The structural and electrophysiological properties of progesterone receptor expressing neurons vary along the anterior-posterior axis of the ventromedial hypothalamus and undergo local changes across the reproductive cycle

Sex hormone levels continuously fluctuate across the reproductive cycle, changing the activity of neuronal circuits to coordinate female behavior and reproductive capacity. The ventrolateral division of the ventromedial hypothalamus (VMHvl) contains neurons expressing receptors for sex hormones and its function is intimately linked to female sexual receptivity. However, recent findings suggest that the VMHvl is functionally heterogeneous. Here, we used whole cell recordings and intracellular labeling to characterize the electrophysiological and morphological properties of individual VMHvl neurons in naturally cycling females. We found that the properties of progesterone receptor expressing (PR+) neurons, but not PR- neurons, depended systematically on the neuron’s location along the anterior-posterior axis of the VMHvl and the phase within the reproductive cycle. Prominent amongst this, the resting membrane potential of anterior PR+ neurons decreased during the receptive phase, while the excitability of medial PR+ neurons increased during the non-receptive phase. During the receptive phase of the cycle, posterior PR+ neurons simultaneously showed an increase in dendritic complexity and a decrease in spine density. These findings reveal an extensive diversity of local rules driving structural and physiological changes in response to fluctuating levels of sex hormones, supporting the anatomical and functional subdivision of the VMHvl and its possible role in the orchestration of different aspects of female socio-sexual behavior.


Introduction
Female reproductive physiology and behavior are under the control of the ovarian sex hormones estrogen and progesterone, whose fluctuating levels act reversibly in the female brain, organizing the activity of neural circuits to synchronize sexual behavior with reproductive capacity (Jennings & de Lecea, 2020;Snoeren, 2018). The ventrolateral division of the ventromedial hypothalamus (VMHvl) is crucial for female reproductive behavior, in particular for the display of lordosis, the female acceptance posture. Nonspecific electrolytic lesions (Pfaff & Sakuma, 1979a) or ablation of genetically delineated neuronal populations (Rissman et al., 1997;Yang et al., 2013) of the VMHvl virtually abolish lordosis, while electrical stimulation at the same location enhances the probability of females displaying it (Pfaff & Sakuma, 1979b). Importantly, VMHvl neurons have rich expression of receptors for estrogen (ER) and progesterone (PR) and therefore are sensitive to the fluctuating levels of these sex hormones across the reproductive cycle (Jennings & de Lecea, 2020;Snoeren, 2018). In accordance, local infusion of sex hormones in the VMHvl increases female receptivity (Rubin & Barfield, 1983a) while For tissue collection, VGat-Cre-tdtomato mice were deeply anesthetized with a lethal amount of a mixture of 12% of ketamine (Imalgene 1000, Merial) and 8% of xylazine (Rompun 2%,Bayer) in saline solution and perfused transcardially with 0.01M PBS followed by 4% PFA in PBS. The brains were removed, fixated in the 4% PFA solution for ~1 hour and transferred to a 30% sucrose (Sigma-Aldrich) in 0.01M Phosphate-Buffer and 0.1% Sodium Azide (ACROS Organics) to allow cryopreservation. Coronal sections with 45μm were cut from the VMH on a sliding microtome (SM2000R, Leica). Sections were imaged with a Zeiss LSM 710 confocal laser scanning microscope with a 10x and a 25x magnification objectives.

Electrophysiological Data Analysis
A total of 144 recorded neurons were used in the present study. Please see Table 1 and Figure 1 for the numbers of recorded PR+ and PR-neurons, across the reproductive cycle and AP axis.
The action potentials recorded in current-clamp mode were analyzed with custom-written MATLAB code. Each spike within the spike trains obtained upon current injection were detected and used to determine the firing frequency, interspike interval (ISI), threshold to spike, latency to spike and coefficient of variation (CV2= 2|ISIn+1 -ISIn|/(ISIn+1 + ISIn)), calculated at a fixed current injection of 150pA) (Holt et al., 1996). The rebound action potentials observed after injecting hyperpolarizing currents were obtained using the same method.
