Exposure to a high fat diet during the perinatal period alters vagal motoneurone excitability, even in the absence of obesity

Abstract

The perinatal period is critically important to the development of autonomic neural circuits responsible for energy homeostasis. Vagal neurocircuits are vital to the regulation of upper gastrointestinal functions, including satiety. Diet-induced obesity modulates the excitability and responsiveness of both peripheral vagal afferents and central vagal efferents but less information is available regarding the effects of diet per se on vagal neurocircuit functions. The aims of this study were to investigate whether perinatal exposure to a high fat diet (HFD) dysregulated dorsal motor nucleus of the vagus (DMV) neurones, prior to the development of obesity. Whole cell patch clamp recordings were made from gastric-projecting DMV neurones in thin brainstem slices from rats that were exposed to either a control diet or HFD from pregnancy day 13. Our data demonstrate that following perinatal HFD: (i) DMV neurones had decreased excitability and input resistance with a reduced ability to fire action potentials; (ii) the proportion of DMV neurones excited by cholecystokinin (CCK) was unaltered but the proportion of neurones in which CCK increased excitatory glutamatergic synaptic inputs was reduced; (iii) the tonic activation of presynaptic group II metabotropic glutamate receptors on inhibitory nerve terminals was attenuated, allowing modulation of GABAergic synaptic transmission; and (iv) the size and dendritic arborization of gastric-projecting DMV neurones was increased. These results suggest that perinatal HFD exposure compromises the excitability and responsiveness of gastric-projecting DMV neurones, even in the absence of obesity, suggesting that attenuation of vago-vagal reflex signalling may precede the development of obesity.

Key points.

• Obesity is recognized as being multifactorial in origin, involving both genetic and environmental factors. The perinatal period is known to be critically important in the development of neural circuits responsible for energy homeostasis and the integration of autonomic reflexes.

• Diet-induced obesity alters the biophysical, pharmacological and morphological properties of vagal neurocircuits regulating upper gastrointestinal tract functions, including satiety. Less information is available, however, regarding the effects of a high fat diet (HFD) itself on the properties of vagal neurocircuits.

• The present study was designed to test the hypothesis that exposure to a HFD during the perinatal period alters the electrophysiological, pharmacological and morphological properties of vagal efferent motoneurones innervating the stomach.

• Our data indicate that perinatal HFD decreases the excitability of gastric-projecting dorsal motor nucleus neurones and dysregulates neurotransmitter release from synaptic inputs and that these alterations occur prior to the development of obesity. These findings represent the first direct evidence that exposure to a HFD modulates the processing of central vagal neurocircuits even in the absence of obesity.

Introduction

The dramatic obesity epidemic has increased interest in understanding the neural regulation of energy balance and homeostasis. Obesity is multifactorial in origin and involves a combination of environmental and genetic factors (Levin, 2006, 2010b). Several lines of evidence have shown that the perinatal environment is critically important in the development of neural circuits responsible for energy and autonomic homeostasis (Levin, 2006, 2010b; Fox & Murphy, 2008). Vagal reflexes are vital to the control, integration and regulation of gastrointestinal (GI) reflexes, including satiety (Schwartz, 2000; Berthoud et al. 2006; Berthoud, 2008). Sensory information from the GI tract is transduced and relayed to the CNS via the afferent vagus nerve; the central terminals of these vagal afferents enter the brainstem via the tractus solitarius and terminate within the nucleus of the tractus solitarius (NTS) using predominately glutamate as a neurotransmitter (Andresen & Yang, 1990; Andresen & Kunze, 1994; Alywin et al. 1997; Baptista et al. 2005a). NTS neurones integrate this visceral sensory information with hormonal, metabolic and neural inputs from other brainstem and higher CNS centres involved in the regulation of autonomic homeostatic reflexes (Loewy, 1991; Rinaman, 2004; Berthoud & Morrison, 2008). The resulting neural signal is transmitted to the adjacent dorsal motor nucleus of the vagus (DMV) which contains the preganglionic parasympathetic motoneurones that supply vagal motor output back to the upper GI tract (Travagli et al. 2006).

Vagal sensorimotor neurocircuits start development from embryonic day 13 (E13); GI-projecting vagal motoneurones can be identified at day E14 and central vagal nuclei appear mature by E18 (Rinaman & Levitt, 1993). Vagal neurocircuits continue to undergo a considerable degree of synaptogenesis, reorganization and differentiation postnatally, however, and do not appear to be fully mature until approximately postnatal days 21–28 (P21–28) (Talley et al. 1997; Vincent & Tell, 1997, 1999; Rinaman, 2003; Dufour et al. 2010). The perinatal period therefore provides a critical opportunity during which autonomic neurocircuits may be re-wired permanently. Several studies have demonstrated the dramatic consequences that environmental and maternal stress during this period can exert upon the development and patterning of neural circuits controlling GI functions (Banihashemi & Rinaman, 2010; Rinaman et al. 2011).

Studies from several laboratories have indicated that diet-induced obesity (DIO) compromises the excitability and responsiveness of peripheral vagal afferents (Covasa et al. 2000a,b; Donovan et al. 2007; Little et al. 2007; Paulino et al. 2009; Daly et al. 2011; de Lartigue et al. 2011) as well as central vagal efferents (Browning et al. 2013b). Roux-en-Y gastric bypass-induced weight loss reversed some, but not all, of the DIO-induced alterations of vagal efferents, suggesting that a high fat diet (HFD) may alter neurones independently of obesity (Browning et al. 2013b). Previous studies have demonstrated that 2 weeks of exposure to HFD is sufficient to attenuate the ability of cholecystokinin (CCK) to inhibit gastric emptying and suppress food intake (Covasa & Ritter, 2000; Swartz et al. 2010), suggesting that HFD may dysregulate vagal reflex control of the upper GI tract even prior to the development of obesity.

The aim of the present study was therefore to use electrophysiological techniques to investigate whether exposure to a HFD during the critical perinatal period can also affect the excitability and responsiveness of central vagal neurocircuits prior to the development of obesity.

Methods

Ethical approval

All experiments were carried out under protocols approved by the Penn State University College of Medicine Institutional Animal Care and Use Committee.

Experimental animals

Timed pregnant Sprague–Dawley rat dams (Charles River, Kingston, NY, USA) were placed on either a control diet (13.5% kcal from fat; Purina Mills, Gray Summit, MO, USA) or HFD (60% kcal from fat; D12491; Research Diets Inc., NJ, USA) at pregnancy day 15. ‘Control’ rats (n = 28) and ‘perinatal HFD rats’ (n = 32) were subsequently weaned onto their respective diets at postnatal day 21. Experiments were carried out at 35–42 days of age. At this point, there were no differences observed between male and female rats, either in body weight or in electrophysiological, morphological or pharmacological properties, and the data were therefore combined.

Neuronal labelling

To label gastric-projecting vagal efferent neurones, the fluorescent neuronal tracer 1,1′dioctadecyl-3,3,3′,3′-tetra-methylinocarbocyanine perchlorate (DiI; Invitrogen, Grand Island, NY, USA) was applied to discrete gastric regions as described previously (Browning et al. 1999). Briefly, 14-day-old rat pups of either sex were anaesthetised with isoflurane (2.5% with air, 600 ml min⁻¹) and the abdominal area cleaned before exposing the stomach via an abdominal laparotomy. The stomach was isolated from the surrounding viscera and crystals of DiI were apposed to the major curvature of the stomach, either the fundus, corpus or antrum/pylorus region. The dye was embedded in place using a fast hardening epoxy resin and the surgical field flushed with warmed sterile saline before closing the abdomen in layers. Rats were allowed to recover for 21–28 days prior to experimentation.

Electrophysiological recordings

On the day of experimentation, brainstem slices were prepared for electrophysiological recording as described previously (Browning et al. 1999, 2013b). Briefly, rats were anaesthetised with isoflurane (5% with air) before being killed via administration of a bilateral pneumothorax. All brainstem slices were prepared between 09.00 and 10.00 h.

