Abstract
Stress impairs fertility, at least in part, via inhibition of gonadotropin secretion. Luteinizing hormone (LH) is an important gonadotropin that is released in a pulsatile pattern in males and in females throughout the majority of the ovarian cycle. Several models of stress, including acute metabolic stress, suppress LH pulses via inhibition of neurons in the arcuate nucleus of the hypothalamus that coexpress kisspeptin, neurokinin B, and dynorphin (termed KNDy cells) which form the pulse generator. The mechanism for inhibition of KNDy neurons during stress, however, remains a significant outstanding question. Here, we investigated a population of catecholamine neurons in the nucleus of the solitary tract (NTS), marked by expression of the enzyme dopamine beta-hydroxylase (DBH), in female mice. First, we found that a subpopulation of DBH neurons in the NTS is activated (express c-Fos) during metabolic stress. Then, using chemogenetics, we determined that activation of these cells is sufficient to suppress LH pulses, augment corticosterone secretion, and induce sickness-like behavior. In subsequent studies, we identified evidence for suppression of KNDy cells (rather than downstream signaling pathways) and determined that the suppression of LH pulses was not dependent on the acute rise in glucocorticoids. Together these data support the hypothesis that DBH cells in the NTS are important for regulation of neuroendocrine and behavioral responses to stress.
- gonadotropin-releasing hormone
- luteinizing hormone
- norepinephrine
- nucleus of the solitary tract
- pulses
- stress
Significance Statement
Stress impairs fertility, at least in part, via inhibition of gonadotropin secretion. The gonadotropin luteinizing hormone (LH) is secreted in a pulsatile pattern in males and females throughout most of the ovarian cycle to support gamete development and sex steroid production in both sexes. Here, we investigated a population of catecholamine neurons in the nucleus of the solitary tract (NTS) for their role in regulation of pulsatile LH secretion. We demonstrate that these neurons are sufficient to suppress pulsatile LH secretion via inhibition of kisspeptin neurons in ovariectomized mice. Moreover, these NTS neurons are also sufficient to augment corticosterone concentrations and induce sickness-like behavior, which raises the possibility that these neurons are important for neuroendocrine and behavioral responses during stress.
Introduction
In mammals, generalized stress responses often include suppression of reproduction via inhibition of the hypothalamic–pituitary–gonadal (HPG) axis, activation of the hypothalamic–pituitary–adrenal (HPA) axis, and changes in behavior. These responses are mediated by key hypothalamic neuroendocrine neurons, which not only respond during stress but play key roles in maintaining homeostatic regulation (McCosh et al., 2019a). For HPA axis activation, stress-induced activation of corticotropin-releasing hormone (CRH) neuron activity in the paraventricular nucleus (PVN) is essential. Regarding the HPG axis, stress reduces activation of neurons in the arcuate nucleus (ARC) that contain kisspeptin, neurokinin B, and dynorphin (termed KNDy cells), resulting in the loss of gonadotropin-releasing hormone (GnRH) pulses and subsequent luteinizing hormone (LH) pulses (Clarkson et al., 2017; Moore et al., 2018). Although critical, our understanding of how these hypothalamic centers are modulated during stress remains limited.
Acute metabolic stress is a physiologically relevant stress type that can be modeled with insulin-induced hypoglycemia (IIH) or glucoprivation induced by 2-deoxyglucose, a glucose antimetabolite. These acute metabolic stress models elicit suppressed LH pulses in humans (Oltmanns et al., 2001), monkeys (Chen et al., 1996), sheep (Clarke et al., 1990), rats (Goubillon and Thalabard, 1996), and mice (McCosh et al., 2019b), via a mechanism involving impairment of upstream pulse generator activity. For example, LH pulsatile suppression during IIH was accompanied by reduced KNDy neurons expressing c-Fos [a marker of neuronal activation (Bullitt, 1990)] in mice (McCosh et al., 2019b) and reduced mediobasal hypothalamic multiunit activity in monkeys (Chen et al., 1996), indicating impairment of KNDy cells during IIH. Furthermore, evidence that the kisspeptin-induced LH increase was not abrogated during IIH suggests uncompromised GnRH cells and gonadotrophs and highlights KNDy cells as a target of inhibition of LH pulses during stress (McCosh et al., 2019b). The premise that KNDy cells are a primary site of suppression is challenged by data in rats with surgical ablation of the area postrema (located in the brainstem) that display normal LH pulses during IIH (Cates and O'Byrne, 2000). This collective work supports the hypothesis that LH pulses are suppressed by an intervening neurocircuit involving the area postrema, but that KNDy cells, GnRH cells, and gonadotrophs are not themselves directly impaired by hypoglycemia.
The area postrema is a circumventricular organ that contains glucose-sensitive neurons, enabling detection of hypoglycemia (Adachi et al., 1991). Although the identity of the putative neurocircuit connecting the area postrema to KNDy neurons is unknown, several lines of evidence support a role for ascending noradrenergic neurons projecting to the PVN as part of this circuit. First, the area postrema innervates the nucleus of the solitary tract (NTS; Cunningham et al., 1994). Second, neurons in the NTS express c-Fos following IIH (Paranjape and Briski, 2005). Third, neurons in the NTS, including catecholamine cells located in and around the NTS, project to several regions implicated in the control of reproduction including the PVN, the ARC, and the preoptic area (the site of GnRH cell bodies; Sawchenko and Swanson, 1982). Regarding the PVN, evidence supports the importance of this intervening hypothalamic area in mediating reproductive suppression. For example, alpha-adrenoreceptor agonists, including norepinephrine, injected into the PVN are sufficient to suppress LH pulses in rats (Tsukamura et al., 1994). Furthermore, ablation of catecholamine neurons that project to the PVN with saponin conjugates prevented the blockade of ovarian cyclicity in rats induced by repeated glucoprivation (I'Anson et al., 2003). Finally, neurons in the NTS that contain dopamine beta-hydroxylase (DBH; a marker for norepinephrine and epinephrine cells) have also been implicated in the activation of PVN-containing CRH neurons following peripheral administration of IL-1β (Li et al., 1996), which likely generalizes to other homeostatic stress types, including IIH.
