Blockade of endogenous insulin receptor signaling in the nucleus tractus solitarius potentiates exercise pressor reflex function in healthy male rats

Insulin not only regulates glucose and/or lipid metabolism but also modulates brain neural activity. The nucleus tractus solitarius (NTS) is a key central integration site for sensory input from working skeletal muscle and arterial baroreceptors during exercise. Stimulation of the skeletal muscle exercise pressor reflex (EPR), the responses of which are buffered by the arterial baroreflex, leads to compensatory increases in arterial pressure to supply blood to working muscle. Evidence suggests that insulin signaling decreases neuronal excitability in the brain, thus antagonizing insulin receptors (IRs) may increase neuronal excitability. However, the impact of brain insulin signaling on the EPR remains fully undetermined. We hypothesized that antagonism of NTS IRs increases EPR function in normal healthy rodents. In decerebrate rats, stimulation of the EPR via electrically induced muscle contractions increased peak mean arterial pressure (MAP) responses 30 min following NTS microinjections of an IR antagonist (GSK1838705, 100 μM; Pre: Δ16 ± 10 mmHg vs. 30 min: Δ23 ± 13 mmHg, n = 11, p = .004), a finding absent in sino‐aortic baroreceptor denervated rats. Intrathecal injections of GSK1838705 did not influence peak MAP responses to mechano‐ or chemoreflex stimulation of the hindlimb muscle. Immunofluorescence triple overlap analysis following repetitive EPR stimulation increased c‐Fos overlap with EPR‐sensitive nuclei and IR‐positive cells relative to sham operation (p < .001). The results suggest that IR blockade in the NTS potentiates the MAP response to EPR stimulation. In addition, insulin signaling in the NTS may buffer EPR stimulated increases in blood pressure via baroreflex‐mediated mechanisms during exercise.

(EPR), the responses of which are buffered by the arterial baroreflex, leads to compensatory increases in arterial pressure to supply blood to working muscle. Evidence suggests that insulin signaling decreases neuronal excitability in the brain, thus antagonizing insulin receptors (IRs) may increase neuronal excitability. However, the impact of brain insulin signaling on the EPR remains fully undetermined. We hypothesized that antagonism of NTS IRs increases EPR function in normal healthy rodents. In decerebrate rats, stimulation of the EPR via electrically induced muscle contractions increased peak mean arterial pressure (MAP) responses 30 min following NTS microinjections of an IR antagonist (GSK1838705, 100 μM; Pre: Δ16 ± 10 mmHg vs. 30 min: Δ23 ± 13 mmHg, n = 11, p = .004), a finding absent in sino-aortic baroreceptor denervated rats. Intrathecal injections of GSK1838705 did not influence peak MAP responses to mechanoor chemoreflex stimulation of the hindlimb muscle. Immunofluorescence triple overlap analysis following repetitive EPR stimulation increased c-Fos overlap with EPR-sensitive nuclei and IR-positive cells relative to sham operation (p < .001). The results suggest that IR blockade in the NTS potentiates the MAP response to EPR stimulation. In addition, insulin signaling in the NTS may buffer

| INTRODUCTION
The role of insulin signaling in the regulation of glucose homeostasis is an extensive area of investigation. 1 However, the actions of insulin in the central nervous system (CNS) are an emerging area of research, given that brain deficiencies in insulin receptor (IR)-mediated signaling may result in aberrant cellular function and brain disorders. 2 Evidence suggests that the actions of brain insulin also extend to alterations in autonomic reflex function. For example, Young et al previously demonstrated that insulin enhances the gain of baroreflex control of muscle sympathetic nerve activity. 3 Additionally, direct injection of insulin into the lateral ventricle of the rat brain also enhances the gain of baroreflex control over heart rate and sympathetic nerve activity. 4 Furthermore, decreased insulin transport capacity into the brain leads to alterations in baroreflex function in rabbits and rats. 5,6 Thus, it is likely that brain insulin significantly modulates neural control of the cardiovascular system.
As the baroreflex interacts with the exercise pressor reflex (EPR) during exercise to regulate arterial pressure and meet oxygen demands of working muscle, [7][8][9] insulin signaling may influence this interaction. During exercise, activation of skeletal muscle sensory fibers evokes the EPR. 10 The EPR is stimulated by group III and group IV primary sensory fibers in working muscle. 11,12 These fibers comprise the mechano-and metabo-reflex components of the EPR, 10 respectively, which are stimulated by mechanical forces and interstitial chemical factors. These fibers can be sensitized by circulating factors, such as insulin, resulting in augmented reflex pressor responses. 13,14 Additionally, experimental development of diabetic rats results in augmented cardiovascular responses to EPR stimulation by hindlimb muscle contractions. 15 This finding can also be in part attributable to sensitization of peripheral sensory fibers. 16 Therefore, sensitization of muscle sensory nerves plays a role in regulating the magnitude of the EPR. To date, however, the role of central insulin signaling in regulation of the EPR has not been fully elucidated but may involve direct actions on hindbrain neuronal excitability. 17 The medullary control centers of the cardiovascular system include the nucleus tractus solitarius (NTS), which is a key integration site for incoming sensory input from working skeletal muscle and arterial baroreceptors. 18 For example, neurons in the NTS are activated by stimulation of the EPR, 19,20 and the baroreflex. 21 Insulin activates potassium channels in hypothalamic neurons in rats, resulting in hyperpolarization. 22 Furthermore, proopiomelanocortin neurons are highly expressed in the NTS cardiovascular control areas 23,24 and are known to be hyperpolarized by enhanced phosphoinositide 3-kinase activity, which is a known target for activation by insulin. 25 Thus, it is logically suggested that pharmacological blockade of central IRs in the NTS should decrease hyperpolarization, resulting in increased excitability of neurons and augmented EPR-stimulated cardiovascular responses in healthy rats. To test the hypothesis, we investigated the impact of acute NTS microinjection of the IR antagonist GSK1838705 on cardiovascular and sympathetic responses to EPR stimulation in normal healthy decerebrated rodents. These studies were done with peripheral baroreceptors intact or surgically ablated to elucidate whether EPR-sensitive and/ or baro-sensitive NTS neurons are influenced by IR blockade. Furthermore, we examined if intrathecal injection of the IR antagonist influences the pressor responses to mechanical or chemical stimulation of sensory afferent fibers. Finally, we determined if the NTS is populated with IRpositive neurons that respond to stimulation by the EPR.

