Sexual dimorphism, estrous cycle and laterality determine the intrinsic and synaptic properties of medial amygdala neurons in rat Free

The posterodorsal subdivision of the medial nucleus of the amygdala (MePD) expresses high levels of androgen, progesterone, and both estrogen α- and β-receptors (Gréco et al., 1998; de Vries and Simerly, 2002; Frankiensztajn et al., 2018). This relates to the occurrence of a local sexual dimorphism and the effects of sex steroids, evidenced by structural and functional differences in prepubertal and adult male and female rats (Cooke and Woolley, 2005; Morris et al., 2008; Rasia-Filho et al., 2004; Zancan et al., 2018). The MePD modulates the display of reproductive behavior in both sexes (Adekunbi et al., 2018), mainly intromission and ejaculation in males (Coolen and Wood, 1999; Coolen et al., 1997; Rasia-Filho et al., 2012a), or the timely secretion of gonadotrophin-releasing hormone for ovulation (Polston et al., 2001; Simerly, 2004) and display of sexual behavior in females (Pfaus and Heeb, 1997; Polston et al., 2001).

The MePD neurons receive direct and indirect projections from the main and accessory olfactory bulbs (AOBs) (Pro-Sistiaga et al., 2007; Kang et al., 2009, 2011; Pereno et al., 2011). Then, they send out projections to the hypothalamic anteroventral periventricular nucleus (AVPV) (de Vries and Simerly, 2002), the medial preoptic area (MPOA), the ventral premammillary nucleus and, to a lesser extent, the ventrolateral part of the ventromedial nucleus, some of them via the bed nucleus of the stria terminalis (BNST) (Dong et al., 2001; Pardo-Bellver et al., 2012; Petrovich et al., 2001). Most outputs from the MePD are considered inhibitory (Choi et al., 2005; Swanson and Petrovich, 1998), and serve to disinhibit and orchestrate the complex hypothalamic function (de Castilhos et al., 2008; Rasia-Filho et al., 2012a). In fact, the medial amygdala (MeA) is largely composed of GABAergic neurons (Carney et al., 2010; Ruiz-Reig et al., 2018). In particular, 80% of the MePD neurons are GABAergic Lhx6 transcription factor-expressing cells in mice (Choi et al., 2005).

As other components of the extended amygdala, the mouse MeA develops from cell clusters coming from the caudoventral pallidal subdivision, the ventral pallium, the commissural preoptic area and the supraoptoparaventricular domain of the hypothalamus (Bupesh et al., 2011; García-López et al., 2008; Ruiz-Reig et al., 2018). Nevertheless, two types of subcortical multipolar neurons have been described in the rat MePD using the Golgi method. They were classified as bitufted or stellate cells, with two or more than two primary dendrites emerging from an ovoid to round cell body, respectively (Dall'Oglio et al., 2008; Rasia-Filho et al., 2004). This cellular homogeneity might explain the low variability of data obtained for the quantification of soma size, soma volume and dendritic spine density of Golgi-impregnated neurons from the MePD of male rats (Rasia-Filho et al., 2012a; Zancan et al., 2017, 2018).

On the other hand, males and females differ in several aspects, making the rat MePD an interesting area to study neural gonadal steroid actions, as occurs for the MPOA and other sex steroid-sensitive interconnected areas (Garelick and Swann, 2014). For example, neurons in males have higher cell body volume (Hermel et al., 2006b) and higher density of proximal dendritic spines than those in females in proestrus or estrus, but not diestrus (Rasia-Filho et al., 2004). Dendritic spines are multifunctional integrative postsynaptic units that receive inputs typically from excitatory synapses (Spruston et al., 2013; Yuste, 2013). Ultrastructural studies further revealed that proestrus females have the highest density of somatic spines (Zancan et al., 2015) and more inhibitory synapses on dendritic shafts in the right MePD than in the left MePD (Brusco et al., 2014). In addition, some aspects of the synaptic processing and plasticity in the MePD are lateralized in rats (Brusco et al., 2014; Hirsch et al., 2018). That is, the frequencies of spontaneous excitatory postsynaptic currents (sEPSCs) and miniature excitatory postsynaptic currents (mEPSCs) are significantly higher in the left MePD in prepubertal males compared to females (Cooke and Woolley, 2005). Males also have 80% more excitatory synapses per MePD neuron in the left hemisphere than females (Cooke and Woolley, 2005).

Therefore, the adult MePD serves as an interface for the integration of environmental chemosensory inputs, circulating levels of gonadal hormones and the activity of the social behavior neural network (Choi et al., 2005; Newman, 1999; Petrulis, 2013; Rasia-Filho et al., 2012b). A better understanding of information processing in the MePD requires detailed characterization of the functional properties and connectivity of its neurons. Although electrophysiological properties of MeA neurons have been described from different divisions in mice (Bian, 2013; Bian et al., 2008; Keshavarzi et al., 2014), no such electrophysiological data are currently available for MePD neurons in adult males and along the estrous cycle of female rats.

Here, we characterized two different types of coexisting neurons and demonstrated the important influence of sexual dimorphism, estrus cycle and lateralization on their intrinsic properties and synaptic inputs in the MePD of adult male and cycling female rats.

First, we evaluated the intrinsic membrane properties and morphology of MePD neurons and investigated whether sexual dimorphism and hemispheric lateralization could influence these cell features.

