All animal handling procedures followed the guiding principles approved by Institutional Animal Care and Use Committee of Jeonbuk National University, CBNU 2016-64, and CBNU 2020-0122. GnRH-green fluorescent protein-tagged (GnRH-GFP; Strain: C57BL/6) mice (31), and ICR mice were housed under stable room temperature (23-26 °C) and an automatic 12-h light/12-h dark photocycle (lights on at 07:00 h) with free access to food and water.

PCOS mouse model was created in two different strains of mice. ICR female mice were used to assess the changes in the estrous cycle, body weight, and pregnancy rate, whereas, GnRH-GFP tagged mice were used for electrophysiological recording. Letrozole 30 µg/ml stock solution was dissolved in 4% aqueous solution of carboxymethylcellulose (CMC) and stored at 4°C. As shown in Figure 1A , ICR female mice of postnatal day (PND) 21 were divided into two groups. For the control group, mice were orally fed with CMC alone by gavage and for the drug-treated group, 1 µg/g body weight of letrozole was orally fed by gavage once a day for 21 days to induce polycystic ovary (8, 16). Body weight was measured every week during the feeding period. For electrophysiological recording, five batch of GnRH-GFP-tagged female mice of PND 21 were orally fed by gavage with 1 µg/g of letrozole dissolved in CMC (n = 6 in each batch) or CMC alone (n = 5 in each batch) for 21 days as shown in Figure 2A . Then, one mouse from either control or letrozole-fed group was alternatively decapitated on daily basis between 11:00 AM to 12:00 PM UTC + 9:00 (Universal Time Coordinate) for the electrophysiological assessment.

The estrous cycle of the mice was assessed for two weeks after completion of letrozole feeding. Estrous phases were determined by monitoring the morphology of the vaginal smear under light microscopy as described by previous study (32). Briefly, vagina was gently flushed 2-3 times with Phosphate Buffered Saline (PBS), and the final flush was collected in the pipette tip and then smeared on the clean glass slides. The dried samples were stained using (0.2%) methylene blue and examined under a light microscope. The phases of the estrus cycle were then identified by cytological features. The predominant presence of nucleated epithelial in individuals or clusters confirmed the proestrus phase. In addition, the estrous phase showed the predominance of cornified epithelial cells, whereas the metaestrous phase showed an increase in the number of leukocytes with epithelial cells. Similarly, the diestrous phase had leukocytes as predominant cells. Metaestrous and diestrous stages showed identical characteristics; thus, the data were pooled as simply diestrous.

For fertility assessment, female mice (PND 60) from the control group (n = 10) and letrozole-fed group (n = 5) were paired with fertile males (a total of 15 breeding pairs), and copulatory plugs were checked daily. Within 4 to 5 days, all mice had a copulatory plug, which was marked as gestational day 1, and all the males were removed from the cages.

Coronal brain slices were prepared using the same procedure as described in a previous study (33). GnRH-GFP tagged female mice from control and letrozole-fed groups were decapitated between 11:00 AM to 12:00 PM UTC + 9:00 (Universal Time Coordinate). Their brains were removed quickly and placed in an ice-cold artificial cerebrospinal fluid (ACSF) with the following composition (in mM): 126 NaCl, 2.5 KCl, 2.4 CaCl₂, 1.2 MgCl₂, 11 D-glucose, 1.4 NaH₂PO₄, 25 NaHCO₃ and pH maintained to 7.3 - 7.4 by bubbling with 95% O₂ and 5% CO₂. Coronal slices (180 µm thickness) containing the preoptic hypothalamic area were cut using a vibratome (VT1200S; Leica biosystem, Wetzlar, Germany) in the ice-cold ACSF. Slices were then allowed to recover in oxygenated ACSF before being transferred to the recording chamber. The images of coronal slices were viewed under an inverted microscope (BX51WI; Olympus, Tokyo, Japan) and displayed on a video monitor. GnRH neurons were identified using 40X objective lens by brief fluorescence illumination and patched under Nomarski differential interference contrast optics. Patch pipettes were pulled from thin-wall borosilicate glass-capillary tubing (PG52151-4; WPI, Sarasota Bay, FL, USA) on a Flaming/Brown puller (P-97; Sutter Instruments Co., Novato, CA, USA). The patch pipette was back-filled with an internal solution containing (in mM): 140 KCl, 1 CaCl₂, 1 MgCl₂, 10 HEPES, 4 MgATP, and 10 EGTA (pH 7.3 with KOH). The tip resistance of the patch pipette filled with an internal solution was approximately 4 to 6 MΩ. The electrode potential was nullified before the gigaseal formation. Then the neurons were voltage clamped at -60 mV, and a gentle suction was applied to rupture the cell’s membrane to make a whole-cell configuration. The whole-cell patch-clamp recordings were performed under voltage clamp mode using the Axopatch 200B amplifier (Molecular Devices, San Jose, CA, USA). The changes in membrane current were sampled online using a Digidata 1322A interface (Molecular Devices) connected to a desktop PC.

