Repeated Pharmacogenetic Catecholamine Neuron Activation in the Ventrolateral Medulla Attenuates Subsequent Glucoregulatory Responses

Glucose is the brain’s obligatory metabolic fuel. Because little glucose is stored in the brain, glucose from peripheral sources must be supplied to the brain by the blood. To protect its glucose supply, protective and corrective responses to glucose deficit, known as counterregulatory responses (CRRs), including increased food intake, corticosterone, and adrenal medullary and glucagon secretion, are elicited by acute glucoprivation (1). However, prior (antecedent) episodes of glucoprivation results in attenuation of these responses, leading to a potentially lethal condition known as hypoglycemia-associated autonomic failure (HAAF) (2,3). HAAF, which can result following inadvertent mismatch of glucose intake/production and insulin dose, is a potentially lethal threat to patients on insulin therapy. Currently, the neuropathologic mechanisms of HAAF are poorly understood.

Previous results strongly support the hypothesis that activation of catecholamine (CA) neurons in the C1 and rostral A1 cell groups in the ventrolateral medulla (VLM) is necessary and sufficient to elicit key CRRs, including increased food intake, corticosterone, and adrenal medullary secretion that preserve, extend, and restore glucose availability (for review, see Ritter [4]). Selective immunotoxin lesion of spinally projecting CA neurons abolishes the adrenal medullary blood glucose response to glucoprivation, while selective lesion of rostrally projecting CA neurons that pass through or terminate in the paraventricular hypothalamic area abolishes glucoprivic feeding and corticosterone responses (5–7).

Although these data contribute to a body of evidence supporting a role for VLM CA neurons in elicitation of certain CRRs, it remains unclear whether dysfunction of these neurons might contribute to HAAF. One obstacle in addressing this question has been that insulin or antiglycolytic drugs such as 2-deoxy-d-glucose (2DG), given systemically to investigate this question, produce glucoprivation and other metabolic alterations throughout the body and brain. Hence, it is difficult to discern whether components of HAAF result from antecedent activation and dysfunction of any specific neuronal populations (e.g., VLM CA neurons) or from more dispersed or distal effects of glucoprivation.

Recently, we applied pharmacogenetic tools (8) to overcome some of these problems. This approach enabled us to test the involvement of specific CA neurons in elicitation of specific CRRs in the absence of glucoprivation (9). We injected a Cre-dependent designer receptor exclusively activated by designer drugs (DREADD) viral construct into the VLM A1 and C1 overlap region (A1/C1) of tyrosine hydroxylase (Th)-Cre⁺ transgenic rats to selectively transfect hM3D(Gq) in VLM CA neurons at the injection site. We found that in transfected rats, systemic administration of the DREADD receptor agonist clozapine-N-oxide (CNO) triggered CRRs that were similar in magnitude and time course to those produced by systemic 2DG-induced glucoprivation. The CNO-elicited responses were specific to transfection site. Moreover, CNO increased c-Fos expression selectively in CA neurons in Th-Cre⁺ rats and was not elicited by CNO in nontransfected rats.

In the current study, we used a pharmacogenetic approach to activate a subpopulation of A1/C1 CA neurons shown previously to increase food intake and corticosterone secretion (9). Increased blood glucose and epinephrine secretion, which require activation of more rostral VLM CA neurons (9,10), were not examined in this experiment. Specifically, we injected a Cre-dependent DREADD construct bilaterally into the VLM A1/C1 overlap region in Th-Cre⁺ transgenic rats, resulting in selective expression of the excitatory DREADD hM3D(Gq) in these neurons. We then selectively and repeatedly activated the transfected neurons using systemic injections of the DREADD receptor agonist CNO in the absence of glucoprivation. The effect of this repeated antecedent activation on the feeding, corticosterone, and c-Fos responses to subsequent glucoprivation was then examined to determine whether the responses were reduced, as they are after repeated glucoprivic episodes (11). Results indicate that prior repeated pharmacogenetic activation of selectively transfected A1/C1 CA neurons significantly reduced their responses to subsequent pharmacogenetic activation of feeding and corticosterone responses. These results are consistent with the possibility that a failure to activate these neurons may contribute to onset of HAAF, which occurs in the aftermath of severe or repeated hypoglycemia.

