen, California, USA) for 30 min at 37°C, and the reaction was stopped with 1 lL 50 mmol/L EDTA (Invitrogen, California, USA) for 10 min at 65°C. The RNA was reverse transcribed into first-strand cDNA using 1.5 lL random hexamer primers 50 lmol/L (Applied Biosystems, California, USA), 200U reverse transcriptase (RevertAid TM M-MuLV, Fermentas, Massachusetts), 4 lL First-Strand Buffer 5x (Fermentas, Massachusetts), 2 lL dNTP mixture 10 mmol/L (Invitrogen, California, USA), and 0.5 lL RNase inhibitor (Ribolock TM , Fermentas, Massachusetts), at a 20 lL final volume reaction. The reverse transcriptase was omitted in control reactions, where the absence of PCR-amplified DNA indicated the isolation of RNA free of genomic DNA. The reaction was carried out at 72°C for 10 min, 42°C for 60 min, and stopped by heating at 70°C for 10 min. cDNA was stored at À20°C until used. Reverse-transcribed cDNA (1:10 diluted in DEPC-treated sterile distilled water) was mixed with 6 lL SYBR Green PCR Master Mix (Applied Biosystems, United Kingdom) for qPCR using 0.3 lmol/L forward and reverse oligonucleotide primers. Primer sequences and cycling parameters are shown in Table 1. Primers for GnRH and GAPDH (glyceraldehyde 3-phosphate dehydrogenase) were successfully employed in vizcacha by Charif et al. (2016) and Gonzalez et al. (2012) respectively. Primers for aromatase were previously published by Galmiche et al. (2006). The primers for GnRH are intron spanning located in exons 2 and 4 of vizcacha, primers for GAPDH are located in exon 5 of mouse and rat, whereas those for aromatase are located in exons 9 and 10 of mouse and rat mRNAs. Oligonucleotides were obtained from Invitrogen (Life Technologies, California, USA). Quantitative measurements were performed using a Stratagene MPX500 cycler (Stratagene, California, USA). Data were collected from the threshold value, taken at the 72°C extension phase, continuously stored during reaction and analyzed by the complementary computer software (MxPro3005P v4.10 Build 389, Schema 85, Stratagene, California, USA). To confirm specificity of the signal, the results were validated based on the quality of dissociation curves generated at the end of the qPCR runs. Relative gene expression was normalized to that of GAPDH as housekeeping gene. For the assessment of quantitative differences in the cDNA target between samples, the mathematical model of Pfaffl was applied (Pfaffl 2001). An expression ratio was determined for each sample by calculating (E target ) ΔCq(target )/(E GAPDH ) ΔCq(GAPDH) , where E is the efficiency of the primer set and Aromatase primers sequences were previously used by Galmiche et al. (2006). GnRH primers were employed in vizcacha by Charif et al. (2016) and GAPDH primers by Gonzalez et al. (2012). Product = Amplified product length. bp, base pairs. ΔCq (quantification cycle) is the difference in the threshold cycle with ΔCq = Cq (normalization cDNA) -Cq (experimental cDNA) . The amplification efficiency of each primer set was calculated from the slope of a standard amplification curve of log (ng cDNA) per reaction versus Cq value (E = 10 À(1/slope) ). Efficiencies of 2.0 AE 0.1 were considered optimal. Six animals were tested per group and each sample was analyzed in triplicate along with nontemplate controls to monitor contaminating DNA. Purity of the amplified products was confirmed by 2% agarose gel electrophoresis (Biodynamics, Buenos Aires, Argentina  (Greenwood et al.1963). Intra-and interassay coefficient of variation was 7.2% and 11.6%, respectively. The detectability limit was 1 pg/100 lL. Protein content of each sample was determined by Bradford method (Bradford 1976). Results were expressed as the ratio between GnRH value obtained by RIA and the protein content. Six animals per group were tested.

