Blunted sympathoinhibitory responses in obesity-related hypertension are due to aberrant central but not peripheral signalling mechanisms

Abstract

The gut hormone cholecystokinin (CCK) acts at subdiaphragmatic vagal afferents to induce renal and splanchnic sympathoinhibition and vasodilatation, via reflex inhibition of a subclass of cardiovascular-controlling neurons in the rostroventrolateral medulla (RVLM). These sympathoinhibitory and vasodilator responses are blunted in obese, hypertensive rats and our aim in the present study was to determine whether this is attributable to (i) altered sensitivity of presympathetic vasomotor RVLM neurons, and (ii) aberrant peripheral or central signalling mechanisms. Using a diet-induced obesity model, male Sprague–Dawley rats exhibited either an obesity-prone (OP) or obesity-resistant (OR) phenotype when placed on a medium high fat diet for 13–15 weeks; control animals were placed on a low fat diet. OP animals had elevated resting arterial pressure compared to OR/control animals (P < 0.05). Barosensitivity of RVLM neurons was significantly attenuated in OP animals (P < 0.05), suggesting altered baroreflex gain. CCK induced inhibitory responses in RVLM neurons of OR/control animals but not OP animals. Subdiaphragmatic vagal nerve responsiveness to CCK and CCK₁ receptor mRNA expression in nodose ganglia did not differ between the groups, but CCK induced significantly less Fos-like immunoreactivity in both the nucleus of the solitary tract and the caudal ventrolateral medulla of OP animals compared to controls (P < 0.05). These results suggest that blunted sympathoinhibitory and vasodilator responses in obesity-related hypertension are due to alterations in RVLM neuronal responses, resulting from aberrant central but not peripheral signalling mechanisms. In obesity, blunted sympathoinhibitory mechanisms may lead to increased regional vascular resistance and contribute to the development of hypertension.

Introduction

Both the severity and incidence of obesity continue to rise, with the latter having escalated to pandemic proportions (Caprio et al. 2008; Lim et al. 2012). As a consequence, a yearly estimate of 2.8 million deaths can be attributed to being overweight or obese (Lim et al. 2012). Up to 78% of newly diagnosed cases of essential hypertension can be attributed to obesity, and hypertension remains the leading global risk factor for death and disability (Moore et al. 2005; Lim et al. 2012). However, despite the vast amount of research that has been conducted in this area, the exact pathophysiology of obesity-related hypertension remains vague and this knowledge is critical to treatment success (Plump, 2010; Jordan et al. 2012). Obesity-related hypertension is likely to be a multifactorial disorder (Esler et al. 2006; Lambert et al. 2010a,b), and the mainstream line of thought is that elevated sympathoexcitation and vasoconstrictor mechanisms are largely responsible for the development of this condition (Esler et al. 2006; Grassi et al. 2010b; Lambert et al. 2010a,b; Huber & Schreihofer, 2011). Nevertheless, there is evidence to suggest that impaired sympathoinhibition may be of equal importance, since human and animal obesity are associated with impaired arterial baroreflex control of heart rate and vascular resistance (Schreihofer et al. 2007; Grassi et al. 2010a; Huber & Schreihofer, 2011). Furthermore, in a model of diet-induced obesity (DIO), we have also demonstrated that the sympathoinhibitory and vasodilator effects of gastrointestinal (GI) hormones involved in blood pressure regulation are attenuated or abolished in obese, hypertensive Sprague–Dawley rats (How et al. 2011; 2013a,b; Sartor, 2013).

Cholecystokinin (CCK) and gastric leptin are GI hormones involved in regulating haemodynamic demand after a meal (for reviews see Sartor & Verberne, 2008; Sartor, 2013) and act directly or indirectly, respectively, to innervate CCK₁ receptors on subdiaphragmatic vagal afferents. This triggers a centrally mediated reflex response that is analogous to that of the baroreflex and ultimately relies on a distinct subset of neurons within the rostroventrolateral medulla (RVLM) (Sartor & Verberne, 2002, 2003, 2008, 2010; Sartor et al. 2006a; How et al. 2013b). The subpopulation of RVLM neurons inhibited by CCK is barosensitive and generally fast firing, fast conducting and predominantly non-catecholaminergic (Sartor & Verberne, 2003). These neurons are thought to govern sympathetic vasomotor outflow to the GI and renal vascular beds, resulting in withdrawal of sympathetic vasomotor tone to these areas to promote vasodilatation, thereby accommodating the increased haemodynamic demand after a meal (Sartor & Verberne, 2002, 2006b, 2008; Sartor, 2013). RVLM neurons are critically important for cardiovascular regulation and the maintenance of sympathetic tone to vascular smooth muscle (Sartor & Verberne, 2002, 2006b, 2008; Guyenet, 2006; Sartor, 2013) and have been shown to contribute to the development of obesity-related hypertension (Stocker et al. 2007), although the latter has not been examined at the single-neurone level. One of the aims in the present study was therefore to examine whether the responsiveness of electrophysiologically identified and characterised RVLM presympathetic vasomotor neurons is affected in obese, hypertensive rats.

The role of impaired sympathoinhibitory signals from the gut in the development of obesity-related hypertension has only recently been considered in the aetiology of this condition, although it remains to be determined whether aberrant peripheral or central mechanisms are responsible (How et al. 2011, 2013b; Sartor, 2013). Studies have shown that rodents exposed to a high fat diet display neuroplastic changes in vagal afferent neurons, and reduced jejunal and vagal afferent responsiveness to gut distension and CCK (Nefti et al. 2009; Daly et al. 2011; de Lartigue et al. 2011). However, possibly due to differences in rodent models/diets used, there are reports of both decreased (Nefti et al. 2009) and increased (Paulino et al. 2009) CCK₁ receptor transcript in the nodose ganglia of high fat-fed animals. To date the in vivo effects of CCK on subdiaphragmatic vagal nerve activity have not been examined in a DIO model in which the specific effects of diet and/or obesity can be evaluated. Therefore to address the hypothesis that aberrant subdiaphragmatic vagal afferent transmission may be responsible for the reduced sympathoinhibitory and vasodilator responses observed in our obese, hypertensive animals, a further aim was to examine the responsiveness of the subdiaphragmatic vagus to CCK administration, and quantitate the expression of CCK₁ receptor mRNA in the nodose ganglia in our DIO model. Both the baro-and the CCK-induced sympathoinhibitory reflex are dependent on vagal afferent excitation for activation of neurons in key cardiovascular medullary centres, including the nucleus tractus solitarii (NTS) and the caudal ventrolateral medulla (CVLM). To evaluate whether central or peripheral mechanisms are involved in reduced sympathoinhibitory responses, a further aim was to determine whether obese, hypertensive animals have reduced Fos-like immunoreactivity (Fos-IR) in conjunction with, or independent from, impaired vagal mechanisms in response to sympathoinhibitory stimuli such as those induced by CCK, in these key cardiovascular regions.

