Leptin suppresses development of GLP-1 inputs to the paraventricular nucleus of the hypothalamus

In neonatal mice, GLP-1 immunoreactive fibers were detected in the PVH as early as P6 (Figure 1A). Although GLP-1 fibers are visible at P6, their density is sparse at this age, and the preautonomic compartment of the PVH is nearly devoid of GLP-1 innervation. By P10, the density of GLP-1 immunoreactive fibers increases in neuroendocrine components of the PVH, with the dorsal region of the medial parvicellular part (PVHmpd) more heavily innervated than the ventral parvicellular (PVHmpv) compartment, and at this age, GLP-1 projections begin to extend into the lateral zone of the posterior magnocellular (PVHpml) compartment (Figure 1B). This overall pattern of GLP-1 fiber distribution was maintained through P16, although the density of GLP-1 fibers increased substantially in both the PVHmpd and PVHpml compartments (Figure 1C). By P24, the density of GLP-1 immunolabeled fibers was elevated further, with a higher density observed in the PVHmpd region, compared with that of the PVHpml (Figure 1D). At 2 months of age, the density and distribution of GLP-1 immunoreactive fibers in the PVH achieves levels that are similar to that of adult mice (Figure 1E). Throughout development, we observed GLP-1 immunoreactive fibers in the most rostral portions of the PVH (PVHap), or within the caudal preautonomic part of the PVH (PVHlp), but to a lesser degree than the neuroendocrine PVHmpd and PVHpml compartments.

To verify expression of LepRb in NTS neurons during postnatal development, we utilized LepRb-Cre mice to target expression of tdTomato (LepRb-Cre::tdTom mice). Moreover, these LepRb neurons appear to provide strong inputs to the PVH. To label terminal fields of NTS LepRb-expressing neurons, an AAV-encoding Cre-dependent EGFP virus was injected into the NTS of LepRb-Cre mice. The virus robustly labeled LepRb-expressing neurons in the NTS (Figure 2F,G) and revealed that the PVH receives direct projections from these neurons (Figure 2H). To confirm that LepRb are functional in NTS neurons during postnatal development we measured pSTAT3 immunolabeling in leptin-treated LepRb-Cre::tdTom mice. Following i.p. leptin injection on P16, 85% of NTS LepRb-Cre::tdTom labeled neurons were positive for pSTAT3 immunoreactivity, in contrast to the lack of pSTAT3 immunolabeling in saline-injected controls (Figure 2A–B,E).

To visualize PPG neurons and their inputs into the PVH, a synaptophysin-tdTomato fusion protein was expressed under the Gcg promoter, resulting in GCG-Cre::SynTom mice. The distribution of SynTom-labeled neurons in GCG-Cre::SynTom mice was nearly identical to previous reports, with expression located exclusively to the medulla (Jin et al., 1988; Vrang et al., 2007; Llewellyn-Smith et al., 2011; Llewellyn-Smith et al., 2013; Gu et al., 2013; Gaykema et al., 2017). The majority of GCG-Cre::SynTom-labeled neurons were observed in the NTS, with a limited number of neurons (approximately 3–4 neurons per section) located in the adjacent intermediate reticular nucleus (IRT). We did not observe GCG-Cre::SynTom cell bodies outside of the medulla (Figure 2—figure supplement 1). In addition, GCG-Cre::SynTom mice were utilized to determine if PPG neurons are responsive to leptin during development. Following i.p. leptin injection at P16, 82% of SynTom-labeled PPG neurons in the NTS displayed pSTAT3 immunoreactivity, as did the majority of PPG neurons located in the IRT, indicating that PPG neurons are responsive to leptin during postnatal development, which may impact targeting of GLP-1 projections to the hypothalamus (Figure 2C–E).

