Hypothalamic Prostaglandins Facilitate Recovery From Severe Hypoglycemia but Exacerbate Recurrent Hypoglycemia in Mice

Abstract

The hypothalamus monitors blood glucose levels and regulates glucose production in the liver. In response to hypoglycemia, glucose-inhibited (GI) neurons trigger counterregulatory responses (CRRs), which stimulate the release of glucagon, epinephrine, and cortisol to elevate blood glucose. Recurrent hypoglycemia (RH), however, reduces the effectiveness of these CRRs. This study examined the role of hypothalamic prostaglandins in glucose recovery during acute hypoglycemia and RH. Imaging mass spectrometry and liquid chromatography/mass spectrometry showed phospholipid and prostaglandin levels in the hypothalamus of C57BL mice were changed after insulin or 2-deoxy-glucose administration. Ibuprofen, a nonsteroidal anti-inflammatory drug, was infused into the ventromedial hypothalamus (VMH) to analyze its effect on glucose production during hypoglycemia, revealing that prostaglandin inhibition decreased glucagon secretion. Additionally, RH-treated mice decreased glucagon release and glucose production during hypoglycemia. Inhibiting prostaglandin production via shRNA against cytosolic phospholipase A2 (cPLA2) in the hypothalamus restored CRRs diminished by RH via increasing glucagon sensitivity. These findings suggest that hypothalamic prostaglandins play a critical role in glucose recovery from acute hypoglycemia by activating VMH neurons and are also crucial for the attenuation of CRRs during RH.

Article Highlights

• Insulin decreases arachidonic acid-containing phospholipids and increases prostaglandin production in the hypothalamus.

• Prostaglandin in the hypothalamus increases glucagon secretion and glucose production during hypoglycemia.

• Recurrent hypoglycemia decreases glucagon secretion and glucose production during hypoglycemia.

• Prostaglandin in the hypothalamus during recurrent hypoglycemia decreases glucagon sensitivity and glucose production.

Graphical Abstract

Introduction

Hypoglycemia, a common adverse effect of glucose-lowering medications, can lead to severe consequences, including lightheadedness, fainting, and even death. Continuous glucose monitoring systems have revealed that hypoglycemic events occur more frequently than previously thought (1). Recurrent hypoglycemia (RH) is defined as experiencing multiple episodes of hypoglycemia, which alters the body’s response to hypoglycemia and increases the risk of severe hypoglycemia (2). Thus, hypoglycemia-related deaths worldwide are assumed to be 4.49 per 1,000 total diabetes deaths (3). Additionally, RH increases the risk of dementia (4). The increasing prevalence of hypoglycemia-related deaths and dementia underscores the need for a deeper understanding of its underlying mechanisms.

To counteract hypoglycemia, the body initiates counterregulatory responses (CRRs) involving the secretion of glucagon, corticosteroids, and catecholamines to stimulate hepatic glucose production and restore blood glucose levels (5). The brain regulates CRRs using two glucose-sensing neurons: glucose-inhibited (GI) and glucose-excited (GE) neurons. GI neurons, which are activated by hypoglycemia and enhance glucose production by modulating CRRs (6), are found in various brain regions, including the arcuate nucleus of the hypothalamus (ARC), lateral hypothalamus, parabrachial nucleus, and ventromedial hypothalamus (VMH) (7–10). Specific neuronal populations (e.g., neuropeptide Y, hypocretin/orexin, and cholecystokinin/leptin receptor–expressing neurons) and certain channels/peptides (e.g., nitric oxide synthase 1, glucokinase, and anoctamin 4) have been identified as GI (11–14). When hypoglycemia occurs, intracellular ATP levels drop, and the AMP-to-ATP ratio increase activates AMP-activated protein kinase and produces nitric oxide to depolarize GI neurons (15). When glucose is unavailable in the brain, neurons use lactate as an alternative fuel. Thus, a lactate infusion into the VMH during hypoglycemia attenuates CRRs (16). In contrast, GE neurons regulate insulin sensitivity in peripheral tissues during hyperglycemia (17). GE neurons also affect CRRs during hypoglycemia via the K_ATP channel (18).