The brainstem was removed and placed in ice-cold oxygenated Krebs’ solution before using a vibratome to cut 4–6 coronal slices (300 μm) spanning the entire rostro-caudal extent of the DVC. Slices were incubated in oxygenated Krebs’ solution at 32 ± 1°C for at least 90 min prior to recording. A single brainstem slice was placed in a custom-made perfusion chamber (volume 0.5 ml) on the stage of a Nikon E600FN microscope equipped with tetramethylrhodamine isothiocyanate (TRITC) epifluorescence filters. The slice was maintained at 32 ± 1°C by continuous perfusion with warmed Krebs’ solution and held in place with a nylon mesh. Once their identity was confirmed, recordings were made from gastric-projecting DMV neurones under bright-field illumination using differential interference contrast (Nomarski) optics.

Whole cell patch clamp recordings were made using patch pipettes of 2–4 MΩ resistance when filled with potassium gluconate solution (see below for composition). Data were acquired using a single electrode voltage clamp amplifier (Axopatch 1D or Axoclamp 200A; Molecular Devices, Sunnyvale, CA, USA) at a rate of 10 kHz, filtered at 2 kHz and digitized via a Digidata 1320A or 1440 interface (Molecular Devices). Data were stored and subsequently analysed using a personal computer and pClamp software (Molecular Devices) or MiniAnalysis software (Synaptosoft, Decatur, GA, USA). Only those neurones with series resistance <15 MΩ were considered acceptable.

When recording evoked inhibitory postsynaptic currents (eIPSCs), neurones were voltage-clamped at −50 mV in the presence of kynurenic acid (1 mm). Synaptic currents were evoked using a tungsten bipolar stimulating electrode, tip separation 120 μm (WPI, Sarasota, FL, USA), placed on the adjacent centralis or medialis subnuclei of the NTS. Pairs of stimuli (0.05–1.0 ms; 10–500 μA, 150–500 ms apart) were applied every 20 s to evoke submaximal eIPSCs.

When recording spontaneous excitatory postsynaptic currents (sEPSCs), neurones were voltage clamped at −60 mV in the presence of the GABA_A receptor antagonist bicuculline (50 μm). When recording spontaneous inhibitory postsynaptic currents (sIPSCs), neurones were voltage clamped at −50 mV in the presence of the non-selective ionotropic glutamate receptor antagonist, kynurenic acid (1 mm) and recordings were made using pipettes filled with a potassium chloride solution (see below for composition); under these recording conditions, sIPSCs were recorded as inward currents. Miniature EPSCs and IPSCs (mEPSCs and mIPSCs, respectively) were recorded in the additional presence of TTX (1 μm) to block action potential-dependent neurotransmitter release.

Neuronal properties

Basic membrane properties: to calculate the membrane input resistance (R_in), the instantaneous current displacement was measured after the membrane was step hyperpolarized from −50 to −60 mV. The membrane capacitance was calculated by an algorithm of the pClamp software whereby repetitive square-wave voltage command pulses (10 mV) were applied to the neurone and the resultant transient current was fitted to C_m = Q_t/ΔV, where C_m is the membrane capacitance, Q_t is the total charge under the transient current response and ΔV is the change in voltage across the neuronal membrane.

Action potential characteristics

DMV neurones were current clamped at approximately −60 mV before depolarizing current pulses were injected (15–30 ms) of sufficient intensity to evoke the firing of a single action potential at pulse offset. The duration of the action potential was measured at threshold, the peak amplitude of the afterhyperpolarization was measured as was the duration of the afterhyperpolarization (decay constant, τ) after being fitted to a single exponential equation. To measure the firing frequency of DMV neurones, longer duration depolarizing current pulses (400 ms) of increasing intensity (30–270 pA) were injected. The number of action potentials fired during the current pulses were counted and expressed as pulses per second.

Voltage-dependent potassium currents

Voltage-dependent potassium currents were measured in the presence of TTX (1 μm) to block voltage-dependent sodium currents and nifedipine (1 μm) to block L-type calcium channels. Note that T-type calcium channels are not present in DMV neurones (K. N. Browning & R. A. Travagli, unpublished observations). To elicit the inactivation curve of the fast transient outward potassium current (I_A), DMV neurones were voltage clamped at −50 mV and hyperpolarizing current pulse steps (400 ms duration) were passed in 10 mV increments to −120 mV before repolarization to −50 mV. The resultant current values were normalized (I_max = 100), averaged and plotted. To elicit the activation curve for I_A, DMV neurones were voltage clamped at −50 mV and first hyperpolarized to −120 mV (400 ms duration) to remove the current inactivation, before being step depolarized from +20 to −70 mV in 10 mV increments (400 ms) and repolarized back to −50 mV. As these current pulse steps may also activate the delayed rectifier current, I_KV, neurones were voltage clamped at −50 mV and step depolarized from +20 to −70 mV without first hyperpolarizing the neurone to remove the I_A inactivation. The resulting delayed rectifier current was measured isochronically at the end of the depolarizing step, averaged and plotted. The outward currents were also subtracted from the control I_A activation currents to allow measurement of the pure I_A current, which was normalized (I_max = 100), averaged and plotted.

Morphological reconstructions

To allow reconstruction of the neuronal morphology, at the end of the experiment, Neurobiotin (2.5%, included in the recording pipette, Vector Laboratories, Burlingame, CA, USA) was injected into the neurone using subthreshold depolarizing current pulses (400 ms pulses at 0.8 Hz for 20 min as described previously; Browning et al. 1999). After removal of the pipette, the neuronal membrane was allowed to reseal for 10–20 min before the brainstem slice was fixed in Zamboni's fixative (see below for composition) at 4°C for at least 24 h.

Brainstem slices were then cleared of fixative by repeated washes in PBS containing Triton X-100 (PBS-Tx; see below for composition). Slices were incubated with avidin D–HRP solution (see below for composition) for 2 h. After repeated washes in PBS-Tx, brainstem slices were incubated in PBS containing diaminobenzidine, cobalt chloride and nickel sulphate for 30 min (see below for composition). Slices were subsequently incubated with 3% H₂O₂ for a period of time sufficient to allow adequate visualization of the neuronal morphology. Slices were mounted on gelatin-subbed slides, air dried and dehydrated through a series of graded alcohols and xylene before being mounted in Permount (Fisher Scientific, Pittsburgh, PA, USA).

Neurolucida software attached to a Nikon E400 microsope was used to create three-dimensional reconstructions of individual Neurobiotin-filled DMV neurones (final magnification ×400). The optical and physical compression of the slice that may have occurred during fixation and processing was corrected by a subroutine of the software that rescaled the section to 300 μm, the original thickness of sectioning.

The morphological features analysed included soma area, form factor (a measure of soma circularity where 1 = perfect circle and 0 = straight line), dendritic branching (number of segments, branch order) and dendritic length (in both the x- and y-axes). To be accepted, the neurones had to have a mediolateral and rostrocaudal branch extension of at least 200 μm, no major branches had to be severed during the initial sectioning, and the neuronal soma must not have been overtly damaged by removal of the recording pipette.

Analysis of space clamp

To examine the contribution that inadequate space clamp may have exerted upon the membrane properties of DMV neurones, a series of experiments were conducted in which the contribution of inadequately clamped potassium ion channels to the overall membrane current was reduced by raising extracellular potassium ion concentrations in the superfusing Krebs’ solution from 2.5 mm to 5 and 10 mm. The response of the neuronal membrane to application of a drug demonstrated previously to act via closure of potassium channels was then assessed. We have demonstrated previously that baclofen induces an outward current in gastric-projecting DMV neurones via activation of potassium channels (Browning & Travagli, 2001a). Neurones were voltage clamped at −50 mV, step hyperpolarized to −120 mV for 1.5 s before being ramp depolarized to +20 mV over a 3 s period; neurones were then hyperpolarized to −50 mV. Baclofen (30 μm) was applied and, once the induced outward current had reached a steady state, the ramp protocol was repeated. The baclofen-induced outward current was obtained by subtraction, and the reversal potential for the baclofen-induced current was calculated. To eliminate the contribution of sodium and calcium currents during the ramp depolarization, all experiments were conducted in the presence of TTX (0.3 μm) and nifedipine (1 μm).