Male transgenic Long-Evans rats expressing Cre recombinase under the control of the Th promoter (Long-Evans-Tg [Th-Cre] 3.1) and their wild-type non-Cre littermates (Th-Cre⁺ and Th-Cre⁻ rats, respectively) were used in the current study. These rats were bred in our vivarium from breeding stock generously provided by Dr. Karl Deisseroth (Stanford University, Stanford, CA) and were 3 months old at the beginning of experiments. Genotyping was performed from an ear punch at 3 weeks of age using PCR (9). Rats were maintained on a 12-h light/12-h dark cycle (lights on at 7 a.m.) with ad libitum access to pelleted rodent food (catalog #5001; LabDiet) and tap water. All experimental procedures conformed to National Institutes of Health guidelines and were approved by the Washington State University Institutional Animal Care and Use Committee.

For intracranial injection of adeno-associated virus (AAV), rats were anesthetized using 1.0 mL/kg of ketamine/xylazine/acepromazine cocktail (50 mg/kg ketamine HCl, Fort Dodge Animal Health; 5.0 mg/kg xylazine, Vedco Inc.; and 1.0 mg/kg acepromazine, Vedco Inc.), placed in a stereotaxic device, and injected with AAV (serotype 2) containing a Cre-dependent doubly floxed inverted open reading frame encoding a reporter gene (mCherry) under a human synapasin I promotor, AAV2-DIO-hSyn-hM3D(Gq)-mCherry (AAV-Gq; 1.6 × 10¹² particles/mL) (Addgene) (8). Following vector transfection, the DREADD receptor is selectively expressed only by Cre-expressing CA neurons proximal to the injection site. Injections (200 nL/site) were delivered bilaterally into the A1/C1 through a pulled glass capillary pipette (30-µm tip diameter) using a Picospritzer. Pipettes entered the brain dorsolateral to the targeted VLM site and were driven ventromedially at a 14° angle to avoid inadvertent damage to or transfection of CA neurons in the dorsomedial medulla by potential diffusion of the viral construct along the pipette tract. The following coordinates were used: 13.75 mm caudal to bregma, 4.0–4.1 mm lateral to midline, and 8.7–8.9 mm ventral to the skull surface (12).

Our previous work has shown that mCherry expression in CA neurons is maximal and stable between 5 and 10 weeks after the virus injection (9,13). Therefore, all testing was conducted within this time interval. Th-Cre⁻ rats with AAV-Gq injection into VLM showed no mCherry expression, verifying that expression of the construct was Cre-dependent.

For immunohistochemistry, rats were euthanized by deep isoflurane-induced anesthesia (Halocarbon Products Corporation). Just prior to cessation of the heartbeat, rats were perfused transcardially with PBS (pH 7.4), followed by freshly made 4% formaldehyde/PBS solution. Brains were rapidly removed and placed in 4% formaldehyde/PBS (5 h), followed by immersion overnight in 25% sucrose/PBS at room temperature. Brains were sectioned coronally at 40-µm thickness using a cryostat. Sections (four serial sets) were collected for analysis of distribution and selectivity of the transfected AAV-Gq using antibodies against mCherry and CA biosynthetic enzyme, dopamine-β-hydroxylase (DBH), and of neuronal activation using c-Fos antibody. For double immunofluorescence staining, sections were incubated with primary antibodies at 4°C (2.5 days) in 10% normal horse serum/PBS, washed, and then incubated in secondary antibodies (4 h) (14,15). The following primary antibodies were used: rabbit anti-DsRed (to detect mCherry; Clontech Laboratories), mouse anti-DBH (Millipore), and goat anti–c-Fos (Santa Cruz Biotechnology). Secondary antibodies were donkey anti-mouse, anti-rabbit, or anti-goat, conjugated with Alexa 488, Cy3, or Alexa 647 (1:500 dilution in 1% normal horse serum/PBS; Jackson ImmunoResearch Laboratories). Sections were mounted and coverslipped with ProLong Gold medium (Thermo Fisher Scientific), examined, and photographed using a Zeiss epifluorescent microscope.