Methods

Ethical approval

All experiments were performed using male Sprague–Dawley from the Animal Resource Centre, Perth, Western Australia. For the obesity studies, a total of 96 rats (153–253 g) were used (3 cohorts of 32 animals); for the Fos saline control study, an additional four animals (349–370 g) were used. Animals were housed in a temperature-controlled animal facility with a 12 h light–dark cycle. This study was approved by the Ethical Review Committee of Austin Health (Heidelberg, Victoria, Australia) and complied with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes. The authors have read and complied with the guidelines for research in rodents outlined for The Journal of Physiology and UK regulations (Drummond, 2009).

For obesity studies, animals were acclimatised for 4 days before being placed on their specified diets as follows: in three separate cohorts of 6-week-old animals (n = 32 per group), 24 of the animals were placed on a moderately high fat diet (MHFD; 32% kcal from fat, SF04–037) and eight on a low fat diet (LFD control animals; 9% kcal from fat, AIN93M) as previously described (How et al. 2011, 2013b). All rats remained on their designated diets for 13–15 weeks. For the saline control group, animals were fed a normal chow diet and were experimented upon within 1 week of delivery.

To accommodate the maximum of four electrophysiological experiments that could be carried out per week, the rats were staggered so that each week only four rats were placed on their respective diets. For each study cohort, rats fed the MHFD (n = 24) were stratified according to weight gain (final weight – initial weight) as follows: the eight rats that gained the most weight were allocated to the obesity prone (OP; n = 8) group, while the eight rats that gained the least weight were allocated to the obesity resistant (OR; n = 8) group as previously described (How et al. 2011, 2013b). The remaining rats were excluded from weight gain analysis. The LFD (n = 8) animals served as controls.

General surgical procedures

Following the feeding cycle, on each experimental day either a LFD control or a MHFD rat was chosen at random to avoid bias. Animals were anaesthetised in a chamber with isoflurane and subsequently tracheotomised and artificially ventilated with 100% O₂ (1 ml (100 g body weight)^–1, 50–60 breaths min^–1) containing 1.5–1.7% isoflurane. The adequacy of anaesthesia was verified by noting absence of withdrawal to firm toe pinch and the absence of an eye blink to gentle corneal probing. These tests were carried out on a 15 min basis for the duration of the experiment. For RVLM recording experiments, paralysis was induced using pancuronium (1–2 mg ml⁻¹, i.v. bolus injection) to prevent muscle twitching during antidromic activation from the spinal cord, in which case a steady arterial pressure (AP) trace and a depressor response to firm toe pinch were used to indicate the adequacy of anaesthesia. Core temperature was maintained between 36–38°C using a rectal probe with input to a servo-controlled heating pad (Coherent Scientific, Hilton, South Australia).

The left jugular vein was cannulated for intravenous administration of drugs and the right brachial artery for measurement of mean AP and heart rate (HR) as previously described (Sartor & Verberne, 2010; How et al. 2011, 2013b). Following midline laparotomy, for ‘close arterial’ administration of drugs, a catheter was inserted into the left carotid artery, fed down the abdominal aorta, and visually placed just upstream from the coeliac artery for administration of CCK (0.05–4 μg kg⁻¹) in random order as previously described (Sartor & Verberne, 2010; How et al. 2011, 2013b).

Excision of fat pads, femur and measurement of naso-anal lengths

At the completion of all experimental procedures, the naso-anal length was measured and epididymal, infrarenal and subcutaneous fat pads excised and weighed as previously described (How et al. 2011, 2013b). For Study 1, the femur was extracted from the right hindlimb, and the length measured using an electronic vernier caliper (Sontax, 150 mm digital caliper).

Study 1: RVLM extracellular single unit recording

In addition to the general surgical procedures described above, an inflatable occlusive cuff was placed around the abdominal aorta caudal to the coeliac artery. Inflation of the cuff was used to elevate AP (aortic occlusion) for induction of a baroreflex response in order to determine whether RVLM neurons were barosensitive and thus vasomotor in nature. Occasionally, aortic occlusion failed to raise AP sufficiently to induce a robust baroreflex response, and in these cases PE (5 μg kg⁻¹, i.v.) was administered. Extracellular recording of RVLM presympathetic neurons was carried out as previously described (Sartor & Verberne, 2002, 2003, 2010). Antidromic activation and the collision test were used to identify RVLM presympathetic neurons (Sartor & Verberne, 2002, 2003, 2010).

Conduction velocities were calculated by dividing the straight-line distance between the recording electrode and the spinal stimulating electrode (in metres), by the antidromic latency (in seconds). Only neurons that met the following criteria were studied: barosensitive (i.e. inhibited by increasing AP); steady basal firing rate (FR) and spinally projecting. Cells were deemed to be CCK sensitive if they responded to CCK administration.

Study 2: subdiaphragmatic vagus nerve preparation and Fos immunohistochemistry

In addition to the general surgical procedures described, the liver was retracted to locate the subdiaphragmatic vagus nerve (SVN) on the anterior portion of the stomach–oesophagus. The nerve was subsequently isolated and placed onto the bared tips of a pair of Teflon-coated silver wires (bare diameter, 250 m; A-M Systems, Everett, WA, Australia) and embedded in silicone sealant (Kwik-Cast; Coherent Scientific, Hilton, South Australia). The wires were then externalised and the abdominal incision sutured closed. The SVN discharge (SVND) was recorded.

In a subset of animals, at the end of the subdiaphragmatic vagus experiments and once all doses of CCK had been administered, animals were left for a 1 h period. After this time, CCK (8 μg kg⁻¹) was administered intraperitoneally (i.p.) for induction of Fos-IR in the hindbrain as described by others (Myers & Rinaman, 2002). SVND was monitored for a further 2 h, after which the animals were immediately transcardially perfused.