The first action potential obtained in each train was used to calculate the spike amplitude and spike half-width to prevent frequency dependent artifacts in the spike shape caused by sodium and potassium channel inactivation.
The resting potential was determined by averaging the membrane voltage values before the current injection and the Ih current amplitude was calculated by the difference between the early and late phase of a hyperpolarizing current pulse.
Voltage clamp test-pulses were analyzed using Clampfit 10.7 Software (Molecular Devices, LLC) and access resistance (Ra), membrane resistance (Rm), time constant ( ) and capacitance were calculated. Only neurons with access resistance 1/10 th lower than the membrane resistance were used for analysis. Access resistance was comparable across location, genotype and the phase of the cycle in the present study.
The passive membrane properties membrane resistance (Rm), membrane capacitance (Cmem), and membrane time constant ( ) were obtained immediately after membrane rupture, using a square voltage step (-10 mV, 100 ms). The access resistance was determined by measuring the amplitude of the current response to the command voltage step and the membrane resistance as the difference between the baseline and the holding current in the steady state after the capacitive decay, by applying the ohm's law.
The membrane time constant was determined by a single exponential fit of the decay phase in response to the square pulse. An approximation of the capacitance was made by using the formula = membrane capacitance * access resistance.

Neuronal Reconstruction and Image Data Analysis
A total of 180 neurons VMHvl neurons were filled with biocytin in acute slices obtained from PR-EYFP female mice using whole-cell recording pipettes. These slices containing biocytin filled neurons were fixed and immunostained for the subsequent reconstruction of neuronal structure. Please refer to Table 2 and Figure 6 for the numbers of filled PR+ and PR-neurons across the reproductive cycle and AP axis.
To determine the Bregma coordinate for each reconstructed neuron and ensure that its location fell within the VMHvl, in a separate experiment we first assessed the extension of the VMHvl, taking advantage of the VGAT-tdTomato. The VMH is organized in a core of glutamatergic neurons (Ziegler et al 2002, Yamamoto et al 2018 and a surrounding shell with higher density of GABAergic neurons (Yamamoto et al 2018). Therefore, we assessed the AP extension of the VMH by counting consecutive slices in which the VMH was detectable, as shown by a decrease of the VGAT-tdTomato signal (Fig. 6A) and multiplying by the thickness of the histological slices. A sharp transition from low to high tdTomato fluorescence was observed both in its anterior limit (near the anterior hypothalamus) as well as with its posterior limit (near the premammilary hypothalamus) (data not shown). The extension obtained matched the extension illustrated by the Paxinos Brain Atlas: Anterior limit= 1.05 to posterior limit= 2.06 from Bregma (that is, the VMHvl spans for ~1 millimeter). For each mouse, we systematically sampled 3 consecutive levels of the VMH in slices of 300 microns of thickness. We thus categorized each sampling level as anterior (Bregma -1.05 to -1.30 mm approx.), medial (Bregma -1.30 to -1.65 mm approx.), and posterior (Bregma -1.65 to -2.00 mm approx.) levels of the VMHv (Fig. 6B-C). Localization of the reconstructed neurons in the brain was assessed by matching in Adobe Illustrator CS6 (Adobe Systems Incorporated) the confocal images of the neurons with adapted sections from the Paxinos brain atlas (Franklin & Paxinos, 2008). Only the neurons inside the defined boundaries of the VMHvl were considered for quantification. We used Simple Neurite Tracer178 (Fiji/ImageJ software179) to analyze the morphological properties and Sholl profiles of the neurons filled with biocytin. The Cell Counter Fiji/ImageJ plugin was used to manually count spines.
Cut and resealed dendrites (identified as a globular thickening in the extreme of a dendrite) were quantified for every neuron and did not reveal significant differences across groups (Number of cut dendrites/total number of dendrites quantified per group for PR+ neurons: 21/126 in 26/100 in D-posterior and 17/100 in PE-posterior). When tested with a Fisher's exact test, neither PR+ nor PR-revealed statistical difference in the proportion of cut dendrites. We thus assume that this unavoidable artifact of our method of neuronal recording and reconstruction did not add any group-specific bias in the structural quantification.