Drug application and statistical analyses

Drugs were applied by superfusion via a series of manually operated valves, at concentrations demonstrated previously to be effective (Browning et al. 2002, 2004; Baptista et al. 2007; Browning & Travagli, 2007). When examining effects on action potential firing frequency, because DMV neurones are known to be spontaneously active (Travagli et al. 1991), current was injected into the neurone sufficient to produce a consistent, low frequency of spontaneous action potential firing. Cholecystokinin-8 s (CCK; 100 nm), methionine-enkephalin (Met-Enk; 10 μm), (2R,4R)-4-aminopyrrolidine-2,4-dicarboxylate (APDC; 100 μm) and (2S)-α-ethylglutamic acid (EGLU; 200 μm) were applied for periods of time sufficient for the response to reach a plateau, or for a minimum of 3 min if the neurone was unresponsive. A neurone was considered responsive if the neuroactive substance induced an alteration in action potential firing frequency of >50% or a change in either current frequency or amplitude >25%. Only responsive neurones were included in the subsequent analyses. When assessing the effects of diet, groups were compared using an unpaired t test or a χ-squared test, when appropriate. When assessing drug effects, each neurone served as its own control and the response in the presence of drug was assessed using Student's paired t test. Results are expressed as mean ± SEM, with significance defined as P < 0.05.

Drugs and solutions

Krebs’ solution (mm): 126 NaCl, 25 NaHCO₃, 2.5 KCl, 1.2 MgCl₂, 2.4 CaCl₂, 1.2 NaH₂PO₄ and 11 dextrose; maintained at pH 7.4 by bubbling with 95/5% O₂/CO₂.

Potassium gluconate intracellular solution (mm): 128 potassium gluconate, 10 KCl, 0.3 CaCl₂, 1 MgCl₂, 10 Hepes, 1 EGTA, 2 Na₂ATP and 0.25 NaGTP; adjusted to pH 7.36 with KOH.

Potassium chloride intracellular solution (mm): 140 KCl, 1 CaCl₂, 1 MgCl₂, 10 Hepes, 10 EGTA, 2 Na₂ATP and 0.25 NaGTP; adjusted to pH 7.36 with HCl.

PBS-TX (mm): 115 NaCl, 75 Na₂HPO₄, 7.5 KH₂PO₄ and 0.15% Triton X-100.

Zamboni's fixative (mm): 1.6% paraformaldehyde, 19 mm KH₂PO₄ and 100 Na₂PO₄ in 240 ml saturated picric acid and 1600 ml water; adjusted to pH 7.4 with NaOH.

Avidin-D–HRP solution: 0.002% avidin D–HRP in PBS containing 1% Triton X-100.

DAB solution: 0.05% DAB in PBS supplemented with 0.025% CoCl₂ and 0.02% NiNH₄SO₄.

TTX was purchased from Alomone Labs (Jerusalem, Israel), neurobiotin and avidin D–HRP were purchased from Vector Laboratories. EGLU and APDC were purchased from Tocris (R&D Systems, Inc., Minneapolis, MN, USA); all other chemicals were purchased from Sigma Chemical Co. (St Louis, MO, USA).

Results

Electrophysiological recordings were made from 219 gastric-projecting neurones from 60 rats. Of these, 103 neurones were control DMV neurones recorded from 28 rats that had been fed normal chow throughout the study. The remaining 116 neurones were from 32 rats whose dams had been fed a HFD from E13; these rats, and the neurones obtained from them, are referred to as perinatal HFD. Recordings were made from rats that were 35–42 days of age; at this time point there was no significant difference in weights of control versus perinatal HFD rats (190 ± 16 vs. 180 ± 9 g, respectively, P > 0.05), i.e. rats were neither overweight nor obese following exposure to the HFD for this period of time.

Perinatal HFD decreases the excitability of gastric-projecting DMV neurones

The input resistance of perinatal HFD gastric-projecting DMV neurones (425 ± 22 MΩ; n = 116) was lower than that of control DMV neurones (501 ± 25 MΩ; n = 103; P < 0.05) while membrane capacitance was increased (103 ± 3.3 vs. 95 ± 2.9 pF; P < 0.05). The duration of action potentials in perinatal HFD gastric projecting DMV neurones was not different from that of control DMV neurones (2.6 ± 0.07 vs. 2.7 ± 0.22 ms; P > 0.05) but the amplitude of the afterhyperpolarization was larger (16.3 ± 0.5 vs. 15.1 ± 0.4 mV; P < 0.05; Fig. 1A). No differences were apparent in the duration of the afterhyperpolarization (128 ± 13.4 vs. 149 ± 13.8 ms; P > 0.05).

Figure 1. Perinatal HFD decreases the action potential firing frequency of gastric-projecting neurones in the DMV.

A, representative traces illustrating the effects of perinatal HFD on action potential properties. DMV neurones were current clamped at −60 mV prior to injection of a short (15 ms) depolarizing current pulse of intensity sufficient to evoke firing of a single action potential at current pulse offset. Note that the amplitude of the action potential afterhyperpolarization was increased significantly by perinatal HFD. B, control and perinatal HFD DMV neurones were current clamped at −60 mV prior to injection of long (400 ms) depolarizing current pulses of increasing magnitude (30–270 pA). Note that perinatal HFD DMV neurones fired fewer action potentials. C, graphical representation of the frequency of action potential firing (expressed as pulses per second, p.p.s.) in DMV neurones from control and perinatal HFD rats. ∗P < 0.05 vs. control.

To determine whether the lower input resistance and larger afterhyperpolarization altered the firing frequency of gastric-projecting DMV neurones, the number of action potentials fired in response to injection of depolarizing current pulses (400 ms duration) was assessed. As illustrated in Fig. 1B and C, the number of action potentials fired in perinatal HFD DMV neurones was significantly lower than in control DMV neurones at all intensities of current pulses investigated.

These results suggest that exposure to HFD during the perinatal period is sufficient to alter the membrane properties of gastric-projecting DMV neurones, even prior to the development of obesity. As a result, perinatal HFD DMV neurones are less excitable and fire fewer action potentials in response to sustained membrane depolarization.

We, and others, have demonstrated previously that the action potential afterhyperpolarization in DMV neurones is mediated, in large part, by the apamin-sensitive (SK) calcium-dependent potassium current, although the fast transient I_A and the delayed rectifier I_KV contribute to a lesser degree (Sah, 1992, 1995; Browning et al. 1999). A charybdotoxin-sensitive (BK) calcium-dependent potassium current is present in DMV neurones, however, following diet-induced obesity (Browning et al. 2013b) or capsaicin-induced neuronal damage (Browning et al. 2013a). To determine whether the larger action potential afterhyperpolarization observed in gastric-projecting perinatal HFD DMV neurones was due to the expression of a charbydotoxin-sensitive calcium-dependent potassium current, neurones were current clamped at approximately −60 mV prior to injection of a short (15–20 ms) depolarizing current pulse of intensity sufficient to evoke the firing of a single action potential at the current pulse offset. Neurones were then superfused with apamin (100 nm) or charybdotoxin (40 nm) and the membrane was re-adjusted to −60 mV before re-evoking a single action potential.

Apamin decreased the afterhyperpolarization amplitude and duration in both control neurones (71 ± 2.8 and 58 ± 5.4% of pre-apamin amplitude and duration, respectively, P < 0.05 vs. baseline for both, n = 6 for each) and perinatal HFD gastric-projecting DMV neurones (78 ± 2.7 and 58 ± 10.5% of pre-apamin amplitude and duration, respectively, P < 0.05 vs. baseline for both, n = 6 for each). As reported previously (Sah, 1992, 1995; Browning et al. 1999), charybdotoxin had no effect on afterhyperpolarization amplitude or duration in control DMV neurones (100 ± 5.7 and 98 ± 12.2% of pre-charybdotoxin amplitude and duration, respectively, P > 0.05 for both, n = 6 for each). Similarly, in perinatal HFD DMV neurones, charybdotoxin had no effect on either afterhyperpolarization amplitude or duration (101 ± 5.1 and 131 ± 25.6%, P > 0.05 vs. baseline for both, n = 6 for each).