Recurrent hypoglycemia (RH) attenuates CRRs (19). Central and peripheral mechanisms have been reported (20,21). In the central nervous system, RH affects GE and GI neurons’ sensitivity to glucose (22,23). Lactate production from glycogen and fatty acid oxidation in astrocytes in the hypothalamus also involves the development of RH (24). Lactate is used as an alternative fuel in the brain after RH (25). However, the mechanism by which the hypothalamus adapts in response to RH has not been fully understood.

Nonsteroidal anti-inflammatory drugs (NSAIDs) are known to develop hypoglycemia by increasing insulin release through the K_ATP channel in pancreatic β-cells (26). NSAIDs also decrease gluconeogenesis by affecting hepatocytes (27). We recently reported that high blood glucose increases prostaglandin (PG) production via phospholipase A2 (PLA2) in the hypothalamus and activates GE neurons (17). Similar to our previous report, insulin-induced hypoglycemia inhibits arachidonic acid (AA) metabolism of phospholipids in the brain (28). Hypoxia and hypoglycemia activate PLA2, increase the release of AA, and increase cFos in neuroblastoma cells (29). From these reports, we hypothesized that hypoglycemia increases the production of PGs through PLA2-mediated AA release, which alters the activities of GI and GE neurons. However, the role of hypothalamic PG in CRRs during hypoglycemia is unclear.

This study aimed to investigate hypothalamic PG’s role in CRRs during acute hypoglycemia and RH. Our findings suggest that hypoglycemia increases PG production in the hypothalamus, which is essential for activating VMH neurons and stimulating glucagon secretion. Furthermore, inhibiting phospholipid metabolism through cPLA2 shRNA in the hypothalamus improves the attenuation of CRRs during RH.

Research Design and Methods

Animals

C57BL/6N male mice (The Jackson Laboratory Japan, Yokohama, Japan) were kept at room temperature with a 12-h/12-h light and dark time cycle. The mice had free access to water and a regular chow diet (Nosan Corporation, Yokohama, Japan). Mice cages were changed once a week, and mice care was performed according to the Animal Care and Use Committee guidelines of Hokkaido University and Kumamoto University.

Recurrent Hypoglycemia

C57BL/6N male mice received an intraperitoneal (i.p.) injection of insulin (2.5 units/kg; Humarin R, Eli Lilly Japan, Hyogo, Japan) at 1 p.m. The high dose of 2.5 units/kg was chosen to induce severe hypoglycemia rapidly and reproducibly, which is also relevant for modeling RH. Blood glucose levels were measured by a handy glucose meter (Nipro FreeStyle, Nipro, Osaka, Japan). When blood glucose reached 40 mg/dL, glucose (3 g/kg) was injected i.p. or subcutaneously to rescue from hypoglycemia. Glucose was administered before the onset of seizures to ensure recovery. The same procedure was repeated 5 times.

Imaging Mass Spectrometry

C57BL male mice received an i.p. injection of insulin (1 unit/kg, 30 min), 2-deoxy-glucose (2DG; 300 mg/kg, 60 min; Sigma-Aldrich, St. Louis, MO), or saline, after which they were sacrificed using CO₂ asphyxiation. The mice brains were immediately placed into ice-cold saline, embedded in 2% sodium carboxymethyl cellulose solution, and frozen with liquid nitrogen. The 10-μm brain sections were prepared by cryostat and immediately mounted onto an indium-tin-oxide-coated glass slide (Bruker Daltonics, Bremen, Germany). The sections on the glass slides were immediately dried and stored at −20°C until imaging mass spectrometry analysis. Phospholipids in the hypothalamus were measured using ultrafleXtreme (Bruker Daltonics), as described previously (17).

Quantification of PGs

C57BL male mice received i.p. injection of insulin (1 unit/kg, 30 min), streptozocin (STZ; 200 mg/kg), 2DG (300 mg/kg, 60 min), or saline, after which they were sacrificed using CO₂ asphyxiation. The hypothalamus was collected and immediately frozen in liquid nitrogen. PGs in the hypothalamus were measured using ultrafleXtreme (Bruker Daltonics), as described previously (17). Briefly, PGs were extracted from homogenized tissue using solid phase extraction and then concentrated. The resulting samples were analyzed by liquid chromatography-mass spectrometry (LC-MS) to quantify the PGs. The analysis used a triple-quadrupole mass spectrometer with electrospray ionization and a reversed-phase column with a step gradient.

