Characterization of GnRH Pulse Generator Activity in Male Mice Using GCaMP Fiber Photometry

Abstract

Kisspeptin neurons located in the hypothalamic arcuate nucleus are thought to represent the GnRH pulse generator responsible for driving pulsatile LH secretion. The recent development of GCaMP6 fiber photometry technology has made it possible to perform long-term recordings of the population activity of the arcuate nucleus kisspeptin (ARN^KISS) neurons in conscious-behaving mice. Using this approach, we show that ARN^KISS neurons in intact male mice exhibit episodes of synchronized activity that last ∼2 minutes and have a mean inter-episode interval of 166 minutes, with a very wide range (43 to 347 minutes). Gonadectomy resulted in dramatic changes in the dynamics of ARN^KISS neuron behavior with temporally distinct alterations in synchronization episode (SE) amplitude (sevenfold increase), inter-SE frequency (range, 2 to 58 minutes), and duration (up to 28 minutes), including the frequent appearance of seemingly unstable clusters of doublet and triplet SEs. The combination of photometry with repeated blood sampling revealed a perfect correlation between ARN^KISS neuron population SEs and LH pulses in intact and short-term gonadectomized (GDX) mice. No differences were detected in SE frequency across 24 hours in either intact or GDX mice. These observations further support a role for ARN^KISS neurons as the GnRH pulse generator and show that it operates in a stochastic manner without diurnal variation in both intact and GDX male mice. The removal of gonadal steroids has multiple time-dependent effects upon ARN^KISS neuron synchronizations, indicating their critical role in shaping pulse generator behavior.

The pulsatile release of LH is essential for normal reproductive function in male and female mammals (1, 2). It is well established that LH pulses arise from an abrupt and transient increase in GnRH secreted into the pituitary portal vasculature (3). This episodic pattern of GnRH release is driven by the synchronous activation of a population of arcuate nucleus kisspeptin (ARN^KISS) neurons that target GnRH neuron distal elements nearby the median eminence (4). Humans and rodents with kisspeptin or kisspeptin receptor mutations exhibit absent or highly abnormal pulsatile LH secretion (5–7). Furthermore, the selective activation or suppression of ARN^KISS neuron activity in vivo modulates pulsatile LH secretion, and these cells exhibit brief periods of synchronization prior to each LH pulse (8, 9). Alongside evidence accumulated over recent years (10), it seems very likely that a subpopulation of ARN^KISS neurons represent the GnRH pulse generator (4).

Studies in a range of species have demonstrated that pulsatile LH secretion in male subjects occurs in an irregular and infrequent manner. In humans, monkeys, sheep, and rodents, an LH pulse occurs approximately once every 3 hours, although the interpulse interval is highly variable (11–15). In gonadectomized (GDX) male mice, LH pulse frequency increases substantially to between one and three pulses per hour depending on species and is uniformly restored to intact frequencies by treatment with testosterone (16–20). Such studies demonstrate the important role of testosterone in suppressing the activity of the GnRH pulse generator in male mammals.

In a recent investigation using GCaMP6 fiber photometry to evaluate synchronous ARN^KISS neuron activity in mice, we were surprised to find that the GnRH pulse generator was active on average every 9 minutes in GDX male mice (9). Although much faster than previously suspected, 94% of the synchronization episodes (SEs) exhibited by ARN^KISS neurons were correlated with a subsequent LH pulse when using 3-minute blood sampling intervals. This suggested that prior blood sampling constraints had been underestimating the frequency of the pulse generator in GDX mice and highlighted the importance of being able to monitor the activity of the GnRH pulse generator directly. In the current study, we have used the same GCaMP6 fiber photometry approach to examine the activity patterns of the GnRH pulse generator in 24-hour continuous recordings in intact male mice, to establish their relationship to pulsatile LH secretion, and to assess how pulse generator behavior changes after gonadectomy.

