Abstract
Neurons in the caudal nucleus of the solitary tract (cNTS) and intermediate reticular nucleus (IRt) that express the glucagon gene (Gcg) give rise to glucagon-like peptide 1 (GLP1)-immunopositive axons in the spinal cord and many subcortical brain regions. Central GLP1 receptor signaling contributes to motivated behavior and stress responses in rats and mice, in which hindbrain GLP1 neurons are activated to express c-Fos in a metabolic state-dependent manner. The present study examined whether GLP1 inputs to distinct brain regions arise from distinct subsets of Gcg-expressing neurons, and mapped the distribution of axon collaterals arising from projection-defined GLP1 neural populations. Using our Gcg-Cre knock-in rat model, Cre-dependent adeno-associated virus (AAV) tracing was conducted in adult male and female rats to compare axonal projections of IRt versus cNTS GLP1 neurons. Overlapping projections were observed in all brain regions that receive GLP1 input, with the caveat that cNTS injections produced Cre-dependent labeling of some IRt neurons, and vice versa. In additional experiments, specific diencephalic or limbic forebrain nuclei were microinjected with Cre-dependent retrograde AAVs (AAVrg) that expressed reporters to fully label the axon collaterals of transduced GLP1 neurons. AAVrg injected into each forebrain site labeled Gcg-expressing neurons in both the cNTS and IRt. The collective axon collaterals of labeled neurons entered the spinal cord and every brain region previously reported to contain GLP1-positive axons. These results indicate that the axons of GLP1 neural populations that innervate the thalamic paraventricular nucleus, paraventricular nucleus of the hypothalamus, and/or bed nucleus of the stria terminalis collectively innervate all central regions that receive GLP1 axonal input.
Significance Statement
Our novel anatomical findings indicate that target-defined populations of forebrain-projecting glucagon–like peptide 1 (GLP1) neurons collectively project to downstream target regions in a widespread sprinkler-type manner, although collateralized axons arising from individual GLP1 projection neurons remain to be defined. Considered together with results from studies investigating the role of central GLP1 receptor signaling pathways in physiology and behavior, these findings support our emerging view that hindbrain Gcg-expressing neurons are positioned to simultaneously modulate synaptic transmission in widespread regions of the spinal cord, brainstem, hypothalamus, and limbic forebrain in a metabolic state-dependent manner.
Introduction
Neurons that express the glucagon gene (Gcg) occupy the caudal nucleus of the solitary tract (cNTS) and the intermediate reticular nucleus (IRt) in rats, mice, and primates, including humans. Gcg-expressing hindbrain neurons produce preproglucagon, which is cleaved to generate glucagon-like peptide 1 (GLP1) and other peptide products. In rats and mice, GLP1-positive cNTS and IRt neurons are activated to express the immediate-early gene product c-Fos after acute cognitive or physical threats, but only when animals are in a state of positive energy balance (Maniscalco et al., 2015; Maniscalco and Rinaman, 2017, 2018). Thus, GLP1 neurons are poised to provide metabolic state-dependent effects on neural circuit activity in regions that receive GLP1 axonal input, including the spinal cord and a large number of subcortical brain areas (Jin et al., 1988; Rinaman, 2010; Vrang and Larsen, 2010; Gu et al., 2013; Llewellyn-Smith et al., 2015). GLP1 receptors (GLP1R) are expressed in regions that receive GLP1 axonal input and also within the cortex and hippocampus (Cork et al., 2015; Farkas et al., 2021).
The collective results of rodent studies in which GLP1R signaling within the central nervous system (CNS) is experimentally enhanced or suppressed support the view that GLP1 acts in many central regions to suppress eating (X.-Y. Chen et al., 2021; Williams, 2022) and other reinforced/rewarding behaviors (Eren-Yazicioglu et al., 2021; Allingbjerg et al., 2023), to increase anxiety-like behavior and activate the endocrine hypothalamic–pituitary–adrenal (HPA) axis (Ghosal et al., 2013; Zheng et al., 2019), and to increase sympathetic outflow (Yamamoto et al., 2002; Hayes et al., 2008; reviewed by Holt and Trapp, 2016). Thus, endogenous central GLP1 signaling pathways may contribute to a wide array of coincident physiological and behavioral responses to stress (Holt and Rinaman, 2022). Apart from the olfactory bulb (Montaner et al., 2023), all GLP1-positive axons in the CNS are assumed to arise from hindbrain GLP1 neurons. However, it is unknown whether separate subpopulations of hindbrain GLP1 neurons provide input to separate postsynaptic CNS targets.
GLP1 neurons are distributed in a ratio of ∼55:45% within the rat cNTS:IRt, respectively (Maniscalco et al., 2015), but nothing is known regarding the potentially unique projections of these two populations of GLP1 neurons. This issue is important, because spatially distinct populations of brainstem reticular neurons that share a common transmitter phenotype (e.g., noradrenergic neurons in the NTS and ventrolateral medulla, serotonin neurons in distinct raphe nuclei, noradrenergic neurons in the locus ceruleus, dopamine neurons in the substantia nigra and ventral tegmental area) have both overlapping and unique axonal projections that help define their function (Kebschull et al., 2016; Poulin et al., 2018; Ren et al., 2019; Waterhouse et al., 2022). A few published studies used classic retrograde tracing approaches to demonstrate that individual GLP1-immunopositive neurons within the cNTS have axon collaterals that target more than one CNS region. For example, ∼40% of cNTS GLP1 neurons are retrogradely labeled from tracer injections into the medial hypothalamus in rats, and ∼20% target both the paraventricular nucleus of the hypothalamus (PVH) and the dorsomedial nucleus of the hypothalamus (DMH; Vrang et al., 2007). Another study found that 32–47% of cNTS GLP1 neurons in rats project to the ventral tegmental area, nucleus accumbens core, and/or nucleus accumbens shell (Alhadeff et al., 2012). Neither study evaluated projections from the IRt population of GLP1 neurons. Given the relatively large proportions of cNTS GLP1 neurons that were retrogradely labeled from each tracer injection site in those reports and considering the large number of distinct brain and spinal cord regions that receive axonal input from hindbrain GLP1 neurons (Jin et al., 1988; Rinaman, 2010; Gu et al., 2013; Llewellyn-Smith et al., 2015), we hypothesize that projection-defined populations of cNTS and IRt GLP1 neurons have axon collaterals that project broadly to modulate synaptic transmission in widespread regions of the brain and spinal cord.
We recently described a novel Sprague Dawley (SD) knock-in rat line, Gcg-Cre, in which cells express iCre recombinase under control of the Gcg promoter (Zheng et al., 2022). The present study used Cre-dependent viral tracing strategies in this rat model to determine whether GLP1 axonal inputs to distinct brain regions arise from both cNTS and IRt populations of hindbrain Gcg-expressing neurons and to identify the CNS distribution of axon collaterals arising from projection-defined subpopulations of GLP1 neurons.
Materials and Methods
Animal work was approved by the Institutional Animal Care and Use Committee of Florida State University and was performed in adherence to the Guide for the Care and Use of Laboratory Animals (National Research Council, 2011). Adult male and female SD Gcg-Cre rats were generated in our local breeding colony as previously reported (Zheng et al., 2022). To generate Gcg-Cre/tdTom reporter rats used in a subset of experiments, adult female homozygous Gcg-Cre rats were bred with male homozygous Long–Evans HsdSage:LE-Rosa26tm1(tdTomato)Sage rats (Envigo), as previously reported (Zheng et al., 2022). Rats were pair- or triple-housed with same-sex siblings in tub cages with wood-chip bedding in a vivarium with a 12:12 h light/dark cycle (lights off from 1700 to 0500 h), with ad libitum access to rodent chow (Lab Diet 5001) and water. Table 1 lists the strain, genotype, and sex of each rat (N = 17) that contributed data to the present report, along with the specific AAV injection(s) that each rat received.
