Estrogenic-dependent glutamatergic neurotransmission from kisspeptin neurons governs feeding circuits in females

The neuropeptides tachykinin2 (Tac2) and kisspeptin (Kiss1) in hypothalamic arcuate nucleus Kiss1 (Kiss1ARH) neurons are essential for pulsatile release of GnRH and reproduction. Since 17β-estradiol (E2) decreases Kiss1 and Tac2 mRNA expression in Kiss1ARH neurons, the role of Kiss1ARH neurons during E2-driven anorexigenic states and their coordination of POMC and NPY/AgRP feeding circuits have been largely ignored. Presently, we show that E2 augmented the excitability of Kiss1ARH neurons by amplifying Cacna1g, Hcn1 and Hcn2 mRNA expression and T-type calcium and h-currents. E2 increased Slc17a6 mRNA expression and glutamatergic synaptic input to arcuate neurons, which excited POMC and inhibited NPY/AgRP neurons via metabotropic receptors. Deleting Slc17a6 in Kiss1 neurons eliminated glutamate release and led to conditioned place preference for sucrose in E2-treated KO female mice. Therefore, the E2-driven increase in Kiss1 neuronal excitability and glutamate neurotransmission may play a key role in governing the motivational drive for palatable food in females.

In contrast to the Kiss1 AVPV/PeN neurons, E2 significantly inhibits Kiss1 mRNA expression in ARH neurons (Lehman et al., 2010;Navarro et al., 2009;Smith et al., 2005). However, the effect of E2 on ion channel expression and excitability is less clear in Kiss1 ARH neurons, and we do not fully understand the mechanism(s) by which E2 affects the excitability of the Kiss1 ARH neuronal population (Zhang et al., 2015). Kiss1 ARH neurons co-express the peptide neurotransmitters kisspeptin (Kiss1), tachykinin 2 (Tac2; a.k.a. neurokinin B; NKB) and dynorphin , which have been proposed to be responsible for pulsatile release of GnRH and LH (Clarkson et al., 2017;Li et al., 2009;Navarro et al., 2009). Indeed, the cellular mechanisms by which Kiss1 ARH neurons synchronize to drive GnRH pulses was recently elucidated and involves a combination of co-released Tac2 excitation via Tac3 receptors and dynorphin presynaptic inhibition via k-opioid receptors . Kiss1 ARH neurons also express the vesicular glutamate transporter 2 (vGluT2) (Cravo et al., 2011;Nestor et al., 2016), an indication that they have the potential to package and release the neurotransmitter glutamate (Herman et al., 2014). Optogenetic activation of Kiss1 ARH neurons evokes glutamatergic EPSCs not only in Kiss1 AVPV/PeN neurons  but also in proopiomelanocortin (POMC) and neuropeptide Y/agouti-related peptide (NPY/AgRP) neurons , which further establishes direct functional connections between Kiss1 ARH neurons and hypothalamic neurons important for the control of reproduction and energy homeostasis. In males, Slc17a6 (encodes vGluT2) mRNA and glutamate release are increased in gonadectomized as compared to intact animals, an indication of inhibitory effects of gonadal steroids on glutamate neurotransmission in these neurons . Based on pronounced male/female differences in feeding behavior and the excitatory glutamatergic input from Kiss1 ARH neurons to Kis-s1 AVPV/PeN neurons in females (Asarian and Geary, 2006;Qiu et al., 2016), we hypothesized that the mRNA expression of Slc17a6 and glutamate release in females in contrast to males, would be amplified by E2 in Kiss1 ARH neurons in order to help maintain reproduction during different energy states. Therefore, we investigated the effects of ovariectomy (OVX) and E2-replacement on neuronal excitability as well as the effect of E2 on mRNA expression of Slc17a6 in Kiss1 ARH neurons and the release of glutamate from these neurons onto Kiss1 AVPV/PeN , POMC and NPY/AgRP neurons. Also, to evaluate the main functional role of glutamate release from Kiss1 ARH neurons in females, we deleted the expression of Slc17a6 in Kiss1 neurons and measured electrophysiological changes in Kiss1 ARH neuronal transmission as well as behavioral changes in vivo.

Results
Estradiol treatment reduces the mRNA expression of neuropeptides while increasing the expression of the T-type calcium channels, HCN channels, and the excitability of Kiss1 ARH neurons It has been shown repeatedly in a number of species, that the mRNA expression of the peptide neurotransmitters kisspeptin, Tac2 and dynorphin, which are co-expressed in Kiss1 ARH neurons, is increased in OVX animals and reduced in E2-treated animals . However, vGluT2 is also expressed in Kiss1 ARH neurons (Cravo et al., 2011;Nestor et al., 2016), and Slc17a6 mRNA levels are reduced in intact as compared to castrated males, an indication of sex-steroid regulation . Currently, we used a sensitive real-time qPCR assay to measure relative quantitative differences of Kiss1, Tac2, Pdyn, Tacr3, Slc17a6 and ion channels in manually-harvested Kiss1 ARH neurons in OVX as compared to E2-treated females (Figures 1-3). Based on the CT values ( Figure 1) and quantitative analysis with Tacr3 as calibrator (see Materials and Methods), Tac2 mRNA was the most highly expressed in Kiss1 ARH neurons from OVX females, such that Tac2 >>> (50-fold) Pdyn !Kiss1 >> (6-fold) Slc17a6, and with E2 treatment, Tac2 >> (18-fold) Pdyn >> (2-fold) Slc17a6 > (2-fold) Kiss1 (one-way ANOVA). These results suggested that there could be a dramatic change in the Kiss1 ARH neuronal synaptic transmission to various target neurons (see below for further quantitative and functional analyses).
Previously, we found that an E2-treatment that induces an LH surge in OVX females increases mRNA expression and/or function of a number of ion channels important for neuronal excitability, including T-type calcium channel subunit Ca V 3.1, sodium channel subunits N aV 1.1a and N aV b2 and the corresponding currents I T , I NaP and I h in Kiss1 AVPV/PeN neurons (Zhang et al., 2015(Zhang et al., , 2013. Since the channels that underlie the pacemaking activity of Kiss1 AVPV/PeN neurons (N aV 1.1a, N aV b2 mRNA and I NaP ) were not altered by E2-treatment in Kiss1 ARH neurons (Zhang et al., 2015), we hypothesized that E2 similarly would not alter Cacna1g mRNA (encoding Cav3.1), Hcn mRNA expression or the associated T-type calcium-and h-currents, respectively, in Kiss1 ARH neurons. Surprisingly, we found that the mRNA expression of Cacna1g and Hcn1 and Hcn2 were significantly increased in Figure 1. qPCR amplification assay illustrating the cycle threshold (CT) for the different neuropeptides and vGluT2 in Kiss1 ARH neurons. Cycle number is plotted against the normalized fluorescence intensity (DRN) to visualize the PCR amplification of Tac2, Pdyn, Kiss1, Tacr3, Slc17a6 and the reference gene Gapdh in 5 cell Kiss1 ARH pools obtained from oil-and E2-treated, OVX animals. The amplification efficiency for each primer pair is listed in Table 1. These efficiencies allowed us to use the comparative DDCT methods for quantification. The cycle threshold (CT; horizontal dashed line) is the point in the amplification from which sample values were calculated using the 2 -DDCT equation as described in the Methods. DOI: https://doi.org/10.7554/eLife.35656.002 The following source data is available for figure 1: Source data 1. OVX + Oil Gene comparisons (fold change generated using OVX + Oil Tacr3 as calibrator) and OVX + E2 Gene comparisons. DOI: https://doi.org/10. 7554/eLife.35656.003 Kiss1 ARH neurons from E2-treated females as compared to oil-treated, OVX females ( Figure 2A). For comparison we also measured Kiss1 and Pdyn mRNAs in the same neuronal pools, and as predicted, Kiss1 and Pdyn mRNA levels were significantly reduced by E2 (Figure 2A). In addition, both T-and h-currents ( Figure 2B) as well as neuronal excitability (measured as rebound excitation; Figure 2C) were increased in Kiss1 ARH neurons with E2-treatment as compared to oil-treatment. Therefore, although E2-treatment downregulates neuropeptide expression in Kiss1 ARH neurons, it significantly increases the intrinsic conductances and hence the excitability of these vital neurons.