The intrinsic membrane properties were analyzed from MePD neurons (n=244) recorded randomly from each hemisphere in males and in females along the three phases of the estrous cycle (Fig. 1A–D). Two coexisting subpopulations of MePD neurons were identified based on the distinct responses to both depolarizing and hyperpolarizing current injections and different firing patterns. These neurons were named as Class I (43% of recording cells) or Class II (57% of recording cells) neurons. The electrophysiological properties of these two classes of neurons are summarized in Table 1 and described below.

MePD Class I neurons characteristically exhibited a reduction in firing rate to a constant stimulus intensity (Fig. 2A). Data were obtained from the MePD of males (21 neurons) and females in diestrus (23 neurons), proestrus (33 neurons) or estrus (28 neurons).

Class I neurons typically had an average resting membrane potential (V_rest) of −50.78±0.68 mV, input resistance (R_in) of 419.48±8.24 MΩ, membrane time constant (τ) of 23.30±0.8 ms, membrane capacitance (C_m) of 59.56±1.06 pF and spike frequency adaptation (SFA) ratio of 5.69±0.51. No statistically significant differences were found for these parameters between sex and cerebral hemisphere (F_3,95=0.93, P=0.42; F_3,95=1.09, P=0.35; F_3,95=0.73, P=0.53; F_3,95=0.48, P=0.69; F_3,95=0.56, P=0.64, respectively; Table 1). These neurons also had a small hyperpolarization-activated voltage hyperpolarization-activated voltage (SAG) (Table 1) and ∼49% of these cells displayed rebound spikes after the hyperpolarizing current steps (Fig. 2A). Furthermore, under depolarizing current steps from 20 pA to 100 pA, these neurons exhibited irregular spike trains with frequency adaptation (Fig. 2A). The firing frequency in left MePD cells at 100 pA/1000 ms depolarizing current step was higher in males than in females in diestrus, proestrus or estrus (F_3,95=4.65, P<0.05; Bonferroni post hoc test, P<0.05 in all cases; Table 1). The same pattern was observed at 60 pA/1000 ms and 80 pA/1000 ms depolarizing current steps and reflects a true difference in excitability of these cells modulated by sex. The action potential (AP) of Class I neurons had an average slow afterhyperpolarization (sAHP) time course and amplitude of 17.4±0.01 ms and 12.48±0.71 mV, respectively, and fast afterhyperpolarization (fAHP) time course and amplitude of 3.2±0.01 ms and 21.15±1.07 mV, respectively. Values for the sAHP time course, sAHP amplitude, fAHP time course and fAHP amplitude parameters were not different between sex and cerebral hemisphere (F_3,95=0.624, P=0.60; F_3,95=0.41, P=0.74; F_3,95=1.04, P=0.37 and F_3,95=1.05, P=0.37, respectively; Table 1). The AP threshold, AP amplitude and AP half-width (HW) were 29.11±2.34 mV, 69.27±1.69 mV and 2.79±0.07 ms, respectively (Table 1).

MePD Class II neurons fired – at most – one or two APs, followed by full adaptation in response to a constant stimulus intensity with no SFA ratio (Fig. 2B). Data were obtained from the MePD of males (32 neurons) and females in diestrus (36 neurons), proestrus (31 neurons) or estrus (40 neurons).

Class II neurons typically had an average V_rest of −49.83±0.46 mV, R_in of 423.46±5.70 MΩ, τ of 23.87±0.88 ms and C_m of 59.36±1.66 pF. No statistically significant differences were observed for these parameters between sex and cerebral hemisphere (F_3,131=0.38, P=0.76; F_3,131=0.78, P=0.50; F_3,131=0.55, P=0.64; F_3,131=2.01, P=0.11, respectively; Table 1). Class II neurons had small hyperpolarization-activated voltage SAG (Table 1). The rebound spikes after the hyperpolarizing current step were observed in ∼83% of the cells (Fig. 2B). Under depolarizing current steps from 20 pA to 100 pA, neurons exhibited one or two APs with full spike frequency adaptation (Fig. 2B).

The AP of Class II neurons had a significantly longer sAHP time course compared to that of Class I neurons (t₁₃₈=4.35, P<0.05; Table 1 and Fig. 2B′). However, the average fAHP time course (4.3±0.01 ms) and amplitude (20.56±0.87 mV) were similar to those of Class I neurons (t₁₃₈=0.91, P>0.05 and t₁₃₈=1.06, P>0.05, respectively; Table 1). Finally, the AP threshold, AP amplitude and AP HW were similar to those of Class I neurons (Table 1). Thus, the sAHP and AP frequency adaptation are the main parameters that differentiate the two classes of neurons registered.

Thus, we described the electrophysiological features of the two coexisting cell subpopulations in the MePD. Only Class I neurons had intrinsic properties influenced by sex differences and hemispheric lateralization, leading to an increased firing frequency in left MePD neurons in males compared to cycling females.

The biocytin-filled neurons in males (11 neurons), and diestrus (11 neurons), proestrus (8 neurons) or estrus females (8 neurons) had clear morphological characteristics of two different MePD cell subpopulations. Based on the cell body shape and the number of primary dendrites, they were classified as bitufted or stellate cells, consistent with previous Golgi studies in adult Wistar rats (Dall'Oglio et al., 2008; de Castilhos et al., 2008; Marcuzzo et al., 2007; Rasia-Filho et al., 1999; Rasia-Filho et al., 2004).