| MATERIALS AND METHODS
All animal studies described were performed in accordance with the US Department of Health and Human Services NIH Guide for the Care and Use of Laboratory Animals. All experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Texas Southwestern Medical Center (no.2019-102849). injection, and immunofluorescence experiments. The animals were housed in temperature (22-24°C) and humidity (40%-60%) controlled chambers (Allentown, Allentown, PA, USA), one to four per cage, under 12 h light/dark cycle with free access to food and water. Only male rats were used as estrous cycles influence both the EPR 26 and baroreflex sensitivity. 27 All rats used in this study were all rendered insentient by the decerebration procedure described below, and subject to microinjections of test solutions into the hindbrain or intrathecal space as outlined. Surgical procedures for all experiments were performed as described previously. 28,29 Briefly, throughout surgical procedures and experimental testing, animal body temperature was kept within a constant range (36.5-37.5°C) with a far-infrared heating pad system (RightTemp; Kent Scientific, Torrington, CT, USA). Prior to surgery, induction of anesthesia was achieved with isoflurane (4%, balanced with oxygen). Adequacy of anesthesia was verified by lack of withdrawal responses to tail and hindlimb pinch. A tracheotomy was then performed, and anesthesia was maintained (2.5% isoflurane) via mechanical ventilation. Both carotid arteries were ligated and stiff polyurethane tubing (RPT-40, Braintree Scientific Inc., Braintree, MA, USA) was retrogradely inserted through the right carotid artery up to the aorta to measure blood pressure. The right jugular vein was also retrogradely catheterized to administer fluids. A constant i.v. infusion of fluids (3-5 mL/h/kg of 2 mL 1 M NaCHO3, and 10 mL 5% dextrose in 38 mL Ringer solution) was used to maintain hydration, plasma volume, and arterial blood pressure.

NTS microinjection studies
The ventral roots were electrically stimulated (2-3 × motor threshold [MT], 0.1 ms pulse duration, 40 Hz, 30s: S88; Grass Instruments) to contract the hindlimb muscles, and some muscles of the upper leg, and stimulate the EPR. First, an L 2 -L 6 laminectomy was performed, and the dura was then cut and reflected. The ventral roots, L 4 and L 5 , of the left hindlimb were then isolated, cut, and laid over a pair of bipolar stainless-steel electrodes to contract the muscles with constant current stimulation (PSIU 6; Grass Instruments, West Warwick, RI, USA). The nerves were protected from drying with a pool of warm mineral oil (37°C). The calcaneal bone of the left hindlimb was then cut and the Achilles tendon was connected via a braided fishing line to a force transducer (FT10; Grass Instruments) clamped onto a 9.5 mm rack and pinion system (Harvard Apparatus, Holliston, MA, USA). The left hindlimb was fixed in place using clamps at the hip and ankle joint. Before muscle contractions the baseline tension was set between 50 and 100 g.
To record changes in sympathetic nerve activity evoked by activation of the EPR, a recording electrode was placed under the renal nerve. The left kidney was first exposed using a retroperitoneal approach to isolate a renal nerve bundle adjacent to the renal artery and vein. A pair of platinum recording electrodes (AM Systems, Sequim, WA, USA) was then placed underneath the nerve and fixed/insulated, with silicone adhesive mold (Kwik-Cast, World Precision Instruments, Sarasota, FL, USA). The recording electrode was connected to a preamplifier (AD Instruments, Colorado Springs, CO, USA) and renal sympathetic nerve activity (RNSA) was quantified by running the signal through a band-pass filter set at 150-1000 Hz (Neuro Amp EX; AD Instruments, USA), then full-wave rectified, and low-pass filtered with a cut-off frequency of 30 Hz.
For all experimental testing rats were placed within a stereotaxic apparatus (David Kopf Instruments, Tujunga, CA, USA) and a precollicular mechanical decerebration procedure was performed as previously described. 28 Intravenous dexamethasone (0.2 mg) was first given to minimize brain edema during the decerebration procedure. The blood pressure was allowed to first stabilize at 1% isoflurane anesthesia before a craniotomy was performed above the sagittal sinus, and a section >1 mm rostral to the superior colliculus was made. Tissues rostral to the section were aspirated and bleeding was minimized by packing the cranial cavity with oxidized regenerated cellulose powder (Surgicel; Ethicon, Raritan, NJ, USA) and cotton around the brainstem. For microinjection experiments the head was then tilted down approximately 15° and fixed, before discontinuing isoflurane anesthesia. A period of at least 60 min was allowed to elapse before beginning experiments, to eliminate the depressive effects of residual isoflurane gas on the EPR. 28 We used ventral root stimulation to evoke muscle contractions and stimulate the EPR in rats receiving bilateral NTS microinjections. The microinjection preparation and procedures were modeled as described previously. 30 Before decerebration, a small amount of bone was carefully removed to aid in visualizing the hindbrain, including the bottom portion of the cerebellum, and exposing the region for microinjections. NTS microinjections were performed with pressurized N 2 gas using a pneumatic Pico pump (PV830; WPI, Sarasota, FL, USA) connected to a glass micropipette (1.0 O.D. × 0.5 I.D.; FHC, Bowden, ME, USA). Borosilicate micropipettes were made with a micropipette puller (P-30, Sutter Instruments, Novato, CA, USA) and the tips were broken back to an O.D. of 10-20 microns. With the aid of a microscope, a series of six injections (50 nL each) were then performed with either the IR antagonist GSK1838705 (Sigma-Aldrich, St. Louis, MO, USA) or the vehicle dimethylsulfoxide (DMSO, Sigma-Aldrich). The injection coordinates were 0.5 mm lateral to each side of the calamus scriptorius, and 0.5 mm rostral and caudal to both those points. Any adjustments to these coordinates were minimized in order not to puncture any superficial blood vessels. A horizontal microscope head equipped with a graduated reticule in one eyepiece was used to track the meniscus and determine the displaced volume during microinjections. To facilitate diffusion of the injectate in the target area a minimum of 2 min was allowed to elapse following each microinjection. In a pilot study the diffusion of pontamine sky blue (1%) dissolved in DMSO was visually assessed to confirm diffusion of the dye in the local region following NTS microinjection procedures.
In a second subset of rats not used for contraction and EPR experiments laminectomy was not performed. In these rats, to activate the skeletal muscle mechanoreflex in the right hindlimb the triceps surae was manually stretched by the calcaneal tendon via the rack and pinion system to a peak tension between 750 and 1200 g for 30s of passive stretch. The choice of right or left hindlimb stimulation was based on surgical skill and speed. To activate the skeletal muscle chemoreflex, a catheter (PE-10) was inserted retrogradely into the left femoral artery until the tip reached the bifurcation point of the abdominal aorta. A silk suture was then used to secure the catheter, and a pneumatic vascular occluder (FST Inc., Foster City, CA, USA) was placed around the right iliac vein to trap the injectate (100 μL, 0.3 μg capsaicin) in the hindlimb circulation. The occluder was deflated after evaluation of the pressor response to capsaicin. For tibial nerve stimulation experiments, the tibial nerve of the right hindlimb was surgically isolated by blunt dissection, and a pair of stainless-steel electrodes was used to stimulate the nerve (10 × MT, 0.75 ms, 20 Hz, for 30 s), eliciting a pressor response. These three maneuvers, at least two for each rat, were done on rats paralyzed with pancuronium bromide (0.2 μg, i.v.) (Sigma-Aldrich) before beginning experimental procedures. Pancuronium bromide administration abolishes sensory input from the EPR completely and the drug has a vagolytic effect which may result in elevated pressor responses to stimulation, relative to the EPR, that can exceed over 100 mmHg. 15 Supplemental doses of pancuronium (0.1 μg) were administered when necessary.
A third subset of rats used for EPR experiments was baro-denervated before beginning EPR experimental procedures. Bilateral baro-denervation was performed using a previously established method known to ensure complete denervation. 31 Briefly, sino-aortic baroreceptor denervation was achieved by first exposing the carotid bifurcation by blunt dissection. Then, the aortic depressor nerve was cut, and the superior laryngeal nerve was sectioned near the junction with the vagus nerve. The superior cervical ganglion was then excised, and the carotid sinus region was stripped of carotid baroreceptor nerves. All cut nerve endings were then painted with a solution composed of 10% phenol and 70% ethanol to ablate any remaining nerve fiber activity. 32 In pilot studies, acute sino-aortic baro-denervation was confirmed in rats under isoflurane anesthesia by evaluating HR responses to i.v. injections of phenylephrine (2 mg/kg) before and immediately after surgical/chemical denervation procedures. Although this dose of phenylephrine allows for determining if denervation procedures are complete, we found this dose to produce deleterious depression of baseline cardiovascular function and therefore these animals were not used for EPR experimentation.