In our hands, as well as in previous studies (Calizo & Flanagan-cato, 2000), the intracellular labeling of VMHvl neurons did not always make their thin axonal process visible, and therefore, the axons identified were not considered for analysis.
Graphs were made using custom made MATLAB code or GraphPad Prism 8 Software.

Statistical Analyses
Statistical analyses were performed using GraphPad Prism 8 Software. Normality of the residuals was tested with the D'Agostino-Pearson omnibus K2 test. When normally distributed, three-way ANOVA tests were performed to compare groups in different phases of the cycle (D vs PE) vs location in the AP axis (anterior vs medial vs posterior) vs genotype (PR+ and PR-neurons), using the Sidak test to correct for multiple comparisons. In the properties whose residuals did not pass the normality test, a logarithmic or square root transformation was applied and normality was consequently reassessed. As specified in the legend, in the properties that the transformations did not make the residuals become normally distributed, we performed a Kruskal-Wallis test followed by a post-hoc test for multiple comparisons: For Figure 1A-F, a Mixed-effects test with repeated measures, and Figure 7A, B a two-way ANOVA with repeated measures were performed. For Figure 4 and 5B, C a Chi-square test was performed.
Whisker plots represent median with interquartile range. Error bars represent mean ± SEM. Significance was noted as *p<0.05.

Results
Local changes in the intrinsic excitability and threshold to spike of PR+ neurons across the reproductive cycle.
In order to investigate the neuronal excitability of PR+ and PR-neurons, we recorded current to voltage input-output response curves (I-V curves) at resting potential in currentclamp mode in acute slices of naturally cycling PR-Cre x EYFP adult female mice. Briefly, the female reproductive stage was determined by vaginal lavage and slices were obtained from females in the least receptive state (diestrus) and the most receptive state (proestrus/estrus). PR+ neurons were identified by their natural fluorescence under the microscope and PR-by the absence of fluorescence (see Methods for details).
In the anterior and posterior VMHvl, the excitability of PR+ neurons was found not to change across the cycle ( Fig. 1A and C). In contrast, medial PR+ neurons of nonreceptive females showed significantly higher excitability when compared to the excitability of medial PR+ neurons originating from receptive females, particularly in response to higher input current (Fig. 1B, strong interaction between the phase of the cycle and the amount of current injected (Mixed-effects test, p<0.0001)). The modulation in the excitability of the medial neurons across the reproductive cycle was specific to the PR+ population since no changes were observed across the cycle in the PR-neurons at any of the AP levels ( Fig. 1D-F).
In addition, we also observed that the PR+ population was less excitable than the PR-in the anterior and posterior levels of the VMHvl (Mixed-effects test, p<0.001 and p<0.01, respectively) but not in the medial subdivision, probably in part due to the increased excitability of medial PR+ neurons from non-receptive females.
It is worth mentioning that the membrane resistance did not vary across experimental groups (Table 1) and only a small decrease was observed in the capacitance of the anterior and posterior PR+ of non-receptive females (three-way ANOVA, location/phase interaction p=0.04). The membrane time constant ( ) also did not vary across the reproductive cycle, however posterior PR+ neurons exhibited a smaller when compared to the anterior and medial PR+ population (three-way ANOVA, location p=0.04).
We observed a small decrease in the capacitance of the anterior and posterior PR+ neurons of non-receptive females (three-way ANOVA, location/phase interaction p=0.04).
In turn, while the does not vary across the reproductive cycle, posterior PR+ neurons have smaller when compared to the anterior and medial PR+ population (three-way ANOVA, location p=0.04).
