Development and prenatal exposure to androgens alter potassium currents in gonadotropin-releasing hormone (GnRH) neurons from female mice

Abstract

Pulsatile gonadotropin-releasing hormone (GnRH) release is critical for reproduction. Disruptions to GnRH secretion patterns may contribute to polycystic ovary syndrome (PCOS). Prenatally androgenized (PNA) female mice recapitulate many neuroendocrine abnormalities observed in PCOS patients. PNA and development induce changes in spontaneous GnRH neuron firing rate, response to synaptic input, and the afterhyperpolarization potential of the action potential. We hypothesized potassium currents are altered by PNA treatment and/or development. Whole-cell patch-clamp recordings were made of transient and residual potassium currents of GnRH neurons in brain slices from 3-week-old and adult control and PNA females. At 3-weeks of age, PNA treatment increased transient current density versus controls. Development and PNA altered voltage-dependent activation and inactivation of the transient current. In controls, transient current activation and inactivation were depolarized at 3-weeks of age versus in adulthood. In GnRH neurons from 3-week-old mice, transient current activation and inactivation were more depolarized in control than PNA mice. Development and PNA treatment interacted to shift the time-dependence of inactivation and recovery from inactivation. Notably, in cells from adult PNA females, recovery was prolonged compared to all other groups. Activation of the residual current occurred at more depolarized membrane potentials in 3-week-old than adult controls. PNA depolarized activation of the residual current in adults. These findings demonstrate the properties of GnRH neuron potassium currents change during typical development, potentially contributing to puberty, and further suggest PNA treatment may both alter some typical developmental changes and induce additional modifications, which together may underlie aspects of the PNA phenotype.

Introduction

Gonadotropin-releasing hormone (GnRH) neurons produce the final central output for the control of reproduction. GnRH release occurs in an episodic pattern, termed pulses^1,2. GnRH acts at the anterior pituitary where low-frequency GnRH pulses favor the secretion of follicle-stimulating hormone (FSH), and high-frequency pulses favor the secretion of luteinizing hormone (LH)^3,4. Disruptions to the secretion patterns of GnRH, LH and/or FSH result in altered reproductive function^5–8. Polycystic ovary syndrome (PCOS) is the leading cause of infertility in females⁹. Patients with hyperandrogenemic PCOS have persistent high-frequency LH pulse secretion and relative FSH deficiency^9–12, suggesting GnRH pulses are also persistently high frequency. Interestingly, PCOS-like symptoms have been observed in peripubertal girls, suggesting the neuroendocrine system is altered even before puberty is complete^13,14.

The etiology of PCOS remains largely unknown. In addition to possible genetic causes^15,16, it has been hypothesized that in utero exposure to high circulating levels of androgens from the mother might reprogram the reproductive neuroendocrine system. This postulate has been supported by animal models including non-human primates, rats, sheep, and mice^17–22. In these models, treatment of pregnant dams with androgens produces offspring that develop neuroendocrine phenotypes similar to those observed in females with hyperandrogenemic PCOS. Adult female prenatally androgenized (PNA) mice exhibit increased circulating testosterone levels, increased LH pulse frequency, and disrupted estrous cycles^20,22–24. PNA causes early vaginal opening, an external index of pubertal onset²⁵, suggesting PNA alters reproductive development. Use of mouse models makes it possible to measure neurophysiologic parameters. Spontaneous GnRH neuron firing rate changes with development in controls and is altered by PNA treatment in both prepubertal and adult mice. Specifically, in PNA females, spontaneous GnRH neuron firing rate is low at 3-weeks of age compared to controls but is increased in adulthood^20,23. The developmental and PNA-induced changes in spontaneous GnRH neuron activity could be the result of altered neurotransmission to these cells and/or changes in their intrinsic properties. PNA treatment increases the frequency of GABAergic neurotransmission to GnRH neurons in 3-week-old and adult females^17,24. Unlike most neurons, GnRH neurons maintain higher intracellular chloride concentration and activation of GABA_A receptors can thus excite these neurons^{26 27}. Despite receiving more frequent GABAergic transmission, however, GnRH neuron activity in 3-week-old PNA females is lower than controls²³, suggesting that GnRH neurons from PNA females at this age respond differently to GABAergic input. Supporting this postulate, local GABA application increased firing in only about 30% of GnRH neurons from 3-week-old PNA females compared to 100% of neurons from 3-week-old control females and control and PNA adults²⁴. The reduced response to GABA in 3-week-old PNA females is not attributable to differences from 3-week-old control in the baseline membrane potential or intracellular chloride concentration²⁴. Together these findings suggest changes intrinsic to GnRH neurons might regulate how they respond to synaptic input and thus contribute to the altered firing rate.

In this regard, GnRH neurons from control and PNA adults fire more action potentials than control and PNA females at 3-weeks of age in response to the same amount of current injection²⁸. Further, in adulthood the afterhyperpolarization potential of the action potential is larger and delayed relative to prepubertal mice²⁸. The ionic changes underlying these differences are unknown. Here, we focus on voltage-gated potassium currents, which play a major role in determining the response of neurons to the inputs they receive, in shaping firing rate and in the repolarization characteristics of neurons^29–32. GnRH neurons display two primary voltage-gated potassium currents, a fast transient current and a residual, delayed-rectifier-like current³³. We tested the hypotheses that their properties change with age and with PNA treatment.

Materials and Methods

All chemicals were acquired from Sigma-Aldrich (St. Louis, MO, USA) unless noted otherwise.

Animals.