Cannula Implantation and Adeno-Associated Virus Injection

C57BL male mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg) or a combination anesthetic (0.3 mg/kg medetomidine, 4.0 mg/kg midazolam, and 5.0 mg/kg butorphanol). A stainless steel cannula (Plastics One, P1 Technologies, Roanoke, VA) was inserted into the VMH using the following coordinates: an anterior-posterior (AP) direction; −1.5 (1.5 mm posterior to the bregma), lateral (L): ±0.4 (0.4 mm lateral to the bregma), dorsal-ventral (DV): −5.7 (5.7 mm below the bregma on the surface of the skull). For intracerebroventricular injection, AP: −0.3, L: 1.0, DV: −2.0. Cannulae were secured on the skulls with cyanoacrylic glue, and the exposed skulls were covered with dental cement. For adeno-associated virus (AAV) injection, the mice were injected in both sides of the VMH with a maximum of 0.3 µL AAV9-RFP-U6-m-PLA2G4A-shRNA (1.0 × 10¹² genome copies [GC]/mL; shAAV-268768, Vector Biolabs, Malvern, PA) or AAV9-U6-shRNA (scrumble, SCRM)-EF1a-GFP (1.0 × 10¹² GC/mL, SL100894, SignaGen Laboratories, Frederick, MD) using the following coordinates: AP: −1.5, L: ±0.5, DV: −5.7.The mice were allowed to recover for 5–7 days before experiments were started.

Quantitative real-time PCR was performed to assess the efficiency of shPLA2 in the hypothalamus. Mice were anesthetized with isoflurane, euthanized by cervical dislocation, and their brains were collected immediately, after which the hypothalamus was trimmed using a brain matrix. The hypothalamus was homogenized with Template Prepper (Nippon Gene, Tokyo, Japan), and the extracted RNA was reverse-transcribed using Oligo(dT)18 primer (Thermo Fisher Scientific) and M-MLV Reverse Transcriptase (Promega, Madison, WI). Pla2g4a and Actin B cDNA levels in the hypothalamus were measured by the ΔΔCt method using SYBR Green reagent (Bio-Rad, Hercules, CA). The information on primers is shown in Supplementary Table 1. Quantitative PCR was performed with diluted cDNAs in a 10 µL reaction volume in duplicate.

Assessment of Cytosolic- or Secretory-Phospholipase A2 Activity

C57BL male mice fed ad libitum were i.p. injected with insulin (1 unit/kg) or saline. The mice were sacrificed 30 min after the injection using CO₂ asphyxiation, a −1.0- to −2.0-mm coronal section from the bregma was excised using the matrix, and the VMH and ARC areas were collected and stored at −80°C until use. Tissues were homogenized and centrifuged at 10,000g for 15 min at 4°C, and supernatants were collected. The activity of cytosolic- or secretory-phospholipase A2 was measured according to the procedures described in the kit manuals (Abcam, Cambridge, U.K.). Phospholipases A1 (Invitrogen, Waltham, MA), phospholipase C (Sigma-Aldrich), and phospholipase D (Sigma) were also measured according to the manufacture’s instruction. Phospholipase activity was normalized to protein concentration in each sample, which is measured using bicinchoninic acid assay kit (Nacalai Tesque, Kyoto, Japan).

2DG-Induced Hyperglycemia and Food Intake

Ibuprofen (50 μg in 0.5 μL) or vehicle (20% DMSO in PBS) was injected through the intracerebroventricular cannulae in C57BL/6N mice in fed state 30 min before the i.p. injection of 2-DG (300 mg/kg). Blood glucose levels were measured by the handheld glucometer (Nipro FreeStyle, Nipro, Osaka, Japan). For food intake, intracerebrovnentricular injection of ibuprofen and i.p. injection of 2-DG or saline were performed as described above. The reduced weight of food was measured.