To determine if leptin is required for normal development of GLP-1 projections to the hypothalamus, the density of GLP-1 immunoreactive inputs to the PVH was evaluated in leptin-deficient mice. Although immunolabeled GLP-1 fibers were clearly apparent in the PVH by P6 and increased in density over time, no statistically significant differences in their density were detected in Lep^ob/ob mice compared to WT controls, in either the PVHmpd or the PVHpml at either P6 or P10 (Figure 1). However, by P16, the density of GLP-1 immunoreactive fibers in the PVHmpd of Lep^ob/ob mice was 32% greater than that of WT mice (Figure 1C,H,M; p=0.037). At this age, no significant differences were detected in the density of GLP-1 immunoreactive fibers in the PVHpml (Figure 1C,H,M). At P24, the density of labeled fibers increased by 46.3% in the PVHmpd,and by 34.2% in the PVHpml, of Lep^ob/ob mice, compared with WT controls (Figure 1D,I,N; p=0.0305 and p=0.0493, respectively). The increase in GLP-1 fiber density in the PVH of Lep^ob/ob mice was maintained into adulthood (P60) in both the PVHmpd and PVHpml (70.2 and 61.3% higher relative to WT mice, respectively; Figure 1E,J,O; p=0.0061 and p=0.0143, respectively). Because expression of GLP-1 immunoreactivity may change in the absence of leptin, we confirmed these findings by using leptin-deficient GCG-Cre::SynTom (GCG-Cre::SynTom::Lep^ob/ob ) mice to visualize axons derived from PPG neurons. In alignment with our immunohistochemical results, at P16 the density of SynTom-labeled inputs was 37.4% higher in the PVHmpd of GCG-Cre::SynTom::Lep^ob/ob mice, and 33.8% higher in the PVHpml, compared with GCG-Cre::SynTom::WT mice (Figure 3A–C; p=0.0308 and p=0.0408, respectively). Furthermore, the increase in the density of GLP-1 axonal labeling in the PVH was maintained into adulthood: GCG-Cre::SynTom::Lep^ob/obmice displayed a 50% increase in the density of GCG-Cre::SynTom inputs to the PVHmpd, and a 40.2% increase in the PVHpml, compared with that of GCG-Cre::SynTom::WT mice at P60 (Figure 3D–F; p=0.0089 and p=0.0045 respectively). Leptin does not affect GLP-1 axonal inputs to the ARH, as no changes were detected in GLP-1 axon density between WT and GCG-Cre::SynTom::Lep^ob/obmice, indicating that leptin does not appear to uniformly impact the projection fields of PPG neurons (Figure 3G–I). Additionally, the increase in the density of SynTom inputs to the PVH of GCG-Cre-SynTom::Lep^ob/ob mice did not appear to be associated with any changes in the density or distribution of PPG neurons (Figure 3J–L), suggesting that leptin impacts targeting of axons to the PVH, but does not alter PPG neuron number.

To determine if targeting of GLP-1 inputs to specific subpopulations of PVH neurons is impacted by leptin, we evaluated the density of GCG-Cre::SynTom terminals onto corticotrophin-releasing hormone (CRH) and Oxytocin neurons in neuroendocrine compartments of the PVH in leptin-deficient mice. CRH and Oxytocin neurons in the PVH were visualized using immunohistochemistry in GCG-Cre::SynTom::Lep^ob/ob mice and GCG-Cre::SynTom::WT controls. In WT mice, GLP-1 inputs to CRH neurons appeared more numerous than those to Oxytocin neurons: the density of GCG-Cre::SynTom-labeled inputs to CRH neurons were approximately 50% greater than those to Oxytocin neurons (Figure 4A,D). Moreover, leptin appeared to preferentially impact the density of GLP-1 inputs onto CRH neurons, as we observed a greater number of GCG-Cre::SynTom-labeled terminals in close association with labeled CRH neurons in adult GCG-Cre::SynTom::Lep^ob/ob mice than in GCG-Cre::SynTom::WT controls (Figure 4D–F; p=0.0017). However, leptin does not appear to alter the density of GLP-1 projections to oxytocin neurons because we did not detect a significant change in the mean number of GCG-Cre::SynTom terminals onto Oxytocin neurons (Figure 4A–C).