Hindbrain CA cell groups are defined as described in The Rat Brain in Stereotaxic Coordinates (12). Cells of groups C1 and A1 are continuously distributed along the rostrocaudal extent of the VLM. We refer to the overlap of caudal C1 with rostral A1 as A1/C1 (14.1–13.4 mm caudal to bregma), the medial portion of C1 as C1m (13.3–12.5 mm caudal to bregma), and rostral C1 as C1r (12.4–11.8 mm caudal to bregma). The number of cells with positive immunostaining of DBH, c-Fos, and mCherry were counted from three sections in A1/C1 with most mCherry expression and averaged.

All feeding tests were conducted in the animals’ home cages between 5 and 10 weeks after AAV injection using standard pelleted rodent diet. At 5 weeks after AAV injection, rats were screened in a CNO-induced feeding test to determine the efficacy of DREADD transfection. Rats with a feeding response >2.0 g during the 4-h test were randomly divided into different groups for further testing. Each rat was tested at 6 and 9 weeks after transfection, as follows. At 6 weeks, each rat was injected once daily for 4 days with CNO (1 mg/kg i.p.) or 2DG (200 mg/kg s.c), doses shown previously to produce similar amounts of food intake (9), or with saline (Sal) as control (0.9% i.p. or s.c.). On days 1–4, food was removed 2 h before and returned 4 h after injection. On day 5, each rat was tested with either the same or a different drug, such that the resulting groups were as follows: repeated (r)CNO-CNO, r2DG-CNO, rSal-2DG, rCNO-2DG, and r2DG-2DG. For tests on day 5, food was removed 2 h prior to the injection and then returned immediately following the injection. Intake was measured 2 and 4 h after the injection. Nine weeks after transfection, rats were randomly assigned to different treatment groups and tested a second time. At the end of the experiment, specificity and efficacy of DREADD transfection in A1/C1 were confirmed by immunohistochemical analysis of mCherry and DBH. In addition, to assure that CNO effects were dependent on activation of transfected CA neurons, CNO was tested in Th-Cre⁻ rats injected in the VLM with the same DREADD construct and in nontransfected Th-Cre⁺ rats. Neither of these control treatments increased feeding.

In this experiment, rats were screened for Sal- and CNO-induced feeding 5 weeks after bilateral AAV-Gq injection into A1/C1, as in experiment 1, then implanted with jugular catheters, and adapted to the blood collection protocol. Rats were injected once daily (at 7 and 10 weeks after AAV injection) with Sal, CNO (1 mg/kg i.p.), or 2DG (200 mg/kg s.c.) for 4 consecutive days in the absence of food. On the 5th day, blood was collected by remote withdrawal from catheters before and after the injection. Samples were centrifuged, and the serum was stored at −80°C. Corticosterone levels were determined from the serum using ELISA kits (IB79175; IBL America). At the conclusion of the experiment, rats were prepared for immunohistochemical analysis of mCherry and DBH, as described above, to verify placement, specificity, and efficacy of DREADD transfection.

In this experiment, rats were screened for Sal- and CNO-induced feeding 5 weeks after bilateral AAV-Gq injection into A1/C1, as in the above experiments. Rats were then randomly divided into three groups and injected once daily for 5 days with Sal (i.p.), CNO, or 2DG (at 6 and 9 weeks after AAV injection). To measure the effects of repeated injections on metabolic parameters and locomotor activity, rats were housed singly in Promethion cages designed for high-definition continuous respirometry measurements (Sable Systems International). The experimental room was maintained at 22.4 ± 1°C, 19–20% humidity, and on a 12-h light, 12-h dark cycle (light period 7 a.m. to 7 p.m.). Food and tap water were available ad libitum, except during the test periods. Rats were acclimatized to the metabolic cages for 3 days before the test. Each group was injected daily at 11 a.m. with one of the following: Sal (i.p.), CNO (1 mg/kg i.p.), or 2DG (250 mg/kg s.c.). Food was removed 2 h before and returned 4 h after each injection. Activity, measured as number of beam breaks in the horizontal (x and y) and vertical (z) planes, was recorded every second. For metabolic measurements, carbon dioxide generated (VCO₂) and oxygen consumed (VO₂) were recorded every 5 min for each rat. Respiratory quotient (RQ) was calculated as the ratio of VCO₂ over VO₂. Energy expenditure (EE) was calculated as EE = (3.815 + 1.232 × RQ) × VO₂. Averages of each parameter per 30-min bin were pooled hourly for statistical comparison. Measurements taken during the period beginning 1 h before until 1 h after the injection, while the experimenter was present in the testing room and animals were being handled, were discarded.