The resting potential of anterior PR+ neurons of receptive females was lower when compared to non-receptive females ( Fig. 2A and B, three-way ANOVA, phase p=0.01, multiple comparison antNon-Rec vs antRec p=0.01). However, this difference was not large enough to produce changes in the threshold to spike (Fig. 2C, Kruskal-Wallis test, p=0.06). The latency to spike of PR+ and PR-neurons did not vary across the cycle ( Fig.   2F and H). In addition, we observed a mild, yet significant, lower latency to spike of the posterior neurons that was independent of the phase of the cycle and the genotype (Fig.   2F,H, three-way ANOVA, location p<0.01).
The regularity of the recorded spike trains was similar across experimental groups, as shown by comparable CV2 values across phase and location for both the PR+ and PRpopulations (regularity was determined at 150pA of injected current, Fig. 2G,I).
In focusing on the action potential shape, we observed a comparable amplitude and halfwidth of the action potentials across the reproductive cycle in PR+ and PR-neurons ( Fig. 3A-E). The spike amplitude of the posterior neurons of both PR+ and PR-populations is moderately but consistently smaller (three-way ANOVA, location p<0.01).
Together, these results suggest that the electrophysiological properties of VMHvl neurons vary across the reproductive cycle, but the observed changes depend on their location in the AP axis. In the sexually receptive phase, anterior PR+ neurons are more hyperpolarized in their resting state and the medial PR+ population exhibits lower excitability.

The proportion of tonic and phasic firing PR+ neurons varies across the anteroposterior axis.
In the present study, we observed two major categories of VMHvl neurons according to their firing patterns: tonic and phasic neurons. Tonic firing neurons exhibited a linear increase in firing rate in response to the increases in the magnitude of current injected.
Interestingly, within this main class of neurons, some neurons did not reach the depolarization block within the range of currents we injected ( Figure 4A Finally, and given that it is well established that hyperpolarizing input can trigger rebound depolarizing responses increasing the firing rate of neurons in a wide variety of brain areas including the hypothalamus (Burdakov et al., 2004;Israel et al., 2008), we sought to characterize the response of VMHvl neurons to hyperpolarizing input. To do so, we quantified the proportion of VMHvl neurons which exhibited rebound firing after hyperpolarization ( Figure 5A, -90pA for 1s). We observed that 6 out of 32 PR+ neurons of non-receptive females and only 2 out of 38 PR+ neurons of receptive females fired rebound action potentials ( Figure 5B). This difference originates mainly from the anterior PR+ neurons (chi-square test, p=0.01), where 4 out of 11 neurons of non-receptive females displayed rebound firing while in receptive females none of the 13 recorded neurons produced rebound action potentials. Furthermore, no significant differences were observed between PR+ neurons across the AP axis nor between the PR+ and the PRpopulations ( Figure 5B and 5C). GABAergic markers were used to delineate the boundaries of the VMHvl, and ensure that the neurons characterized in this study were indeed within the nucleus ( Fig. 6B and C).
The somatic area was unaltered across the reproductive cycle and was not different between the PR+ and PR-populations (Table 2). We observed that PR+ neurons of receptive females exhibit a robust increase in the number of dendrites per neuron across the AP axis (Table 2, three-way ANOVA, genotype/phase interaction p<0.01), and a higher number of branching points per neuron compared to neurons originating from nonreceptive females ( Table 2, three-way ANOVA, genotype/phase interaction p=0.01). The changes across the reproductive cycle are specific for the PR+ population, as they were not observed in PR-neurons (Table 2). Both in medial PR+ and PR-neurons, independent of the phase of the cycle, we observed a moderately lower number of primary dendrites per neuron, compared to the anterior and posterior neurons which yielded small yet significant differences across the AP axis (three-way ANOVA, location p=0.04).