Experimental cases
Cre-dependent viral tracing
To confirm or refine the stereotaxic coordinates used to deliver AAV into each brain region (described below) and to directly visualize the tissue area covered by 500 nl injection volumes, we conducted pilot studies using the surgical, perfusion, and immunohistochemical approaches described below. Rats received 500 nl of either Fluoro-Gold (0.05%; Fluorochrome) or cholera toxin beta (CTb; 0.05–0.1%; List Biological Laboratories #104) into each targeted brain region and were perfused on the same day or the following day. Brains were processed for either direct visualization (i.e., FG immunofluorescence) or immunocytochemical localization of CTb (immunolabeling described below) to reveal injection sites. In other rats, CTb (0.01%) was mixed with retrograde AAV (AAVrg), and CTb immunolabeling used to identify injection sites 3 weeks later.
Experiment 1: Cre-dependent anterograde tracing of the axonal projections of cNTS versus IRt GLP1 neurons
Rats were anesthetized by isoflurane inhalation (1–3% in oxygen; Halocarbon Laboratories) and placed into a stereotaxic device. The head was ventroflexed by 20° using an adjustable gas anesthesia adaptor (David Kopf Instruments; Model 929-B). The skin on top of the neck was shaved, disinfected, and incised down the midline (0.75–1.0 cm) just caudal to the occipital bone. The neck muscles were then incised down the midline, bluntly dissected laterally, and held in place with surgical hooks to reveal the dura overlying the cisterna magna. The dural layer was scraped clean using a dental plaque remover, incised using a sterile needle, and reflected to visualize obex.
Table 2 lists the Cre-dependent adeno-associated viruses (AAVs) used in the present report, including their Addgene product numbers. The AAV1 serotype is optimized for anterograde transport (Haery et al., 2019). In some rats, AAV1 expressing enhanced yellow fluorescent protein (EYFP) or enhanced green fluorescent protein (EGFP) reporter in a Cre-dependent manner (Addgene 20298 or 51502, respectively) was delivered bilaterally into the cNTS (0.3 mm lateral, 0.5 mm ventral to obex). In other rats, AAV1 expressing EGFP in a Cre-dependent manner (Addgene 51502) was delivered bilaterally into the IRt (1.7 mm lateral, 1.7 mm ventral to obex). For cNTS injections, AAV1 was delivered through a pulled glass micropipette tip (20 µm outer tip diameter) connected to a 10 μl Hamilton syringe. IRt injections were achieved using a 10 µl Hamilton syringe with a 33 Ga beveled stainless steel injector tip (bevel facing rostrally). In both cases, AAV1 was delivered by pressure (500 nl/injection, 500 nl/min), with the injector tip angled anteriorly by 10° to avoid the occipital bone and caudal cerebellum. Injection speed and volume were controlled by a digital stereotaxic microinjector (Quintessential Stereotaxic Injector, catalog #53313). The injector tip was left in place for 3 min after AAV1 delivery and then withdrawn. The retracted dorsal neck muscles were sutured closed, and then the overlying skin was sutured. Rats were injected subcutaneously with ketoprofen (2 mg/kg BW; Covetrus) and buprenorphine (0.03 mg/kg BW; Covetrus) and then returned to their home cages after full recovery from anesthesia.
AAVs and antibodies
Experiment 2: forebrain delivery of Cre-dependent AAVrg to label the axon collaterals of projection-specific GLP1 neurons
The retrograde serotype of AAV (AAVrg) is optimized to permit retrograde access to projection neurons whose axons occupy the AAVrg injection site. Fluorescent reporters expressed by each of the two Cre-dependent AAVrgs used in the present study (Table 2) are fused to a humanized channelrhodopsin that is transported throughout the processes of transduced neurons. As a result, fluorescent reporter labels the axon collaterals of Cre-expressing neurons projecting to the AAVrg injection site.
For AAVrg injections, the skin on top of the skull was shaved, disinfected, incised along the midline (0.75–1.0 cm) and reflected laterally, and the head was adjusted to achieve a flat-skull position. After identifying bregma, the rostrocaudal and mediolateral position of each injection target [i.e., PVH, anterior ventrolateral bed nucleus of the stria terminalis (vlBST), or paraventricular thalamic nucleus (PVT)] was marked on the skull, and a small hole over each injection site was drilled using a Dremel tool. Stereotaxic coordinates (from bregma) were as follows: PVH, 1.90 mm caudal, 0.4 mm lateral, and 8.6 mm ventral; vlBST, 0.30 mm caudal, 1.70 mm lateral, and 8.1 mm ventral; PVT, 1.9 mm caudal, 0.2 mm lateral, and 6.1 mm ventral.
In some rats, Cre-dependent AAVrg expressing EYFP (Addgene 20298) was injected unilaterally into either the PVH, vlBST, or PVT. In other rats, Cre-dependent AAVrg expressing mCherry (Addgene 20297) was injected unilaterally into either the PVH or the vlBST, and in the same rats, the EYFP-expressing AAVrg (Addgene 20298) was injected ipsilaterally into the other target (i.e., vlBST or PVH). In these dual-targeted rats, 0.01% CTb was added to the vlBST-targeting AAVrg solution for postmortem visualization of injection sites.
At each injection site, AAVrg was delivered by pressure (500 nl over 1 min) using a 10 µl Hamilton syringe with a 33 Ga beveled stainless steel injector (bevel facing medially), with injection speed and volume controlled as described for cNTS and IRt injections. Three minutes postinjection, the injector was slowly removed. The second injection was then performed in rats receiving dual PVH- and vlBST-targeted injections. After the final injection was complete, a small piece of sterile Gelfoam Dental Sponge (Henry Schein) was inserted into the drilled skull opening(s), and the skin overlying the skull was closed using stainless steel Autoclips (Kent Scientific). Rats were injected subcutaneously with ketoprofen (2 mg/kg BW) and buprenorphine (0.03 mg/kg BW) and then returned to their home cages after full recovery from anesthesia.
Perfusion fixation and tissue sectioning
Rats were perfused 3–4 weeks after AAV injection. For this, rats were injected with pentobarbital sodium (Euthasol, 130 mg/kg, i.p.; Covetrus) and then transcardially perfused with ∼50–75 ml of physiological saline followed by ∼250–300 ml of ice-cold 4% paraformaldehyde in 0.1 M sodium phosphate buffer (PB). Brains and spinal cords were postfixed in situ overnight, then extracted, postfixed an additional 2–6 h, and cryoprotected in 20% sucrose solution for 24–72 h at 4°C. Using a freezing sliding microtome, brains were sectioned coronally into six series of 25- or 35-µm-thick sections. Spinal cords were divided into cervical, thoracic, lumbar, and sacral segments and then sectioned horizontally into three series of 40-µm-thick sections. Tissue sections were immediately processed for immunocytochemical amplification of viral reporter labeling (described below) or stored at −20°C in cryopreservant solution (Watson et al., 1986) for later processing.
Immunocytochemistry
Tissue sections were rinsed in PB and then pretreated for 20 min in 1.0% sodium borohydride and 15 min in 0.3% H2O2 with an intervening 30 min rinse in 0.1 M PB. Primary and secondary antisera (listed in Table 2) were diluted in PB containing 0.3% Triton X-100 and 1% normal donkey serum. Tissue sections were rinsed for 30–60 min in several changes of PB between incubation steps.
