The subfornical organ (SFO) is a critical circumventricular organ involved in the control of cardiovascular and metabolic homeostasis. Despite the plethora of circulating signals continuously sensed by the SFO, studies investigating how these signals are integrated are lacking. In this study, we use patch-clamp techniques to investigate how the traditionally classified “cardiovascular” hormone ANG II, “metabolic” hormone CCK and “metabolic” signal glucose interact and are integrated in the SFO. Sequential bath application of CCK (10 nM) and ANG (10 nM) onto dissociated SFO neurons revealed that 63% of responsive SFO neurons depolarized to both CCK and ANG; 25% depolarized to ANG only; and 12% hyperpolarized to CCK only. We next investigated the effects of glucose by incubating and recording neurons in either hypoglycemic, normoglycemic, or hyperglycemic conditions and comparing the proportions of responses to ANG (n = 55) or CCK (n = 83) application in each condition. A hyperglycemic environment was associated with a larger proportion of depolarizing responses to ANG (χ2, P < 0.05), and a smaller proportion of depolarizing responses along with a larger proportion of hyperpolarizing responses to CCK (χ2, P < 0.01). Our data demonstrate that SFO neurons excited by CCK are also excited by ANG and that glucose environment affects the responsiveness of neurons to both of these hormones, highlighting the ability of SFO neurons to integrate multiple metabolic and cardiovascular signals. These findings have important implications for this structure’s role in the control of various autonomic functions during hyperglycemia.
- subfornical organ
- autonomic regulation
- angiotensin II
to maintain homeostasis, it is important that the brain is able to integrate the various signals circulating in the blood, which represent information about the “milieu interieur”. Many of these signals, however, are unable to cross the blood-brain barrier (BBB) due to their large and/or lipophobic nature. One way through which this circulating information can be communicated to the central nervous system (CNS) is via the circumventricular organs (CVOs). What allow these unique CNS structures to sense these circulating signals are their dense vascularization, containing fenestrated capillaries (15, 35), and the expression of receptors for many of these signals (for review, see Ref. 32). This information is then communicated via neuronal efferents to areas in the brain protected by the BBB.
The subfornical organ (SFO) is a forebrain sensory CVO with neuronal projections to various hypothalamic nuclei critical for maintaining normal cardiovascular and metabolic homeostasis. These include, but are not limited to, magnocellular neurons of the paraventricular (PVN) and supraoptic (SON) nuclei that secrete vasopressin and oxytocin into the circulation, which directly constrict blood vessels and promote water reabsorption in the kidneys (33, 34). Parvocellular neurons in the PVN project to the medulla and spinal cord, which activate the sympathetic nervous system to affect heart rate, blood pressure, and water/sodium reabsorption in the kidney (4, 19, 28, 40, 44, 51). Additionally, the SFO projects to the lateral hypothalamus and the arcuate nucleus, regions that play important roles in energy balance (16, 53).
Consequently, electrical stimulation of the SFO has been shown to cause direct increases in blood pressure (12), increase plasma vasopressin concentrations (11), and elicit increased water (47) and food (50) intake in rats. More recent literature, however, has suggested that the SFO’s role in food intake is not as clear as its effects on drinking. Oka et al. (37) used optogenetics to stimulate excitatory [calcium/calmodulin-dependent protein kinase II (CAMKII) positive] or inhibitory (vesicular GABA transporter positive) subpopulations of SFO neurons and showed that water intake by these animals could be elicited and inhibited, respectively, without effects on food intake (37). Conversely, Nation et al. (36) showed that chronic stimulation of CAMKII-positive SFO neurons using Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) caused robust increases in water intake accompanied by a small increase in body weight (36). Interestingly, when looking at how the transcriptome of the SFO is modified in both dehydrated and fasted rats compared with controls, it was found that the number of gene transcripts changed in animals that were deprived of food were 15 times higher than those that were water deprived (20). Together, these studies provide clear evidence of a critical role for the SFO in regulation of the cardiovascular system and an influential role in energy metabolism.
To sense circulating signals, the SFO must express receptors for these signals. From the aforementioned neuroanatomy, connectivity, and physiological roles of the SFO, it may come as no surprise that this sensory CVO expresses receptors for and can respond to various circulating cardiovascular signals, such as osmolarity (45, 46), sodium (21), ANG (26, 31), apelin (9), and atrial natriuretic peptide (6, 43), as well as metabolic signals, including glucose (30), ghrelin (41, 52), leptin (48), amylin (42), adiponectin (2), and CCK (1). This being said, it is important to emphasize that although the literature classifies hormones as “metabolic” or “cardiovascular,” many have been shown to also affect systems different from those that they were originally classified (for review, see Ref. 8). For example, leptin, a “satiety hormone,” produced and secreted by adipose tissue (for review, see Ref. 14), has also been shown to decrease blood pressure when microinjected into the SFO of young lean rats (49).