To determine if leptin alters the ability of PPG neurons to convey visceral sensory information to the hypothalamus, we injected CCK i.p. to activate vagal afferents and measured the density of Fos immunostained nuclei in the NTS and PVH of GCG-Cre::SynTom::Lep^ob/ob mice and WT controls. Consistent with previous findings in rats (Rinaman, 1999; Maniscalco and Rinaman, 2013), CCK administration results in an induction of Fos immunoreactivity in the majority of PPG neurons. Injection of CCK to GCG-Cre::SynTom::WT mice caused an induction of Fos immunoreactivity in 95% of labeled PPG neurons (located in the NTS and IRT), compared with saline-injected controls (Figure 5 insets A, B, D). Similar levels of Fos labeling were observed in PPG neurons of GCG-Cre::SynTom::WT and GCG-Cre::SynTom::Lep^ob/ob mice following CCK injection (Figure 5C inset,D). However, CCK administration resulted in a significantly higher number of Fos immunolabeled nuclei in the PVHmpd of GCG-Cre::SynTom::WT mice than in saline-injected GCG-Cre::SynTom::WT controls (Figure 5A–D; p<0.0002). Further, injection of CCK resulted in a 64% increase in the number of Fos immunolabeled nuclei in the PVHmpd of GCG-Cre::SynTom::Lep^ob/ob mice, compared with CCK-injected GCG-Cre::SynTom::WT mice (Figure 5A–D; p<0.0001). To test whether leptin alters the activity of postsynaptic neurons in the PVH that receive GLP-1 inputs, we used whole-cell patch-clamp electrophysiology to record miniature excitatory postsynaptic currents (mEPSCs) from PVH neurons that express GLP-1 receptors (GLP-1 R), visualized by crossing Glp1r-Cre mice with the Cre-dependent fluorescent reporter tdTomato in leptin-deficient mice (GLP-1 R-Cre::tdTom::Lep^ob/ob mice). Leptin did not impact the kinetics of glutamate signaling, as there were no differences measured in the rise and decay times of averaged mEPSC events normalized to peak current in GLP-1 R-Cre::tdTom::WT controls and GLP-1 R-Cre::tdTom::Lep^ob/ob mice (Figure 6G). Consistent with the increased levels of Fos immunoreactivity detected in the PVH of leptin-deficient mice, we observed a significant and selective increase in the frequency of mEPSCs in GLP-1 R neurons in slices isolated from GLP-1 R-Cre::tdTom::Lep^ob/ob mice, compared with mEPSCs recorded from labeled neurons in GLP-1 R-Cre::tdTom::WT controls (Figure 6H). The increase in mEPSC frequency observed in GLP-1 R-Cre::tdTom::Lep^ob/ob mice appeared to be due to a 41% decrease in inter-event intervals in leptin-deficient mice (Figure 6I; p<0.0001). Because PPG neurons are glutamatergic, these results are consistent with enhanced excitatory neurotransmission in GLP-1 R neurons of leptin-deficient mice. There was no difference in mEPSC amplitude between GLP-1 R-Cre::tdTom::Lep^ob/ob mice and WT controls (Figure 6J), which indicates that leptin does not directly alter the sensitivity of postsynaptic GLP-1 R neurons in the PVH to glutamate.