GnRH-GFP (Tg(Gnrh1-EGFP)51 Sumo MGI:6158457) mice were bred in our colony³⁴. All mice were provided with ad libitum water and Harlan 2916 (non-breeders) or 2919 (breeders) chow. Mice were held on a 14:10 light/dark cycle with lights on at 3AM Eastern Standard Time. To generate female PNA mice, female GnRH-GFP transgenic mice on a C57Bl/6J background and a CD1 female were bred with a C57Bl/6J male. The CD1 dam assists in providing maternal care and nutrition. Females were monitored daily for a copulatory plug (day 1 of pregnancy). On days 16–18 of pregnancy, GnRH-GFP dams were injected subcutaneously with 225μg/day of dihydrotestosterone (DHT) in 50μl sesame oil for PNA treatment or sesame oil for controls. Litter sizes were adjusted to 4–15 pups by culling the CD1 pups, identifiable by size and coat color, to standardize nutrition. Adult female mice were studied on the morning of diestrus, determined by vaginal cytology and confirmed by uterine mass. All procedures were approved by the Institutional Animal Care and Use Committee of the University of Michigan.

Verification of PNA phenotype.

Electrophysiological studies were done on 3-week-old (18–21 d, before vaginal opening, VO) and adult (84–130 d) diestrous females. PNA-induced differences were confirmed for the prepubertal mice in littermates raised to adulthood, as the PNA phenotype is consistent among littermates. PNA status was measured by monitoring age at which VO occurred, estrous cyclicity via vaginal lavage for 14 consecutive days and measuring anogenital distance (AGD) between 70–73d of age (average of three successive daily measures).

Brain slice preparation.

All solutions were bubbled with 95% O₂/5% CO₂ for at least 15 min before use and throughout experimental recordings. Brain slices were prepared as described³⁵. Brains were removed and placed in ice-cold sucrose saline containing the following (in mM): 250 sucrose, 3.5 KCl, 26 NaHCO₃, 10 D-glucose, 1.25 Na₂HPO₄, 1.2 MgSO₄, and 3.8 MgCl₂ (350mOsm). Coronal slices (300μm) through the preoptic and anterior hypothalamic region were cut with a Leica VT1200S Microtome (Leica Biosystems, Buffalo Grove, IL). Slices were incubated for 30 min at room temperature (~21–23°C) in 50% sucrose saline and 50% artificial cerebrospinal fluid (ACSF, containing (in mM): 135 NaCl, 3.5 KCl, 26 NaHCO₃, 10 D-glucose, 1.25 Na₂HPO₄, 1.2 MgSO₄, 2.5 CaCl₂, 315 mOsm, pH 7.4). Slices were then transferred to 100% ACSF at room temperature for 0.5–5 h before recording. No changes in properties were associated with time from slice preparation to recording. A minimum of five mice from at least four litters were studied per group; up to three recordings were used per mouse.

Recording solutions and data acquisition.

Brain slices were transferred to a recording chamber continuously perfused with carboxygenated ACSF (3mL/min) maintained at 30–31°C with an inline-heating unit (Warner Instruments Model SH-27B). Potassium currents were pharmacologically isolated by blocking fast sodium (2μM tetrodotoxin; Tocris Bioscience or Abcam) and calcium (300μM NiCl₂) channels as well as ionotropic receptors for GABA (100μM picrotoxin) and glutamate (20μM D-APV and 10μM CNQX; Tocris Bioscience). The pipet solution contained (in mM): 120 K gluconate, 20 KCl, 10 HEPES, 5 EGTA, 0.1 CaCl₂, 4 MgATP, and 0.4 NaGTP (305 mOsm, 7.2 pH with NaOH). This solution was based on the native intracellular chloride concentrations in GnRH neurons that was determined using gramicidin perforated-patch recordings²⁶. A 14.5mV liquid junction potential was negated online before each recording. GFP-positive GnRH neurons were visualized with infrared differential interference contrast and fluorescence microscopy on an Olympus BX51WI microscope. Recordings were made using an EPC-10 patch-clamp amplifier and a computer running PatchMaster software (HEKA Elektronik). All current measurements were acquired at 20kHz and filtered at 10kHz. Recording quality and passive properties were monitored throughout experiments from the averaged membrane current response to 16 hyperpolarizing voltage steps from −70mV (5mV, 20ms). Data were analyzed using WaveMetrics IgorPro (Sutter Instruments). Only recordings with an input resistance of >500MΩ, stable compensated series resistance of <15MΩ and a stable capacitance (8pF-20pF) were used for analysis.

Activation/inactivation of voltage-gated potassium currents.

GnRH neurons display two major voltage-gated potassium currents, a fast transient (I_A-like) current and a slow residual (I_K-like) current. The distinct voltage dependence of inactivation of the transient and residual currents was used to isolate them. GnRH neurons were given a prepulse step to −100mV or −40mV for 500ms. The −40mV or −100mV prepulse was followed by a family of voltage steps from −100mV to +50mV (500ms) in 10mV increments, and then a final step to −10mV for 50ms. Pilot studies showed that a 500ms prepulse at −40mV results in full inactivation of the transient component, leaving the residual current mostly unchanged, and the prepulse at −100mV for 500ms completely removes I_A inactivation, yielding total current. The transient current was mathematically isolated by subtracting the current with the −40mV prepulse from that obtained with the −100mV prepulse at each voltage step. Activation of the transient current was estimated using the peak current during the final test pulse to −10mV that followed the family of voltage steps. Leak current was calculated from an average of 4 sweeps at 1/8 amplitude (modified P/−8) from a baseline potential of −70mV³⁶ and subtracted online.