Thus, it is clear that the SFO is continuously sensing a plethora of circulating signals that influence a number of different physiological systems, leading to the obvious suggestion that single SFO neurons will sense and integrate information from multiple signals before relaying it to the rest of the brain. Despite the numerous studies showing clear effects of individual signals on neurons, data examining the interaction and integration of these signals in single SFO neurons are limited. Previous electrophysiology studies have shown that a subpopulation of angiotensin-responsive neurons in the SFO can also respond to vasopressin (3), another cardiovascular hormone, and another ANG-responsive group also responds to oxytocin (22), a traditionally classified reproductive hormone. When investigating the integration of two different satiety hormones, amylin and leptin were shown to both influence the same subpopulation of SFO neurons (48), whereas ghrelin and amylin affected different neurons (41). There is only one study to our knowledge that shows an interaction between a traditionally classified cardiovascular and a metabolic signal in single SFO neurons. This study published last year by Young et al. (57) showed that selective deletion of angiotensin type 1a receptors (AT1Rs) in the SFO attenuated leptin-induced weight loss in mice by abolishing leptin-induced sympathetic activation of brown adipose tissue thermogenesis. Collectively, these data suggest that circulating signals are integrated in the SFO and may ultimately influence physiological function, emphasizing the need for further investigation of its integrative capacity.
Thus, the present study was undertaken to investigate the interaction between ANG, CCK, and glucose in single SFO neurons, the former being perhaps the best understood cardiovascular signal that influences SFO neurons, and the latter two being hormone and nutrient signals that also influence SFO neurons (1, 30). Using perforated patch-clamp techniques, we have carried out recordings from dissociated SFO neurons to examine whether single SFO neurons respond to both CCK and ANG, hormones traditionally considered to influence separate physiological systems. Second, we have examined whether glycemic/metabolic state influences the responsiveness of SFO neurons to the cardiovascular signal ANG and/or the metabolic signal CCK.
MATERIALS AND METHODS
Ethics approval, animals, and subfornical organ neuron dissociation.
All experimental procedures in this study were carried out in accordance with the ethical criteria established by the Canadian Council on Animal Care, after approval from the Animal Care Committee at Queen’s University (Kingston, ON, Canada); protocol no. 2013-032. Dissociated SFO neurons were prepared from male Sprague-Dawley rats (100–150 g) purchased from Charles River Laboratories (Montreal, QB, Canada). Upon arrival to the Queen’s Animal Care facility, rats were housed in pathogen-free conditions on a reversed 12:12-h light-dark cycle with free access to food and water for a minimum of 4 days before use. The protocol used for dissociation of these cells was adapted from those previously described (10). For each dissociation, three rats were decapitated, and their brains were removed and immediately placed in an oxygenated (95% O2-5% CO2), ice-cold artificial cerebral spinal fluid (aCSF) solution containing (in mM) 124 NaCl, 2.5 KCl, 1.3 MgSO4 hexahydrate, 1.24 KH2PO4, 20 NaHCO3, 10 glucose, and 2.27 CaCl2. First, a block of tissue containing the SFO hanging below the hippocampal commissure was removed from the brain. Then the SFO, which is readily distinguished from the adjacent tissue, was microdissected and placed in a drop of Hibernate-A media (Thermo Fisher Scientific) supplemented with B-27 (GIBCO, Gaithersburg, MD). After the three SFOs were extracted, all were placed in a solution, containing 10–16 mg of papain (lyophilized; Worthington Biochemical, Lakewood, NJ) dissolved in 5 ml of Hibernate-A, for 30 min at 31°C to break down the connective tissues. They were then rinsed twice with the Hibernate-A/B27 solution to remove the papain. The SFOs then underwent three cycles of trituration, which requires using a 1-ml pipette and turbulent flow to break neurons apart. The tube with the solution now containing single dissociated neurons was centrifuged for 8 min at 100 g to form a pellet of neurons at the bottom of the tube. The supernatant was then carefully removed and the pelleted cells were resuspended in a Neurobasal-A solution (Thermo Fisher Scientific, Waltham, MA). This solution was ordered without d-glucose or sodium pyruvate, and then supplemented with B-27, 227 μM sodium pyruvate (Thermo Fisher Scientific), 100 U/ml penicillin-streptomycin (Thermo Fisher Scientific), 0.4 mM l-glutamine (Thermo Fisher Scientific), and glucose (final concentration either 1, 5, or 10 mM glucose, depending on the experiment). The solution was then aliquoted onto 35-mm tissue-culture-treated plastic-bottom Corning dishes (10–12 dishes; Sigma, St. Louis, MO) at a low density to avoid development of synaptic connections. Plates were incubated for 2.5 h to allow for the dissociated neurons to sink and adhere to the bottom of the plate. The incubator (Forma Scientific, Marietta, OH) was maintained at 37°C and 5% CO2 balanced with ambient air. Then, ~2 ml of the previously described supplemented Neurobasal-A solution (containing either 1, 5, or 10 mM glucose) were added to each plate. Plates were maintained in culture for a minimum of 24 h before use, and all recordings were performed within 4 days of the dissociation process.