Here, we examined activation of DBH neurons in the NTS during acute metabolic stress in ovariectomized (OVX) female mice and used chemogenetics to test the effects of NTSDBH cell activation on the neural control of LH pulsatile secretion, necessary for both males and females.
Materials and Methods
Animals
OVX females were used in this study because we previously identified robust suppression of LH pulses in this model during acute metabolic stress (McCosh et al., 2019b). Adult female mice were housed under standard laboratory conditions with a 12 h light/dark photocycle in a double barrier vivarium at the University of California, San Diego. For Experiment 1, female C57/BL6 mice (Envigo) were used. For Experiments 2, 3, and 4, female DBH-Cre-positive mice (and Cre-negative littermates) were used (Roman et al., 2016); a subset of these mice carried the Kiss1-hrGFP transgene, in which GFP is expressed under the Kiss1 promoter (RRID:IMSR_JAX:023425; Cravo et al., 2013). Mice had free access to feed (irradiated chow #2920X, Harlan) and water, except when noted below. Animals were OVX at ∼8 weeks of age via bilateral lumber laparotomy under isoflurane anesthesia with aseptic technique. Acclimatization to blood sampling and intraperitoneal (i.p.) injection procedures were performed daily for 5 weeks before experiments, as described previously (McCosh et al., 2018). Blood samples were collected between 8:30 A.M. and 12 P.M., with lights on at 6 A.M. All animal procedures were performed in accordance with the University of California, San Diego Institutional Animal Care and Use Committee policies and in accordance with the National Institutes of Health guidelines for the care and use of research animals.
Neurosurgery, virus, and chemogenetic activation
Mice (∼10 weeks of age) were anesthetized with ketamine and xylazine and placed in a stereotaxic frame (David Kopf Instruments) for bilateral neural injection of virus with aseptic technique. The post of the stereotaxic frame was angled 10° caudal and fitted with a 10 µl Hamilton syringe with a 33 gauge needle. Burr holes were drilled through the skull to allow injection at −5.7 mm from the bregma, ±0.5 mm lateral, and 5.06 mm below the surface of the skull (Fig. 2A). All adeno-associated viruses (AAV) were produced in our laboratory using standard techniques using plasmids pAAV2/1 [a gift from James M. Wilson (Addgene plasmid #112862; http://n2t.net/addgene:112862; RRID:Addgene_112862)], pAdDeltaF6 [a gift from James M. Wilson (Addgene plasmid #112867; http://n2t.net/addgene:112867; RRID:Addgene_112867)], and pAAV-hSyn-DIO-hM3D(Gq)-mCherry [a gift from Bryan Roth (Addgene plasmid #44361; http://n2t.net/addgene:44361; RRID:Addgene_44361)] to create the designer receptor exclusively activated by designer drug (DREADD) virus, or pAAV-hSyn-DIO-mCherry [a gift from Bryan Roth (Addgene plasmid #50459; http://n2t.net/addgene:50459; RRID:Addgene_50459)] to create the mCherry control virus. Three groups of mice were generated: (1) DBH-Cre-negative mice that received the DREADD virus, (2) DBH-Cre-positive mice that received the mCherry control virus, and (3) DBH-Cre-positive mice that received the DREADD virus. On experimental days, mice received i.p. injection of saline or clozapine N-oxide (CNO; BML-NS105, Enzo Life Sciences, 1 mg/kg mouse body weight, dissolved in saline). The ability of this chemogenetic approach (i.e., the same DBH-Cre mouse line and the same viral vector delivered via stereotaxic injection of AAV1 into the NTS) to increase cell firing in NTSDBH cells has been demonstrated previously (Roman et al., 2016). In the present study, we observed robust c-Fos staining in NTSDBH cells transduced with the hM3D(Gq) virus, confirming that these neurons were indeed activated in response to CNO.
Neural tissue collection and processing
Mice were deeply anesthetized with pentobarbital (Fatal-Plus) and perfused with heparinized saline and 4% paraformaldehyde in phosphate-buffered saline (PBS). Neural tissue was collected, stored in 4% paraformaldehyde overnight, and then transferred to 30% sucrose for at least 2 d. Brains were sectioned at 40 µm into three series (each section 120 µm apart) and stored in a cryopreservative solution (Watson et al., 1986) at −20°C until immunohistochemistry (IHC) was performed.