For immunofluorescence, native EGFP or EYFP reporter labeling was enhanced using chicken anti-EGFP (1:20 K) followed by peroxidase-conjugated donkey anti-chicken IgG (1:500) and tyramide-conjugated fluorescein (1:450). Native fluorescent mCherry reporter labeling was enhanced using guinea pig anti-red fluorescent protein (RFP; 1:15 K) followed by Cy3-conjugated donkey anti-guinea pig IgG (1:500). Dopamine beta hydroxylase (DBH, to identify noradrenergic neurons and processes) was labeled using mouse anti-DBH (1:8 K) followed by Alexa Fluor 647-conjugated donkey anti-mouse IgG (1:500). Sections from rats that received a single-site AAV injection were double-labeled for EGFP and RFP or for EGFP and DBH. In rats that received dual PVH/vlBST injections (in which CTb was included in the AAVrg mixture to mark vlBST injection sites), one set of tissue was processed for triple immunofluorescent localization of CTb (goat anti-CTb; 1:5 K), mCherry reporter (guinea pig anti-RFP; 1:15 K), and EYFP reporter (chicken anti-EGFP; 1:10 K), followed by incubation in a cocktail of species-specific donkey anti-IgG antibodies conjugated to Alexa Fluor 647, Cy3, and Alexa Fluor 488 (each at 1:500), respectively.
For quantitative analysis of cNTS and IRt cell counts and semiquantitative analysis of axonal projections, EYFP/EGFP reporter labeling was localized using nickel-enhanced immunoperoxidase to facilitate hindbrain cell counting and brain-wide inspection of labeled axons without concern regarding fluorescent bleaching. For this, sections were incubated in chicken anti-EGFP (1:10 K) followed by biotinylated donkey anti-chicken IgG (1:1,000). Sections were then treated with Elite Vectastain ABC reagents (Vector Laboratories) and reacted with nickel sulfate, diaminobenzidine (DAB), and H2O2 in a 0.1 M sodium acetate buffer to produce a blue–black reaction product. To label all Gcg-expressing cNTS and IRt neurons, we processed hindbrain tissue sections from three adult, surgically naive Gcg-Cre/tdTom reporter rats (two males, one female) similarly using guinea pig anti-RFP followed by biotinylated donkey anti-guinea pig IgG and nickel-enhanced DAB to label tdTom-positive cNTS and IRt neurons. A separate set of tissue sections from each AAV-injected rat included in this report was similarly processed for nickel-enhanced immunoperoxidase localization of reactive microglia using rabbit anti-Iba1 (1:10 K, FUJIFILM Wako Laboratory Chemicals, Product Number 019-19741) followed by biotinylated donkey anti-rabbit IgG. This facilitated visualization of needle tracks and injection sites after rats were perfused several weeks after surgery.
Immunofluorescent and immunoperoxidase-labeled tissue sections were mounted onto Superfrost Plus microscope slides, dehydrated in ascending ethanol solutions, defatted with xylene, and coverslipped using Cytoseal 60 (VWR International).
Microscopic imaging
Low-magnification images of immunofluorescent or immunoperoxidase labeling were collected with a Keyence (BZ-X700) microscope using a 2× objective, with or without tiling. Tiled images were processed using “Full Focus” and “XY-Stitching” features of the Keyence Analyzer BZ-H4XD software to obtain Z-max projection images. Higher-magnification images of immunoperoxidase labeling were collected using a 10× or 20× objective. In some cases, slides containing hindbrain sections were imaged at 20× using tiling and then stitched to display the full complement of immunoperoxidase-labeled cNTS and IRt GLP1 neurons. Confocal images of immunofluorescence labeling were collected with a Leica TCS SP8 confocal microscope using 20× air and 40× or 100× oil objectives. Alexa Fluor 647 was excited using a Diode 638 laser. Cy3 and fluorescein/Alexa Fluor 488 were excited using OPSLs 552 and 488, respectively. Collected image tiles were merged, and Z-max projection images were generated using the Leica LAS version 4.0 image software. To construct figures for publication, we imported images into Adobe Photoshop and corrected their brightness and contrast.
Quantitative analysis of labeled hindbrain neurons
Labeled cNTS and IRt neurons were counted in AAV1 and AAVrg injection cases using hindbrain tissue sections processed for nickel–DAB immunoperoxidase labeling to localize EGFP/EYFP reporter. To determine the total number of Gcg-expressing cNTS and IRt neurons, counts of tdTom immunoperoxidase-labeled neurons were made in surgically naive adult rats (i.e., no AAV injections). In each case, the number of coronal hindbrain tissue sections (spaced 150 or 210 μm apart) that contained the rostrocaudal extent of cNTS and IRt labeling was recorded to derive averaged counts of labeled cNTS and IRt cells per section.
Semiquantitative analysis of axonal labeling
The CNS-wide distribution of labeled axonal projections was inspected in analyzed cases using tissue sections processed for nickel–DAB immunoperoxidase labeling to localize EGFP/EYFP reporter. For this analysis, labeling in every slide-mounted section (i.e., coronal brain sections spaced by 150 or 210 μm, horizontal spinal segment sections spaced by 120 μm) was analyzed using 10–20× microscope objectives (Keyence microscope). The presence and location of axonal labeling were noted within each section, with close reference to a rat brain atlas (Swanson, 2018). In addition, the relative density of EGFP-positive axonal labeling in each CNS region was assigned a semiquantitative value: “+” to represent sparse labeling, “++” to represent moderate labeling, and “+++” to represent dense labeling.
Results
We previously reported that Cre is efficiently and selectively expressed by GLP1-immunopositive neurons within the cNTS (and IRt) that express Gcg (Zheng et al., 2022). For the present study, the total number of Gcg-expressing cNTS and IRt neurons was determined by counting tdTom immunoperoxidase-labeled neurons in surgically naive adult Gcg-Cre/tdTom reporter rats (i.e., with no AAV injections; N = 2 male, 1 female). Reporter-labeled neurons were present in an average of 11.6 ± 1.7 hindbrain coronal sections (35 μm thick, spaced by 210 μm). The cNTS contained 10.5 ± 1.6 tdTom-positive cells per section (average total number, 119.3 ± 12.2 cNTS cells), and the IRt contained 9.7 ± 1.9 tdTom-positive cells per section (average total number, 110.0 ± 16.7 IRt cells).