These results suggest that the larger afterhyperpolarization amplitude observed in perinatal HFD gastric-projecting DMV neurones is not due to the appearance of a charybdotoxin-sensitive (BK) calcium-dependent potassium current.

Perinatal HFD does not alter voltage-dependent potassium currents

The voltage-dependent activation and inactivation of the fast transient outward potassium current, I_A, was calculated for control and perinatal HFD DMV neurones (n = 18 for I_A inactivation; n = 80 for I_A activation). As illustrated in Fig. 2A and B, perinatal HFD had no significant effect upon the voltage dependency of either the inactivation or the activation of I_A. This suggests that, unlike diet-induced obesity, HFD itself does not open a window current, suggesting that alterations in the voltage dependency of I_A activation or inactivation are not responsible for the slower action potential firing frequency observed in perinatal HFD DMV neurones.

Figure 2. Perinatal HFD does not alter voltage-dependent potassium currents in gastric-projecting DMV neurones.

A, left: representative traces illustrating the activation of the ensemble outward potassium currents (upper trace, black) or the delayed rectifier potassium current (I_KV, red trace) evoked by voltage clamping the neuronal membrane from −120 mV or −50 mV to +20 mV. The current trace obtained by subtraction of I_KV from the ensemble current (blue) is composed principally of the transient outward potassium current, I_A. Right: to examine the voltage dependency of I_A inactivation, gastric-projecting DMV neurones were voltage clamped at −50 mV before being step hyperpolarized (400 ms duration) in 10 mV increments to −120 mV before being repolarized to −50 mV. The resultant current values were normalized (I_max = 100) and averaged. For the purposes of clarity, only the traces from −20, −50 and −120 mV are shown. B, graphical representation of the voltage-dependence of I_A activation and inactivation. Note that the voltage dependence of I_A activation was unaltered by perinatal HFD and there were only minor alterations in the voltage dependency of I_A inactivation. ∗P < 0.05. C, graphical representation of the normalized I–V relationship of the delayed rectifier potassium current, I_KV. Note that perinatal HFD had no effect on the magnitude of I_KV at any potential examined.

The delayed rectifier, I_KV, was also assessed in control and perinatal HFD DMV neurones (n = 26–39); when normalized relative to membrane capacitance (pA pF^–1), there was no significant difference in I_KV at any membrane potential tested (Fig. 2C).

In contrast to the effects of diet-induced obesity (Browning et al. 2013b), these results suggest that perinatal HFD exposure per se has no effect upon I_A or I_KV voltage-dependent potassium currents that were independent of changes in neuronal size or membrane capacitance.

Perinatal HFD does not alter properties of spontaneous or miniature excitatory or inhibitory currents

s/mEPSCs and s/mIPSCs were recorded in gastric-projecting DMV neurones voltage clamped at −60  and −50 mV, respectively. Briefly, there were no differences in any of the spontaneous current properties analysed between control and perinatal HFD neurones (frequency, amplitude, rise time, rise time_10–90, decay, half-width or total current, defined as frequency × area). Data are summarized in Table 1. Similarly, other than a decrease in mEPSC amplitude in perinatal HFD neurones (29.4 ± 2.5 vs. 44.8 ± 7.0 pA; P < 0.05) there were few differences in the miniature current properties compared to control neurones. These data suggest that perinatal HFD did not alter the synaptic release properties of terminals impinging upon gastric-projecting DMV neurones, although note that the relatively modest sample size of the current study may be too restrictive to make definitive statements regarding perinatal HFD-induced alterations in action potential-dependent and -independent synaptic transmission.

Table 1.

Properties of spontaneous and miniature excitatory and inhibitory currents

	sEPSCs		mEPSCs
	Control (n = 27)	Perinatal HFD (n = 26)	Control (n = 23)	Perinatal HFD (n = 24)
Frequency (p.p.s.)	2.1 ± 0.4	2.7 ± 0.5	0.8 ± 0.1	1.4 ± 0.2
Amplitude (pA)	33.2 ± 2.6	29.5 ± 2.0	44.8 ± 7.0	29.4 ± 2.5∗
Rise time (ms)	2.5 ± 0.1	2.1 ± 0.1	2.5 ± 0.1	2.2 ± 0.1
Rise time10–90 (ms)	1.7 ± 0.1	1.5 ± 0.1	1.9 ± 0.2	1.7 ± 0.2
Decay (ms)	3.2 ± 0.3	2.6 ± 0.2	3.60 ± 0.3	2.9 ± 0.1
Half width (ms)	3.1 ± 0.3	2.6 ± 0.2	3.7 ± 0.3	3.0 ± 0.1
Current (frequency × area)	213.4 ± 37.5	210.2 ± 56.2	216.2 ± 87.0	108.9 ± 21.1

	sIPSC		mIPSCs
s/mIPSCs	Control (n = 27)	Perinatal HFD (n = 24)	Control (n = 24)	Perinatal HFD (n = 21)
Frequency (p.p.s.)	1.5 ± 0.2	1.8 ± 0.3	1.1 ± 0.2	1,6 ± 0.3
Amplitude (pA)	66.2 ± 4.6	81.8 ± 6.6	71.1 ± 5.7	79.7 ± 6.1
Rise time (ms)	3.0 ± 0.1	2.8 ± 0.2	3.0 ± 0.2	2.8 ± 0.1
Rise time10–90 (ms)	2.0 ± 0.1	2.0 ± 0.4	2.0 ± 0.2	1.7 ± 0.1
Decay (ms)	7.2 ± 0.5	5.6 ± 0.4	7.8 ± 0.8	6.5 ± 0.3
Half width (ms)	6.5 ± 0.4	5.3 ± 0.4	7.6 ± 1.0	5.9 ± 0.3
Current (frequency × area)	511.6 ± 109.9	706.9 ± 158.7	590.3 ± 125.8	814.8 ± 150.6

Perinatal HFD affects the presynaptic, but not postsynaptic, response to CCK

Previous studies have demonstrated that gastric-projecting DMV neurones respond to a variety of satiety-related neuropeptides and neurohormones including CCK, and that this response is attenuated or lost under conditions of diet-induced obesity (Tong et al. 2011; Browning et al. 2013b). To determine whether perinatal HFD alters the responsiveness of gastric-projecting DMV neurones to CCK, neurones were current clamped to a membrane potential sufficient to allow a slow, steady spontaneous firing frequency. Neurones were then superfused with CCK (100 nm) and the alteration in firing frequency was analysed over a 30 s period of peak response. Seven of 10 control DMV neurones responded to CCK with an increase in action potential firing frequency (15 ± 3 vs. 155 ± 50 action potentials per minute, i.e. 936 ± 199% of baseline firing frequency, P < 0.05; Fig. 3A) while 11/12 perinatal HFD DMV neurones responded with an increase in firing frequency (9 ± 2 vs. 159 ± 35 action potentials per minute, i.e. 1433 ± 498% of baseline firing frequency, P < 0.05; P > 0.05 vs. proportion of responding control neurones; Fig. 3B).

Figure 3. Perinatal HFD does not affect the postsynaptic responsiveness of DMV neurones to CCK, but decreases the proportion of presynaptic terminals that increase glutamate release.