Evaluation of transduction efficiency
Tissue was processed for IHC to assess transduction efficiency using mCherry. IHC was performed on tissue from 7.08 to 7.44 mm caudal to the bregma (contains activated DBH cells identified in Experiment 1) for detection of DBH and mCherry. All steps were performed at room temperature with gentle mixing unless otherwise noted. First, tissue was washed 12 times in 0.1 M phosphate buffer (PB) and then incubated in a boiling antigen retrieval solution (Citra buffer, Thermo Fisher Scientific) for 10 min, twice. Tissue was then washed four times in 0.1 M PBS (standard washing step), then incubated in 1% H2O2 for 10 min, and washed again. Tissue was then placed in a blocking solution [4% normal goat serum (Jackson Laboratory) in 0.1 M PBS with 0.4% Triton X-100] for 1 h, and then in rabbit anti-DBH (ImmunoStar, 22806), and diluted 1:16,000 in a blocking solution for 18 h at 4°C. Tissue was then incubated in biotinylated goat anti-rabbit diluted 1:500 in a blocking solution, ABC elite (dilution 1:500 in PBS, Vector Laboratories), biotinylated tyramide (1:250 dilution in PBS, with 0.003% H2O2, Akoya Biosciences), and then streptavidin conjugated to Alexa 647 (diluted 1:200 in PBS, Life Technologies), with intervening wash steps. For detection of mCherry, tissue was incubated in a protein blocking solution for 1 h and then in RFP booster-Alexa568 nanobody (ChromoTek, rb2AF58-50) diluted 1:1,000 in a protein blocking solution for 18 h at 4°C. Tissue was then washed in 0.1 M PB, mounted on superfrost slides, and coverslipped with gelvatol (Harlow and Lane, 2006). This DBH antibody has been used extensively in mouse neural tissue (Toyoda et al., 2022; Choi et al., 2023); importantly no staining with this antibody is noted in Dbh knock-out mice (Lustberg et al., 2022). DBH-Cre-positive animals that had fewer than five total mCherry neurons in the analyzed region of the NTS were excluded from further analysis.
Experiment 1: are NTSDBH cells activated during acute metabolic stress?
To determine if NTSDBH cells are activated during acute metabolic stress, neural tissue was collected 2 h after insulin injection as described previously (McCosh et al., 2019b). Briefly, OVX mice were fasted for 5 h and then received an i.p. injection of saline (n = 7) or insulin (n = 6; 0.75 mU/g body weight, Humulin R, NDC 0002-8215-01), and fixed neural tissue was collected 2 h later. This time point was chosen to capture the change in the c-Fos expression (∼60 min for protein expression) in response to maximal hypoglycemia, which occurs ∼60 min following insulin injection (McCosh et al., 2019b).
IHC was performed on one complete series of tissue from the brainstem, encompassing the rostral–caudal extent of the NTS, for detection of c-Fos and DBH. Since both antibodies were produced in rabbits, the first antibody was used at an exceeding low concentration, and the signal was amplified for detection, prohibiting the second secondary from detecting the first primary (Shindler and Roth, 1996). Evidence that this method is effective is demonstrated by the virtual absence of c-Fos stating in saline-treated animals (Fig. 1). All steps were performed at room temperature with gentle mixing unless otherwise noted. First, staining for DBH was performed as described in the “Evaluation of transduction efficiency” section above. Then, for detection of c-Fos, tissue was incubated in a protein blocking solution for 1 h and then in a rabbit anti-c-Fos antibody (Cell Signaling Technology, Clone 9F6 #2250) diluted 1:1,000 in a protein blocking solution for 18 h at 4°C. Tissue was then washed and incubated in goat anti-rabbit IgG conjugated to Dylight 488 (Life Technologies) diluted 1:200 in a protein blocking solution. Finally, tissue was washed in 0.1 M PB, mounted on superfrost slides, and coverslipped with gelvatol (Harlow and Lane, 2006). Due to poor tissue quality or missing sections, some animals were excluded from individual regions.
Experiment 2: does chemogenetic activation of NTSDBH cells alter pulsatile LH secretion, corticosterone concentrations, or KNDy cell activation?
Experiment 2A: LH secretion pattern
To determine if chemogenetic activation of NTSDBH cells alters LH secretion patterns, three groups of OVX mice were generated: (1) DBH-Cre-negative mice that received the DREADD virus (n = 5), (2) DBH-Cre-positive mice that received the mCherry control virus (n = 6), and (3) DBH-Cre-positive mice that received the DREADD virus (n = 14; Fig. 2A). Two weeks after neurosurgery, mice were randomly assigned to receive saline or CNO; 2 weeks later, each animal received the alternate treatment in a cross-over design (the order of treatments was randomized). Blood samples during each run of the cross-over experiment were collected at 6 min intervals for 60 min before and 90 min after injection for determination of LH concentrations (Fig. 2B). Mice that exhibited only one LH pulse in the preinjection period or DBH-Cre-positive mice that had insufficient mCherry staining were excluded from analysis, resulting in five, four, or five mice in each group, respectively.
Experiment 2B: corticosterone concentrations
This experiment was conducted in a cross-over design with animals receiving saline or CNO in a randomized order in OVX DBH-Cre-positive animals that received the DREADD virus (n = 7) or the mCherry control virus (n = 5). A separate cohort of animals was generated for this experiment (i.e., animals from Experiment 2A were not reused). The experimental design was similar to Experiment 2A except blood samples were collected at 15 min intervals from 15 min before until 60 min after injection of saline or CNO and at 30 min intervals for an additional 2 h for determination of corticosterone concentrations. Animals that had corticosterone concentrations >400 ng/ml in the baseline sample were excluded from analysis, resulting in the analysis of four mice per group.