For perforated patch-clamp recordings, gramicidin (Sigma-Aldrich, St. Louis, MO, USA) was first dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich) to a concentration of 2.5–5 mg/ml. Then diluted to a final concentration of 2.5–5 μg/ml in the pipette solution and sonicated for 20 min before the use. Access resistance was monitored in the initial experiments, and experiments began when the resistance stabilized at 50–90 mΩ. Typically, after 15–30 min of gigaseal formation, the cell’s resting membrane potential (RMP) reaches a stable level below −45 mV. Any spontaneous membrane rupture was evident by sudden increases in membrane potentials above 0 mV. All the recordings were made at room temperature.

All the chemical reagents for ACSF and pipette solutions, including letrozole (L6545), γ-aminobutyric acid (GABA) (A2129), 4,5,6,7-tetrahydroisoxazolo[5,4-c]pyridine-3-ol (THIP) hydrochloride (T101), tetrahydrodeoxycorticosterone (THDOC) (P2016), bicuculline (14340), kainic acid (K0250), dehydroepiandrosterone (DHEA) (252805) and metformin (PHR1084) were purchased from Sigma-Aldrich (St. Louis, MO, USA) whereas, muscimol (0289) was obtained from Tocris Bioscience. In addition, 2,4-Thiazolidinedione (SC-216281) was purchased from Santa Cruz Biotechnology (Finnell Street, Dallas). Kisspeptin-10 (048-56) was obtained from Phoenix Pharmaceuticals, INC (Burlingame, CA, USA). All the chemicals were dissolved in distilled water except THDOC, Thiazolidinedione, and DHEA, which were dissolved in DMSO. Stocks were diluted (usually 1,000-fold) in ACSF to desired final concentrations before bath application.

All statistical values were expressed as mean ± standard error of the mean. The Shapiro-Wilk test was used to determine whether the data had a normal distribution. The unpaired t-test was used to compare the data from a normal distribution, while the Mann-Whitney test was used to compare data from a non-normal distribution. In addition, the percentage of the estrous cycle and pregnancy outcomes were analyzed using the chi-square test. Statistical analysis was performed in Origin software (OriginLab Corp, Northampton, MA, USA), and the tests performed are indicated in the figure legends. Statistical significance was defined as a p-value < 0.05. Acquisition and subsequent analysis of the acquired data were performed using Clampex software (Axon Instruments). The traces were plotted using Origin 8 software (OriginLab Corp., Northampton, MA, USA).

In this study, all control females had 4 to 5 days of the normal estrous cycle, but letrozole-fed mice showed irregularity in the estrous cycle ( Figures 1B, C ). The pie chart shows the average percentage of three phases of the estrous cycle in control and letrozole-fed mice in two weeks period ( Figure 1D ). In both groups, the proestrous phase remained almost constant, but the letrozole-fed mice had a shorter estrous phase (control: 28.6%; letrozole: 12.8%) and prolonged diestrous phase (control: 44.3%; letrozole: 60.0%) than control. Next, the average days spent in three phases of an estrous cycle that occurred in two weeks were compared. Control and letrozole-fed mice had nearly similar proestrus duration (control: 1.28 ± 0.12 days, letrozole: 1.46 ± 0.14 days, p > 0.05, unpaired t-test) whereas the estrous phase duration was reduced (control: 1.28 ± 0.12 days, letrozole: 0.75 ± 0.17 days, *p < 0.05, unpaired t-test), and the diestrous phase duration was prolonged (control: 2.06 ± 0.22 days, letrozole: 3.23 ± 0.45 days; *p < 0.05, unpaired t-test) in the letrozole-fed mice ( Figure 1E ). Additionally, the body weight of letrozole-fed mice was significantly increased in the third week of letrozole treatment (control: 24.6 ± 0.29 gm, n = 15; letrozole: 26.3 ± 0.27 gm, n = 15; *p < 0.05, unpaired t-test, Figure 1F ).