Two weeks after metabolic measurements, rats were then divided into groups that received a daily injection of Sal (i.p.) or CNO (1.0 mg/kg i.p.) in the absence of food each day for 5 days, as described above. On day 5, rats were euthanized and perfused 2 h after the injection, and brain tissue was collected for immunostaining of c-Fos, mCherry, and DBH.

All results are presented as mean ± SEM. One-way or two-way repeated-measures ANOVA was used for statistical analysis. After significance was determined, multiple comparisons between individual groups were tested by a post hoc Student-Newman-Keuls test. P < 0.05 was considered to be statistically significant.

The data sets generated and analyzed during this study are available from the corresponding author upon reasonable request.

As shown in Fig. 1, a single CNO injection in bilaterally transfected A1/C1^AAV-Gq rats significantly increased 2- and 4-h feeding compared with Sal-injected rats (P values <0.001). Single injection of 2DG following repeated Sal injections produced a feeding response that was similar in magnitude to that produced by CNO injection. Daily injection of CNO for 4 days in the absence of food attenuated CNO-induced feeding on day 5 (P values <0.001, rCNO-CNO vs. CNO). Similarly, four daily 2DG injections reduced 2DG-induced feeding on day 5 (P values <0.01, r2DG-2DG vs. rSal-2DG). Furthermore, repeated 2DG injections on days 1–4 in the absence of food reduced CNO-induced feeding on day 5 (P values <0.001, r2DG-CNO vs. CNO). However, repeated CNO injections did not reduce subsequent 2DG-induced feeding (P values >0.9, rCNO-2DG vs. rSal-2DG).

Subsequently, all rats were euthanized, and viral transfection was confirmed by immunochemical staining for DBH and mCherry in the hindbrain. A typical example is shown in Fig. 2. Transfected neurons were localized at the A1/C1 injection site (Fig. 2B), where nearly half (47%) of the CA neurons, but no non-CA neurons, in A1/C1 were transfected (Fig. 2C), a number similar to that reported in our previous study (9).

As shown in Fig. 3, corticosterone concentrations in samples collected on day 5 from A1/C1^AAV-Gq rats injected once daily with CNO on days 1–4 (in the absence of food) were reduced in rCNO-CNO group compared with the response in rSal-CNO group (P values <0.05 at 1 h and 2 h after the injection), revealing a pattern of corticosterone responses similar to the feeding response. In contrast, neither single nor repeated CNO injections in rats transfected at this A1/C1 site increased plasma glucose, glucagon, or epinephrine levels (data not shown), as we also observed previously in our CNO mapping of the VLM (9). Repeated CNO or 2DG injection reduced corticosterone concentrations in rCNO-2DG and r2DG-2DG groups compared with rSal-2DG group (P values <0.01 at 2 h).

Before catheter-implantation surgery, these A1/C1^AAV-Gq rats were tested for food intake after Sal and CNO injection. Food intake was significantly enhanced by CNO compared with Sal treatment at 2 h (0.6 ± 0.2 g vs. 1.7 ± 0.2 g; P < 0.001) and 4 h (0.8 ± 0.2 g vs. 3.5 ± 0.2 g; P < 0.001), respectively. At the end of the experiment 2, viral transfection was further confirmed by double staining of DBH and mCherry. As in experiment 1, nearly half (47%) of the A1/C1 CA neurons near the injection site were transfected.

In the absence of food, a single injection of CNO or 2DG decreased RQ in A1/C1^AAV-Gq rats at 2–4 h after the injection (P values <0.05) (Fig. 4), indicating increased utilization of fat and decreased utilization of carbohydrate on day 1. However, these decreases in RQ were completely blocked on day 5 after repeated daily injection of either CNO or 2DG on the 4 previous days. CNO and 2DG had no effects on EE and locomotor activity (Fig. 4).