Experiment 2C: KNDy cell activation
A subset of mice used in Experiments 2A and 2B, which carried the Kiss1-hrGFP transgene, were used in this experiment. Two weeks after the final blood sampling session in Experiment 2A or 2B, mice were randomly assigned to receive saline (n = 4) or CNO (n = 6), and fixed neural tissue was collected 2 h later. To assess activation of KNDy cells, tissue from the hypothalamus was rinsed 12 times in 0.1 M PB and then incubated in a boiling antigen retrieval solution for 10 min, twice. Tissue was then washed, incubated in 1% H2O2 for 10 min, and washed again. Tissue was then placed in a blocking solution for 1 h and then in rabbit anti-c-Fos (1:1,000, Cell Signaling technology, Clone 9F6 #2250) for 18 h at 4°C. Tissue was then washed, incubated in goat anti-rabbit conjugated to Alexa 647, and then washed again. Tissue was placed in a blocking solution for 1 h and then in rabbit anti-GFP conjugated to Alexa 488 (Life, A21311) in a blocking solution for 18 h at 4°C. Tissue was then washed in 0.1 M PB, mounted on superfrost slides, and coverslipped with gelvatol (Harlow and Lane, 2006).
Experiment 3: are GnRH cells directly inhibited by NTSDBH cell activation?
The LH response to exogenous kisspeptin was determined in OVX DBH-Cre-positive mice that received the stimulatory DREADD virus following i.p. injection of saline or CNO in a cross-over design (the order of treatments was randomized). Blood samples were collected at 6 min intervals for 60 min before and 90 min after i.p. injection of saline or CNO to confirm suppression of LH pulses by NTSDBH cell activation. Then, 90 min after saline or CNO exogenous injection, kisspeptin was administered via i.p. injection [2 µg/g body weight (McCosh et al., 2019b; Makowski et al., 2020); rat kisspeptin-10; 4243, Tocris Bioscience], and an additional blood sample was collected 10 min later. Two weeks later, the experiment was repeated except that each animal received the opposite treatment (CNO or saline; Fig. 6A). This cross-over experiment included eight mice.
Experiment 4: is suppression of LH secretion dependent upon elevated corticosterone secretion?
To determine if adrenal hormones such as corticosterone are necessary for suppression of LH pulses during NTSDBH cell activation, LH secretion was monitored during NTSDBH cell activation before and after adrenalectomy. The experimental design is shown in Figure 7A. Two weeks after OVX, DBH-Cre-positive mice received the stimulatory DREADD virus, and blood samples were collected at 6 min intervals for 60 min before and 90 min after i.p. injection of CNO. Five days after the first sampling period, mice were adrenalectomized (ADX) via lumber laparotomy under isoflurane anesthesia using aseptic technique and subsequently received 0.9% saline for drinking water. Nine days after adrenalectomy, the blood sampling procedure and CNO injection was repeated. To confirm successful adrenalectomy, corticosterone concentrations were measured at 30 min intervals during 1 h of restraint stress 2–3 d prior to each blood sampling period. This experiment included eight mice; however, one animal was excluded from analysis based on having only one LH pulse in a pre-CNO sampling period, and another mouse was excluded for having no apparent mCherry staining in the NTS.
Hormone enzyme-linked immunosorbent assays
LH concentrations were determined in 3 µl tail-tip whole blood samples that were immediately diluted 1:20 into an assay buffer and stored on ice until frozen (−20°C). Samples were assayed in singleton using a protocol and antibodies based on the method of Steyn et al. (2013), as well as a recently published detection LH antibody validated for use in our laboratory (Kreisman et al., 2021, 2022). Samples in Experiment 2 were assayed with anti-LH antiserum from the National Hormone and Peptide Program (AFP240580Rb). The functional sensitivity was 0.2 ng/ml, and the intra- and interassay coefficients of variation were 2.5 and 7.4%, respectively. Samples for Experiments 3 and 4 were assayed with anti-LH from Medix Biochemica (SPRN-5), the functional sensitivity was 0.067 ng/ml, and the intra- and interassay coefficients of variation were 2.1 and 4.5%, respectively.
For detection of corticosterone, 12 µl tail-tip whole blood samples were centrifuged at 5,000 rpm for 15 min, and serum was harvested and stored at −20°C until assay. Serum (4 µl) was assayed using the DetectX Corticosterone EIA kit (K014; Arbor Assays), as described previously (Luo et al., 2016). Intra-assay and interassay coefficients of variation were 3.8 and 9.4, respectively, and the assay sensitivity was 18.6 pg/ml.
Microscopy
Images were collected using a Ti2-E inverted microscope with a DS-Qi2 monochrome camera, operated with Nikon Instruments Software (NIS) Elements software system. For each animal, images were taken of every section of a complete series of sections, containing the NTS. Images were ordered using the Franklin and Paxinos mouse brain atlas. For Experiment 2C, images were collected throughout the ARC for assessment of GFP and c-Fos. The number of cells per hemisection was counted with ImageJ by an observer blinded to treatment.
Statistical analysis
Mean LH was calculated as the mean of all LH concentrations in the sampling period. LH pulses were identified using the PulsaR software (Porteous et al., 2021), using the following parameters: smoothing, 0.7; G1 = 2.2; G2 = 2.7; G3 = 1.9; G4 = 1.5; G5 = 1.2; and peak split = 2.5. The program was run on data from pre- and postinjection (saline or CNO) periods separately; peaks that occurred at either the first or last sample were excluded. LH pulse amplitude was calculated as the difference between concentrations at the peak and the preceding nadir. Pulse frequency was calculated as the number of pulses that occurred per hour.