Immunofluorescence with confocal microscopy
After removal, brains were coronally sectioned in blocks of 5-7 mm thick, fixed in cold 4% paraformaldehyde (PFA) in 0.1 mol/L PBS (pH 7.4) for 72 h, dehydrated through a graded series of ethanol solutions, and embedded in paraffin. For each specimen, the brain region containing the hypothalamus was entirely cut to serial coronal sections (5 lm thick) and mounted onto coated slides. Sections were dewaxed in xylene and rehydrated through a decreasing series of ethanol solutions (100%, 95%, and 70%). One on every 10 sections was separated to perform classical Nissl technique staining to localize hypothalamic nuclei, according to previous description (Dorfman et al. 2011). Adjacent sections were used for immunohistochemical assays. Antigen retrieval was performed by boiling sections in 10 mmol/L sodium citrate buffer (pH 6) for 20 min, followed by 20 min cooling at room temperature. GnRH pulsatility was measured in vitro as previously described (Catalano et al. 2010;Charif et al. 2016 . During incubation, the medium from each tube was collected at 7.5-min intervals, replaced with fresh medium, and stored at À20°C. A depolarizing concentration of potassium chloride (KCl, 100 mmol/L) was added to the last tube (30 min) to test tissue viability by identifying a marked peak of GnRH release. GnRH content of each collected medium was determined by RIA as described above. GnRH pulsatile parameters were determined using the computer algorithm Cluster8 developed by Veldhuis and Johnson (1986) (Pulse_XP software, http://mljohnson.pha rm.virginia.edu/home.html). A 2 9 2 cluster configuration and a t-statistic of 2 for the upstroke and down stroke, to maintain false-positive and false-negative error rates below 10%, were used as suggested by Mart ınez de la Escalera et al. (1992). Tissues from five different animals were tested per group.

ELISA of estradiol and estrone delivered to the medium
To verify the inhibition of aromatase activity by letrozole, both estradiol (E 2 ) and estrone (E 1 ) content was determined in all the aliquots of the collected medium from the hypothalamic explants. Quantification was performed using EIA Estradiol ELISA Kit (EIA2693, DRG Int., Germany) and EIA Estrone ELISA Kit (EIA4174, DRG Int., Germany) according to the manufacturer's instructions. Direct solid-phase enzyme immunoassays that detect a range 16-2000 pg/mL of E 2 or 13-1000 pg/mL of E 1 were developed. Intra-and interassays coefficients of variation were below 10.3% for both measurements. The absorbance of the solutions, including the samples and the experimental blank (medium that has not come in contact with the brain explant), was measured at 450 nm (lQuant Microplate Spectophotometer, Bio-tek Instruments Inc., Winooski, VT) and inversely related to the concentration of E 2 or E 1 . Calculation of E 2 and E 1 content was made by reference to the respective calibration curve and expressed as pg/mL/ hypothalamic weight to normalize steroid content to the weight of the corresponding hypothalamic explant.

Statistical analysis
Values are expressed as mean AE standard deviation (SD).
All the experiments were performed by duplicate. Results were evaluated using t-test for comparisons between two groups, or one-way analysis of variance (ANOVA) followed by Newman-Keuls and Bonferroni's Multiple Comparison tests was employed for comparisons among more than two groups. Statistical analysis was performed using Prism 4.0 (GraphPad Software Inc., San Diego, California, USA). Differences were considered significant when P < 0.05.

Results
Hypothalamic aromatase and GnRH display a similar expression pattern throughout gestation in L. maximus Hypothalamic aromatase expression varied significantly throughout gestation. A significant increase in mid-pregnant vizcachas compared to nonpregnant, early-pregnant, and term-pregnant females was observed, and this profile was consistent at both mRNA and protein levels ( Fig. 1A-B). On the other hand, GnRH displayed a pattern similar to that of aromatase; a significant sharp increment at mid-pregnancy that fell down at term pregnancy reaching values close to those exhibited by earlypregnant females was observed. Both mRNA and protein values showed a similar pattern of variation throughout gestation ( Fig. 1C-D). The analysis of the aromatase PCR product sequence of L. maximus was performed using the Blast algorithm (NCBI, NIH, USA). It showed 96% homology with rat (Rattus norvegicus), 93% with mouse (Mus musculus), 91% with naked mole-rat (Heterocephalus glaber), and 90% with Chinese hamster (Cricetulus griseus) and guinea pig (Cavia porcellus).