| Experimental protocols for NTS microinjection experiments
To assess the effects of acute NTS IR blockade on the EPR in both baro-intact and baro-denervated rats the left hindlimb was contracted before, then 30 min following, NTS microinjections of the antagonist GSK1838705 (100 μM) (baro-intact, n = 11; baro-denervated n = 8), or the vehicle control (baro-intact, n = 9; baro-denervated n = 9). The mechano-and chemoreflex can be activated independent of skeletal muscle contraction using passive stretch of the triceps surae or i.a. capsaicin injections, respectively. Additionally, the afferent fibers innervating the skeletal muscles can be activated by direct electrical stimulation using the tibial nerve stimulation maneuver. This maneuver bypasses skeletal muscle contractions and directly activates both mechano-and metabo-sensitive afferent fibers, and furthermore precludes muscular fatigue, allowing for full activation of the sensory afferent nerve fibers. For NTS microinjection experiments with mechanoreflex, chemoreflex, and tibial nerve stimulation, GSK1838705 (10 μM) (n = 16) or vehicle (n = 13) were given as described for EPR stimulation experiments. The MT was determined before skeletal muscle paralysis. At least two of the three maneuvers were performed on each of these rats.

| Intrathecal injection studies
In a fourth set of rats, before decerebration, a limited laminectomy, L 2 -L 3 , was performed and the dura was cut at the L 2 vertebrae level. An intrathecal catheter (SUBL-140, Braintree Scientific Inc) connected to a Pinport (InsTech, Plymouth Meeting, PA, USA) was then inserted into the spinal cord cavity down to the level of the L 5 vertebrae. Petroleum jelly was then used to cover the exposed spinal nerves and intrathecal catheter. The injections (40 μL, 10 min) were performed with a digitally controlled pump (Micro4/UMP3 controller and pump, WPI, USA). The stereotaxic frame was tilted and held at 15° before beginning experiments to minimize the injectate from migrating to the medulla.
As in the NTS microinjection experiments, to stimulate the mechanoreflex the left gastrocnemius muscles were passively stretched to a tension between 750 and 1200 g for 30 s. Likewise, the chemoreflex was stimulated by i.a. capsaicin (100 μL, 0.1 μg) (Sigma-Aldrich), injected through a retrograde superficial epigastric artery catheter inserted down to the bifurcation with the right femoral artery. As described above, a vascular occluder placed around the right iliac vein was used to trap the injectate in the hindlimb. A lower dose of capsaicin, and injections into right hindlimb, was administered in these experiments to attenuate any potential desensitization, or impairment of afferent responsiveness due to repeated administration of the chemical over a prolonged experimental period. The total dose of capsaicin (0.6ug) was equal to that used in the study above, which was administered in two doses and diluted over a larger tissue area and within an equal interval. Prior to beginning experimentation rats were paralyzed before beginning mechano-and chemoreflex experiments, as described above, thus precluding EPR experiments.

| Experimental protocols for intrathecal injection experiments
Intrathecal injections of the antagonist GSK1838705 (100 μM) (n = 8), or the vehicle (n = 7) (0.1% DMSO diluted in artificial cerebrospinal fluid [aCSF, Harvard Apparatus]), were administered to determine if IR antagonism in the spinal cord influences the pressor responses to activation of metabo-and mechanoreflex. Pressor responses to each stimulus were first determined before i.t. injections, 5 min after injections, and then at every 30 min interval for a period of 2 hrs total post-injection.