Time dependence of inactivation and recovery following inactivation for transient current.

To measure the time dependence of inactivation, a 500ms prepulse at −100mV was used to remove inactivation. Then, membrane potential was stepped to −40mV for 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, and 1024ms, followed by a test pulse at −10mV to measure the peak current. The non-inactivating component at −10mV after the 1024ms inactivation step was subtracted from all of the other traces to isolate the transient current. To measure the time dependence of recovery from inactivation of the transient current, membrane potential was held at −40mV (500ms) to fully inactivate the fast transient component. Then, membrane potential was stepped to −100mV for 1, 2, 4, 8, 16, 32, 64, 129, 512, and 1024ms, followed by a test pulse at −10mV to measure the peak current. The non-inactivating component at 1ms recovery was subtracted from each of the other traces to isolate the transient component.

Data analysis.

Current density was calculated by dividing the maximum current by the capacitance for each recording. The driving force was calculated using the Goldman-Hodgkin-Katz (GHK) equation^37,38:

$$GHK\left(V\right)=V\left(\frac{zF}{RT}\right)\left(e^{\frac{zF}{RT}\left(V-E_K\right)}-1\right)/\left(e^{\frac{zF}{RT}V}–1\right)$$

where V is the step potential and E_K was calculated to be −92.6mV after activity corrections³⁹. Potassium conductance was estimated by dividing peak current by driving force. Conductance values were used to plot the voltage-dependent activation of transient and sustained potassium current components.

These data were plotted as a function of voltage and fit with the Boltzmann equation:

$$I\left(V\right)=\frac{1}{1+e^{-\frac{kF}{RT}\left(V-V0.5\right)}}$$

where V is the command membrane potential of the step, V_0.5 is the potential at which half the current was activated or inactivated, and k is the “slope factor”, F is the Faraday constant, R is the gas constant, and T is the absolute temperature. For inactivation, the driving force is the same for all cells because it is measured at the same membrane potential (test pulse 0mV); because of this, there is no need to convert to conductance for comparison. The Boltzmann equation was thus fit to the normalized peak current as a function of the inactivating prepulse (−100mV to −30mV). The Boltzmann fit was used to estimate the V_0.5 activation, V_0.5 inactivation and the slope factors.

To characterize the time dependence of inactivation and recovery, the relevant current was normalized to the maximum peak current for each cell and plotted as a function of duration of the prepulse. This was fit to a double exponential equation to account for a fast and slow components:

$$Y=Y_{\mathit{\text{end}}}+\left(Y_{\mathit{\text{start}}}-Y_{\mathit{\text{end}}}\right)*\left(F_{\mathit{\text{fast}}}*e^{\left(-K_{\mathit{\text{fast}}}*{\tau }_{\mathit{\text{fast}}}\right)}\right)+\left(1-F_{\mathit{\text{fast}}}\right)*e^{\left(-K_{\mathit{\text{slow}}}*{\tau }_{\mathit{\text{slow}}}\right)}$$

In this equation, Y_start is the Y value when X (time; ms) is zero; this value is held constant either at 1 for time-dependence of inactivation or 0 for the time-dependence of recovery from inactivation. Y_end is the y-value when X is 1024ms; this is held constant at either 0 for the time-dependence of inactivation or 1 for the time-dependence of recovery from inactivation. F_fast, the percent of the fast component, was held constant at 90% for both time-dependence fits. K_fast and K_slow are the rate constants, expressed in the reciprocal of the X-axis time units. Values for the fast (τ_fast) and slow (τ_slow) time constants are reported.

Statistics.

Data were tested for normal distribution with the Shapiro-Wilk test. Data are reported as the mean ± SEM or median ± IQR, as appropriate, with the individual values shown when practical. Statistical tests were chosen based on experimental design and data distribution and are specified in the results. Statistical comparisons and all graphs were made in Prism 10.1.1 (GraphPad Software), except for the statistical analysis of the proportion of time spent in each estrous cycle phase, for which RStudio was used^40–42 as Prism is unable to run the appropriate chi-squared tests^43,44. Analysis in R (R version 4.3.1 “Beagle Scouts”) using RStudio (RStudio Team, 2023) and a combination of open-sourced packages including, rstatix⁴³ and flextable⁴⁴, for which the “chisq_test”, “chisq_descriptives”, “row_wise_fishers_test”, and the “p.adjust.method” function were used to run the chi-squared tests. Significance was set at p<0.05.

Results

Verification of prenatal androgenization phenotype.

PNA-induced differences were confirmed in this study. Consistent with previous observations^{17,20,23,28,45} vaginal opening (VO) occurred at a younger age (Mann Whitney U-test) and a lower body mass (Mann Whitney U-test) in PNA (n=24) than control (n=28) females (Figure 1A, B; Table 1). Anogenital distance (AGD) was increased in adult PNA females (two-tailed, unpaired Student’s t-test, Figure 1C; Table 1) and estrous cycles were disrupted. Specifically, PNA females spent more time in diestrus and less time in proestrus and estrus than controls (χ², Figures 1D, E; Table 1).

Figure 1.

Confirmation of PNA phenotype. A-C. Individual values and median + the interquartile range (IQR) for age at vaginal opening (VO; A) and body mass at VO (B). Individual values and mean ± SEM for anogenital distance in adulthood (AGD, C). D, Representative estrous cycles over 14d. P, proestrus; D, diestrus; E, estrus (top, vehicle; bottom, PNA). E, Individual values and mean ± SEM for days in each cycle over 14d. Statistical parameters are in Table 1.

