Efferent projections of Nps-expressing neurons in the parabrachial region

Abstract

In the brain, connectivity determines function. Neurons in the parabrachial nucleus (PB) relay diverse information to widespread brain regions, but the connections and functions of PB neurons that express Nps (neuropeptide S) remain mysterious. Here, we use Cre-dependent anterograde tracing and whole-brain analysis to map their output connections. While many other PB neurons project ascending axons through the central tegmental tract, NPS axons reach the forebrain via distinct periventricular and ventral pathways. Along the periventricular pathway, NPS axons target the tectal longitudinal column and periaqueductal gray, then continue rostrally to target the paraventricular nucleus of the thalamus. Along the ventral pathway, NPS axons blanket much of the hypothalamus but avoid the ventromedial and mammillary nuclei. They also project prominently to the ventral bed nucleus of the stria terminalis, A13 cell group, and magnocellular subparafasciular nucleus. In the hindbrain, NPS axons have fewer descending projections, targeting primarily the superior salivatory nucleus, nucleus of the lateral lemniscus, and periolivary region. Combined with what is known already about NPS and its receptor, the output pattern of Nps-expressing neurons in the PB region predicts roles in threat response and circadian behavior.

Graphical Abstract

We mapped the efferent projections of NPS neurons in the Parabrachial Nucleus and Peri-Locus Coeruleus using a genetic approach combined with a neural network algorithm (BoutonNet). NPS neurons project via a periventricular and ventral pathway and target the PVT, BST, DMH, and TLC. Lighter labeling was seen in descending projections to the brainstem.

Introduction

The parabrachial nucleus (PB) is a brainstem region that conveys sensory information to a wide variety of brain regions (Cechetto et al., 1985; Herbert et al., 1990; Saper & Loewy, 1980). Glutamatergic neurons in this region belong to two mutually exclusive macropopulations defined by the transcription factors Atoh1 and Lmx1b (Karthik, Huang, Delgado, Laing, Peltekian, Iverson, Grady, Miller, McCann, Fritzch, et al., 2022). These two macropopulations have distinct projection pathways: Lmx1b neurons project axons to the amygdala and cerebral cortex, primarily via the central tegmental tract, while Atoh1 neurons project axons to the hypothalamus and other targets, primarily via a ventral pathway (Karthik, Huang, Delgado, Laing, Peltekian, Iverson, Grady, Miller, McCann, Fritzch, et al., 2022). Each macropopulation has distinct connections (Huang et al., 2020; Huang et al., 2021; Pauli et al., 2022) and neurons within both macropopulations have distinct functions, which include thermoregulation, pain, itch, sodium appetite, and conditioned taste aversion (Carter et al., 2015; Gasparini et al., 2019; Geerling et al., 2016; Mu et al., 2017; Nakamura & Morrison, 2008, 2010; Shin et al., 2011).

The Atoh1 PB macropopulation includes a subpopulation of neurons that express Nps and produce neuropeptide S (NPS; Huang et al., 2022). The PB region contains two clusters of these neurons – a rostral NPS group near the lateral lemniscus and a caudal NPS group near the locus coeruleus. These rostral and caudal clusters are present in mice (Clark et al., 2011; Huang et al., 2022; Liu et al., 2011) and in humans (Adori et al., 2015). Using Nps Cre-reporter mice, we also found neurons extending rostrally, alongside the lateral lemniscus, and we identified novel populations in the nucleus incertus, lateral habenula, and anterior hypothalamus (Huang et al., 2022).

While little is known about the neurons that produce NPS, possible functions have been inferred from genetic and pharmacologic studies of this neuropeptide and its receptor. In rodents, injecting NPS into brain increases locomotion, arousal, and body temperature, while also reducing food intake and producing analgesia (Ensho et al., 2017; Leonard et al., 2008; Li et al., 2009; Peng, Han, et al., 2010; Peng, Zhang, et al., 2010; Rizzi et al., 2008; Smith et al., 2006; Xu et al., 2004). The NPS receptor (NPSR1) is required for its arousal-promoting effect in mice (Duangdao et al., 2009; Fendt et al., 2011; Ruzza et al., 2012; Zhu et al., 2010), and human genetic studies have linked NPSR1 gain-of-function variants to sleep deprivation, anxiety, asthma, and panic disorder (Domschke et al., 2011; Donner et al., 2010; Gottlieb et al., 2007; Laitinen et al., 2004; Xing et al., 2019).

In contrast to these extensive pharmacologic and genetic findings, the connections and functions of NPS neurons remain unclear. Connectivity determines function, so mapping their axonal projections could predict additional functions and should help focus existing ideas about neural circuit mechanisms of NPS-modulated behaviors. Previous studies identified NPS-immunoreactive fibers in the septum, hypothalamus, and midline thalamus (Adori et al., 2016; Clark et al., 2011), but our identification of several clusters of Nps-expressing neurons outside the PB (Huang et al., 2022) made it necessary to clarify which NPS neurons send output to each target.

Given the diverse patterns of axonal projections that arise from separate subsets of PB neurons, elucidating the role of NPS neurons requires a comprehensive map of axonal projections from Nps-expressing neurons in the PB region. The goal of this study is to analyze the projection pattern of NPS neurons in the PB region as a framework for future studies exploring their function. To accomplish this goal, we combined Cre-conditional axonal tracing and presynaptic labeling (Carter et al., 2013; Opland et al., 2013) with semi-automated bouton plotting (Grady et al., 2022) in the Nps-2A-Cre mouse strain we generated and characterized (Huang et al., 2022). Based on their genetic identity as Atoh1-derived neurons, which send output to the hypothalamus (Karthik, Huang, Delgado, Laing, Peltekian, Iverson, Grady, Miller, McCann, Fritzch, et al., 2022), we hypothesized that NPS neurons project axons primarily via a ventral pathway from the PB to the hypothalamus, and not via the central tegmental tract, through which the axons of neighboring, Lmx1b-expressing PB neurons travel to reach the amygdala and cerebral cortex (Huang et al., 2020; Huang et al., 2021; Karthik, Huang, Delgado, Laing, Peltekian, Iverson, Grady, Miller, McCann, Fritzsch, et al., 2022). Our results support this hypothesis and reveal robust axonal projections through an additional, periventricular pathway to a novel pattern of output targets.

Materials and Methods

Mice.

All mice were group-housed in a temperature- and humidity-controlled room on a 12/12-hour light/dark cycle with ad libitum access to water and standard rodent chow. Overall, we used n=7 Nps-2A-Cre, n=4 Nps-2A-Cre;R26-LSL-L10GFP, and n=4 C57BL6/J mice ranging in age from 8 to 10 weeks (20–40 g body weight, male and female). Detailed information about each knockin strain is provided in Table 1. All mice were hemizygous and maintained on a C57BL6/J background. Nps-2A-Cre mice allow expression of Cre-conditional genetic constructs selectively in cells expressing 2A-Cre, which was inserted immediately downstream of the last exon of the endogenous Nps gene (Huang et al., 2022). These mice are available through The Jackson Laboratory (Jax #038113). Crossing Nps-2A-Cre mice to the R26-LSL-L10GFP Cre-reporter strain produces green fluorescence in cells that express Cre throughout the life of the mouse. The ribosomal L10 subunit (Rpl10)-eGFP fusion protein is concentrated in the cell soma. All experiments were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee at the University of Iowa.

Table 1.

Cre driver and reporter mice used in this study.

