Electrophysiological properties and projections of lateral hypothalamic parvalbumin positive neurons

Cracking the cytoarchitectural organization, activity patterns, and neurotransmitter nature of genetically-distinct cell types in the lateral hypothalamus (LH) is fundamental to develop a mechanistic understanding of how activity dynamics within this brain region are generated and operate together through synaptic connections to regulate circuit function. However, the precise mechanisms through which LH circuits orchestrate such dynamics have remained elusive due to the heterogeneity of the intermingled and functionally distinct cell types in this brain region. Here we reveal that a cell type in the mouse LH identified by the expression of the calcium-binding protein parvalbumin (PVALB; LHPV) is fast-spiking, releases the excitatory neurotransmitter glutamate, and sends long range projections throughout the brain. Thus, our findings challenge long-standing concepts that define neurons with a fast-spiking phenotype as exclusively GABAergic. Furthermore, we provide for the first time a detailed characterization of the electrophysiological properties of these neurons. Our work identifies LHPV neurons as a novel functional component within the LH glutamatergic circuitry.


Introduction
The hypothalamus contains a diverse collection of intermingled cell types defined by the expression of classical neurotransmitters and neuropeptides [1,2]. While extensive research has been done on hypothalamic neurons that express neuropeptides such as agouti-related peptide (AGRP), pro-opiomelanocortin (POMC) [3][4][5][6], hypocretin (orexin; HCRT), melaninconcentrating hormone (MCH) [7,8], or on larger populations defined by neurotransmitter expression (e.g. vesicular GABA (γ-aminobutyric acid) transporter (SLC32A1 commonly known as VGAT) and vesicular glutamate transporter 2 (SLC17A6 commonly known as VGLUT2)) [9], much less attention has been given to a small collection of neurons expressing the calcium-binding protein parvalbumin (PVALB; PV neurons) in the lateral hypothalamus (LH PV neurons) [10,11]. Throughout the central nervous system, PV-expressing neurons are typically GABAergic interneurons with fast-spiking characteristics [12] that are necessary and sufficient for the generation of network oscillations in both the neocortex and hippocampus [13,14]. However, previous studies using immunohistochemistry and gene expression analysis showed that within the LH, PV-expressing neurons colocalize with glutamate in both rats and mice [11,15]. Nevertheless, quantitative measurements of the numbers of co-expressing neurons and their ability to release glutamate and form functional excitatory synapses have yet to be determined. Furthermore, the fundamental electrophysiological properties, synaptic connections, and the functional roles of LH PV neurons have remained largely uncharacterized. Here, we used a combination of selective targeting, electrophysiology, single-cell RT-qPCR, and in situ hybridization assays to characterize the intrinsic properties of LH PV neurons and map their axonal projections.

Animals
All experimental protocols were conducted in accordance with U.S. National Institutes of Health Guidelines for the Care and Use of Laboratory Animals and with the approval of the National Institute on Drug Abuse Animal Care and Use Committee. Two-to five-monthold male and female C57BL/6J (wild-type, Strain 664, The Jackson Laboratory, ME, USA), Pvalb IREScre (C57BL/6 background, Strain 8069, The Jackson Laboratory), and Pvalb IREScre crossed with Rosa26 LSL-tdTomato (C57BL/6 background, Strain 7909, The Jackson Laboratory) mice were used in this study. Prior to stereotaxic viral injection, mice were group housed with littermates in temperature-and humidity-controlled rooms with ad libitum access to water and rodent chow (PicoLab Rodent Diet 20, 5053 tablet, LabDiet/Land O'Lakes Inc., MO, USA) on a 12 h light/dark cycle.
For mapping LH PV axonal projections, eight-week-old Pvalb IREScre heterozygous mice were bilaterally injected with 30 nl of an adeno-associated virus (rAAV2/9-hEF1α-DIO-hSyn-mCherry; titer: 1.0×10 13 genomic copies/ml; Massachusetts Institute of Technology Viral Gene Transfer Core, Boston, MA, USA) [16] into the lateral hypothalamus as described above. After surgery, mice were individually housed for six weeks for post-surgical recovery and viral transduction.