Aromatase and GnRH colocalize in the hypothalamus of L. maximus
Cytosolic aromatase and GnRH localization was determined in neurons of the mPOA of nonpregnant females ( Fig. 2A-B). Two subgroups of GnRH neurons, high-or low-labeled, were found (Fig. 2B). Qualitatively, lowlabeled neurons were more abundant than high-labeled neurons, being the latter scattered throughout the mPOA. Strikingly, neurons with high signal for GnRH did not show aromatase expression while neurons with low density of GnRH depicted cytoplasmic aromatase colocalization (Fig. 2C). In addition, randomly distributed GnRH immunopositive axonal varicosities were observed (not shown). Aromatase-or GnRH-specific labeling was not detected after preabsorption of the primary antibodies with aromatase or LHRH synthetic peptides, in adjacent sections, or after omission of the primary antibodies (not shown).
The colocalization of aromatase and GnRH in the mPOA was detected in all experimental groups (Fig. 3). All the low-density GnRH immunoreactive neurons showed aromatase expression without significant variations among the evaluated groups (Fig. 4A). However, additional aromatase immunoreactive neurons lacking GnRH immunolabeling were also observed (Fig. 3I, L). The abundance of aromatase-expressing cells lacking GnRH expression showed significant differences throughout gestation, increasing from less than 5% at early gestation to approximately 60% at term pregnancy (Fig. 4B).

Neuroestradiol is involved in GnRH hypothalamic release
GnRH delivery by hypothalamic explants supplemented with the aromatase inhibitor letrozole was evaluated. GnRH pulsatile pattern in letrozole-treated explants was compared against its basal pulsatility in hypothalamic explants without letrozole (Control) throughout six hours long. Both letrozole treatment and control condition showed five pulses of GnRH throughout the experiment (Fig. 5A-B). Parameters concerning the GnRH pulsatile delivery (Fig. 5C-E) and the GnRH-secreted mass ( Fig. 5F-H) were evaluated. Pulsatile parameters concerning GnRH delivery as pulse frequency, mean intervals between pulses, and mean pulse width did not show significant variations between letrozole and control conditions ( Fig. 5C-E respectively). In contrast, a significant increase was observed in the GnRH-secreted mass when hypothalamic explants were supplemented with letrozole versus control condition. This parameter is reflected in the mean mass of GnRH delivered by pulse (Fig. 5F), the maximum mass of GnRH secreted by pulse (Fig. 5G), and the total GnRH mass secreted during the six hours of the experiment (Fig. 5H). To confirm that letrozole inhibited the aromatase activity in hypothalamic explants, E 1 and E 2 content was determined in all the aliquots corresponding to GnRH peaks collected throughout experiment (Fig. 6). Letrozole induced a significant decrease in E 1 and E 2 of hypothalamic origin with respect to control condition ( Fig. 6A-B). This confirmed the decrease in aromatase activity induced by letrozole action.

Discussion
This study shows the first evidence of expression of aromatase enzyme in the mPOA and its hypothalamic activity in Lagostomus maximus. Variations in aromatase expression during pregnancy, with a concordant pattern with GnRH expression, suggest the action of neuroestradiol in GnRH expression at mid-gestation. In addition, two subgroups of GnRH neurons were detected according to their labeling of GnRH by immunohistochemistry. The action of neuroestradiol on GnRH modulation and its possible role in ovulation in the South American plains vizcacha during gestation is described. Fertility depends on a complex steroid feedback mechanism that regulates the activity of GnRH neurons. E 2 is known as a major regulator of GnRH neuron activity by bimodal feedback mechanisms. Aromatase is a monooxygenase that belongs to the cytochrome P450 superfamily and is responsible for the aromatization of androgens into estrogens. This enzyme is highly conserved among vertebrates. Its localization and activity in the hypothalamus of several species like rat, monkey, Japanese quail, and zebra fish has been reported (Roselli et al. 1985;Roselli and Resko 1987;Schumacher and Balthazart 1987;Vockel et al. 1990;Naftolin et al. 2001). In this work, we described the localization of aromatase in the mPOA of the hypothalamus of the South American plains vizcacha and its colocalization with GnRH suggesting the direct modulation of neuroestradiol on GnRH expression. Two subpopulations of GnRH neurons have been described in the rostral hypothalamus of rats. They are composed of abundant low-intensity-stained GnRH neurons (more than 80% of total GnRH neurons) and some scattered highly-intensity-stained GnRH neurons (Wray et al. 1989). In this work, we show the existence of these both groups in the mPOA of the vizcacha. Rats and mice also show aromatase-expressing neurons in the neighborhood of GnRH cells in the mPOA (Wagner and Morrell 1996), but not the coexpression of aromatase and GnRH in the same neurons. The cytoplasmic colocalization of GnRH with aromatase in the mPOA of the vizcacha suggests the capacity of these cells to convert testosterone into E 2 , which may directly affect GnRH expression. Strikingly, we noted that this colocalization was exclusively detected in the low-stained GnRH neurons, whereas highly stained GnRH neurons were devoid of aromatase expression. It has been hypothesized that GnRH neuron population was divided into pulse and surge cells and differences in neuronal activity may depend on E 2 modulation (Caraty et al. 1998;Boukhliq et al. 1999). In this context, the two subpopulations of GnRH neurons found in the mPOA of the vizcacha may be related with differences in the modulation of pulse and surge of GnRH by E 2 action. In addition, it seems important to note that both neurons and glia have been described to synthesize E 2 in vitro (Zwain and Yen 1999), however, in this study aromatase was only localized in neurons. This is in accordance with immunohistochemical studies performed in Japanese quail, rat, monkey, and humans that show neurons as the primarily aromatase-expressing cells in mammals (Naftolin et al. 1996) while glial expression of aromatase is induced by brain damage with neuroprotective effects (Saldanha et al. 2009). The action of neuroestradiol-modifying GnRH release has been previously shown using microdialysis method in monkeys (Kenealy et al. 2013). The changes in GnRH pulsatile parameters induced by aromatase inhibition here described confirm the influence of neuroestrogens on GnRH release. The increased level of pulsatile parameters resulting from blocking aromatase activity with letrozole, confirmed by E 1 and E 2 decrease, suggests that GnRH delivery may be negatively affected by a feedback action of these steroids. Concordantly, the positive feedback of estradiol on GnRH could be also confirmed during pregnancy when increased levels of E 2 at mid-gestation are correlated with increased levels of hypothalamic GnRH content and serum LH (Inserra et al. 2017).
The bimodal effect of E 2 over GnRH during the estrous cycle of mammals appears to occur at different anatomical localization and it is generally accepted that it is mediated by kisspeptin in mice. This peptide mediates the E 2 -negative feedback in the arcuate nucleus and its positive feedback in the anteroventral periventricular nucleus (Smith et al. 2006 (Dungan et al. 2007) suggesting alternative ways of GnRH delivery modulation. Other factors such as the neuromodulators c-aminobutiric acid (GABA) and glutamate may be also involved (Roman o et al. 2008). In line with this, the negative feedback exerted by E 2 over its own receptor decreases GnRH dynamics. ER expression changes during the estrous cycle, being decreased during the proestrous stage, when E 2 levels are the highest (Morrison et al. 2006). In addition,  Figure 5. Neuroestradiol is involved in GnRH hypothalamic release. Representative GnRH pulsatile profiles from female hypothalamic explants control (A) and supplemented with letrozole (B), determined by radioimmunoassay (RIA). Five pulses were detected in both conditions during 6 h of treatment. The number of each pulse is indicated above the graph, whereas the bottom line indicates the time lapse of the GnRH release study. Several pulsatil parameters were evaluated: Pulse frequency (C), mean interval between pulses (D), mean pulse width (E), mean mass of GnRH delivered by pulse (mean area of the peaks (F), the maximum mass of GnRH secreted by pulse (mean pulse height), (G), and the total GnRH mass secreted during the 6 h of the experiment (under curve area), (H). Significant increase was determined in maximum mass of GnRH delivered by pulse, the mass of GnRH secreted by pulse, and total GnRH delivered in letrozole-supplemented hypothalamus related to control. Different letters indicate significant differences between groups with P < 0.05. N = 5 per group. several in vivo and in vitro studies developed in rat and mouse have shown that estrogen decreases GnRH gene expression (Zoeller and Young 1988;Wray et al. 1989;Wolfe et al. 1996;Spratt and Herbison 1997). The action of E 2 on GnRH neurons through ERa has been checked over for approximately 40 years by a variety of histological methods. Most data indicate that in a large variety of mammalian species like mink, sheep, ewe, guinea pig, mouse, and rat, GnRH neurons do not express ERa but does ERb (Watson et al. 