Materials and Methods

Animals

Adult male heterozygous KISS1-Cre^+/− or KISS1-Cre^−/− (21) mice (mixed 129S6Sv/Ev C57BL6 genetic background), alone or crossed onto the Cre-dependent reporter Ai9-CAG-tdTomato line, were housed under a 12-hour light/dark schedule (lights on at 6:00 am) with ad libitum access to food and water. All experiments were approved by the University of Otago Animal Welfare and Ethics Committee.

Stereotaxic injections of adeno-associated virus

Cre-dependent adeno-associated virus (AAV) encoding GCaMP6s (AAV9-CAG-FLEX-GCaMP6s-WPRE-SV40, 1.3 × 10¹³ mol/mL; University of Pennsylvania Vectorcore, Philadelphia, PA) was injected bilaterally into the mouse arcuate nucleus (ARN) as reported previously (9). In brief, mice were anesthetized with 2% isoflurane and placed in a stereotaxic apparatus with Carprofen analgesia (5 mg/kg body weight, subcutaneously). A custom-made bilateral Hamilton syringe apparatus holding two 29-gauge Hamilton syringes with needles held 0.9 mm apart was used to perform AAV bilateral injections (anterior-posterior to bregma, −1.2 mm; depth, 5.9 mm). The needles were lowered into place over 2 minutes and left in situ for 3 minutes before the injection was made. AAV (1 μL) was injected bilaterally at a rate of 100 nL/min, and the needles were left in situ for 10 minutes before being withdrawn.

Fiber photometry and blood sampling

Immediately after AAV injection, an optical fiber (400 μm diameter, 0.48 numerical aperture; Doric Lenses, Quebec, QC, Canada) was positioned directly above the ARN using the same coordinates as for the AAV injections but with a depth of 5.8 mm. Mice were housed individually and, after a 1-week recovery period, received daily handling habituation to the fiber photometry recording setup for at least 3 weeks. Over a period of 4 to 12 weeks after surgery, fiber photometry experiments were undertaken to record fluorescence signals from freely behaving mice in their home cages for 24-hour periods using previously reported methodology (9). In brief, excitation light (50 μW at tip) was provided by violet (referred to as 405-nm) and blue (referred to as 490-nm) fiber-coupled light-emitting diodes passing light down the optic fiber at different sinusoidally modulated frequencies (531 and 211 Hz, respectively). A scheduled mode of light emission (5 seconds on/15 seconds off) was used for all long-term recordings, and continuous light was used for short recordings aimed at evaluating the detailed temporal dynamics of SEs. Emitted fluorescence was collected by the same fiber, passed through a 500- to 550-nm emission filter, and focused onto a Newport 2151 photoreceiver. The two GCaMP6s emissions were recovered by demodulating the 405-nm (531-Hz) and 490-nm (211-Hz) signals. The 405-nm signal reports the excitation/emission of GCaMP6s at its isosbestic point and was used to determine background (non–calcium-dependent) fluorescence levels. The 490-nm signal reports calcium-dependent GCaMP6s excitation/emission and was used to monitor population changes in [Ca²⁺]_i. The signals were modulated/demodulated and recorded at 10 Hz with custom software (Tussock Innovation, Dunedin, New Zealand).

For analysis, fluorescence signals (490 to 405 nm) were collected and converted to ΔF/F (%) values as follows: ΔF/F = 100 × (F − Fb)/F, where Fb is the basal fluorescence signal between episodes, and F is the recorded fluorescence. An SE was defined as a sharp elevation in calcium signal with a peak >10% of Fb. In GDX mice, the appearance of multiple SEs occurring close together without a return to Fb were termed “clusters” and could be composed of doublet, triplet, and quadruplet episodes.

To assess the relationship of calcium episodes with pulsatile LH secretion, freely behaving mice were attached to the fiber photometry system, and LH secretion was measured by obtaining 3-μL blood samples every 3 to 6 minutes from the tail tip over a period of 120 to 240 minutes. Blood sampling and subsequent ELISA measurement of LH were undertaken as reported previously (22). The assay sensitivity was 0.04 ng/mL, with intra-assay and interassay coefficients of variation of 9.3% and 10.5%, respectively.

