Multi-dimensional oscillatory activity of mouse GnRH neurons in vivo

The gonadotropin-releasing hormone (GnRH) neurons represent the key output cells of the neural network controlling mammalian fertility. We used GCaMP fiber photometry to record the population activity of the GnRH neuron distal projections in the ventral arcuate nucleus where they merge before entering the median eminence to release GnRH into the portal vasculature. Recordings in freely behaving intact male and female mice revealed abrupt ~8-min duration increases in activity that correlated perfectly with the appearance of a subsequent pulse of luteinizing hormone (LH). The GnRH neuron dendrons also exhibited a low level of unchanging clustered, rapidly fluctuating baseline activity in males and throughout the estrous cycle in females. In female mice, a gradual increase in basal activity that exhibited ~80-min oscillations began in the afternoon of proestrus and lasted for 12 h. This was associated with the onset of the LH surge that ended several hours before the fall in GCaMP signal. Abrupt 8-min duration episodes of GCaMP activity continued to occur on top of the rising surge baseline before ceasing in estrus. These observations provide the first description of GnRH neuron activity in freely behaving animals. They demonstrate that three distinct patterns of oscillatory activity occur in GnRH neurons. These are comprised of low-level rapid baseline activity, abrupt 8-min duration oscillations that drive pulsatile gonadotropin secretion and, in females, a gradual and very prolonged oscillating increase in activity responsible for the preovulatory LH surge.

Methods
We used fiber photometry to record GCaMP fluorescence signal in freely behaving mice for 6 or 24 h periods using previously described methodology (Han, Kane et al. 2019, Han, Morris et al. 2023). This included a custom-built photometry system using Doric components (Doric Lenses, QC, Canada) and National Instrument data acquisition board (TX, USA) based on a previous design (Lerner, Shilyansky et al. 2015). Blue (465-490 nm) and violet (405 nm) LED lights were sinusoidally modulated at frequencies of 531 and 211 Hz respectively and were focused onto a single fiber optic connected to the mouse. The light intensity at the tip of the fiber was 30-80 microwatts. Emitted fluorescence signal from the brain was collected via the same fiber, passed through a 500-550 nm emission filter and focused onto a fluorescence detector (Doric, QC, Canada). The emissions were collected at 10 Hz and the two GCaMP6 emissions were recovered by demodulating the 465-490 nm signals (calcium-dependent) and 405 nm (calcium-independent) signals. Signals were either recorded in a continuous mode or a scheduled 5s on/10s off mode.

Analysis was performed in MATLAB with the subtraction of 405 signal from 465-490 signal to extract the calcium-dependent signal followed by an exponential fit algorithm used to correct for baseline shift. The signal was converted to DF/F (%) values using the equation DF/F=(Frecorded-Fbaseline)/ Fbaseline) x 100. The Findpeaks algorithm was used to detect dSEs, and the duration and the time of SE to the half width full maximum was determined. For deconvolution of signals from 24-h proestrus recordings, the movmean algorithm was used to extract 30-min rolling average.

With this approach, slow oscillatory shifts in baseline during the proestrous surge were detected. The onset and offset of the surge signals were determined by the points at which there was a >5% increment in ΔF relative to all previous timepoints, and when the calcium signal returned to 90% of the baseline value, respectively. The highest value observed during the slow oscillatory phase of the surge was determined to be the peak, and FWHM values were determined from the onset, peak and offset values. The time taken between one trough to another in individual oscillations was calculated as the duration of each oscillation. The second phase of the deconvolution detected all dSE signals using ‘findpeaks’ algorithm and signals between -60 to 360 s around each peak were separated to visualize the remaining low amplitude ‘residual’ signal. A threshold of 5% above baseline was used to extract the residual signal. Residual signals with peaks occurring > 420s from the preceding peak were considered to represent separate clusters.

To examine the relationship between calcium episodes with LH pulses, freely behaving mice were attached to the fiber photometry system, and 4-mL blood samples were obtained every 5 to 10 minutes from the tail tip over a period of 120 to 240 minutes. To assess the relationship between the long calcium increment during proestrus evening and the LH surge, female mice were attached to the fiber photometry system in the morning of proestrus and blood samples (3-mL) collected every 3 hours for 18 hours until the morning of estrus. Levels of LH were measured by in-house LH ELISA (Steyn, Wan et al. 2013) with an assay sensitivity of 0.04 ng/mL and intra-assay coefficient of variation of 8.2%.

Immunohistochemistry

Adult GnRH-Cre,GCaMP6s mice were given a lethal overdose of pentobarbital (3mg/100mL, i.p.) and perfused transcardially with 4% paraformaldehyde. Brains were processed for dual GFP and GnRH immunofluorescence. For GFP immunostaining,  anti-chicken GFP (1:5000, Aves Lab) was used followed by AlexaFluor 488-conjugated goat-anti-chicken (1:1000). For GnRH cell body immunostaining, GA2 guinea pig anti-GnRH antisera (1:3000, gift from G.Anderson, New Zealand) was used in combination with AlexaFluor 647-conjugated goat anti-guinea pig immunoglobulin (1:500, Thermo Fisher Scientific, USA). For GnRH dendron immunostaining, rabbit anti-GnRH (1:20,000, LR1, gift of R.Benoit, Montreal) antisera was used followed by biotinylated goat anti-rabbit immunoglobulin (1:1000, Jackson Immunoresearch) and AlexaFluor 568-conjugated Streptavidin (1:400, Thermo Fisher Scientific, USA). Imaging was performed using a Leica SP8 Laser Scanning Confocal Microscope (Leica Microsystems) at the Cambridge Advanced Imaging Center and analyzed using ImageJ.

Statistical Analysis

All statistical analyses were performed in Prism 10 (GraphPad software Inc.). All values given in this study are mean ± SEM, and significance is defined as P <0.05* or P<0.01**. For inter-peak interval analysis in females across the estrous cycle and the residual cluster activity analysis, Kruskal Wallis ANOVA followed by Dunn’s post-hoc tests was used.