Long-term Recordings of Arcuate Nucleus Kisspeptin Neurons Across the Mouse Estrous Cycle

Abstract

The arcuate nucleus kisspeptin (ARNKISS) neurons represent the GnRH pulse generator that likely drives pulsatile gonadotropin secretion in all mammals. Using an improved GCaMP fiber photometry system enabling long-term continuous recordings, we aimed to establish a definitive profile of ARNKISS neuronal activity across the murine estrous cycle. As noted previously, a substantial reduction in the frequency of ARNKISS neuron synchronization events (SEs) occurs on late proestrus and extends into estrus. The SE amplitude remains constant throughout the cycle. During metestrus, we unexpectedly detected many multipeak SEs where many SEs occurred rapidly, within 160 seconds of each other. By applying a machine learning-based, k-means clustering analysis, we were further able to detect substantial within-stage variability in the patterns of pulse generator activity. Estrous cycle-dependent changes in SE activity occurred around the time of lights on and off. We also find that a mild stressor such as vaginal lavage reduces ARNKISS neuron SE frequency for up to 3 hours. These observations provide a comprehensive account of ARNKISS neuron activity across the estrous cycle, highlight a new pattern of multipeak SE activity, and introduce a new k-means clustering approach for analyzing ARNKISS neuron population behavior.

Keywords: GnRH pulse generator; arcuate nucleus; estrous cycle; k-means clustering; kisspeptin; luteinizing hormone; pulsatility.

Figure 1.

ARN^KISS neuron pulse generator activity across the estrus cycle. (A-D) Representative 24-hour GCaMP photometry traces recorded from the same female mouse transitioning from diestrus to proestrus (A), proestrus to estrus (B), estrus to metestrus (C), and from metestrus to diestrus (D). The shaded area indicates the period of lights off (from 7 Pm to 7 Am). The dotted line shows the threshold for detecting synchronization events. Note the emergence of the multipeak synchronization events (mpSEs) in the estrus and metestrus stages (C, D). Numbers above the traces indicate the number of individual peaks in each mpSE. Estrous stage was determined at the beginning and at the end of each recording. Each stage of the estrous cycle was defined as beginning at midnight.

Figure 2.

ARN^KISS neuron pulse generator activity across the estrus cycle. (A-D) Representative 24-hour GCaMP photometry traces recorded from a different female mouse transitioning from diestrus to proestrus (A), proestrus to estrus (B), estrus to metestrus (C), and from metestrus to diestrus (D). The shaded area indicates the period of lights off (from 7 Pm to 7 Am). The dotted line shows the threshold of detecting synchronization events. Note the emergence of the multipeak synchronization events (mpSEs) in the estrus and metestrus stages (C, D). Numbers above the traces indicate the number of individual peaks in each mpSE. Estrous stage was determined at the beginning and at the end of each recording. Each stage of the estrous cycle was defined as beginning at midnight.

Figure 3.

Bar graphs showing the frequency and other features of synchronization events (SEs) of the ARN^KISS neurons across the estrous cycle. (A) Average number of SEs per hour plotted in 6-hour bins (N = 10-16 mice/stage) over 24 hours (N = 10, 14, 15, and 16 recordings started in D, P, E, and M, respectively). The “lights off” period (from 7 Pm to 7 Am) is indicated by shading. *P < .05 shows significant difference vs other 6-hour periods indicated by lines above the bars. Abbreviations: D, diestrus; E, estrus; M, metestrus; P, proestrus. Note that the 1 Am to 7 Am and 7 Am to 1 Pm bars in diestrus are double plotted. (B) Amplitude of SEs normalized to the highest SE amplitude value across all recordings from the same mouse over the estrous cycle (>72 hours, N = 4-10 mice/stage). (C) Width of SEs measured at the half-maximum of the amplitude (N = 10-16 mice/stage). (B, C) The analysis was performed in DP, PE, EM, and MD, meaning 24-hour recordings during which mice were transitioning from proestrus to estrus (PE), estrus to metestrus (EM), metestrus to diestrus (MD), and diestrus to proestrus (DP). (A-C) All recordings were started between 9 and 11 Am. (D-F) Changes in SE frequency over three 3-hour time blocks in diestrous mice that were undisturbed (N = 5); (D), following “sham lavage” handling (N = 5); (E) or vaginal lavage (N = 6); (F) occurring at the start of the second 3-hour period, respectively. The 3-hour time bins started at 7 Am, 10 Am, and 1 Pm. **P < .01 and *P < .05. Data are represented as mean ± SEM. For post hoc analysis, Holm-Sidak multiple comparison test was used.