| End experiment procedures and solutions
For animals used in all the physiological experiments described above (contraction, stretch, and i.a. capsaicin injections) hexamethonium bromide (60 mg/kg) (Sigma-Aldrich) was administered intravenously to verify postganglionic nerve activity when a signal was successfully obtained. Background noise was removed from the RSNA data by subtracting the measured signal up to 30 min following a bolus infusion of potassium chloride (4 M, 2 mL/kg, i.v.). Additionally, for animals given intrathecal injections, coomasie blue was used to verify that the experimental injectate was infused into spinal fluids postmortem and did not reach the level of the medulla. In all animals, heart, lung, and body mass were measured as well as tibial bone length.
In all studies DMSO was used to first solubilize GSK1838705 as a stock solution (100 mM), and as the vehicle control. The stock solution was diluted with aCSF (0.1% DMSO) to final working concentration. Hexamethonium and pancuronium bromide were diluted in PBS solution. Capsaicin (Sigma-Aldrich) was dissolved in 50% ethanol and then diluted with 0.9% saline to working concentration.

| NTS c-Fos activation studies
In a final set of rats the EPR was repeatedly stimulated (surgical procedures described above) to induce c-Fos activation in the NTS of decerebrated rats, as previously described. 20

| Experimental protocols for NTS c-Fos activation experiments
In the repetitive EPR stimulation group (n = 7), the L 4 and L 5 ventral roots were electrically stimulated for 30 s, followed by 30 s of rest, then repeated for a 60 min period, followed by 60 min of rest. In contrast, a sham surgery group (n = 5) received no EPR stimulus throughout the same period. Subsequently, the rats were given Euthasol (390 mg/mL/kg i.p.) anesthesia before beginning transcardial perfusions with saline (200 mL) and 4% PFA solution (400 mL), then harvesting the brain. The cerebellum and forebrain were removed from the hindbrain (pons and medulla), and the hindbrain was submerged in freshly prepared 4% PFA solution (Sigma-Aldrich) overnight until further processing. Onset and adequacy of anesthesia was verified by the depressive effects of Euthasol on heart rate and blood pressure and absence of cardiovascular responses to tail pinch.

| Immunofluorescence staining and triple labelling studies
Hindbrain sections were examined by immunofluorescence to determine the degree of IR, c-Fos, and DAPI overlap. When c-Fos is activated, it translocates into the nucleus, and DAPI counterstaining reduces false-positive signals. Tissues previously submerged overnight in 4% PFA solution were dehydrated in 10% sucrose for 24 hr, followed by 20% sucrose for 24 hr. The hindbrain was then submerged in OCT medium and frozen on dry ice before cutting slices (35 mm) on a cryostat (2800 Frigocut; Reichert-Jung/Leica, Deer Park, IL, USA). Slices 1 mm below through 1 mm above the calamus scriptorius were then collected. The sections were then washed in PBS (4 × 10 min) before beginning immunostaining procedures. First, sections were blocked and permeabilized (1 × 10 min) in antibody buffer (PBS, 0.5% Triton-X100, 5% NGS, 10 mg/mL BSA), then probed overnight with primary antibodies raised against c-Fos and IR (recombinant rabbit anti-c-Fos monoclonal antibody, 1:1000, cat. no. 2250, RRID#AB_2247211, Cell Signaling Technology, Danvers, MA, USA; mouse anti-IR β monoclonal antibody, 1:100, cat. no. sc-57 342, RRID#AB_784102, Santa Cruz Biotechnology, Dallas, TX, USA) diluted in antibody buffer. We previously used this IR antibody, from the same lot, to characterize IR binding in high power magnification rat DRG neurons. 14 Sections were then washed (4 × 10 min) with antibody buffer solution before secondary antibody incubations (1 × 60 min). Fluorescence-conjugated secondary antibodies (goat antirabbit polyclonal Alexa Fluor Plus 488 conjugate, 1:500, cat. no. A32731, RRID#AB_2633280, Thermo Fisher Scientific, Waltham, MA, USA; goat anti-mouse polyclonal Cy3 conjugate, 1:500, cat. no. 115-165-146, RRID#AB_2338690, Jackson ImmunoResearch Laboratories, West Grove, PA, USA) were diluted with antibody buffer solution. Subsequently, sections were washed (4 × 10 min) with antibody buffer solution before being fixed onto charged slides, rinsed with water (1 × 5 min), applying mounting medium with DAPI (ProLong Gold anti-fade with DAPI, cat. no. P36931, Thermo Fisher Scientific), and coverslipping with micro cover glass and nail varnish. Negative control experiments were done to verify that secondary antibodies did not react non-specifically with primary antibodies or with tissue. Slide-mounted sections were then used for immunofluorescence imaging. Normal goat serum and bovine serum albumin were purchased from Jackson ImmunoResearch Laboratories, and Triton-X100 from Sigma-Aldrich.
For image analysis of the NTS a standard multi-channel epifluorescence microscope (Zeiss, Axio Imager A2, Oberkochen, Germany) was used, as similarly described before. 20 Briefly, a series of grayscale photomicrographs at 10× magnification were captured for coordinates −15.4 to −13.5 mm from bregma, for each fluorophore. The images were then processed step-by-step for triple overlap analysis using image analysis software (FIJI; National Institute of Health, Maryland, USA). First, the region compromising the NTS was cropped out, and the image background was then subtracted by the rolling ball method, before then applying individualized thresholding methods for each image/coordinate and corresponding channels (threshold method-Triangle, c-Fos; threshold method-Otsu, IR and DAPI). After thresholding, a binary image was created, and a minimum particle size filter (5-71 pixel 2 , 0.6-1.0 circularity) was applied for each channel to remove the remaining background signal left over after thresholding. A region of interest of the remaining signal was then saved for each channel. For each image, the area overlap of each fluorophore with the other corresponding fluorophores was then calculated by the software. Grayscale photomicrographs were pseudo colored for illustration (see Figure 6 for example).