Slice preparation and electrophysiology
After cervical dislocation, mice were decapitated and their brains were rapidly removed and placed into an ice-cold N-methyl-D-glucamine (NMDG)-based slicing solution [17]  Characterization of the intrinsic electrophysiological properties of lateral hypothalamic parvalbumin (LH PV ) neurons (n = 34) was performed using Pvalb IREScre ;Rosa26 LSL-tdTomato mice (Pvalb cre/+ ;Rosa26 tom/tom ). Parvalbumin-tdTomato-positive lateral hypothalamic neurons were located in brain slices, first with epifluorescence, followed by infrared differential interference contrast (IR-DIC) optics, using an upright Olympus BX51WI microscope (Olympus Corporation, MA, USA). Whole-cell current-clamp recordings were performed using a MultiClamp 700B amplifier (5 kHz low-pass Bessel filter and 10 kHz digitization using a 1440A Digidata Digitizer) with pClamp 10. For channelrhodopsin (ChR2)-assisted circuit mapping (CRACM) of neurons synaptically connected to LH PV neurons, Pvalb IREScre heterozygous mice were bilaterally injected with an adeno-associated virus into the LH as previously described. Horizontal slices containing the LH from AAV-injected Pvalb IREScre mice were used and ChR2:tdTomato-containing axons visualized in the LH. Whole-cell voltage-clamp recordings of parvalbumin-negative LH neurons (n = 75) were performed using patch pipettes (3.0-4.5 MΩ) containing (in mM): 117 cesium methanesulfonate, 20 HEPES, 0.4 EGTA, 2.8 NaCl, 5 TEA-Cl, 4 Mg-ATP, 0.4 Na-GTP, 3 QX-314, and 0.2% biocytin (pH adjusted to 7.3 using CsOH, and osmolality of 287 mOsm/ kg H 2 O). Pipette capacitance was compensated for immediately after the pipette was placed into the recording solution, and whole cell capacitance was compensated for after obtaining whole cell access. To compensate for cell membrane depolarization associated with cesium methanesulfonate-based solutions, holding currents ranging from −15 pA to −40 pA were applied, and the series resistance compensation was at least 70%. Series resistance (10-25 MΩ) was monitored with a -5 mV hyperpolarizing pulse given every 10 s, and only recordings that remained stable over the period of data collection were used. Recorded cells were held at -70 mV and photocurrents were evoked by 1 ms blue (473 nm) light pulses (diode-pumped solidstate laser; OptoEngine LLC, UT, USA) delivered at a frequency of 0.1 Hz to determine synaptic connectivity. Light-evoked glutamatergic currents were blocked by perfusing the ionotropic glutamate receptor antagonist, 6,7-dinitroquinoxaline-2,3-dione (DNQX; 10 μM) and D-(-)-2-amino-5-phosphonopentanoic acid (D-AP5; 50 μM). The liquid junction potential for these measurements was not corrected. All recordings were made at 32˚C. All chemicals were obtained from Sigma-Aldrich (MO, USA) or Tocris Bioscience (Bristol, UK).

Electrophysiological analysis
Intrinsic membrane properties of LH PV neurons were characterized in current-clamp configuration as previously described [18,19]. Briefly, the resting membrane potential (V rmp ) and capacitance of the cell membrane (C mem ) were measured directly after obtaining whole-cell configuration. The action potential (AP) properties were determined from the first AP evoked by applying 500 ms depolarizing current steps in a range of 20-100 pA in 20 pA increments. The respective AP parameters were determined from the first evoked AP. Both the action potential (AP) and fast after-hyperpolarization (fAHP) amplitudes were determined relative to the AP threshold (i.e. membrane potential (V m ) at which dV m /dt first reached 20 V/s). The fAHP latency was determined as the time difference between the AP threshold level and the largest fAHP peak. AP latency was measured as the time difference between the current onset time and the time when the AP peak was reached, while the AP half-width (HW) was determined as the halfway duration between AP peak and the AP threshold level. AP broadening was measured at twice current amplitude of the first evoked AP and calculated according to (HW2 -HW1)/HW1, where HW1 and HW2 correspond to the HW of the first and second APs, respectively. Further membrane properties and firing rate were determined by applying 500 ms current step injections ranging from −100 pA to 1500 pA in 20-50 pA increments. Input resistance (R in ) was calculated from the slope of a linear regression fit to the steady-state voltage-current relation using a range of hyperpolarizing current steps. The membrane time constant (τ) was determined by single exponential fit to 63.2% of the rising phase of a mean voltage response to a −100 pA current step. The sag ratio was determined according to V steady / V hyp , where V steady and V hyp correspond to the voltage response measured at the end of a −100 pA hyperpolarizing current step and the hyperpolarization peak, respectively. The maximal firing frequency was determined as the spike frequency response to the largest depolarizing current step (500 ms) below that where AP failures were observed. To generate the I-f curves, we measured the firing rate by counting the number of spikes elicited in response to step depolarizing current injections during a 500 ms window. Note that each neuron was tested for a subset of current amplitudes from 0−1200 pA.
Images were taken with an AxioZoom.V16 fluorescence microscope and z-stacks were collected using an LSM700 laser scanning confocal microscope (Carl Zeiss Microscopy LLC, NY, USA). Stacks were imported into Vaa3D 3D visualization-assisted analysis software [20], and fluorescent neurons were counted bilaterally from every section (n = 3 mice).