Statistical analysis was performed using JMP Pro, version 16 (SAS Institute). For Experiment 1, differences in the total number of DBH cells, the total number of c-Fos cells, and the percentage of DBH cells that contained c-Fos were compared between groups (saline vs insulin) with separate Mann–Whitney tests for each anatomical position in the brainstem. Nonparametric tests were performed due to the unequal variance between groups that could not be ameliorated with transformations. For Experiments 2A and 2B, mean LH, LH pulse frequency, LH pulse amplitude, and corticosterone concentrations were each compared between groups with two-way ANOVA; the model included treatment (saline vs CNO) and time (before vs after injection). For Experiment 2C, the number of kisspeptin cells and the percentage of these that contained c-Fos were compared between groups (saline vs CNO) with t tests for each aspect of the ARC (i.e., rostral, middle, and caudal). For Experiment 3, LH concentrations were compared between groups with two-way ANOVA; the model included pretreatment (saline vs CNO) and sampling time (0 min before vs 10 min after kisspeptin). For Experiment 4, mean LH, LH pulse frequency, LH pulse amplitude, and corticosterone concentration were each compared between groups with two-way ANOVA. The model included treatment (adrenal intact vs ADX) and time (before vs after CNO). For confirmation of adrenalectomy, corticosterone concentrations were compared between groups with two-way ANOVA; the model included treatment (restraint vs control) and time after treatment. Where appropriate, post hoc analysis was performed using Tukey's HSD test.
Results
Experiment 1: are NTSDBH cells activated during acute metabolic stress?
Representative images of DBH and c-Fos IHC in mice treated with saline (A, C, E) or insulin (B, D, F) are shown in Figure 1. In both groups, DBH cells were apparent throughout the rostral–caudal extent of the NTS, but the number of DBH cells did not differ between groups (p > 0.05 in all regions; data not shown). In contrast, the total number of c-Fos cells were significantly increased in response to insulin in a restricted region of the NTS corresponding to −6.84, −7.08, −7.20, −7.32, and −7.44 mm from the bregma (Z = 2.34, p = 0.019; Z = 2.86, p = 0.004; Z = 3.08, p = 0.002; Z = 2.78, p = 0.005; and Z = 2.23, p = 0.026, respectively; Fig. 1G). Similarly, the percentage of DBH cells that contained c-Fos was also restricted to −6.84, −6.96, −7.08, −7.2, −7.32, and −7.44 mm from the bregma following insulin (Z = 2.43, p = 0.015; Z = 2.19, p = 0.029; Z = 2.77, p = 0.006; Z = 3.08, p = 0.002; Z = 2.40, p = 0.016; and Z = 2.31, p = 0.021, respectively; Fig. 1H).
Acute hypoglycemia causes activation of NTSDBH cells. Representative IHC images of DBH (top panels), c-Fos (middle panels), or merged images (bottom panels) in the NTS (−7.32 mm from the bregma) in OVX female mice that received an i.p. injection of saline (A, C, E) or insulin (0.75 mU/g body weight; B, D, F). Main panels, scale bar = 100 µm; inset panels, scale bar = 20 µm. Total number of c-Fos immunoreactive cells throughout the NTS (G). Percent of DBH cells containing c-Fos throughout the NTS (H). Open circles, saline (n = 7); closed circles, insulin (n = 6). *p < 0.05 versus saline-injected animals at the same anatomical location.
Experiment 2: does chemogenetic activation of NTSDBH cells alter pulsatile LH secretion, corticosterone concentrations, or KNDy cell activation?
Histological and behavioral observations
Stereotaxic injection of virus resulted in transduction of cells in the NTS (Fig. 2), but not the ventrolateral medulla or the locus ceruleus, in DBH-Cre-positive mice. Three mice were excluded from further analysis in Experiment 2 based on having no apparent mCherry expression in the NTS. In the NTS of DBH-Cre-positive mice injected with the stimulatory DREADD or mCherry control virus, 222.1 ± 77.9 DBH cells were counted per mouse in which 35.4 ± 1.8% of DBH cells contained mCherry and 92.1 ± 7.6% of mCherry cells contained DBH (representative image in Fig. 2D,F,H). In contrast, virtually no mCherry expression was observed in DBH-Cre-negative mice (representative image in Fig. 2C,E,G; two mCherry cells in four DBH-Cre-negative mice). Of note was the clear sickness-like behavior exhibited by DBH-Cre-positive mice that received the stimulatory DREADD virus and injected with CNO. Observed behaviors included lethargy, hunched posture, and orbital tightening that were similar to mice injected with insulin (Experiment 1). These behaviors were not observed in any other condition (i.e., following saline injection, DBH-Cre-negative mice, or those that received the mCherry control virus).
Bilateral neurosurgical approach (A) and experimental design including timeline (B) for Experiment 2A and 2B. N.Sx, neurosurgery. Representative images of transduction in the NTS. DBH (C, D), mCherry (E, F), and merged images (G, H) for a DBH-Cre-negative mouse injected with the stimulatory DREADD virus (C, E, G) and a DBH-Cre-positive mouse injected with the stimulatory DREADD virus (D, F, H). Arrowheads denote DBH cells with mCherry. Main panels, scale bar = 100 µm; inset panels, scale bar = 10 µm.
Experiment 2A: LH secretion pattern
Figure 3 shows LH concentration patterns during chemogenetic activation of NTSDBH cells and associated controls. LH pulses were evident in all mice in the preinjection sampling periods (Fig. 3A). Mean LH, number of pulses per hour, and pulse amplitude were not altered by saline injection in any group of mice (Fig. 3B–J). In DBH-Cre-negative mice that received the stimulatory DREADD virus, or DBH-Cre-positive mice that received the mCherry virus, neither mean LH (Fig. 3B,E), number of pulses per hour (Fig. 3C,F), nor pulse amplitude (Fig. 3D,G) was altered by CNO treatment. In contrast, in DBH-Cre-positive mice that received the stimulatory DREADD virus, both mean LH and the number of pulses per hour were significantly reduced following CNO injection compared with all other sampling periods (F1,12 = 10.63, p = 0.007 and F1,12 = 35.56, p ≤ 0.001, respectively; Fig. 3H,I); pulse amplitude was not altered in the three animals expressing a pulse post-CNO (Fig. 3J).