Glutamate, not GABA, is released from Kiss1 ARH neurons Using whole-cell, voltage-clamp recordings in slices from OVX female mice, we initially examined evoked (photostimulated) PSCs in ARH NPY/AgRP neurons from Kiss1 Cre:GFP ::Npy GFP mice or in POMC neurons from Kiss1 Cre:GFP ::Pomc EGFP mice that had received a bilateral injection of AAV1-ChR2-mCherry (or ChR2-YFP) into the ARH ( Figure 4A,B). Subsequently, we performed blind wholecell patch recordings from POMC and NPY/AgRP neurons in Kiss1 Cre:GFP mice that had received a bilateral injection of AAV1-ChR2-mCherry (or ChR2-YFP) into the ARH. For the latter group we could segregate the medially located, higher input resistance (1.6 ± 0.2 GW), lower capacitance (~18.4 ± 1.0 pF) NPY/AgRP neurons from the more laterally located, lower input resistance (~1.2 ± 0.2 GW), higher capacitance (~24.4 ± 1.3 pF) POMC neurons (Qiu et al., 2006;Smith et al., 2013). In total, seventy-seven cells showed an inward current following photostimulation with a mean amplitude of 29.8 ± 3.4 pA, a mean latency to peak of 4.0 ± 0.1 ms, and an average decay time constant of 5.5 ± 0.6 ms. This light-induced response was blocked with the AMPA-and NMDAreceptor blockers, CNQX and AP5, respectively ( Figure 4C). Therefore, glutamate, and not GABA, is the primary amino acid neurotransmitter released from Kiss1 ARH neurons in females. Moreover, although TTX (1 mM) abrogated the photostimulated post-synaptic inward current, we could rescue the light-induced response with the addition of the potassium channel blocker 4-AP (100 mM to the . Estradiol regulation of Slc17a6 mRNA expression in Kiss1 ARH neurons. Quantitative real-time PCR measurements of Slc17a6, Tac2 and Tacr3 mRNAs in Kiss1 ARH neuronal pools (3-6 pools of 5 cells each per animal) from OVX oil-and E2-treated mice (n = 4-7 animals per group). Note that E2 increased the mRNA expression of Slc17a6, but as expected decreased the mRNA expression of Tac2 and Tacr3 in the same Kiss1 neuronal pools. Bar graphs represent the mean ±SEM (for Slc17a6, Unpaired t-test, t (8) = 4.522, p<0.001; Tac2, Unpaired t-test, t (8) = 6.350, p<0.001; Tacr3, Unpaired t-test, t (6) = 7.161, p<0.001). ***p<0.001. DOI: https://doi.org/10.7554/eLife.35656.006 The following source data is available for figure 3: bath) in both POMC (n = 3) and NPY/AgRP (n = 2) neurons ( Figure 4D,E), which is biophysical evidence for direct synaptic signaling from Kiss1 ARH neurons to POMC and NPY/AgRP neurons (Cousin and Robinson, 2000;Petreanu et al., 2009). Following recording, the cytoplasm of numerous responsive cells was collected and analyzed for mRNA transcripts (Pomc, Npy, and Kiss1) using scRT-PCR to confirm our targeting strategy described above ( Figure 4D,E, Insets). Based on this analysis, there was no difference in the amplitude of the evoked fast PSCs between POMC and NPY neurons (post-hoc identified) (25.8 ± 4.8 pA, n = 36 POMC neurons versus 33.3 ± 4.9 pA, n = 41 NPY/AgRP neurons). Therefore, it appears that when Kiss1 ARH neurons in the female are firing at low Steroid-dependent glutamate release from Kiss1 ARH neurons E2-treatment of OVX females increased the mRNA expression of Slc17a6 in Kiss1 ARH neurons, and synaptic glutamate release is tightly coupled to vGluT2 expression (Herman et al., 2014). Therefore, we explored the differences in evoked glutamate release from Kiss1 ARH neurons using a pairedpulse-ratio paradigm (PPR: ratio of the amplitude of the second pulse over the amplitude of the first pulse) (Zucker and Regehr, 2002) in control (oil) and E2-treated OVX Kiss1 Cre:GFP , Kiss1 Cre:GFP :: Npy GFP and Kiss1 Cre:GFP ::Pomc EGFP female mice that received a bilateral injection of AAV1-DIO-ChR2:mCherry (or AAV1-DIO-ChR2:YFP) into the ARH. Using a photostimulation PPR protocol of two 5 ms LED stimulations separated by 50 ms ( Figure 5A,B), we found that POMC and NPY/AgRP neurons in E2-treated, OVX female mice had a significantly lower PPR compared to oil-treated, OVX females in both POMC neurons ( Figure 5C,D) and in NPY/AgRP neurons ( Figure 5E,F). These results indicate that there is a greater probability of glutamate release with the first stimulus in E2treated, OVX animals ( Figure 5B), which is consistent with higher mRNA expression of Slc17a6 in Kiss1 ARH neurons following E2-treatment. Collectively, these data indicate that there is a striking male/female difference concerning the effects of sex steroids (testosterone versus estradiol) on Slc17a6 mRNA expression and glutamate release in Kiss1 ARH neurons . Previously, we found that Kiss1 ARH neurons send projections to and excite Kiss1 AVPV/PeN neurons via glutamate release . Currently we confirmed that ChR2-mCherry injections labeled Kiss1 ARH neurons within the arcuate nucleus only ( Figure 6A1), and sent extensive fiber-projections rostrally to the AVPV/PeN area in the vicinity of immunoreactive Kiss1 neurons ( Figure 6A2,  , illustration of a paired-pulse regime (two blue light pulses of 5 ms duration separated by 50 ms); fast glutamatergic inward currents (P1 and P2) were recorded to measure the probability of neurotransmitter release in postsynaptic neurons. (C), AAV1-DIO-ChR2:mCherry was bilaterally injected into ARH of Kiss1 Cre ::Npy GFP mice or Kiss1 Cre ::Pomc EGFP mice. Using a paired-pulse regime, fast glutamatergic inward currents were recorded in POMC EGFP neurons (V hold = À60 mV) from both oil-treated, OVX (upper trace, black) and E2-treated, OVX (lower trace, red) females. The averaged responses (50 sweeps) are shown. (D) E2-treatment significantly decreased the paired-pulse ratio (P2/P1; indicating that there was a higher probability of glutamate release from Kiss1 Cre:GFP -ChR2 neurons (Unpaired t-test, t (20) = 4.184, p<0.001). (E) similarly using a paired-pulse regime, fast glutamatergic inward currents were recorded in NPY GFP neurons (V hold = À60 mV) from both oil-treated, OVX (upper trace, black) and E2-treated, OVX (lower trace, red) females. The averaged responses (50 sweeps) are shown. (F) E2-treatment significantly decreased the paired-pulse ratio (P2/P1) indicating that there was a higher probability of glutamate release from Kiss1 Cre:GFP -ChR2 neurons (Unpaired t-test, t (14) = 3.255, p<0.01). **p<0.01, ***p<0.001. DOI: https://doi.org/10.7554/eLife.35656.009 The following source data is available for figure 5: Source data 1. The paired-pulse ratio (P2/P1) for Figure 5D   showing the pronounced projections of ChR2:mCherry fibers to the preoptic area including the PeN following bilateral injections of AAV1-DIO-ChR2: mCherry in the ARH of Kiss1 CreGFP V2 mice (note that the GFP was not visible in the V2 animals). Therefore, some POA sections from the same animals were stained for kisspeptin using the Caraty 564 antibody and revealed immunoreactive Kiss1 neurons in the PeN (green cells) (A3). Essentially none of the POA somas including the Kiss1 cells expressed ChR2-mCherry. Scale bars = 100 mM (A1,A2); 50 mm (A3). (B) schematic of photostimulation of the terminals of Kiss1 ARH neurons and recording of Kiss1 AVPV/PeN neurons. (C) following AAV1-DIO-ChR2:YFP (or mCherry) injection into the ARH, a fast inward current was recorded in Kiss1 AVPV/PeN neurons following blue light stimulation (green trace). The response was antagonized by CNQX (10 mM) and AP5 (50 mM) (not shown) and was abrogated in the presence of TTX (1 mM, black trace) but rescued with the addition of the K + channel blocker 4-AP (100 mM, magenta trace; n = 4 neurons). (D) using a paired-pulse regime (two blue light pulses of 5 ms duration separated by 50 ms), fast glutamatergic inward currents were recorded in Kiss1 AVPV/PeN neurons (V hold = À60 mV) from both oil-treated, OVX (upper trace, black) and E2- Figure 6 continued on next page A3). Notably, ChR2-mCherry expressing cell bodies were not found in the AVPV/PeN. Thus, we determined whether this glutamatergic projection was also regulated by E2. We used a coronal slice preparation that contained only the ARH Kiss1 Cre:GFP -AAV1-DIO-ChR2:mCherry (or ChR2:YFP) fiber projections to the AVPV/PeN ( Figure 6A2,A3) and explored the light-evoked glutamate release from Kiss1 ARH neurons ( Figure 6B). Photostimulation-induced an inward current in Kiss1 AVPV/PeN neurons, which was blocked in the presence of TTX (1 mM) but rescued with addition of 4-AP in the presence of TTX (n = 4) ( Figure 6C). Therefore, this is biophysical evidence for direct synaptic signaling from Kiss1 ARH neurons to Kiss1 AVPV/PeN neurons. Using a photoactivation-PPR protocol of two 5 ms LED stimulations as described above, we found that AVPV/PeN neurons, including Kiss1 AVPV/PeN neurons, from E2-treated females had reduced PPR, an indication that Kiss1 AVPV/PeN neurons, similar to POMC and NPY/AgRP neurons, receive an enhanced glutamatergic input from Kiss1 ARH neurons in E2-treated females ( Figure 6D,E).