Strain	Reference	Source information	Key gene
Nps-2A-Cre	Huang, D., Zhang, R., Gasparini, S., McDonough, M. C., Paradee, W. J., & Geerling, J. C. (2022). Neuropeptide S (NPS) neurons: Parabrachial identity and novel distributions. J Comp Neurol, 530(18), 3157–3178. https://doi.org/10.1002/cne.25400	Available from originating investigators	2A-Cre inserted immediately after terminal exon in endogenous neuropeptide S gene
R26-LSL-L10GFP reporter	Krashes, Michael J., et al. “An excitatory paraventricular nucleus to AgRP neuron circuit that drives hunger.” Nature 507.7491 (2014): 238.	Available from originating investigators http://www.informatics.jax.org/allele/MGI:5559562	Floxed transcription STOP cassette followed by EGFP/Rpl10 fusion reporter gene under control of the CAG promoter targeted to the Gt(ROSA)26Sor locus

Stereotaxic injections.

Mice were anesthetized with isoflurane (0.5–2.0%) and placed in a stereotactic frame (Kopf 1900 or 940). We made a midline incision and retracted the skin to expose the skull. Through a pulled glass micropipette (20–30 μm inner diameter), we injected 50–100 nL of AAV8-hEf1a-DIO-synatophysin-mCherry (AAV8-DIO-Syp-mCherry, x 2.5 10¹³ pfu/mL; purchased from Dr. Rachel Neve at the Massachusetts Institute of Technology McGovern Institute for Brain Research Viral Vector Core). In Nps-2A-Cre mice used for formal analysis in the coronal plane, we used separate coordinates to target the rostral or caudal cluster of NPS neurons unilaterally in each case. Rostral coordinates were: 1.90 mm right of midline, 4.80 mm caudal to bregma, and 3.90 mm deep to bregma. Caudal coordinates were: 1.05 mm right of midline, 5.35 mm caudal to bregma, and 4.00 mm deep to bregma. In Nps-2A-Cre;R26-LSL-L10GFP mice used for histology in the sagittal plane, we targeted the rostral PB bilaterally, using the following coordinates: left and right 1.80 mm lateral to midline, 4.80 mm caudal to bregma, and 3.90 mm deep to bregma. Each injection was made over a 5-minute period, using picoliter air puffs through a solenoid valve (Clippard EV 24V DC) pulsed by a Grass stimulator. The pipette was left in place for an additional 3 minutes, then withdrawn slowly before closing the skin with Vetbond (3M). Carprofen (5 mg/kg) was provided prior to incision and again 24 hours later for postoperative analgesia. AAV-injected mice were allowed to survive for 3–5 weeks after surgery to allow for production of Cre-conditional proteins.

Perfusion and tissue sections.

Mice were anesthetized with ketamine (150 mg/kg) and xylazine (15 mg/kg, dissolved with ketamine in sterile 0.9% saline and injected i.p.). They were then perfused transcardially with phosphate-buffered saline (PBS, prepared from 10X stock; P7059, Sigma), followed by 10% formalin-PBS (SF100–20, Fischer Scientific). After perfusion, we removed each brain and fixed them overnight in 10% formalin-PBS. We sectioned each brain into 40 μm-thick coronal slices using a freezing microtome and collected tissue sections into separate, 1-in-3 series. Sections were stored in cryoprotectant solution at −30 °C until further processing.

Immunohistology.

For brightfield labeling (immunohistochemistry), we removed tissue sections from cryoprotectant and rinsed them in PBS. To quench endogenous peroxidase activity, we incubated all tissue sections for 30 minutes in 0.3% hydrogen peroxide (#H325–100, Fisher) in a PBS solution containing 0.25% Triton X-100 (BP151–500, Fisher). After 3 washes in PBS, we loaded sections into a primary antibody solution prepared in PBT-NDS-azide, which is a PBS solution containing 0.25% Triton X-100, 2% normal donkey serum (NDS, 017–000-121, Jackson ImmunoResearch), and 0.05% sodium azide as a preservative (14314, Alfa Aesar). This PBT-NDS-azide solution contained either rabbit anti-dsRed or rat anti-mCherry, and we incubated tissue sections overnight, at room temperature, on a tissue shaker. On the following morning, after 3 PBS washes, we incubated sections for 2 hours in a PBT-NDS-azide solution containing a 1:500 dilution of either donkey anti-rabbit (#711–065-152) or donkey anti-rat (#712–065-153; Jackson ImmunoResearch) secondary antibody. Sections were then washed 3 times and placed for 1 hour in biotin-avidin complex (Vectastain ABC kit PK-6100; Vector), washed 3 times in PBS, and incubated in nickel-diaminobenzidine (NiDAB) solution for 10 minutes. Our stock DAB solution was prepared by adding 100 tablets (#D-4418, Sigma, Saint Louis, MO) into 200 mL ddH2O, then filtering it. We used 1 mL of this DAB stock solution, with 300 μL of 8% nickel chloride (#N54–500, Fisher Chemical) per 6.5 mL PBS. After 10 minutes in NiDAB, we added hydrogen peroxide (0.8 μL of 30% H2O2 per 1 mL PBS-DAB) and swirled sections for 4–5 min until observing dark color change. After two rapid PBS washes, we wet-mounted one or more sections, checked them in a light microscope to ensure optimal staining, and in rare cases replaced sections for up to one additional minute for additional enzymatic staining. Finally, after washing an additional 3 times in PBS, we mounted tissue sections on glass slides (#2575-plus; Brain Research Laboratories). Slides were air-dried, then dehydrated in an ascending series of alcohols and xylenes, coverslipped immediately with Cytoseal 60 (#8310–16 Thermo Scientific) and stored at room temperature until microscope imaging.

For immunofluorescence labeling, we removed tissue sections from cryoprotectant, rinsed them in PBS and loaded them into a PBT-NDS-azide solution containing one or more primary antisera (Table 1). We incubated these sections overnight at room temperature on a tissue shaker. The following morning, the sections were washed 3 times in PBS, then incubated for 2 hours at room temperature in a PBT-NDS-azide solution containing species-specific donkey secondary antibodies. These secondary antibodies were conjugated to Cy3, Cy5, Alexa Fluor 488, or biotin (Jackson ImmunoResearch #s 711–065-152, 711–165-152, 711–175-152, 705–065-147, 705–545-147, 713–545-147, 706–545-148, 706–165-148, 706–065-148, 715–065-15, 712–165-153; each diluted 1:1,000 or 1:500). When biotin was used, sections were washed 3 times in PBS, then incubated for an additional 2 hours in streptavidin-Cy5 (#SA1011; Invitrogen) prepared in PBT-NDS-azide. These sections were then washed 3 times in PBS and mounted on glass slides (#2575-plus; Brain Research Laboratories), then coverslipped using Vectashield with DAPI (Vector Labs). Slides were stored in slide folders at 4 °C until microscope imaging.

Nissl counterstaining.

After whole-slide microscope imaging (described below), all NiDAB-immunolabeled sections were Nissl-counterstained and re-imaged. First, coverslips mounted with Cytoseal were removed by soaking slides in xylenes for up to a week. After rehydrating tissue through 1-minute dips in a graded series of alcohols, we rinsed the slides in water and dipped them in a 0.125% thionin solution (Fisher Scientific) for 1 minute. Slides were then rinsed in distilled water until the solution cleared, and then dehydrated in a series of ethanol solutions: 50% EtOH, 70% EtOH, 400mL of 95% EtOH with 10 drops of glacial acetic acid, 95% EtOH, 100% and 100% EtOH. Finally, after the second of two xylene solutions, slides were coverslipped immediately with Cytoseal.

Microscope imaging and figures.