Radioactive in situ hybridization and immunohistochemistry
Wildtype mice were deeply anesthetized with isoflurane and transcardially perfused with 0.1 M phosphate buffer (PB) followed by 4% PFA in 0.1 M PB. Whole brains were removed and post-fixed in 4% PFA/PB for 2 h at 4˚C. Samples were washed with 0.1 M PB (2 × 30 min each) at 4˚C, and then transferred to 18% sucrose in 0.1 M PB. Free-floating coronal cryosections (14 μm) were sliced using a Leica CM3050 S cryostat. In situ hybridization was performed as previously detailed [21,22]. Steps are at room temperature unless otherwise noted. Sections were imaged with brightfield and epiluminescence microscopy using a Leica DMR microscope with a 20× objective and cellSens Standard v1.11 software (Olympus Corporation). Neurons observed within the LH PV region were manually assessed for the co-expression of Vglut2 mRNA or Vgat mRNA with anti-parvalbumin immunolabeled cells. Brightfield was used to determine whether a parvalbumin-immunolabeled (brown DAB product) neuron contained the aggregates of silver grains for Vglut2 mRNA or Vgat mRNA, which were viewed under epiluminescence.

Single-cell gene expression profiling and analysis
Single-cell cytoplasm harvesting was performed as previously described [23]. Brain slices were prepared from Pvalb cre/+ ;Rosa26 tom/tom mice as described above. First, the fast-spiking firing pattern of an LH PV tdTomato-positive neuron was recorded during whole-cell configuration. Subsequently, the cytoplasm of the recorded neuron was harvested into the recording pipette. The stability of the gigaseal (i.e. seal between the neuron and the pipette) was constantly monitored to avoid extracellular contamination. The total recording time and harvesting of the intracellular content did not exceed more than 4 min. The content of the pipette tip containing the harvested cytoplasm was then expelled into an RNase-free PCR tube. For positive control samples, we patched fast-spiking parvalbumin-expressing (tdTomato-positive) basket cells in the hippocampus and performed the same procedure as described above. Single-cell extracts were processed using the Ambion 1 Single Cell-to-CT Kit (Thermo Fisher Scientific, MA, USA). Briefly, samples were incubated in lysis solution with DNase I. Reverse transcription was then performed followed by preamplification of all target genes using pooled TaqMan Gene Expression Assays (Thermo Fisher Scientific). Target genes in LH PV cells were detected using the recommended best coverage assays for Pvalb (Mm00443100_m1), Slc17a6 (Vglut2, Mm00499876_m1), Slc32a1 (Vgat, Mm00494138_m1), Kcnc1 (Kv3.1, Mm00657708_m1), Kcnc2 (Kv3.2, Mm01234232_m1), and Hcn2 (Mm00468538_m1). qPCR was performed in 10 μl reactions in 96-well plates using an Applied Biosystems 7500 Fast Real-Time PCR System (Invitrogen, CA, USA) with the following cycling parameters: (1) 50˚C for 2 min, (2) 95˚C for 10 min, and (3) 40 repeats of 95˚C for 15 s, 60˚C for 1 min. Reactions included TaqMan Gene Expression Master Mix, TaqMan probe, and cDNA according to manufacturer's protocol (Thermo Fisher Scientific). Each plate was run with a negative control (no cDNA template) and a positive control (hippocampal basket cell positive for Pvalb, Vgat, Kv3.1, Kv3.2, and Hcn2). Technical replicates (triplicate) were performed for each sample-gene combination. Cycle threshold (Ct) values were determined using Applied Biosystems 7500 v2.0.6 software (Invitrogen). Amplification Ct values higher than 37 and samples lacking any amplification curves were designated below the limit of detection. Three out of eight LH PV cells lacked amplification curves for Kv3.2. These samples were excluded from the Kv3.2 analysis. For all other target genes, n = 8 cells. Gene expression was normalized to Pvalb and Vglut2 using the 2 −ΔΔCt method [24].