High frequency stimulation of glutamate release from Kiss1 ARH neurons excites POMC neurons and inhibits NPY neurons via metabotropic receptors
Although optogenetic activation of Kiss1 ARH neurons at low frequency caused similar ionotropic stimulation of both POMC and NPY/AgRP neurons, high-frequency stimulation of Kiss1 ARH neurons/ fibers, while blocking the fast AMPA/NMDA input with CNQX + AP5, had differential actions on these two different populations of ARH neurons as measured in slices from E2-treated females. High-frequency (20 Hz) stimulation of Kiss1 ARH neurons/fibers generated a slow EPSC in POMC EGFP neurons, and in current clamp the stimulation depolarized POMC neurons and increased their firing rate ( Figure 7A-C). In contrast, high-frequency optogenetic stimulation generated a slow IPSC in NPY GFP neurons, and in current clamp the same stimulus hyperpolarized and inhibited NPY neurons ( Figure 7A,D,E). This high frequency response was blocked approximately 90% by the mGluR7 antagonist ADX71743 (IPSP reduced from 5.9 ± 1.2 mV to 0.6 ± 0.2 mV, n = 5; paired t-test, t (4) = 4.281, p=0.013). The change in membrane potential following Kiss1 ARH stimulation was significantly different between POMC and NPY/AgRP neurons ( Figure 7F). Therefore, high-frequency activity in Kiss1 ARH neurons, which through presumably glutamate spillover to extrasynaptic mGluRs (Nietz et al., 2017;Watanabe and Nakanishi, 2003), excites POMC but inhibits NPY/AgRP neurons.
To pharmacologically elucidate the postsynaptic metabotropic glutamate response in NPY neurons, we first utilized the group II mGluR agonist DCG-IV (10 mM), and found that it hyperpolarized and inhibited firing in NPY neurons ( Figure 10A). In addition, DCG-IV in the presence of fast sodium channel, ionotropic glutamatergic and GABAergic blockade induced an outward current in NPY/ AgRP neurons ( Figure 10B). The I-V relationship for the DCG-IV induced current exhibited inward rectification and a reversal potential at E K + (À95 mV), the hallmark of activation of G protein-coupled inwardly rectifying K + (GIRK) channels ( Figure 10C). Moreover, 10 mM DCG-IV was more efficacious to hyperpolarize NPY/AgRP neurons in E2-treated OVX females, indicating that there was a more robust E2-induced inhibition of NPY/AgRP neurons ( Figure 10G). DCG IV, however, had no effect on POMC neuronal excitability (data not shown). Since mGluR7 was the most highly expressed in NPY/AgRP neurons (Figure 8), we also tested the mGluR7-selective agonist AMN082 (10 mM) (Ren et al., 2011) for its postsynaptic actions on NPY/AgRP neurons. Indeed, AMN082 hyperpolarized and inhibited firing in NPY/AgRP neurons ( Figure 10D). It generated about a 2-fold greater outward current than DCG-IV, which reversed near E K + ( Figure 10E,F), and the magnitude of the hyperpolarization was significantly increased in E2-treated, OVX females ( Figure 10H). However, AMN082 (10 mM) had no effect on a POMC neuron that exhibited a robust response to DHPG (data not shown). In addition, the mRNA expression of mGluR7 was significantly increased in NPY neurons obtained from E2-versus oil-treated OVX females ( Figure 10I). Therefore, E2 significantly increases the receptor expression and response of NPY/AgRP to group II/III mGluR's (Gi,o-coupled), which would contribute to the inhibition of these orexigenic neurons in heightened E2-driven reproductive states.

Conditional knockout of vGluT2 in Kiss1 ARH neurons eliminates glutamate release
To better understand the function of glutamate in Kiss1 neurons, we ablated vGluT2 specifically in Kiss1 neurons. Given that Slc17a6 is not expressed in Kiss1 AVPV/PeN neurons in our mouse model, this ablation was specific for Kiss1 ARH neurons and, perhaps, for other Kiss1 neurons not yet shown to express Slc17a6, including the amygdala and bed nucleus of stria terminalis (BNST) Kiss1 neurons  Qiu et al., 2016). To delete vGlut2 from Kiss1 neurons we crossed Slc17a6 lox/ lox mice with Slc17a6 +/D ::Kiss1 CreGFP/+ mice, which yielded mice that were knockout (KO) or heterozygotes for Slc17a6 (Het) in Kiss1 neurons, as well as Slc17a6 Het and wild-type (WT) mice that lacked Cre. We confirmed the Slc17a6 deletion in KO mice using scRT-PCR of harvested Kiss1 ARH neurons ( Figure 11A). For identification of Kiss1 ARH neurons in E2-treated animals and for functional studies, AAV1-DIO-ChR2:YFP was injected bilaterally in the ARH of Slc17a6 KO, Slc17a6 Het and control Kiss1 Cre:GFP females. Importantly based on the scRT-PCR analysis, none of the YFP cells in KO females expressed Slc17a6, whereas 100% expressed Tac2 ( Figure 11A). The majority of cells also expressed Kiss1 mRNA, although this transcript was more difficult to detect in E2-treated females (see Figure 1). Based on previous findings, the slow EPSP underlying Kiss1 ARH neuronal synchronization, is dependent on the release of Tac2 and dynorphin in Kiss1 ARH neurons . To explore a potential role of glutamate in this function, we generated a slow EPSP using high-frequency (20 Hz, 10 s) optogenetic stimulation in slices obtained from oil-and E2-treated, OVX females . High-frequency optogenetic stimulation generated a similar slow EPSP amplitude in arcuate Kiss1 neurons from oil-treated, OVX Kiss1 control and oil-treated OVX Slc17a6 KO Kiss1 mice ( Figure 11C,D,G). In contrast, high-frequency optogenetic stimulation of Kiss1 neurons in E2- , ionotropic glutamatergic (CNQX, 10 mM and AP5, 50 mM) and GABAergic (picrotoxin, 100 mM) blockers (V hold = À60 mV). (C) I-V relationship for DCG-IV-induced current showed a reversal at E K+ (À95 mV). (G) DCG-IV was more efficacious to hyperpolarize NPY neurons in E2-treated versus oil-treated, OVX females (Unpaired t-test, t (32) = 2.261, p=0.031). *p<0.05. (D) the mGluR7 allosteric agonist AMN082 (10 mM) hyperpolarized and inhibited firing of an NPY GFP neuron. (E) AMN082 generated a 25 pA outward current in a NPY GFP neuron in the presence of fast sodium channel and ionotropic glutamatergic blockade (V hold = À60 mV). (F) I-V relationship for AMN082-induced current showed a reversal close to E K+ . (H) AMN082 was more efficacious than DCG-IV to hyperpolarize NPY neurons in E2-treated versus oil-treated, OVX females (Unpaired t-test, t (17) = 3.747, p=0.002). (I) Quantitative real-time PCR measurements of Grm7 mRNA in NPY GFP neuronal pools (4 pools of 5 cells each per animal) from oil-and E2-treated, OVX mice (n = 4-5 animals per group). Bar graphs represent the mean ± SEM (Unpaired t-test, t (7) =6.020, p=0.0005). ***p<0.001. DOI: https://doi.org/10.7554/eLife.35656.019 The following source data is available for figure 10: Source data 1. DCG-IV was more efficacious to hyperpolarize NPY neurons in E2-treated versus oil-treated, OVX females ( Figure 10G). DOI: https://doi.org/10.7554/eLife.35656.020 treated, OVX control and E2-treated, OVX Slc17a6 KO mice revealed a significantly reduced slow EPSP in Slc17a6 KO as compared to control Kiss1 ARH neurons ( Figure 11E,F,G). Therefore even though the slow EPSP is generated mainly by Tac2 in OVX females , glutamate may contribute to the slow EPSP in Kiss1 ARH neurons in E2-treated, OVX females.