Results

Pulse generator activity in intact male mice

We have previously shown that 51% to 67% of ARN^KISS neurons express GCaMP6 using the combination of Kiss-Cre mice and AAVs used here and that 89% to 96% of the GCaMP6-expressing cells in the ARN are kisspeptin neurons (9). Mice with fiberoptic probes located immediately above the midcaudal regions of the ARN exhibited detectable SEs and were used for studies.

Intact mice (n = 5) were connected to the fiber photometry system, and ARN^KISS neuron population calcium signals measured under the scheduled light mode for 24 hours. Abrupt increases in calcium signal were detected to occur in an irregular manner across this period (Fig. 1A and 1B). The total duration of these SEs was 2.3 ± 0.07 minutes, consisting of an initial rapid increase (0.88 ± 0.07 minutes) followed by a slower decline (1.43 ± 0.09 minutes). Although SEs occurred on average every 166 ± 6 minutes, the intervals between episodes ranged from 43 to 347 minutes without any modal distribution pattern (Fig. 1C). No differences were detected in the frequency of SEs during the two 12-hour light/dark periods (P > 0.05, paired t test) (Fig. 1A, 1B, and 1D).

Relationship of synchronized calcium episodes to pulsatile LH secretion in intact mice

We next examined the relationship of these ARN^KISS neuron SEs to pulsatile LH secretion by taking 3- to 6-minute tail-tip blood samples for a 240-minute period while monitoring calcium signals. Because pulse generator activity is slow and unpredictable in intact male mice, we took regular 6-minute blood samples until an SE was detected and switched to 3-minute blood sampling. This revealed that episodes were perfectly correlated with pulsatile LH secretion; every SE was followed by an LH pulse (n = 9 episodes in five mice) (Fig. 2A and 2B), and no increments in LH were detected without a preceding SE. The mean increment in LH levels from baseline at the time of a pulse was 1261% (range, 255% to 3963%). Analysis of the temporal relationship between calcium episodes and LH secretion demonstrated that the time from the peak of an SE to the peak of an LH pulse was highly consistent at 4.5 ± 0.3 minutes (n = 9 episodes in five mice) (Fig. 2C).

Pulse generator activity in GDX male mice

To generate a detailed profile of pulse generator activity in the absence of gonadal steroids, mice GDX for 4 to 8 weeks (n = 5) were connected to the fiber photometry system, and ARN^KISS neuron population calcium signals were measured under the scheduled light emission mode for 24 hours. These recordings showed a pattern of synchronized episodic behavior (Fig. 3A and 3B) that was very different from that observed in intact mice (compare with Fig. 1). The intersynchronization episode (inter-SE) interval was reduced to a mean of 16.0 ± 0.7 minutes, although, as in intact mice, a wide range of intervals was detected (2 to 58 minutes) (Fig. 3C). We also observed that the number of SEs was not different between the 12-hour light and dark periods (Fig. 3D).

Unexpectedly, GDX mice often exhibited SEs (Fig. 3A and 3B) that appeared as clusters of doublets, triplets, or quadruplets (Fig. 4A). Whereas intact mice only display singlet episodes, GDX animals have 43 ± 8% of their episodes as singlets, 42 ± 6% as doublets, 12 ± 4% as triplets, and 2 ± 1% as quadruplets (Fig. 4B). The duration of singlet SEs was not significantly different (P = 0.06) between intact (2.31 ± 0.07) and GDX (3.10 ± 0.45) mice (Fig. 4C); however, the dynamics were altered. Intact mice exhibited episodes with the initial rise to peak taking 0.88 ± 0.07 minutes, whereas GDX mice showed a faster incline (0.32 ± 0.04 minutes; P = 0.001) (Fig. 4C). The doublet/triplet clusters in GDX mice displayed the same fast onset (0.38 ± 0.04 minutes) with prolonged duration (8.27 ± 1.24 minutes, P = 0.0001). These differences were most clearly visualized using continuous light emission mode photometry providing high (10-Hz) temporal resolution data (Fig. 4D).

