Disruption of estradiol regulation of orexin neurons: a novel mechanism in excessive ventilatory response to CO2 inhalation in a female rat model of panic disorder

Panic disorder (PD) is ~2 times more frequent in women. An excessive ventilatory response to CO2 inhalation is more likely during the premenstrual phase. While ovarian hormones appear important in the pathophysiology of PD, their role remains poorly understood as female animals are rarely used in pre-clinical studies. Using neonatal maternal separation (NMS) to induce a “PD-like” respiratory phenotype, we tested the hypothesis that NMS disrupts hormonal regulation of the ventilatory response to CO2 in female rats. We then determined whether NMS attenuates the inhibitory actions of 17-β estradiol (E2) on orexin neurons (ORX). Pups were exposed to NMS (3 h/day; postnatal day 3–12). The ventilatory response to CO2-inhalation was tested before puberty, across the estrus cycle, and following ovariectomy. Plasma E2 and hypothalamic ORXA were measured. The effect of an ORX1 antagonist (SB334867; 15 mg/kg) on the CO2 response was tested. Excitatory postsynaptic currents (EPSCs) were recorded from ORX neurons using whole-cell patch-clamp. NMS-related increase in the CO2 response was observed only when ovaries were functional; the largest ventilation was observed during proestrus. SB334867 blocked this effect. NMS augmented levels of ORXA in hypothalamus extracts. EPSC frequency varied according to basal plasma E2 levels across the estrus cycle in controls but not NMS. NMS reproduces developmental and cyclic changes of respiratory manifestations of PD. NMS disrupts the inhibitory actions of E2 on the respiratory network. Impaired E2-related inhibition of ORX neurons during proestrus is a novel mechanism in respiratory manifestations of PD in females.


Introduction
Many panic disorder (PD) patients experience respiratory symptoms, including hyperventilation, sleep apnea, chest pain, and dyspnea [1][2][3] . According to Klein's "False suffocation alarm hypothesis", excessive sensitivity to respiratory stimuli is at the core of PD 4 . CO 2 inhalation can trigger intense fear, autonomic, and ventilatory responses 5 and the probability of experiencing the panicogenic effects of CO 2 is greater in PD patients than healthy subjects 6,7 . Located in a hypothalamic region initially known as the "panic area", orexin-producing neurons (ORX) regulate arousal and the intensity of respiratory reflexes [8][9][10] . ORX concentration in the cerebrospinal fluid of PD patients is higher than in healthy subjects 11 and activation of ORX neurons and ORX-1 receptors are both necessary to observe a panic-prone state in rats and in humans [11][12][13] .
Early life adversities are an important risk factor for PD; clinical and pre-clinical data show that unstable parental conditions or experimental disruption of maternal care augment respiratory and behavioral responses to CO 2 [14][15][16] .
Much like chronic stress, neonatal maternal separation (NMS) augments ORX function [17][18][19] . Together, these data strongly argue that NMS-related dysregulation of ORX function is an important mechanism in excessive ventilatory response to CO 2 inhalation in PD patients. The mechanisms responsible for abnormal ORX-modulation of the ventilatory control system are not fully understood, but to make significant progress that will ultimately influence clinical practice, pre-clinical research requires animal models that are close to the clinical reality 20,21 .
The prevalence of PD is 2-3 fold greater in women than in men, yet most of our basic knowledge arises from experiments performed on males despite strong evidence indicating that endocrine regulation of ORX neurons is sex-specific 22,23 . This is an important issue because clinical observations point to an important role of ovarian hormones in the pathophysiology of PD. The incidence of PD rises at puberty and in adult women, the responsiveness to CO 2 inhalation is highest during the premenstrual phase. Together, these observations suggest that PD patients are more sensitive to hormonal fluctuation than healthy subjects [24][25][26] . To address this issue, we first tested the hypothesis that early life adversities (in the form of NMS) disrupts hormonal regulation of the ventilatory response to CO 2 . Specifically, we determined whether NMS-related increase in the responsiveness to CO 2 inhalation evolves with reproductive status of females. Non-invasive respiratory measurements were performed prior to puberty, across each phase of the estrus cycle, and following ovariectomy. We then used a pharmacological approach to determine whether activation of ORX 1 -receptors is necessary to NMS-induced enhancement of the CO 2 response. The evidence indicating that 17β-estradiol (E 2 ) inhibits ORX neurons being indirect 27,28 , we then used whole-cell patch-clamp recordings to assess E 2 's effects on green fluorescent protein (GFP)-labeled ORX neurons in females. Finally, we tested the hypothesis that NMS disrupts E 2 's inhibitory actions on the ORX system and its influence on CO 2induced respiratory manifestation of PD in females.