Based on the intrinsic membrane properties, 18 neurons visualized under confocal microscopy were classified as Class I and 20 as Class II. Interestingly, most Class I neurons were bitufted cells (16 of 18 neurons; Fig. 2E), with two primary dendrites arising from opposite poles of a fusiform cell body (Fig. 2C). Most Class II neurons were stellate cells (17 of 20 neurons; Fig. 2F), with three or more primary dendrites and variable cell body shapes (Fig. 2D). Both cell types had dendritic spines with various shapes and homogeneous distribution from proximal to distal branches (Fig. 2C′,D′). Morphological features of the two cell subpopulations described here were not affected by brain hemisphere, i.e. Class I and II neurons were observed equally in both MePD sides.

Subsequently, we further investigated whether sexual dimorphism and hemispheric lateralization could influence the excitatory and inhibitory synaptic transmission onto MePD neurons.

Neurons were randomly recorded, independent of the neuronal subpopulation, from the right and left MePD, respectively, in males (22 and 28 neurons) and females in diestrus (24 and 22 neurons), proestrus (27 and 31 neurons) or estrus (22 and 27 neurons). Fig. 3A,B show the location and distribution of the recorded neurons in the MePD of all studied groups.

Representative traces of sEPSCs and mEPSCs for males and cycling females are shown in Fig. 3C–F. Notably, the sEPSC amplitude in MePD cells in the right and left hemispheres was significantly smaller in females in proestrus and estrus than in males and females in diestrus (F_3,98=58.57, P<0.05; Bonferroni post hoc test, P<0.001 in all cases; Table 2). Cumulative probability plots and Kolmogorov–Smirnov analysis further confirmed the reduction in the sEPSC amplitude in the MePD neurons from females in both proestrus and estrus (P<0.001; Fig. 3G,I). In addition, the sEPSC frequency in neurons of the left MePD was higher in males than in females in the three phases of the estrous cycle (F_3,98=6.75, P<0.05; Bonferroni post hoc test, P<0.05 in all cases; Table 2). Cumulative probability plots and Kolmogorov–Smirnov analysis further confirmed the higher sEPSC frequency in neurons in the left MePD in males than in cycling females (P<0.001; Fig. 3H,J). No statistically significant difference in the sEPSC frequency in right MePD neurons was found between the groups (F_3,98=0.23, P=0.87).

In contrast to the sEPSC amplitude and frequency, no statistically significant differences between groups or brain hemisphere were found for the mEPSC amplitude (F_3,98=1.14, P=0.06) or frequency (F_3,98=1.11, P=0.13) in the MePD neurons (Table 2). Cumulative probability plots and Kolmogorov–Smirnov analysis further confirmed these results (P>0.05 in all cases; Fig. 3K–N). The comparative plots of sEPSC and mEPSC amplitude and frequency are shown in Fig. 3O–R.

The sEPSC decay time constant in MePD cells from males and females in diestrus was longer than that in MePD cells from females in proestrus or estrus (F_3,98=66.96, P<0.05; Bonferroni post hoc test, P=0.004 and P=0.01, respectively), with no effect of hemispheric lateralization (F_3,98=3.48, P=0.06) (Table 2 and Fig. 4A,I). Furthermore, the sEPSC 10–90% rise time in the left MePD neurons was significantly longer in males than in females along the three phases of the estrous cycle (F_3,98=9.51, P<0.05; Bonferroni post hoc test, P<0.05 in all cases; Table 2 and Fig. 4B,I). No such difference was found in the right MePD (F_3,98=1.08, P=0.35). Different from what was seen for the sEPSC, no statistically significant differences between groups or brain hemisphere were found for the mEPSC decay time constant (F_3,98=1.23, P=0.10) or 10–90% rise time (F_3,98=1.55, P=0.20) in the MePD neurons (Table 2 and Fig. 4C,D,J).

Altogether, these data indicate that MePD neurons receive more excitatory synaptic input in males than in females in diestrus, proestrus and estrus, depending on the brain hemisphere; and females in diestrus receive a higher magnitude of excitatory synaptic inputs compared to females in proestrus or estrus.

Neurons were recorded randomly, independent of the neuronal subpopulation, from the right and left MePD, respectively, in males (21 and 20 neurons) and females in diestrus (23 and 22 neurons), proestrus (24 and 26 neurons) or estrus (22 and 24 neurons). Fig. 5A,B show the location and distribution of the recorded neurons in the MePD of all studied groups.

First, we compared the kinetic properties of spontaneous inhibitory postsynaptic currents (sIPSCs) and miniature inhibitory postsynaptic currents (mIPSCs) in MePD neurons. The sIPSC decay time constant (F_3,83=1.10, P=0.35) and 10–90% rise time (F_3,83=0.50, P=0.68; Table 3 and Fig. 4E,F,K) in MePD cells were similar in all groups. However, both the mIPSC decay time constant and 10–90% rise time in the MePD cells were shorter in females in proestrus than in males and females in diestrus or estrus (F_3,83=16.18, P<0.05 and F_3,83=13.81, P<0.05, respectively; Bonferroni post hoc test, P<0.001 in all cases). No statistically significant effect of hemispheric lateralization was observed (F_3,83=0.14, P=0.70 and F_3,83=2.98, P=0.06, respectively) (Table 3 and Fig. 4G,H,L).