To further study the actions of kisspeptin on NPY/AgRP and POMC neurons, we applied kisspeptin directly on these neurons. Kisspeptin (200 nM) inhibited firing and hyperpolarized NPY GFP neurons even in the presence of the GABA A blocker bicuculline (10 mM) ( Figure 13A,B). Also, similar to the GABA B receptor agonist baclofen, kisspeptin induced an outward, albeit smaller current with a reversal potential close to E K + (À90 mV) ( Figure 13C). In addition to binding to the Kiss1 receptor (GPR54), kisspeptin is also known to bind to and activate neuropeptide FF receptors 1 and 2 (NPFFR1 and NPFFR2), and these Gai/o coupled receptors are both expressed in the ARH (Elhabazi et al., 2013;Rønnekleiv et al., 2014). We used scRT-PCR to document the expression of Npffr1 mRNA in NPY neurons ( Figure 13D). Finally, the RFamide-related peptide-3 (RFRP-3) (10 mM), a selective agonist for NPFFR1 and NPFFR2 (Bonini et al., 2000), hyperpolarized and inhibited the firing of NPY/AgRP neurons ( Figure 13E). The I/V plot of the RFRP-3 induced current showed a reversal potential close to E K + (À85 mV) ( Figure 13F). Therefore, it is probable that kisspeptin and RFRP3 activate the same receptor on NPY/AgRP neurons. We also explored the actions of kisspeptin on POMC neurons after first identifying that Kiss1r mRNA (GPR54) is expressed in a subpopulation of these neurons using scRT-PCR ( Figure 14A). In agreement with a previous publication (Fu and van den Pol, 2010), we found that kisspeptin (200 nM) depolarized and increased the firing frequency of POMC neurons ( Figure 14B). In addition, the I/V plot showed that kisspeptin activated a non-selective cationic channel that reversed at À10 mV ( Figure 14C). Female mice lacking Slc17a6 in Kiss1 neurons develop a condition place preference for sucrose Ad libitum access to standard, low-fat mouse chow did not affect body weight in mice lacking Slc17a6 in Kiss1 neurons versus control females over the limited time-course of our study (data not shown). Therefore, we hypothesized that the lack of glutamate release from Kiss1 ARH neurons might increase motivation for palatable food in Slc17a6 KO females due to diminished regulatory synaptic input onto POMC, NPY/AgRP neurons or other target neurons. For this analysis we used a conditioned place preference (CPP) paradigm (Figure 15), that has been used extensively to evaluate . Protocol for inducing CPP with sucrose. The protocol for conditioning and preference testing consisted of four phases over the course of 11 days (sucrose habituation, a baseline place preference (BPP) test, sucrose conditioning, and a conditioned place preference (CPP) test). Food-motivated behavior was assessed during the dark cycle using an unbiased procedure. On the day before BPP, Day À1, sucrose habituation occurred where mice received sucrose pellets overnight (O/N) in their home cage to prevent neophobia. The initial BPP (black vs. white chamber) was assessed on Day 1 in a three-chamber place preference apparatus and the chamber pairing was assigned in an unbiased manner. During sucrose conditioning, mice were given access to sucrose-filled (CS+, Days 2, 4, 6, 8) or empty (CS-, Days 3, 5, 7, 9) lids on alternating days. Mice were given access to sucrose-filled lids in one chamber (e.g. white), then on alternating days they were presented with empty lids in the other chamber (e.g. black). Mice were tested for acquisition of a CPP to sucrose on Day 10, which was indicated by increased time spent in the sucrose-conditioned chamber. Animals were fed ad lib standard low-fat chow in their home cage throughout the study. For cyclical estradiol treatment, animals were given a priming (0.25 mg) and a surge (1.5 mg) dose of 17b-estradiol Benzoate (E2) at 9 AM prior to the BPP and prior to the CPP as indicated. During the sucrose-conditioning (phase 3), the animals were treated twice with a 1 mg maintenance dose of E2. DOI: https://doi.org/10.7554/eLife.35656.026 drugs of abuse as well as the preference for palatable food (Cunningham et al., 2006;Harris et al., 2005;Prus et al., 2009;Sinclair et al., 2017). While several groups have used sucrose pellets/solutions to produce a CPP (Alderson et al., 2001;Duarte et al., 2003;Papp et al., 2002) many investigators have found food restriction as a necessary adjuvant (Baunez et al., 2005;Figlewicz et al., 2001). However, even with short-term food deprivation, homeostatic processes can be engaged that prevent selective study of hedonic pathways, as evidenced by the ability of standard chow to produce a CPP following 20 h/day of food restriction (Papp et al., 2002;Popik et al., 2003). Therefore, to study hedonic feeding, we used ad-libitum fed mice and tested them during the dark phase when mice normally eat most of their food.
Food-motivated, behavioral analysis revealed that E2-treated, OVX Kiss1 female mice failed to exhibit a change in place preference for sucrose, a natural food reward ( Figure 16A, Figure 16-figure supplement 1). In contrast, E2-treated, OVX Slc17a6 KO mice significantly increased their occupancy of the sucrose-paired chamber ( Figure 16A). E2-treated, OVX Heterozygous (Slc17a6 +/lox :: Kiss1 CreGFP/+ ) mice showed a trend to increase their occupancy in the sucrose-paired chamber ( Figure 16A). In addition, we measured the amount of sucrose consumed by each experimental group during conditioning. The E2-treated, OVX Kiss1 controls did not increase their sucrose intake at any time ( Figure 16B, Figure 16-figure supplement 1). The Slc17a6 KO females slightly increased their sucrose intake on day 6 of the test (third day of sucrose exposure; Figure 15) and this was significantly augmented by day 8 of the test (fourth day of sucrose exposure) ( Figure 16B, Figure 15). The Slc17a6 Het mice exhibited a smaller, but significant increase in sucrose consumption on day 6 of the test ( Figure 16B). Together, these findings indicate that E2-treated, OVX Kiss1 control females had a consistent diminished motivation for sucrose, whereas abolishing glutamate release from Kiss1 neurons enhanced the motivational response for sucrose. Interestingly, attenuating glutamate release from Kiss1 neurons in Slc17a6 Het females nearly recapitulated the phenotype of the Slc17a6 KO animals suggesting a gene-dosage/threshold effect. We also measured the body weight gain over the ten-day period of the CPP test. Although both the Slc17a6 KO and Het mice gained weight in comparison to the E2-treated, OVX Kiss1 controls, the difference was only statistically significant in the Slc17a6 Het mice ( Figure 16C). Based on these findings, we measured the mRNA expression of Slc17a6 in Kiss1 ARH neurons in the heterozygous E2-treated, OVX females (Slc17a6 Het) as compared to control E2-treated, OVX Kiss1 Cre females. As illustrated ( Figure 16D), Slc17a6 mRNA in Het mice was~35% of that of controls (Slc17a6 Het mice, 5 pools each from 4 animals; Kiss1 control mice, 5 pools each from 5 animals), and Slc17a6 mRNA was not detected in Kiss1 neurons from Slc17a6 KO females. Interestingly intact males, which have reduced Slc17a6 mRNA expression in Kiss1 ARH neurons and reduced glutamate release compared to castrates  and E2-treated females, demonstrated a motivation for sucrose based on CPP ( Figure 16-figure supplement 1).