Hyperinsulinemic-Hypoglycemic Clamp

C57BL male mice were anesthetized with premixed ketamine (100 mg/kg) and xylazine (10 mg/kg) or a combination anesthetic (0.3 mg/kg medetomidine, 4.0 mg/kg midazolam, and 5.0 mg/kg butorphanol). Polyethylene catheters were implanted into the right carotid artery and jugular vein. The tubes were inserted subcutaneously and protruded from the neck skin (30). The mice were allowed to recover for 3–5 days, and tubes were flushed with heparinized saline daily.

The mice were fasted for 4 h, and experiments were initiated in a free-moving condition. Ibuprofen (250 μmol/L, 0.2 μL) or vehicle (5% DMSO in PBS) was injected bilaterally into the VMH 15 min before measuring initial blood glucose levels. After initial blood glucose levels were measured, hypoglycemic clamp was performed, as described previously (30). After the blood sample was collected at t = 120 min, the mice were anesthetized with isoflurane, euthanized by cervical dislocation, and blood samples were collected from the heart to measure CRR hormones (30).

Glucagon and Norepinephrine Injection

C57BL/6N male mice were bilaterally injected in the VMH with shPLA2 or shSCRM as described above. Three weeks later, these mice experienced RH for five days. On day 6, mice received i.p. injection of glucagon (0.1 unit/kg) in ad libitum fed condition. Blood glucose levels were measured. On another day, the same mice received norepinephrine (NE; 1 mg/kg) i.p. injection, and blood glucose levels were measured at the same time points following the injection.

Measurement of Serum Hormones

After the clamp was performed, the blood samples were centrifuged for 10 min at 1,000g and maintained at −30°C until hormones were measured. The glucagon, corticosterone, epinepheline, and NE concentrations were measured with each ELISA kit (Enzo Life Sciences, Farmingdale, NY). All of the protocols followed the instructions provided by the kit.

Immunohistochemistry

C57BL/6 male mice received saline or ibuprofen (30 mg/kg body wt) per os (p.o.) using a gastric probe 30 min before i.p. injection of saline or insulin (0.75 units/kg). The mice were sacrificed using CO₂ asphyxiation and perfused with heparinized saline transcardially at 60 min after i.p. injection. Brain sections (50-µm each) containing both VMH and ARC were collected. The floating sections were incubated with rabbit–anti-cFos antibody (1:1000; no. 2250S, Cell Signaling Technology, Danvers, MA) in blocking solution (0.1 mol/L phosphate buffer [PB] containing 4% normal guinea pig serum, 0.1% glycine, and 0.2% Triton X-100) overnight at room temperature. Sections were rinsed with PB and incubated in a secondary antibody (1:500, Alexa 488 secondary antibody, A11008; Invitrogen, Waltham, MA) for 2 h at room temperature. The stained sections were washed with PB three times and mounted on glass slides with DAPI Fluoromount-G (Southern Biotech, Birmingham, AL). Cells were automatically counted using an ImageJ software plugin (Analyze Particles). The sections were incubated with rabbit–anti-cFos antibody (1:1000; no. 2250S, Cell Signaling Technology), and guinea pig–anti-glutamate transporter (VGluT2) antibody (1:1000; MSFR106280, Nittobo Medical, Tokyo, Japan), goat–anti-glutamate decarboxylase (GAD67/65) antibody (1:1000; MSFR101600, Nittobo Medical), or mouse–anti-estrogen receptor-α (ERα) antibody (1:250; Sc-8002, Santa Cruz Biotechnology, Dallas, TX) to assess the neuronal characterization of insulin-activated neurons.

Intracerebrovnentricular Administration of PGs

Saline, PGE2, or PGF2α (25 ng in 0.5 µL, 10% DMSO in PBS) was injected through the intracerebrovnentricular cannulae in C57BL/6N mice in fed state 60 min before the perfusion. To determine which PGs are involved in hypothalamic neuronal activation, cFos immunohistochemistry was performed as described above.