Table 1.

Statistical parameters characterizing the PNA phenotype.

Property	VEH median and interquartile range (IQR)	PNA median and interquartile range (IQR)	Mann-Whitney U test	Two-tailed p-value
age at vaginal opening (d)	31.0 (IQR 29.25–32.5)	24.0 (IQR 21.25–27.00)	U=43	p<0.0001
body mass (g) at vaginal opening	14.7 (IQR 14.13–15.4)	10.15 (IQR 9.38–13.08)	U=71.50	p<0.0001

Property	Mean ± SEM for AGD
AGD (mm)	VEH	PNA
	4.996 ± 0.07	6.195 ± 0.08
	Unpaired, two-tailed Student’s t test	mean difference (VEH-PNA)	Effect size r 2
	t(10.98), df=48; p<0.0001	Diff[Cl,0.979,1.418] 1.198 ± 0.11	r2=0.72

	Mean ± SEM for days spent across estrous cycle state
Cycle Stage	VEH	PNA
estrus	6.00 ± 0.381	4.583 ± 0.645
diestrus	5.833 ± 0.333	9.292 ± 0.636
proestrus	2.125 ± 0.202	0.125 ± 0.069

Property	χ2 test
Estrous cycle stage distribution	χ2= 66.194, n=671, df=2, p<0.0001
	Estrus	Diestrus	Proestrus
Std. residual	2.736304	−6.388093	6.823246
Fisher’s exact test; Bonferroni adjusted	p=0.02	p<0.0001	p<0.0001

Bold indicates p<0.05.

Recording quality parameters and passive properties of GnRH neurons.

Passive properties and series resistance were used to assess the quality of electrophysiological recordings (Figure 2). There were no differences among these parameters among any of the recording protocols, and all recordings were thus combined for quality assessment (two-way ANOVA; Table 2). There were no differences in passive properties or series resistance among groups.

Figure 2.

Recording quality and passive property parameters. A-D, Individual values and mean ± SEM for input resistance (A), compensated series resistance (B), capacitance (C), and holding current (D). Numbers and statistical parameters are in Table 2.

Table 2.

Descriptive statistics and statistical parameters from two-way ANOVA for recording quality parameters and passive properties.

Protocol	VEH 3-weeks	PNA 3-weeks	VEH adults	PNA adults
Voltage-dependence of activation & inactivation (transient & residual currents)	n=14 cells from 7 mice from 5 litters	n=16 cells from 10 mice from 7 litters	n=14 cells from 7 mice from 6 litters	n=14 cells from 10 mice from 10 litters
time dependence of inactivation (transient current)	n=13 from 8 mice from 4 litters	n=13 cells from 8 mice from 7 litters	n=13 cells from 7 mice from 6 litters	n=15 cells from 10 mice from 10 litters
Time dependence of recovery following inactivation (transient current)	n=13 cells from 8 mice from 4 litters	n=13 cells from 8 mice from 5 litters	n=13 cells from 8 mice from 7 litters	n=14 cells from 9 mice from 9 litters

Descriptive Statistics (mean ± SEM)
Property	VEH 3-weeks	PNA 3-weeks	VEH adults	PNA adults
series resistance (MΩ)	10.343 ± 0.316	9.449 ± 0.319	10.376 ± 0.278	10.141 ± 0.420
input resistance (MΩ)	877.413 ± 89.167	830.047 ± 62.029	841.480 ± 89.866	821.913 ± 67.695
capacitance (pF)	13.529 ± 0.443	12.717 ± 0.701	13.813 ± 0.740	14.691 ± 0.591
holding current (pA)	−43.060 ± 4.929	−43.564 ± 5.130	−32.774 ± 5.355	−45.236 ± 5.546

Two-way ANOVA
Property	Age	Treatment	Interaction
series resistance (MΩ)	Diff, −0.3643 [Cl, −1.541, 0.2098] F(1,62)=1.176, p=0.2824	Diff, 0.5641 [Cl, −0.1040, 1.232] F(1,62)=2.848; p=0.0965	Diff, 0.6590 [Cl, −0.6774, 1.995] F(1,62)=0.9716, p=0.3281
input resistance (MΩ)	Diff, 22.03 [Cl, −136.2, 180.3] F(1,62)=0.08615; p=0.7749	Diff, 34.47 [Cl, −124.8, 191.7] F(1,62)=0.1788; p=0.6739	Diff, 27.80 [Cl, −288.7, 344.3] F(1,62)=0.03083; p=0.8612
capacitance (pF)	Diff, −1.129 [Cl, −2.380, 0.1214] F(1,62)=3.258; p=0.0759	Diff, −0.03315 [Cl, −1.284, 1.217] F(1,62)=0.002808; p= 0.9579	Diff, 1.690 [Cl, −0.8115, 4.191] F(1,62)=1.824; p=0.1818
holding current (pA)	Diff, −4.307 [Cl, −14.78, 6.163] F(1,62)=0.6762; p=0.4141	Diff, 6.483 [Cl, −3.986, 16.95] F(1,62)=1.532; p=0.2204	Diff, −11.96 [Cl, −32.90, 8.890] F(1,62)=1.303; p=0.2580

Voltage-dependent properties of the transient potassium current change with development and PNA treatment.