The multiple SEs occurring quickly within a cluster allowed us to analyze whether a refractory period exists for ARN^KISS neuron synchronization. To assess this, we plotted the inter-SE interval against the normalized amplitude of the second episode compared with the first episode (Fig. 4E and 4F). This revealed an absolute refractory period of 78 seconds, during which a subsequent SE did not occur (Fig. 4F). After this period, the amplitude of the second episode was approximately the same as the first episode regardless of the inter-SE interval (Fig. 4E).

Temporal changes in pulse generator dynamics after gonadectomy

To examine in detail how pulse generator activity changes with time after gonadectomy, we made repeated 90-minute recordings from mice when intact and 3, 5, 7, 14, and 21 days after gonadectomy (n = 5). Because SEs are infrequent in intact mice (Fig. 1), the 90-minute recording period showed only one or no episodes (Fig. 5A and 5B). Hence, the inter-SE in intact mice was estimated by measuring the time between an episode and either end of the recording (i.e., at 0 or 90 minutes), whichever had a longer interval, or was set at 90 minutes for mice with no episodes. This value (71 ± 8 minutes) is a substantial under-representation of actual inter-episode interval (166 ± 16 minutes, from 24-hour recordings) but enabled statistical analysis of the data across days.

The frequency and amplitude of SEs changed with different temporal dynamics after gonadectomy (Fig. 5A and 5B). As soon as GDX day 3, the inter-ES interval was significantly reduced to 17.46 ± 1.74 minutes (P < 0.01; one-way repeated measures ANOVA followed by post hoc Tukey test) and remained at this level to GDX day 21 (P < 0.005; one-way repeated measures ANOVA followed by post hoc Tukey test) (Fig. 5C). In parallel with this time course, doublet and triplet synchronization clusters were present by GDX day 3 and comprised ∼40% to 50% of all episodes through to GDX day 21 (Fig. 5D). In contrast, the amplitude of SEs increased gradually over the 21-day period. rising from a 22 ± 9% increase in fluorescence in the intact state to a 133 ± 41% change at GDX day 21 (Fig. 5E).

Relationship of synchronized calcium episodes to pulsatile LH secretion in short-term GDX mice

We previously assessed the relationship of synchronized calcium episodes to pulsatile LH secretion in long-term GDX mice and found that 94% of episodes were followed by an LH pulse (9). We suspected that this imperfect relationship resulted from the LH pulses in long-term GDX mice being too fast (9-minute interval) for a 3-minute blood sampling regimen to assess. Because the inter-SE interval is longest (17 minutes) during the first week after gonadectomy, we undertook conjoint photometry and 3-minute pulse bleeding experiments in day 5 GDX mice. This showed a perfect relationship between SEs and pulsatile LH secretion in all five mice assessed (Fig. 6A and 6B). Every SE was followed by an increase in LH concentration, with no increments in LH detected without a preceding calcium event (Fig. 6A and 6B). Using the SEs as a guide, any increment in LH levels above 12% was found to identify an LH pulse. The interval from the peak of the SE to the peak of the LH pulse was highly consistent across all GDX mice, being 4.2 ± 0.2 minutes (33 pulses, n = 5) (Fig. 6C). We noted that the 3-minute handling of mice for tail-tip bleeding resulted in the disappearance of clusters containing multiple episodes, with only single episode synchronizations being recorded (Fig. 6A and 6B).

Discussion

The GCaMP fiber photometry technique enables the activity of the GnRH pulse generator to be measured over long time periods in freely behaving mice. Using this approach, we report that the ARN^KISS neurons exhibit SEs that occur in a highly irregular manner, with the inter-SE interval varying from a mean of 166 minutes in intact male mice to a mean of 16 minutes in GDX mice. A perfect correlation was observed between ARN^KISS neuron SEs and pulsatile LH release, reinforcing the role of these cells as the pulse generator. The dramatic changes observed in SE dynamics after gonadectomy provide insights into the synchronization mechanism and indicate the importance of gonadal steroids in modulating different aspects of GnRH pulse generator function.