Animals and ethical approval
Experiments were performed on sexually mature female Sprague-Dawley rats and pre-pubertal rat pups (14-15 days old) of both sexes. Rats were preferred to mice because, unlike mice, enhancement of the CO 2 response due to early life adversities is sex-specific in this species 15,29 . Details about age, body weight, and animal distribution amongst the experimental groups are reported in Supplemental Table 1 and in the figures. All animals were born and raised in our animal care facilities. Rats were supplied with food and water ad libitum and maintained in standard conditions (21°C, 12:12-h dark-light cycle: lights on at 06:00 and off at 18:00). Animal Care Committee of Université Laval approved all the experimental procedures and protocols, which were in accordance with the guidelines of the Canadian Council on Animal Care.

Neonatal maternal separation (NMS)
The NMS protocol was identical to the one used in our previous studies 29,30 . Briefly, virgin females were mated and 3 days after delivery, each litter was separated daily from their mother 3 h/day (09:00-12:00) from postnatal day 3-12. Control animals were undisturbed during the same period; see the Supplement for details. Rats were weaned and raised under standard animal care conditions until experiments were performed. For each group, rats originated from multiple litters to avoid litter-specific effects.

Whole-body plethysmography and tissue sampling
Ventilatory variables were measured in unrestrained, unanesthetized rats using whole-body plethysmography according to standard procedures 30 . 45 to 60 min prior to the recording, females were injected either with vehicle or with the selective ORX 1 receptor antagonist (SB334867; 15 mg/kg). Ventilation was recorded at rest (room air) followed by hypercapnic exposure (5% CO 2 , balance air; 10 min). CO 2 levels used to assess ventilatory and behavioral responsiveness varies between 5 and 35% in animals and humans 31,32 . The level chosen here ensured a robust ventilatory response with minimal change in behavior to avoid movement artefacts that interfere with respiratory measurements. At the end of the experiment, rats were deeply anesthetised; as NMS increases vaginal sensitivity 33 , determination of the estrus cycle by vaginal smear was performed only at this time to avoid influencing results. Blood and brains were then harvested to obtain post-CO 2 samples immunohistochemistry and quantification of plasma E 2 levels. Experiments were performed between 13:00 and 15:00. Note that blood and brains were also obtained from a distinct group exposed to room air to obtain baseline data for E 2 and quantification of ORX A in hypothalamus extracts. For setup, protocol, and data collection details, see the Supplement.
Ovariectomy and 17β-estradiol (E 2 ) replacement Ovariectomy (OVX) was used to reduce circulating ovarian hormones chronically; surgery was performed according to standard procedures 34 . Two weeks after surgery, the CO 2 response was measured and compared between NMS and controls. A distinct group of OVX females received either vehicle (peanut oil; 100 µl) or one E 2 injection (3 or 10 µg) every 4 days to restore E 2 level within physiological range and mimic cyclic fluctuations. The effects of higher E 2 supplementation was tested in another group of OVX females by injecting with 25 µg. The last injections were performed on the day of the experiments. Based on the evidence suggesting that E 2 inhibits ORX neurons 27,28 , only E 2 was used in the replacement.

c-Fos/orexin-A immunohistochemistry
We first used c-Fos protein expression to determine whether NMS augments neuronal activation of ORX neurons in females. Based on ventilatory measurements, this initial evaluation was performed during the proestrus phase only. 40 μm coronal brain sections were doublelabeled with primary antibodies against c-Fos and ORX A to confirm cell phenotype. Since the physiological function of ORX neurons differs between hypothalamic sub regions [35][36][37] , single (ORX A only) and double-labeled cells were counted in the dorsomedial and lateral hypothalamus (DMH and LH, respectively) and the perifornical area (PeF). See Supplement for details.

Whole-cell patch-clamp recording of orexin neurons Identification of orexin neurons with an adeno-associated virus (AAV)
Four weeks-old females were injected with an AAV construct that expresses a green fluorescent protein (GFP) under the control of an ORX promoter. During surgery, rats received unilateral injections (1 µl/side) of the ORX: GFP virus in the following stereotaxic coordinates (from Bregma: RC: −2.6 mm; ML: 1.2 mm; DV: −9.0 mm). A 4week recovery period ensured consistent GFP labeling; the intensity of GFP labeling observed in the PeF area was more apparent than in adjacent areas.