First, we characterized the amplitude, density, and voltage-dependences of activation and inactivation of the transient current (Figure 3, Table 3). Figure 3A shows representative traces. Neither development nor PNA treatment alter the maximum peak current measured in GnRH neurons at −10 mV (Figure 3B; two-way ANOVA/Fishers least significant difference (LSD). There was, however, an interaction between development and PNA treatment on transient current density. This difference appears to be driven by a PNA-induced increase in this parameter at 3 weeks of age (Figure 3C two-way ANOVA/Fisher’s LSD). Figure 3D shows the normalized current inactivation (left) and normalized conductance activation (right) as a function of voltage for the transient current. The V_0.5 of activation and V_0.5 of inactivation were more depolarized in cells from 3-week-old than adult control mice (Figures 3E, 3F; two-way ANOVA/Fisher’s LSD). PNA treatment hyperpolarized the V_0.5 of inactivation at 3 weeks of age. Neither development nor PNA treatment altered the slope factor of activation or inactivation of the transient current (Figures 3G, 3H; two-way ANOVA/Fisher’s LSD).

Figure 3.

Characterization of the transient potassium current. A, Representative traces illustrating mathematical isolation of the transient current (right) by subtracting the −40mV prepulse traces in the middle panel from the −100mV prepulse traces in the left panel. Only three voltage-steps are shown for clarity. B, C, Individual values and mean±SEM for maximum current (B) and current density (C). D, Voltage-dependence of inactivation (left) and activation (right). Fraction of maximum refers to normalized current for inactivation and normalized conductance for activation. Boltzmann fits of the mean data are shown for each group. E-H, Individual values and mean±SEM V_0.5 activation (V_0.5 act, E), V_0.5 inactivation (V_0.5 inact, F), inactivation slope factor (G), and activation slope factor (H). For B, C and E-H, two-way ANOVA/Fisher’s LSD; statistical parameters are in Table 3; tx = treatment. Error bars are often within the size of the symbol.

Table 3.

Descriptive statistics and statistical parameters from two-way ANOVA/Fisher’s LSD for the voltage-dependent properties of the transient current.

Descriptive Statistics (mean ± SEM)
Property	VEH 3-weeks	PNA 3-weeks	VEH adults	PNA adults
V0.5 activation (mV)	−33.427 ± 1.181	−36.269 ± 0.841	−38.167 ± 1.367	−35.744 ± 1.464
Activation slope factor	3.765 ± 0.295	3.674 ± 0.320	3.888 ± 0.321	3.789 ± 0.543
V0.5 inactivation (mV)	−58.976 ± 0.005	−62.269 ± 0.942	−63.671 ± 1.119	−61.736 ± 1.158
Inactivation slope factor	−5.750 ± 0.103	−5.750 ± 0.095	−5.804 ± 0.106	−5.504 ± 0.095
Maximum current (pA)	5.222 ± 0.314	5.771 ± 0.346	5.901 ± 0.310	5.710 ± 0.432
Maximum current density (pA/pF)	0.377 ± 0.022	0.479 ± 0.029	0.442 ± 0.033	0.417 ± 0.034

Two-way ANOVA
Property	Age	Treatment		Interaction
V0.5 activation (mV)	Diff, 2.108 [Cl, −0.3297, 4.564] F(1,54) =3.006 ; p=0.0887	Diff, 0.2094 [Cl, −2.228, 2.647] F(1,54) = 0.029965; p=0.8638		Diff, 5.265 [Cl, 0.3891, 10.14] F(1,54) = 4.687; p=0.0348
	Fisher’s LSD	3-week VEH versus 3-week PNA	3-week VEH versus adult VEH	3-week PNA versus adult PNA	adult VEH versus adult PNA
		p=0.0988	p=0.0089	p=0.7578	p=0.1712
Activation slope factor	Diff, −0.1189 [Cl, −0.3199, 0.08204] F(1,54) = 1.408; p=0.2406	Diff, 0.09526 [Cl, −0.1057, 0.2962] F(1,54) = 0.9030, p=0.3462		Diff, −0.007502 [Cl, −0.3944, 0.4094] F(1,54) = 0.0014, p=0.9703
V0.5 inactivation (mV)	Diff, 2.081 [Cl, −0.0300, 4.192] F(1,54) = 3.906; p=0.0532	Diff, 0.6790 [Cl, −1432, 2.790] F(1,54) = 0.4158; p=0.5218		Diff, 5.228 [Cl, 1.005, 9.450] F(1,43) = 6.162; p=0.0162
	Fisher’s LSD	3-week VEH versus 3-week PNA	3-week VEH versus adult VEH	3-week PNA versus adult PNA	adult VEH versus adult PNA
		p=0.0287	p=0.0030	p=0.7175	p=0.2064
Inactivation slope factor	Diff, −0.09627 [Cl, −0.2965, 0.1040] F(1,54) = 0.9292, p=0.3394	Diff, −0.1496 [Cl, −0.3498, 0.05064] F(1,54) = 2.243, p=0.1400		Diff, 0.2993 [Cl, −0.1011, 0.6998] F(1,54) = 2.246, p=0.1398
Maximum current (pA)	Diff, −0.3096 [Cl, −01.020, 0.4009] F(1,54) = 0.7634, p=0.3861	Diff, −0.1793 [Cl, −0.8898, 0.5312] F(1,53) = 0.2559, p=0.6150		Diff, −0.7400 [Cl, −2.161, 0.6810] F(1,54) = 1.0909, p=0.3011
Maximum current density (pA/pF)	Diff, 0–0.0018 [Cl, −0.06185, 0.05832] F(1,54) = 0.0035, p=0.9532	Diff, −0.03836 [Cl, −0.09844, 0.02172] F(1,54) = 1.6838, p=0.2060		Diff, −0.1283 [Cl, −0.2464, −0.0061] F(1,54) = 4.439, p=0.0398
	Fisher’s LSD	3-week VEH versus 3-week PNA	3-week VEH versus adult VEH	3-week PNA versus adult PNA	adult VEH versus adult PNA
		p=0.0183	p=0.1376	p=0.1468	p=0.5674

Bold indicates p<0.05.