Figure 4.

Histograms showing the distribution of the inter-SE intervals as the percentage of all inter-SE intervals in the given estrous stage in 5-minute time bins between 0 and 180 minutes. (A-A′) The 24-hour recordings during which mice were transitioning from diestrus to proestrus (n = 337 events, N = 10 mice), (B-B′) proestrus to estrus (n = 238, N = 14), (C-C′) estrus to metestrus (n = 372, N = 15), and (D-D′) and from metestrus to diestrus stages (n = 562, N = 16). (A′, B′, C′, and D′) Histograms plotting all inter-SE intervals less than 3 minutes for each stage in 20-second time bins. Note that short inter-SE intervals were detected almost exclusively in estrous and metestrous stages (C′ and D′). All photometry recordings were started between 9 and 11 Am. Estrous stage was determined at the beginning and at the end of each recording.

Figure 5.

Features and distribution of multipeak synchronization events (mpSEs) exhibited by ARN^KISS neurons. (A-B) Representative GCaMP photometry traces showing mpSEs recorded in metestrus-estrus stages. The gray vertical lines above the traces point to the individual peaks of mpSEs that occur with SE intervals in <160-second intervals. Horizontal arrows with numbers show the time course of the mpSEs. (C) Scatter plot indicating the number of mpSEs over 24-hour recordings, during which mice were transitioning from diestrous to proestrous (DP, N = 10), proestrous to estrous (PE, N = 14), estrous to metestrous (EM, N = 15), and metestrous to diestrous (MD, N = 16) stages. (D-E) Bar graphs plotting the number of mpSEs per hour (D), and the average number of peaks per mpSEs (E) in 6-hour time blocks over 48 hours during transitioning from estrus across metestrus (M, N = 15) to diestrus (D, N = 16). The “lights off” period (from 7 Pm to 7 Am) is indicated by shading. All photometry recordings were started between 9 and 11 Am. Estrous stage was determined at the beginning and at the end of each recording. Each stage of the estrous cycle was defined as beginning at midnight. Data are represented by mean ± standard error of the mean. *P < .01 vs proestrus, and #P < .05 (Dunn multiple comparisons test).

Figure 6.

Unsupervised k-means clustering reveals dynamic changes in the activity profile of the ARN^KISS neurons with high temporal resolution. (A) Radar plot demonstrating parameters used for the cluster assignments with the following axes: (top) SD of inter-SE intervals, (top right) number of multipeak SEs (mpSEs), (bottom right) number of single SEs (with mpSEs counted as a single SE), (bottom left) number of peaks in mpSEs, (top left) duration of mpSEs. (B) Cluster centroid values for the normalized parameters used for k-means clustering. All axes have a minimum value of 0 and maximum value of 1. (C) Hourly k-means cluster assignments for 96-hour continuous photometry recordings from 5 mice. Colored fields represent the assigned clusters for the center (fourth hour) of the 7-hour time windows (see Methods). In each 24-hour segment of the recording, the first and the last 3 hours of each day were extrapolated from the assignment of the fourth and 21st hours, respectively. The exact time of vaginal lavage is indicated by an arrow (D, diestrus; E, estrus; M, metestrus; P, proestrus). Note that results of the vaginal cytology were used for later reference, not for the k-means clustering. (D) K-means cluster assignments for 24-hour segments of the whole dataset (N = 71) during transitioning from proestrus to estrus (PE, N = 14), estrus to metestrus (EM, N = 15), metestrus to diestrus (MD, N = 16), diestrus to proestrus (DP, N = 10) as well as during recordings when mice stayed in a stage for >24 hours, in diestrus (DD, N = 14) and estrus (EE, N = 2). (E) Bar graph indicating the proportion of the cluster assignments for each stage of the estrus cycle.