Fiber Photometry

Male C57BL/6 mice (8–9 weeks old) received an injection of AAV1-syn-jGCaMP8s-WPRE (5.7 × 10¹² GC/mL; Addgene, no.162374) into the VMH (AP −1.5 mm, lateral 0.53 mm, DV 5.75 mm) at a volume of 0.3 µL. Three weeks later, an optic fiber (400 µm in diameter) was implanted at the viral injection site (DV 5.60 mm). After a recovery period of at least 1 week, the mice underwent the experiments. Fiber photometry recordings were performed using a wireless fiber photometry device (TeleFipho, BioResearch Center, Aichi, Japan). Five minutes after device attachment, saline or ibuprofen (30 mg/kg) was administered orally, followed by i.p. injection of insulin (0.75 units/kg) 30 min later (0 min). Photometry signals were recorded at a sampling rate of 100 Hz. The raw signal was fitted with an exponential curve using all data acquired during the recording. ΔF/F was z-scored using data from 1 to 5 min prior to insulin injection. Both signal recording and analysis were performed using TeleFipho software (BioResearch Center), and plots were generated using the open-source software pMAT (31).

Statistical Analysis

For repeated-measures analysis, two-way ANOVA was used when values over different times were analyzed, followed by the Šidák multiple comparisons tests. For the statistical analysis between multiple independent groups, one-way ANOVA was used, followed by Tukey multiple comparisons tests. When only two groups were analyzed, statistical significance was determined by the unpaired Student t test (two-tailed P value). GraphPad Prism 10 software (GraphPad Software, San Diego, CA) was used for these statistical analyses. A value of P < 0.05 was considered statistically significant. All data are shown as mean ± SEM.

Data and Resource Availability

The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request

Results

Glucose Deprivation, Induced by Insulin and 2DG, Decreases AA-Containing Phospholipids in the Hypothalamus in Mice

To check the role of hypoglycemia in hypothalamic phospholipid metabolism, the signal intensity of 11 different phospholipids located in both the VMH and the ARC were measured at 30 min after the i.p. insulin injection (Fig. 1A). After the insulin injection, blood glucose levels became 71.6 ± 6.5 mg/dL (Fig. 1B). Insulin injection decreased phosphatidylinositol (PI) (16:0/20:4), PI (18:0/20:4), PI (18:1/20:4), phosphatidylethanolamine (PE) (18:0/16:1), PE (18:0/20:4), and phosphatidylserine (PS) (18:0/16:0) in the VMH compared with the saline group (Fig. 1C–F) and PI (16:0/20:4), PI (18:0/20:4), and PI (18:1/20:4) in the ARC (Fig. 1C, G–I). We measured the activities of PLA1, PLA2, PLC, and PLD, which are enzymes that metabolize phospholipids. Insulin injection activated cPLA2 but not PLA1, secretory PLA2, PLC, and PLD in the hypothalamus (Fig. 1J–N).

Figure 1.

Insulin injection decreases phospholipids in the VMH. A: A schematic showing the VMH and ARC nucleus of the hypothalamus, and imaging mass spectrometry (IMS). B: Blood glucose levels 30 min after the saline or insulin i.p. injection. C: Representative results of IMS showing the distribution of hypothalamic PI (16:0/20:4), PI (18:0/20:4), and PI (18:1/20:4) in male mice after injection with saline (left) or insulin (right). The white line shows the position of the VMH and ARC. Scale bar: 200 µm. Relative intensities of PI (D and G), PE (E and H), and PS (F and I) in the VMH and ARC 30 min after the saline (n = 5) or insulin (n = 5) i.p. injection. Enzymatic activity of PLA1 (J), cytosolic PLA2 (K) and secretory PLA2 (L), PLC (M), and PLD (N) in the hypothalamus 30 min after the saline (n = 5) or insulin (n = 4) i.p. injection. All experiments were performed using male mice. All results are shown as mean ± SEM. Student t test (two-tailed P value) and one-way ANOVA, followed by Tukey multiple comparison tests, were used for statistical analysis. *P < 0.05, **P < 0.01, ****P < 0.0001.

STZ-induced disruption of insulin secretion elevated blood glucose level to 442.8 ± 35.5 mg/dL 4 days after STZ treatment (Fig. 2A). When phospholipid intensity was examined in the VMH and ARC, no significant differences were observed in any types of phospholipids, including PI, PE, and PS in either hypothalamic nucleus (Fig. 2B–G). Meanwhile, we evaluated the effect of insulin-independent hypoglycemia induced by 2-DG on hypothalamic phospholipid metabolism. The i.p. injection of 2DG increased blood glucose levels after 60 min (Fig. 2H). 2DG decreased the signal intensity of PI (18:0/20:4) and PE (p18:0/20:4) in the VMH compared with mice receiving a saline injection (Fig. 2I–K). 2DG injection also decreased PI (18:1/20:4), PE (p18:0/20:4), and PS (18:0/16:0) in the ARC (Fig. 2L–N). These data suggest that not only insulin but also hypoglycemia enhances AA-containing phospholipid metabolism in the hypothalamus, at least in part.