PCOS is considered to have a significant impact on pregnancy outcomes. Compared with control females, letrozole-fed mice had a lower pregnancy rate. On gestational day 21, 9 out of 10 (90%) control females delivered pups, while 2 out of 5 (40%) letrozole-fed females delivered pups (p < 0.05, chi-square test, Figure 1G ). Furthermore, letrozole-fed females delivered less number of pups per litter (n = 7.0 ± 2.0, N = 2) compared to controls (n = 9.4 ± 0.76, N = 9) but this difference was statistically insignificant (figure not shown, ‘n’ represents pups number and ‘N’ represents litter number).

GABA and glutamate receptor-mediated responses were recorded using the whole-cell patch-clamp technique from GnRH neurons distributed in the hypothalamic preoptic regions of GnRH-GFP- mice. To see if there were any variations in the response produced by excitatory neurotransmitters in letrozole-fed mice, we compared the responses elicited by GABA (100 µM), muscimol (3 µM), and kainate (10 µM) in control and letrozole-fed groups. Figure 2A , shows the experimental paradigm for the electrophysiological assessment. The traces of GABA-induced inward current from control and letrozole-fed mice are shown in Figure 2B . With respect to control, mean inward current induced by GABA was remarkably decreased in letrozole-fed mice (control: -970.0 ± 164.0 pA, n = 15; letrozole: -251.0 ± 46.0 pA, n = 20; *p < 0.05, unpaired t-test, Figure 2C ). Furthermore, GABA_A receptor agonist muscimol-mediated response was dramatically reduced in the letrozole-fed mice ( Figure 2D ). The mean inward current induced by muscimol was significantly lower in letrozole-fed mice compared to the control (control: -453. ± 56.0 pA, n = 12; letrozole: -276.0 ± 28.19 pA, n = 18; *p < 0.05, unpaired t-test, Figure 2E ). Furthermore, the glutamate agonist kainate had a negligible effect on the GnRH neurons of letrozole-fed mice ( Figure 2F ). The mean inward current induced by kainate was significantly decreased in letrozole-fed mice compared to control (control: -18.3 ± 2.94 pA, n = 13; letrozole: -8.22 ± 1.73 pA, n = 19, ^#p < 0.05, Mann-Whitney test, Figure 2G ).

Next, we compared the response induced by bath application of endogenous neurosteroids on GnRH neurons among control and letrozole-fed mice. In GnRH neurons, neurosteroids activates the extrasynaptic GABA_A receptors and mediates tonic form of GABAergic signaling (34). Under whole-cell voltage-clamp mode, neurosteroids THDOC and THIP mediated response on GnRH neurons were recorded from control and letrozole-fed mice. THDOC, a potent allosteric modulator of GABA_A receptor, elicited a larger response in GnRH neurons from the control (-25.2 ± 1.90 pA, n = 7) compared to the letrozole-fed mice (-13.8 ± 1.79 pA, n = 6, *p < 0.05, unpaired t-test, Figures 3A, C ). Next, the effect of thiazolidinedione on THDOC-induced response in GnRH neurons of the control and letrozole-fed mice was assessed ( Figure 3B ). Thiazolidinedione is a medicine used for treating type 2 diabetes and PCOS (35). Pretreatment of the slice with thiazolidinedione for 10 to 15 minutes resulted in slight decrease of THDOC-mediated response in GnRH neurons from the control group (THDOC: -25.2 ± 1.90 pA, n = 7, THDOC + Thiazolidinedione: -19.9 ± 3.95 pA, n = 5; p > 0.05, unpaired t-test, Figure 3C ). However, in a letrozole-fed mice, thiazolidinedione pretreatment did not change the THDOC-mediated response in GnRH neurons (THDOC: -13.8 ± 1.79 pA, n = 6, THDOC + thiazolidinedione: -19.9 ± 3.33 pA, n = 5; p > 0.05, unpaired t-test, Figure 3C ).