All slides were imaged using an Olympus VS120 slide-scanning microscope. For brightfield images of NiDAB and then Nissl-counterstained sections, we used a 20x UPLSAPO 20x air objective (numerical aperture 0.75) and the extended focal imaging (EFI) to collect and combine in-focus images from 11 focal planes through the tissue. To collect epifluorescence images, we first used a 10x (NA 0.40) objective, then collected additional EFI or multifocal image stacks at 20x in regions of interest. After reviewing whole-slide images in OlyVIA (Olympus), we collected additional EFI or multifocal image stacks at higher magnifications in some regions of interest.

To plot Syp-mCherry-labeled boutons for illustrations from case 5569, we used BoutonNet (Grady et al., 2022). This method uses full-resolution source images (346 nm/pixel OME/TIFF exports from VS-ASW) and combines two separate algorithms to label boutons: an intensity-based “proposal” algorithm, followed by a convolutional neural-network based “confirmation” algorithm. In separate layers in Illustrator, we aligned the PNG output file containing plotted boutons, the NiDAB source image, and a Nissl-counterstained image of the same section. We used the aligned Nissl cytoarchitecture to trace borders, white matter tracts, and ventricles in an additional illustration layer. We inspected each plot to verify accuracy and remove any automated bouton detection atop histological artifacts rather than actual NiDAB-labeled boutons. Finally, we exported the illustration layer and BoutonNet PNG together as a TIF for the final figure layout.

For Figures containing brightfield or fluorescence images, we first used Adobe Photoshop to crop bitmap images from VS-ASW or cellSens (Olympus), adjust brightness and contrast, and combine raw fluorescence data for multicolor combinations. We used Adobe Illustrator to add lettering, trace scale bars atop calibrated lines from OlyVIA (to produce a clean white or black line) and to make all illustrations.

Whole-brain analysis of Syp-mCherry labeling.

To compare the spatial distribution of neurons transduced in the PB region, we first plotted Syp-mCherry-expressing neurons across five rostrocaudal levels from each case with a brain sliced in the coronal plane (approximately −4.7 to −5.6 mm caudal to bregma). We overlaid these injection site plots in Adobe Illustrator and used images of the same sections after Nissl counterstaining to illustrate brainstem borders and major white matter tracts in each section.

Next, we immunolabeled, imaged, and reviewed every tissue section from a 1-in-3 series of sections through the full brain (olfactory bulbs to cervical spinal cord). We began by identifying all regions with Syp-mCherry-labeled boutons in NiDAB-labeled images (without Nissl counterstaining). We then compared each region side-by-side with Nissl-counterstained images of the same sections to identify cytoarchitectural loci. For each region, we scored the density of Syp-mCherry-labeled boutons using a semi-quantitative scale (0–4), and for brain regions with minimal or “trace” labeling, we assigned an intermediate designation (0.5), as in previous work (Huang et al., 2021). Three independent raters (RZ, DH, & JCG) scored every region from every brain and reviewed the results to reach consensus.

Nomenclature.

We use the term “boutons” to refer to punctate Syp-mCherry labeling. For mouse genes, we used Mouse Genome Informatics (MGI) nomenclature. For mouse proteins, we used common abbreviations from the published literature. For neuroanatomical structures and cell populations, where possible, we used and referred to nomenclature defined in peer-reviewed neuroanatomical literature. In some instances, we used or referred to nomenclature derived from rodent brain atlases (Dong, 2008; Paxinos & Franklin, 2013), with preference to the taxonomy used in the publicly available Allen Brain Mouse Atlas (Wang et al., 2020).

Results

Brain-wide distribution of NPS neurons.

We began by examining the brain-wide distribution of NPS neurons in parasagittal tissue sections from Nps-2A-Cre;R26-LSL-L10GFP mice (Figure 1 and Supplemental Video 1). The PB region contained the most prominent clusters of neurons expressing the L10GFP reporter, similar to our previous report in coronal tissue sections (Huang et al., 2022). The sagittal plane of section better highlighted the rostrocaudal ribbon of NPS neurons in the lateral habenular nucleus and epithelioid cells running along the central canal of the spinal cord (Figure 1e). Also matching our previous description, we identified fewer neurons scattered in the hindbrain reticular formation, nucleus incertus region, anterior hypothalamus, and medial amygdala.

Figure 1.

Neurons with L10GFP expression in Nps Cre-reporter mice (Nps-2A-Cre;R26-lsl-L10GFP) in a sagittal (a), coronal (b), and horizontal (c) planes. Each dot represents approximately 1–5 neurons. Additional non-neuronal L10GFP-expressing cells were found along the central canal of the spinal cord (d,e). All scalebars are 100 μm.

NPS-immunoreactive fibers.

We immunolabeled NPS and observed fiber labeling in a pattern largely consistent with a previous report (Clark et al., 2011). The anterior paraventricular thalamic nucleus contained the densest labeling (Figure 2b). We also found immunoreactive fibers in the preoptic area (Figure 2a), sparse labeling in the dorsal midbrain (Figure 2c), moderately dense labeling in the subparaventricular zone (Figure 2d), and a thin, dense horizontal band dorsal to the third ventricle (Figure 2d). Further caudally in the medial hypothalamus, we found a broad patch of labeling in the DMH and dorsal to it (Figure 2e), as well as in the periventricular region alongside the mammillary recess of the third ventricle (Figure 2f). All mice had this same overall pattern of fiber labeling, but the intensity of labeling varied between cases.

Figure 2.

NPS immunofluorescence labeling in the BST (a), PVT (b), PAG (c), PVH (d), caudal hypothalamus (e), and posterior periventricular hypothalamic nucleus (f) varied between cases. Scalebars in every panel are 200 μm.

Injection sites.

We used Cre-conditional anterograde tracing to determine which NPS target regions receive input from the PB, which contains the densest concentration of Nps-expressing neurons. Into the PB region of Nps-2A-Cre mice, we injected a vector that delivers a Cre-conditional presynaptic marker (AAV8-hEf1a-DIO-synaptophysin-mCherry, n=7). We targeted the rostral NPS cluster in three cases (5495, 5568, 5569) and the caudal NPS cluster in four cases. Two caudal injections failed to transduce any neurons, and we analyzed the remaining two cases (5429, 5432).

Rostral-to-caudal plots of each injection site are shown in Figure 3, and rostral-to-caudal images of the full injection site in each case are shown in Supplementary Figure 1. Among the three rostral cases, 5568 transduced the most neurons, 5569 transduced an intermediate amount, and 5495 transduced the fewest. Among the two caudal cases, 5429 transduced neurons mainly in the caudal cluster and few in the lateral PB, and 5432 transduced neurons exclusively in the caudal NPS cluster. No injection sites transduced any neurons in the nucleus incertus or reticular formation. We sliced these five brains in the coronal plane and used them for manual, brain-wide analysis of Syp-mCherry NiDAB labeling, as well as BoutonNet plots and immunofluorescence labeling in select target regions.

Figure 3.

Injection site plots of Syp-mCherry transduced neurons in all Nps-2A-Cre cases. Rostral cases (5569, 5568, 5495) had labeling in the rostral PB with no labeling caudally. In caudal cases (5432, 5429), somatic labeling was primarily located caudally, near the locus coeruleus, and caudal case 5429 had some labeling in the lateral parabrachial nucleus.

As a supplement to this primary analysis in the coronal plane, we injected the same AAV vector into the rostral PB of additional Nps-2A-Cre;R26-LSL-L10GFP mice (n=4). We sliced these four brains in the sagittal plane and used their parasagittal sections for immunofluorescence labeling. An injection site from one of these cases is shown in Figure 4.

Figure 4.

Parasagittal injection site of Cre-conditional Syp-mCherry in case 6730. L10GFP expression reveals cells in the rostral lateral PB with a history of Nps expression (green, in a and c). Syp-mCherry expression (red in b) was exclusive to a subset of these neurons (yellow in c). NPS neurons are immediately dorsal and caudal to the noradrenergic A7 group of neurons, which are immunoreactive for tyrosine hydroxylase (TH; magenta, in a and c). Scalebar in (b) is 200 μm and applies to (a). Scalebar in (c) is 200 μm.