Representative traces of sIPSCs and mIPSCs for males and cycling females are shown in Fig. 5C–F. The sIPSC amplitude in the MePD cells was significantly smaller in males and in females in diestrus and estrus than in females in proestrus (F_3,83=128.24, P<0.05; Bonferroni post hoc test P<0.001 in all cases), with no statistically significant effect of hemispheric lateralization (F_3,83=2.70, P=0.10; Table 3). Cumulative probability plots and Kolmogorov–Smirnov analysis further confirmed these results (P<0.001; Fig. 5G,I). In addition, the sIPSC frequency in the MePD neurons from the right hemisphere was significantly lower in males compared to females in the three phases of the estrous cycle (F_3,83=6.61, P<0.05; Bonferroni post hoc test, P<0.001 in all cases; Table 3). In contrast to the right MePD, the average sIPSC frequency in the left MePD neurons was lower in females in estrus and diestrus than in males and females in proestrus (F_3,83=14.30, P<0.05; Bonferroni post hoc test, P<0.001 in all cases; Table 3). Cumulative probability plots and Kolmogorov–Smirnov analysis further confirmed these data (P<0.001; Fig. 5H,J).

No statistically significant differences in the mIPSC amplitude (F_3,83=1.86, P=0.06; Table 3) or frequency (F_3,83=1.51, P=0.06; Table 3) in the MePD neurons were found between groups. Cumulative probability plots and Kolmogorov–Smirnov analysis further confirmed these results (P>0.05; Fig. 5K–N). The comparative plots of sIPSC and mIPSC amplitude and frequency are shown in Fig. 5O–R.

Taken together, these data indicate that the MePD neurons receive less inhibitory synaptic input in males than in females, depending on the brain hemisphere, whereas females in proestrus receive a higher magnitude of inhibitory input compared to males and females in diestrus or estrus.

Our results demonstrate the presence of two coexisting subpopulations of spiny neurons in the MePD of adult male and female Wistar rats, with two main differences in spike frequency patterns during depolarizing current. Both MePD neuronal subpopulations had electrophysiological features affected by sex and estrous cycle. The Class I neurons were also affected by hemispheric lateralization, as evidenced by the higher firing frequency of left MePD neurons in males than in cycling females. Furthermore, excitatory and inhibitory inputs to MePD cells were modulated by sex, estrous cycle and hemispheric lateralization. There was an overall increment in the excitatory input to MePD neurons of males compared to cycling females and, between estrous phases, in diestrus compared to proestrus or estrus. On the other hand, there was an increase in the inhibitory input to MePD neurons of females in proestrus compared to males and females in diestrus or estrus. These results depend on the hemisphere, with some findings observed in both sides or in just one of them.

These neurons were identified as Class I neurons, predominantly bitufted-shaped cells exhibiting irregular spikes with frequency adaptation, and Class II neurons, predominantly stellate-shaped cells exhibiting full spike frequency adaptation. Class II neurons had a longer sAHP time course than Class I neurons and, consequently, tended to fire one or two APs and present full spike frequency adaptation. The spike frequency adaptation could be the reflex of the intrinsic neuronal membrane properties or a result of synaptic input to the cell (Peron and Gabbiani, 2009). Since the sAHP is one of the membrane properties that can influence spike frequency and is considered an index of neuronal excitability (Sah, 1996), a potential mechanism responsible for sAHP differences could be related to variability in expression and distribution of functional sAHP channels in these two classes of cells. However, there is no evidence to support that the functional sAHP channels are uniformly distributed over the membrane of MePD neurons. In addition, the sAHP can shape temporal integration of synaptic inputs by shunting excitatory postsynaptic potentials (Sah and Bekkers, 1996). This shunting mechanism could be attributed to the Ca²⁺ influx through the NMDA receptors and voltage-gated Ca²⁺ channels that would activate sAHP channels, hyperpolarizing the membrane potential that facilitates the voltage-dependent Mg²⁺ re-block of the NMDA receptors, consequently reducing temporal summation during a period of intense synaptic events (Wu et al., 2004). Therefore, we cannot rule out the influence of different patterns of synaptic stimuli on these neuronal responses. Indeed, we observed that Class I neurons from the left MePD exhibited a higher firing rate in males than in females. Interestingly, the frequency of sEPSCs to neurons from the left MePD was also higher in males than in females. Our results could contribute to future studies to investigate whether the increased excitatory synaptic tone onto the MePD of males could lead to the significant increase in firing rate in Class I neurons, since the passive membrane properties of these cells were similar for both sexes (Table 1). In addition, both classes of neurons were found in males and females along the three phases of the estrous cycle. These data suggest that the identification of two subpopulations of MePD neurons represent a constitutive feature of the MePD rather than just a finding related to a functional characteristic restricted to one sex.

Both classes of MePD neurons had SAG and rebound depolarization following hyperpolarizing pulses. These two features were reported previously for the MeA neurons of mice (Bian, 2013; Carney et al., 2010; Keshavarzi et al., 2014). Class I and Class II neurons typically had high R_in and SAG, suggestive of the presence of a hyperpolarization-activated cation current (I_h) current in these cells (Bian et al., 2008; Keshavarzi et al., 2014). In addition, the rebound depolarization can provide cells the capacity to regulate their firing pattern by generating spike bursts, which increase the rate and duration of spike output, and by regulating the first spike latency and precision (Molineux et al., 2008; Sangrey and Jaeger, 2010). The biophysical mechanisms of spike generation enable individual neurons to encode different inputs into distinct spike patterns (Kepecs and Lisman, 2003), shaping the information transferred between pre- and postsynaptic neurons. A period of inhibitory input, at one or multiple sites, results in a postsynaptic firing that most often occurs through a rebound depolarization (Engbers et al., 2011; Pedroarena, 2010). The neuronal firing pattern depends on the magnitude of the preceding period of inhibition (Sangrey and Jaeger, 2010). The ion channel mechanisms underlying the neural coding through rebound responses is not fully identified and the role of rebound depolarization in the MePD neurons in rodents remains unclear at this point.