A, representative trace from a gastric-projecting DMV neurone illustrating the membrane depolarization and increase in action potential firing frequency in response to superfusion with CCK (100 nm). Neurones were current clamped to allow a low frequency of spontaneous action potential firing before superfusion of 100 nm CCK for a period of time sufficient for the response to reach a plateau. B, graphical representation of the proportion of gastric-projecting control and perinatal HFD DMV neurones responding to CCK with a membrane depolarization and increase in action potential firing (left bars) and the magnitude of the response (% increase in action potential firing rate; right bars). Note that perinatal HFD did not alter either the proportion of responding neurones or the magnitude of their response. C, CCK increases the frequency, but not amplitude, of mEPSCs in control and perinatal HFD DMV neurones. Six consecutive traces from a representative control DMV neurone voltage clamped at −60 mV, under control conditions (upper traces; black), during CCK (100 nm) perfusion (red traces, middle) and following wash-out (blue traces; lower). CCK increased the frequency of mEPSCs in 12/19 control DMV neurones, but only 4/22 perinatal HFD DMV neurones (P < 0.05). D, cumulative response graphs from the same neurone as in C illustrating the actions of CCK to increase mEPSC frequency (decrease in inter-event interval) but not mEPSC amplitude.

The ability of CCK to alter the neurotransmitter release from terminals impinging upon gastric-projecting DMV neurones was also assessed. In the presence of TTX (1 μm) to block action potential-dependent synaptic release, 6/14 control DMV neurones responded to CCK (100 nm) with an increase in frequency (from 1.1 ± 0.4 to 2.0 ± 0.6 pulses per second (p.p.s.), P < 0.05 vs. baseline) but not amplitude (58 ± 7.6 to 55 ± 8.5 pA, P > 0.05 vs. baseline) of mIPSCs, suggesting actions at presynaptic sites to increase GABA release. Similarly, 5/20 perinatal HFD DMV neurones (P > 0.05 vs. control DMV neurones) responded to CCK with an increase in mIPSC frequency (1.0 ± 0.2 to 1.5 ± 0.3 p.p.s., P > 0.05 vs. baseline) but not amplitude (85 ± 15.0 to 84 ± 16.3 pA, P > 0.05 vs. baseline).

In 12/19 control DMV neurones, CCK (100 nm) increased the frequency (from 1.7 ± 0.5 to 2.5 ± 0.8 p.p.s., P < 0.05) but not amplitude (26 ± 2.9 to 25 ± 2.5 pA, P > 0.05) of miniature glutamatergic EPSCs (mEPSCs; Fig. 3C–E). In contrast, only 3/15 perinatal HFD DMV neurones (P < 0.05 vs. control DMV neurones) responded to CCK with an increase in mEPSC frequency (0.4 ± 0.10 to 1.1 ± 0.49 p.p.s., P < 0.05) with no effect on amplitude (39 ± 9.4 to 36 ± 10.0 pA, P > 0.05; data not shown).

These data suggest that, in the absence of obesity, HFD does not alter either the proportion of gastric-projecting DMV neurones that respond to CCK with an increase in excitability, or the magnitude of that response. In contrast, perinatal HFD appears to have differential effects upon synaptic terminals impinging upon gastric-projecting DMV neurones, attenuating the ability of excitatory but not inhibitory nerve terminals to respond to CCK.

Perinatal HFD uncovers the ability of Met-Enk to inhibit GABAergic synaptic transmission

We have demonstrated previously that neurotransmitters/neuromodulators negatively coupled to adenylate cyclase, such as Met-Enk, inhibit excitatory glutamatergic, but not inhibitory GABAergic, synaptic transmission to gastric-projecting DMV neurones due, in part, to low levels of cAMP–protein kinase A activity within GABAergic brainstem terminals (Browning et al. 2002, 2004). In the present study, Met-Enk decreased mEPSC frequency (1.4 ± 0.3 to 0.6 ± 0.2 p.p.s., P < 0.05 vs. baseline) but not amplitude (30 ± 1.8 to 30 ± 2.1 pA, P > 0.05 vs. baseline) in 5/6 control DMV neurones as well as 3/5 perinatal HFD DMV neurones (2.4 ± 1.0 to 1.8 ± 0.9 p.p.s. and 21 ± 1.6 to 20 ± 0.3 pA for frequency and amplitude, P < 0.05 vs. baseline for both, respectively; P > 0.05 vs. control DMV neurones).

Furthermore, as described previously (Browning et al. 2002), in control DMV neurones, Met-Enk had no effect upon mIPSC frequency (0.8 ± 0.2 to 0.7 ± 0.2 p.p.s., P > 0.05 vs. baseline) or amplitude (89 ± 10.0 to 93 ± 14.0 pA, P > 0.05 vs. baseline) in any of the six neurones tested (Fig. 4 A, B). In contrast, in 7/8 perinatal HFD DMV neurones, Met-Enk decreased the frequency (1.1 ± 0.2 to 0.6 ± 0.1 p.p.s., P < 0.05 vs. baseline) but not the amplitude (78 ± 11.3 to 85 ± 14.1 pA, P > 0.05 vs. baseline) of mIPSCs, suggesting actions at presynaptic sites to inhibit GABAergic synaptic transmission (Fig. 4 A, B). Furthermore, Met-Enk decreased the amplitude of evoked IPSC (eIPSCs) in 7/8 neurones from 193 ± 32 to 158 ± 27 pA (i.e. a 19 ± 2% decrease, P < 0.05 vs. baseline; Fig. 4C), in contrast to previous studies in which Met-Enk did not alter evoked inhibitory currents in DMV neurones (Browning et al. 2002, 2004).

Figure 4. Perinatal HFD uncovers the ability of methionine-enkephalin (Met-Enk) to modulate GABA release.

A, Met-Enk decreases mIPSC frequency in perinatal HFD, but not control, DMV neurones. Six consecutive traces from representative control (left; black) and perinatal HFD (right; red) neurones voltage clamped at −50 mV in baseline conditions (top), during Met-Enk (10 μm) perfusion (middle) and following wash-out (bottom). Met-Enk decreased the frequency, but not amplitude, of mIPSCs in perinatal HFD, but not control, DMV neurones. B, graphical summary of the effects of Met-Enk to decrease mIPSC frequency in perinatal HFD (red) but not control (black) DMV neurones. Note that Met-Enk decreased the frequency, but not amplitude, of mIPSCs, indicating a presynaptic site of action. ∗P < 0.05 vs. pre-Met-Enk perfusion. C, Met-Enk decreases the amplitude of eIPSCs in perinatal HFD DMV neurones. Neurons were voltage clamped at −50 mV; electrical stimulation of the adjacent NTS was used to evoked pairs of IPSCs separated by 150–500 ms. Superfusion with Met-Enk (10 μm) decreased eIPSC amplitude in 7/8 perinatal HFD neurones. In contrast, Met-Enk had no effect on eIPSC amplitude in any of the 6 neurones tested.

Previously, we demonstrated that the low levels of cAMP within GABAergic brainstem terminals was due to tonic activation of group II metabotropic glutamate receptors (mGluRs) via glutamate released from vagal afferent terminals (Browning et al. 2006; Browning & Travagli, 2007). To determine whether disruption of this tonic activation was responsible for the ability of Met-Enk to inhibit GABA release, we examined the ability of group II mGluR agonists and antagonists to modulate synaptic transmission to gastric-projecting DMV neurones. To determine whether group II mGluRs are still present within brainstem circuits following perinatal HFD, the ability of the group II mGluR agonist APDC (100 μm) to modulate mEPSCs and mIPSCs was assessed.

APDC decreased the frequency (1.9 ± 0.8 to 1.0 ± 0.3 p.p.s., P < 0.05 vs. baseline) but not the amplitude (31 ± 2.0 to 29 ± 2.0 pA, P > 0.05 vs. baseline) of mEPSCs in 6/8 control DMV neurones while in perinatal HFD DMV neurones, APDC decreased the frequency (2.6 ± 0.8 to 1.8 ± 0.6 p.p.s., P < 0.05 vs. baseline) but not amplitude (22 ± 4.1 to 22 ± 3.4 pA, P > 0.05 vs. baseline) of mEPSCs in 4/6 neurones tested (P > 0.05 vs. control DMV neurones).