Neurons in the VMHvl are characterized by having a long primary dendrite (LPD) that can be several fold longer than its short primary dendrites (SPD) (Calizo & Flanagan-cato, 2002). Therefore, we analyzed LPD and SPD separately (Table 2). While the length of the LPD of PR+ was unchanged across the reproductive cycle, PR-neurons exhibited longer LPDs in non-receptive females (three-way ANOVA genotype/phase interaction p=0.01). No overall significant differences were found in the SPD length of PR+ and PRneurons across the cycle and AP axis. Interestingly, the posterior neurons were shown to have shorter LPDs than the anterior and medial neurons regardless of the phase of the cycle or the genotype (three-way ANOVA, location p<0.01). No differences in the length of LPDs and SPDs were found between the PR+ and PR-populations.
In addition to the analysis of dendritic length, we sought to investigate whether the complexity of the dendritic trees was modified across the reproductive cycle. To do so, we used the Sholl method that measures the number of dendritic processes intersections as a function of the radial distance from the soma. No changes were observed in the maximum number of dendritic intersections of PR+ and PR-neurons across the phase of the cycle ( Figure 7A-F) nor across their location in the VMHvl, indicating that these neurons reach comparable maximum complexities in their dendritic trees. Nevertheless, the analysis of the Sholl profiles of these neurons revealed that the posterior PR+ neurons of non-receptive females had a significantly lower complexity of their dendritic tree (Fig.   7C, repeated measures ANOVA p<0.0001) compared to those from females in the receptive phase. In addition, the medial PR-neurons of non-receptive females had a higher complexity than their counterparts in receptive females (Fig. 7E, repeated measures ANOVA p<0.0001).
The increased complexity observed in the posterior PR+ neurons of females in the receptive phase was large enough to provide differences when tested with the same method specifically in the proximal branching (Fig. 7C, <250μm from the soma), while the increased complexity of the medial PR-neurons of non-receptive females were observed at more distal parts of the soma (240 to 340μm) and were large enough to yield multiple comparisons significant changes (Fig. 7E).
We also observed that medial PR+ neurons have lower dendritic branching than medial PR-neurons ( Figure 7B and 7E, repeated measures ANOVA p<0.0001), particularly in the distal parts from the soma (320 to 380μm). The posterior PR+ neurons have lower proximal dendritic branching than the posterior PR-neurons ( Figure 7C and 7F, repeated measures ANOVA p<0.01). These observations indicate that PR+ neurons have a different dendritic complexity proximal to the cell body compared to that of the PRpopulation, independent of the phase of the reproductive cycle.
To summarize, similarly to what we report for the electrophysiological properties, the structural properties of PR+ and PR-neurons are diverse, with some varying across the AP axis and the reproductive cycle.

Dendritic spine density largely differs between the PR+ and PR-populations.
Previous studies have shown that externally primed estrogen exerts effects on the dendritic spine density of VMHvl neurons that are different depending on the type of dendrite, specifically, an increase in spine density on SPDs of VMHvl neurons (Calizo & Flanagan-cato, 2000) and a decrease of the spine density on LPDs of VMHvl neurons expressing the receptor for estrogen (Calizo & Flanagan-cato, 2002). In order to determine if such modulation of the spine density of VMHvl neurons is present in the physiological range of sex hormone fluctuation, the dendritic spine density of PR+ and PR-neurons was analyzed depending not only on their location in the VMHvl, but also on the primary dendrite category across the reproductive cycle of naturally cycling females.
Although overall PR+ and PR-neurons have comparable spine densities on SPDs across the reproductive cycle ( Fig. 8A and C) and location, posterior VMHvl neurons in nonreceptive females have higher spine density on SPDs compared to VMHvl neurons in the receptive phase resulting in a significant interaction between the phase of the cycle and location within the VMHvl (three-way ANOVA, phase/location interaction p=0.02).
Furthermore, the PR+ population has significantly lower spine density on the SPDs than PR-neurons (three-way ANOVA, genotype p<0.0001).
Similar to what was observed for the spine densities of SPDs, for LPDs we observed that the posterior VMHvl neurons of receptive females present a decreased spine density compared to posterior neurons from females that were non-receptive, which in the case of LPDs the change is specific to PR+ neurons ( Fig. 8B and D, three-way ANOVA, phase/location/genotype interaction p<0.01, multiple comparison posterior PR+ Non-Receptive vs PR+ Receptive, p=0.01).