Perforated-cell current-clamp recordings from dissociated SFO neurons were collected and filtered at 2 kHz using a Multiclamp 700B patch-clamp amplifier (Molecular Devices, Sunnyvale, CA). Analog signals of membrane potential (Vm) and whole cell current data from the amplifier were digitized at 10 kHz using a Cambridge Electronics Design (CED, Cambridge, UK) Micro1401 interface. The software used to control recording parameters and display the current- and voltage-clamp data were Spike2 (version 8.01) (CED) and Signal (version 6.01) (CED). The perforating agent used for all parts of this study was amphotericin B (AMPHB) (Sigma-Aldrich; cat. no. A4888), an antifungal from Streptomyces species. A stock of AMPHB was made on the basis of previously described instructions (27) by adding 20 μl of DMSO per mg of AMPHB and vortexing for 2 min. This solution was aliquoted and stored in the −20°C freezer for a maximum of 48 h. Immediately before recording, ~6–8 μl of this stock was defrosted and added to the internal pipette solution, vortexed for 30 s, and sonicated for 2–5 s (final concentration 400 μg/ml). This AMPHB pipette solution is made routinely every 2–3 h.
The internal pipette solution used for all recordings contained (in mM) 125 potassium gluconate (C6H11KO7), 10 KCl, 2 MgCl2 hexahydrate, 10 HEPES, 1 EGTA, 0.3 CaCl2 dihydrate, and 2 NaATP. KOH was added until the solution reached a pH of 7.3. The solution was then sterilized via a 0.22-μm filter, and the osmolarity was checked (accepted range = 280–300 mosmol/l). The internal solution was then aliquoted and stored at −80°C for a maximum of 3 mo. Recording electrodes made from borosilicate glass (World Precision Instruments, Sarasota, FL) were shaped using a Flaming Brown micropipette puller (P97; Sutter Instruments, Novato, CA) and had a resistance between 3 and 4 MΩ after being filled with the internal solution and placed in the bath. The extracellular bath solution used for recordings contained (in mM) 140 NaCl, 5 KCl, 2 CaCl2 dihydrate, 1 MgCl2 hexahydrate, and 10 HEPES. NaOH was added until the solution reached a pH of 7.3. The glucose concentration of the external was matched to the glucose concentration of the Neurobasal-A solution that cells were incubated in, of either 1, 5, or 10 mM glucose. For the 1- and 5-mM glucose external solutions, mannitol was added to compensate for the difference in osmolarity (9 mM and 5 mM, respectively). The acceptable osmotic range of the external solution was between 280 and 300 mmol/kg. Dissociated SFO neurons were perfused with the external bath solution (heated to 37°C) at an average rate of 1.0 ml/min, using a gravity perfusion system, and fluid was removed from the dish via a vacuum pump. The electrode was dipped into the internal solution for half a second and then back-filled with 3 μl of the AMPHB pipette solution. The pipette tip was lowered onto the selected cell using an MP-225 micromanipulator (Sutter Instruments). Once contact with the cell membrane was made, slight negative pressure was applied to form a GΩ seal. Time was then allowed for AMPHB to form pores in the membrane until the series resistance stabilized (~15–20 min). Only those cells that reached a series resistance of <20 MΩ were used. Cells were identified as neurons if fast activating and inactivating inward sodium currents could be evoked by a series of voltage steps in the voltage-clamp configuration (using the Signal program). Only when individual cells could fire action potentials >60 mV in the current-clamp configuration (either spontaneously or evoked by an injection of current) were they selected for experimentation. A calculated liquid junction potential of −15.7 mV was added to the reported Vm of all neurons recorded.