Leptin receptors are expressed on nearly all PPG neurons, but they are also expressed in a number of regions that provide afferents to the NTS and PVH. Therefore, we used a combined loss of function/gain of function molecular genetic approach to examine the site of action for the developmental regulation of PPG inputs to the PVH by leptin. Immunohistochemistry was utilized to visualize GLP-1 projections in functionally leptin-receptor null LepRb^TB/TB mice, which effectively lack expression of LepRb throughout the CNS (LepRb^TB/TB mice) due to the insertion of a LoxP-flanked transcription-blocking cassette in the Lepr gene (Berglund et al., 2012). The density of GLP-1 immunolabeled inputs to the PVH of LepRb^TB/TB mice was approximately 47% higher than in WT mice (Figure 6A,B,D; p=0.0022). Furthermore, the increase in the density of immunolabeled GLP-1 inputs observed in LepRb^TB/TB mice was almost identical to that observed in leptin-deficient mice when compared with WT mice. To restore LepRb signaling specifically in PPG neurons, LepRb^TB/TB mice were crossed with GCG-Cre mice (resulting in GCG-Cre::LepRb^TB/TB mice) and the density of GLP-1 immunoreactive fibers was measured in the PVH. Targeted expression of LepRb in PPG neurons appeared to restore the density of GLP-1 fibers in both the PVHmpd and PVHpml to levels that were similar to those observed in WT controls (Figure 6C,D), suggesting that LepRb functions cell autonomously to reduce targeting of GLP-1 projections to the PVH. Leptin does not alter PPG neuron number in either GCG-Cre::LepRb^TB/TB or LepRb^TB/TB mice, a finding nearly identical to that of the density and distribution of PPG neuron number in leptin-deficient mice (WT: 19.67 ± 0.8; LepRb^TB/TB: 20.02 ± 0.2; GCG-Cre::LepRb^TB/TB: 19.78 ± 0.7 PPG neurons per section). To assess whether manipulation of LepRb on PPG neurons impacts viscerosensory transmission from the gut to the brain, vagal afferents were stimulated by i.p. injection of CCK and the number of Fos immunoreactive nuclei in the PVH of WT, LepRb^TB/TB, and GCG-Cre::LepRb^TB/TB mice was quantified. Injection of CCK in WT mice resulted in a significant increase in the number of Fos immunoreactive nuclei in the PVH compared with WT mice that received saline injections (Figure 6E,H; p=0.0156). Following CCK injection, a significant 65% increase in immunolabeled Fos nuclei was detected in global leptin-receptor null LepRb^TB/TB mice compared with WT mice (Figure 6E,F,H; p=0.0213), consistent with our observations in Lep^ob/ob mice, which suggests leptin specifies GLP-1 projections into the PVH through LepRb signaling. Further, restoration of LepRb signaling specifically in PPG neurons on an otherwise functionally LepRb null background normalized densities of Fos-labeled nuclei in the PVH of GCG-Cre::LepRb^TB/TB mice, compared to those observed in WT mice (Figure 6E,G,H; p=0.4031, ns). This finding suggests that the activity of GLP-1 neural circuitry is regulated by LepRb signaling in a cell-autonomous manner. Taken together with results obtained in leptin-deficient Lep^ob/ob mice (see Figure 5), our observations in leptin receptor-deficient LepRb^TB/TB mice suggest that leptin alters transmission of viscerosensory information primarily by acting directly on PPG neurons during development of their ascending projections.

In order to determine the role of LepRb on PPG neurons in regulating food intake and anxiety-like behaviors, we analyzed meal patterns, performance on the elevated zero maze test and novelty-suppressed feeding test, and measured blood serum corticosterone levels in GCG-Cre::LepRb^TB/TB mice with targeted expression of LepRb on PPG neurons, global LepRb-deficient (LepRb^TB/TB mice), and WT controls. Leptin-receptor null (LepRb^TB/TB mice) demonstrated increased body weight and food intake compared with WT mice (Figure 7A,B; p=<0.0001 and p=0.0004, respectively), consistent with previous reports (Berglund et al., 2012). However, targeted expression of LepRb on PPG neurons did not normalize body weight or food intake to wild-type levels, and these GCG-Cre::LepRb^TB/TB mice remained obese and ate significantly more food than WT mice (Figure 7A,B; p=0.0046). Furthermore, targeted expression of LepRb on PPG neurons did not normalize any meal pattern parameters analyzed to those characteristic of wild-type mice (Figure 7C,D). GCG-Cre::LepRb^TB/TB mice displayed no difference in meal number or meal size compared to LepRb^TB/TB mice, although both of these groups displayed a significant decrease in meal number and increase in meal size compared with WT mice (Figure 7C,D; p=0.0013 and p=0.0004, respectively). Because GLP-1 is involved in the regulation of stress and anxiety (Tauchi et al., 2008; Terrill et al., 2018), we also tested whether re-expression of LepRb on PPG neurons regulates anxiety-like behavior by performance on the elevated zero maze. As expected, LepRb^TB/TB mice displayed increased anxiety-like behaviors by increased percent of time spent in the closed arm (Figure 7F; p=0.0252), decreased number of transitions into the open zone (Figure 7G; p=0.0013) and significantly decreased total distance traveled compared with WT mice (Figure 7H; p=0.0498). However, targeted expression of LepRb on PPG neurons did not normalize any aspects of performance on the elevated zero maze, with GCG-Cre::LepRb^TB/TB mice displaying no differences compared to LepRb^TB/TB mice (Figure 7F–H). In addition, blood levels of serum corticosterone were elevated in LepRb^TB/TB mice compared with WT mice (p=0.0329), but corticosterone levels in GCG-Cre::LepRb^TB/TB mice did not differ from those of LepRb^TB/TB mice (Figure 7I). The novelty-suppressed feeding test was used to assess ingestive behaviors in an aversive environment and the latency to initiate a feeding bout and amount consumed was measured. Consistent with the elevated anxiety-like behavior displayed by LepRb^TB/TB mice in the elevated zero maze, these animals also showed significantly longer time to initiate feeding (p<0.0001) and decreased total distance traveled (p=0.0014) compared with WT mice, although total amount of food consumed was not different between the groups (Figure 7J–L). Restoration of LepRb on PPG neurons did not normalize performance on the novelty-suppressed feeding test, as GCG-Cre::LepRb^TB/TB mice did not differ in their responses compared to LepRb^TB/TB mice (Figure 7J–L).