In addition, by comparing PR+ with PR-neurons, we observed that PR+ neurons have lower spine density on the LPDs in almost all groups, except for the anterior neurons of receptive females and posterior neurons of non-receptive females, in which the opposite is observed ( Figure 8B and D, three-way ANOVA, genotype p<0.01).
Altogether, we observed that neurons expressing progesterone receptor have an overall lower spine density when compared to their neighboring PR-neurons. Across the reproductive cycle, posterior PR+ neurons undergo more pronounced structural changes, exhibiting even lower spine density in the receptive phase.

Discussion
We combined whole-cell recordings with labeling of individual neurons to investigate the structural and electrophysiological properties of PR+ and PR-neurons along the AP axis of the VMHvl and across the reproductive cycle of naturally cycling female mice. We show that PR+ cells are distinct from PR-in three major aspects: first, due to structural properties (PR+ neurons have lower spine densities in general) (Fig. 9B); second, due to the great extent to which the properties of PR+ neurons vary across the AP axis (Fig. 9C); and third due to local changes in the properties of PR+ across the reproductive cycle (which are minimal for the PR-population) (Fig. 9D). Our results further support the existence of subdivisions in the VMHvl and its role in coordinating female behavior with the internal reproductive state.
The impact of the reproductive cycle on VMHvl function has been interrogated in vivo and in vitro, but most studies were performed in females whose ovaries were removed and then supplemented with estrogen and progesterone (Calizo & Flanagan-cato, 2000, 2002Griffin et al., 2010;Griffin & Flanagan-Cato, 2008;Millhouse, 1973;Rubin & Barfield, 1980, 1983b, 1983a The expression of hormone receptors, such as the progesterone receptor, are under the control of sex hormone levels as well (MacLusksy & McEwen, 1978), meaning that that the neuronal response is affected at the level of the receiver (receptor) and message (sex hormone). Persistent hormonal replacement of ovariectomized females leads to tumor development (Kordon et al., 1993) further suggesting that the hormonal treatment leads to undesired physiological effects. Finally, the sexual behavior that hormonally treated females exhibits differs from the behavior of naturally cycling females (Zipse et al., 2000), suggesting that the treatment fails to fully recapitulate the effects of natural sex hormones levels. To the best of our knowledge, this is the first study interrogating the intrinsic properties of neurons across the reproductive cycle of naturally cycling females and therefore the results observed reflect endogenous changes. However, we cannot claim that the changes we observe across the reproductive cycle are an effect of sex hormone levels because those were not directly manipulated. Manipulations with more naturalistic levels of sex hormones or local manipulations in the expression levels of sex hormone receptors should be employed to establish a direct causal link between hormonal levels, the structural and physiological properties of VMHvl neurons and the changes observed.
Our results indicate that PR+ cells are distinct from their neighbors, a difference that is probably established early in development, and later on by the fact that their intrinsic properties of this population can be directly affected by estrogen and progesterone, as Interestingly, even though anterior PR+ neurons had a comparable excitability profile across the reproductive cycle, the resting potential of neurons in the receptive phase was significantly hyperpolarized. This change in resting potential likely underlies a moderate increase in the threshold to fire in these neurons, which required more current to start generating action potentials. Even if this trend did not reach significant differences with our current statistical power, the visual difference between anterior non-receptive and receptive PR+ neurons can hardly be overlooked (Fig. 2C). In addition, the anterior PR+ neurons of receptive females do not show hyperpolarized-induced rebound firing. The fact that these neurons have a more hyperpolarized resting potential, thus more distant from the firing threshold, might explain the fact that the same magnitude of Ih current fails to evoke rebound firing in anterior PR+ at the receptive phase (Fig. 5B). We also report a robust increase of the firing rate under same current injections in medial VMHvl neurons obtained from females in the non-receptive phase compared to those in the receptive phase. These changes were not accompanied by changes in capacitance or membrane resistance across the reproductive cycle of medial neurons, thus it is unlikely that the changes in firing rate are caused by differences in the passive propagation of current into the neuron and instead points towards a modulation of voltage sensitive channels in the medial PR+ neurons, that may be mediated by progesterone (Scharfman & MacLusky, 2006). Overall, these results suggest that the anterior and medial VMHvl have the potential to generate more action potentials in non-receptive females, which is in seeming disagreement with the enhanced-male evoked responses that we previously observed, at the population level, in the VMHvl of sexually receptive female mice (Nomoto & Lima, 2015). However, the results of this study were obtained with extracellular recordings of non-identified neurons and therefore we do not know if they were PR+ or PR-. Also, in the present study, we investigated the intrinsic properties of VMHvl neurons and cannot make any claim regarding the driving input they receive (synaptic and/or neuromodulatory). The VMHvl receives indirect input from the vomeronasal organ, which includes some neurons whose activity is increased in response to male stimuli when females are sexually receptive (Dey et al., 2015). It is conceivable that the enhanced VMHvl activity that we previously observed reflects modifications at the sensory level.