Control saline group

The identical procedures to those described above were carried out in a separate group of normal chow-fed rats (n = 4) to determine whether experimental preparation and procedures contributed to c-Fos expression, with the exception that saline was administered i.p. instead of CCK. Immediately prior to the animals being killed, a local anaesthetic gel lignocaine (2%) was applied generously onto the subdiaphragmatic vagus to establish the zero level of nerve discharge. Resting SVND was determined at the beginning of the experiment over a 10 min period. Upon subsequent computer analysis, the zero level of noise discharge was subtracted from this 100% level of SVND, eliminating residual noise from the signal. SVND was analysed off-line, full-wave rectified, and averaged over 1 s intervals. Nerve discharge was therefore quantified as arbitrary units of activity and calculated using the following formula:

Transcardial perfusion

At the completion of the experiment, rats were deeply anaesthetised (4% isoflurane in inspired air) and blood was flushed from their vascular systems with tissue culture medium (DMEM/F12; Sigma Chemical, St Louis, MO, USA), and transcardially perfused with 1 litre of 4% formaldehyde in 0.1 m phosphate buffer, pH 7.4. Brains were postfixed overnight in the same fixative containing 20% sucrose, and then placed in a 30% sucrose solution for 16–24 h.

Peroxidase immunohistochemistry for light microscopy

Brain sections were cut using a cryostat (30 μm). Every second section was collected throughout the medulla and stored in phosphate-buffered saline. Sections were then permeabilised for immunostaining and pre-incubated in 10% normal horse serum in immunobuffer (10% NHS-IB) as previously described (Burman et al. 2004). Sections were then incubated in primary antibody solution (c-Fos, K-25: sc253, 1:20,000; Santa Cruz Biotechnology Inc, Santa Cruz, CA, USA) diluted in 10% NHS-IB for at least 24 h (up to 7 days) at room temperature in a humidity chamber. Between each incubation, all sections were rinsed 3 times with TPBS (Triton X-100 in PBS). Sections were rinsed and transferred to a secondary antibody solution (1:500; Biotin-SP-Conjugated AffiniPure Donkey Anti-Rabbit IgG; Jackson's ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and incubated for 24 h or overnight. The sections were then rinsed and incubated in avidin–horseradish peroxidase complex (1:1500 in IB) for 4 h prior to being reacted for 10–15 min with nickel-intensified diaminobenzidine (Sigma Aldrich, Castle Hill, NSW, Australia) to give sections a black amorphous reaction product (Burman et al. 2004). The reaction was completed with several TPBS rinses and sections were mounted onto gelatin slides to air dry. Sections were delipidated in xylene (Fronine Laboratory Supplies, Scoresby, VIC, Australia) and coverslipped using DPX Mountant for histology solution (Sigma Aldrich, Castle Hill, NSW, Australia).

Fos-IR analysis

Tissue sections were examined under a light microscope (Olympus BX60, Tokyo, Japan). Total number of Fos-positive neurons in the NTS and CVLM were counted in cardiovascular regions of the caudal medulla from plates 76 (Bregma −14.60 mm) to 72 (Bregma –13.68 mm) as described by us and others (Weston et al. 2003; Viltart et al. 2006; Llewellyn-Smith & Verberne, 2011) and according to the Rat Brain Atlas (Paxinos et al. 1999), and represented as average number of Fos positive neurons/section. Quantification of the Fos-IR neurons was performed bilaterally for both the NTS and the CVLM.

Study 3: determination of CCK₁ receptor expression in nodose ganglia

Following the general surgical procedures and collection of baseline AP and HR parameters, animals were deeply anaesthetised, following which the nodose ganglia were immediately excised, snap-frozen in liquid nitrogen and stored at –80°C until further use. Total RNA was isolated from the nodose ganglia of all animals by phenol chloroform method as previously described (Davey et al. 2000). Total RNA (5 μg) was treated with one unit of DNase I (DNase free kit, Ambion), and cDNA was subsequently synthesised from 1 μg of DNase-treated RNA using random primers and M-MLV transcriptase (Promega, Madison, WI, USA) according to the manufacturer's instructions. Quantitative real-time PCR (q-PCR) was performed in duplicate (200 ng cDNA per 20 μl) using an Applied Biosystems 7500 Real Time PCR System and an Applied Biosystems Taqman CCK₁ receptor (CCK₁R) gene assay (Assay ID: Rn00562164_m1), as described previously (Davey et al. 2008). Absolute expression was calculated using the ΔΔCT method with values for the gene of interest normalised to 18S rRNA (Applied Biosystems, CA, US; Assay ID 4331182) and expressed relative to a control sample.

Data analysis and statistics

All data are expressed as means ± standard error of mean (SEM), with P < 0.05 set as the significance level. Baseline RVLM neuronal discharge rates and SVND were measured prior to injection of any drugs or manipulations of AP. RVLM neuronal activity, SVND, AP and HR were monitored following close arterial CCK (0.05–4 μg kg⁻¹) administered as a single bolus, in random order. RVLM and SVN responses to CCK were measured immediately following infusion, during the nadir of the depressor response, as previously described (Sartor & Verberne, 2002, 2003, 2006b, 2007, 2010; Sartor et al. 2006a). AP and HR were allowed to return to baseline prior to subsequent injections. AP, HR, RVLM neuronal activity and SVND were stored and analysed using a Cambridge Electronic Design, computer-assisted data acquisition system and Spike2 software (version 5.13, Cambridge, UK). For in vivo experiments, following completion of surgical procedures, a minimum of 15 min was allowed for stabilisation and post-surgical recovery before subsequent experimental protocols were initiated.

All statistical analyses were carried out using GraphPad Instat (version 3.05 for Windows 95 GraphPad Software, San Diego, CA, USA). One-way analysis of variance followed by the Tukey–Kramer multiple comparison test was used to analyse the data between animals grouped as OP, OR or LFD. When the normality test failed, the Kruskal–Wallis test or the non-parametric Mann–Whitney test was used to determine significance. For correlation analysis, linear regression with a Pearson correlation was used for comparisons between variables, and r values were calculated as a measure of fit.

Materials

The following drugs/hormones were used: CCK octapeptide (CCK-8; sulphated form; American Peptide Co., Sunnyvale, CA, USA), lignocaine gel (Astra Zeneca Pty Ltd, North Ryde, NSW, Australia) and phenylephrine (PE; Sigma Aldrich, Castle Hill, NSW Australia).

Results

General results

All animals had similar initial weights and baseline HR (P > 0.05; see Table 1). In contrast, OP animals had significantly higher AP, weight gain, adiposity index, naso-anal length and femur length when compared to OR and control animals (P < 0.05 for all; see Table 1). AP was positively correlated with adiposity (r value = 0.475; P = 0.007), total fat pads (r value = 0.542; P = 0.002), weight gain (r value = 0.529; P = 0.002), femur length (r value = 0.391; P = 0.030) and naso-anal length (r value 0.437; P = 0.014).