Mice expressing Cre recombinase under control of the leptin receptor promoter (LepRb-Cre mice) were provided by Dr. Martin Myers, Jr., University of Michigan (Leshan et al., 2006). Transgenic BAC mice expressing Cre recombinase under control of the Gcg promoter (GCG-Cre mice) were generated at the University of Texas Southwestern and validated by Scott and colleagues (Gaykema et al., 2017). Mice expressing the Cre-dependent fluorescent reporters tdTomato (tdTom mice; Ai14D-Gt(Rosa)26Sor; stock number: 007914) and synaptophysin-tdTomato (SynTom mice; Ai34D-Rosa-CAG-LSL-Synaptophysin-tdTomato-WPRE; stock number: 012570) were obtained from The Jackson Laboratory. Knockin mice expressing an IRES-Cre fusion protein under control of the Glp1r promoter (GLP-1 R-Cre mice) were generated by Dr. Stephen Liberles, Harvard Medical School (Williams et al., 2016) and obtained from The Jackson Laboratory (stock number: 029283). Mice containing a Cre-dependent Lox-P flanked transcription blocker (loxTB) sequence between exons 16 and 17 of the leptin receptor gene (LepRb^TB/TB mice) were obtained from The Jackson Laboratory (stock number: 018989) and validated by Elmquist and colleagues (Berglund et al., 2012). Experimental mice were generated through heterozygous intercrosses to generate homozygous, leptin-deficient offspring (Lep^ob/ob mice). WT littermates with normal leptin expression were used as controls. To visualize neurons in the NTS that express leptin receptors and neurons in the PVH that express GLP-1 receptors, LepRb-Cre mice and GLP-1 R-Cre mice were crossed with tdTom mice to generate LepRb-Cre::tdTom mice and GLP-1 R-Cre::tdTom mice. To visualize PPG inputs to the PVH, GCG-Cre mice were crossed with SynTom mice to generate GCG-Cre::SynTom mice. These mouse lines were then bred onto the leptin-deficient background to generate GLP-1 R-Cre::tdTom::Lep^ob/ob mice and GCG-Cre::SynTom:Lep^ob/ob mice. WT controls were generated from the same litters.

All animal care and experimental procedures were performed in accordance with the guidelines of the National Institutes of Health and the Institutional Care and Use Committee of Vanderbilt University. Mice were housed at 22°C on a 12:12 hr light:dark cycle (lights on at 6:00 am:lights off at 6:00 pm). Mice were provided ad libitum access to a standard chow diet (PicoLab Rodent Diet 20 #5053). Mice were weaned at P22 and maintained with mixed genotype littermates until males were used for experiments.

LepRb-Cre mice were anesthetized with isoflurane (1.5–2.0%, 1 L/minute in O₂) and placed in a stereotaxic frame (Kopf Instruments, Tujunga, CA). The skull was exposed through a dorsal midline incision in the skin. The fourth ventricle and obex were identified and used as geographical landmarks to determine injection site in the NTS. The stereotaxic coordinates used were: A/P, −0.16; M/L, ±0.2; D/V, −0.2 from the obex. Viral injections were performed using glass micropipettes in a pressurized picospritzer system (General Valve Corporation, Fairfield, NJ; 40 pounds pressure per square inch; average duration 4msec). Then, 0.5 μL AAV pCAG-FLEX-EGFP-WPRE virus (Addgene, Cambridge, MA) was injected into the NTS. The tip of the glass micropipette was left in the brain for 5 min, and then slowly retracted. Mice were given analgesic and allowed to recover on a heating pad for 2 days post-surgery. They were then transcardially perfused as described above 14 days post-surgery.