Vm was obtained for a minimum of 5 min, and after a steady baseline was established, 2–3 ml of 10 nM CCK (sulfated octapeptide, cat. no. 069–03; Phoenix Pharmaceuticals, Burlingame, CA), dissolved in the external solution, were bath applied to cells. Sulfated CCK-8 was selected as it binds to both CCK1 and CCK2 receptors. After washout of CCK and a return to baseline if the cell responded, 2–3 ml of 10 nM ANG II (cat. no. 002–12; Phoenix Pharmaceuticals) were applied to the same cell. A concentration of 10 nM (10−8 M) was selected for both peptides, as it falls within the effective concentration ranges previously described (1, 38). For the first set of experiments, peptides were applied in a bath solution containing 5 mM glucose. For the rest of the experiments, the glucose concentrations were either 1, 5, or 10 mM glucose.
“Response” characterization and data analysis.
Responsiveness of SFO neurons was determined by analyzing the average Vm before, during, and after peptide application in 100-s bins (100 × 10K = one million data points to calculate each mean), using a script in Spike2. The neuron was considered responsive if the change in Vm during or after the application exceeded two times the SD of 100 s of control baseline Vm before the peptide hit the bath, and if it were followed by a return toward baseline Vm following washout of the peptide. In the rare cases where the SD of the baseline Vm was exceedingly large as a result of high-frequency action potential firing in control conditions, a change in firing frequency >75% from the baseline firing frequency was used to classify the neuron as responding to the hormone. The values reported are the changes in Vm, representing the largest 100-s change during or after peptide application. After classification of responses based on these criteria, group response averages are expressed as means ± SE.
χ2 analyses of contingency tables were used to determine whether proportions of responses differed significantly between SFO neurons incubated and recorded in different glucose concentrations. One-way ANOVA tests were used to compare magnitudes of responses or resting membrane potentials (Vr) of neurons before peptide application between the 10, 5, and 1 mM glucose conditions. All analyses were performed using GraphPad Prism 6 software and used a 95% confidence interval.
Subfornical organ neurons integrate metabolic and cardiovascular hormones.
Recent electrophysiology studies have shown that significant proportions of SFO neurons respond to the metabolic signal, CCK (1), or the cardiovascular signal, ANG (26). However, whether these peptides affect different neuronal subpopulations or not has yet been elucidated. To answer this question, we performed perforated-cell current-clamp recordings from dissociated SFO neurons incubated and recorded in 5 mM glucose. After a minimum 100-s stable baseline Vm was established, bath application of CCK (10 nM) was applied via a gravity perfusion system. If the neuron was classified as responsive to CCK (see “Response” characterization and data analysis in materials and methods), sufficient time was allowed for recovery of the Vm, and then ANG II (10 nM) was applied to the same dissociated SFO neuron. Sequential application of each peptide was achieved in a total of 30 dissociated SFO neurons, which were prepared from nine separate dissociations (27 rats total). The average Vr of neurons before hormone application was −66.6 ± 1.0 mV. All neurons showed a return to baseline Vm. In some cases, however, a full recovery after ANG application was not observed, as some of the responses to CCK and/or ANG could each last more than 45 min, and recordings were not maintained long enough to see a full recovery to both.
A subpopulation of SFO neurons respond to both CCK and ANG, while two other groups of neurons respond to either hormone exclusively.
Consistent with previous studies, we found that bath application of CCK on dissociated SFO neurons caused both depolarizing and hyperpolarizing responses (1), whereas ANG caused depolarizing responses only (10, 39). A total of 16/30 neurons tested were classified as responsive, of which 63% depolarized to both CCK and ANG (n = 10/16; CCK mean Vm Δ = 7.2 ± 1.1 mV; ANG mean Vm Δ = 11.0 ± 2.1 mV), as seen in example traces shown in Fig. 1A and summarized in Fig. 1D. Interestingly, all of the neurons that depolarized to CCK also depolarized to ANG, although in contrast, a small proportion of the ANG-responsive neurons did not respond to CCK (Fig. 1B). This ANG-only responsive group of neurons made up 25% of total responding neurons (Fig. 1D) (n = 4/16; mean Vm Δ = 10.4 ± 2.6 mV). In contrast, the neurons that hyperpolarized to CCK (12%) did not respond to ANG (Fig. 1, C and D) (n = 2/16; mean Vm Δ = −7.2 ± 0.1 mV). There was no significant difference between the average Vr of cells that depolarized or hyperpolarized to CCK (−68.0 ± 2.3 mV vs. −66.4 ± 0.5 mV; paired t-test, P = 0.77). These data suggest three separate subpopulations of SFO neurons exist: one that responds to CCK and ANG, one that responds only to ANG, and another that responds only to CCK (Fig. 1D).