Time-dependence of inactivation and recovery from inactivation of the transient potassium current change with development and PNA treatment.

Next, we characterized the time-dependence of inactivation (Figure 4A–D) and recovery from inactivation (Figure 4E–H) of the transient current (Table 4). Representative traces are in Figures 4A and E, and the time-dependence of inactivation and removal of inactivation in Figures 4B and F. Neither age nor PNA treatment altered the fast time constant of inactivation (Figures 4C, D). In contrast, there was an increase in the fast time constant for recovery in PNA adults compared to both 3-week-old PNA mice and adult controls (Figures 4G, H; two-way ANOVA/Fisher’s LSD). Neither the slow time constant for inactivation nor recovery were affected by any treatment.

Figure 4.

Characterization of the time dependence of inactivation and recovery from inactivation of the transient current in GnRH neurons. A, E, Representative traces illustrating the time dependence of transient current inactivation (A) and recovery from inactivation (E). Only three voltage-steps are shown for clarity (bottom). B, F, Current was normalized to the maximum and plotted as a function of time (B, inactivation; F, recovery). The solid lines represent the double exponential equation evaluated using the mean fast and slow time constants for each group. C, D, G, H, Individual values and mean±SEM for the fast time constant of inactivation (inact; C), the slow time constant of inactivation (inact; D), the fast time constant of recovery (recov; G), and the slow time constant of recovery (recov; H). C, D, G, H, two-way ANOVA/Fisher’s LSD; statistical parameters are in Table 4.

Table 4.

Descriptive statistics and statistical parameters from two-way ANOVA/Fisher’s LSD for the time-dependence of inactivation and recovery from inactivation for the transient current.

Descriptive Statistics (mean ± SEM)
Property	VEH 3-weeks	PNA 3-weeks	VEH adults	PNA adults
fast inactivation time constant (ms)	61.065 ± 4.456	54.449 ± 5.563	48.896 ± 4.670	56.888 ± 5.254
slow inactivation time constant (ms)	439.708 ± 29.126	325.946 ± 61.406	329.105 ± 64.934	301.855 ± 46.740
fast recovery time constant (ms)	14.802 ± 0.682	16.295 ± 1.039	16.088 ± 0.978	20.023 ± 1.389
slow recovery time constant (ms)	61.596 ± 13.174	73.841 ± 26.838	62.848 ± 11.798	92.188 ± 19.666

Two-way ANOVA
Property	Age	Treatment			Interaction
fast inactivation time constant (ms)	Diff, 2.185 [Cl, −7.626, 12.00] F(1,45)=0.2102; p=0.6559	Diff, 1.992 [Cl, −7.819, 11.80] F(1,45)=0.1672; p=0.6845			Diff, 19.97 [Cl, 0.3468, 39.59] F(1,45)=4.201; p=0.0462
	Fisher’s LSD	3-week VEH versus 3-week PNA	3-week VEH versus VEH adult		3-week PNA versus PNA adult	VEH adult versus PNA adult
		p=0.0854	p=0.0876		p=0.2581	p=0.2574
slow inactivation time constant (ms)	Diff, 67.35 [Cl, −35.48,173.5] F(1,45)=1.734; p=0.1945	Diff, 70.51 [Cl, −32.49, 173.5] F(1,45)=1.901; p=0.1748			Diff, 86.51 [Cl, −119.5,292.5] F(1,45)=0.7155; p=0.4021
fast recovery time constant (ms)	Diff, −2.507 [Cl, −4.602,−0.4113] F(1,45)=5.805; p=0.0201	Diff, −2.714 [Cl,−4.810,−0.6188] F(1,45)=6.806; p=0.0123			Diff, 2.443 [Cl, −1.748, 6.634] F(1,45)=1.379; p=0.2465
	Fisher’s LSD	3-week VEH versus 3-week PNA		3-week VEH versus VEH adult	3-week PNA versus PNA adult	VEH adult versus PNA adult
		p=0.3110		p=0.3824	p=0.0158	p=0.0111
slow recovery time constant (ms)	Diff, −9.799 [Cl, −47.37,27.77] F(1,45)=0.2760; p=0.6019	Diff, −20.79 [Cl, −58.36,16.78] F(1,45)=1.242; p=0.2709			Diff, 17.10 [Cl, −58,04, 92.24] F(1,45)=0.2100; p=0.6490

Bold indicates p<0.05.

Development and PNA treatment interact to alter the voltage-dependence of activation of the residual current.

We next tested if development and/or PNA treatment alter the residual current (Figure 5A). Neither maximum peak current nor current density were altered by development or PNA treatment alone (Figures 5B, C; two-way ANOVA, Table 5). Figure 5D shows the normalized conductance as a function of voltage estimated from the residual current. There was an age by treatment interaction on the V_0.5 of activation. This was attributable to a development-related hyperpolarization of V_0.5 activation in control mice and a PNA-induced depolarization of V_0.5 activation in adult mice (Figure 5E). Neither development nor PNA treatment altered the slope factor of activation of the residual current (Figure 5F; two-way ANOVA).

Figure 5.