1992;Herbison et al. 1993;Laflamme et al. 1998;Warembourg et al. 1998;Herbison and Pape 2001;Smith and Jennes 2001;Wintermantel et al. 2006). However, some works showing ERa expression in GnRH neurons of the POA of rats (Butler et al. 1999) and in GT1-7 and GN11 cell lines that express GnRH have been reported (Shen et al. 1998;Roy et al. 1999;Ng et al. 2009). Adding to the controversy, we have recently shown ERa expression in GnRH neurons of medial POA and SON of the South American plains vizcacha (Inserra et al. 2017). It was hypothesized that the direct action of E 2 , by means of ERa on GnRH neurons, may represent a differential reproductive strategy of vizcacha to guarantee GnRH synthesis during pregnancy where GnRH neurons may sense circulating estradiol (Inserra et al. 2017). Concordantly, the locally produced E 2 by aromatase, in the same neurons where ERa is expressed, may contribute to the direct regulation on GnRH synthesis and delivery allowing ovulation during gestation. Both genomic and nongenomic actions of estradiol interact to induce responses. Estradiol may act by classical and nonclassical receptors. Classical receptors as ERa and ERb were described to be located in the cell nucleus with transcriptional effects, or to be embedded in cell membranes with rapid effects as the control of intracellular calcium oscillations in GnRH neurons (Roman o et al. 2008). In mouse, for example, it was described that E 2 -induced increase in intracellular calcium concentration stimulates firing activity of GnRH neurons through ERb (Sun et al. 2010). In addition, nonclassical receptors as GPR30 and ER-X localize in the plasma membrane and can rapidly activate a wide variety of intracellular signaling pathways like modulations of intracellular calcium concentration and rapid protein phosphorylation in several tissues including hypothalamus (Gu et al. 1996;Mermelstein et al. 1996;Moss et al. 1997;Abrah am et al. 2004). High doses of estradiol are required to activate nongenomic responses in relation with doses required to induce genomic effects (Cornil et al. 2006). Considering that aromatase activity is markedly affected by calciumdependent phosphorylation (Balthazart et al., 2003) and that it may be induced by E 2 nongenomic effects, actions of ERa in the vizcacha may be involved in changes in aromatase expression during pregnancy directly affecting autoregulation of E 2 synthesis.
A succession of effects predominates in the E 2 feedback process with early rapid E 2 effects followed by classical later genomic responses (Herbison 1998). The increase in brain aromatase activity, which may contribute with a rapid increment of in situ synthetized E 2 , could potentially activate responses that are insensitive to the lower concentrations of estrogens obtained from the peripheral circulation and be the initial step to trigger GnRH synthesis and surge. This idea has been well established in the case of negative feedback (Herbison 2009); however, its critical role on positive feedback actions had not been completely elucidated. Present data contribute with the analysis of both E 2 regulatory pathways and allow to hypothesize that the synthesis of neuroestradiol might act as the trigger for a prompt GnRH availability enabling Estrone (pg/mL/hypothalamic weight) A B Figure 6. Hypothalamic synthesis and delivery of estrone and estradiol is decreased by letrozole. Hypothalamic synthesis and delivery of E 1 (A) and E 2 (B) was significantly decreased by inhibition of aromatase activity as a result of treatment with the specific inhibitor letrozole. E 1 and E 2 content was expressed as pg/ mL/hypothalamic weight in order to normalize steroid content to the weight of the corresponding hypothalamic explant. For each evaluated group, values represent the average of E 1 or E 2 in the incubation media corresponding to the aliquots where each GnRH pulse showed its maximum height. Different letters indicate significant differences between groups with P < 0.05. N = 5 per group. H.H.O. activity at mid-gestation. The activation of H.H.O allows preovulatory follicle formation followed by the development of accessory corpora lutea with steroidogenic activity that produces the necessary progesterone to maintain gestation to term and guarantee the reproductive success.
In conclusion, local synthesis of E 2 by aromatase may drive to higher E 2 levels than those obtained from systemic circulation alone, with fluctuations in a rapid manner. This could play a crucial role in regulating GnRH delivery being part of the peculiar reproductive strategy of the vizcacha to ensure GnRH availability. The contribution of aromatase activity to GnRH release may represent a fast mechanism for the H.H.O. axis activity regulation.