Subfornical organ neurons integrate glucose with cardiovascular and metabolic hormones.
These initial recordings demonstrated that in 5 mM glucose <50% of the 30 SFO neurons tested responded to bath application of ANG II (n = 14/30), a significantly smaller proportion of cells when compared with previous findings in the literature that consistently report a responsiveness between 61 and 72%: 61% (26); 63% (10); 66% (38); and 72% (39). A major difference between these studies and ours is that, traditionally, these experiments were performed at 10 mM glucose, whereas the neurons in our experiment were recorded at 5 mM glucose. In these experiments, the use of 5 mM glucose represented an attempt to mimic the normoglycemic circulating environment to which the SFO is exposed in these animals (55), but it evolved to set the stage for the second set of our experiments, which sought to investigate whether glycemic condition could affect the responsiveness of these cells to either a cardiovascular and/or metabolic signal, such as ANG and CCK.
To investigate this, neurons were dissociated in the same way as described above, but instead were incubated at either 1 mM (9 dissociations; 27 rats), 5 mM (14 dissociations; 42 rats), or 10 mM glucose (5 dissociations; 15 rats) for a minimum 24-h period, up to 4 days thereafter. The glucose concentration in the external recording solution was also matched to the concentration at which the cells were incubated. In this study, 5 mM was considered normoglycemic on the basis of the average circulating glucose concentrations of healthy rats (55). This is a very similar range to that present in healthy humans (17). Our lower and higher concentrations of glucose were determined on the basis of previous studies, demonstrating that the SFO responds electrophysiologically to changes in bath glucose concentrations interchanging between 5, 10, and 1 mM glucose (30): normal, extremely high, and low concentrations of blood glucose, respectively (55). Also, as previously mentioned, we chose 10 mM glucose for our hyperglycemic condition, as it is what has been used for recordings in the past. We have chosen to mimic circulating glucose concentrations, as opposed to glucose concentrations in brain areas that are protected by the BBB, which are significantly lower, on the basis of the fact that the SFO lacks a blood-brain barrier and is, therefore, consistently bathed in circulating glucose concentrations.
An increasing glycemic environment increases the responsiveness of SFO neurons to ANG.
Therefore, we next examined the effects of 10 nM ANG on SFO neurons incubated and recorded in a hyperglycemic environment (10 mM glucose) and found that under these conditions, 73% of neurons tested depolarized (n = 8/11; mean Vm Δ = 6.5 ± 0.9 mV), as demonstrated in the example trace Fig. 2A and summarized in Fig. 2D. This proportion is consistent with the aforementioned studies investigating the effects of ANG on SFO neurons recorded at 10 mM glucose (61–72% ANG-responsive SFO neurons) (10, 26, 38, 39). As previously mentioned, neurons incubated and recorded in 5 mM glucose caused 50% of neurons to depolarize in response to bath application of ANG (Fig. 2, B and D) (n = 16/32; mean Vm Δ = 10.6 ± 1.4 mV), and after lowering the glucose to 1 mM, only 33% depolarized (Fig. 2, C and D) (n = 4/12; mean Vm Δ = 6.2 ± 2.2 mV). As demonstrated in Fig. 2D, increasing the glycemic environment of neurons, from 1 → 5 → 10 mM glucose was associated with an increase in the proportion of depolarizing neurons, from 33% → 50% → 73%, to bath application of 10 nM ANG. Statistical analysis of the proportions of neurons responding to ANG at 1, 5, and 10 mM glucose [including the population previously tested in our laboratory at 10 mM glucose (10)] using χ2 analysis of contingency tables showed a significant correlation between glucose concentrations and response profiles of SFO neurons (P < 0.05). A one-way ANOVA was conducted to compare the mean depolarization magnitudes to ANG in each of the glycemic conditions and determined that glucose did not have an effect on response magnitude, P = 0.10. Also, there was no significant difference between the average Vr of cells before ANG application in the 10 mM (−66.3 ± 2.7 mV), 5 mM (−66.5 ± 1.4 mV), or 1 mM glucose (−64.2 ± 2.3 mV) condition; one-way ANOVA, P = 0.71. Altogether, these data provide evidence for a positive relationship between increasing glucose concentration and the percentage of responding neurons to ANG.