Table 1.

Characteristics of animals grouped according to weight gain (obesity prone (OP), obesity resistant (OR)) compared with control animals on a LFD

	Control	OP	OR
Baseline AP (mmHg)	95 ± 2***	112 ± 2	97 ± 2†††
Baseline HR (beats min–1)	335 ± 5	355 ± 7	356 ± 7
Initial weights (g)	231 ± 2	225 ± 5	232 ± 2
Final weights (g)	520 ± 10***	611 ± 8	520 ± 6†††
Weight gain (g)	289 ± 9***	385 ± 6	288 ± 6†††
Total fat pad mass (g)	37 ± 2***	64 ± 3	45 ± 3†††
Adiposity Index	7.7 ± 0.4***	11.6 ± 0.5	9.5 ± 0.6†
Naso-anal length (cm)	26.7 ± 0.2***	27.8 ± 0.2	26.7 ± 0.2†††
Femur length (mm)	41.5 ± 0.5*	43.1 ± 0.3	41.1 ± 0.4††

Data represent means ± SEM of 22–24 rats/group (except for femur length, for which there are 8 animals/group). Control vs. OP:

^*P < 0.05,

^***P < 0.001; OP vs. OR:

^†P < 0.05,

^††P < 0.01,

^†††P < 0.001.

Study 1: RVLM recording experiments

We recorded from 19 presympathetic vasomotor RVLM neurons in total (control n = 6; OP n = 8; OR n = 5). In OP and control groups, we occasionally recorded from more than one neuron per animal, and the results include neurons from a total of four control and six OP animals, respectively; only one neuron per animal was recorded in the OR group. Two neurons were barosensitive but unresponsive to CCK and these were only evaluated for barosensitivity. Typically presympathetic vasomotor neurons inhibited by CCK are fast firing and fast conducting (conduction velocities > 1 m s^–1; Sartor & Verberne, 2002, 2003; Sartor et al. 2006a) and these were typical characteristics of neurons examined in the present study. As per our previous studies and those of others (Schreihofer & Guyenet, 1997; Sartor & Verberne, 2002, 2003), RVLM neurons selected for the present study were spontaneously active and were inhibited by increases in AP in a time-locked fashion, and were bulbospinal as determined by the collision test. Two of the neurons studied for barosensitivity (in the OR group; see below) were lost before the effects of CCK could be tested. RVLM neurons in OR and control animals were inhibited in a dose-dependent manner by CCK (Fig. 1A). In contrast, RVLM neurons in OP animals were only slightly inhibited or activated by CCK administration (Fig. 1A and B). The percentage change of RVLM neuronal activity from baseline in response to CCK administration was directly correlated to the resting AP of the animals, i.e. animals with higher AP tended to have excitatory responses, and those with lower AP inhibitory responses, to CCK (Fig. 2; see Table 2).

Figure 1.

A, CCK administration (0.05–4 μg kg⁻¹ close arterially) induced dose-dependent inhibitory responses in RVLM neurons of control/OR animals that were significantly reduced/abolished in OP animals (OP n = 8; OR n = 3; control n = 6). OP vs. control: *P < 0.05, **P < 0.01, ***P < 0.001; OP vs. OR: †P < 0.05. B, representative traces of RVLM neuronal responses in a control (upper traces; latency: 6 ms, conduction velocity, 1.9 m s^–1; firing rate, 7 spikes s^–1) and an OP (lower traces; latency, 3.5 ms, conduction velocity, 8.8 m s^–1; firing rate, 18 spikes s^–1) animal. Pulse triggered histograms (to the right of the traces) demonstrate poor pulse-synchronicity in the OP animal compared to the control animal. Proof that neurons are bulbospinal is determined by the positive collision tests (to the far right-hand side of the traces). While both neurons are barosensitive and inhibited by increases in arterial pressure induced by aortic occlusion (AOc), the OP animal is less sensitive to these changes. The neuron from the control animal is clearly inhibited by CCK administration (1 μg kg⁻¹) whereas in the OP animal, CCK induces a modest excitatory response.

Figure 2.

Animals with higher AP (predominantly OP animals) had reduced inhibitory, or excitatory, responses to CCK (at 0.1 μg kg⁻¹), whereas those with lower APs had clear inhibitory responses to CCK.

Table 2.

Correlation analyses: the effect of CCK on RVLM neuronal activity versus arterial pressure and weight gain

	AP (mmHg)		Weight gain (g)
[CCK]	r value	P value	r value	P value
0.05 μg kg−1	0.454	0.089	0.783	0.001
0.1 μg kg−1	0.672	0.004	0.749	0.001
0.5 μg kg−1	0.689	0.006	0.682	0.007
1 μg kg−1	0.592	0.016	0.640	0.008
2 μg kg−1	0.507	0.065	0.637	0.014
4 μg kg−1	0.700	0.011	0.734	0.007

Correlation analyses demonstrating the relationship between the effect of CCK administration on RVLM neuronal activity, and resting AP or weight gain of the animals. Significant correlations appear in bold.

OP animals clearly demonstrated a smaller inhibitory response to increases in AP, and we chose 20% inhibition for comparative purposes, as this was the maximum inhibition that was consistently attained for this group of animals. Both the absolute AP (OP 158 ± 8 mmHg; OR 127 ± 8 mmHg; control 124 ± 6 mmHg) and the change in AP from baseline (OP 43 ± 4 mmHg; OR 24 ± 6 mmHg; control 27 ± 2 mmHg) for 20% inhibition in neuronal firing rate were significantly higher (P < 0.05) in OP animals compared to OR or controls (Fig. 3A). We were able to inhibit only two of the neurons in OP animals to a level above 20% (one by 26% and the other by 45%; average neuronal inhibition 24 ± 3% with AP elevated to 160 ± 8 mmHg), and none were silenced. These neurons also demonstrated poor pulse synchronicity as demonstrated in Fig. 1B. On the other hand, 20% inhibition was the minimum level of inhibition observed in neurons from control and OR animals, with two neurons totally silenced. Upon raising AP to higher levels than required for 20% inhibition, the average neuronal inhibition for controls was 47 ± 11% (AP elevated to 141 ± 6 mmHg), and 53 + 14% (AP elevated to 145 ± 8 mmHg) for OR. Collectively for control and OR animals, the RVLM inhibitory response ranged from 20 to 100% (with 10/11 neurons displaying >20% inhibition), and maximal elevation of AP to an average of 143 ± 5 mmHg induced an average 50 ± 8% decrease in neuronal firing rate, with neurons displaying clear pulse synchronicity as demonstrated in Fig. 1B. While the percentage change in neuronal FR with maximal AP increase appeared lower in OP animals, this was not statistically significant when compared to OR and control animals (P > 0.05). Baroreflex-induced changes in HR were not significant amongst the groups (control −7 ± 3 beats min^–1; OR −7 ± 3 beats min^–1; OP −13 ± 5 beats min^–1; P > 0.05). RVLM vasomotor neurons were less responsive to blood pressure changes in OP animals when compared to those in OR or controls: i.e. both the change in AP from baseline and the absolute AP required for a 20% reduction in FR of these neurons, was significantly greater in OP animals (P < 0.05; Fig. 3A).While there was a tendency for RVLM neurons in OP to have higher firing rates than OR or controls, this did not reach statistical significance (P > 0.05; Fig. 3B). Even though the discharge rate for neurons in OP animals tended to be higher, there was no significant difference in the absolute change in discharge rate for 20% inhibition (control 3.4 ± 0.8 spikes s⁻¹; OR 2.8 ± 1.2 spikes s⁻¹; OP 4.5 ± 0.9 spikes s⁻¹; P > 0.05). Conduction velocities of neurons were also not significantly different between the groups (OP 5.5±0.8 m s^–1; OR 4.0 ± 0.8 m s^–1; control 4.9 ± 0.2 m s^–1).