LH concentrations during activation of NTSDBH cells. Representative LH pulse profiles from three OVX female mice in this cross-over experiment (A). Top, DBH-Cre negative with stimulatory DREADD virus; middle, DBH-Cre positive with mCherry control virus; and bottom, DBH-Cre positive with DREADD virus. Arrows indicate time of i.p. injection of saline (left panels) or CNO (right panels). Darkened data points represent pulses. Group means ± SEM for LH concentration (B, E, H), number of pulses per hour (C, F, I), and pulse amplitude (D, G, J) before (open bars) and after CNO (gray bars). Groups included: DBH-Cre negative with stimulatory DREADD virus, n = 5; DBH-Cre positive with mCherry virus, n = 4; and DBH-Cre positive with stimulatory DREADD virus, n = 5. Bars with different letters are significantly different.
Experiment 2B: corticosterone concentrations
Figure 4 shows mean corticosterone concentrations in DBH-Cre-positive mice that received either the control or stimulatory DREADD virus and were treated with saline or CNO. In mice that received the control virus (Fig. 4A), corticosterone concentrations did not differ between saline and CNO treatment (main effect of treatment p > 0.05), nor was there a significant interaction between time and treatment (p > 0.05). Corticosterone concentrations were significantly, though modestly, elevated at 15 min relative to baseline (−15 and 0 min) in response to both saline and CNO treatment (F8,51 = 6.55; p < 0.001). In DBH-Cre-positive mice that received the stimulatory DREADD virus, there was a significant interaction between time and treatment (F8,51 = 7.058; p < 0.001). When treated with saline, corticosterone concentrations did not differ from baseline samples at any time after injection (Fig. 4B). In contrast, corticosterone concentrations were significantly elevated relative to baseline by 30 min after CNO injection and remained elevated for all subsequent samples.
Corticosterone concentrations during chemogenetic activation of NTSDBH cells in OVX female DBH-Cre positive with mCherry control virus mice (A) and DBH-Cre positive with DREADD virus mice (B). Group means ± SEM of n = 4 mice per group. Arrows indicate time of i.p. injection of saline (white data points) or CNO (black data points). *p < 0.05 versus baseline (A). Data points with different letters are significantly different (B).
Experiment 2C: KNDy cell activation
Figure 5 shows representative IHC for GFP (marker of kisspeptin cells, top panels), c-Fos (second row), and merged images (bottom two rows) in the mARC and quantification of neuronal activation in mice treated with saline (A, C, E, G) or CNO (B, D, F, H). The total number of GFP cells did not differ between mice that received saline and CNO in any region of the ARC (data not shown). The percentage of GFP cells that contained c-Fos was significantly reduced in CNO-treated animals compared with saline in the rARC (T = 9.94; p < 0.0001), mARC (T = 0.005; p = 0.0028), and cARC (T = 5.95; p = 0.004; Fig. 5I).
Activation of arcuate kisspeptin cells following chemogenetic activation of NTSDBH cells in OVX female mice. Representative image of IHC for hrGFP (marks kisspeptin cells; A, B), c-Fos (C, D), and merged images (E, F; higher magnification insets: G, H) from a mouse treated with saline (A, C, E, G) or CNO (B, D, F, H). Arrows denote GFP cells that contain c-Fos. Scale bar (all panels) = 50 µm. I, Percent of kisspeptin cells that contain c-Fos throughout the ARC. Group means ± SEM of n = 4 (saline) and 6 (CNO) mice per group. *p < 0.05. 3V, third ventricle.
Experiment 3: are GnRH cells directly inhibited by NTSDBH cell activation?
LH concentrations before and after kisspeptin injection in mice pretreated with saline or CNO are shown in Figure 6 (experimental design represented in Fig. 6A). Similar to Experiment 2A, LH concentrations were significantly lower following CNO compared with saline across time (prekisspeptin: saline vs CNO; F1,21 = 18.72; p = 0.003; Fig. 6B). Kisspeptin induced a significant increase in LH concentrations in mice, regardless of saline or CNO pretreatment (F1,21 = 27.36; p < 0.001). The response to kisspeptin was not different across treatments as indicated by a lack of significant interaction between treatment (saline vs CNO) and time (pre- vs postkisspeptin; F1,21 = 2.3; p = 0.145). Furthermore, the fold increase was similar, 5.8 ± 3.0 versus 6.3 ± 1.6, saline versus CNO, p > 0.05.
Kisspeptin challenge during chemogenetic activation of NTSDBH cells. A, Experimental timeline, details in text. B, LH concentrations 0 min before (white bars) and 10 min after i.p. kisspeptin injection (gray bars) in OVX female mice pretreated with saline (left) or CNO (right) 90 min prior. Group means ± SEM of n = 8 mice. No significant time × treatment interaction, *p < 0.05 pre- versus postkisspeptin.
Experiment 4: is suppression of LH secretion dependent upon elevated corticosterone secretion?