Similarly, another neuroactive steroid, THIP, which activates extrasynaptic GABA_A receptors containing δ-subunit, exerts a similar response as THDOC in control and letrozole-fed mice. Bath application of THIP-induced larger response in the GnRH neurons of control (-40.4 ± 6.74 pA, n = 5) compared to the letrozole-fed mice (-16.4 ± 1.59 pA, n = 6, *p < 0.05, unpaired t-test, Figure 4A, C ). Next, the effect of DHEA, a PCOS inducer (36), on THIP-induced response in GnRH neurons of the control and letrozole-fed mice was assessed ( Figure 4B ). Pre-treatment of the slice with DHEA for 10 to 15 minutes resulted in the decrease of THIP- mediated response in GnRH neurons of control mice (THIP: -40.4 ± 6.74 pA, n = 5, THIP + DHEA: -29.2 ± 3.44 pA, n = 5; p > 0.05, unpaired t-test, Figure 4C ). However, pretreatment with DHEA did not affect the THIP response in the letrozole-fed mice (THIP: -16.4 ± 1.59 pA, n = 6, THIP + DHEA: -18.4 ± 1.70 pA, n = 6; p > 0.05, unpaired t-test, Figure 4C ).

Next, to investigate whether letrozole affects the kisspeptin sensitivity on adult GnRH neurons, 30 nM kisspeptin was bath applied in perforated patch-clamp mode. Here, we compared the kisspeptin sensitivity on adult GnRH neurons from control and letrozole-fed mice. In control mice, all tested GnRH neurons (n = 17, 100%) responded to kisspeptin with either membrane depolarization or membrane depolarization accompanied by the continuous firing of action potential (AP). However, only 64.3% (n = 9/14) of GnRH neurons from letrozole-fed mice responded to kisspeptin with similar characteristics to control, while the remaining neurons did not respond to kisspeptin. A significant difference was found in the percentage of GnRH neurons showing excitation between the control and letrozole group (p < 0.05; chi-square test, Figure 5B ). Figure 5A represent the traces for kisspeptin-induced excitation of GnRH neurons from control and letrozole-fed mice. Although the percentage of GnRH neurons responding to kisspeptin was lower in letrozole-fed mice, the mean membrane depolarization induced by kisspeptin was not significantly different from control (control: 4.94 ± 0.74 mV, n = 17; letrozole: 3.57 ± 1.03 mV, n = 9; p > 0.05, unpaired t-test).

Next, kisspeptin evoked excitation on GnRH neurons from both the control and letrozole-fed mice was recorded in the presence of metformin and thiazolidinedione, both drugs are used in treatment of PCOS and type 2 diabetes mellitus (35). Metformin and thiazolidinedione were bath applied to the GnRH neurons for at least 10 to 15 min before the kisspeptin perfusion. In the presence of metformin and thiazolidinedione, all the GnRH neurons from control mice responded to kisspeptin with membrane depolarization accompanied by spontaneous AP firings, whereas, in letrozole-fed mice, the majority of the neurons displayed AP firing alone ( Figures 5C, E ). In the presence of metformin, kisspeptin evoked depolarization was significantly smaller in the letrozole-fed mice compared to control (control: 4.02 ± 0.5 mV, n = 5, letrozole: 1.43 ± 0.37 mV, n = 5; *p < 0.05, unpaired t-test, Figure 5D ). Similarly, in the presence of thiazolidinedione, kisspeptin evoked membrane depolarization was also significantly smaller in the letrozole-fed mice compared to the control (control: 4.02 ± 0.89 mV, n = 6, letrozole: 1.12 ± 0.26 mV, n = 6; *p < 0.05, unpaired t-test, Figure 5F ).