Some membrane properties of MePD neurons of rats differ from those of mice. First, most Class I and II MePD neurons exhibited a V_rest near −50 mV and a R_in of ∼420 MΩ in both male and female Wistar rats. On the other hand, MePD neurons from male mice exhibited a V_rest of ∼70 mV and an R_in of ∼700 MΩ or 260 MΩ when recorded in the same experimental conditions (Bian, 2013). These differences could be attributed to neurotransmitter release, receptor expression or intrinsic membrane properties of different neuronal subpopulations between species (Pardo-Bellver et al., 2012). Second, we did not find cells with characteristics of GABAergic interneurons in the MePD of rats, whereas a previous study on MeA neurons of GAD67-GFP mice demonstrated the presence of this neuronal type in the most medial part of the MeA (Bian, 2013). Here, we recorded cells from the MePD, avoiding the ultimate borders of this subnucleus (de Olmos et al., 2004; Nishizuka and Arai, 1983a,b). Third, the two subpopulations of adult rat MePD neurons were multipolar spiny cells, confirming previous morphological data (Cooke et al., 2007; Dall'Oglio et al., 2008; Rasia-Filho et al., 2004; Zancan et al., 2018). Different from the MeA of mice, we have not observed pyramidal (Niimi et al., 2012) or pyramidal-like (Bian, 2013) neurons in the MePD of rats (Marcuzzo et al., 2007). Our sampled biocytin-filled neurons did not have the morphological and electrophysiological features of pyramidal cells (DeFelipe and Fariñas, 1992; Le Bé et al., 2007; McCormick et al., 1985). Most Class I neurons were predominantly bitufted-shaped cells, whereas Class II neurons were predominantly stellated-shaped cells.

Nevertheless, previous reports indicate that rat and mice MePD neurons can be inhibitory cells (Choi et al., 2005; Swanson and Petrovich, 1998) expressing calbindin, nitric oxide synthase, Forkhead box transcription factor or somatostatin (Mugnaini and Oertel, 1985; Carney et al., 2010; Ruiz-Reig et al., 2018). These neurons have distinct characteristics from neurons of most of the posteroventral subdivision of the MeA, which express vGlut2 (also known as Slc17a6) and are glutamatergic projection neurons (Bian et al., 2008; Keshavarzi et al., 2015; Ruiz-Reig et al., 2018). Although it is currently not clear whether Class I or Class II MePD neurons are used for a specific functional role in the rat social behavior network, it is conceivable to propose that their dendritic structure can provide different biophysical and integrative properties for synaptic inputs (Dall'Oglio et al., 2008).

During adulthood, males and females in diestrus have more proximal dendritic spines than females in proestrus and estrus (Rasia-Filho et al., 2004). Interestingly, we found that, in both hemispheres, the MePD neurons of males and females in diestrus receive a higher magnitude of excitatory synaptic inputs compared to neurons of proestrus and estrus females. However, the density of dendritic spines decreases at the peak of circulating estrogen and progesterone during proestrus (Rasia-Filho et al., 2004). Estrogen reduces the axonal excitability that reaches the MeA of female rats (Yoshida et al., 1994). As spines are mostly postsynaptic sites, it is highly likely that part of the incoming information to the MePD converges and acts upon dendritic spines in a quantitatively different manner depending on the sex and the hormonal status of the animal (Brusco et al., 2014; Rasia-Filho et al., 2004).

Furthermore, sex differences in excitatory inputs to MePD neurons could be due to distinct sources and number of afferences to this subnucleus and/or direct effects of different levels of androgen and estradiol/progesterone (Cooke et al., 2007; Rasia-Filho et al., 2004). Accordingly, glutamatergic inputs to MePD neurons from AOB mitral cells (Quaglino et al., 1999; Scalia and Winans, 1975) and from cortical amygdala (Mugnaini and Oertel, 1985) are higher in males than in females (Pérez-Laso et al., 1997; Vinader-Caerols et al., 1998). We also demonstrated that sexual dimorphism significantly affects the sEPSC kinetics of MePD neurons. Sex steroid-mediated signaling influences the synaptic transmission strength and kinetics (Hansberg-Pastor et al., 2015; Sellers et al., 2015), likely modulating the afferent information to the MePD neurons (Ervin et al., 2015). Different locations and subtle alterations in the glutamate receptor subunits or in the kinetics of the glutamate transporters might change sEPSC kinetics (Wall et al., 2002). Dendritic spines in the MePD are immunoreactive to glutamate α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and N-methyl-D-aspartate receptor (NMDA) receptors (Dalpian et al., 2015). Thus, it likely that dendritic spines contribute to the differences in excitatory inputs recorded in the adult male MePD. These data reinforce the plastic changes in dendritic spines and synaptic processes in the MePD after puberty in rats (Cooke and Woolley, 2009; Zancan et al., 2018).