After a repeated Sal injection, a single injection of CNO enhanced c-Fos expression in the A1/C1 region, where CA neurons were transfected by injection of AAV-Gq (Fig. 5). After repeated CNO injections for 4 consecutive days, the total c-Fos and c-Fos/mCherry cell numbers in A1/C1 were dramatically decreased in rCNO-CNO rats on day 5 compared with rSal-CNO rats (P values <0.001).

As shown in Fig. 6A, food intake was significantly enhanced by CNO, but not Sal treatment, at 2 h and 4 h (P values <0.001), respectively. As in other experiments cited in this study, nearly half (46%) of the A1/C1 CA neurons near the injection site were transfected, as determined by double staining of DBH and mCherry (Fig. 6B).

It is well known that prior severe or recurrent glucose deficit attenuates CRRs to a subsequent glucoprivic episode, comprising the pathophysiological condition known as HAAF (2,3). In the current study, we show that 4 days of repeated pharmacogenetic activation of A1/C1 CA neurons in DREADD-transfected Th-Cre⁺ rats reduced subsequent counterregulatory feeding, corticosterone secretion, RQ, and c-Fos responses to a subsequent CNO injection. The effects of repeated CNO treatment on these CRRs were similar to those produced by repeated subcutaneous 2DG injections but occurred in the absence of prior glucoprivation. These results are consistent with our previous work demonstrating the essential contribution of VLM CA neurons in the elicitation of key CRRs (for review, see Ritter [4]). However, the fact that glucoprivation itself is not required for CRR attenuation suggests that prior intense activation of CA neurons is a fundamental cause of the subsequent impairment of feeding and corticosterone responses.

Food intake and corticosterone secretion are not traditionally included in the clinical definition of HAAF, which is defined on the basis of autonomic deficits in glucoregulatory functions (2). However, it is clear that both food intake and corticosterone secretion are critical CRRs (4). If glucose stores are depleted, food intake is the sole means of restoring euglycemia, and failure to detect and/or respond to feeding cues allows glucose deficit to persist and worsen. Corticosterone secretion is also important, as it rapidly elevates fatty acid mobilization and utilization, thus conserving available glucose for use by the brain while supplying an alternative energy substrate for peripheral use. Cleary, both food intake and corticosterone secretion are protective and restorative CCRs that support maintenance of euglycemia.

Several lines of work have demonstrated the importance of CA neurons for elicitation of increased feeding, a critical CRR. Glucoprivation activates hindbrain CA neurons and increases c-Fos expression predominantly in the VLM CA neurons (5). Selective immunotoxin lesion of CA neurons that project rostrally abolishes CRRs, including food intake, while lesion of spinally projecting CA neurons eliminates blood glucose and plasma epinephrine responses to glucose deficit (6,7). Concurrent silencing of cotransmitter genes dbh and npy in the A1/C1 area of the VLM also suppresses glucoprivic feeding (16). Moreover, early work revealed that acute 2DG activates the CA synthetic enzyme Th and that repeated glucoprivation diminishes Th activity in forebrain terminals (17). These and other findings are consistent with the hypothesis that detrimental effects of repeated VLM CA neuron activation are fundamental to development of the severe glucose deficits resulting in HAAF.

It is important to note that repeated 2DG reduced the feeding response to both subsequent 2DG, as expected from previous results, and to systemic CNO injections. However, repeated CNO reduced the response to subsequent CNO, but not to 2DG (shown in Fig. 1). We attribute this seeming disparity to the fact that the DREADD construct transfected only 47% of A1/C1 CA neurons specifically at the injection site, while systemic 2DG likely activated the transfected and additional nontransfected A1/C1 CA neurons that are involved in these glucoregulatory responses.