LH and corticosterone concentrations in DBH-Cre-positive mice with the stimulatory DREADD virus sampled before and after adrenalectomy are shown in Figure 7. Before adrenalectomy, placement in a restraint device induced a significant increase in corticosterone concentrations (Fig. 7B). In contrast, after adrenalectomy, corticosterone concentrations in response to restraint did not differ from the prestress value, confirming complete removal of the adrenal gland, statistically revealed by a significant interaction between adrenal status (before vs after adrenalectomy) and response to restraint (F2,25 = 29.93; p < 0.0001). Representative LH pulse profiles from two mice sampled before and after adrenalectomy and injected with CNO are shown in Figure 7C. Regardless of adrenal status, both mean LH concentration (Fig. 7D; F1,15 = 75.94; p < 0.001) and number of pulses per hour (Fig. 7E; F1,15 = 102.40; p < 0.001) were significantly reduced following CNO injection. There was no effect of adrenalectomy (mean LH: F1,15 = 0.04, p = 0.85; number of pulses: F1,5 = 0.58, p = 0.46), nor was there a significant interaction between adrenal status and response to CNO (mean LH: F1,5 = 0.04, p = 0.84; number of pulses: F1,15 = 1.43, p = 0.25).
LH concentrations during chemogenetic activation of NTSDBH cells in ADX mice. A, Experimental timeline, details in text. B, Corticosterone concentrations in response to a restraint stress test, conducted before ADX (white data points) and after ADX (gray data points). Data points with different letters are significantly different. C, Representative LH pulse profiles before ADX (left panel) and after ADX (right panel); CNO administered i.p. at time 0. Black data points represent pulses. Group means ± SEM of n = 6 for: (D) Mean LH, (E) number of pulses per hour. No significant time × treatment interaction, *p < 0.05 pre versus post-CNO injection.
Discussion
The data generated in the present study provide evidence that NTSDBH cells are a mediator of stress-like neuroendocrine responses. First, we found that a subpopulation of NTSDBH cells is activated (expressed c-Fos) during acute hypoglycemia. Next, we showed that chemogenetic activation of NTSDBH cells suppressed pulsatile LH secretion and induced a stress-like rise in corticosterone concentrations. Activation of NTSDBH cells caused a reduction in the percentage of KNDy cells that were activated, but did not alter responses to Kiss1 receptor activation, suggesting impairment upstream of GnRH cell activity. Finally, we determined that NTSDBH cell activation suppressed LH pulses in ADX mice, indicating that the observed suppression of LH pulses is not mediated by corticosterone.
The NTS is a heterogenous brain region with cells that produce numerous receptors, neurotransmitters, and neuropeptides. Accordingly, a variety of physiological processes have been attributed to these cells including regulation of reproduction, stress responses, and the control of feeding (Holt, 2022). A marker of neuronal activation, c-Fos, is increased in the NTS following a variety of stress types including metabolic stress in rats (Paranjape and Briski, 2005). Here, we detected an increase in the total number of c-Fos cells as well as the percentage of DBH cells that contained c-Fos in mice following acute metabolic stress. Of note, we observed that increased c-Fos was restricted to a region of the caudal NTS (−7.08 to −7.32 mm from the bregma) and that only a subpopulation (<50%) of DBH neurons in these areas was activated. This may reflect the relatively imprecise nature of activity assessment by c-Fos or that only a small portion of NTSDBH neurons are sensitive to acute hypoglycemia. Whether the same DBH cells that were activated in response to IIH here are also activated during other types of stress or visceral sensation, or if discrete subpopulations respond to separate stress types, remains an outstanding question.
Our chemogenetic data demonstrate that NTSDBH cells are sufficient to suppress LH secretion in mice. We observed a suppression of LH pulse frequency, but not LH pulse amplitude, suggesting an impairment of KNDy cells since these cells are responsible for organizing pulsatile secretion patterns (Herbison, 2018; Moore et al., 2018). Further support for inhibition of KNDy cells comes from the observation that a reduced percentage of Kiss1 cells in the ARC expressed c-Fos following activation of NTSDBH cells, indicating suppression of endogenous KNDy cell activity. A similar suppression of KNDy cell activation has been observed following psychosocial (Yang et al., 2017), immune (Makowski et al., 2020), and metabolic stress (McCosh et al., 2019b). Although the precise mechanism for the inhibition of KNDy cells remains unknown, evidence in which monosynaptic tracing of afferents to KNDy cells labeled few cells in the brainstem (Moore et al., 2019) suggests that NTSDBH cells do not act directly on KNDy cells. Instead, NTSDBH cells likely signal via the PVN since norepinephrine (NE) injected into this region suppressed LH pulses in rats (Tsukamura et al., 1994). Furthermore, the inhibitory effects of NE in the PVN can be blocked with a nonselective CRH receptor antagonist, raising the possibility that neurons that produce ligands for either CRH receptor (i.e., CRH, urocortin 1, urocortin 2, or urocortin 3) project to and act on KNDy cells (Tsukamura et al., 1994; Squillacioti et al., 2019).