Figure 3.

A, arterial pressure (AP) required for 20% inhibition in neuronal firing rate of RVLM neurons in control, OP and OR animals. Horizontal line represents mean resting AP for animals from Study 1 (OP n = 8; OR n = 5; control n = 6). Both the absolute AP (†P < 0.05, ††P < 0.01) and the change in AP from baseline (*P < 0.05) for 20% inhibition in neuronal firing rate were significantly higher in OP animals compared to OR or controls. B, dot plot representing individual basal firing rates (spikes s^–1) of neurons characterised in A for OP, OR or control animals. Horizontal lines represent mean values.

Study 2: SVN experiments and c-Fos immunohistochemistry

Administration of close arterial CCK induced a dose-dependent excitatory response in SVND that was not significantly different between OP, control and OR animals (P > 0.05 for all; Fig. 4). The number of Fos-IR neurons per section in response to i.p. CCK administration was compared between OP, controls and OR animals in the NTS and CVLM. In the NTS, only OP animals had significantly fewer Fos-IR neurons compared to control animals (P < 0.01; Fig. 5A; refer to Fig. 6 for representative photomicrographs). When evaluated according to specific rosto-caudal levels, Fos-IR in the NTS remained statistically significant between OP and control animals at almost every level examined (P < 0.01; Bregma −13.68 mm to –13.8 mm; P < 0.05, Bregma –14.08 mm to −14.30 mm). In the CVLM, there were significantly fewer Fos-IR neurons in OP animals compared to both control and OR animals (P < 0.05 for both; Fig. 5B). A similar number of sections were evaluated for each experimental group (control 23 ± 2; OP 25 ± 1; OR 23 ± 2; P>0.05). In the saline control group, Fos-IR was minimal in both the NTS (average 1.1 ± 0.6 neurons per section) and the CVLM (average 1.2 ± 0.7 neurons per section), and was significantly lower when compared to the Fos-IR in the diet-fed animals (OP, OR or controls; P < 0.05 for all in both the NTS and CVLM; ANOVA).

Figure 4.

A, CCK administered close to the coeliac artery induced a dose-dependent increase in SVND that was not significantly different between OP (n = 7), OR (n = 7) or control (n = 8) animals. B, SVND responses to varying doses of CCK in a control animal, demonstrating the dose-dependent effects.

Figure 5.

Values represented in graphs are means ± SEM of Fos immunoreactive (Fos-IR) neurons per section in the nucleus of the solitary tract (NTS; A) and caudal ventrolateral medulla (CVLM; B) between Bregma −14.6 to −13.68 according to the Rat Brain Atlas (Paxinos et al. 1999). In the NTS, there were significantly fewer Fos-IR neurons in OP compared to control animals. In the CVLM, there were significantly fewer Fos-IR neurons in OP compared to either control or OR animals. Values are means ± SEM; n = 4 per group; *P < 0.05, **P < 0.01.

Figure 6.

CCK administered at 8 μg kg⁻¹ in control (left-hand sections), OP (middle sections) and OR animals (right-hand sections). A–C, Fos-IR in the nucleus of the solitary tract (NTS); D–F, Fos-IR in the caudal ventrolateral medulla (CVLM). In both the NTS and the CVLM there were fewer Fos-IR neurons in OP compared to control and/or OR animals. Scale bar = 100 μm; magnification ×10. Schematic representation to the far right indicates typical areas from which Fos-IR was quantified (roughly within shaded areas) according to the Rat Brain Atlas (Paxinos et al. 1999).

Study 3: CCK₁ receptor mRNA in nodose ganglia

CCK₁ receptor mRNA in nodose ganglia did not differ between control (100 ± 37), OP (129 ± 45) and OR (128 ± 57) animals (n = 7–8 per group; P > 0.05).

Discussion

Our study is the first to demonstrate at the single-neuronal level, that the effect of CCK and baroreflex activation on RVLM neurons critical to cardiovascular regulation is blunted in diet-induced obese, hypertensive animals, most likely due to impaired central signalling mechanisms. Specifically, CCK-induced inhibitory responses observed in RVLM neurons of OR/control animals were significantly blunted or were excitatory in OP animals. Similarly, the barosensitivity of RVLM neurons was significantly attenuated in OP animals compared to control or OR animals, suggesting altered baroreflex gain. While the subdiaphragmatic vagal afferent response to CCK remained intact in obese animals, central neuronal activation, as determined by Fos-IR, was diminished in these animals when compared controls and/or OR animals. Our results indicate that attenuation of CCK-induced sympathoinhibitory mechanisms in obese, hypertensive animals may result from impaired signalling pathways in the brainstem, within the reflex circuit between vagal afferents and presympathetic RVLM neurons. This impairment may occur at one or all of the areas involved in the tri-synaptic reflex, i.e at (i) the excitatory synapses between the vagal afferents and NTS neurons, (ii) the excitatory synapses between NTS and CVLM neurons, or (iii) the inhibitory synapse between CVLM and RVLM neurons.