Brain-wide distribution of Syp-mCherry labeled boutons.

We analyzed the brain-wide distribution of Syp-mCherry labeling in each coronal case (Figure 5). There are fewer NPS neurons in the caudal cluster, near the locus coeruleus, and cases with injections into this caudal region also had sparser overall labeling than cases with injections targeting the larger, rostral cluster. Brain regions with prominent labeling in rostral cases also contained at least trace labeling in caudal cases, but many regions received heavier labeling in rostral cases than in caudal cases, as described below. In all cases, labeling ipsilateral to the injection site was stronger. To plot the brain-wide distribution of Syp-mCherry-labeled boutons, we performed BoutonNet analysis on tissue sections from a representative case (5569; Figure 6).

Figure 5.

Density of Syp-mCherry labeling in 150 brain regions in each case, grouped by rostral vs. caudal injection site. Light grey indicates absence of labeling and darker shades of blue represent increasing density of labeling. For all abbreviations, see List of Abbreviations.

Figure 6.

Brain-wide distribution of Syp-mCherry labeling in case 5569. These 26 illustrated sections (a–z) were chosen to best represent the output pattern of NPS neurons in the PB region. Transduced neurons in the injection site are represented by black dots in panels (s–u). All abbreviations are found in List of Abbreviations.

Cerebral cortex and olfactory bulb.

Except for a few, scattered boutons in some cases, we did not find labeling in the cerebral cortex or olfactory bulb. The anterior olfactory nucleus contained trace labeling in several cases, and the tenia tecta contained trace labeling only in rostral cases.

Septum.

The lateral septum contained labeling in all cases. This labeling concentrated in rostral and ventral subregions, with virtually no labeling dorsally (Figure 6c–d). In rostral cases, the medial septal nucleus also contained light labeling, but no other region of the septum contained labeling in other cases.

Basal ganglia.

In every case, we found light labeling in the caudal nucleus accumbens shell, where it merges with the bed nucleus of the stria terminalis (Figure 6d-e). Rostral to this, the nucleus accumbens shell contained no more than trace labeling (Figure 6c). Trace labeling in the striatum, globus pallidus, and subthalamic nucleus was an inconsistent finding in rostral cases.

Amygdala.

The amygdala received very little input. The central nucleus of the amygdala contained trace Syp-mCherry labeling in some cases, and the medial amygdalar nucleus had trace labeling in all cases (Figure 6l). Only case 5568 had trace labeling in the basomedial nucleus and cortical amygdalar area.

Basal forebrain.

Most basal forebrain nuclei received no input. Some cases had light labeling in the substantia innominata or trace labeling in the diagonal band (Figure 6c-h).

Bed nucleus of the stria terminalis.

All cases had dense labeling in the bed nucleus of the stria terminalis (BST), ventral to the anterior commissure (Figure 6e-f). Lighter labeling covered a larger expanse of the anterolateral and anteromedial BST while largely avoiding the oval subnucleus and posterior subnuclei. We immunolabeled AgRP and CGRP to contextualize the anterograde terminal field in this interesting area. Syp-mCherry-labeled boutons overlaid a focal cluster of CGRP-labeled axons, which identify the fusiform BST subnucleus, but Syp-mCherry medium-density labeling also extended medially through ventral BST subnuclei that receive dense AgRP-immunoreactive fibers. In contrast, the oval BST subnucleus, which contains denser CGRP labeling, had significantly less Syp-mCherry labeling (Figure 7).

Figure 7.

In the ventral BST, Syp-mCherry immunofluorescence (red, a) partly overlapped CGRP-immunoreactive fibers (green, b) and extensively overlapped AgRP-immunoreactive fibers (blue, c–e). AgRP, CGRP, and mCherry combined immunolabeling in panel (e). All sections are from case 5568. All scalebars are 200 μm. Scalebar in (d) also applies to (a–c).

Lamina terminalis.

Both circumventricular organs of the lamina terminalis, the subfornical organ (SFO, Figure 6g) and vascular organ of the lamina terminalis (OVLT), had light labeling in several cases. Between them, the median preoptic nucleus contained moderate labeling in each case (Figure 6f).

Hypothalamus.

In every case, the hypothalamus received more labeling overall than any other brain region (Figure 6g-o). However, this labeling was not uniform across the hypothalamus. The medial hypothalamus contained much more Syp-mCherry labeling than the lateral hypothalamus. Within the medial hypothalamus, however, labeling was sparse in the mammillary, ventromedial, and suprachiasmatic nuclei. Also, a band extending dorsolaterally from the medial preoptic nucleus through the posterior BST received relatively less labeling than surrounding parts of the anterior hypothalamus.

Near the third ventricle and dorsal to the suprachiasmatic nucleus, we found a dense patch of Syp-mCherry labeling in a focal subregion of the subparaventricular zone (SPZ), in-between TH-immunoreactive neurons and axons in the PVH and GRP-immunoreactive fibers projecting dorsally from the SCN (Figures 6i & 8). The paraventricular hypothalamic nucleus (PVH) itself contained only light labeling, but a thin, horizonal band immediately dorsal to the ventral third ventricle contained denser labeling (Figure 8).

Figure 8.

Dense Syp-mCherry immunofluorescence (red, a) lies immediately ventral to TH-immunoreactive neurons and fibers in the paraventricular hypothalamic nucleus (green) and immediately dorsal to GRP-immunoreactive fibers in the SPZ (b–c). Combined mCherry, TH, and GFP labeling is shown in panel (c; case 5569). This level of the anterior hypothalamus also contained dense Syp-mCherry labeling in a band dorsal to the third ventricle. Both scalebars are 200 μm. Scalebar in (b) also applies to (a).

The dorsomedial hypothalamic nucleus (DMH) contained the densest labeling in the hypothalamus, and the parasubthalamic nucleus (PSTN) also contained dense labeling. Labeling in the rest of the lateral hypothalamic area and posterior hypothalamus was less dense. The dense blanket of labeling that covered the DMH extended dorsally to encompass the A13 cluster of catecholaminergic-GABAergic neurons (Negishi et al., 2020) in the zona incerta, at the dorsal margin of the hypothalamus (Figure 9).

Figure 9.

Syp-mCherry labeling (red, b–c) in a parasagittal section though the thalamus and hypothalamus (case 6730). L10GFP Cre-reporter for Nps (green; a, c) identifies NPS neurons in the anterior hypothalamic nucleus and lateral habenula, while TH immunolabeling identifies catecholaminergic neurons and fibers (magenta; a, c). Anterogradely labeled axons produced dense concentrations in the paraventricular thalamic nucleus (PVT), magnocellular subparafasicular nucleus (mSPF), A13 group, and several hypothalamic subregions, while most of the thalamus and much of the mPOA, VMH, and MB lacked labeling. For remaining abbreviations, see List of Abbreviations. Scalebars are 800 μm and the scalebar in (b) also applies to (a).

Thalamus.

The paraventricular nucleus of the thalamus (PVT) contained some of the densest labeling in every case. Profuse labeling enveloped FoxP2-immunoreactive neurons in the anterior PVT and extended laterally to surround the paratenial nucleus (Figure 10). This dense terminal field extended caudally through the full PVT. The reunions, rhomboid, xiphoid, and lateral habenular nuclei also contained modest labeling.

Figure 10.