With evidence that KNDy cells are impaired during NTSDBH cell activation, we tested whether GnRH cells and gonadotrophs are themselves inhibited or lack stimulatory input from KNDy cells, using a kisspeptin challenge. We observed no significant difference in the effect of exogenous kisspeptin to stimulate LH secretion during NTSDBH cell activation versus controls and interpret these findings to indicate that signaling pathways downstream of Kiss1 receptor activation are largely intact. LH concentrations were significantly lower when mice were pretreated with CNO compared with saline; however, the fold change in response to kisspeptin was not different between pretreatment with CNO and saline. We believe that the analysis of the fold change is an appropriate manner of comparing these data in which the baseline (prekisspeptin) values are clearly not similar. We recognize that the total amount of LH released when animals were pretreated with CNO is significantly lower than following saline pretreatment and suspect this is due to a lack of kisspeptinergic input to GnRH cells, rather than a specific reduction in the response to kisspeptin. A potential caveat to this interpretation is the possibility that the high dose of kisspeptin maximally stimulated these pathways, masking subtle impairment of GnRH cell or gonadotroph function. Indeed, previous anatomical (Miller and Zhu, 1995; Turi et al., 2003; Todman et al., 2005; Campbell and Herbison, 2007; Vastagh et al., 2016) and electrophysiological (Han and Herbison, 2008) data raised the possibility that NE acts directly on GnRH neurons to suppress pulsatile secretion. In contrast, our data are consistent with the pharmacological data described above (Tsukamura et al., 1994), indicating that NE acts upstream of GnRH cells.
Elevated circulating glucocorticoids are a central hallmark of stress responses and have been hypothesized to mediate the stress-induced suppression of gonadotropin secretion. In mice, chronic corticosterone is sufficient to suppress LH pulses in estradiol-replaced OVX mice but not in OVX mice (Kreisman et al., 2019), demonstrating estradiol dependence for this effect in mice, similar to reports in sheep (Oakley et al., 2009a,b). In the present study, activation of NTSDBH cells induced a stress-like increment in corticosterone concentrations that occurred concomitantly with the decrease in LH pulses. To determine if elevated corticosterone is necessary for the suppression of LH pulses, we performed adrenalectomy and stimulated NTSDBH cells. The suppression of pulsatile LH secretion was similar to that documented prior to adrenalectomy, demonstrating that corticosterone is not necessary for suppressed LH secretion in response to NTSDBH cell activation. Based on our observations that corticosterone concentrations remained at baseline during restraint stress following adrenalectomy, we are confident that surgical removal of the source of glucocorticoids was complete. Collectively, our data support the notion that NTSDBH cells regulate both the HPG axis via inhibition of KNDy cells and the HPA axis, presumably by activating CRH neurons in the PVN, yet the resulting rise in corticosterone is not necessary for LH suppression.
Our data demonstrating DBH cells in the NTS are sufficient to alter LH, and corticosterone secretion extends previous work which identified a role for NE signaling but lacked anatomical precision regarding the source of NE-containing cells (I'Anson et al., 2003). Additional questions generated from this work remain. For example, future work is necessary to determine whether activation of DBH cells in the ventrolateral medulla or locus ceruleus is also sufficient to suppress LH pulses and augment corticosterone secretion, as well as to test the necessity of each of these NE-containing populations for these neuroendocrine responses during stress. Another question pertains to the signaling mediator conveying the effects of NTSDBH cell activation, a caveat related to the chemogenetic approach used here. Although cells that contain NE were targeted (using a DBH-Cre transgenic mouse line), it is not clear which signaling molecule(s) released induce the observed effects. NE is the most likely mediator of these effects given the pharmacological data described above; however, NTSDBH cells are also glutamatergic, and subpopulations of these cells express a variety of neuropeptides (Holt, 2022).
In the course of studying the endocrine effects of NTSDBH cell activation, we observed a robust sickness-like behavioral phenotype. The behavior was apparent ∼20 min after CNO injection and lasted for the duration of the blood sampling period (90 min after CNO injection). Although this behavioral response was not mentioned in previous report utilizing this DBH-Cre mouse line with the same stimulatory DREADD construct delivered into the NTS (Roman et al., 2016), there are several possible reasons for this discrepancy. First, the previous study used a lower CNO dose (0.8 mg/kg) than in the present study (1 mg/kg). Second, the minor stress associated with tail-tip blood sampling may have exacerbated otherwise minor behavioral responses. Finally, the blood sampling procedure enabled constant and close monitoring of animals including handling of control animals that would otherwise be inactive during the lights-on period, which represent experimental differences compared with the previous report.
In conclusion, the data presented here support the hypothesis that NTSDBH cells are sufficient to induce stress-like neuroendocrine effects in female mice. Moreover, these data support the hypothesis that the suppression of LH pulses we observed was caused by the inhibition of KNDy cells, and not downstream GnRH or gonadotroph cell inhibition, but not mediated by elevated corticosterone secretion. Future studies are needed to fully elucidate the connections between NTSDBH cells and KNDy cells for the suppression of gonadotropin secretion during stress as well as understand the generalizability of this pathway for reproductive suppression across stress types.
Footnotes
We thank Dr. Alexander Kauffman (UC San Diego) and Dr. Shannon Stevens (Albany Medical College) for the critical equipment (stereotaxic frame) and training in the delivery of virus. We acknowledge Dr. Janet Roser (Department of Animal Science, University of California, Davis) for the final characterization, preparation, and distribution of 518B7 anti-bovine luteinizing hormone antibody and Quidel for the initial development. We also acknowledge the contributions of Dr. Al Parlow and the National Hormone and Peptide Program for the development of AFP240580Rb. Finally, we thank the Nikon Imaging Center at UC San Diego for the support on microscopy experiments. This work was supported by NIH Grants R01 HD086100, R01 HD103725, R21 HD105103, and P50 HD012303 and the UCSD Health Sciences Senate. R.B.M. was supported by NIH Grants K99/R00 HD104994, F32 HD096811, and T32 HD007203.
The authors declare no competing financial interests.
- Correspondence should be addressed to Kellie M. Breen Church at kbchurch{at}ucsd.edu.