Cre-dependent tracing of axonal projections arising from cNTS GLP1 neurons
Figure 1 includes images of Cre-dependent EGFP labeling at multiple rostrocaudal levels through the caudal medulla in a representative homozygous male Gcg-Cre rat that received bilateral AAV1 injections (Addgene 20298) into the cNTS (Table 1). DBH double labeling confirmed the absence of Cre-dependent EGFP reporter in noradrenergic neurons located within cNTS injection sites. The labeled axons of EGFP-positive neurons were prevalent within the medullary reticular formation (Fig. 1). Extended Data Figure 1-1D shows an image of Iba1 immunolabeling identifying the caudal dorsomedial medullary injection site, centered on the cNTS. We speculate that labeling of IRt GLP1 neurons observed after cNTS-targeted injections of Cre-dependent AAV1 is due to Cre-dependent anterograde transsynaptic delivery from transduced cNTS GLP1 neurons to postsynaptic IRt GLP1 neurons (see Discussion). Table 3 reports the number of EGFP-positive neurons counted in the cNTS and IRt in rats after bilateral cNTS injections of AAV1, with approximately three times as many EGFP-positive cell bodies present within the cNTS injection site versus the IRt. The average number of AAV1 reporter-labeled cNTS cells per section (9.1 ± 2.3) was similar to the average number of tdTom reporter-labeled cNTS neurons (10.5 ± 1.6) in surgically naive adult Gcg-Cre/tdTom reporter rats. EGFP fluorescence labeled the cell bodies and dendrites of AAV-transduced neurons, as well as their axons throughout the CNS (Fig. 1). No EGFP-positive cell bodies were observed outside of the cNTS and IRt. Thus, EGFP-positive axons observed in all other CNS regions are assumed to arise from labeled hindbrain GLP1 neurons. In this regard, and as summarized in Table 4, the central distribution of labeled axons was consistent with the published distribution of GLP1-immunopositive fibers in rats (Jin et al., 1988; Rinaman, 2010; Gu et al., 2013). The labeled axons of AAV-transduced GLP1 neurons ascended through the reticular formation of the medulla, pons, and midbrain to reach the hypothalamus and subcortical limbic forebrain. Figure 2 includes representative images of labeled axons within the pontine parabrachial nucleus, locus ceruleus, dorsal raphe nucleus, and midbrain periaqueductal gray. Figure 3 includes images of axonal labeling within the diencephalic PVT, PVH, lateral hypothalamic area, supraoptic nucleus, arcuate nucleus, and median eminence, while Figure 4 shows labeled fibers within the anterior BST and medial preoptic area. Table 4 lists all CNS regions in which EGFP immunoperoxidase-labeled axons were consistently observed in each of three male Gcg-Cre rats after bilateral cNTS-targeted injections and indicates the relative density of axonal labeling in each region.
Cre-dependent EGFP reporter immunolabeling in sections through multiple rostrocaudal levels of the caudal medulla in a representative homozygous male Gcg-Cre rat that received bilateral cNTS-centered injections of AAV1 (Addgene 20298). EGFP-immunopositive neurons and processes are labeled in green, while DBH-immunopositive noradrenergic neurons and processes are labeled in red. Panel A is the most caudal section depicted (bregma, −16.0 mm), and H is the most rostral (bregma, −14.2 mm). In B, the IRt region containing the letter “i” is shown at higher magnification in panel I. In E, the cNTS region containing the letter “j” is shown at higher magnification in panel J. J1 shows the same image (J) with EGFP labeling alone, and J2 shows the same image with DBH immunolabeling alone to demonstrate that no DBH-positive noradrenergic neurons are EGFP-positive. Scale bars: A–H, 500 μm; I, J1, 100 μm. c, central canal; AP, area postrema; DMX, dorsal motor nucleus of the vagus; IRt, intermediate reticular nucleus. See Extended Data Fig. 1-1D for an image of microglial Iba1 labeling within the caudal dorsomedial medulla after bilateral cNTS injections of AAV1.
Figure 1-1
Iba1 immunoperoxidase labeling in sections through injection sites in representative AAV-injected cases. A, PVH (higher magnification view of boxed region shown in a; image from case 22-422; unilateral AAVrg); B, vlBST (higher magnification view of boxed region shown in b; image from case 19-370; unilateral AAVrg); C, PVT (higher magnification view of boxed region shown in c; image from case 23-73; unilateral AAVrg); and D, cNTS (higher magnification view of boxed region shown in d; image from case 19-01; bilateral AAV1). In a-c, arrowheads point out Iba1-positive glia marking needle tracks (tissue disturbance) approaching and/or within injection sites. Note that the beveled tip of the Hamilton injection needle used for forebrain injections extends 0.3 mm beyond the delivery port. In D and d, general microglial reactivity is present within the caudal dorsomedial medullary region containing bilateral cNTS injection sites. Approximate bregma levels of each image: A, -1.5 mm; B, -0.2 mm; C, -1.9 mm; D, -14.3 mm. cNTS, caudal nucleus of the solitary tract; PVH, paraventricular nucleus of the hypothalamus; PVT, paraventricular thalamic nucleus; vlBST, ventrolateral bed nucleus of the stria terminalis. Download Figure 1-1, TIF file.
Cre-dependent EGFP reporter immunolabeling in sections through the pons (A,C; bregma, −9.25 mm) and midbrain (B,D; bregma, −7.6 mm) in a representative homozygous male Gcg-Cre rat that received bilateral cNTS-centered injections of AAV1 (Addgene 20298). EGFP-immunopositive axons are labeled green. In A, the pontine PBN region labeled “c” is shown at higher magnification in panel C. In B, the PAG region labeled “d” is shown at higher magnification in panel D. Scale bar: A, 500 μm; B,C, 200 μm; D, 100 μm. Aq, cerebral aqueduct; IV V, fourth ventricle; PAG, periaqueductal gray; PBN, parabrachial nucleus (m, medial; la, lateral); DR, dorsal nucleus raphé; scp, superior cerebellar peduncle.
Cre-dependent EGFP reporter immunolabeling in sections through three rostrocaudal levels of the diencephalon (A–D, bregma, −1.55 mm; E,F, bregma, −3.25 mm; G,H, bregma, −4.0 mm) in a representative homozygous male Gcg-Cre rat that received bilateral cNTS-centered injections of AAV1 (Addgene 20298). EGFP-immunopositive axons are labeled in green. In A, the small letter “b” includes the region containing the supraoptic nucleus of the hypothalamus, shown at higher magnification in panel B. In A, the small letter “c” includes the region containing the PVT, shown at higher magnification in panel C. In A, the small letter “d” includes the region containing the PVH, shown at higher magnification in panel D. In E, the small letter “f” includes the region containing the DMHp, shown at higher magnification in panel F. In G, the small letter “h” includes the region containing the PVp, shown at higher magnification in panel H. Scale bars: A, E, G, 500 μm; C, D, F, H, 100 μm; B, 50 μm. DMHp, dorsomedial hypothalamic nucleus, posterior part; fx, fornix; mt, mammillothalamic tract; opt, optic tract; PT, paratenial nucleus; PVH, paraventricular nucleus of the hypothalamus; PVp, periventricular hypothalamic nucleus, posterior part; PVT, paraventricular thalamic nucleus; SON, supraoptic nucleus of the hypothalamus; III V, third ventricle.
Cre-dependent EGFP reporter immunolabeling in sections through the rostral diencephalon (bregma ,−0.5 mm) in a representative homozygous male Gcg-Cre rat that received bilateral cNTS-centered injections of AAV1 (Addgene 20298). EGFP-immunopositive axons are labeled in green. In A, the small letter “b” includes the region containing the MPN, MPO, and AVP, shown at higher magnification in panel B. Scale bars: A, 500 μm; B, 100 μm. ac, anterior commissure; AVP, anteroventral preoptic nucleus; MPN, medial preoptic nucleus; MPO, medial preoptic area; och, optic chiasm; III V, third ventricle.
Figure 4-1
Immunoperoxidase labeling of AAVrg-encoded EGFP reporter illustrates the basis of Table 4 semiquantitative axonal labeling density values. The left column depicts labeling within the PVH (+++), dorsal raphe (++) and superior colliculus (+) of a representative homozygous female Gcg-Cre rat (19-366) that received a unilateral injection of AAVrg (Addgene 20298) into the anterior vlBST. The center column depicts similar axonal labeling in a representative heterozygous female Gcg-Cre rat (22-430) that received a unilateral injection of AAVrg (Addgene 20298) into the PVH. The right column depicts similar axonal labeling in a representative heterozygous female Gcg-Cre rat (23-73) that received a unilateral injection of AAVrg (Addgene 20298) into the PVT. PVH, paraventricular nucleus of the hypothalamus; PVT, paraventricular thalamic nucleus; vlBST, ventrolateral bed nucleus of the stria terminalis. Download Figure 4-1, TIF file.