Characterization of the residual potassium current in GnRH neurons. A, Representative traces illustrating the activation of the residual current (top). Only three voltage-steps are shown for clarity (bottom). Individual values and mean±SEM for maximum current at −10 mV. (B), current density (C). D, Voltage-dependence of activation plotted as fraction of maximum conductance; lines are Boltzmann fits of the mean data for each group. Individual values and mean±SEM for V_0.5 activation (V_0.5 act, E), and activation (act) slope factor (F). B, C, E, F two-way ANOVA/Fisher’s LSD. Statistical parameters are in Table 5.

Table 5.

Descriptive statistics and statistical parameters from two-way ANOVA/Fisher’s LSD for the voltage-gated properties of the residual current.

Descriptive Statistics (mean ± SEM)
Property	VEH 3-weeks	PNA 3-weeks	VEH adults	PNA adults
V0.5 activation (mV)	−8.352 ± 0.905	−10.842 ± 0.937	−11.838 ± 1.423	−8.193 ± 1.172
Activation slope factor	3.724 ± 0.061	3.850 ± 0.077	3.632 ± 0.097	3.640 ± 0.137
Maximum current (pA)	4.337 ± 0.248	4.492 ± 0.224	4.731 ± 0.318	4.769 ± 0.236
Maximum. current density (pA/pF)	0.313 ± 0.017	0.376 ± 0.024	0.351 ± 0.027	0.329 ± 0.023

Two-way ANOVA
Property	Age	Treatment		Interaction
V0.5 activation (mV)	Diff, 0.4182 [Cl, −1.826, 2.663] F(1,54) = 1396, p=0.702	Diff, −0.5773 [Cl, −2.882, 1.667] F(1,54) = 0.2660, p=0.6082		Diff, 6.135 [Cl, 1.646, 10.62] F(1,54) = 7.508, p=0.0083
	Fisher’s LSD	3-week VEH versus 3-week PNA	3-week VEH versus adult VEH	3-week PNA versus adult PNA	adult VEH versus adult PNA
		p=0.1157	p=0.0347	p=0.0947	p=0.0275
Activation slope factor	Diff, 0.1510 [Cl, −0.04228, 0.3444] F(1,54) = 2.454, p=0.1231	Diff, −0.06698 [Cl, −0.2603, 0.1264] F(1,54) = 0.4824, p=0.4903		Diff, −0.1181 [Cl, −0.5047, 0.2686] F(1,54) = 0.3749, p=0.5429
Maximum current (pA)	Diff, −0.3357 [Cl, −0.8520, 0.1806] F(1,54) = 01.700,p=0.1979	Diff, −0.09675 [Cl, −0.6131, 0.4195] F(1,54) = 0.1412, p=0.7086		Diff, −0.1177 [Cl, −1.150, 0.9149] F(1,54) = 0.05218, p=0.8202
Maximum. current density (pA/pF)	Diff, 0.004584 [Cl, −0.04177, 0.05094] F(1,54) = 0.0393, p=0.8436	Diff, −0.02063 [Cl, −0.06698, 0.02573] F(1,54) = 7961, p=0.3762		Diff, −0.08477 [Cl, −0.1775, 0.0079939] F(1,54) = 3.361, p=0.0723

Bold indicates p<0.05.

Discussion

The pattern of GnRH/LH release can change in ways that promote reproduction, such as through the pubertal transition, and in ways that impair reproduction, such as persistent high-frequency release in women with PCOS. Changes in GnRH neuron firing rate are likely to underlie changes in GnRH release⁴⁶, and potassium currents play a major role in sculpting neuronal firing activity^29–32. Here we demonstrate that changes in voltage-dependent potassium currents of GnRH neurons occur during pubertal development and that PNA treatment further alter some characteristics of these currents.

GnRH neurons express a fast transient potassium current^33,35 that responds to subthreshold changes in voltage and plays a role in regulating neuronal excitability and firing properties⁴⁷. Transient current density was increased in GnRH neurons from PNA mice at 3-weeks of age compared to 3-week-old controls. This finding extends prior work in this PNA model that showed that GnRH neurons from PNA females have a lower spontaneous firing rate than controls at 3-weeks of age by providing a possible mechanistic explanation; greater transient potassium current density typically resists depolarization⁴⁸, which may lead to cells initiating firing bouts less frequently in the absence of other changes. Further, the increased transient current density in 3-week-old PNA females could contribute to the blunted firing response to exogenous GABA exposure observed in GnRH neurons from PNA mice compared to controls at 3-weeks of age. Of interest, the increased transient current density in 3-week-old PNA females may be a compensatory mechanism initiated in GnRH neurons by the increased excitatory GABAergic neurotransmission at this age as a way of providing neuroprotection in PNA mice^49–52.

In addition to current density, the voltage and time dependence of transient potassium currents help sculpt firing in excitable cells. Regarding the former, a clear development-associated hyperpolarization of both the inactivation and activation of the transient current occur in control mice. Hyperpolarization of inactivation in adults would tend to make less current available in the subthreshold range, perhaps contributing to increased excitability of adult GnRH neurons compared to those from prepubertal mice²⁸. The hyperpolarization of activation in adult controls may sculpt aspects of the action potential spike, including the increased afterhyperpolarization potential²⁸. In PNA females, there were no changes in either voltage-dependence of inactivation or activation with development. In contrast to the effects of development on voltage dependence of inactivation, PNA treatment had a more marked effect on its time dependence. Most notably, recovery from inactivation took longer in PNA adults in comparison to other groups. This may be a contributing factor to the increased spontaneous activity observed in GnRH neurons from PNA mice in adulthood²³. Together these data indicate changes in transient current properties occur concomitant with pubertal development, and that prenatal androgen exposure both disrupts typical developmental changes in this current and induces an independent shift in properties.