Similarly, APDC decreased the frequency but not amplitude of mIPSCs in both control (1.2 ± 0.2 to 0.7 ± 0.1 p.p.s., P < 0.05 vs. baseline and 79 ± 8.5 to 89 ± 0.6 pA, P > 0.05 vs. baseline; n = 9/11 neurones tested) and perinatal HFD DMV neurones (0.8 ± 0.2 to 0.5 ± 0.1 p.p.s., P < 0.05 vs. baseline and 88 ± 15.6 to 88 ± 18.0 pA, P > 0.05 vs. baseline; n = 5/5 neurones tested; P > 0.05 vs. control neurones; Fig. 5A, B). Furthermore, APDC decreased the amplitude of eIPSCs in 4/5 perinatal HFD DMV neurones tested (245 ± 65 vs. 194 ± 2 pA in the presence of APDC, P < 0.05 vs. baseline; Fig. 5C) as described previously for control DMV neurones (Browning et al. 2006; Browning & Travagli, 2007). These results suggest that group II mGluRs are constitutively present on both excitatory and inhibitory terminals impinging upon gastric-projecting DMV neurones and that exposure to perinatal HFD does not alter their expression.

Figure 5. Perinatal HFD attenuates the tonic activation of presynaptic group II mGluRs on inhibitory terminals.

A, the group II mGluR agonist APDC (100 μm) decreased mIPSC frequency in both control and perinatal HFD DMV neurones. Six consecutive traces from a representative perinatal HFD DMV neurone voltage clamped at −50 mV in control conditions (black; top), during APDC perfusion (red; middle) and following wash-out (black; bottom). Note that APDC decreased the frequency, but not amplitude, of mIPSCs. B, graphical summary of the effects of APDC to decrease mIPSC frequency in control (black) and perinatal HFD (red) DMV neurones. Note that APDC decreased the frequency, but not amplitude, of mIPSCs, indicating a presynaptic site of action. ∗P < 0.05. C, APDC decreased the amplitude of eIPSCs in perinatal HFD DMV neurones. Neurons were voltage clamped at −50 mV. Electrical stimulation of the adjacent NTS was used to evoke IPSCs. Superfusion with APDC (100 μm) decreased eIPSC amplitude in 4/5 perinatal HFD DMV neurones tested. D, the group II mGluR antagonist EGLU (200 μm) increased mIPSC frequency in control, but not perinatal HFD, DMV neurones. Six consecutive traces from a representative perinatal HFD DMV neurone voltage clamped at −50 mV in control conditions (black; top), during EGLU perfusion (red; middle) and following wash-out (black; bottom). Note that EGLU had no effect on mIPSC frequency or amplitude. E, graphical summary of the effects of EGLU to increase mIPSC frequency in control (black) but not perinatal HFD (red) DMV neurones. Note that EGLU increased the frequency, but not amplitude, of mIPSCs, indicating a presynaptic site of action. ∗P < 0.05. F, EGLU did not alter the amplitude of eIPSCs in perinatal HFD DMV neurones. Neurons were voltage clamped at −50 mV. Electrical stimulation of the adjacent NTS was used to evoke IPSCs. Superfusion with EGLU (200 μm) had no effect upon eIPSC amplitude in 6/7 perinatal HFD DMV neurones tested.

To determine whether group II mGluRs on central vagal neurocircuits were activated tonically, the ability of the group II mGluR antagonist EGLU (200 μm) to modulate mEPSCs and mIPSCs was assessed. As shown previously (Browning & Travagli, 2007), EGLU had no effect on mEPSC frequency (0.9 ± 0.2 to 0.8 ± 0.2 p.p.s., P > 0.05 vs. baseline) or amplitude (22 ± 2.5 to 21 ± 2.8 pA, P > 0.05 vs. baseline) in any of 7 control DMV neurones tested. Similarly, EGLU had no effect upon mEPSC frequency or amplitude (2.3 ± 0.5 to 2.3 ± 0.6 p.p.s. and 22 ± 2.5 to 21 ± 2.8 pA, P > 0.05 vs. baseline for both; n = 6/7 neurones) in perinatal HFD DMV neurones (suggesting that while these receptors are present on glutamatergic terminals, they are not activated tonically).

In contrast, while EGLU increased mIPSC frequency in 8/11 gastric-projecting DMV neurones (0.9 ± 0.2 to 1.3 ± 0.3 p.p.s., P < 0.05 vs. baseline), EGLU increased mIPSC frequency in only 1/7 perinatal HFD neurones and had no effect on the remaining 6 neurones (2.2 ± 1.1 to 2.4 ± 1.1 p.p.s., P > 0.05 vs. baseline; Fig. 5E, D). Additionally, EGLU had no effect on eIPSC amplitude in any of the 5 perinatal HFD DMV neurones tested (210 ± 31 vs. 203 ± 24 pA in the presence of EGLU, P > 0.05 vs. baseline; Fig. 5F) in contrast to previous studies in which EGLU increased eIPSC amplitude in control DMV neurones (Browning et al. 2006; Browning & Travagli, 2007). These data suggest that, following perinatal HFD, group II mGluRs present on GABAergic brainstem terminals are no longer activated tonically. Our previous studies have demonstrated that monosynaptic vagal afferent inputs activate presynaptic group II mGluRs on inhibitory GABAergic brainstem terminals. Removal of these inputs, or blocking their effects with group II mGluR antagonists, increases cAMP levels within these terminals, allowing the modulation of GABAergic synaptic transmission by other neurotransmitters/neuromodulators (Browning et al. 2004, 2006; Browning & Travagli, 2007). The data presented herein data suggest that perinatal HFD induces some alteration in either DVC neurones or in vagal afferent inputs to the brainstem, ultimately increasing cAMP levels and thus allowing modulation of GABAergic inhibitory transmission to gastric-projecting DMV neurones.

Perinatal HFD increases the size and dendritic extent of gastric-projecting DMV neurones

Morphological reconstructions were obtained from 17 control and 36 perinatal HFD gastric-projecting DMV neurones. As summarized in Table 2, perinatal HFD increased the soma area and dendritic extent, in both the x- and the y-axes, of gastric-projecting DMV neurones. In contrast, soma diameter and form factor were decreased, suggesting the DMV neurones were flatter and more elongated following perinatal HFD. An increase in segment length, segment ends and branch order suggests that perinatal HFD increased dendritic branching (Fig. 6).

Table 2.

Morphological properties of gastric-projecting DMV neurones

	Control neurones (n = 17)	Perinatal HFD neurones (n = 36)
x-axis (μm)	235 ± 27	506 ± 83∗
y-axis (μm)	141 ± 19	289 ± 39∗
Soma area (μm2)	228 ± 14	260 ± 13∗
Soma diameter (μm)	43 ± 11	24 ± 1.0∗
Form factor	0.69 ± 0.5	0.48 ± 0.05∗
Segment ends	4.5 ± 0.49	9.7 ± 0.75∗
Segment length	134 ± 12	254 ± 19∗
Branch order	2.0 ± 0.25	3.8 ± 0.23∗

^∗P < 0.05 vs. control neurones.

Figure 6. Perinatal HFD alters the morphological properties of gastric-projecting DMV neurones.

Computer-aided reconstructions of representative neurobiotin-filled DMV neurones from control (left) and perinatal HFD (right) rats. Note that, in comparison with neurones from control rats, perinatal HFD increased the soma size and extent of dendritic arborization.

The morphological alterations induced by perinatal HFD do not result in space clamp problems

To determine whether the altered neuronal membrane excitability and responsiveness following exposure to perinatal HFD occurred in response to space clamp issues resulting from the increased soma size and dendritic extent of gastric-projecting DMV neurones, a series of experiments were conducted in which the response to superfusion of the GABA_B receptor agonist baclofen (30 μm) was assessed in the presence of different extracellular potassium ion concentrations (2.5, 5 and 10 mm).