Although the pulse generator exhibits an SE on average every 166 minutes in intact male mice, an exceptionally wide range of inter-SE intervals exists, varying from 50 to 350 minutes. Although these intervals are much shorter in GDX mice, there is still considerable variation (2 to 60 minutes). This unpredictable, stochastic pattern of the pulse generator activity has also been indicated by a few long-duration studies examining pulsatile LH secretion in male humans and mice (11, 15). In men, the mean LH pulse interval is 119 minutes, but the interpulse interval ranges from 30 to 480 minutes (11). The mechanisms underlying the stochasticity of pulse generation are currently unknown. In agreement with the present observations, auto-correlation analyses of LH pulse intervals in human male subjects have not found any striking relationship between the time interval of one pulse and the next (23). This indicates that the timing of an LH pulse is independent of any “memory” within the pulse generator (23), although this may not be the case in GDX female monkeys (24).

We found no differences in the pattern of pulse generator activity across the 24-hour recording period in either intact or GDX male mice. This is similar to observations in human male subjects where, on average, there are no significant differences in pulsatile LH secretion detected across the day and night, although individual men were sometimes found to display elevated LH pulse amplitudes at night (11). To our knowledge, the profile of LH pulses across 24 hours has not been assessed in rodents, but an investigation of ARN multiunit activity (MUA), likely to reflect the activity of ARN^KISS neurons (4), in OVX rats detected a small 7% increase in MUA frequency during the night (25). Thus, if they exist, circadian changes in pulse generator activity are subtle. In support of the apparent lack of circadian regulation of LH pulses in male mice, ARN^KISS neurons in this sex have recently been shown to exhibit little if any direct response to either vasopressin or vasoactive intestinal polypeptide, the two primary circadian neuropeptides (26).

Previous work has demonstrated a 94% correlation between ARN^KISS neuron GCaMP calcium SEs and pulsatile LH secretion in the long-term GDX male mouse (9). We now extend this observation by showing a perfect correlation between SEs and pulsatile LH secretion in both intact and 5-day GDX male mice. This improved correlation most likely arises from our use of short-term GDX mice in which the pulse frequency is more appropriate for a 3-minute blood sampling protocol. We have been unable to undertake faster pulse bleeding because 2-minute sampling appears overly stressful to the mice. Indeed, we note that the doublet/triplet clusters in 5-day GDX mice are replaced by singlet SEs even with the 3-minute blood sampling regimen. The reasons for this are unknown at present, but we note that the administration of corticotropin-releasing factor reduces MUA duration in rhesus monkeys, suggesting that stress may affect pulse generator duration (27). Together these observations of a near-perfect correlation between SEs and LH pulses in intact and in short- and long-term GDX male mice provide further strong evidence for the ARN^KISS neurons being the GnRH pulse generator (4).

The peak of an SE occurs 4.2 and 4.5 minutes before the peak of the LH pulse in GDX and intact mice, respectively. This is compatible with the relatively slow activation of the GnRH neuron distal dendron by kisspeptin (1 to 2 minutes) (28) and time required for GnRH to pass through the portal circulation to activate the gonadotrophs and generate an increment in LH release at the tail (1 to 2 minutes) (29).

Gonadectomy resulted in remarkable changes in the frequency, amplitude, and dynamics of the SEs. Whereas the inter-episode interval had quickly reduced by GDX day 3, the amplitude of episodes changed much more gradually, reaching an approximately sixfold higher level at GDX day 21. Multi-episode synchronizations also appeared as soon as GDX day 3. These time differences suggest that independent mechanisms are responsible for modulating different aspects of pulse generator activity. The neural mechanisms responsible for these different components of pulse generation are not established at present.