| Data handling and statistical analyses
Contractions and passive stretch maneuvers were performed for a period of 30 s following a period of 30 s baseline preload tension (50-100 g). Likewise, the stimulation period for tibial nerve stimulation was 30s. In contrast, for stimulation with capsaicin the peak cardiovascular responses typically occur within the first 5 s of stimulation. Cardiovascular and sympathetic responses were measured during each stimulation maneuver. Data for arterial blood pressure, heart rate, RSNA, and tension were obtained with LabChart data acquisition software (AD Instruments) and the Powerlab analog-to-digital converter (Powerlab8/30; AD Instruments) at a 1 kHz sampling rate. For cardiovascular, sympathetic and tension measurements, the baseline values were subtracted from the values measured during stimulation. One second averages were measured, and the baseline sampling period was considered as 100% baseline RSNA, with stimulation induced changes in RSNA expressed as a percentage of this baseline.
Physiological data were analyzed by two-way ANOVA repeated measure (RM) design and included main effects for (1) injection as a between-subject factor (DMSO vs. GSK1838705), (2) time as a within-subject factor (Pre, 30 min), and (3) an injection × time interaction. Then, if significance was detected the analysis was followed by Holms-Sidak post-hoc analysis. Histological data were analyzed by two-way ANOVA design and included main effects for (1) groups as a between-subject factor (Sham, EPR), (2) coordinate as a within-subject factor (−15.4 to −13.5 mm from bregma), and (3) a group × coordinate interaction. A p-value of <0.05 was defined as statistically significant. All data analysis was done using GraphPad Prism9 software and represented as means ± SD. Table 1 summarizes morphometric characteristics and baseline hemodynamics before (1% isoflurane) and after decerebration in each experimental procedure. Resting blood pressures following decerebration were within physiological range in all animals, thus all data were included in subsequent analysis. Post-hoc analysis following two-way ANOVA analysis did not demonstrate significant differences in resting blood pressure between groups before or after decerebration. Post-hoc analysis following two-way ANOVA did not demonstrate significant differences in resting heart rates before decerebration between groups. However, resting heart rates after decerebration were different between select groups (not shown) depending on surgical interventions such as laminectomy and baro-denervation. Figure 1 illustrates the influence of vehicle control or GSK1838705 NTS microinjections on the pressor response to EPR stimulation in a normal, representative baro-intact rat. Control microinjections did not noticeably influence the peak pressor response after 30 min ( Figure 1A,B). In contrast, GSK1838705 microinjections augmented the peak pressor response ( Figure 1C,D). Further analysis of group data demonstrated that GSK1838705 NTS microinjections significantly potentiated the peak pressor response to activation of the EPR (Pre: Δ15 ± 8 mmHg, 30 min: Δ15 ± 8 mmHg in n = 9 control vs. Pre: Δ16 ± 10 mmHg, 30 min: Δ23 ± 13 mmHg in n = 11 GSK1838705, p = .026 for ANOVA injection × time interaction, Figure 2A) However, the integrated pressor response during EPR stimulation was not statistically significant (Pre: 148 ± 69 mmHg × s, 30 min: 154 ± 138 mmHg × s in control vs. Pre: 174 ± 137 mmHg × s, 30 min: 272 ± 190 mmHg × s in GSK1838705, p = .249 for ANOVA injection × time interaction, Figure 2B). Furthermore, the peak (p = .354 for ANOVA injection × time interaction, Figure 2C) and integrated HR response (p = .652 for ANOVA injection × time interaction, Figure 2D) did not differ. Likewise, the peak RSNA (p = .249 for ANOVA injection × time interaction, Figure 2E) and integrated RSNA response (p = .230 for ANOVA injection × time interaction, Figure 2F) did not change. Peak tension (p = .800 for ANOVA injection × time interaction, Figure 2G) and integrated tension (p = .395 for ANOVA injection × time interaction, Figure 2H) during EPR stimulation did not differ 30 min following microinjections in either control or GSK1838705 groups.