Sections through the PVH with were examined on a laser scanning confocal microscope (Zeiss 710 or 800) and cytoarchitectonic features of the PVH, visualized with the cytoplasmic neuronal marker HuC/D (Biag et al., 2012) were used to define matching regions of interest (ROI) for quantitative analysis. Because the PVH contains functionally discrete subcompartments, we quantified the density of GLP-1 immunolabeled or GCG-Cre::SynTom-labeled fibers in the PVH in anatomically defined ROIs, as well as onto specific cell types. Thus, we quantified GLP-1 input density in the PVHmpd subcompartment, which is involved in the control of hormone secretion from the anterior pituitary, and in the PVHpml, because these cells innervate regions of the brainstem known to regulate autonomic neurons. Moreover, these subcompartments contain the highest density of labeled GLP-1 fibers, as reported previously in the rat (Tauchi et al., 2008). In addition to subnuclear quantification, we also measured the density of GLP-1 inputs onto specific cell types within in the PVH, as leptin has been shown to impact peptidergic innervation of neuroendocrine and sympathetic preautonomic neurons with both cell type and target-neuron specificity (Bouyer and Simerly, 2013). Because it was reported that PPG neurons primarily innervate CRH and Oxytocin neurons in the PVH (Tauchi et al., 2008; Katsurada et al., 2014; Kanoski et al., 2016), we measured inputs to CRH and Oxytocin neurons the PVH at higher magnification. Confocal image stacks were collected for each ROI through the entire thickness of the PVH at a frequency of 0.4 μm using the 40x objective and a frequency of 1.19 μm using a 20x objective. Velocity visualization software (PerkinElmer, Waltham, MA) was used to prepare 3D reconstructions of each multichannel set of images. To quantify the overall densities of labeled fibers in each ROI of the PVH, a previously validated methodology was utilized (Bouyer and Simerly, 2013). First, images were binarized, skeletonized, and the total fiber length was summed for each ROI throughout the image stack to obtain an estimate of the total density of GLP-1 fibers within that ROI.

The density of labeled inputs from PPG neurons onto identified CRH and Oxytocin neurons in the PVH was measured in GCG-Cre::SynTom::Lep^ob/ob mice and GCG-Cre::SynTom::WT controls by using Imaris image analysis software (Bitplane V9.2, Salisbury Cove, Maine). Sections containing immunolabeled oxytocin, CRH, and genetically labeled GCG-Cre::SynTom fibers were segmented in each image stack by establishing an intensity threshold. Three-dimensional (3D) reconstructions of image volumes were then rendered for each multi-channel image stack using Imaris. Surface renderings of CRH or Oxytocin neurons were created and GCG-Cre::SynTom-labeled fibers and terminals were defined by using the Spots function. Spots identified as being in close apposition (voxel to voxel apposition) to the neuronal surface renderings were quantified. Voxels containing GCG-Cre::SynTom-labeled neuronal processes that did not make contact with the surface renderings of CRH or oxytocin neurons were excluded from analyses. Numbers of GCG-Cre::SynTom-labeled spots (terminals) were summed and divided by the total number of CRH or oxytocin neurons contained in each image stack to estimate the density of GCG-Cre::SynTom inputs onto each population of PVH neurons. The terms ‘inputs’ or ‘terminals’ used in this study refer to the visualization of synaptophysin-tdTomato (SynTom) labeled axonal structures in close apposition to neuronal cell bodies that were labeled histochemically with CRH, Oxytocin, or the pan-neuronal cell marker HuC/D antibodies. The terms ‘fibers’, ‘axons,’ or ‘projections’ include, but are not limited to SynTom-labeled axon structures, but also GLP-1 immunohistochemical labeling, which can be detected in axon terminals as well as fibers en passant. Quantification of these axonal elements were only considered to be in close apposition onto a target neuron in the PVH if they were touching by voxel-to-voxel contact, with no unlabeled voxels between the presynaptic and postsynaptic elements of the presumptive synapse. Super-resolution optical microscopy, or ultrastructural visualization, will be necessary to directly measure synaptic contacts onto postsynaptic cells, but the density of genetically targeted and immunohistochemically labeled inputs to CRH and Oxytocin neurons appears to correlate with previous estimates of the density of synapses onto identified cell types within the PVH (Bouyer and Simerly, 2013).