Baclofen induced a significant outward current (55 ± 7.3 pA) in 6/9 control DMV neurones in the presence of 2.5 mm K⁺ Krebs’ solution. The magnitude of this outward current was reduced to 28 ± 6.1 and 17 ± 4.8 pA in the presence of 5 and 10 mm K⁺ Krebs’ solution, respectively (Fig. 7). In perinatal HFD DMV neurones, the same concentration of baclofen induced an outward current of magnitude 62 ± 16.9, 32 ± 24.5 and 16 ± 3.4 pA in 2.5, 5 and 10 mm K⁺ Krebs’ solution, respectively, in 13/18 neurones (P > 0.05 vs. proportion of control DMV neurones; P > 0.05 vs. magnitude of response in control DMV neurones; Fig. 7). The reversal potential of the baclofen-induced outward current in control neurones was −113.2 ± 8.2, −92.5 ± 1.4 and −79.6 ± 4.2 mV, which was not significantly different from that of either perinatal HFD DMV neurones (−116.9 ± 1.7, −94.4 ± 1.6 and −74.1 ± 2.8 mV; Fig. 7) or the theoretical Nernstian reversal potential for a potassium-dependent response (−105, −87 and −69 mV for 2.5, 5 and 10 mm K⁺ solutions, respectively).

Figure 7. The morphological alterations induced by perinatal HFD do not result in space clamp problems.

A, neurons were voltage clamped at −50 mV before being step hyperpolarized to −120 mV (1.5 s) before being ramp depolarized to +20 mV over a 3 s period. Neurons were then repolarized to −50 mV. Representative traces from a control DMV neurone illustrating the current induced in response to this protocol in control conditions (black) and following superfusion with baclofen (30 μm; red), which induced an outward current. The baclofen-induced current, obtained by subtraction, is illustrated below (gray). B, graphical summary of the reversal potential for the baclofen-induced outward current in control (black) and perinatal HFD (red) neurones in the presence of 2.5, 5 and 10 mm extracellular [K⁺] alongside the theoretical reversal potential (blue) for a potassium-dependent current. Note that neither the control nor the perinatal HFD reversal potentials deviated significantly from each other or from the theoretical reversal potential, suggesting that the alterations in membrane properties observed following perinatal HFD were not due to space clamp problems secondary to increased dendritic arborization in these neurones. C, graphical summary of the baclofen-induced outward current in control (black) and perinatal HFD (red) neurones in the presence of 2.5, 5 and 10 mm extracellular [K⁺]. Note that the magnitude of the baclofen-induced outward current did not differ between control and perinatal HFD DMV neurones.

These results suggest that the alterations in excitability and responsiveness of gastric-projecting DMV neurones observed in response to perinatal HFD exposure are not due to space clamp problems arising from the increased soma size and dendritic arborization.

Discussion

The results of the present study suggest that perinatal exposure to a HFD alters the excitability and responsiveness of gastric-projecting DMV neurones even in the absence of increased body weight or obesity. Specifically, the present study demonstrated that perinatal HFD exposure: (i) decreases gastric-projecting vagal motoneurone excitability; (ii) decreases the responsiveness of glutamatergic, but not GABAergic, nerve terminals impinging upon gastric-projecting DMV neurones to CCK; (iii) uncovers the ability of Met-Enk to inhibit GABAergic synaptic transmission due, at least in part, to (iv) decreased tonic activation of presynaptic group II mGluR; and (v) alters DMV neuronal morphology. While DIO decreases the excitability and responsiveness of central vagal motoneurones (Browning et al. 2013b), the present study suggests that perinatal HFD exposure is sufficient to alter motoneurone behaviour and suggests that diet alone may modulate the ability of central vagal neurocircuits to regulate upper GI functions. It is important to note that the present study investigated only gastric-projecting DMV neurones and it remains to be determined whether similar alterations in membrane, pharmacological and morphological properties are also present in vagal motoneurones projecting to non-gastric regions, including the small intestine and pancreas.

While the causes of obesity are recognized as being multi-factorial, including both genetic and environmental factors, ultimately it is due to an energy imbalance, where intake consistently exceeds output (Berthoud & Morrison, 2008; Levin, 2010a). Vagal sensory afferents play critically important roles in the regulation of many GI functions, including satiety, and visceral afferents are considered to be important principally in short-term satiety signalling (Schwartz, 2000; Berthoud, 2008; Abizaid & Horvath, 2008). Consequently, less attention has been paid to the role of vagal neurocircuits in the longer term regulation of energy balance, satiation and the development of obesity. Vagal neurocircuits continue to develop postnatally, and are not fully patterned until approximately P21–28 (Rinaman et al. 2000; Rinaman, 2004; Dufour et al. 2010) which may render them especially vulnerable to developmental alterations in response to dietary manipulations. In both humans and rodents, exposure to a perinatal HFD can result in long-term adverse effects in offspring, including alterations in the regulation of energy balance and a tendency to develop metabolic syndrome, diabetes and obesity (Levin, 2006, 2010b). In the present study, rats were exposed to a HFD during three distinct phases of perinatal development: in utero, postnatal pre-weaning and the immediate post-weaning period. Future studies will be required to determine whether particular developmental periods are more or less vulnerable to the adverse effects of HFD exposure.

The behaviour and activity of vagal afferents is known to be remarkably labile; pathophysiological conditions such as insult or injury modulate afferent activity (Moore et al. 2000; Bielefeldt et al. 2002a,b; Gebhart et al. 2002; Kollarik & Undem, 2002, 2004; Myers et al. 2002; Dang et al. 2004; Tolstykh et al. 2004; Hermes et al. 2008). Similarly, the behaviour of vagal afferents is impacted adversely by diet-induced obesity (Covasa et al. 2000a,b; Donovan et al. 2007; Paulino et al. 2009; Daly et al. 2011; de Lartigue et al. 2011; Kentish et al. 2012). The activity and behaviour of vagal afferents appears surprisingly dynamic, however, and respond to ongoing physiological demands (Burdyga et al. 2006; Dockray, 2009) even perhaps on a minute-to-minute basis (Babic et al. 2012).

Obesity is not merely a disease of the peripheral nervous system, however, and while few studies have examined plasticity within vagal efferent motoneurones themselves, their activity, sensitivity and responsiveness are also affected by insult and injury (Browning et al. 2013a) as well as by DIO (Browning et al. 2013b). Obesity-induced GI dysregulation, including altered motility and accelerated gastric emptying, may therefore result from vago-vagal reflexes that are dysfunctional at multiple sites. As demonstrated recently, however, weight loss reverses some, but not all, of the obesity-induced effects on vagal efferent neurones, suggesting that the effects of diet on vagal neurocircuits may be distinct from those of obesity (Browning et al. 2013b). In the present study, we demonstrated that perinatal exposure to a HFD induces alterations in the biophysical, pharmacological, morphological and synaptic properties of vagal efferent motoneurones, even in the absence of obesity. It remains to be determined whether these adverse outcomes occur in response to similarly short periods of exposure to a HFD in postnatal rats and whether these effects are reversible or if they represent irreversible reprogramming of central vagal neurocircuits. Certainly, previous studies have demonstrated that changes in baroreflex control and heart rate variability are altered after remarkably short periods (1–4 days) of exposure to a HFD (Iliescu et al. 2013) suggesting that autonomic neurocircuits may be particularly susceptible to the adverse effects of dietary manipulation.

In the present study, perinatal exposure to a HFD decreased DMV neuronal excitability due, in part, to an increase in afterhyperpolarization amplitude. This is in agreement with our previous studies, demonstrating an increase in afterhyperpolarization amplitude in diet-induced obese rats or following neuronal damage (Browning et al. 2013a,b). In contrast to these previous studies, however, in the present study the increased afterhyperpolarization amplitude was not due to activation of a charbydotoxin-sensitive calcium-dependent potassium (BK) current. While the cause of the increased afterhyperpolarization amplitude in perinatal HFD neurones remains to be determined, it is of interest that previous studies have suggested the involvement of an apamin- and charbydotoxin-insensitive calcium-dependent potassium current that is associated with the transient outward potassium current (I_A) in regulating the action potential repolarization and afterhyperpolarization of vagal motoneurones (Sah, 1992, 1995). Given the increased amplitude of afterhyperpolarization observed in this study and others (Browning et al. 2013a,b), and the prominent role of this current upon DMV neuronal firing, more in-depth studies into its characterization are warranted.