A hyperglycemic environment increases the proportion of hyperpolarizing SFO neurons to CCK and decreases the proportion of depolarizing neurons.
Almost identical to the results demonstrated in Ahmed et al. (1), which is the only other electrophysiology experiment investigating the effects of CCK on SFO neurons, our data show that 38% of neurons recorded at 5 mM glucose depolarized in response to bath application of 10 nM CCK (Fig. 3, C and G) (n = 17/45; mean Vm Δ = 8.2 ± 0.9 mV) and that 11% hyperpolarized (Fig. 3, D and G) (n = 5/45; mean Vm Δ = −6.3 ± 0.6 mV), with about half of the neurons not responding to CCK application (Fig. 3G) (n = 23/45 or 51%; mean Vm Δ = 0.7 ± 0.5 mV). Increasing the concentration of glucose to 10 mM to which cells were incubated and recorded caused 19% of the neurons to depolarize (Fig. 3, A and G) (n = 4/21; mean Vm Δ = 8.4 ± 2.5 mV), 29% of neurons to hyperpolarize (Fig. 3, B and G) (n = 6/21; mean Vm Δ = −7.3 ± 1.3 mV), and 52% of neurons not to respond (Fig. 3G) (n = 11/21; mean Vm Δ = −0.3 ± 0.7 mV) to bath application of CCK. After lowering the glucose to 1 mM, 35% of neurons depolarized (Fig. 3, E and G) (n = 6/17, mean Vm Δ = 7.5 ± 0.9 mV), 6% hyperpolarized (Fig. 3, F and G) (n = 1/17, Vm Δ = −13.5 mV), and 59% did not respond (Fig. 3G) (n = 10/17, mean Vm Δ = −1.0 ± 0.8 mV) to bath application of CCK. As demonstrated in Fig. 3G, increasing the glycemic environment of neurons, from 1 → 5 → 10 mM glucose was associated with a decrease in the proportion of depolarizing neurons (35% → 38% → 19%) and an increase in the proportion of hyperpolarizing neurons (6% → 11% → 29%) to bath application of 10 nM CCK, without changing the proportion of nonresponding neurons (59% → 51% → 52%). Statistical analysis of the proportions of neurons responding to CCK at these different glucose concentrations [including the population previously tested in our laboratory at 5 mM glucose (1)] using χ2 analysis of contingency tables showed a significant correlation between glucose concentrations and response profiles of SFO neurons (P < 0.01). One-way ANOVAs were conducted to compare the mean depolarization and hyperpolarization magnitudes to CCK in the 1, 5, and 10 mM glucose conditions and determined that glucose did not have a significant effect on either response magnitude (depolarizations: P = 0.91; hyperpolarizations: P = 0.08). Also, there was no significant difference between the average Vr of cells before CCK application in the 10 mM (−65.3 ± 1.7 mV), 5 mM (−67.9 ± 1.0 mV), or 1 mM glucose (−64.1 ± 2.3 mV) condition; one-way ANOVA, P = 0.16. Altogether, these data suggest that a hyperglycemic environment modifies the responsiveness of SFO neurons to CCK.
Interestingly, a hyperglycemic environment appears to have an opposite effect on these cardiovascular and metabolic regulating hormones, by increasing the proportion of depolarizing neurons to ANG and decreasing the proportion of depolarizing neurons to CCK, while increasing the proportion of hyperpolarizing neurons to CCK (Fig. 4).
The present study provides compelling evidence for the integrative capacity of single SFO neurons. Our results demonstrate that the population of SFO neurons that depolarize in response to the metabolic hormone, CCK, also depolarize to the cardiovascular hormone, ANG. In addition, the responses of neurons to both of these signals are altered in hyperglycemic states. To our knowledge, this study is the first to demonstrate direct excitation of the same SFO neuron by both a cardiovascular and metabolic hormone, as well as the ability of circulating glucose concentrations to modulate the way SFO neurons respond to hormonal stimuli.