Deletion of Slc17a6 in Kiss1 neurons does not alter the estrous cycle
We have shown previously and currently that Kiss1 ARH neurons exhibit direct communication with Kiss1 AVPV/PeN neurons via glutamate, which project to and excite GnRH neurons via kisspeptin release . Therefore, we hypothesized that glutamate released from Kiss1 ARH neurons would be involved in the excitation of Kiss1 AVPV/PeN neurons and the induction of the GnRH surge. As a consequence, deletion of Slc17a6 in Kiss1 ARH neurons might lead to disruption of the estrous cycle and the GnRH surge and thus affect fertility. However, we found that the estrous cycle was normal in Slc17a6 KO females (one-way ANOVA; p=0.68; comparing cycle length between: wild types (WT), n = 5, 5.55 ± 0.42; Het, n = 3, 4.89 ± 0.48; and KO, n = 7, 5.43 ± 0.44), although the staging of the cycle was not confirmed by steroid hormone measurements. Similarly, the time to conception when placed with a fertile male was not different between the three groups (one-way ANOVA; p=0.68 comparing time (days) to conception between: WT, n = 9, 1.75 ± 0.40; Het, n = 3, 2.83 ± 1.64; KO, n = 8, 1.89 ± 0.67).

Discussion
It is well known that the peptide neurotransmitters in Kiss1 ARH neurons are negatively regulated by E2, and these neurons are responsible for pulsatile release of GnRH and reproduction (Oakley et al., 2009). Our current findings that E2 increases the expression of Slc17a6 and glutamate release reveal that the amino acid and peptide neurotransmitters are regulated differentially by E2 in Kiss1 ARH neurons in females in contrast to our findings in males . We also found that E2 increased the expression of T-type calcium and h-currents in female Kiss1 ARH neurons, which led to increased neuronal excitability, concomitant with E2-induced inhibition of the expression of the peptide neurotransmitter mRNAs Kiss1, Tac2 and Pdyn. Optogenetic activation of Kiss1 ARH neurons revealed direct frequency-dependent glutamatergic and peptidergic neurotransmission to POMC and AgRP neurons in females, an indication that Kiss1 ARH neurons may play a role in regulating feeding behavior. The glutamatergic outputs were lost in females with conditional ablation of Slc17a6 in Kiss1 ARH neurons, whereas excitatory and inhibitory kisspeptin responses could be evoked in POMC and NPY/AgRP neurons, respectively. Experiments in vivo revealed that Slc17a6 KO females did not gain weight on normal mouse chow, but the motivation to ingest sucrose was increased in females lacking vGluT2 in Kiss1 neurons. Overall, these and other findings support the idea that Kiss1 ARH neurons may provide E2-and frequency-dependent signals not only to POMC and NPY/AgRP neurons, but also to Kiss1 AVPV/PeN neurons to help coordinate feeding and reproduction in females ( Figure 17).
Kiss1 ARH neurons are known to co-express the peptide neurotransmitters Kiss1, Tac2 and Dynorphin (Goodman et al., 2007;Lehman et al., 2013;Navarro et al., 2009). Using sc-qPCR we observed that Tac2, and not Kiss1, is by far the most highly expressed peptide mRNA in Kiss1 ARH neurons. Even in the presence of high physiological levels of E2, the mRNA expression of Tac2 was many-fold higher than Kiss1 under the same treatment. Since Tac2 plays a key role in synchronous firing of Kiss1 ARH neurons , which underlies the pulsatility of GnRH release that Figure 17. Working Model. KNDy (Kisspeptin, NKB, Dynorphin) neurons in the ARH express Ca V 3 (I T ) and HCN (I h ) channels (currents) that are upregulated by E2 and contribute to increased excitability of Kiss1 ARH neurons. Kiss1 AVPV/PeN neurons also express Ca V 3 (I T ), HCN (I h ) and Nav (I NaP ) channels that are highly up-regulated by E2 along with Kiss1 mRNA expression. Notably, E2 induces spontaneous, repetitive burst firing activity in Kiss1 AVPV/PeN neurons necessary for the release of GnRH (Wang et al., 2016;Zhang et al., 2015). E2 also directly excites POMC neurons via inhibition of GIRK current, but inhibit NPY/AgRP neurons via activation of GIRK current . These congruent actions of E2 on POMC and NPY/AgRP neurons contribute to the control of homeostatic feeding. High frequency photo-stimulation (focal light stimulation of channel rhodopsin, ChR2) in Kiss1 ARH neurons releases glutamate to further excite POMC neurons via mGluRs group I and inhibit NPY/AgRP neurons via mGluRs group II/III; and excite Kiss1 AVPV/PeN neurons via NMDA/AMPA receptors. Ablating Slc17a6 from Kiss1 ARH neurons, results in the abrogation of glutamate release onto POA and ARH neurons. The lack of glutamate release from Kiss1 ARH neurons appears to have little or no effect on estrous cyclicity, an indication that the direct effects of E2 to increase the excitability of Kiss1 AVPV/PeN neurons is sufficient to drive the reproductive cycle. However, E2-treated Slc17a6 KO Kiss1 mice develop a condition place preference for sucrose indicative of positive motivational effect of sucrose in these females. DOI: https://doi.org/10.7554/eLife.35656.031 drives pituitary LH secretion (Clarkson et al., 2017), it is not surprising that Tac2 was found to be the most highly expressed peptide in Kiss1 ARH neurons.
Given that the peptide neurotransmitters in Kiss1 ARH neurons are primarily down-regulated by E2, at least in rodents, the Kiss1 ARH neurons are believed to be under inhibitory control by E2 and are important for negative-feedback regulation of GnRH and LH secretion Navarro et al., 2009;Smith et al., 2005). However, our past (Gottsch et al., 2011) and current findings that these Kiss1 ARH neurons express T-type calcium and pacemaker h-currents and are keenly sensitive to excitation by glutamate are indications that these neurons have pacemaker electrophysiological properties similar to other CNS neurons (Bal and McCormick, 1993;Lüthi and McCormick, 1998). Additionally, in contrast to the neuropeptides, E2 increased Slc17a6 mRNA expression in Kiss1 ARH neurons and increased glutamate release onto Kiss1 AVPV/PeN neurons in the POA, and onto POMC and NPY/AgRP neurons in the ARH. This is a clear indication that the amino acid and peptide neurotransmitters are regulated differentially by E2 in Kiss1 ARH neurons in females. Interestingly, we have reported that Slc17a6 mRNA expression in Kiss1 ARH neurons and the probability of glutamate release are decreased along with the neuropeptides in intact versus castrated males . Therefore, there is a significant male/female difference in sex-steroid regulation of glutamate signaling by Kiss1 ARH neurons . This could underlie the sex differences in feeding behavior between males and females, namely that testosterone increases feeding in castrated males, whereas E2 reduces feeding in OVX females and during the peri-ovulatory phase of the estrous cycle when E2 levels are maximal (Asarian and Geary, 2006;Asarian and Geary, 2013).