Regarding the initiation of an SE, we show that both the speed of onset and the magnitude of an episode are elevated substantially in GDX mice. According to the KNDy hypothesis, the initiation and amplitude of the pulse generator is driven by neurokinin B (NKB) signaling among ARN^KISS neurons (10, 30). Whereas insitu hybridization studies have found that levels of Tac2 and its receptor Tacr3 mRNA are upregulated in GDX mice (31, 32), the effects of NKB and NKB receptor agonists on male ARN^KISS neuron excitability in vitro are mixed, with reports of no change or elevated excitability after gonadectomy (33–35). Given that amplitude changes occur slowly over weeks, it is likely to be important to examine specific time points after gonadectomy in future investigations. Nevertheless, these slow changes in the amplitude of SEs may reflect slowly elevating NKB synthesis and releasable pools and/or increased receptor expression among the ARN^KISS neuron population after the removal of gonadal steroids.

The most unexpected observation of these studies was the appearance of multi-episode doublet, triplet, and quadruplet synchronization clusters. These appear to reflect a highly unstable situation whereby the mechanism that normally restrains or terminates an episode is relatively ineffective and allows repeated SEs to occur within a cluster. We found the absolute refractory period for the generation of an SE to be 78 seconds, with episodes generated after this time having a similar amplitude. This all-or-none phenomenon suggests the presence of feedback and/or desensitizing signaling cascades within the ARN^KISS neurons that restrict the maximum frequency of SEs. This is likely to be separate from the unknown mechanism that generates the stochastically generated interval between SEs.

The KNDy hypothesis posits that the termination of an SE occurs as a result of inhibitory dynorphin signaling within the ARN^KISS neuronal network (10). This would predict that the presence of long-duration, multi-episode clusters in GDX mice results from relatively ineffective dynorphin signaling among the ARN^KISS neurons. Evidence for this is equivocal; although ARN^KISS neurons are reported to be less sensitive to dynorphin in GDX male mice (34), other studies have found increased Pdyn mRNA levels in ARN^KISS neurons of GDX mice (31, 32), indicating the potential for enhanced dynorphin signaling in the absence of gonadal steroids. Thus, the role of dynorphin signaling at the ARN^KISS neuron in pulse termination remains unclear. This is further questioned by the apparent absence of dynorphin in ARN^KISS neurons in human male subjects (36).

Fiber photometry records the population activity of the ARN^KISS neurons. As such, it cannot resolve whether the multi-episode synchronizations observed in GDX mice represent the simultaneous repeated synchronizations of the same cells or the possibility that different subpopulations are bursting at slightly different time points around a common initiation signal. The answer to this question will likely only come from the ability to measure the activity of single ARN^KISS neurons in vivo. Unfortunately, the low firing rates exhibited by individual ARN^KISS neurons in the acute brain slice in vitro (35, 37–39) bear little resemblance to their SEs in vivo. Indeed, spontaneous SEs have not been observed in the brain slice, and the overall excitability of individual ARN^KISS neurons is reported to be unchanged, slightly increased, or slightly decreased in GDX mice (31, 35, 37). This suggests that the in vivo synchronization mechanism and its modulation by gonadectomy may derive principally from changes in ARN^KISS neuron network behavior rather than alterations in the intrinsic electrical properties of individual cells.

In summary, we demonstrate the utility of GCaMP fiber photometry for assessing GnRH pulse generator activity in the male mouse. The high temporal resolution of the technique provides a detailed description of the ARN^KISS neuron SEs that precede every LH pulse. This reveals the stochastic nature of pulse generation and the important role of gonadal steroids in modulating multiple parameters of ARN^KISS neuron SEs in male mice and hints at the complexity of pulse generator operation in vivo.

Abbreviations:

Abbreviations:
 
• AAV

  adeno-associated virus

 
• ARN

  arcuate nucleus

 
• ARN^KISS

  arcuate nucleus kisspeptin

 
• GDX

  gonadectomized

 
• inter-SE

  intersynchronization episode

 
• MUA

  multiunit activity

 
• NKB

  neurokinin B

 
• SE

  synchronization episode

Acknowledgments

We thank Prof. William Colledge (University of Cambridge, UK) for the generous provision of the Kiss-Cre mouse line.

Financial Support: This work was supported by the New Zealand Health Research Council.

Disclosure Summary: The authors have nothing to disclose.

References