The speculative attribution of the effects of specific properties of the transient current to prior observations in GnRH neuron physiology in these same animal models is done with caution for several reasons. First, different properties of the same macroscopic current can have different predicted effects on neuronal output in the same animal model. For example, changes in voltage-dependence of inactivation of the transient current from 3-weeks of age to adults in control mice might increase excitability, whereas the lack of difference in transient current density between these same groups might predict no change. Second, the direction of change induced by a particular property in isolation is a postulate. For example, increased transient current can be associated with reduced excitability³⁵ but can also induce membrane potential changes that enhance recovery from inactivation of the fast sodium current and thus increase firing rate⁵³. Third, the effects of the transient potassium current on firing and excitability do not occur in isolation but are integrated with changes in other GnRH neuron intrinsic properties that may occur as the animal goes through puberty and/or in response to PNA treatment. In this regard, we observed age-dependent changes in the activation of the residual potassium current in controls but PNA treatment-dependent shifts in this property in adults. Activation of this current typically occurs at the depolarized membrane potentials achieved during an action potential. The residual current is important in repolarizing the membrane, sculpting action potential shape^54,55 and determining the generation of subsequent action potentials in a burst⁵⁴. In adult control females, hyperpolarized residual current activation might contribute to reduced spontaneous firing activity of GnRH neurons in the group compared to 3-week-old controls and to adult PNA females. Fourth, there are documented changes in GABAergic fast synaptic input with age and PNA treatment that will also contribute to determining the overall firing rate of these cells^17,24. Finally, other factors, such as neuromodulator input, may also differ among groups. Of interest in this regard, no difference was observed in the spontaneous firing rate of arcuate kisspeptin neurons among these same four groups⁵⁶, but interactions with other neuromodulatory neurons and with non-neuronal cells could alter GnRH neuron output in a PNA-treatment and/or age-dependent manner. Computational modeling would be one way to address these interactions, but there are no data on the other intrinsic properties of GnRH neurons in these animal models, limiting the utility of this approach at present.

The observed changes in the voltage- and time-dependent properties of potassium currents in this study extend work in other brain regions indicating these currents change across development. For example, in acutely-dissociated large aspiny neurons, transient current inactivation is depolarized and occurs faster with increasing age; this may contribute to reduced spike latency in mature versus young neurons⁵⁷. Developmental changes in voltage-dependent potassium currents are associated with establishing reliable spiking in the calyx of Held⁵⁸. Potassium current alterations can also underlie neuropathological conditions, including pain⁵⁹, neurodegeneration⁶⁰, and epilepsy⁶¹. Within the reproductive neuroendocrine system, changes in potassium currents, in particular the transient current, are associated with changes in GnRH and arcuate kisspeptin neuron physiology. In a model for estradiol positive feedback in which GnRH neurons are more active and more excitable^62,63, transient current density is suppressed and other properties shifted in comparison to a gonadectomized open feedback loop condition³⁵. In contrast, no differences were found between the open loop and a model of negative feedback; there were also no differences in excitability between these groups³³. Estradiol also reduces transient current amplitude in arcuate kisspeptin neurons; in these cells, reducing this current enhances synaptic response and is associated with altered firing activity⁶⁴. Of note, most prior studies on GnRH and kisspeptin neuron currents were done in adults, thus no developmental data are available. The steroid regulation of these currents is nonetheless applicable to the present work. The increased production of sex steroids that occurs with puberty might be the catalyst behind the developmental change observed in potassium currents observed in cells from control mice, as well as excitability and the GnRH neuron transcriptome^28,45. Steroidogenesis is also altered in adult PNA mice, which have increased testosterone levels^17,21,65. While many studies have examined estrogen effects on potassium currents in neuronal populations^{33,35,55,64,66}, we found no studies of androgen effects on central neurons. Work in the cardiovascular system suggests androgen-associated changes in potassium currents. Cardiovascular disease, including arrhythmias, are more common in men⁶⁷. Electrocardiogram QTc intervals are similar among prepubertal males and female and adult females, whereas adult males have shorter QTc intervals^68–70. In rat cardiomyocytes, dihydrotestosterone (DHT), a non-aromatizable androgen, upregulates the delayed-rectified potassium channel and shortens the electrocardiogram QT interval, which correspond to the action potential duration in ventricular cardiomyocytes, compared to castrated males without DHT treatment⁷¹. In immortalized pituitary GH3 cells, androgens rapidly and reversibly block a sustained current⁷², and in cancer cells, voltage-gated potassium currents are associated with cell proliferation⁷³. These studies support the postulate that age and PNA-induced changes in sex steroid milieu in the present studies could be the catalysts underlying changes observed in voltage-gated potassium currents.

The changes in voltage-dependent potassium currents in GnRH neurons described in this study likely play a role in shaping the activity of GnRH neurons through typical pubertal development and the changes induced by prenatal androgen exposure. Untangling these underlying mechanisms has implications for increased GnRH/LH release frequency and altered pubertal progression in hyperandrogenemic adolescents and individuals with PCOS.

Abbreviations:

GnRH

Gonadotropin-releasing hormone

PNA

prenatal androgenization

LH

luteinizing hormone

FSH

follicle stimulating hormone

PCOS

polycystic ovary syndrome

DHT

dihydrotestosterone

GABA

gamma aminobutyric acid

GFP

green-fluorescent protein

VO

vaginal opening

AGD

anogenital distance

LSD

least significant difference

tx

treatment

act

activation

inact

inactivation

recov

recovery