On the other hand, the mEPSCs in MePD neurons of prepubertal rats have a higher frequency on the left, but not the right, hemisphere in males compared to females (Cooke and Woolley, 2005). We did not find sex or hemispheric lateralization effects on mEPSC properties during adulthood. Since spontaneous synaptic currents represent the synaptic function in neural network and miniature synaptic currents reflect quantal transmitter release (Tian et al., 2012), our results suggest the existence of differences at a circuit level rather than changes to receptors or release machinery at excitatory synapses (Yamasaki et al., 2006). In fact, there is increasing evidence that spontaneous and miniature neurotransmitter releases are controlled by different intrapresynaptic mechanisms (Wakita et al., 2015). Therefore, local excitatory neurotransmitter release or synaptic strength on each site seems to be unaltered in the adult MePD.

According to our findings, the stronger excitatory input to MePD neurons in males and females in diestrus could modulate the neural coding and information output, but that will depend on the spatiotemporal strength of the inhibitory input onto MePD neurons. We found that the MePD neurons in the right hemisphere exhibited a lower sIPSC frequency in males than in females in diestrus and estrus, and that the inhibition prevails during proestrus. In addition, in proestrus, the MePD neurons receive more inhibitory inputs at the same time that the magnitude of excitatory inputs reduces. The sIPSC amplitude to MePD neurons was significantly smaller in males and in females in diestrus and estrus than in females in proestrus in both hemispheres. Different from sIPSCs, we did not find sex or hemispheric lateralization effects on mIPSC amplitude or frequency in the MePD neurons, again suggesting differences at the circuit level, similar to excitatory inputs. However, faster decay and rise time of the mIPSCs were observed in neurons in the MePD in females in proestrus than in males or females in diestrus or estrus. These changes in kinetics suggest an increase in inhibitory synapses close to the soma or a loss of more distal inhibitory synapses (Kobayashi and Buckmaster, 2003; Shao and Dudek, 2005). The MePD of both hemispheres in adult male and female rats receives inhibitory inputs at multisynaptic contacts (Brusco et al., 2014). Inhibitory synapses, when targeting dendrites, tend to generate inhibitory postsynaptic currents (IPSCs) with distinct kinetics and dynamics from perisomatic IPSCs. The dendritic IPSCs usually have a slower decay and rise time than the perisomatic IPSCs (Ropert et al., 1990). The faster kinetics of mIPSCs in MePD neurons of females in proestrus found here may also result from a shift in the GABA_A receptor subunits, changes in GABA transporter kinetics (Calcagnotto et al., 2002) or synaptic compartmentalization along the neuronal membrane (Kobayashi and Buckmaster, 2003). Accordingly, proestrus females have more somatic spines (Zancan et al., 2015), assumed to be inhibitory (Kubota et al., 2016), and more direct dendritic shaft synapses with symmetric GABA-immunoreactive terminals in the MePD (Brusco et al., 2014) than do males or diestrus and estrus females.

Previous studies demonstrated that the morphology and function of the MePD is lateralized (Arpini et al., 2010; Brusco et al., 2014). Here, we found that the hemispheric lateralization influenced the excitatory inputs, with no effect on the inhibitory inputs. That is, the left MePD neurons had a higher frequency of sEPSCs in adult males compared to cycling females as mentioned above. One possible reason is that the left MeA is specialized for chemosensory and/or steroid negative feedback regulation of neuroendocrine secretion (Cooke and Woolley, 2005).

Our electrophysiological results provide additional data to investigate the role of excitatory and inhibitory inputs to the right and left MePD of sexually mature, adult rats. The association between neuronal morphology, intrinsic properties, gonadal hormone modulation of synaptic processing, and connectivity will be relevant to understand the complex spatiotemporal synaptic elaboration within the circuitries that modulate chemosensory processing, emotion and behavior display in male and female rats. Interestingly, sex-related hemispheric lateralization of human amygdala volume and function also exists in humans (Cahill, 2006; Kim et al., 2012). Human MeA neurons also exhibit complex dendritic spines and evidence for an elaborated connectivity, including multisynaptic spines contacting both excitatory and inhibitory terminals (Zancan et al., 2015).

In males, the MePD activation during mating most likely disinhibits brain areas involved in sexual behavior (Coolen et al., 1997; Rasia-Filho et al., 2012a; Petrulis, 2013). Here, we demonstrated that the two classes of MePD inhibitory neurons in males received predominant excitatory inputs. Previous studies have shown that MePD neurons project to the BNST, the main efferent pathway in males (de Olmos et al., 2004; Pardo-Bellver et al., 2012), containing mostly GABAergic cells (Poulin et al., 2009; Yamamoto et al., 2018) and projecting to the hypothalamus (Dong et al., 2001). Therefore, the MePD neurons could disinhibit the hypothalamic cells via an indirect pathway, by inhibiting BNST GABAergic cells.

In females, a dynamic elaboration of synaptic processing occurs in the MePD along the estrous cycle (Oro et al., 1988; Lehmann and Erskine, 2005; Zancan et al., 2015). Since our results demonstrated that the inhibitory inputs are predominant onto inhibitory MePD neurons in the proestrus phase, we hypothesized that the morphological and functional organization of MePD neurons result in disinhibition of hypothalamic circuits via a direct pathway, leading to neuroendocrine secretion, ovulation and sexual receptivity (Carney et al., 2010; Simerly, 2004). On the other hand, during diestrus phase, the resultant excitation of inhibitory MePD neurons could induce the inhibition of hypothalamic neurons. Overall, we hypothesized that the efferent information of the MePD is sent to the hypothalamus via distinct pathways in a sex-dependent manner, probably via an indirect pathway in males and a direct pathway in females.