In the thalamus, dense Syp-mCherry labeling (red, a) concentrated in the anterior PVT and arced laterally around the paratenial nucleus (case 5495). These thalamic nuclei contained strong nuclear immunoreactivity for the transcription factor FoxP2 (green, b–c). Scalebar in (b) applies to (a) and is 200 μm. Scalebar in (c) is 200 μm.

We also found a small, dense patch of labeling in the caudal, ventral thalamus. The density of labeling in this small, round spot was similar to the density of Syp-mCherry labeling in the PVT. Syp-mCherry labeling here avoided neighboring catecholamine neurons and instead targeted a small spot referred to as the magnocellular subparafasciular nucleus (mSPF; Figure 6m, Figure 9c). In contrast, the parafasicular thalamic nucleus had sparse labeling, and the parvocellular subparafasicular region had light labeling that extended laterally toward the geniculate nuclei and peripeduncular nucleus.

In the caudal, lateral thalamus, we found sparse labeling in the intergeniculate leaflet, ventral lateral geniculate nucleus, and a region referred to as the posterior limitans nucleus (Paxinos & Franklin, 2013; Paxinos & Watson, 2007). Labeling in these regions was sparse in rostral cases and absent in caudal cases. No cases contained labeling in the dorsal lateral geniculate nucleus. Immediately ventral to the medial geniculate nucleus, we found moderately dense labeling in the peripeduncular nucleus.

Midbrain.

In rostral cases, the tectum contained one of the densest targets in the brain. There, we found a thin, dense stripe of Syp-mCherry-labeled boutons spanning the midline of the superior and inferior colliculi. This terminal field, immediately dorsal to the PAG, was apparent in serial coronal sections (Figure 11a-b) and even more prominent in the sagittal plane (Figure 11c-d). This rostrocaudally elongated zone appears to be the “tectal longitudinal column” (TLC) described in rats (Saldana et al., 2007). The TLC was one of the most prominent targets in every rostral case, but we found virtually no labeling here in caudal cases. Labeled axons and boutons reached the TLC via a periventricular pathway through the periaqueductal gray matter, which also contained moderately dense labeling.

Figure 11.

(a–b) Dense labeling in the deep midline SC (case 5569; coronal section, NiDAB labeling for Syp-mCherry). (c) Sagittal view of median collicular labeling with L10GFP Cre-reporter for Nps (green) plus immunofluorescence labeling for TH (magenta; case 6731) and magnified area of interest in (d). Scalebars are (a) 500 μm, (b) 100 μm, (c) 1 mm, and (d) 100 μm.

Separate from this periventricular pathway, the midbrain received moderate labeling along a ventrolateral pathway. Syp-mCherry-labeled axons in the ventral pathway coursed alongside the medial lemniscus and dorsal to the substantia nigra, without arborizing, until producing a concentration of branches and boutons in the retrorubral field and extending rostrally into the peripeduncular and subparafasciular thalamic target zones described above.

Cerebellum.

No part of the cerebellar cortex or deep cerebellar nuclei contained labeling in any cases.

Hindbrain.

NPS neurons supply a descending pathway, but these projections did not produce wide-ranging labeling in the hindbrain. A thin rim of moderately dense labeling encased the superior olivary nucleus, trapezoid bodies, and ventral nucleus of the lateral lemniscus. We also found light labeling in the ventrolateral pontine reticular nucleus and in a rim around the facial motor nucleus. Immediately rostral and dorsolateral to the facial motor nucleus, a moderately dense collection of Syp-mCherry labeling appeared to target the focal cluster of medium-large neurons in the superior salivatory nucleus. The nucleus of the solitary tract had no more than trace labeling. A region of the caudal ventrolateral medulla that included the A1 group and caudal pressor area (CPA) contained sparse labeling.

Discussion

Glutamatergic neurons in the PB form two, mutually exclusive groups, an Atoh1-derived macropopulation and an Lmx1b-expressing macropopulation. Collectively, they project axons via four pathways: a periventricular pathway, the central tegmental tract, a ventral pathway, and a descending pathway (Figure 12a). The Lmx1b macropopulation projects primarily via the central tegmental tract, while the Atoh1 macropopulation projects primarily via the ventral pathway. As an Atoh1 subpopulation, Nps-expressing PB neurons project axons via the ventral pathway, and our results reveal an unexpectedly robust contribution to the periventricular pathway (Figure 12b). Their axons heavily target the midline thalamus, medial hypothalamus, and midline tectum, with lighter projections to several other brain regions. In contrast, we found little to no labeling in the cerebral cortex, olfactory bulb, amygdala, and cerebellum. After discussing limitations of our study, we compare our results to previous work and forecast the function of NPS neurons in the PB region.

Figure 12.

Projection pathways and targets of (a) PB glutamatergic neurons in comparison to (b) the Nps-expressing subset of glutamatergic neurons. Nps-expressing neurons project prominently via a periventricular pathway, as well as a ventral pathway. These pathways converge in the anterior hypothalamus, and individual axons may reach diencephalic targets through either pathway.

Limitations.

Syp-mCherry tracing with BoutonNet algorithmic processing has several advantages compared to previous anterograde tracing techniques (Grady et al., 2022). Syp-mCherry concentrates in the presynaptic bouton, while previous tracers uniformly distribute in the axoplasm. However, a small portion of the putative boutons labeled by our method likely represent non-synaptic portions of axons cut in cross-section because Syp-mCherry must travel through the axon shaft to reach presynaptic boutons. To mitigate this limitation, we visually inspected Syp-mCherry labeling in every brain region and applied semi-quantitative density standards used in previous tracing studies (Huang et al., 2020; Huang et al., 2021). Despite these efforts to identify boutons from morphology, determining whether individual boutons form structural synapses would require the use of electron microscopy (Wouterlood & Jorritsma-Byham, 1993), and determining whether they form functional synapses would require the use of electrophysiology (Petreanu et al., 2007). Neither of these approaches is practical at the scale of this study.

We genetically targeted NPS neurons using a knockin-Cre strain with a 2A “self-cleaving” peptide inserted after the last exon of the endogenous Nps gene (Huang et al., 2022). We used a Cre-conditional construct (DIO-Syp-mCherry), so that only neurons with sufficient Cre expression between the time of injection and perfusion were transduced. We performed these experiments in adult mice, so the results reflect axonal projections of neurons with adult Nps expression, not developmental expression, which may be more widespread.

Additionally, the Syp-mCherry promoter is different from the Nps gene promoter. Therefore, neurons with sufficient Cre expression activate the DIO construct and produce large amounts of Syp-mCherry that do not reflect the varying levels of Nps expression in those same neurons. Thus, while our approach has high sensitivity for neurons that express Nps, the axonal labeling density in our results does not reflect Nps expression level in those neurons. This may be a relevant consideration due to the known variations in NPS protein and Nps mRNA expression with physiological state, as well as the broader distribution of neurons that expressed a Cre-reporter for Nps at the fringes of each neuronal cluster with dense Nps mRNA and NPS peptide (Huang et al., 2022).