Slice electrophysiology
Hypothalamic slices (300 µm) containing GFP-labeled ORX neurons were used for whole-cell patch-clamp recording of basic electrophysiological properties, excitatory postsynaptic currents converging onto ORX neurons, and responses to E 2 application (100 nM; 10 min). E 2 concentration was based on the literature 38 . Slice preparation and recording procedures were performed as described previously 39,40 . Since ORX neurons of the LH do not contribute to cardiorespiratory regulation 37 , recordings were performed in the PeF/DMH. The estrus cycle was determined after the brain was harvested; data were compared between NMS and controls. See the Supplement for details.

Statistics
Multifactorial analysis of variance (ANOVA) assessed the effects of NMS and estrous cycle on respiratory variables, 17β-estradiol (E 2 ), immunohistochemical, and electrophysiological data. CO 2 exposure was also considered for analyses of respiratory data and E 2 ; a repeated measures design was used when appropriate; equality of variance was tested. The relationship between E 2 and the ventilatory response to CO 2 was assessed using analysis of co-variance (ANCOVA) and a correlation z-test. Since patch-clamp data often originate from multiple cells from the same animal, a mixed ANOVA model (mixed-effect model) was used to ensure that between-group differences were not attributable to a specific subject 41 . ANOVA results are reported in the figures for clarity and conciseness. All data are presented as means ± SEM. A significant ANOVA results (P ≤ 0.05) was followed by a post hoc test (Fisher's least significant difference) to identify specific differences; a Bonferroni correction was applied when multiple comparisons were performed. Analyses were performed using Statview 5.0 (SAS Institute, Cary, NC, USA) and JASP (version 0.13; University of the Netherlands).

Results
The ventilatory response to CO 2 and 17β-estradiol (E 2 ) levels across the female's reproductive status CO 2 inhalation induced a rapid and robust hyperpnoea, which was greater in NMS females than controls (Fig. 1A, B). The intensity of the respiratory frequency and minute ventilation responses varied across the different phases of the estrus cycle and the largest difference between NMS and controls were observed during proestrus (Fig. 1A, B and Supplementary Fig. 1B). Baseline E 2 levels fluctuated across the estrus cycle and peaked during proestrus; NMS did not affect those values (Fig. 1C). During proestrus, CO 2 exposure augmented E 2 levels in control, but not NMS females (Fig. 1D). The E 2 levels measured following CO 2 exposure during proestrus were inversely correlated with the intensity of the breathing frequency response in NMS but not controls (Fig. 1E).
Prior to puberty, the hypercapnic ventilatory response was modest 42 ; the response did not differ between sexes and was unaffected by NMS ( Fig. 2A and see Supplementary Fig. 1C, D for effects on other respiratory variables). In adults, OVX reduced E 2 levels ( Supplementary  Fig. 2) and the CO 2 response such that the intensity of the hyperpnoea was now similar between groups (Fig. 2B). In OVX females, the first E 2 supplementation protocol restored plasma E 2 to physiological levels (Fig. 3C) and increased the respiratory frequency response in NMS but not controls. In both groups, the response was directly proportional to E 2 levels (Fig. 2B, C and Supplementary  Fig. 1). While the E 2 levels achieved with the second supplementation protocol exceed normal values (Fig. 2D), this treatment demonstrated that at that elevated E 2 inhibits the ventilatory response to CO 2 ( Fig. 2D and Supplementary Fig. 1E, F).
None of the ventilatory variables and other indicators of metabolism measured at rest differed between Fig. 1 Comparison of the hyperventilatory response to CO 2 inhalation (5% CO 2 ; 10 min) between adult females raised under standard conditions or subjected to neonatal maternal separation (NMS) across the different phase of the estrus cycle. A Original plethysmographic recording comparing ventilatory activity at rest and at the onset of CO 2 inhalation (blue arrow) in a female raised under control conditions (top trace) versus a female subjected to neonatal maternal separation (bottom trace; NMS: 3 h/day; postnatal days 3-12). B The breathing frequency response expressed as a percentage change from baseline (room air). 17β-estradiol (E 2 ) levels across the estrus cycles measured C while breathing room air (baseline) and D 30 min following CO 2 inhalation test (5% CO 2 ; 10 min). E Regression analysis comparing the relationship between E 2 levels during proestrus (post-CO 2 ; blue box) and the intensity of the breathing frequency response to CO 2 between NMS (red circles) and control females (white circles). Box plots: the top and bottom boundaries of the box indicate the 25 th and 75 th percentile, respectively. Within the box, the black bar indicates the median; the other bar (red or white) indicate the mean. The error bars indicate the 10 th and 90 th percentiles. The numbers in brackets below the boxes indicate the number of replicates in each group. Post hoc pairwise comparisons were performed only when warranted by ANOVA. *Indicates a value different from corresponding proestrus value at P < 0.05. † Indicates a value different from corresponding control value at P < 0.05. groups. Breathing frequency of "resting" OVX females was slightly higher than intact females but this was not sufficient to augment minute ventilation (Supplementary Table 1).