Number of reporter-labeled neurons in the cNTS and IRt after injection of Cre-dependent AAV1 or AAVrg
Distribution and relative density of EGFP-positive axonal labeling in rats after injection of Cre-dependent AAV1 or AAVrg (see Key below table)
Cre-dependent tracing of axonal projections arising from IRt GLP1 neurons
Cre-dependent AAV1 (Addgene 51502) injected bilaterally into the IRt led to EGFP reporter labeling of IRt and cNTS neurons, dendrites, and axonal projections (Fig. 5). Extended Data Figure 5-1C shows the approximate location and size of a 500 nl injection into the IRt region. As stated above, we speculate that labeling of cNTS GLP1 neurons after IRt-targeted injections of Cre-dependent AAV1 is due to anterograde transsynaptic delivery to postsynaptic cNTS GLP1 neurons (see Discussion). Approximately three times as many EGFP-positive cell bodies were present within the IRt versus the cNTS (Table 3). The average number of AAV1 reporter-labeled IRt cells per section (9.3 ± 2.2) was similar to the average number of tdTom reporter-labeled IRt neurons (9.7 ± 1.9) in surgically naive adult Gcg-Cre/tdTom reporter rats.
Cre-dependent EGFP reporter immunolabeling in sections through multiple rostrocaudal levels of the caudal medulla in a representative female Gcg-Cre/tdTom reporter rat that received bilateral IRt-centered injections of AAV1 (Addgene 51502). EGFP-immunopositive neurons and processes are labeled in green, while red RFP immunolabeling identifies tdTom-positive neurons and processes (i.e., Gcg-expressing cells). Panel A is the most caudal section depicted (bregma, −16.0 mm), and J is the most rostral (bregma, −13.76 mm). In E, the cNTS region indicated by the letter “k” is shown at higher magnification in panel K. In I, the IRt regions indicated by the small letters “l” and “m” are shown at higher magnification in panels L and M, respectively. K1 shows the same cNTS image as panel K but with EGFP labeling alone, and K2 shows the same cNTS image with RFP immunolabeling alone, to demonstrate that all EGFP-positive cNTS neurons are RFP-positive (i.e., Gcg-expressing). L1 shows the same IRt image as panel L but with EGFP labeling alone, and L2 shows the same IRt image with RFP immunolabeling alone, to demonstrate that all EGFP-positive IRt neurons are RFP-positive (i.e., Gcg-expressing). M1 shows the same IRt image as panel M but with EGFP labeling alone, and M2 shows the same IRt image with RFP immunolabeling alone, to demonstrate that all EGFP-positive IRt neurons are RFP-positive (i.e., Gcg-expressing). Scale bars: A–J, 500 μm; K–M, 100 μm. c, central canal; AP, area postrema; DMX, dorsal motor nucleus of the vagus. See Extended Data Fig. 5-1C to visualize the location and size of a 500 nl injection into the IRt region.
Figure 5-1
A, the location and size of a PVT-focused CTb injection is shown across several sections from a 1:6 series, corresponding to approximate bregma levels -1.5 to -1.4 mm (rat perfused the day after injection). B, the location of a PVH-focused injection of AAVrg containing CTb is shown at higher magnification (rat perfused 3 wks after injection), corresponding to approximate bregma level -1.5 mm. CTb labeling identifies dense clusters of neurons within the PVH injection site. Fluorescent CTb labeling was captured at high resolution using tiling, and then the resulting image inverted to better reveal tissue landmarks. C, the location and size of an IRt-focused injection of FG in a Gcg-Cre/tdTom reporter rat (rat perfused day of injection), corresponding to approximate bregma level -14.2 mm. FG native fluorescence (green) is centered slightly dorsolateral to the position of tdTom + (i.e., Gcg-expressing) IRt neurons (red), leading to subsequent adjustment of stereotaxic coordinates for IRt injections. Gcg-expressing neurons within the cNTS (red) also are visible, completely outside of the FG injection site. cNTS, caudal nucleus of the solitary tract; IRt, intermediate reticular nucleus; PVH, paraventricular nucleus of the hypothalamus; PVT, paraventricular nucleus of the thalamus. Download Figure 5-1, TIF file.
Images in Figure 5 show Cre-dependent EGFP labeling at multiple rostrocaudal levels through the caudal medulla in a representative female Gcg-Cre/tdTom reporter rat that received bilateral AAV injections into the IRt (Table 1). Double labeling for RFP in the same medullary tissue sections (Fig. 5) identified tdTom reporter labeling of Gcg-expressing neurons. As expected, all EGFP-positive neurons were RFP-positive, evidence for the phenotypic specificity of Cre-mediated AAV reporter expression (Fig. 5).
After bilateral IRt injections, the distribution of EGFP-positive axons throughout the brain was qualitatively indistinguishable from that observed in rats after bilateral cNTS injections. The relative density of axonal labeling in some CNS regions was less robust after IRt versus cNTS injections, whereas axonal labeling in other CNS regions was more robust after IRt versus cNTS injections (Table 4). Table 4 lists each CNS region in which EGFP immunoperoxidase-labeled axons were consistently observed in each of the three female Gcg-Cre/tdTom rats after bilateral IRt injections and indicates the relative density of axonal labeling in each region.
Forebrain delivery of Cre-dependent AAVrg to label the axon collaterals of GLP1 projection neurons
After Cre-dependent AAVrg was microinjected into either the PVH, vlBST, or PVT, reporter-labeled neuronal cell bodies were observed in both the cNTS and IRt and in no other CNS regions. The number of labeled neurons within the cNTS and IRt was determined in each analyzed AAVrg injection case using hindbrain tissue sections processed for nickel–DAB immunoperoxidase labeling to localize EGFP/EYFP reporter (Table 3; Extended Data Figs. 6-1–6-3). The average number of labeled neurons in each hindbrain region was quite similar across the three forebrain injection sites, with approximately twice as many labeled cNTS versus IRt neurons per section (Table 3).
Fluorescent reporter labeling filled the dendrites of retrogradely labeled cNTS and IRt neurons and also filled their axons. Labeled axons were present within each forebrain injection target, as expected. Surprisingly, labeled axons also were present within every CNS region that contained axonal labeling in rats that received injections of Cre-dependent AAV1 directly into the cNTS or IRt. Labeling in a representative female Gcg-Cre rat that received a unilateral PVT-targeted injection of Cre-dependent AAVrg is shown in Figure 6, which depicts labeled cell bodies and processes in the cNTS and IRt and axons within the PVT injection site, PVH, BST, and thoracic spinal cord. Extended Data Figures 6-1–6-3 show slide scans of retrogradely labeled cNTS and IRt neurons after AAVrg injections into the PVT, PVH, or vlBST, respectively. Table 4 lists each CNS region in which EGFP immunoperoxidase-labeled axons were observed consistently in each male and female Gcg-Cre or Gcg-Cre/tdTom rat after unilateral AAVrg injections into any of the three forebrain sites and indicates the relative density of axonal labeling in each region. Notably, the distribution of axonal labeling was similar across rats, regardless of the injection site. Extended Data Figure 4-1 provides a visual guide for the semiquantitative assessment of axonal labeling densities.