As noted above, corticosterone plays a critical role in control of peripheral metabolism, including the response to glucose deficit, in part by increasing fat metabolism. A shift from glucose utilization to fat oxidation is a hallmark of the integrated response to glucose deficit. Accordingly, we found that RQ was reduced in both 2DG- and CNO-injected rats on day 1 of CNO injection, and, importantly, this response was abolished after repeated injection of either 2DG or CNO. These changes were not attributable to changes in EE or activity levels, both of which decreased during the test in each group but did not differ between groups and were similar on days 1 and 5.

Although corticosterone is recognized for its critical role in control of peripheral metabolism during glucopenia, early data also strongly implicated increased glucocorticoid levels as a cause of deficient counterregulation after antecedent hypoglycemia in nondiabetic experimental animals and in humans. For example, CRR elicitation is impaired by prior administration of the synthetic glucocorticoid dexamethasone in rats (18). In addition, attenuation of CRRs by prior glucoprivation is diminished in adrenalectomized humans (19) and rats (18).

Previously, we showed that immunotoxin lesion of rostrally projecting CA neurons profoundly impaired glucoprivation-induced corticosterone secretion and induction of corticotropin-releasing hormone (Crh) heteronuclear RNA and c-fos mRNA in the paraventricular hypothalamus (PVH), without impairing basal Crh mRNA expression, circadian corticosterone release, or the corticosterone response to swim stress (7). Thus, particular CA projections are required for the corticosterone response to glucoprivation but are dispensable for responses to certain other stressors. However, because CRH secretion is increased by diverse stressors activated via diverse neural inputs to the paraventricular nucleus, corticosterone-induced impairment of CA function may not be limited to feedback from glucoprivic stress. For example, immobilization stress also increases corticosterone secretion and reduces CA synthesis and release both at rest and during stress (20), as well as reducing glucoprivic feeding (17). These observations, and the fact that neurons of VLM CA subgroups that project to hypothalamus express glucocorticoid receptors (21), raise the possibility that endogenous glucocorticoids contribute to development of HAAF by suppression of VLM CA neurons that control their release in response to glucose deficit.

The reciprocity between A1/C1 CA neurons and corticosterone is intriguing, and the physiology of their interaction may be critical for control of the CRRs mediated by VLM CA neurons. Our finding that CNO-induced activation of selectively transfected A1/C1 neurons elicits corticosterone secretion in the absence of glucoprivation strongly supports the hypothesis that the neural activation of PVH CRH neurons via CA neurons elicits and is sufficient for activation of the corticosterone response. Finally, immunotoxin lesion of this CA innervation profoundly impaired glucoprivation-induced corticosterone secretion and induction of Crh heteronuclear RNA and c-fos mRNA in the PVH, without impairing basal Crh mRNA expression, circadian corticosterone release, or the corticosterone response to swim stress (7). Thus, particular A1/C1 projections are required for the corticosterone response to glucoprivation but are dispensable for responses to certain other stressors. These findings alert us to the possibility that elevation of circulating corticosterone by nonglucoprivic stressors may, in some cases, contribute to development of HAAF. Further work examining these possible interactions will be necessary.

In conclusion, results of this study reveal that antecedent activation of A1/C1 CA neurons results in reduced subsequent responsiveness of these neurons to glucoprivation, as indicated by their reduced glucoprivation-induced c-Fos expression. Reduced CA neuron responsiveness is accompanied by attenuated feeding and corticosterone secretion, which typically are elicited by glucoprivic activation of A1/C1 neurons. These effects of antecedent pharmacogenetic CA neuron activation mimic important components of HAAF but are produced in the absence of actual glucoprivation. The ability to selectively activate discrete portions of the neural circuitry that participate in CRR using pharmacogenetics provides a much-needed new opportunity to dissect the circuitry and neuropathologic mechanisms underlying this life-threatening condition.

Acknowledgments. The authors thank Dr. R.C. Ritter (Washington State University, Pullman, WA) for helpful discussions.

Funding. This study was supported by the National Institutes of Health (grant R01-DK114187 to S.R.) and American Diabetes Association (grant 118IBS156 to S.R.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.-J.L. and S.R. designed the experiments. A.-J.L. and Q.W. performed the experiments. A.-J.L. and S.R. analyzed the data and wrote the manuscript. All authors revised the manuscript and approved the final version. A.-J.L. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.