Excessive ORX modulation contributes to NMS-related increase in CO 2 response
Anomalies in ORX neurotransmission contributes to the pathophysiology of a panic-prone state in rats and in Fig. 2 Importance of ovarian function on the breathing frequency response to CO 2 inhalation (5% CO 2 ; 10 min). Breathing frequency response to CO 2 inhalation (5% CO 2 ; 10 min) in A pre-pubertal rat pups (14-15 days old) and B adult females with intact ovaries or following (OVX) with and without 17-β estradiol (E 2 ) replacement. Data from intact adult females include females without surgical procedure (from Fig. 1) and females that subjected to sham surgery that received vehicle injections (peanut oil; 100 µl). Ovariectomy or sham surgeries were performed 2 weeks prior to ventilatory measurements. E 2 replacement reproduced cyclic fluctuations by performing a daily injection every 4 days over 12 days prior to the experiments (4 injections in total). Each injection contained either vehicle (peanut oil, 100 µl) or E 2 (3 or 10 µg; normal levels or 25 µg; high levels). The histograms represent the frequency responses expressed as a percentage change from baseline (room air). Data are compared between rats raised under standard conditions (white bars) or subjected to neonatal maternal separation (NMS; red bars). Box plots: the top and bottom boundaries of the box indicate the 25 th and 75 th percentile, respectively. Within the box, the black bar indicates the median; the other bar (red or white) indicate the mean. The error bars indicate the 10 th and 90 th percentiles. The numbers in brackets below the boxes indicate the number of replicates in each group. Post hoc pairwise comparisons were performed only when warranted by ANOVA. *Indicates a value different from the corresponding intact value at P < 0.05. † Indicates a value different from the corresponding control value at P < 0.05. The relationship between plasma E 2 levels and the intensity of the hyperventilatory response in OVX females supplemented with C normal E 2 (3-10 µg) or D high E 2 (25 µg).
humans [11][12][13] . To determine whether enhanced ORX contributes to NMS-related increase in CO 2 responsiveness in females, we first quantified ORX A levels in hypothalamus extracts; values obtained in NMS females were 51% higher than controls (Fig. 3A). We then inactivated ORX 1 receptors by pre-treatment with SB334867. The treatment did not affect breathing at rest (Supplementary Table 1) and generally had limited effects on the CO 2 response; however, SB334867 prevented the excessive ventilatory response of NMS rats during proestrus ( Fig. 3C and Supplementary Fig.  3). As physiological data indicate that NMS-related anomalies in respiratory control are more important during proestrus (Fig. 1), we used c-Fos immunolabeling to determine if ORX neurons of NMS females were more active than control during that phase (Fig. 4). By comparing data with OVX females, we evaluated the sensitivity to E 2 withdrawal between groups. In the DMH, the percentage of ORX neurons expressing c-Fos was greater in NMS than Fig. 3 Effects of neonatal stress on ORX A in the hypothalamus and effectiveness to the selective orexin-1 receptor antagonist SB334867 on the hyperventilatory response to CO 2 . A Comparison of basal ORX A levels in hypothalamus extracts from females subjected to neonatal maternal separation (NMS; red bars) or raised under control conditions (white bars). Data were obtained from different each phase of the estrus cycle; however, data are pooled since the effect of ovarian status was not significant. B Effect of pre-treatment with SB334867 on the breathing frequency response of control females and C females subjected to NMS. Data were obtained during each phase of the estrus cycle; breathing was measured at the end of the CO 2 inhalation test (5% CO 2 ; 10 min) and expressed as a percentage change from baseline (room air). Data from SB334867-treated rats (15 mg/kg; I.P.; blue bars) are compared with those that received an injection of vehicle (50 mM HP-β-CD-10% + DMSO-2%) 45-60 min before start the plethysmography recordings. Box plots: the top and bottom boundaries of the box indicate the 25 th and 75 th percentile, respectively. Within the box, the black bar indicates the median; the other bar (red, white, or blue) indicate the mean. The error bars indicate the 10 th and 90 th percentiles. The numbers in brackets below the boxes indicate the number of replicates in each group. Post hoc pairwise comparisons were performed only when warranted by ANOVA. *Indicates a value statistically different from corresponding proestrus value at P < 0.05. † Indicates a value different from corresponding control value at P < 0.05. # Indicates a value different from corresponding vehicle value at P < 0.05. control during proestrus. Since OVX increased this ratio only in controls, the level achieved was now similar between groups (Fig. 4D). While a similar trend was observed in the PeF, neither NMS nor OVX affected the number of doublelabeled cells in the LH (Fig. 4E and F, respectively).