Cre-dependent EGFP reporter immunolabeling (green) in a representative female Gcg-Cre rat that received a unilateral PVT-targeted injection of AAVrg (Addgene 20298). Thus, the EGFP-labeled axons present in every CNS region arise from the subset of cNTS and/or IRt neurons that were retrogradely labeled from the PVT injection site. Red immunolabeling identifies DBH-positive noradrenergic neurons and processes. In the caudal medullary section shown in panel A (bregma, −14.5 mm), the small letter “b” is located adjacent to the cNTS, shown at higher magnification in panel B. In A, the small letter “c” is positioned in the medullary reticular formation near the IRt, shown at higher magnification in panel C. EGFP-positive neurons within the cNTS and IRt are distinct from DBH-positive noradrenergic neurons. In D, a section through the diencephalon (bregma, −1.78 mm) includes the region of the PVT-centered AAVrg injection site with dense EGFP-positive axons and also includes EGFP-positive axon collaterals within the PVH. The PVH and PVT from the same case (but in a different tissue section; bregma, −1.54 mm) are shown at higher magnification in panels E and F, respectively. Panel G depicts a more rostral section with EGFP-labeled axons in the anterior vlBST and MPN (bregma, −0.46 mm). Panel H is a horizontal section through the thoracic spinal cord, showing EGFP-labeled axons traveling in the white matter and branching within the gray matter. Scale bars: A, D, G, 500 μm; all other panels, 100 μm. ac, anterior commissure; MPN, medial preoptic nucleus; och, optic chiasm; PVH, paraventricular nucleus of the hypothalamus; PVT, paraventricular thalamic nucleus; vlBST, ventrolateral bed nucleus of the stria terminalis; III V, third ventricle. See Extended Data Fig. 1-1B to visualize Iba1-positive microglia marking an injection needle track into the PVH, Extended Data Fig. 5-1A to visualize the location and size of a 500 nl injection into the PVT, and Extended Data Fig. 6-1 for a slide scan of retrogradely labeled GLP1 neurons within the cNTS and IRt in a representative case.
Figure 6-1
Slide scan of immunoperoxidase EGFP reporter labeling in a 1:6 series of hindbrain sections in a representative heterozygous female Gcg-Cre rat (23-73) that received a unilateral injection of AAVrg (Addgene 20298) into the paraventricular thalamic nucleus (PVT). Retrogradely-labeled neurons within the cNTS and IRt can be viewed at higher magnification by zooming in. cNTS, caudal nucleus of the solitary tract; IRt, intermediate reticular nucleus. Download Figure 6-1, TIF file.
Figure 6-2
Slide scan of immunoperoxidase EGFP reporter labeling in a 1:6 series of hindbrain sections in a representative heterozygous female Gcg-Cre rat (22-430) that received a unilateral injection of AAVrg (Addgene 20298) into the paraventricular nucleus of the hypothalamus (PVH). Retrogradely-labeled neurons within the cNTS and IRt can be viewed at higher magnification by zooming in. cNTS, caudal nucleus of the solitary tract; IRt, intermediate reticular nucleus. Download Figure 6-2, TIF file.
Figure 6-3
Slide scan of immunoperoxidase EGFP reporter labeling in a 1:6 series of hindbrain sections in a representative homozygous female Gcg-Cre rat (19-366) that received a unilateral injection of AAVrg (Addgene 20298) into the anterior ventrolateral bed nucleus of the stria terminalis (vlBST). Retrogradely-labeled neurons within the cNTS and IRt can be viewed at higher magnification by zooming in. cNTS, caudal nucleus of the solitary tract; IRt, intermediate reticular nucleus. Download Figure 6-3, TIF file.
Figures 7⇓⇓⇓–11 depict Cre-dependent reporter labeling in a representative homozygous female Gcg-Cre rat that received unilateral injections of AAVrg into both the PVH and ipsilateral vlBST. The AAVrg injection sites in PVH and vlBST (the latter also marked by CTb immunolabeling) are illustrated in Figure 7, which also depicts the rostrocaudal distribution of labeled cNTS and IRt neurons. Many cNTS and IRt neurons were double-labeled, expressing fluorescent reporter from both forebrain injection sites (Fig. 7). Representative images of labeling are included for the pons and midbrain (Fig. 8), thalamus (Fig. 9), hypothalamus (Figs. 9–11), and BST (Fig. 11). Single- and double-labeled axon collaterals arising from retrogradely labeled hindbrain GLP1 neurons were observed in every CNS region that contained anterograde axonal labeling in rats that received cNTS- or IRt-centered AAV1 injections (Table 4).
RFP and EGFP reporter immunolabeling in sections through multiple rostrocaudal levels of the caudal medulla (A–J) and through PVH and BST injection sites (K,L) in a representative female homozygous Gcg-Cre rat. This rat received unilateral injection of a Cre-dependent AAVrg expressing mCherry (Addgene 20297) into the PVH and an ipsilateral injection of Cre-dependent AAVrg expressing EGFP (Addgene 20298) into the anterior vlBST; CTb was included in the vlBST injectate to mark the injection site. EGFP-immunopositive neurons and processes are labeled in green, while mCherry (RFP)-immunopositive neurons and processes are labeled in red. Panel A is the most caudal medullary section depicted (bregma, −15.46 mm) and H the most rostral (bregma, −14.16 mm). In A, the IRt region containing the letter “j” is shown at higher magnification in panel J. In C, the cNTS region containing the letter “i” is shown at higher magnification in panel I. The colored arrows in panel I point out cNTS neurons that are single-labeled for EGFP (green), for RFP (red), or double-labeled for both fluorophores (yellow). Panels I1, I2, and I3 show higher-magnification images of these single- and double-labeled cNTS neurons. Panel J shows high-magnification images of a single-labeled IRt neuron (green arrow) and two double-labeled IRt neurons (yellow arrows). K, RFP-positive axonal labeling in the PVH injection site. The area indicated by the arrow is enlarged in the inset to show EGFP fiber labeling (i.e., axon collateral from the vlBST injection site). L, CTB (blue) immunolabeling marks the anterior vlBST site (arrow) where EGFP-expressing AAVrg plus CTb was injected. EGFP axonal labeling in the injection site can be seen at higher magnification in the inset. Red/yellow RFP-positive and double-labeled (RFP + EGFP) axons can also be seen. Scale bars: A–H, 500 μm; I, 50 μm; J, 20 μm. ac, anterior commissure; AP, area postrema; c, central canal; III V, third ventricle. See Extended Data Fig. 1-1A to visualize Iba1-positive microglia marking an injection needle track above the PVH, Extended Data Fig. 5-1B to visualize the location and size of a PVH injection site, and Extended Data Figs. 6-2 and 6-3 for slide scans of retrogradely labeled GLP1 neurons within the cNTS and IRt in representative cases that received PVH- or anterior vlBST-centered injections of AAVrg.
RFP (mCherry) and EGFP reporter immunolabeling in pontine sections in a representative female homozygous Gcg-Cre rat that received unilateral injection of Cre-dependent AAVrg expressing mCherry (Addgene 20297) into the PVH and an ipsilateral injection of Cre-dependent AAVrg expressing EGFP (Addgene 20298) into the anterior vlBST. In A (bregma, −8.85 mm), the PBN region indicated by the small letter “c” is shown at higher magnification in panel C. In B (bregma, −8.5 mm), the region of the LDT indicated by the small letter “d” is shown at higher magnification in panel D. Scale bars: A, B, 500 μm; C, D, 100 μm. Aq, cerebral aqueduct; LDT, laterodorsal tegmental nucleus; PAG, periaqueductal gray; PBN, parabrachial nucleus (la, lateral, m, medial); scp, superior cerebellar peduncle; IV V, fourth ventricle.