Statistical analysis
Data are reported as mean ± s.e.m. or mean ± s.d. unless otherwise noted. Individual data points are shown for qPCR. Data from electrophysiological recordings were analyzed with Clampfit v10.6 (Molecular Devices LLC, CA, USA), Origin Pro v9.2 (OriginLab Corporation, MA, USA) and MATLAB R2015A (The MathWorks Inc., MA, USA). Analyses were performed using the Analysis ToolPak of Microsoft Excel 2016 and OriginPro v9.2.

LH PV neurons display a fast-spiking action potential phenotype
We first characterized the electrophysiological properties of LH PV neurons under currentclamp conditions. These neurons form a compact and small cluster in the LH medial to the optic tract (340 ± 10 neurons, n = 3 mice; bilateral; Fig 1A), and the number of cells was consistent with previous studies estimating~400 PV-immunoreactive neurons in the LH of mice [11]. We performed whole-cell current clamp recordings from LH PV neurons (n = 34) identified by tdTomato fluorescence in brain slices sectioned horizontally from Pvalb IREScre ; Rosa26 LSL-tdTomato mice and observed that these neurons fired action potentials at high-frequency and with little accommodation in response to depolarizing current injections (Fig 1B). This fast-spiking action potential phenotype and other electrophysiological characteristics such as the resting membrane potential (V rmp ), action potential half-width (AP HW ), and the maximal firing frequency (Table 1) are similar to the properties of both hippocampal and neocortical PV-positive GABAergic interneurons [13,14]. Previous studies showed that fastspiking interneurons can be characterized by the expression of a specific combination of ion channels that confer these electrophysiological properties as for example delayed rectifier voltage-gated potassium channels KCNC1 (KV3.1) and KCNC2 (KV3.2) and hyperpolarizationactivated cation channels (HCNs) [12]. Therefore, we sought to determine whether the expression of these channels is similar in LH PV neurons. Single-cell RT-qPCR analysis revealed that LH PV neurons express Kv3.1, Kv3.2, and Hcn2 subunit genes with comparable relative abundance (Fig 2A and 2B). Together, these results demonstrate that LH PV neurons are equipped with ion channels that are implicated in setting precise pacing of spiking activity to minimize accommodation during prolonged depolarization with some electrophysiological diversity (Fig 2C and 2D) possibly attributable to the variations in the relative contributions of specific ion channels.

LH PV neurons release the excitatory neurotransmitter glutamate and provide excitatory inputs onto neurons within the LH
Thus far, our results indicate that LH PV neurons have several features in common with PVpositive GABAergic interneurons in other brain regions. Therefore, we next sought to determine whether LH PV neurons are GABAergic. Surprisingly, previous studies have shown that parvalbumin can be found colocalized with glutamate immunohistochemically in both, rat and mouse LH [11,15]. However, a quantification of the number of LH PV neurons expressing specific markers for glutamate and their ability to release the neurotransmitter and form functional synapses have not been determined. Therefore, we used several approaches to determine the neurotransmitter used by LH PV neurons. We first performed channelrhodopsin (ChR2)assisted circuit mapping (CRACM) [25,26] to examine whether LH PV neurons are synaptically connected to other cells within the LH (Fig 3A). We stereotaxically injected a Cre recombinase-dependent viral vector [26] bilaterally into the LH of Pvalb IREScre transgenic mice to target channelrhodopsin-2 (ChR2) fused to the fluorophore tdTomato (ChR2:tdTomato) specifically to LH PV neurons. We next performed whole-cell recordings from individual PV-negative neurons within the LH (n = 75) under voltage-clamp configuration ( Fig 3A) and observed that photostimulation of ChR2-expressing LH PV neurons and axons evoked excitatory postsynaptic currents (EPSCs; 69.0 ± 5.0 pA) in synaptically connected LH neurons (n = 13; 17.33% connected). These EPSCs were also blocked by bath applied selective antagonists of glutamate receptors (2.5 ± 0.4 pA). This demonstrates that rather than releasing GABA, LH PV neurons release the excitatory neurotransmitter glutamate. To determine whether LH PV neurons contained markers of glutamate or GABA neurons, we performed in situ hybridization assays to measure the expression of Slc17a6 (Vglut2; vesicular glutamate transporter 2) and Slc32a1 (Vgat; vesicular GABA transporter) in these neurons (Fig 3B and 3D). We observed that Pvalb mRNA was predominantly detected in neurons that express Vglut2 (95% Pvalb + /Vglut2 + ; 5% Pvalb + /Vgat + ; n = 4). Together, these results describe a cluster of glutamatergic fast-spiking LH PV neurons that provide excitatory inputs to neurons within the LH and send long range projections throughout the brain (n = 3, Fig 4A-4H).