| Influence of NTS insulin receptor antagonism on the EPR in baro-intact and baro-denervated rats
Sino-aortic baro-denervation was performed in a subset of rats ( Figure 3). In this case, the peak pressor response to EPR stimulation following NTS microinjections in either control or GSK1838705 groups did not differ (Pre: Δ34 ± 17 mmHg, 30 min: Δ33 ± 27 mmHg in control vs. Pre: Δ27 ± 16 mmHg, 30 min: Δ28 ± 21 mmHg in GSK1838705, p = .901 for ANOVA injection × time interaction, Figure 3A). The integrated pressor response to EPR stimulation was not different in either group (Pre: 346 ± 459 mmHg × s, 30 min: 307 ± 296 mmHg × s in control vs. Pre: 271 ± 134 mmHg × s, 30 min: 436 ± 380 mmHg × s in GSK1838705, p = .307 for ANOVA injection × time interaction, Figure 3B). Furthermore, the peak (p = .989 for ANOVA injection × time interaction, Figure 3C) and integrated HR response (p = .491 for ANOVA injection × time interaction, Figure 3D) was also not changed. Likewise, the peak RSNA (p = .861 for ANOVA injection × time interaction, Figure 3E) and integrated RSNA (p = .197 for ANOVA injection × time interaction, Figure 3F) were not different. Peak tension (p = .218 for ANOVA injection × time interaction, Figure 3G) and integrated tension (p = .615 for ANOVA injection by time interaction, Figure 3H) during EPR F I G U R E 1 Representative peak pressor response to EPR stimulation and the influence of acute NTS insulin receptor antagonism. The peak pressor responses to EPR stimulation are highlighted in italics. Peak pressor response to EPR stimulation is increased following acute NTS microinjections of the insulin receptor antagonist GSK1838705 (C and D), but not vehicle (A and B). The x-axis represents time, and y-axis arterial blood pressure. The dotted line and gray box indicate the contraction period. AP, arterial pressure; s, seconds. stimulation also did not differ 30 min following microinjections in either control or GSK1838705 groups. Figure 4 demonstrates the peak pressor and cardioaccelator responses to select stimulation of primary sensory nerve fibers before and following NTS microinjection of control or GSK1838705. In mechanoreflex experiments, the peak pressor response to passive stretch stimulation decreased in the control group, but it remained unchanged in the GSK1838705 group (Pre: Δ31 ± 20 mmHg, 30 min: Δ18 ± 19 mmHg in control vs. Pre: Δ24 ± 11 mmHg, 30 min: Δ28 ± 14 mmHg in GSK1838705, p = .003 for ANOVA injection × time interaction, Figure 4A). The peak HR response (p = .131 for ANOVA injection × time interaction, Figure 4B) was not different among groups. However, in chemoreflex experiments the peak pressor response to i.a. capsaicin stimulation was not different in either group (Pre: Δ43 ± 15 mmHg, 30 min: Δ35 ± 20 mmHg in control vs. Pre: Δ51 ± 22 mmHg, 30 min: Δ58 ± 16 mmHg in GSK1838705, p = .105 for ANOVA injection × time interaction, Figure 4C). On the contrary, the peak HR response significantly increased in the GSK1838705 group (Pre: Δ10 ± 10 bpm, 30 min: Δ7 ± 5 mmHg in control vs. Pre: Δ5 ± 4 mmHg, 30 min: Δ11 ± 6 bpm in GSK1838705, p = .023 for ANOVA injection × time interaction, Figure 4D). In tibial nerve stimulation experiments, the peak pressor response was augmented in the GSK1838705 group (Pre: Δ47 ± 20 mmHg, 30 min: Δ38 ± 21 mmHg in control vs. Pre: Δ47 ± 30 mmHg, 30 min: Δ64 ± 30 mmHg in GSK1838705, p = .004 for ANOVA injection × time interaction, Figure 4E). The peak HR response (p = .147 for ANOVA injection × time interaction, Figure 4F) did not differ. Finally, the peak tension to passive stretch did not change (p = .405 for ANOVA injection × time interaction) following NTS microinjections of either vehicle  Figure 5 demonstrates the influence of vehicle or GSK1838705 i.t. injections on the peak pressor, cardioaccelerator, or RSNA responses to stimulation by either passive stretch or i.a. capsaicin administration. The peak pressor response to passive stretch did not change in either group (p = .662 for ANOVA injection × time interaction, Figure 5A). Furthermore, the corresponding F I G U R E 2 Influence of acute insulin receptor antagonism in the NTS on the EPR in normal rats. Vertical bars represent average pressor (A and B), cardioaccelerator (C and D), and sympathetic (E and F) responses to contraction (G and H), before (Pre) and after (30 m) NTS microinjections of vehicle or GSK1837085 (100 μM) in normal rats. Open bars represent the means in the vehicle treated groups and gray bars represent the means in the GSK1838705 groups. Analyzed by two-way RM ANOVA and Sidak's multiple comparison test. **p < .01. Bars are mean ± SD. changes in peak HR (p = .180 for ANOVA injection × time interaction, Figure 5B) and RSNA (p = .199 for ANOVA injection × time interaction, Figure 5C) responses did not change. Similarly, the peak pressor response to stimulation by i.a. capsaicin did not change in either group (p = .754 for ANOVA injection × time interaction, Figure 5D). Likewise, the corresponding changes in peak HR (p = .293 for ANOVA injection × time interaction, Figure 5E) and RSNA (p = .445 for ANOVA time by injection × time interaction, Figure 5F) responses did not change. Finally, the peak tension to passive stretch was not different following i.t. injections (p = .983 for ANOVA injection × time interaction) of either vehicle (Pre: 0.7 ± 0.1 kg, 5 min: 0.7 ± 0.2 kg, 30 min: 0.7 ± 0.3 kg, 60 min: 0.8 ± 0.5 kg, 90 min: 0.7 ± 0.2 kg, 120 min: 0.7 ± 0.3 kg) or GSK1838705 (Pre: 0.7 ± 0.1 kg, 5 min: 0.7 ± 0.6 kg, 30 min: 0.7 ± 0.6 kg, 60 min: 0.6 ± 0.1 kg, 90 min: 0.7 ± 0.5 kg, 120 min: 0.7 ± 0.6 kg, data not shown in figure). Figure 6 illustrates an increased c-Fos protein expression (green channel) induced by repetitive EPR stimulation relative to sham control in the rostral NTS. Neuronal cells (blue channel) expressing IRs (red channel) also co-express c-Fos (merged channel; IR/c-Fos/ DAPI). Note the abundance of IR-positive neurons in the NTS, the high primary and secondary antibody titers used in these experiments ensured adequate saturation and illumination of the target. Remnants of the area postrema (−14.28 to −13.68 bregma) can be observed, as this area often tears off during processing or antibody incubations. Figure 7A demonstrates the increased overlap area between IR + /c-Fos + /& DAPI + in the rats receiving repetitive EPR stimulation versus sham controls (p < .001 for ANOVA group).