Neonatal maternal separation disrupts E 2 regulation of orexin neurons of the PeF/DMH
As our data indicate that NMS-related potentiation of ORX action on respiratory control is the greatest during proestrus, we then used whole-cell patch-clamp recording to assess the impacts of NMS and E 2 on ORX neurons. In naturally cycling females, membrane potential (Vm) did not change across the estrus cycle and values recorded in NMS females were slightly lower than controls. However, OVX augmented cell capacitance and the resting Vm of ORX neurons in NMS but not controls (Supplementary Table 2). None of the other basic cell properties was influenced by NMS or ovarian function (Supplementary Table 2). Ovarian status influenced the excitatory postsynaptic currents (EPSC) frequency, especially in controls (Fig. 5A, C). In those females, EPSC frequency was inversely proportional to the basal level of E 2 associated with each phase of the estrus cycle (Fig. 5C). In contrast with controls, ORX neurons from NMS females had the highest EPSC frequency during proestrus, when E 2 levels peaked (Fig. 5A, C). EPSC amplitude was not affected by NMS or the estrus cycle (Fig. 5A, E). Bath application of Fig. 4 Neonatal maternal separation (NMS) augments hypothalamic levels of orexin A (ORX A ) and augments c-Fos immunolabelling in ORX A neurons in adult female rats. A Schematic representation of frontal brain sections illustrating the hypothalamic region (blue square) that was isolated and homogenised to measure basal ORX A levels by ELISA assay. The schematic also illustrates the regions containing ORX neurons that were delineated to quantify c-Fos immunolabelling (adapted with permission from ref. 72 ). Photomicrographs comparing c-Fos (Texas red) and ORX A (FITC) immunolabelling between tissue sections from females in proestrus B 1 raised under control conditions (CTRL) or C 2 previously subjected to NMS (3 h/day, postnatal days 3-12). The third ventricle (3 V) and fornix (f) are identified as landmarks. Panels B 2 and C 2 were obtained at higher magnification for CTRL and NMS, respectively and white arrows identify double-labeled neurons. D Comparison of the proportion of ORX A positive neurons expressing c-Fos in the dorsomedial hypothalamus (DMH) between NMS and controls during proestrus and two weeks following ovariectomy (OVX). Similar analyses were performed in E the perifornical area (PeF) and F the lateral hypothalamus. Box plots: the top and bottom boundaries of the box indicate the 25 th and 75 th percentile, respectively. Within the box, the black bar indicates the median; the other bar (red or white) indicate the mean. The error bars indicate the 10 th and 90 th percentiles. The numbers in brackets below the boxes indicate the number of replicates in each group. Post hoc pairwise comparisons were performed only when warranted by ANOVA. † Significantly different from corresponding control value at P ≤ 0.05. *Indicates a value statistically different from corresponding proestrus value at P < 0.05. E 2 reduced EPSCs frequency in both control and NMS; however, the largest drop was observed during proestrus in NMS females (−0.63 versus −4.09 Hz for control and NMS, respectively). E 2 reduced EPSC amplitude in ORX cells from control but not NMS (Fig. 5F, G-I).