It has been previously reported that hormonal treatment in ovariectomized rats reduced the amount and length of secondary dendrites of a non-defined VMHvl neuronal population (Griffin & Flanagan-cato, 2009). Even if in seeming contradiction with our findings, since we report the opposite effect in dendritic complexity and unchanged dendritic lengths, we would like to point out several possible reasons for such difference.
First, the changes we observed are specific to the posterior PR+ population (and absent in our sample of PR-neurons), therefore, sampling VMHvl neurons independently of their genotype could have masked a specific plasticity process in posterior PR+ neurons.
Second, in our dataset we observed that medial PR-neurons undergo a reduction in dendritic complexity that goes in line with previous reports. Thus, pooling samples from different levels of the VMHvl may mask changes, or alternatively sampling exclusively in a single AP region of the VMHvl could lead to findings that are limited to that level alone, and thus not generalizable to the whole VMHvl. Interestingly, posterior PR+ neurons did not only undergo changes at the level of their dendritic complexity, but also at the level of spine density, which was reduced in the receptive phase specifically in their LPDs. The fact that such dendritic spine changes are specific to LPDs may suggest that some pathway specific plasticity is happening during the reproductive cycle. Even though it has been previously hypothesized that SPDs may preferentially integrate local excitatory inputs, while LPDs may preferentially interact with long range inputs, inhibitory neurons of the shell and neuropeptides (Griffin & Flanagan-cato, 2011;Yamamoto et al., 2018), as of yet, no circuit mapping study has quantitatively explored what is the anatomical origin of the inputs. Thus, both the anatomical origin of those inputs and their possible relevance for the dendritic coding presented here needs to be further investigated. Our findings suggest a modification of the overall synaptic weights that the LPDs and SPDs will have in the dendritic integration for output generation, with a bias towards the inputs contacting SPDs.
It is worth noting that while the spine densities we observed are comparable to the ones reported in previous studies (Calizo & Flanagan-cato, 2002), the dendritic lengths we report both for PR+ and for PR-neurons are several fold longer than those previously reported (Calizo & Flanagan-cato, 2002). These changes may be explained by the differences in the technical approaches used to investigate neuronal structure. While previous reports have used fixed histological slices (100-150 microns) filled a posteriori with lucifer yellow, in the present study we have used acute slices for ex vivo electrophysiological recordings (300 microns), in which neurons were filled with biocytin during their electrophysiological monitoring. Thus, several key differences may explain the contradictory results: first, the thickness of the slice, which could allow us to track dendrites for longer extensions; second, the fact that neurons in acute slices for ex vivo recordings reseal their membranes, producing a visible thickening in the extremity of their cut dendrites, thus allowing us to recognize incomplete dendrites with our technical approach.
In summary, we have observed a surprising diversity of structural and physiological plasticity processes both along the AP axis of the VMHvl, and across the reproductive          Posterior neurons undergo structural modulations across the cycle, such as changes in proximal dendritic complexity and spine density.