Unlike previous studies showing a decrease in the proportion of vagal efferent neurones excited by CCK (Browning et al. 2013b), in the present study, there was no alteration in either the proportion of responding gastric-projecting DMV neurones or the magnitude of that response. Perinatal HFD exposure did, however, decrease the proportion of glutamatergic, but not GABAergic, nerve terminals impinging upon gastric-projecting vagal efferent motoneurones that increased neurotransmitter release in response to CCK. GI neurohormones such as CCK act primarily in a paracrine manner following their release from intestinal enteroendocrine cells (Raybould & Tache, 1988; Holzer et al. 1994; Raybould, 2007). Subsequent activation of peripheral vagal afferent terminals initiates vago-vagal reflex gastric relaxation, decreased gastric motility and emptying (Raybould & Tache, 1988; Holzer et al. 1994). The actions of these GI neurohormones may not be restricted to peripheral vagal afferents, however; the dorsal vagal complex (DVC, i.e. NTS, DMV and area postrema) is a circumventricular organ (Cottrell & Ferguson, 2004; Fry & Ferguson, 2007) and circulating neurohormones may act directly at the level of the DVC to modulate central vagal neurocircuits (Baptista et al. 2007). Indeed, CCK excites NTS and DMV neurones both directly and indirectly, by increasing the release of glutamate from presynaptic terminals (Appleyard et al. 2005; Baptista et al. 2005b; Zheng et al. 2005; Browning et al. 2011). In the present study, the decreased ability of CCK to modulate glutamatergic neurotransmission to gastric-projecting vagal efferent motoneurones suggests that, even in the absence of direct effects to modulate DMV neuronal membrane excitability, exposure to a perinatal HFD may decrease the ability of CCK to act presynaptically to increase DMV neurone excitability indirectly. Further experiments will be required to determine if the perinatal HFD exposure has similar effects on other satiety neurohormones known to modulate central vagal neurocircuits (Browning & Travagli, 1999; Browning et al. 2005; Wan et al. 2007, 2008; Browning & Hajnal, 2012).

It is well recognized that the activity of DMV neurones, and hence vagal efferent control of gastric motility and tone, is regulated by tonic GABAergic inputs, principally from the NTS (Sivarao et al. 1998; Travagli et al. 2006). As described previously, however, neurotransmitters and neuromodulators are unable to modulate GABAergic neurotransmission to DMV neurones due to the low resting levels of cAMP within inhibitory nerve terminals. Raising cAMP levels, regardless of the mechanism, allows previously inactive neurohormones such as insulin (Blake & Smith, 2014) or neurotransmitters/neuromodulators acting at μ-opioid, pancreatic polypeptides, 5-HT_1A receptors, α2 and oxytocin OT₁ receptors to inhibit GABAergic neurotransmission (Browning & Travagli, 2001b, 2006, 2009; Browning et al. 2004; Holmes et al. 2013). Further studies determined that the low levels of cAMP within NTS inhibitory nerve terminals were due to the ongoing activation of group II mGluRs via the tonic release of glutamate from vagal afferent terminals (Browning et al. 2006; Browning & Travagli, 2007). In the present study we demonstrated that, following perinatal HFD exposure, application of the μ-opioid receptor agonist Met-Enk decreased eIPSC amplitude and decreased mIPSC frequency, even in the absence of exogenous elevation of cAMP levels. Unlike control DMV neurones, the group II mGluR antagonist EGLU failed to increase eIPSC amplitude or mIPSC frequency following perinatal HFD exposure. The group II mGluR agonist APDC, however, decreased eIPSC amplitude and mIPSC frequency, indicating that the presynaptic inhibitory action of Met-Enk appears due to the loss of tonic activation of group II mGluRs, as observed previously following selective vagal deafferentation (Browning & Travagli, 2007; Holmes et al. 2013). These data suggest the possibility that the levels of cAMP–protein kinase A activity in the DVC are increased following HFD, possibly resulting from loss of tonic activation of group II mGluRs.

Studies from several laboratories have demonstrated that the excitability and responsiveness of vagal afferent neurones and peripheral terminals is decreased following the development of DIO (Covasa et al. 2000a,b; Donovan et al. 2007; Paulino et al. 2009; Daly et al. 2011; de Lartigue et al. 2011; Kentish et al. 2012). Results from the present study would suggest that, even in the absence of obesity, exposure to perinatal HFD also decreases the release of glutamate from central vagal afferents and attenuates the tonic activation of presynaptic group II mGluRs. Consequently, cAMP levels within presynaptic inhibitory terminals are elevated, and neurotransmitters or neuromodulators such as opioid peptides may modulate GABAergic neurotransmission to gastric-projecting DMV neurones. Given the prominent role that GABAergic inputs play in setting the activity state of DMV neurones regulating the functions of the upper GI tract, this suggests that central integration and modulation of vagal efferent outflow may be altered dramatically following exposure to a HFD, even prior to the development of obesity.

The morphological properties of DMV neurones are correlated with, but do not control, their biophysical and electrophysiological properties (Browning et al. 1999; Martinez de la Pena y Valenzuela et al. 2004). The branching pattern of dendrites must be sufficient to sample, and process, its converging input signals and reflects interactions between developmental as well as environmental cues. The dendritic pattern will therefore determine the amount of innervation, particularly from excitatory inputs, that each neurone receives. We have demonstrated previously that DIO increased the size and dendritic extent of vagal efferent motoneurones; these changes were not reversed following weight loss, however, suggesting that they occurred in response to the diet (Browning et al. 2013b). In the present study, similar alterations in the morphological properties of gastric-projecting DMV neurones were observed, suggesting perhaps that the increased neuronal size and dendritic arborization is a characteristic response of vagal efferent motoneurones in response to a HFD. Interestingly, previous studies have also shown that the dendritic pattern of cortical neurones is rearranged following deafferentation as neurones begin to ‘turn’ towards active areas to form new connections (reviewed by Macias, 2008). It is therefore tempting to speculate that the increased dendritic arborization observed in the present study may be an attempt to retain synaptic innervation despite a loss of neuronal excitability and a probable decrease in excitability of vagal afferent inputs (Daly et al. 2011; Kentish et al. 2012).

Summary and conclusions

The results of the present study demonstrate that perinatal HFD compromises central vagal neurocircuits even in the absence of increased body weight or obesity. The actions of perinatal HFD to alter the sensitivity and responsiveness of gastric-projecting DMV neurones largely mimic those observed in DIO in adult rats, suggesting that disrupted vagal efferent signalling occurs in advance of obesity. The perinatal period represents a time of continuing development that may render vagal neurocircuits especially vulnerable to disruption by dietary manipulation. Future studies will be required to determine whether the adverse effects of perinatal HFD on gastric-projecting DMV neurones are reversible or whether they represent irreversible damage and reprogramming of central vagal neurocircuits, which may have significant implications for the longer term neural regulation of energy balance in offspring.

Glossary

CCK

cholecystokinin

DIO

diet-induced obesity

DMV

dorsal motor nucleus of the vagus

DVC

dorsal vagal complex (i.e. NTS DMV plus area postrema)

E13

embryonic day 13

eIPSC

evoked inhibitory postsynaptic current

HFD

high fat diet

GI

gastrointestinal

IA

fast, transient outward A-type potassium current

IKV

delayed rectifier potassium current

Met-Enk

methionine-enkephalin

mGluR

metabotropic glutamate receptor

NTS

nucleus of the tractus solitarius

P21–28

postnatal days 21–28

s/mEPSCs

spontaneous/miniature excitatory postsynaptic currents

s/mIPSCs

spontaneous/miniature inhibitory postsynaptic currents

Competing interest

The authors declare no competing financial interests.

Author contributions

Conception and design of experiments: K.N.B. Collection, analysis and interpretation of data: R.B., S.R.F., K.N.B. Drafting the article or revising it critically for important intellectual content: R.B., S.R.F., K.N.B.

Funding

Funding was provided by the National Science Foundation IOS1138978 (to K.N.B.).