Discussion
Puberty and cyclic fluctuations in ovarian hormones are normal physiological processes but in a subpopulation of women, these events contribute to the onset and cyclic exacerbation of PD [43][44][45] . Thus, elucidating how ovarian function affects respiratory manifestations of PD is of utmost importance to our understanding of the pathophysiology of this disorder. Early life adversities are a significant risk factor for PD and the sex-specific enhancement of the ventilatory response to CO 2 inhalation in rats is an attractive model to address this question 29 . Here, we show that in NMS females endogenous release of E 2 during CO 2 inhalation is insufficient to maintain the ventilatory response within a normal range. These data support our main hypothesis and indicate that mature (functional) ovaries are necessary to observe an excessive ventilatory response to CO 2 inhalation in NMS females. This, and the involvement of the ORX system in  Fig. 2A measured in each stage are reported for comparison; values from NMS and controls were pooled since they are not statistically different. E Reports EPSC amplitudes. Note that since recordings in diestrus were performed only in controls (see F, G), data for this stage could not be compared between groups. Effect of E 2 application (10 min; 100 nM) on EPSC F frequency and G amplitude in slices from control females. EPSC frequency and amplitude data from NMS females are reported in (H, I), respectively. Within histogram bars, the numbers in brackets indicate the number of replicates in each group. Data are expressed as the mean ± SEM. Post hoc pairwise comparisons were performed only when warranted by ANOVA. † Significantly different from corresponding control value at P ≤ 0.05. *Indicates a value statistically different from corresponding proestrus value at P < 0.05.
this process strengthens the model and our demonstration that NMS disrupts E 2 regulation of ORX neurons point to a novel mechanism in the pathophysiology of PD in females.
Neonatal maternal separation affects E 2 signaling regulating the response to CO 2 inhalation Progesterone and E 2 both fluctuate across the estrus cycle. However, the largest NMS-related increase in CO 2 response coincided with the peak in E 2 (proestrus) and following OVX, restoring normal E 2 levels alone was sufficient to reinstate an excessive ventilatory response in NMS rats. We, therefore, conclude that E 2 plays a primary role in the enhanced response to CO 2 . NMS does not affect the cyclic changes in E 2 or progesterone under basal conditions ( Fig. 1C; 46 ) but attenuates E 2 release following CO 2 exposure, especially during proestrus. The sympathetic system regulates ovarian E 2 secretion 47 , and ovarian aromatase expression peaks during proestrus 48 . Since plasma E 2 closely reflects brain levels 49 , such impairment in E 2 signalling within the brain of NMS females is likely. The inability of NMS females to augment E 2 in response to CO 2 during proestrus suggests that NMS reduced E 2 synthesis capacity and thus compromise the response to an acute challenge. The relationships between E 2 levels and the intensity of the ventilatory response suggest that impairment of E 2 signalling contributes to the abnormal respiratory phenotype of NMS females. Experiments performed on OVX females show that E 2 can attenuate the excessive CO 2 response of NMS females; however, higher levels are necessary.