Elevated AP has been associated with alterations in the vascular walls; however, using the same DIO model upon which we have based our studies, Stocker et al. (2007) demonstrated that vascular reactivity (as determined by ED₅₀ responses to phenylephrine and norepinephrine (adrenaline)) is not different between OP, OR and control animals. Nevertheless, they found that inhibition of the RVLM using muscimol induced a greater decrease in arterial blood pressure in OP animals compared to OR or controls, and concluded that elevated sympathetic outflow in obese, hypertensive animals was mediated via the RVLM (although they did not record neuronal function). Their data would in fact suggest that RVLM neurons may not be less responsive to inhibitory input per se, but that the significantly reduced responses observed in our experiments could be attributed to a smaller level of activation in NTS and CVLM, which is also supported by our Fos data. The inhibitory effects of CCK on RVLM neurons and on sympathetic nerve activity are baro independent, but instead are dependent on activation of CCK₁ receptors at subdiaphragmatic vagal afferents (Sartor & Verberne, 2002, 2003, 2007, 2008; Verberne & Sartor, 2004; Sartor et al. 2006a,b). Presympathetic vasomotor RVLM neurons inhibited by CCK are thought to drive splanchnic and renal sympathetic outflow that affects blood flow to the relative vascular beds. We recently demonstrated that OP hypertensive rats had blunted renal sympathoinhibitory and vasodilator responses to CCK and gastric leptin when compared to normotensive controls or OR animals (How et al. 2013b; Sartor, 2013). In addition, the typical sympathoinhibitory and vasodilator responses induced by CCK in the splanchnic nerve and mesenteric vascular bed of control animals were significantly blunted or became sympathoexcitatory/vasoconstrictor in OP, hypertensive animals (How et al. 2011, 2013b). Since sympathetic vasomotor activity is dependent on input from RVLM presympathetic vasomotor neurons, one of the aims of the current study was therefore to determine, at the single neuronal level, whether these changes are due to altered responses in the subset of RVLM neurons hypothesised to be responsible for sympathetic vasomotor outflow to these vascular regions (Sartor & Verberne, 2002, 2003).

We recorded from barosensitive, fast-firing, fast-conducting (>1 m s^–1) RVLM neurons that are typically inhibited by CCK; in our experience, this subset of neurons is not activated by CCK (Sartor & Verberne, 2002, 2003; Sartor et al. 2006a). Instead, RVLM barosensitive neurons with slow-firing, slow-conducting characteristics are the subtype typically activated by CCK, and these have been hypothesised to drive sympathetic vasomotor outflow to the lumbar region, thus promoting vasoconstriction through sympathoexcitatory mechanisms (Sartor & Verberne, 2002, 2003). There were no significant differences in the firing rates or conduction velocities between OP, OR or control animals, indicating that the subset of neurons from which we recorded belonged to the subset of typically CCK-inhibited neurons. The present study correlated well with our previous studies in which OP animals had reduced/reversed sympathoinhibitory/vasodilator responses to CCK compared to controls or OR animals (How et al. 2011, 2013b), and similarly demonstrated that CCK induced a weaker inhibitory, or excitatory, response in RVLM neurons of these animals. We have previously hypothesised that CCK-sensitive RVLM presympathetic vasomotor neurons receive both direct excitatory inputs (from NTS or another source) and indirect inhibitory inputs (NTS–CVLM–RVLM) (Sartor & Verberne, 2006b). Activation or inhibition of these neurons by CCK is likely to depend on the proportion of excitatory versus inhibitory signals they receive. Thus, while CCK-inhibited neurons receive predominantly inhibitory inputs, CCK-activated neurons receive predominantly excitatory inputs. Removal of the inhibitory arm of the CCK reflex (by muscimol/kynurenate injections into CVLM) reverses the splanchnic sympathoinhibitory response to a sympathoexcitatory one, possibly due to unmasking of excitatory signals (Sartor & Verberne, 2006b). Similarly, in the present study, the subpopulation of fast-firing, fast-conducting RVLM presympathetic vasomotor neurons normally inhibited by CCK instead displayed predominantly excitatory responses to the hormone in OP animals (refer to Fig. 1A). One explanation for this is that the inhibitory arm of the CCK reflex is compromised in these animals, with the excitatory component remaining intact.

We previously demonstrated that impaired vasodilator responses to CCK were associated with higher AP, weight gain and adiposity of the animals, and animals with higher resting AP tended to have vasoconstrictor rather than vasodilator responses (How et al. 2013b). In direct accord with this, our current study showed that RVLM neuronal responses to CCK also correlated with resting AP and weight gain. Animals with lower resting AP or weight gain (such as OR and control animals) tended to have inhibitory responses to CCK, whereas animals with higher resting AP (mainly OP animals) had either reduced inhibitory or excitatory responses to CCK (Fig. 2). Elevated AP per se is unlikely to be affecting the response of RVLM neurons to CCK, as both peripheral nerve and RVLM neuronal responses to CCK remain intact under conditions of elevated AP (e.g. following phenylephrine administration; authors’ unpublished observations). Together with our previous studies, this data strongly suggests that reduced renal/splanchnic nerve and regional vascular responses to CCK in OP animals are due to changes in central reflex mechanisms.

Many have demonstrated altered baroreflex gain to be associated with obesity and hypertension in both animal (Huber & Schreihofer, 2010; Zhao et al. 2012) and human studies (for reviews see Grassi, 2004; Hall et al. 2007). A recent study by Fardin et al. (2012) demonstrated attenuated baroreflex-mediated changes in renal SNA in Wistar rats fed a high fat diet. However, baroreflex-mediated changes in a DIO model have to our knowledge, never been examined at the single neuronal level. In the current study we therefore sought to determine whether, apart from the CCK-induced sympathoinhibitory reflex, other sympathoinhibitory reflexes such as the baroreflex, are also affected at the single neuronal level in OP, hypertensive animals. Both the absolute AP and the change from baseline required to inhibit RVLM vasomotor neurons were greater in OP animals compared to control or OR animals, correlating well with other studies that have associated altered baroreflex gain with obesity hypertension (Huber & Schreihofer, 2010; Zhao et al. 2012). Furthermore, in our study, OP animals demonstrated poor pulse synchronicity when compared to control animals. Nevertheless, the fact that both the bradycardic response and maximal inhibitory response of RVLM neurons to baroreflex activation were not significantly different between OP, OR and controls in our study, indicates a baroreflex curve shift without a change in range, and may indicate differences between our rat model and that used by Fardin et al. (2012). It remains to be determined whether altered baroreflex control in obesity-related hypertension occurs as a compensatory mechanism to deal with increased sympathetic nerve activity, or whether it contributes to elevated sympathoexcitation itself (Hall et al. 2007).