Discussion
The roles of specific, genetically-distinct lateral hypothalamic neuronal types in regulating circuit function have not been thoroughly identified. Here, we performed for the first time a comprehensive characterization of the electrophysiological properties of a subclass of LH neurons known as LH PV cells, and we implicate them in the regulation of activity dynamics within the LH. We find that LH PV neurons exhibit a fast-spiking action potential phenotype and have electrophysiological characteristics similar to hippocampal and neocortical parvalbumin-positive GABAergic interneurons [13,14]. Moreover, like those neurons, we reveal that LH PV cells express a combination of ion channels that confer these electrophysiological properties [12]. Our single-cell RT-qPCR analysis demonstrates that LH PV neurons are equipped with ion channels that are implicated in setting and regulating fast-spiking activity [12]. Interestingly, heterogeneous firing patterns were observed in a preliminary study attempting to record spontaneous activity from LH PV neurons extracellularly in anaesthetized rats [28]. However, these experiments were performed with traditional extracellular electrophysiology methods in which an intrinsic limitation is the accurate identification of the recorded cell types. Therefore, further work using a combination of optogenetics and electrophysiological methods is needed to determine the activity patterns of LH PV neurons in vivo and during behavior.
In the LH, immunohistochemical colocalization of parvalbumin and glutamate has been reported [11,15]. However, here we provide a quantitative measurement of the number of glutamatergic LH PV neurons. Furthermore, we demonstrate that these cells release glutamate and provide excitatory control of local neuronal circuits within the LH. Remarkably, our findings and those of others [29,30], challenge long-standing conceptualizations that fast-spiking neurons are exclusively GABAergic, suggesting conservation of the fast-spiking phenotype across at least two neurotransmitter systems.
Our results showing that LH PV neurons send long range axonal projections to several brain regions, including the lateral habenula (LHb), submedius thalamic nucleus (Sub), retromamillary nucleus (RMM), and periaqueductal gray (PAG) provide further support for the idea that the LH and its diverse collection of genetically-distinct cell types are crucial for orchestrating a variety of motivated behaviors. In particular, the Sub has been implicated in learning and decision making [31], and the PAG mediates fear-related freezing behavior [32]. Interestingly, a recent study showed that glutamatergic lateral hypothalamic inputs to the LHb regulate reward  and feeding-related behaviors [33]. However, as all glutamatergic neurons in the LH were targeted during the study, the contribution of specific genetically-distinct glutamatergic LH neurons is yet to be determined. Thus, it is possible that LH PV neurons regulate circuit function through connections in the LHb. Further experiments using a combination of optogenetics and behavioral assays are needed to determine whether LH PV neurons regulate reward-related behaviors. Similarly, further work is needed to determine functional inputs to LH PV neurons. Inhibitory inputs specifically innervating and suppressing glutamatergic neurons in the LH have been shown to promote feeding behaviors [9]. Based on what we have just described, our work identifies LH PV neurons as a novel functional component within the LH glutamatergic circuitry. Thus, these cells could possibly regulate consummatory and appetitive behaviors.
In summary, our work has revealed that LH PV neurons form functional excitatory synapses with other lateral hypothalamic neurons, which directly implicates LH PV neurons in the regulation of activity dynamics within the LH. Thus, our findings will serve as a basis for future models of lateral hypothalamic circuitry that regulate behaviors essential for survival.