In conclusion, the adult rat shows sexual dimorphism, estrous cycle and hemispheric lateralization features that determine the synaptic inputs and electrophysiological features of two coexisting subpopulations of bitufted and stellate neurons in the MePD of males and cycling females. Our data indicate the influence of sexual dimorphism, estrous cycle and hemispheric lateralization at cellular and synaptic levels in the adult rat MePD.

All efforts were made to minimize the number of animals studied and their suffering. All animal procedures were conducted in accordance with international laws and guidelines for the care and use of laboratory animals (European Communities Council Directive of 24 November 1986, 86/609/EEC; NIH Publication No. 85–23, revised 1985, USA) and were approved by local Animal Ethics Committee of Universidade Federal do Rio Grande do Sul, Brazil (CEUA-UFRGS protocol no. 28885).

Adult Wistar rats (3 months old) were sourced from the Central Institute for Experimental Animals (CREAL-UFRGS). Rats were housed in groups (up to three animals per standard cage of 41×34×16 cm) under standard laboratory conditions with free access to food and water, at room temperature (RT) (∼21°C) and under a light/dark cycle of 12 h (lights on at 06:00). Virgin male and female animals were used. Daily vaginal smears were obtained during the morning over 2 consecutive weeks prior to the beginning of the experiment. The regularity of the estrous cycle was determined according to cytological criteria (Rasia-Filho et al., 2004). Only normally cycling females were included and studied in the diestrus, proestrus and estrus phases (Zancan et al., 2015).

Acute brain slices were prepared for physiological recordings as described previously (Cobos et al., 2005; Vendramin Pasquetti et al., 2017). Males and females were anesthetized with ketamine and xylazine (80 mg/kg and 10 mg/kg intraperitoneally, respectively) and were quickly decapitated. The brains were rapidly removed and placed into ice-cold oxygenated slicing artificial cerebrospinal fluid (sACSF) containing 220 mM sucrose, 3 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 26 mM NaHCO₃, 1 mM CaCl₂ and 10 mM dextrose (295–305 mOsm, pH 7.4). The MePD was identified in coronal slices from 3.0 mm to 3.60 mm posterior to the bregma, taking the optic tract and the stria terminalis as anatomical references (Paxinos and Watson, 1998). Serial sections (300 µm) were cut using a vibrating microtome (VTS-1000; Leica, Germany) at 4°C oxygenated sACSF. The slices from the left and right hemispheres were separated and immediately transferred to a chamber to remain immersed in oxygenated normal-recording artificial cerebrospinal fluid (nACSF) consisting of 124 mM NaCl, 3 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 26 mM NaHCO₃, 1 mM CaCl₂ and 10 mM dextrose (295–305 mOsm, pH 7.4) (Cobos et al., 2005; Vendramin Pasquetti et al., 2017). The slices were bubbled with 95% O₂/5% CO₂ and maintained at 37°C for 45 min and, afterwards, at RT.

Each slice containing the MePD of either the left or right hemisphere was moved to a recording chamber perfused at RT, in oxygenated nACSF at a rate of 3 ml/min. Whole-cell recordings were obtained from visually identified neurons in the MePD, using an infrared differential interference contrast (IR-DIC) video-microscopy system with a microscope (BX51WI, Olympus). To be considered, neurons must be located within the MePD, from its intermediate to lateral columns, avoiding ultimate medial, lateral and inferior borders (de Olmos et al., 2004). The medial ‘molecular layer’, which is a cell-sparse region close to the optic tract, mainly formed by axons from the accessory olfactory pathway (Pro-Sistiaga et al., 2007; Scalia and Winans, 1975), was excluded from the recordings.

Patch electrodes (3–7 MΩ) were pulled from borosilicate glass capillary tubing (TW150-3, World Precision Instruments) using a computer-controlled micropipette puller (P-1000, Sutter Instrument) and fire polished.

Synaptic currents onto MePD neurons were recorded in a whole-cell voltage clamp, and the series resistance and whole-cell capacitance were continually monitored and compensated throughout the course of the experiment. Recordings were rejected if values of whole-cell access resistance changed by >25% (or exceeded 20 MΩ). The sEPSCs were recorded at −70 mV with an internal solution containing 135 mM CsCl₂, 10 mM NaCl, 2 mM MgCl₂, 10 mM HEPES, 10 mM EGTA, 2 mM Na₂ATP and 0.2 mM Na₂GTP (285–290 mOsm, pH 7.2). To isolate glutamatergic currents, slices were perfused with nACSF containing (–)-bicuculline methiodide (10 µM; Sigma-Aldrich, 14343), a GABA_A receptor antagonist. The sIPSCs were recorded at 0 mV with an internal solution containing 117.5 mM Cs-Gluconate, 11 mM CsCl₂, 1 mM MgCl₂, 10 mM HEPES, 11 mM EGTA, 2 mM Na₂ATP, 0.5 mM Na₂GTP and 1.25 mM QX-314 (285–290 mOsm, pH 7.2). To isolate GABAergic currents, slices were further perfused with nACSF containing 6,7-dinitroquinoxaline-2,3-dione (DNQX; 20 µM; Sigma-Aldrich, D0540) and D-(–)-2-amino-5-phosphonovaleric acid (D-AP5; 50 µM; Sigma-Aldrich, A5282), the non-NMDA and NMDA receptor antagonists, respectively. Both mEPSCs and mIPSCs were recorded in nACSF containing the appropriate receptor antagonist plus tetrodotoxin (TTX; 1 µM; Cayman Chemical, 14964). Intrinsic membrane properties were recorded in whole-cell current-clamp mode in other sets of cells from males and cycling females with an intracellular patch pipette solution containing 120 mM K-gluconate, 10 mM KCl, 1 mM MgCl₂, 0.025 mM CaCl₂, 10 mM HEPES, 0.2 mM EGTA, 2 mM Na₂ATP, 0.2 mM Na₂GTP and 0.1 mM biocytin (0.25%; Sigma-Aldrich, B4261) (285–290 mOsm, pH 7.2). Firing properties were obtained ∼3 min after obtaining whole-cell configuration. Membrane potential was adjusted to −65 mV by passing a constant holding current, and 500 ms and 1000 ms current step injections were delivered in 20 pA increments from −100 to +100 pA.