Since RVLM neurons in OP animals had blunted responses to both CCK administration and to baroreflex activation, our results suggest a non-specific decrease in the response of RVLM neurons to inhibitory inputs. Whether increased drive from the RVLM in hypertension is due to increased firing rate of individual neurones or more active neurones under resting conditions remains a longstanding and unresolved issue. Interestingly there was a tendency for RVLM neurons in OP to have higher firing rates than OR or controls in our study, although this did not reach statistical significance (P > 0.05; Fig. 3B). A limitation of the present study is that cell recording for baroreflexes represents a large sampling bias, as neurons were selectively chosen for fast conduction velocities. A further limitation was the small neuronal sample size, although the study met the minim requirement of five rats per group required to detect significant differences in AP between experimental groups (at a significance of 0.05 and a power of 80%). Nevertheless, to adequately address whether obesity and hypertension are associated with elevated basal activity of RVLM neurons, a dedicated and extensive future study examining a much larger number of RVLM neurons and encompassing the different subpopulations, will be required.

Whether reduced sympathoinhibitory/augmented sympathoexcitatory mechanisms linked to hypertension are due to aberrant peripheral or central mechanisms, remains a contentious issue (Fink, 2010; Plump, 2010; Huber & Schreihofer, 2011; Sartor, 2013). Huber and Schreihofer recently demonstrated that attenuation of sympathoinhibitory reflexes including the baroreflex, in obese Zucker rats was due to changes occurring within the medulla and not due to aberrant vagal afferent signalling (Huber & Schreihofer, 2011). Similarly, baroreceptor-linked autonomic reflex pathways involved in controlling heart rate have been shown to be compromised in obesity due to centrally occurring deficits (McCully et al. 2012). In contrast, although not in obese animal models, others have demonstrated that diminished responsiveness of peripheral baroreceptor afferents may contribute to the development of hypertension (Gordon & Mark, 1984). It is possible that obesity-related hypertension may have a separate aetiology, since high fat diets and increased food intake known to cause obesity, have been shown to be associated with inflammation that may affect signalling mechanisms in the gut (de La Serre et al. 2010) and the brain (Cai & Liu, 2012). Therefore, a further aim in our study was to determine whether altered sympathoinhibitory/vasodilator responses to CCK in obesity stem from changes occurring within the central reflex circuitry, or at the level of subdiaphragmatic vagal afferents.

Diets high in fat content (≥45% total fat content) have been associated with changes in gut microbiota that promote inflammatory processes altering GI neuronal function and vagal afferent transmission (de Lartigue et al. 2011). Stemming from this and other evidence in the literature (Nefti et al. 2009; Daly et al. 2011; de Lartigue et al. 2011), we initially hypothesised that the reduced sympathoinhibitory and vasodilator responses to GI hormones in OP animals observed in our studies may be due to changes in SVN signalling. While we used an intact SVN in our experiments and did not isolate afferent traffic, it should be noted that the subdiaphragmatic vagus carries predominantly afferent fibres that outnumber efferent fibres by up to 16-to-1 (Undem & Weinreich, 2005). Nevertheless, while our recordings are likely to represent predominantly afferent nerve activity, it should be noted that absolute differences in afferent versus efferent responses could not be conclusively determined. We used our DIO model to examine the effects of CCK administration on SVN responsiveness in the different groups of animals, and contrary to our hypothesis, found no differences between OP, OR and control animals (Fig. 4). Consistent with this finding, using molecular techniques we demonstrated that CCK₁ receptor expression in the nodose ganglia of our animals did not differ between the groups. In contrast, others have demonstrated decreased (Nefti et al. 2009) or increased (Paulino et al. 2009) CCK₁ receptor expression in the nodose ganglia of high fat-fed animals, but these studies differed from ours in species used, higher and un-physiological fat content (>32%) and duration of diet exposure. This issue therefore remains unresolved in the literature. We can however, conclude that, at least in our physiologically relevant DIO model, impaired vagal afferent signalling is unlikely to be responsible for the reduction in RVLM neuronal, sympathoinhibitory and vasodilator responses observed in our studies.

To determine whether a defect within the central reflex circuitry was instead responsible, we examined the brainstem response to intraperitoneal CCK administration using the proto-oncogene marker of neuronal activation c-Fos. Similar to the baroreflex, the CCK-induced sympathoinhibitory reflex involves activation of medullary NTS and CVLM neurons; the latter in turn relay inhibitory signals to the subset of RVLM presympathetic vasomotor neurons that are inhibited by CCK. We therefore compared Fos-IR in the areas of the caudal brainstem activated by CCK, and found that OP animals had significantly less Fos-IR neurons in both these regions compared to controls and/or OR animals (Figs 5 and 6). In the NTS, this significant difference was only apparent between OP and control animals, whereas the expression in OR animals was not different to that of either OP or controls. In the CVLM on the other hand, the Fos expression in OP animals was significantly lower compared to both OR and control animals. This discrepancy could be due to the fact that the NTS is involved in a diverse range of CCK-related signalling mechanisms other than cardiovascular, including those related to satiety. In the latter case, peripheral signals are transmitted via the NTS to higher brain regions such as the paraventricular nucleus to regulate feeding (Blevins & Baskin, 2010). On the other hand, the sympathoinhibitory and cardiovascular effects of CCK are associated with a separate central pathway that necessitates the reflex activation of CVLM neurons, the latter of which is not involved in the feeding-related response to this hormone. Our results therefore highlight a defect in the CCK-induced sympathoinhibitory reflex pathway that is specific to OP animals, but suggests that feeding (or other) related changes in the CCK-induced central signalling pathways may occur due to diet. Others have also demonstrated reduced Fos expression in the NTS of high fat-fed obese animals in response to CCK (Covasa et al. 2000a) or substances known to cause CCK release (Covasa et al. 2000b), although expression in the CVLM was not examined in these studies due to its lack of relevance to satiety. The latter studies also failed to stratify animals on a high fat diet into OP or OR. To our knowledge ours is the first study in the literature to examine the response to CCK in specific cardiovascular controlling centres in the medulla oblongata, in a DIO model. Although we were unable to distinguish whether NTS neurons activated by CCK were involved in regulating satiety or sympathetic vasomotor tone, we were able to demonstrate a distinct decrease in Fos-IR in response to CCK in the CVLM of OP animals, an area that is specific to vasomotor regulation. Our results indicate that attenuation of CCK-induced sympathoinhibitory and vasodilator responses in obese, hypertensive animals may result from impaired signalling mechanisms in the brainstem. While the exact location or mechanism responsible for the central signalling defects was not determined in the present study, it is possible that the responsiveness of neurons within the central neurocircuitry involved in the baro-and CCK-induced sympathoinhibitory reflexes (NTS, CVLM and RVLM) to respective neurotransmitters, may be reduced in obese animals. This will need to be determined in future studies.