We first investigated whether neurons that responded to the metabolic signal, CCK, could also respond to the cardiovascular signal, ANG. This was based on previous studies showing that the SFO expresses receptors for, and single neurons respond to, a number of metabolic and cardiovascular signals (for review, see Ref. 32). As stimulation of the SFO has been shown to increase both fluid (37, 47) and food (50) intake, we hypothesized that the neurons that responded to these signals may represent separate subpopulations within the SFO with distinct electrical circuits controlling these physiological behaviors. Interestingly, only 25% of the total number of responding neurons depolarized to ANG exclusively, and only 12% exclusively responded to CCK, these neurons being hyperpolarized by this peptide. To our surprise, all of the neurons that underwent a depolarization of the Vm in response to bath application of CCK were also depolarized by ANG. This group of neurons made up 63% of the total responding neurons. These results were consistent with previous electrophysiology reports showing that CCK causes both depolarizing and hyperpolarizing responses in the SFO, the majority being depolarizing responses (1), as opposed to ANG, which consistently causes a depolarization of SFO neurons (10, 26, 38, 39). One possible explanation for why the SFO exhibits heterogeneous responses to CCK application is that it expresses both CCK1 and CCK2 receptors (1). As these are both G protein-coupled receptors, each receptor could be mediating one type of response by affecting different downstream ion channels via second messenger signaling pathways. Resting membrane potentials before peptide application were not significantly different between depolarizations and hyperpolarizations and, therefore, do not likely explain the heterogeneous responses to CCK.
Unlike CCK, the actions of ANG at the SFO have been thoroughly investigated in vivo. Microinjection of ANG into the SFO elicits drinking and pressor responses in rats (18, 29) (for review, see Ref. 13). Despite CCK’s traditional role in controlling food intake and digestion, the extremely large overlap of neurons that depolarize to both CCK and ANG raises the possibility that CCK actions in the SFO may also influence fluid intake or cardiovascular control. Interestingly, in accordance with this theory, intravenous administration of CCK-33 in rats has been shown to increase systolic and diastolic blood pressure (56). A more recent study showed that intravenous administration of CCK-8 in anesthetized rats caused a triphasic response of blood pressure characterized by an immediate fall, then a brief increase after 15 s, followed by a maximum decrease at 3 min (23). The authors suggested that the decreases in blood pressure were mediated via CCK1R activation on vagal afferents, whereas the pressor responses were likely mediated by CCK1R activation in the central nervous system (23). These findings, in addition to our data showing that CCK depolarizes the majority of ANG-responsive neurons in the SFO that have well-described cardiovascular effects when activated, warrant future investigation of CCK’s potential effects on blood pressure via signaling at the SFO.
The next set of our experiments evolved from the observation that <50% of SFO neurons responded to bath application of ANG, compared with the literature, which reproducibly reports a significantly higher responsiveness, between 61 and 72%, in both slice and dissociated neurons (10, 26, 38, 39). One major difference between our protocol and those previously published was that we used 5 mM glucose during incubation and recording of neurons, as opposed to the traditional 10 mM glucose that has been used for decades. Thus, our second set of experiments sought to establish whether circulating glucose concentration could affect the responsiveness of SFO neurons to ANG or CCK in vitro; two hormones traditionally involved in the regulation of the cardiovascular system and energy metabolism, respectively.
Our study revealed a positive correlation between responsiveness of SFO neurons to ANG and increasing glucose concentration, where the percentage of depolarizing responses increased from 33% → 50% → 73% when glucose was increased from 1 mM → 5 mM → 10 mM. It is possible that the high-glucose environment during the minimum 24-h incubation time caused upregulation of AT1R expression—the most abundant angiotensin receptor in the SFO, which has been shown to mediate ANG’s cardiovascular effects at this site (24). Increased transcription and translation of AT1Rs could increase the total percentage of responding neurons by recruiting neurons that normally would not express enough AT1Rs to elicit a response in normal glucose conditions, into the ANG-responsive population of SFO neurons. Using quantitative RT-PCR (qPCR) techniques, Wang et al. (54) demonstrated that incubation of rat glomerular mesangial cells in a high-glucose medium increased the expression of AT1R mRNA by 1.5 times when compared with the normal glucose condition, in as little as 24 h. On the basis of their results from various qPCR and Western blot experiments paired with the use of blockers, it was suggested that the hyperglycemic environment caused overproduction of reactive oxygen species that stimulated p102 synthesis, a transcription activator, which, in turn, upregulated AT1R expression in these cells (54). Similar studies using the same techniques also showed a dose-dependent positive correlation between increasing glucose concentrations and AT1R gene and protein expression in an insulin-producing pancreatic β-cell line (25) and an immortalized proximal tubule epithelial cell line from rabbit kidneys (5).
By extrapolating our results and this concept to the human condition, one could theorize that the increased responsiveness of SFO neurons to ANG seen in hyperglycemic conditions in our experiment could contribute to an underlying neuronal mechanism responsible for the comorbidities of hyperglycemia and hypertension experienced by individuals suffering from obesity.