The mechanism by which E2 signals in Kiss1 ARH neurons to inhibit the expression of neuropeptides appears to be via ERa, given that ablation of ERa in Kiss1 ARH neurons prevents E2-induced suppression of Kiss1 mRNA, and also global KO of ERa prevents suppression of Tac2 and Pdyn by E2 in the ARH (Dubois et al., 2016;Yang et al., 2017). The mechanism by which E2 enhances glutamatergic expression and transmission in Kiss1 ARH neurons is currently unknown. Potentially, E2 acting via different E2 receptors and/or signaling pathways in Kiss1 ARH neurons could be responsible for the diverse actions of E2 (decrease in the expression of neuropeptides versus the increase in expression of excitatory ion channels and glutamate release) similar to what has been reported previously in hippocampal neurons and in hypothalamic NPY/AgRP neurons (Boulware et al., 2005;Smith et al., 2013,Smith et al., 2014. The excitatory glutamatergic inputs to NPY/AgRP neurons play a key role in the response to fasting, although the origin of this excitatory input has not been determined (Liu et al., 2012). Similarly, it has been shown that glutamatergic neurons other than POMC neurons within the arcuate nucleus are responsible for a fast acting satiety signal to suppress feeding even after a 24 hr fast in male mice (Fenselau et al., 2017). In addition, our current findings using optogenetic stimulation of Kis-s1 ARH neurons in vitro have documented direct excitatory projections to both POMC and NPY/AgRP neurons by low-frequency stimulation, whereas high-frequency stimulation activated POMC and inhibited NPY/AgRP neurons in both males  and females (current findings). The low-frequency excitation was blocked by TTX, but reinstated by application of the potassium channel blocker 4-AP, which is biophysical evidence of direct synaptic input from Kiss1 ARH neurons to POMC and NPY/AGRP neurons in agreement with our previous findings in males . High-frequency optogenetic stimulation (20 Hz) of Kiss1 ARH neurons, which mimics a firing rate that is observed in these ARH neurons in vivo (Moss et al., 1975), excited POMC neurons. This appears to be generated by an activation of group I metabotropic glutamate receptors 1 and 5, both of which we have documented are expressed in POMC neurons. Also, we found that the group I mGluR agonist DHPG excited POMC neurons in the presence of TTX and ionotropic glutamate receptor blockers. Collectively these data demonstrate that there is a direct excitatory input to POMC neurons from Kiss1 ARH neurons via glutamate actions on ionotropic and metabotropic receptors. In contrast, we found that high-frequency stimulation generated an inhibition of NPY/AgRP neurons, which was mediated by glutamatergic type II/III metabotropic receptors. Indeed, we found that NPY/AgRP neurons express Grm7 mRNA (encoding mGluR7), and to a lesser extent Grm2 mRNA (encoding mGluR2) and that the high frequency response was inhibited by the mGluR7 antagonist ADX71743. In addition, these neurons were inhibited by bath application of the group II mGluR agonist DCG-IV and the group III mGluR agonist AMN082 in the presence of synaptic blockade by TTX, and ionotropic glutamate and GABA A inhibitors, providing evidence for a direct inhibitory Kiss1 ARH glutamatergic input to NPY/AgRP neurons. The response in POMC neurons to group I metabotropic agonist DHPG, which activates a TRPC current (Tozzi et al., 2003), was not different between oiland E2-treated, OVX females. In contrast, the GIRK current in NPY/AgRP neurons, which was activated by the group II and the group III metabotropic glutamate receptor agonists, was augmented by E2-treatment, as was the mRNA expression of mGluR7. Congruent with these findings, we have shown that E2-treatment increases KCNQ 5 channel expression and the corresponding inhibitory M-current in NPY/AgRP neurons (Roepke et al., 2011) and augments postsynaptic GABA B receptor coupling ( Smith et al., 2013, Smith et al., 2014. Clearly, E2 inhibits the orexigenic NPY/AgRP neurons via multiple mechanisms and thereby helps to prevent hyperphagia (Roepke et al., 2010).
The evidence that Kiss1 ARH neurons are involved in the regulation of GnRH and LH pulsatility via their peptide neurotransmitters is well documented (Clarkson et al., 2017;Goodman et al., 2013;Navarro et al., 2009;Qiu et al., 2016;Wakabayashi et al., 2010). Although it is known that Slc17a6 is expressed in Kiss1 ARH neurons (Cravo et al., 2011;Nestor et al., 2016), the role of glutamate neurotransmission from these neurons had not been elucidated. Currently, we deleted vGluT2 (Slc17a6) in Kiss1 neurons and found that glutamate release was completely abrogated in the ARH population of Kiss1 neurons. Given that we have documented previously that Kiss1 AVPV/PeN neurons do not express Slc17a6  and do not release glutamate, we believe that the KO would be specific for Kiss1 ARH neurons. However, Kiss1 neurons are expressed in other brain regions, including the medial amygdala and BNST , and although Slc17a6 has not yet been documented in these neurons, we cannot rule out the possibility that vGluT2 has been deleted in extra-hypothalamic kisspeptin neurons.
Within the Kiss1 ARH neurocircuitry, the lack of glutamate resulted in a diminished slow EPSP in E2-treated animals. The slow EPSP is attenuated by E2 treatment , and it was further reduced in E2-treated Slc17a6 KO females. Given that the slow EPSP underlies Tac2-dependent engagement of synchronous activity in Kiss1 ARH neurons , these findings support the idea that glutamate may play a role in Kiss1 ARH synchronous firing activity and LH pulsatility in the presence of high circulating levels of E2, when peptide neurotransmitters are low and glutamate levels are high in female Kiss1 ARH neurons. Interestingly, females with ablation of vGluT2 in Kiss1 neurons appeared to exhibit a normal ovulatory cycle, an indication that glutamatergic neurotransmission from Kiss1 neurons may not be necessary to support reproductive function. However, deletion of ERa in all kisspeptin neurons (KERKO) significantly attenuates glutamatergic synaptic input (spontaneous EPSCs) to the Kiss1 AVPV/PeN neurons, which could be due to lack of E2-stimulated glutamate release from Kiss1 ARH neurons (Wang et al., 2017). The net result is an increase in LH pulse frequency in KERKO mice, which was proposed to represent a lack of negative feedback (Wang et al., 2017). Clearly further experiments, including deletion of Slc17a6 specifically in adult females, which would prevent potential developmental compensation, are necessary to determine whether glutamate and/or peptide neurotransmission from Kiss1 ARH neurons to Kiss1 AVPV/PeN neurons are involved in reproductive functions.
High-frequency stimulation of Kiss1 ARH neurons in slices from E2-treated OVX Slc17a6 KO females induced a low-amplitude, inward current in POMC neurons and a low-amplitude, outward current in NPY/AgRP neurons, an indication of residual excitatory and inhibitory inputs, respectively to these neurons. Given that the high frequency stimulation has been documented to induce peptide release , this is an indication that kisspeptin release from Kiss1 ARH neurons evoked the excitation of POMC neurons and inhibition of NPY/AgRP neurons. In agreement with previous studies (Fu and van den Pol, 2010), bath application of kisspeptin activated POMC neurons most likely by signaling via the Kiss1 receptor, GPR54, which is expressed in POMC neurons. Similarly, kisspeptin inhibited firing and hyperpolarized NPY/AgRP neurons via activation of a GIRK current. The Kiss1 receptor is not expressed in NPY/AgRP neurons. However, NPFFR1 is expressed, and its agonist RFRP-3 inhibited firing and hyperpolarized NPY/AgRP neurons via activation of a GIRK current. Based on evidence that both RFRP-3 and kisspeptin bind to and activate NPFFR1 (Bonini et al., 2000), we propose that the two different neuropeptides inhibit NPY/AgRP neurons by action on the same receptor. Collectively, these data indicate that Kiss1 ARH neurons may use both kisspeptin and glutamate to activate POMC neurons but inhibit NPY/AgRP neurons, and compensatory kisspeptin input could be one of the reasons why deleting vGluT2 was not effective to alter body weights on a normal chow diet over the time-course of our study.