In conjunction with the present study, our research suggests that sympathoinhibitory and vasodilator responses are affected in obesity and appear to be associated with hypertension (How et al. 2011, 2013b). Impairment of reflex sympathoinhibitory mechanisms may lead to a net increase in vascular resistance and contribute to the aetiology of the disease. The importance of such changes is highlighted by studies demonstrating increased vascular resistance in the splanchnic vascular bed in hypertension (Osborn et al. 2011). Normally, the role of GI hormones is considered relevant in short term cardiovascular regulation that occurs during the postprandial period, but obese individuals often consume large quantities of high energy, high fat foods that take longer to digest (Prentice & Jebb, 2003). Under these circumstances gut hormones may take on a more critical role in order to sustain the increased haemodynamic demand of the kidneys and GI tract (Sartor, 2013). In obesity this phenomenon may be additive to the sympathetic pressor effects associated with various other factors including the renin–angiotensin–aldosterone systems and high circulating adipose-derived leptin, thereby indirectly contributing to obesity-related hypertension.

Our results suggest that central brainstem processing of sympathoinhibitory reflexes is compromised in obese, hypertensive animals, although the exact cause of this remains to be determined. There is increasing evidence for immune-to-brain signalling in pathological conditions such as obesity and hypertension (Lumeng & Saltiel, 2011; Cates et al. 2012). It has been suggested that neurogenic hypertension in spontaneously hypertensive rats may result from brainstem inflammation and hypo-perfusion which adversely affect neuronal function in critical cardiovascular centres such as the NTS (Waki et al. 2010, 2011, 2013; Cates et al. 2012). Other inflammatory markers have been shown to be up-regulated in obesity-related animal studies in response to over-nutrition and have been linked to brain inflammation (Lumeng & Saltiel, 2011; Cai & Liu, 2012). However, studies carried out thus far have been focused on metabolic pathways in the hypothalamus, and the role of inflammatory cytokines in medullary cardiovascular-related pathways is yet to be explored (see Lumeng & Saltiel, 2011; Cai & Liu, 2012). To date, such studies have only been conducted independently in either obese (normotensive) or hypertensive (non-obese) models. Given the compelling evidence suggesting that brainstem hypo-perfusion and inflammation may be important contributors to neurogenic hypertension in spontaneously hypertensive rats and the strong link between obesity and brain inflammation, it is possible that the reduced sympathoinhibitory responses observed in our studies result from inflammatory processes adversely affecting medullary neuronal function. Exploration of this possibility may encourage an exciting new area of research.

Perspectives

Despite the growing problem of obesity and related cardiovascular complications, we are yet to fully understand the pathophysiology underlying obesity-related hypertension. While sympathoexcitatory mechanisms are thought to play a major role in the development of hypertension in obesity, we and others have begun to focus on the role of sympathoinhibitory mechanisms that are likely to be equally as important. The latter notion stems from our previous studies in which we found renal sympathoinhibitory and vasodilator responses to CCK to be significantly blunted in OP hypertensive animals, as well as other studies that have demonstrated impaired baroreflex function in various models of obesity and hypertension. Whether central or peripheral signalling mechanisms are responsible for these impairments, however, remains somewhat contentious. In this respect, our current study has demonstrated that, at least with respect to the CCK-induced sympathoinhibitory reflex, these impairments occur centrally. The RVLM plays a critical role in sympathetic vasomotor control, and our study is the first, to our knowledge, that has examined the impact of obesity and hypertension on a subpopulation of these neurons at the single neuronal level. While we found that these neurons have diminished responses to both baroreflex-and CCK-induced stimuli in obese, hypertensive animals, the present study was unable to determine whether this results from an intrinsic change in neuronal properties or a downstream effect due to aberrant signalling from NTS and CVLM. Future studies will be required to evaluate these possibilities.

Key points

• Using a diet-induced obese rat model, we examined two sympathoinhibitory reflexes: the baroreflex and the reflex induced by the gastrointestinal hormone cholecystokinin (CCK).

• The change in neuronal discharge of presympathetic vasomotor neurons in the rostroventrolateral medulla (RVLM) to both sympathoinhibitory stimuli was significantly blunted in obesity-prone (OP) hypertensive animals when compared to obesity-resistant (OR) animals or controls on a low fat diet, at the single neuronal level.

• CCK₁ receptor expression in the nodose ganglia, and subdiaphragmatic vagal afferent responses to CCK were not significantly different between OP, OR or control animals.

• OP hypertensive animals had significantly reduced Fos-like immunoreactivity in the nucleus of the soliltary tract and the caudal ventrolateral medulla in response to CCK when compared to controls and/or OR animals, indicative of impaired signalling pathways in the brainstem within the reflex circuit between vagal afferents and presympathetic RVLM neurons.

• Blunted sympathoinhibitory responses in obesity-related hypertension are associated with blunted responses in RVLM neurons as a result of aberrant central but not peripheral signalling mechanisms.

Glossary

AP

arterial pressure

CCK

cholecystokinin

CVLM

caudal ventrolateral medulla

DIO

diet-induced obesity

Fos-IR

Fos-like immunoreactivity

GI

gastrointestinal

HR

heart rate

LFD

low fat diet

MHFD

moderately high fat diet

NHS-I

normal horse serum in immunobuffer

NTS

nucleus tractus solitarii

OP

obesity prone

OR

obesity resistant

RVLM

rostroventrolateral medulla

SVN

subdiaphragmatic vagus nerve

SVND

subdiaphragmatic vagus nerve discharge

Additional information

Competing interests

No conflicts of interest are declared by the authors.

Author contributions

D.M.S. was primarily responsible for the conception and the design of the study, with contributions from J.M.-Y.H. The electophysiological experiments were performed by D.M.S, S.A.W and S.I.A. The immunohistochemical studies were conducted by S.I.A and J.M-Y.H. Molecular studies were carried out by J.M-Y.H under the guidance of R.A.D. The manuscript was prepared by J.M-Y.H and D.M.S. and edited by all authors, who approved of the submitted version of the manuscript. All experiments were conducted at Austin Health, Heidelberg, Victoria, Australia.

Funding

This study was supported by the National Health and Medical Research Council of Australia, the Sir Edward Dunlop Medical Research Foundation and the University of Melbourne Research Grant Support Scheme (to D.M.S.).