To measure the density of Fos labeling in the NTS and PVH that resulted from i.p. injection of CCK, the number of Fos immunoreactive nuclei was counted in maximum projection confocal images through each region aided by Volocity software. Images of sections through the caudal NTS containing labeled GCG-Cre::SynTom neurons were matched using cytoarchitectonic features of the caudal brainstem such as the area postrema, large motor neurons of the DMX, and the central canal and used for quantification. The number of GCG-Cre::SynTom-expressing neurons in each section was counted manually and the number of Fos immunoreactive nuclei was counted using the object counting function in Volocity. The number of Fos immunoreactive nuclei that were colocalized to GCG-Cre::SynTom-expressing neurons in the NTS was expressed as a percentage of the total number of GCG-Cre::SynTom-expressing neurons counted within the caudal NTS. Numbers of Fos immunoreactive nuclei were also counted in the same ROIs through the PVH as described previously for GLP-1 fiber density measurements. Images containing the PVHmpd and PVHpml regions were identified and Volocity software was used to count the total number of Fos positive nuclei in the ROIs in the PVHpml and PVHmpd regions that receive dense GLP-1 inputs. Similarly, the number of pSTAT3 immunoreactive nuclei in LepRb- (LepRb-Cre::tdTom) and in PPG (GCG-Cre::SynTom) neurons was quantified in sections through the NTS by using object counting features of Volocity software and expressed as a percentage.

In order to measure spontaneous synaptic currents, acute brain slices were prepared from GLP-1 R-Cre::tdTom::Lep^ob/ob and GLP-1 R-Cre::tdTom::WT mice. Mice were anesthetized with isoflurane (5%) and perfused transcardially with ice-cold Choline-Cl slicing solution containing 105 mM Choline-Cl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 7 mM MgCl₂, 25 mM NaHCO₃, 25 mM Glucose, 11 mM Na-ascorbate, 3 mM Na-pyruvate, and 0.5 mM CaCl₂. Mice were decapitated, the brains rapidly removed from the skull and 300 μm-thick coronal sections cut into cold slicing solution on a Leica VT1200S vibratome. Slices were stored in oxygenated artificial cerebrospinal fluid (aCSF) and maintained at 32°C throughout the experiment. Whole-cell voltage clamp recordings (at a holding potential of −60 mV) were obtained from the cell bodies of fluorescently identified GLP-1 R-Cre::tdTom neurons. Miniature excitatory postsynaptic currents (mEPSCs) were recorded using a Multi-Clamp 700B amplifier. Signals were digitized through a Digidata 1550B, interfaced via pCLAMP 10 software (Molecular Devices, San Jose, CA). All recordings were performed at 32°C, maintained with a TC-344B temperature controller connected to an inline heater (64–0102) and heated PH-1 platform (Warner Instruments, Hamden, CT). Patch pipettes were pulled from borosilicate glass capillaries with resistances ranging from 3 to 5 MΩ when filled with pipette solution. The bath solution contained 124 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 1.25 mM NaH₂PO₄, 25 mM glucose, 25 mM NaHCO₃, 1 mM Na-ascorbate and 1 mM Na-pyruvate. Postsynaptic AMPA-receptor mediated currents in response to spontaneous release of glutamate-containing vesicles from presynaptic terminals were recorded in the presence of the sodium channel blocker TTX (1 µM). The pipette solution contained 120 mM Cesium Methanesulfonate (CsMeSO₄), 5 mM CsCl, 3 mM Na-ascorbate, 4 mM MgCl₂, 10 mM HEPES, 2 mM ethylene glycol-bis-(aminoethyl ethane)-N,N,N’,N’-tetraacetic acid (EGTA), 4 mM Na-ATP, 0.4 mM Na-GTP, 10 mM phosphocreatine, 0.5 mM CaCl₂, 5 mM glucose, 5 mM QX-314 pH 7.25 (CsOH). All signals were digitized at 20 kHz, filtered at 2 kHz, and analyzed offline with Clampfit software (Molecular Devices, San Jose, CA). Images of patched cells were captured on a Zeiss Axioskop 2 FS Plus microscope equipped with a Ximea XiQ USB3 Vision camera with a 40x objective.

Group data are presented as mean values ± SEM. Statistical significance was determined using GraphPad Prism software. Student’s t-test was used to compare data within two groups, using paired and unpaired tests where appropriate. One-way ANOVA followed by a pairwise post-hoc test was used to test for comparisons between three groups. p-values<0.05 were considered statistically significant.