Finally, although our combined cases labeled projection patterns of rostral and caudal NPS neurons, we did not use genetic targeting to distinguish subpopulations of NPS neurons. Rostral Nps-expressing neurons are concentrated between the tip of the superior cerebellar peduncle and the medial edge of the lateral lemniscus, resembling the cytoarchitecturally defined “extreme lateral” subnucleus in the rat PB (Fulwiler & Saper, 1984). This rostral group also includes neurons scattered caudally through the lateral PB. Additional neurons expressing an Nps Cre-reporter extend rostrally, along the medial edge of the lateral lemniscus, just dorsal to location of Pth2-expressing (TIP39) neurons in the medial paralemniscal nucleus (Dobolyi et al., 2003; Sun et al., 2022; Varga et al., 2008). These far-rostral NPS neurons overlap a location described as the “semilunar nucleus” in rats (Gómez-Martínez et al., 2023). While neurons in the main (“extreme lateral”) NPS cluster express high levels of Nps mRNA, other rostral neurons are identifiable only by Cre-reporter expression and contain very little Nps mRNA. These neurons may form separate subpopulations with separate connections and functions, but discriminating differences in projection patterns would require identifying additional genetic markers that distinguish NPS subpopulations. As in the rostral PB, most NPS neurons near the locus coeruleus express Foxp2, but a small number of neurons in Barrington’s nucleus expressed a Cre-reporter for Nps and lacked FoxP2 (Huang et al., 2022). Differentiating the projection patterns of NPS Cre-reporter neurons in Barrington’s nucleus, which lack FoxP2, from those of peri-LC NPS neurons, which express Foxp2, may be possible with intersectional genetic targeting. Rostral and caudal cases in our study had largely similar patterns, and without additional genetic markers specific to the rostral or caudal cluster of NPS neurons, we hesitate to draw firm conclusions about separate efferent projection patterns.

Comparison with previous neuroanatomical literature.

Generally, the pattern of Syp-mCherry labeling described above approximated the pattern of NPS fiber immunolabeling we observed (Figure 2), which matched the description in a previous study using the same NPS antiserum (Clark et al., 2011). This implies that neurons in the PB region supply most of the NPS peptide in the mouse brain.

Neighboring PB neurons express Cck and project heavily to the VMH (Garfield et al., 2014; Grady et al., 2020), and a previous study of NPS immunoreactivity reported fiber labeling in this hypothalamic nucleus (Clark et al., 2011). In our study, however, the VMH contained virtually no NPS-immunoreactive fibers, and it was largely devoid of Syp-mCherry labeling (Figure 2e). Our NPS labeling was highly similar to the previously published images, and this difference likely reflects separate neuroanatomical interpretations regarding the location and boundaries of the VMH.

Additionally, we identified dense output projections to regions with sparse NPS immunolabeling. Regions with prominent Syp-mCherry labeling despite sparse NPS immunolabeling included the ventral BST, mSPF, and TLC. A previous study in rats described retrograde labeling in the “extreme lateral” PB subnucleus after a tracer injection spanning the rostral BST, nucleus accumbens, and lateral septum (Fulwiler & Saper, 1984). Another previous study identified retrograde labeling in the PB region after tracer injections into the mSPF (Wang et al., 2006). We are not aware of any previous retrograde tracing studies that examined connectivity from the PB region to TLC, but this midline region of the tectum has Npsr1 expression as part of a broader pattern of expression in the superior colliculi (Clark et al., 2011).

NPS neurons are glutamatergic, Atoh1-derived, and Foxp2-expressing, but they do not express Pdyn or Cck (Huang et al., 2022; Pauli et al., 2022; Xu et al., 2007). The present results, including the lack of labeling in the cerebral cortex and amygdala, are largely consistent with our previous reports of efferent projections from the larger population of Foxp2-expressing PB neurons in rats and mice (Huang et al., 2020; Shin et al., 2011). NPS neurons also do not express Lmx1b, and the present results contrast the known projections of the Lmx1 PB macropopulation, including its subpopulation of Calca-expressing neurons, which prominently target the amygdala and cerebral cortex (Huang et al., 2021).

Most projections identified here represent a subset of the efferent projections previously identified in our study of Foxp2-expressing neurons in the PB region (Huang et al., 2020). However, NPS neurons also appear to target a small number of brain regions, including the TLC and NLL, that were less evident in our previous study. In subsequent Foxp2-IRES-Cre tracing cases with injection sites centered rostrally, at the level of the NPS neurons, the TLC and NLL contained dense Syp-mCherry labeling similar to the rostral Nps-2A-Cre tracing cases described here (unpublished observations, S.G. and J.C.G.). We also found dense fiber labeling in the TLC in Atoh1-Cre mice with a tdTomato Cre-reporter (see Figure 20b of Karthik, Huang, Delgado, Laing, Peltekian, Iverson, Grady, Miller, McCann, Fritzch, et al., 2022). Together, these findings indicate that a rostral, Nps- and Foxp2-expressing subset of Atoh1-derived neurons project to these auditory-associated brainstem nuclei. Future, retrograde tracing experiments should help clarify whether NPS output projections to the TLC and NLL arise from neurons within the rostral-lateral PB or from NPS neurons that surround the NLL just beyond the rostral limit of the PB.

The TLC is a median component of the superior colliculus and contains (Saldana et al., 2007) ventral (TLCv) and dorsal (TLCd) components (Aparicio & Saldaña, 2014). The densest output from PB Nps-expressing neurons appears to target the TLCv, with less labeling in the TLCd (Figure 11). Interestingly, the TLCv is reciprocally connected with the periolivary region, which also receives input from rostral NPS neurons (Saldaña et al., 2009; Viñuela et al., 2011).

During the final revision of this article, two studies appeared with slightly different descriptions of NPS output connectivity (Angelakos et al., 2023; Xing et al., 2024). Each used a separate strain Nps-Cre mice and slightly different axonal labeling methods. Angelakos et al. injected a pair of AAVs that delivered both Cre-conditional GFP and Cre-conditional synaptophysin-mRuby into the lateral PB (n=2) or peri-LC (n=2). Xing et al. injected a single AAV to deliver Cre-conditional channelrhodopsin-eYFP into the lateral PB, peri-LC, and nucleus incertus region (n=3).

Angelakos et al. reported mRuby-labeled puncta from the lateral PB in several regions that received substantive labeling in our rostral cases, including the PVT, POA, LHA, PAG, RR, and NTS. However, they did not report any labeling in the BST, SPZ, DMH, or TLC, where we found dense Syp-mCherry labeling in every case with an injection site in the rostral lateral PB. From the peri-LC, Angelakos et al. reported that the strongest projections targeted the cap of Kooy within the dorsal inferior olive (IOK), with additional labeling in the BST, PVT, “posterior subthalamic nucleus” (likely meaning the parasubthalamic nucleus), VTA, prepositus hypoglossal nucleus, and NTS, but they did not report any projections to the MnPO, DMH, or RR. In contrast, none of our cases had any labeling in the IOK. Some of these discrepancies may have resulted from injection site differences, but their report did not include any plots or images showing the full distribution of neurons transduced in any injection site. Of note, one of our caudal injection sites transduced neurons exclusively in the peri-LC cluster (next to the locus coeruleus; case 5432), without transducing any neurons in the nucleus incertus region, while Angelakos et al. reported that their peri-LC injections transduced neurons in the nucleus incertus, possibly due to larger AAV injection volumes (400 nL in their study versus 50–100 nL in ours). No axonal labeling is shown in any target region, which makes it difficult to comment further on differences between our results and theirs.

Xing et al. reported projections from the PB to the PVT/PVA, LPOA, MnPO, DMH, LHA, and PAGvL (Xing et al., 2024). From the peri-LC, they reported projections to the PVA, LPOA, PAGvL, LHA, PPN. Our study identified output projections to all of these regions, and several others described above. From the nucleus incertus region (central pontine gray), they described output projections to the LHA, PAGvL, and DR. No descending projections are mentioned (including whether or not IOK received input in peri-LC or nucleus incertus cases), and no plots or images are shown from any injection site, so it is difficult to comment further on any specific differences between studies.

Functional implications.