Voltage and current were recorded with a Multiclamp 700B amplifier (Molecular Devices) and monitored with an oscilloscope and pClamp 10.3 software (Molecular Devices). Whole-cell voltage-clamp data were low-pass filtered at 1 kHz and digitally sampled at 20 kHz using a Digidata 1440A (Molecular Devices).

Neurons from the left and right MePD of male and cycling females were visualized using an IR-DIC video-microscopy system and filled with biocytin during current-clamp recording to correlate the cell morphology, well described previously (Arpini et al., 2010; Brusco et al., 2010; Brusco et al., 2014; Dall'Oglio et al., 2008; de Castilhos et al., 2008; Hermel et al., 2006a; Hermel et al., 2006b; Rasia-Filho et al., 2004; Rasia-Filho et al., 2012b), with the intrinsic properties. At the end of each recording session, each slice with biocytin-filled neurons was fixed overnight in 4% paraformaldehyde at 4°C. After 3×10 min washes with 0.1 M phosphate buffer solution (PBS; pH 7.4), the slices were blocked in PBS containing 2% bovine serum albumin and 0.1% Triton X-100 for 1 h at RT. Afterwards, the slices were incubated with streptavidin-conjugated AlexaFluor-488 (1:500; Thermo Fisher Scientific, S11223) (Moraes et al., 2013) for 4 h at RT. Each brain slice was mounted on slides with Fluoromount (Sigma-Aldrich, F4680) and a coverslip, and kept in the dark at 4–8°C. About 24 h later, neurons were imaged using a confocal laser-scanning microscope (FV1000 Spectral, Olympus) with a planapochromatic 63×/1.4 NA water-immersion objective lens. Spectral detectors were adjusted to capture emission from lasers at a wavelength of 488 nm. The z-stack acquisition was performed at 0.1 μm using a resolution of 1024×1024 pixels with 4× zooming (Brusco et al., 2010). For each slice, the biocytin-positive neuron was identified and z projections were collected sequentially to reconstruct and study the cellular morphology (da Silva et al., 2015). These images were three-dimensionally reconstructed using Photoshop CS3 software (Adobe Systems).

At least 100 individual events were manually selected for each cell to analyze the synaptic currents (EPSCs or IPSCs) (Cobos et al., 2005; Cubelos et al., 2010; Vendramin Pasquetti et al., 2017). The EPSC and IPSC frequency, amplitude, 10–90% rise time and decay time constant were measured using Mini Analysis 6.0.7 (Synaptosoft Software).

Neurons were classified according to their specific intrinsic membrane properties. For this purpose, the following electrophysiological parameters were measured and compared between cell types: V_rest, R_in, τ, C_m, AP threshold, AP amplitude, AP HW, frequency, fAHP and sAHP. The fAHP and sAHP were measured on the first APs from multiple APs evoked by the 1000 ms depolarization step. We also measured the hyperpolarization-activated voltage (SAG) and the SFA. The SAG amplitude was calculated from the voltage responses obtained after injecting a hyperpolarizing current (–100 pA, 1000 ms) between the potential at SAG peak and the potential at the steady state (Vandecasteele et al., 2011). The SFA was calculated using the formula: SFA ratio=F_init/F_final, where F_init is the initial instantaneous spike frequency (1/first interspike interval) and F_final is the instantaneous frequency calculated from the last interspike interval (1/last interspike interval) (Vandecasteele et al., 2011). These intrinsic membrane properties were analyzed using pClamp 10.3 software (Molecular Devices).

Results are expressed as means±s.e.m. Statistical tests were performed using the SPSS 22.0 program (IBM Software). Two-way analysis of variance (ANOVA) was used to analyze the results between different cells followed by the Bonferroni post hoc test. The independent variables sex (males and females in each estrous cycle phase) and cerebral hemisphere were used for the analysis of synaptic properties and intrinsic membrane properties of each cell type. Two-tailed Student's t-test was used to analyze the intrinsic membrane properties between different cell types. The Kolmogorov–Smirnov test was used to compare the cumulative distributions of amplitude and inter-event interval of postsynaptic currents between groups. Statistically significant differences were considered at P<0.05.

We thank colleagues from the Center of Microscopy and Microanalysis of UFRGS (CMM-UFRGS) for skillful technical support.