| DISCUSSION
The major findings of the current investigation were (1) antagonism of IRs in the NTS potentiates the peak pressor response to EPR stimulation via hindlimb contraction in baro-intact but not baro-denervated rats, (2) the peak pressor response to tibial nerve stimulation is potentiated following IR blockade in the NTS, (3) the peak pressor responses to mechano-and chemoreflex stimulation are not influenced by intrathecal injections F I G U R E 3 Influence of acute insulin receptor antagonism in the NTS on the EPR in baro-denervated animals. Vertical bars represent average pressor (A and B), cardioaccelerator (C and D), and sympathetic (E and F) responses to contraction (G and H), before (Pre) and after (30 m) NTS microinjections of vehicle or GSK1837085 (100 μM) sino-atrial baro-denervated rats. Open bars represent the means in the vehicle treated groups and gray bars represent the means in the GSK1838705 groups. Analyzed by two-way RM ANOVA and Sidak's multiple comparison test. Bars are mean ± SD. of the IR antagonist, and (4) the NTS is populated with IR-positive neurons that are sensitive to activation of the EPR.
The peak pressor responses to EPR activation during muscle contraction or tibial nerve stimulation were increased following NTS microinjections of the IR antagonist GSK1838705. Moreover, acute sino-aortic barodenervation abolished the potentiated pressor response to EPR stimulation. Additionally, the cardioaccelator response to i.a. capsaicin injections were also augmented following GSK1838705 microinjections, which lends further support to the hypothesis that acute IR inhibition of the cardiovascular control areas in the NTS enhance exercise induced cardiovascular responsiveness. These findings suggest that insulin signaling maintains the capacity to buffer cardiovascular responses to EPR stimulation, perhaps through interactions with NTS baroreflex neurons.
Antagonism of neuronal IRs in the NTS did not enhance RSNA responses to stimulation of muscle sensory fibers via skeletal muscle contractions. It is of note that in previous studies it was found that in rodents lumbar sympathetic nerve activity is particularly sensitive to central insulin signaling relative to renal or adrenal sympathetic nerve activity. [33][34][35] Since we did not measure differential sympathetic outflows in the current study, a role for an enhanced sympathetic excitatory mechanism originating from blockade of NTS IR-positive neurons cannot be excluded. Additionally, we were only successful in measuring the RSNA response to contraction in a limited number of rats (five in vehicle microinjections vs. four in GSK1838705 microinjections, Figure 2E,F), though there was no apparent trend in either peak or integrated responses. Therefore, we were limited in assessing the effect of NTS microinjection of IR antagonist on the sympathetic responses evoked by hindlimb afferent sensory nerve stimulations.
The first synapse in the EPR neural axis is in the dorsal horn of the spinal cord. 36,37 Augmented EPR responses may occur because of alterations in neurochemistry at this level of the EPR neuroaxis. 38 Additionally, dorsal and ventral horn neurons express IRs 39 and sequestering intrathecal insulin with antibodies in rats produces slowed motor nerve conduction and atrophy of axonal fibers similar to that seen in diabetic neuropathy. 40 In light of these findings, we tested whether acute inhibition of insulin signaling by i.t. injections of GSK1838705 alters the pressor, cardioaccelator and sympathetic responses to stimulation by passive stretch or i.a. capsaicin injections. However, we did not observe changes in any of these measures. These findings are congruent with our previous observations that GSK1838705 prevented acute insulin induced sensitization of mechanically and chemically evoked responses in both mouse DRG and rat muscle nerve preparations. 13,14 Thus, acute exposure of spinal IRs to GSK1838705 may not have any effect on peripheral nerve neurotransmission.
It is our contention that the volume of the experimental solution injected into the spinal fluids, as well as the experimental period, were sufficient to allow for exposure of the IR antagonist to the dorsal horn, spinal cord roots and dorsal root ganglion tissues. It was not experimentally practical, nor possible, to combine intrathecal injection and ventral root stimulation methods. Consequently, it F I G U R E 4 Influence of acute insulin receptor antagonism in the NTS on the pressor response to afferent stimulation in normal rats. Vertical bars represent the average peak pressor (A, C, and E) and cardioaccelerator (B, D, and F) responses to hindlimb afferent nerve stimulation, before (Pre) and after (30 m) NTS microinjections of vehicle or GSK1837085 (10 μM). The mechanoreflex (A and B) was induced by passive stretch. The chemoreflex (C and D) was induced by i.a. capsaicin. The tibial nerve stimulation (E and F) was performed at 10 × MT. Open bars represent the means in the vehicle treated groups and gray bars represent the means in the GSK1838705 groups. Analyzed by two-way RM ANOVA and Sidak's multiple comparison test. Statistical differences noted with asterisks. *p < .05, **p < .01. Bars are mean ± SD. remains to be definitively determined if acute inhibition of insulin signaling in the dorsal horn influences EPR responses to hindlimb contraction. Furthermore, spinal insulin acts as a neurotrophic factor and promotes normal sensory function by maintaining synthesis of neuromodulator proteins and peptides, 41 thus a chronic model of IR antagonism may reveal additional roles for insulin signaling in regulating EPR responses in healthy rats. Taken together, we must conclude that acute inhibition of spinal insulin signaling does not influence cardiovascular or sympathetic responses evoked by stimulation of hindlimb sensory afferents innervating skeletal muscle in healthy rats.
Peripheral sensory nerves converge onto dorsal horn neurons in the spinal cord, which have axonal projections to the NTS. 18 Consistent with these findings, immunostaining experiments have demonstrated that hindlimb muscle contraction and passive stretch increases neuronal c-Fos expression in the NTS of Wistar rats 20 and cats. 19 Therefore, we examined whether these same neurons in the NTS are also positive for IRs, and if the distribution of these neurons mirrored previous findings. Consistent with our previous work, 20 the current investigation demonstrated similar distribution patterns of c-Fos-positive nuclei co-expressing IRs in the NTS of Sprague-Dawley rats following repetitive EPR activation. Thus, our findings extend previous work, though further studies are warranted to determine the full physiological relevance of insulin signaling in the NTS. Particularly in the context of the direct impacts of insulin microinjections in the brain on the EPR function in healthy versus insulin resistant rats.