The PB integrates interoceptive sensory inputs. The genetic identity and location of neurons in the PB determine their different projections and functions. For example, Pdyn-expressing neurons in the dorsolateral PB relay warm thermosensory information (Geerling et al., 2016), Satb2-expressing neurons in the medial PB relay taste information (Fu et al., 2019; Jarvie et al., 2021), and Calca-expressing neurons in the external lateral subdivision induce taste aversion and inhibit appetite (Carter et al., 2015; Carter et al., 2013; Palmiter, 2018). In contrast to these populations, the precise function of Nps-expressing neurons is unknown. They represent a neuroanatomically distinct population with a novel projection pattern, targeting a number of brain regions that, among other things, regulate circadian rhythm and threat response.

The SCN is the master circadian regulator, and among brain regions with direct connectivity to the SCN, NPS axons project to the SPZ, PVT, DMH, and IGL. The SPZ is the main output target of the SCN and plays an important role in the circadian regulation of locomotor activity, body temperature, and sleep (Vujovic et al., 2015). Evidence exists for a relationship between NPS and sleep: a human NPSR1 gain-of-function mutation was associated with substantially reduced sleep, and mice carrying this mutation also slept less (Xing et al., 2019). Additionally, central injection of an NPSR1 antagonist reduced wakefulness in rats (Oishi et al., 2014), and more recent studies indicate that separate clusters of NPS neurons modulate REM sleep and breathing (Angelakos et al., 2023; Xing et al., 2024). Another dense target region that may play a role in circadian function is the mSPF. We were surprised to find dense labeling in the mSPF, whose neurons are medial and ventral to the fasciculus retroflexus and were not described previously as receiving substantial input from the PB region. This compact and unusual thalamic nucleus contains GABAergic neurons and sends a prominent output projection to the IGL (Vrang et al., 2003), which receives direct input from the retina and sends direct output to the SCN (Moore et al., 2000). The IGL itself received light axonal input from the NPS neurons in our cases with rostral injection sites. The anterior PVT and DMH also play important roles in circadian regulation, and each receives heavy input from NPS neurons. It is likely that NPS projections to these regions influence circadian patterns, and this influence may help explain sleep/wake changes associated with NPS and its receptor.

Separately, adaptive responses to threats are fundamental to survival. NPS axons target several brain regions that are implicated in modulating threat responses, including the DMH, ventral BST, PVT, and medial tectum. NPS axons produced particularly dense labeling in the DMH, while avoiding the VMH. In contrast, Cck-expressing neurons in the superior lateral PB project to the VMH and are activated by noxious, aversive input from the spinal cord (Hermanson et al., 1998), and neurons in the ventrolateral VMH cause aggressive attack behavior in male mice (Lin et al., 2011). However, the lack of Syp-mCherry labeling in the VMH and the mutual exclusivity of Cck and Nps expression suggest that NPS neurons do not regulate these behaviors. Dense labeling in the DMH spreads dorsally and laterally into the zona incerta and encompasses the A13 catecholaminergic cell group. Electrically stimulating this medial region of the hypothalamus triggers escape behaviors, whereas stimulating the lateral hypothalamus triggers exploratory locomotion (Lammers et al., 1988). The DMH contains Npsr1 expression, and central NPS injections increase locomotion, so it is tempting to speculate that NPS-DMH connectivity promotes escape responses.

NPS axons avoid the dorsal BST, while broadly targeting a subregion of the ventral BST that also receives AgRP input from the arcuate hypothalamic nucleus. Stimulating the AgRP (GABAergic) projection to this region of the BST increases food intake, whereas central injections of NPS suppress food intake, suggesting that these convergent peptidergic inputs have opposing influences on hunger (Betley et al., 2013; Smith et al., 2006). The BST also plays a role in sustained responses to threat (Davis et al., 2010; Gewirtz et al., 1998; Walker et al., 2003) and is implicated in generalized anxiety disorder in humans (Lebow & Chen, 2016; Yassa et al., 2012). Based on associations between human NPSR1 gain-of-function mutations with panic and anxiety disorders (Domschke et al., 2011; Donner et al., 2010; Okamura et al., 2007; Pape et al., 2010), NPS projections to the BST may play a role in regulating long-term responses to threats.

The PVT is the most prominent region with Syp-mCherry labeling, and it contains similarly prominent NPS-immunoreactive fiber labeling and Npsr1 mRNA expression (Clark et al., 2011). Notably, activating overall PB projections to the PVT is aversive (Zhu et al., 2022). Also, deleting either Nps or Npsr1 attenuates conditioned fear responses, and these responses can be rescued by injecting NPS into the PVT in Nps knockout mice (Garau et al., 2022). Therefore, NPS neuron connectivity to the PVT may be important for relaying aversive information.

In the midbrain, we found a dense patch of labeling in the midline of the tectum. This labeling was most prominent in cases with rostral injections sites, and it was virtually absent in caudal cases. This area of the brain, identified in rats as the TLC, contains neurons that respond to auditory stimuli and project to the superior olivary complex (Saldana et al., 2007). Additional work suggests that neurons in this tectal subregion encode threat salience, as opposed to underlying neurons in the dorsal PAG that regulate escape behavior (Evans et al., 2018). We postulate that rostral NPS axons targeting the medial tectum (TLC) are part of a neural circuit that codes for threat salience.

Descending projections to the hindbrain skirt around the edges of the facial motor nucleus and appear to target the superior salivatory nucleus. Via their peripheral projections to the pterygopalatine ganglion, SSN neurons control the lacrimal gland, along with mucous secretion and dilation of cerebral and retinal blood vessels. Along with their output to other target regions implicated in aversive responses, this connection to the SSN may allow NPS neurons to drive autonomic responses to distress, such as mucous secretion from the nasal mucosa and tear secretion from the lacrimal glands.

Conclusion.

Pharmacological studies found that NPS promotes arousal and increases locomotion, and more recent, cell-type specific activation of Nps-expressing neurons in the PB region identified changes in REM sleep and breathing. In humans, NPS receptor mutations are associated with reduced sleep time, as well as panic and anxiety disorders. In this study, we mapped and analyzed the output projections of NPS neurons in the PB region. Their target regions predict that NPS neurons influence circadian function and compute responses to threat.

Supplementary Material

Video 1

Supplemental Video 1. Rotations of a 3D model of NPS neuron locations in the mouse brain.

2

Supplemental Figure 1. Full rostral-to-caudal Nissl-counterstained sections with Syp-mCherry NiDAB immunolabeling in all injection sites cut in the coronal plane.

Table 2.

Antisera.

Antigen	Immunogen Description	Source, host species, RRID	Concentration
AgRP	Human AgRP 83–132 Amide	Phoenix Pharmaceuticals, rabbit polyclonal, cat#: H-003–53, lot#: 01825–2, RRID AB_2313908	1:5,000
CGRP	Synthetic peptide corresponding to rat CGRP aa23–37 (sequence VKDNFVPTNVGSEAF, C-terminal) conjugated to gamma globulin.	Abcam, goat polyclonal, cat# ab36001, lot #: GR3330475–13, RRID: AB_725807	1:1,000
DsRed	DsRed-express, a variant of Discosoma sp. red fluorescent protein	Takara Bio, rabbit polyclonal, cat#: 632496, lot# 1805060, RRID: AB_10013483	1:2,000
FoxP2	E. coli-derived recombinant human FoxP2 isoform 1 Ala640-Glu715	R&D Systems, sheep polyclonal cat# AF5647; lot#: CCUB0109061, RRID: AB_2107133	1:3,000
GFP	Full length green fluorescent protein from the jellyfish Aequorea victoria	Invitrogen, chicken polyclonal, cat# A10262, lot# 2591656, lot# 1972783, RRID: AB_2534023	1:3,000
GRP	Synthetic (human) bombesin coupled to bovine thyroglobulin with glutaraldehyde.	Immunostar, rabbit polyclonal, cat# 20073, lot# 1420001, RRID: AB_572221	1:4,000
mCherry	Full length mCherry protein	Invitrogen, rat monoclonal, cat# M11217, lot#: VA 293192, lot# W6337875, lot# VH 307508, RRID: AB_2536611	1:2,000
NPS	Synthetic NPS peptide	Abcam, rabbit polyclonal, #ab18252, lot: GR198504–1, RRID: AB_776718	1:2000
TH	Denatured tyrosine hydroxylase from rat pheochromocytoma in rabbit host	Millipore, rabbit polyclonal, AB152, 3199177, RRID: AB_390204	1:10,000
TH	Native tyrosine hydroxylase from rat pheochromocytoma in sheep host	Millipore, sheep polyclonal, cat#: Ab1542, lot#: 34756625, RRID: AB_90755	1:1,000