As the Kiss1 Slc17a6 KO females did not appear to gain weight on standard low-fat mouse chow diet, we used a place preference paradigm to determine the motivation for sucrose in Slc17a6 KO as compared to control Kiss1 and Slc17a6 Het females (Prus et al., 2009). These experiments revealed a preference for the sucrose-paired chamber by the E2-treated, OVX mice lacking Slc17a6 in Kiss1 neurons compared to Kiss1 Cre and Het females. The Slc17a6 Het females did not exhibit a CPP to sucrose, but did consume more sucrose in comparison to control Kiss1 females. Importantly, the E2-treated Slc17a6 Het females expressed low levels of Slc17a6 mRNA in Kiss1 ARH neurons, an indication that glutamate release was attenuated (Herman et al., 2014), which could be why the Slc17a6 Het females behaved more like Slc17a6 KO than control Kiss1 animals thereby exhibiting a behavioral gene-dosage/threshold effect. While reward-seeking behaviors have not been specifically associated with Kiss1 ARH neurons, NPY/AgRP and POMC neurons are involved in hedonic feeding in addition to controlling homeostatic food intake (Betley et al., 2013;Hayward et al., 2006;Lippert et al., 2014;Pandit et al., 2014,Pandit et al., 2015Rubinstein and Low, 2017;Stuber and Wise, 2016) Although it is clear that homeostatic feeding is regulated by NPY/AgRP and POMC neurons, the motivational drive to eat palatable foods are more complex behaviors that are regulated by multiple interconnected neural systems (Chen et al., 2016;Denis et al., 2015;Fenselau et al., 2017;Rossi and Stuber, 2018). Therefore, further analyses of the neural circuits underlying these physiological responses are essential for understanding such complex behavior. Regardless, our findings have revealed a novel role for glutamate neurotransmission by Kiss1 ARH neurons ( Figure 17), which appears to be sex-specific based on our findings that glutamate signaling is regulated differently in female and male  mice. (1:10,000)

Continued on next page
Kiss1 Cre:GFP version 1 (V1) and Kiss1 Cre:GFP V2 The offspring from Kiss1 Cre;GFP V1 animals crossed with a reporter line like lacZ, sometimes showed ectopic expression (Gottsch et al., 2011). Subsequently, the Palmiter group produced Kiss1 Cre:GFP V2 animals in which the Cre was less efficient (Padilla et al., 2018), thereby significantly decreasing the likelihood of ectopic expression (Song and Palmiter, 2018). However, Kiss1 Cre:GFP V2 animals express less GFP and, therefore, the individual Kiss1 neurons are more difficult to see in a slice preparation unless injected with Cre-dependent ChR2-mCherry (or ChR2-YFP) to label Kiss1 neurons. Currently, we have used both Kiss1 Cre:GFP lines and we have found no differences between the two in terms of the molecular, electrophysiological and optogenetic characterization.

AAV delivery
Viruses were prepared at the University of Washington according to published methods (Gore et al., 2013). Fourteen to twenty-one days prior to each experiment, Kiss1 Cre:GFP , Kiss1 Cre: GFP, V2, Kiss1 Cre:GFP ::Pomc EGFP or Kiss1 Cre:GFP ::Npy GFP female mice (>60 days old) received bilateral ARH injections of a Cre-dependent adeno-associated viral (AAV; serotype 1) vector encoding ChR2 fused to yellow fluorescent protein, YFP (AAV-EF1a-ChR2:YFP) or mCherry (AAV-EF1a-ChR2:mCh) as described . Briefly, using aseptic techniques, anesthetized mice (1.5% isoflurane/O 2 ) were placed in a Kopf stereotaxic apparatus (Tujunga, CA) and received a medial skin incision to expose the surface of the skull. Two holes were drilled into the skull at designated coordinates from bregma (x: ± 0.30 mm; y: -1.200 mm). For the viral injections, a glass micropipette (Drummond Scientific #3-000-203-G/X; Broomall, PA) was fabricated with a Narishige PE-2 puller and beveled (tip diameter = 45 mm), filled with mineral oil and loaded with an aliquot of AAV using a Nanoject II (Drummond Scientific). The pipette tip was positioned at x: -0.30 mm; y: -1.200 mm and lowered to z: -5.800 mm (surface of brain z = 0.0 mm). The AAV (2.0 Â 10 12 particles/ml) was injected at a rate of 100 nl/min (300 nl total), raised to À5.700 mm for a second injection (200 nl total) and then left in place for 10 min post-injection before the pipette was slowly removed from the brain. The other hemisphere injection was done at the same y and z coordinates, but at x: +0.30 mm. The skin incision was closed using skin adhesive, and each mouse received analgesia (Carprofen; 5 mg/kg; subcutaneous) for two days post-operation. These arcuate injections only labeled Kiss1 cell bodies in the ARH as evaluated using immunocytochemistry and scRT-PCR (see Figure 6)   .

Ovariectomy (OVX) and estradiol treatment
We were interested in exploring the actions of preovulatory levels of 17b-Estradiol (E2) on neurotransmitter expression, neuronal activity and behavior in females. Therefore, we used an E2-treatment regimen that we have documented induces a preovulatory surge of luteinizing hormone (LH) in GnRH GFP and Kiss1 GFP females Zhang et al., 2013). When necessary, at least seven days prior to each experiment, ovaries were removed as described previously while under isoflurane inhalation anesthesia (Piramal Enterprises Limited, Andhra Pradesh, India) (Zhang et al., 2013). Each mouse received analgesia (Carprofen; 5 mg/kg; subcutaneous) on the day of operation. 17b-Estradiol benzoate (E2) treatments were as described previously (Zhang et al., 2013). Briefly, each animal was injected on day 5 following OVX with 0.25 mg E2, followed on day 6 with 1.50 mg E2 and used for experiments on day 7. High circulating (proestrous) levels of 17b-estradiol (E2) were verified by the uterine weights (>100 mg) at the time of the hypothalamic slice preparation.
qPCR and scRT-PCR qPCR and scRT-PCR were conducted as previously described . Briefly, the ARH was microdissected from 240 mm basal hypothalamic coronal slices (3-4 slices/per mouse) from OVX oil-and estradiol-treated female Kiss1 Cre:GFP mice (n = 5-7 animals/group; qPCR), POMC EGFP , and NPY/AgRP GFP mice (n = 5 animals each/group; qPCR). The dispersed cells were visualized using a Leitz inverted fluorescent microscope, patched, and then harvested with gentle suction to the pipette using a Xenoworks digital micromanipulator system (Sutter Instrument; Novato, CA) and expelled into a siliconized 0.65 ml microcentrifuge tube containing a solution of 1X Invitrogen Superscript III Buffer, 15 U of RNasin (Promega), 10 mM dithiothreitol and diethylpyrocarbonate-treated water in a total of 5 ml for a single cell (1 cell/tube for scRT-PCR) or 8 ml for pooled cells (5 cells/tube for qPCR). cDNA synthesis was performed on single cells and pools of 5 cells as previously described  and stored at À20˚C. Controls included non-fluorescent cells, artificial CSF (aCSF), single cells without reverse transcriptase (RT) and tissue controls with and without RT. Primers for the genes that encode for Kiss1 (Kiss1), NKB (Tac2), NKB receptor (Tacr3) Dynorphin (Pdyn), POMC (Pomc), NPY (Npy), AgRP (Agrp), vGluT2 (Slc17a6), vGAT (Slc32a1), mGlur1 (Grm1), mGlur2 (Grm2) mGlur3 (Grm3), mGlur5 (Grm5), mGlur7 (Grm7), GAPDH (Gapdh), b-actin (Actb), Ca V 3.1 (Cacna1g), HCN1 (Hcn1) and HCN2 (Hcn2) were designed using Clone Manager software (Sci Ed Software) to cross at least one intron-exon boundary and optimized as previously described Nestor et al., 2016). See Table 1 for primer sequences and optimal annealing temperature for all genes, as well as amplification efficiency for each gene used for quantitative analysis. Note that for qPCR, the annealing temperature for all genes was set at 60˚C. Primers for qPCR were further tested for efficiency (E = 10 (À1/m) -1; Table 1) (Livak and Schmittgen, 2001;Pfaffl, 2001). qPCR was performed on a Quantstudio 7 Flex Real-Time PCR System (Life Technologies) using the Power Sybrgreen (Life Technologies) mastermix method according to established protocols . The comparative DDCT method (Livak and Schmittgen, 2001;Pfaffl, 2001) was used to determine values from duplicate or triplicate samples of 4 ml for the target genes and 2 ml for the reference gene. The relative linear quantity was determined using the 2 -DDCT equation . In order to determine the relative expression levels of target genes in Kiss1 ARH neurons obtained from OVX and estradiol-treated animals, the mean DCT for the target genes from the OVX female samples were used as the calibrator, and the data were expressed as nfold change in gene expression normalized to the reference gene Gapdh (Kiss1 neurons) or Actb (NPY neurons) relative to the calibrator. For quantification differences between the mRNA expression of Tac2 (NKB), Pdyn, Kiss1 and Slc17a6 in Kiss1 ARH neurons, the mean D CT for the target gene Tacr3 from the OVX female samples were used as the calibrator. For quantification of Grm7 in NPY/ AgRP neurons, the mean DCT for the target genes from the OVX female samples were used as the calibrator as described above. scRT-PCR was performed on 3 ml of cDNA in a 30 ml reaction volume and amplified 50 cycles using a C1000 Thermal Cycler (Bio-Rad).The PCR product was visualized with ethidium bromide on a 2% agarose gel.