While the distribution of IR mRNA and radio-labeled insulin has been characterized in different regions of the rat brain, 42,43 the physiological relevance of IRs in the brain remains an active area of study. To our knowledge, this study is the first to elucidate a physiological role for insulin signaling in NTS neurons receiving input from muscle sensory afferent fibers. One possible mechanism of action could be the result of increased sensitivity of all the NTS neurons involved in regulating the response to EPR stimulation, resulting in a net enhancement of peak pressor responses. These neurons could include GABAergic inhibitory interneurons (baro-insensitive), and glutamatergic excitatory interneurons (baro-sensitive) which may crosstalk with one another as well as communicate with NTS output neurons that project to caudal ventrolateral medulla and sympathetic premotor neurons. 18 Increased excitability of these neurons may occur in central hypoinsulinemia by decreased F I G U R E 5 Acute influence of intrathecal GSK1838705 injections on the mechano-and chemoreflex in normal rats. Vertical bars represent the average peak pressor (A and D), cardioaccelerator (B and E) and sympathetic (C and F) responses to hindlimb stimulation by passive stretch or capsaicin injections, before (Pre) and after (5,30,60,90, and 120 m) intrathecal injections (vehicle or GSK1838705 100 μM). Open bars represent the means in the vehicle treated groups and gray bars represent the means in the GSK1838705 groups. Analyzed by two-way RM ANOVA. Bars are mean ± SD. hyperpolarization of neurons, the proposed mechanism has been discussed elsewhere. 17 The EPR is thought to activate spinal dorsal horn neurons that synapse onto GABAergic interneurons in the NTS. 18 As discussed above, the responsiveness of these inhibitory interneurons may be increased following acute IR inhibition. Thus, the potentiation of peak pressor responses to EPR stimulation following IR inhibition may be explained by enhanced inhibitory neurotransmission. This mechanism of action may involve crosstalk between GABAergic and glutamatergic interneurons, since barodenervation abolished the potentiated EPR following acute IR inhibition. Characterization of the overlap between IRpositive neurons with GABAergic interneurons in the NTS regions sensitive to EPR stimulation was beyond the scope of the current investigation. However, the elevated peak pressor response observed following pharmacologic inhibition of IRs is consistent with this line of reasoning.
The NTS receives multiple inputs from higher brain centers which may potentially influence current results. 44 For example, both the mesencephalic locomotor region and hindlimb group III sensory afferents influence the activity of NTS baro-sensitive interneurons. 45 However, the NTS may not receive much input from this region due to the precollicular decerebrate preparation used in this study. Since the GSK1838705 potentiated pressor response was abolished by baro-denervation, it is likely that acute alterations in neurotransmission mediated by IR inhibition are limited to changes F I G U R E 6 The EPR increases c-Fos expression in the NTS of normal rats. Insulin receptors are expressed on the surface of NTS neurons (A and B red channel), while activated c-Fos (C and D green channel) translocates into the nucleus (E and F blue channel). c-Fos expression in the rostral NTS of sham treated rats (C) appears decreased relative to c-Fos expression in the NTS (rostral to the c.s.) of EPR treated (D) rats. Grayscale photomicrographs were acquired for each channel at 10× magnification, merged into colored images (G and H), and then the contrast and brightness were adjusted to illustrate differences. The original photomicrographs were used for analysis. Scale bar (lower right within images) is 100 microns.
in the activity of baro-and EPR-sensitive NTS interneurons. Similarly, local astrocytes and glial cells in the NTS may also have a role in explaining the current results. For example, NTS astrocytes play a role in regulating other cardiovascular reflexes, such as the baroreflex. 46 Previous studies have shown that the vehicle DMSO has the potential to decrease the EPR when injected into the hindlimb circulation, 47 which may be a confounding factor in the current study. However, we did not observe an attenuation of the EPR in NTS microinjection control experiments with 0.1% DMSO containing aCSF solution. This could represent differences due to the site of injection (peripheral tissues vs. CNS tissues), the total injectate volume, a transient effect of DMSO, or a combination of these. To address this a 30 min period following the last microinjection allowed for any confounding effect of the vehicle itself to dissipate, as well as allowing time for sufficient diffusion of the experimental agent within the cardiovascular control region of the NTS. Thus, the depressive effect on the peak pressor response to passive stretch following NTS vehicle microinjections ( Figure 4A) may be explained by local tissue edema/bruising arising from the mechanical insult of the microinjection, thereby influencing the responsiveness to mechanoreflex stimulation. Although the use of filter paper as a medium to deliver the drug could have been used to prevent such issues, this strategy increases the diffusion gradient and the area effected by the drug, as well as the filter media itself binding the drug. That being said, we cannot fully differentiate the influence of local tissue damage from that of DMSO on neurotransmission in the NTS, especially those neurons in the NTS regions where the mechano-sensitive afferents have synapses, 20 which are near the injection sites.
There are additional limitations to the current investigation. First, peak responses in blood pressure were the only variable in which significant differences were observed, however, blood pressure is the primary variable and there is less variation in the peak pressor response than in the integrated pressor response since the latter is typically not sustained over the period of hindlimb afferent stimulation. Thus, the peak pressor response may be a more suitable measure for assessment of EPR magnitude. Decerebration also renders the preparation without central command input and decreased vagal tone, thus detection of changes in the cardioaccelerator response to EPR activation may be more difficult. Second, we limited the current investigation to the antagonist GSK1838705 (10 μM or 100 μM) since this chemical compound has the highest IC50 for the insulin receptor (1.6 nM) and can block the actions of physiological and supraphysiological levels of insulin at low concentrations in patch clamp studies. 13 This drug may also interact with neurons that express IGF-1 receptors; however, this assessment was beyond the scope of the present study. Third, the sequestration of endogenous insulin with foreign bodies or siRNA may require an extended period 48 or multiple days for the desired effects to manifest, 41 respectively. Therefore, acute pharmacological blockade may be the most effective method to observe acute changes with insulin signaling without simulating a disease model. Lastly, our interpretation of the results is limited to male rats only. In female rats, estrous cycles influence both the EPR 26 and the baroreflex. 27 We do not know if antagonism of insulin receptors in the NTS of female rats would influence the EPR or the baroreflex the same as in male rats. As such, future studies are warranted to determine if sex differences exist.
The implications of these findings suggests that brain insulin signaling is significantly altered in T2DM. 17 For example, insulin binding to brain capillaries is reduced in genetically obese hyperinsulinemic Zucker rats as compared to lean Zucker rats. 49 Furthermore, in dogs, increased adiposity induced by high-fat feeding is associated with reduced insulin delivery to the CNS. 50 These reports suggest that the chronic peripheral hyperinsulinemia associated with insulin resistance in T2DM results in hypoinsulinemia in the CNS. Understanding the consequences of central hypoinsulinemia may lead to improved or novel treatment paradigms. In the current study we show that decrease insulin signaling in the NTS via pharmacological antagonism of IRs alters central processing of sensory information, resulting in potentiation of the EPR. Rescuing central insulin signaling by restoring endogenous insulin transport into the brain may be a potential therapy to treat or reverse autonomic dysfunction associated with peripheral insulin resistance.
In conclusion, we have for the first time to our knowledge, demonstrated that the NTS is populated with IRpositive neurons that are sensitive to EPR stimulation. Inhibition of insulin signaling in this region of the NTS may increase neuronal excitability, perhaps by decreasing hyperpolarization, and altering normal processing of peripheral sensory afferent information. The net result is potentiation of the EPR. These results may shed light on the mechanisms in which EPR responses are augmented in rodent models of T2DM. 15