Neonatal maternal separation augments ORX activation under basal conditions
Having previously shown that NMS does not affect the carotid body's response to CO 2 in male and female rats 50 , our investigation focused on central mechanisms regulating the ventilatory response to CO 2 . Orexin neurons project to key medullary areas regulating breathing, including those that generate respiratory rhythm and contribute to CO 2 chemosensitivity 9 . The clinical evidence implicating ORX in PD is important 11,51 and our results showing that NMS augments ORX A levels in the hypothalamus are consistent with clinical and preclinical data. A similar increase has been reported in males 18 but since regulation of ORX neurons likely differs between sexes 19,22 , testing this effect in females was necessary. ORX synthesis is activity-dependent and highly plastic 52 . In light of the close relationship between the hypothalamo-pituitary adrenal (HPA) axis and the ORX system 19,53 the enhancement of basal HPA activity commonly reported in animals and humans who experienced early life adversities could explain the higher ORX A level in hypothalamic extracts [54][55][56] . In adult rats, however, disruption of HPA axis function by NMS is significant only in males 57,58 . Thus, another mechanism should be considered to explain this result.
In control females, comparison of c-Fos expression between females experiencing high (proestrus) and low (OVX) E 2 clearly supports an inhibitory action of estrogens on ORX neurons. Conversely, the high c-Fos/ ORX A ratio observed in NMS females, regardless of the ovarian function, suggests a generally higher degree of basal activity and a reduced sensitivity to E 2 and/or insufficient levels. This result therefore provides a plausible explanation for greater level of ORX A in hypothalamic extracts and the larger ventilatory response to CO 2 inhalation. The latter interpretation is supported by fact that NMS augmented the c-Fos/ORX A ratio in PeF/DMH areas that, unlike the LH, regulate cardiorespiratory homeostasis 37 . Moreover, pre-treatment with SB334867 prevented NMS-related increase in the ventilatory response to CO 2 .
Neonatal maternal separation disrupts E 2 regulation of ORX neurons ORX neurons are essential to several homeostatic functions. To the best of our knowledge, this is the first study documenting the impact of natural fluctuations in ovarian hormones (and OVX) on basic properties and excitatory synaptic inputs in females. While natural fluctuations in ovarian hormones have no impact on basic properties, the Vm and capacitance values obtained during the natural E 2 nadir (metestrus) differs from those recorded following OVX. While OVX is the gold standard in preclinical research for evaluating gonadal hormone effects in females 59 , these results remind us that the changes induced by OVX may be more complex than a simple reduction of circulating hormones. Keeping that limitation in mind, the opposing effects of OVX on Vm between NMS and controls nonetheless indicate that NMS affects the way ovarian hormones influence this important property of ORX neurons.
Orexin neurons are the target of multiple afferent signals form diverse origins 10,60,61 . The frequency and amplitude of spontaneous EPSC's reflect the number and the strength of excitatory synaptic inputs acting on ORX neurons, respectively. E 2 acts via both membrane and nuclear receptors and the results reported here provide valuable insights into the mechanisms by which endogenous E 2 contributes to inhibition of ORX neurons. EPSC frequencies measured in controls were inversely related to basal plasma E 2 levels associated with natural cyclic fluctuations or OVX; exogenous E 2 elicited a similar decrease in frequency in the minutes that followed its application onto slices. This implies that both E 2 receptor types could regulate the number of synapses converging onto ORX neurons. Conversely, the fact that only acute E 2 reduced EPSCs amplitude suggests that regulation of synaptic strength by E 2 signalling is rapid but transient. E 2 is generally known to promote dendritic spine formation, potentiate excitatory synaptic transmission, and reduce the efficacy of GABAergic inhibition [62][63][64][65] . In the cortex and the hippocampus, however, application of LY 3201 (a selective agonist of the nuclear receptor ERβ) can elicit an opposite response by reducing dendritic spines and increasing expression of glutamic acid decarboxylase (GAD) 66 . This increase in GAD expression, combined with a reduction in the expression of NMDA receptors shifts the balance between excitatory and inhibitory neurotransmission in favor of inhibition 66 . Together, these effects explain the anxiogenic and anxiolytic actions of ERα and ERβ, respectively 67 . Since E 2 inhibits expression of ERs 68 , region-specific changes in the relative expression of ERα and ERβ likely contribute to the phasedependent effects reported here. While NMS affects ERβ expression in the hippocampus of males 69 , its impact in females is yet to be tested.
At the system level, the CO 2 responses measured in controls indicate that as E 2 declines across the cycle, the increased activation of ORX cells is dampened to prevent excessive hyperpnoea. Obviously, this mechanism is not fully functional in NMS females. Regulation of ORX neurons is a complex process that involves an important local network of neurons and astrocytes 60 but obviously, NMS reduces E 2 's actions on these cells. In fact, the CO 2 response following E 2 replacement in OVX NMS females suggests that depending on the concentration administered, E 2 may have excitatory or inhibitory effects. Interestingly, PD is rare following menopause but E 2 replacement therapy has been linked with the development of panic attacks in some patients 70 .

Limitations and conclusion
Inadequate modeling of human disease hinders translation of basic knowledge into effective treatment for human 21 . While our study shows that NMS closely reproduces developmental and cyclic changes in the respiratory manifestations of PD and enhancement of ORX modulation, we must keep in mind that animal research cannot reproduce the complex psychosocial reality often associated with PD. Furthermore, the estrus cycle in rodents is not equivalent to the menstrual cycle in humans 59 . That being said, NMS nonetheless meets key criteria expected from an animal model, including timedependent and sex-specific effects on respiration 20,71 . The results reported here, therefore, offer valuable insights into the basic mechanism in this neurological disorder affecting female rats. Our demonstration that NMS disrupts the inhibitory actions of E 2 on respiratory control are significant as they offers new avenues to alleviate PD.