As glucose concentration represents energy status in an organism, that is, a fasted, fed, or obese state, it seemed logical that it could influence the actions of a metabolic hormone in a structure that has been shown to influence food intake. Investigation of the effects of glycemic condition on CCK responsiveness in our study revealed that, in contrast to what was observed with ANG, a negative correlation between depolarization of SFO neurons to CCK and increasing glucose concentration was seen (5 mM, 38% → 10 mM, 19%). This was accompanied by an increase in the percentage of hyperpolarizing responses by more than twofold from the normal to hyperglycemic condition, from 11% at 5 mM glucose → 29% at 10 mM glucose. It is possible that SFO neurons that hyperpolarize to CCK may be excitatory neurons projecting to areas of the hypothalamus involved in the control of food intake, such as the LH and AgRP/NPY-producing neurons in the arcuate nucleus (16, 53). If so, CCK’s increased inhibition to these hypothalamic areas normally involved in stimulation of food intake, via the SFO, could, therefore, decrease the desire to eat, producing a larger satiating effect in hyperglycemic conditions.
It is unlikely that this shift from depolarizing to hyperpolarizing responses to CCK applied in hyperglycemic conditions is attributable to glucose causing changes in the Vr, which has been shown in the canonical β-cell model (for review, see Ref. 7), as there was no significant difference between the baseline Vr of neurons in each glucose concentration in our experiment, nor between neurons that depolarized or hyperpolarized within glucose groups. What may explain our findings, similar to what was previously discussed with ANG, is that glucose may modify expression of the CCK receptors. Ahmed et al. (1) used CCK1 and CCK2 receptor blockers with peripheral administration of CCK and measured expression of c-Fos and p-ERK in the SFO to determine that CCK2 receptors are the dominant excitatory receptor in the rat SFO, and also used PCR to show that the SFO expresses both CCK1 and CCK2 receptors. These data suggest that the CCK2 receptors may mediate the depolarizing responses seen electrophysiologically, while CCK1 receptors may mediate the hyperpolarizing responses. To our knowledge, studies investigating CCK receptor expression changes after exposure to varying levels of glucose, even in nonneuronal tissues of the body, have not been performed. What has been studied, though, is the change in the expression of various metabolic and cardiovascular hormone receptors in the SFO after fluid and food deprivation (20). This study provides evidence for the plasticity of metabolic and cardiovascular receptors in the SFO, maintaining the ability of these receptors to adapt to changes in cardiovascular and metabolic status.
Perspectives and Significance
In conclusion, our findings demonstrate that a subpopulation of SFO neurons respond to both ANG and CCK, hormones traditionally considered to influence cardiovascular and metabolic regulation, respectively, and that glycemic state influences the responsiveness of SFO neurons to both of these signals. As the physiological role of CCK in the SFO has not yet been investigated, our demonstration that a large percentage of SFO neurons depolarize to both CCK and ANG may suggest a role for CCK in increasing fluid intake and/or blood pressure via activation of the SFO, complementary to the well-understood actions of ANG. Additionally, our results show that increasing glucose concentrations from 1 to 5 and 10 mM are associated with increased responsiveness of SFO neurons to ANG, potentially revealing a neuronal mechanism contributing to the comorbidities of hyperglycemia and hypertension presented in obesity. In contrast, a hyperglycemic environment causes an increase in the proportion of SFO neurons that hyperpolarize to CCK, which may demonstrate an attempt to decrease food intake in an energy-fulfilled state. These results provide compelling evidence for the ability of a metabolic signal, glucose, to modify the way SFO neurons respond to both a metabolic and a cardiovascular hormone. Altogether, our findings highlight the SFO as an integrative site where various metabolic and cardiovascular signals may interact to influence autonomic function.
This work was supported by Canadian Institutes of Health Research Grant MOP12192 (to A. V. Ferguson) and an Ontario Graduate Scholarship (to N. M. Cancelliere).
No conflicts of interest, financial or otherwise, are declared by the authors.
N.M.C. and A.V.F. conception and design of research; N.M.C. performed experiments; N.M.C. analyzed data; N.M.C. and A.V.F. interpreted results of experiments; N.M.C. prepared figures; N.M.C. drafted manuscript; N.M.C. and A.V.F. edited and revised manuscript; N.M.C. and A.V.F. approved final version of manuscript; N.M.C. and A.V.F. agree to be accountable for all aspects of this work.
- Copyright © 2017 the American Physiological Society