Abbreviations

A1

A1 catecholaminergic group

A8

A8 catecholaminergic group

ac

anterior commissure

AcbSh

accumbens nucleus, shell

ACA

anterior cingulate cortical area

AgRP

agouti-related peptide

AHA

anterior hypothalamic area

AI

agranular insular cortex

AmbC

ambiguus nucleus compact part

AON

anterior olfactory nucleus

AP

area postrema

APir

amygdalopiriform transition area

APN

anterior pretectal nucleus

Arc

arcuate nucleus

AVP

arginine vasopressin

BLA

basolateral amygdalar nucleus

BMA

basomedial amygdalar nucleus

BST

bed nucleus of the stria terminalis

BSTaL

BST anterolateral subnucleus

BSTam

BST anteromedial subnucleus

BSTfu

BST fusiform subnucleus

BSTov

BST oval subnucleus

BSTp

BST posterior subnucleus

CeA

central nucleus of the amygdala

CeAc

CeA capsular subdivision

CeAl

CeA lateral subdivision

CeAm

CeA medial subdivision

CGRP

calcitonin gene-related peptide

Cla

claustrum

CLi

central linear raphe

CM

central medial thalamic nucleus

CoA

cortical amygdalar nucleus

CPA

caudal pressor area

Cun

cuneiform nucleus

DCN

deep cerebellar nuclei

DCO

dorsal cochlear nucleus

Dl

dysgranular insular cortex

DMH

dorsomedial hypothalamic nucl

DP

dorsal peduncular cortical area

DR

dorsal raphe nucleus

DTN

dorsal tegmental nucleus

Ent

entorhinal cortical area

Epd/Epv

endopiriform nucleus, dorsal/ventral subdivisions

Epy

ependymal cell layer of cerebral aqueduct

EW

Edinger-Westphal nucleus

FL

flocculus of the cerebellum

fr

fasciculus retroflexus

fx

fornix

GI

granular insular cortex

GiRN

gigantocellular reticular nucleus

GPe

globus pallidus external

GPi

globus pallidus internal

GRP

gastrin-related peptide

Hc

hippocampus

IA

intercalated amygdalar nucleus

IC

inferior colliculus

IFV

interfascicular trigeminal nucleus

IG

induseum griseum

IGL

intergeniculate leaflet

ILA

infralimbic cortical area

IMD

intermediodorsal thalamic nucleus

IO

inferior olivary complex

IPN

interpeduncular nucleus

IRN

intermediate reticular nucleus

L6 PFC

layer 6 prefrontal cortex

LAr

lateral amygdaloid nucleus

LDT

laterodorsal tegmental nucleus

LGd

lateral geniculate nucleus, dorsal subdivision

LGv

lateral geniculate nucleus, ventral subdivision

LHA

lateral hypothalamic area

LHb

lateral habenular nucleus

LM

lateral mammillary nucleus

LPOA

lateral preoptic area

LRN

lateral reticular nucleus

LSi

lateral septum, intermediate subdivision

LSr

lateral septum, rostral subdivision

LSv

lateral septum ventral subdivision

MA

magnocellular nucleus

MB

mammillary bodies

MD

mediodorsal thalamic nucleus

MeA

medial amgydalar nucleus

MGN

medial geniculate nucleus

MHb

medial habenular nucleus

MM

medial mammillary nucleus

MnPO

median preoptic nucleus

MOB

main olfactory bulb

MPN

medial preoptic nucleus

mPOA

medial preoptic area

MR

median raphe

MRF

midbrain reticular formation

MS

medial septum

mSPF

magnocellular subparafascicular nucleus

NBM

nucleus basalis of Meynert

NDB

nucleus of the diagonal band

NeoCtx

neocortex

NLL

nucleus of the lateral lemniscus

NPS

neuropeptide S

NTB

nucleus of trapezoid body

NTS

nucleus of the solitary tract

NTS-Lat

NTS lateral subdivision

NTS-Med

NTS medial subdivision

OVLT

organum vasculosum of lamina terminalis

PAG

Periaqueductal gray matter

PAG-L

PAG lateral subdivision

PAGdl

PAG dorsolateral subdivision

PAGdm

PAG dorsomedial subdivision

PAGvL

PAG ventrolateral subdivision

PaRN

parvicellular reticular nucleus

PB

parabrachial nucleus

PC

paracentral thalamic nucleus

PeriRh

perirhinal cortical area

PF

parafascicular thalamic nucleus

PG

pontine grey

PH

posterior hypothalamic nucleus

Pir

piriform cortical area

PL

prelimbic cortical area

PLi

posterior limitans thalamic nucleus

PMd/v

posterior mammillary nucleus, dorsal/ventral

Po

posterior thalamic complex

PP

peripeduncular nucleus

PPN

pedunculopontine tegmental nucleus

PreT

pretectal region

PRN

pontine reticular nucleus

PS

parastrial nucleus

PSTN

parasubthalamic nucleus

PSV

principal sensory trigeminal nucl

PT

paratenial thalamic nucleus

PVH

paraventricular hypothalamic nucleus

PVT

paraventricular thalamic nucleus

RCA

retrochiasmatic area

Re

reuniens thalamic nucleus

Rh

rhomboid thalamic nucleus

RLi

rostral linear raphe

RMg/Ob/Pa

raphe magnus/obscurus/pallidus

RMTg

rostromedial tegmental nucleus

RN

red nucleus

RR

retrorubral area

SC

superior colliculus

SC-Lat

SC lateral aspect

SC-Sen

SC sensory-related

SCN

suprachiasmatic nucleus

SCO

subcommissural organ

SEZ

subependymal germinal zone

SF

septofimbral nucleus

SFO

subfornical organ

SHi

septohippocampal nucleus

SI

Substiantia innominata

SNc

Substiantia nigra pars compacta

SNr

Substiantia nigra pars reticularis

SOC

superior olivary complex

SON

supraoptic nucleus

SPFp

subparafasicular nucleus, parvicellular part

SpV

spinal trigeminal nucleus

SPZ

subparaventricular zone

SSN

superior salivatory nucleus

STN

subthalamic nucleus

Str (CP)

striatum (caudoputamen)

Str/GP bdr

striatum/pallidum border

SUM

supramammillary nucleus

TH

tyrosine hydroxylase

TLC

tectal longitudinal column

TM

tuberomammillary nucleus

TRN

tegmental reticular nucleus

TR

amygdalopiriform transition area

TT

taenia tecta

V

trigeminal motor nucleus

VII

facial motor nucleus

VLPO

ventrolateral preoptic nucleus

VM

ventromedial thalamic nucleus

VMH

ventromedial hypothalamic nucl

VNC

vestibular nuclear complex

VPpc

parvicellular ventral posterior thalamic nucleus

VTA

ventral tegmental area

Xi

xiphoid thalamic nucleus

XII

hypoglossal motor nucleus

ZI

zona incerta

Data Availability.

The data that support the findings of this study are available from the corresponding author upon reasonable request.