Estrous cycle and fertility
Prior to monitoring the estrous cycle, females were habituated to handling for at least 3 days, and during the study, each cage was only handled by the investigator. To track the estrous cycle, a vaginal lavage with sterile 0.1M PBS was performed daily between 0700 hr to 0800 hr, and evaluated by light microscopy for the appearance of: cornified epithelial cells, leukocytes, and nucleated epithelial cells. The loss of leukocytes and presence of an abundance of nucleated cells was scored as proestrus, while the transition to primarily cornified epithelial cells was scored as estrus. Note one acyclic mouse was excluded from both the wild type and homozygous groups. For fertility measurements, females were mated with a sexually-experienced wild type male on day 0. Litters were tracked from day 16 through 30. The length of time until conception was calculated as the number of days until parturition minus a 20 day gestation period. Note, two wild type and two homozygous animals were excluded because they did not give birth within the 30 day trial.
Conditioned place preference (CPP) behavioral assay for vGluT2 deleted and control Kiss1 animals Slc17a6 Het (Slc17a6 +/lox ::Kiss1 Cre:GFP/+ , n = 4) and KO (Slc17a6 D/lox ::Kiss1 Cre:GFP/+ , n = 8) females were initially injected with AAV1-DIO-ChR2:YFP in the arcuate nucleus as described above in order to do optogenetic experiments in vitro following in vivo conditioned place preference (CPP) tests. Control Kiss1 (Kiss1 Cre:GFP/+ ) females (n = 8), destined for CPP analysis, received sham surgery, but without viral injection. In addition, we prepared another set of control females (Kiss Cre:GFP/+ , n = 7), which similar to Slc17a6 Het and KO females received viral injection prior to the CPP test. Both Kiss1 control groups exhibited similar results in the CPP test, indicating that the presence of unstimulated ChR2::YFP fusion protein had no effect on behavior.
The animals were OVX at the time of viral injections or sham surgery and received equal cyclical (~every 3-5 days) priming 0.25 mg and/or a surge 1.0-1.5 mg dose of 17b-estradiol benzoate (E2)  prior to and throughout the CPP procedure (see Figure 15). The E2-treatment started on days 5-6 following OVX with a 2 mg dose, followed 6-7 days later with another 2 mg dose during the recovery period from viral or sham injections. Thereafter, the cyclical E2-treatment was initiated with a 0.25 mg dose followed the day after with 1.5 mg E2, a treatment shown to induce a LH surge in GnRH mice  and Kiss1 mice (Zhang et al., 2013). As indicated in Figure 15, we used this E2-treatment immediately prior to the baseline place preference (BPP; initiated 18-20 days after OVX) and prior to the CPP in all groups. The animals were handled daily over a 7 day period prior to the CPP procedure in order to adapt to the investigator and minimize handling stress. The protocol for inducing CPP with sucrose consisted of four phases over the course of 11 days (sucrose habituation, a BPP test, sucrose conditioning, and a CPP test) as depicted in Figure 15. During sucrose habituation (Day À1), mice received sucrose in their home cage to prevent neophobia. Animals received 0.5 g of sucrose reward pellets (TestDiet, Richmond, IN) in a scintillation vial lid left in cages overnight. The next morning cages were changed so no sucrose remained. All behavior testing was conducted in a separate room kept on the same light schedule as the colony room. Mice were transferred into the animal testing room 1 hr prior to dark onset (1700 hr). The CPP behavioral assay was conducted in the dark cycle (beginning at 1800 hr). During the baseline place preference test (Day 1), mice were placed in the central chamber of a three-chamber place preference apparatus (Product Number: MED-CPP-3013, Med Associates Inc, Fairfax, VT) to determine pretest values. Doors were left open for mice to explore all three chambers for 15 min. An unbiased assignment was used where mice for each experimental group Kiss1 Cre:GFP controls, vGluT2 KO (Slc17a6 D/lox ::Kiss1 Cre:GFP/+ ), and heterozygous vGluT2 animals (Slc17a6 +/lox ::Kiss1 Cre: GFP/+ ) were subdivided into counterbalanced subgroups based on their assigned conditioning chamber (white or black chamber) (Cunningham et al., 2003(Cunningham et al., , 2006. During the sucrose conditioning phase, mice received sucrose-filled lids (CS+,Days 2,4,6,8) in one chamber or empty lids Days 3,5,7,9) in the other chamber on alternating days ( Figure 15). Mice were confined to a chamber (black or white) with access to sucrose (0.5 g) or without it for 30 min. Lids were stabilized in the chamber with heavy-duty magnets. We had separate sets of lids and magnets for paired (sucrose access) and unpaired days (no sucrose access). To test for acquisition of a CPP for sucrose (Day 10), mice were placed in the central chamber of the apparatus with door open for 15 min. Time spent in the sucrose-paired chamber was recorded pretest (Day 1) and posttest (Day 10). After each animal, the chambers were wiped down with 70% ethanol and paper towels located at the bottom of cages were replaced. At the end of the testing period, animals were returned to the colony room. All animals had free access to normal chow in their home cage.

Imaging
Photomicrographs of ARH YFP or mCherry expression were acquired using an Olympus BX51W1 upright microscope equipped with a Rolera XR Fast 1394 camera and a Nikon E800 fluorescent microscope equipped with DS-U2 camera. Confocal photomicrographs were acquired using a Zeiss LSM 510 and a Zeiss LSM 780 confocal microscopes, each equipped with 20x and 40x (NA 0.8) APO objectives with Zen software.

Data Analysis
For qPCR three to seven Kiss1 or NPY/AgRP neuronal pools (5 cells/pool) from each animal were run in duplicate or triplicate for mRNAs that encode for Kiss1, Tac2, Pdyn, Tacr3, vGluT2, vGAT, Ca V 3.1, HCN1, HCN2, mGluR7 and GAPDH or b-actin and the mean value of each gene from each animal was used for statistical analysis. Data are expressed as Mean ± SEM and were analyzed using a oneway ANOVA and Tukey's multiple comparison post-hoc test, or an unpaired student's t-test. Since mRNA for vGAT was below the level of detectability (CT 36) in ARH Kiss1 neurons, we report it as not detectable. For scRT-PCR the number of Kiss1 GFP , POMC EGFP and NPY/AgRP GFP neurons expressing each transcript was counted for each animal and the mean number of neurons/animal was determined and used for further analysis of Mean, SEM, and percentage expression. Electrophysiology data are expressed as Mean ± SEM and were analyzed using either an unpaired Student's t-test (paired-pulse experiment) or a one-way ANOVA and Newman-Keuls multiple comparison post-hoc test (POMC and AgRP pharmacology experiments).
Behavioral data are depicted as Mean ± SEM. CPP and sucrose consumption were analyzed using a two-way ANOVA (experimental group as the between-subject factor and CPP protocol days, pretest and posttest, as the within-subject factor) followed by Bonferroni multiple comparison post-hoc test. Body weight gain over the 10 day CPP period was analyzed using a one-way ANOVA followed by Tukey's multiple comparison post-hoc test.