RFP (mCherry) and EGFP reporter immunolabeling in diencephalic sections in a representative female homozygous Gcg-Cre rat that received unilateral injection of Cre-dependent AAVrg expressing mCherry (Addgene 20297) into the PVH and an ipsilateral injection of Cre-dependent AAVrg expressing EGFP (Addgene 20298) into the anterior vlBST. In A (bregma, −1.78 mm), the PVT region indicated by the small letter “b” is shown at higher magnification in panel B, the PVH region indicated by the small letter “d” is shown at higher magnification in panel D, and the SON region indicated by the small letter “c” is shown at higher magnification in panel C. The boxed PVH region (labeled “e”) in panel D is shown at higher magnification in panel E, in which single- and double-labeled axons are indicated by colored arrows. Panels E1 and E2 show these axons at higher magnification and slightly rotated in 3D maximum projection images. Scale bars: in A, 500 μm; in B, 100 μm; in C and E, 20 μm; in D, 200 μm. opt, optic tract; PVH, paraventricular nucleus of the hypothalamus; PVT, paraventricular thalamic nucleus; SON, supraoptic nucleus of the hypothalamus; III V, third ventricle.
RFP (mCherry) and EGFP reporter immunolabeling in caudal diencephalic sections in a representative female homozygous Gcg-Cre rat that received unilateral injection of Cre-dependent AAVrg expressing mCherry (Addgene 20297) into the PVH and an ipsilateral injection of Cre-dependent AAVrg expressing EGFP (Addgene 20298) into the anterior vlBST. In A (bregma, −3.8 mm), the small letter “d” indicates the region of the posterior DMH, shown at higher magnification in panel D. In B, the small letter “e” indicates the region of the ARH, shown at higher magnification in panel E. In C, the posterior hypothalamic region indicated by the small letter “f” is shown at higher magnification in panel F. Scale bars: B, C, 500 μm; D–F, 100 μm. ARH, arcuate nucleus of the hypothalamus; cp, cerebral peduncle; DMHp, posterior dorsomedial nucleus of the hypothalamus; mt, mammillothalamic tract; pc, posterior commissure; III V, third ventricle.
RFP (mCherry) and EGFP reporter immunolabeling in a rostral diencephalic section in a representative female homozygous Gcg-Cre rat that received unilateral injection of Cre-dependent AAVrg expressing mCherry (Addgene 20297) into the PVH and an ipsilateral injection of Cre-dependent AAVrg expressing EGFP (Addgene 20298) into the anterior vlBST. In A (bregma, −0.2 mm), the boxed anterior vlBST region indicated by the letter “c” is shown at higher magnification in panel C, and the boxed MPO region indicated by the letter “b” is shown at higher magnification in panel B. Colored arrows in panel C point out single- and double-labeled axons, which are shown at higher magnification in slightly rotated 3D maximum projection images in panels C1 and C2. Scale bars: A, B, 500 μm; C, 20 μm. ac, anterior commissure; och, optic chiasm; MPN, medial preoptic nucleus; vlBST, ventrolateral bed nucleus of the stria terminalis.
Discussion
Results from this study demonstrate that populations of hindbrain Gcg-expressing neurons that provide axonal input to the PVH, PVT, and anterior vlBST in rats collectively send axons to CNS regions extending rostrocaudally from the olfactory tubercle to the spinal cord. Indeed, every central region previously reported to contain GLP1-immunopositive axons (and several additional regions) contained labeled axon collaterals arising from thalamic, hypothalamic, and limbic forebrain projection-defined populations of GLP1 neurons. As reviewed elsewhere, broadly projecting neuromodulatory systems originating in the brainstem reticular formation (which includes the cNTS and IRt) are organized to support exteroceptive and interoceptive sensory-driven transitions in the behavioral state (Horn et al., 2020; Collins et al., 2023). Similarly, the collateralized axonal projections of collective groups of GLP1 neurons that innervate the PVH, PVT, or anterior vlBST position these neural populations to broadcast signals across widely distributed central postsynaptic targets, potentially in a metabolic state-dependent manner.
As summarized in Table 4, Cre-dependent viral tracing identified axonal labeling in a large number of CNS regions, including some that (to our knowledge) have not previously been specified as containing GLP1-immunopositive fibers. These additional regions include the olfactory tubercle, lateral habenula, several thalamic nuclei, the supramammillary nucleus, the inferior colliculus and anterior pretectal nucleus, the red nucleus, the substantia nigra, and the medullary parapyramidal nucleus. While the presence of labeled axons in these regions represents a new finding, it is unsurprising, given that the cNTS and medullary reticular formation are known to project to these brain regions, some of which are known to express GLP1R and are implicated in GLP1R-mediated behavioral functions (Guevara-Aguilar et al., 1987; Ruggiero et al., 1998; Llewellyn-Smith et al., 2013; Cork et al., 2015; Tuesta et al., 2017; Graham et al., 2020; López-Ferreras et al., 2019, 2020; Farkas et al., 2021; X.-Y. Chen et al., 2021). The distribution of axons observed in the present rat study also is similar to the distribution of genetically encoded, reporter-labelled axons arising from Gcg-expressing hindbrain neurons in mice (Llewellyn-Smith et al., 2011, 2013).
It appears that all central regions that receive axonal input from GLP1 neurons contain GLP1 binding sites, neurons that express Glp1r mRNA, and/or axonal input from such neurons (Göke et al., 1995; Graham et al., 2020; Farkas et al., 2021). In this regard, GLP1R protein is often localized in the membrane of axon varicosities and terminals (Farkas et al., 2021), consistent with evidence for GLP1R-mediated presynaptic modulation of neurotransmitter release in many brain regions (Secher et al., 2014; Liu and Pang, 2016; Rebosio et al., 2018; Péterfi et al., 2021; Povysheva et al., 2021; X.-Y. Chen et al., 2021). Glp1r also is expressed by some cells within the cortex and hippocampus (Hsu et al., 2015; Graham et al., 2020; Farkas et al., 2021), but the present results are consistent with previous reports that the axons of hindbrain GLP1 neurons do not reach these brain areas. It has been suggested that GLP1 released from hindbrain neurons might access hippocampal receptors through volume transmission (Hsu et al., 2015). Alternatively, Glp1r-expressing cortical and hippocampal neurons may innervate subcortical regions that receive GLP1 axonal input.
The “sprinkler-like” distribution of axon collaterals arising from collective populations of projection-defined GLP1 neurons in the present study is consistent with experimental evidence that GLP1R signaling across multiple CNS regions produces a constellation of behavioral and physiological responses that can be grouped together under the umbrella of homeostatic threat or stress responses. Given that stimulus-induced activation of hindbrain GLP1 neurons depends on the animal's metabolic state (discussed further, below), our anatomical findings support the view that central GLP1 signaling pathways are arranged to modulate synaptic transmission in widespread central regions in a state-dependent rather than a site-dependent manner. Stress and metabolism are closely linked, and the axons of GLP1 neurons reach most (if not all) central regions of relevance to energy balance, including regions that control food intake, sympathetic outflow, and the endocrine HPA axis (Holt and Trapp, 2016; Maniscalco and Rinaman, 2017; Holt and Rinaman, 2022; Table 4). We discovered that GLP1 neurons express c-Fos in rats after acute exposure to diverse cognitive and interoceptive threats (Rinaman, 1999a, 1999b; Rinaman and Comer, 2000; Maniscalco et al., 2015), but not after exposure to physiologically significant but nonstressful stimuli, such as consuming a large anticipated meal (Rinaman, 1999a). Additional findings from our laboratory suggest that central GLP1 signaling provides a link between metabolic and emotional states (Maniscalco et al., 2015; Maniscalco and Rinaman, 2018). For example, overnight fasting completely blocks the ability of acute stress to activate c-Fos in GLP1 neurons and also attenuates threat-induced c-Fos in the downstream projection site (Maniscalco et al., 2015). Fasting also reduces acoustic startle and increases open-arm activity in the elevated-plus maze, supporting a causal link between reduced GLP1 signaling and reduced arousal/vigilance and behavioral inhibition/avoidance threat responses (Maniscalco et al., 2015; Maniscalco and Rinaman, 2018). Furthermore, knockdown of GLP1R in the anterior vlBST reduces behavioral responses to stress, similar to the effects of caloric deprivation (Zheng et al., 2019).
Published studies using pharmacological approaches indicate that GLP1R signaling in a number of brain regions suppresses food intake through mechanisms that include increased gastrointestinal satiation and/or satiety, induction of nausea, stress-induced hypophagia, and reduced motivation for food reward (Skibicka, 2013; Eren-Yazicioglu et al., 2021; Williams, 2022). The effects of central GLP1R signaling to reduce motivation for food reward are evident in both rodents and humans, in which GLP1R agonists also reduce the rewarding/reinforcing properties of addictive drugs such as cocaine, amphetamine, opioids, alcohol, and nicotine (Hayes and Schmidt, 2016; Brunchmann et al., 2019; Eren-Yazicioglu et al., 2021; Klausen et al., 2022). It is known that caloric restriction enhances both food and drug reward-driven behaviors in rats and mice (Hsu et al., 2018; Carr, 2020). We speculate that this effect is at least partly due to reduced recruitment of hindbrain GLP1 neurons whose axons collateralize extensively to target central reward-related areas (Table 4), including the substantia nigra, ventral tegmental area, laterodorsal tegmental nucleus, PVT, ventral striatum (i.e., olfactory tubercle and nucleus accumbens), and lateral septum.
Our anatomical findings support the view that central GLP1 signaling pathways are arranged to modulate synaptic transmission in widespread central regions in a state-dependent rather than a site-dependent manner. This has significant implications in the design and interpretation of results from studies that manipulate central GLP1 signaling. GLP1 neurons are glutamatergic (Zheng et al., 2015) and are synaptically coupled in rats (Card et al., 2018), suggesting that recruitment of a subset of GLP1 neurons increases the likelihood that additional GLP1 neurons will be recruited. Indeed, our AAV1 tracing results support the view that cNTS GLP1 neurons provide synaptic input to IRt GLP1 neurons and vice versa (discussed further below). Considering this, and also given the demonstrably widespread axon collateralization of at least some populations of Gcg-expressing hindbrain neurons, care should be taken when interpreting the behavioral or physiological impact of optogenetic or chemogenetic stimulation of GLP1 cell bodies within the cNTS or IRt or the impact of stimulating GLP1 axons in different central regions. Such approaches may activate GLP1 axon collaterals in other CNS regions, and may depolarize or induce action potentials in additional hindbrain GLP1 neurons, which could lower the threshold for synaptic release of GLP1/glutamate in most if not all CNS regions to which these hindbrain neurons project.
Study limitations
A limitation of our results is that they describe the axonal projections of tracer-labeled subpopulations of GLP1 projection neurons. While the apparent majority of Gcg-expressing cNTS and IRt neurons were labeled after Cre-dependent AAV1 was injected directly into each hindbrain region, smaller subsets of Gcg-expressing GLP1 neurons were retrogradely labeled after site-specific injections of AAVrg into the PVH, PVT, or anterior vlBST (Fig. 7, Table 3). This could reflect the presence of Gcg neurons within the cNTS and/or IRt that do not project to the PVH, PVT, or anterior vlBST or could be due to technical reasons such as insufficient axonal uptake of AAVrg for viral transduction and/or injection sites that did not cover the neuroanatomical boundaries of targeted nuclei. We also could not trace the axon collaterals of individual GLP1 neurons, some of which may project to only one or a limited subset of central targets. These limitations might be addressed in future work through approaches such as multiplex analysis of projections by sequencing or Cre-dependent Brainbow reporter expression (Kebschull et al., 2016; Han et al., 2018; Sakaguchi et al., 2018; X. Chen et al., 2019; Wu et al., 2021) to identify the axonal branches and target sites of individual GLP1 neurons within the cNTS and IRt.
A second limitation of our study is that we could not determine whether or not the axonal projections of GLP1 neurons in the cNTS overlap completely with the axonal projections of GLP1 neurons in the IRt. We observed that bilateral cNTS-centered AAV1 injections efficiently labeled most Gcg-expressing neurons in the cNTS but also labeled neurons within the IRt. Similarly, bilateral IRt injections of AAV1 consistently labeled most Gcg-expressing IRt neurons but also labeled cNTS neurons. The stereotaxic coordinates for IRt injections were >1 mm lateral and ventral to the cNTS, such that each injection avoided direct involvement of the alternate region (Extended Data Figs. 1-1C, 2-1). The dendrites and axons of Gcg-expressing neurons extend into the medullary reticular formation surrounding each targeted hindbrain region, perhaps contributing to Cre-mediated labeling of GLP1 neurons with cell bodies outside of the injection site. However, similar 500 nl injections of other Cre-dependent AAV serotypes (i.e., AAV8, AAV2, and AAV5) into the cNTS in Gcg-Cre rats do not lead to viral reporter expression by IRt neurons (Zheng et al., 2022 and unpublished observations; Lorena-Lopez et al., 2023). Conversely, the AAV1 serotype used for hindbrain injections in the present study can be transported both retrogradely (Haery et al., 2019) and transsynaptically in the anterograde direction (Zingg et al., 2022) to generate Cre-dependent reporter labeling. Thus, labeling of Gcg-expressing cNTS neurons after IRt injections of AAV1 and labeling of IRt neurons after cNTS injections of AAV1 could reflect Cre-dependent reporter expression after retrograde and/or transsynaptic anterograde transport among GLP1 neurons in each hindbrain region. We previously reported that Gcg-expressing IRt neurons do not appear to provide synaptic input to Gcg-expressing cNTS neurons in mice (Holt et al., 2019), but this has not been examined directly in rats.
Summary
Results from the present study challenge the existence of projection-specific groups of hindbrain GLP1 neurons that target only one brain region, such as the hypothalamic PVH, the thalamic PVT, or the limbic forebrain (BST). Instead, collective populations of GLP1 projection neurons in the cNTS and IRt that innervate these three forebrain regions give rise to highly collateralized axons that innervate all central regions that receive GLP1 axonal input. Our results do not negate the possibility that some GLP1 neurons within the cNTS or IRt have only local projections or that some GLP1 neurons project exclusively to one or a few central regions; this remains to be determined. Regardless, the present anatomical findings along with our previously published work support our emerging view that collective populations of projection-defined GLP1 neurons broadcast their neuromodulatory influence over synaptic signaling in widespread central regions in a metabolic state-dependent manner.
Footnotes
Reported research was funded by the National Institutes of Health Grant MH059911.
The authors declare no competing financial interests.
- Correspondence should be addressed to Linda Rinaman at lrinaman{at}fsu.edu.