Figure 2.

Glucose deprivation, but not insulin deficiency, changes phospholipids in the hypothalamus. A: Blood glucose levels 4 days after the saline or STZ injection (200 mg/kg) in normal mice. Relative intensities of phospholipids in the VMH (B–D) and ARC (E–G) 4 days after i.p. injection with saline (n = 4) or STZ (n = 5) in normal mice. H: Blood glucose levels in 30 min after the saline or 2DG i.p. injection in normal mice. Relative intensities of phospholipids in the VMH (I–K) and ARC (L–N) 60 min after i.p. injection with saline (n = 4) or 2DG (n = 3) in normal mice. All experiments were performed using male mice. All results are shown as mean ± SEM. Student t test (two-tailed P value) and one-way ANOVA, followed by Tukey multiple comparison tests, were used for statistical analysis. *P < 0.05.

Insulin Increases PG Production in the Hypothalamus

AA is known to be used as a substrate of PGs. Thus, we measured all types of PGs in the whole hypothalamus by LC-MS 30 min after insulin injection when blood glucose became 46.7 ± 9.8 mg/dL (Fig. 3A). Insulin injection increased the amount of 6-keto-PGF1α, 11-β-13,14-dihydro-15-keto-PGF2α, PGF2α, PGE2, and 20-hydroxy-PGF2α in the hypothalamus (Fig. 3B–G). Therefore, these results suggest that insulin injection induces AA release from phospholipids to produce PGs in the hypothalamus.

Figure 3.

Insulin increases PGs in the hypothalamus. A: Blood glucose levels 30 min after the saline or insulin i.p. injection. B: Relative amounts of hypothalamic PGs mediated by cyclooxygenase showing mean fold-change in color. Fold-change of 6-keto-PGF1α (C), 11-β-13,14-dihydro-15-keto-PGF2α (D), PGF2α (E), PGF2 (F), and 20-hydroxy-PGF2α (G). All experiments were performed using male mice. All results are shown as mean ± SEM. The unpaired Student t test (two-tailed P value) was used for statistical analysis. *P < 0.05; **P < 0.01.

Ibuprofen Blocks CRRs and Activity of GI Neurons

To investigate the role of PGs in the brain in the regulation of systemic glucose metabolism, we injected ibuprofen intracerebroventricularly and measured glucose production induced by i.p. injection of 2DG. 2DG injection activates CRRs and increased blood glucose levels (Fig. 4A and B). Intracerebroventricular injection of ibuprofen lowered the 2DG-induced hyperglycemia (Fig. 4A and B). 2DG is also known to increase food intake. However, ibuprofen injection did not change feeding behavior compared with saline-injected mice (Fig. 4C).

Figure 4.

Central ibuprofen injection impairs glucose production in 2DG-induced glucose deprivation and hypoglycemic clamp. A: 2DG-induced hyperglycemia in mice that received ibuprofen (Ibu) or vehicle intracerebroventricular (icv) injection. B: Area under the curve of blood glucose in A. C: Food intake after intracerebroventricular saline and i.p. saline (Vehi + Sal), intracerebroventricular ibuprofen and i.p. saline (Ibu + Sal), intracerebroventricular saline and i.p. 2DG (Vehi + 2DG), and intracerebroventricular ibuprofen and i.p. 2DG (Ibu + 2DG). D: Blood glucose levels during hyperinsulinemic-hypoglycemic clamp in mice that received ibuprofen or vehicle injection into the VMH bilaterally. Insulin infusion was started at t = 0 min. E: Glucose infusion rate to maintain hypoglycemia at 50 mg/dL during clamp. Concentration of serum hormones, glucagon (F), corticosterone (G), epinephrine (H), and NE (I), corrected at 120 min of the clamp experiment. All experiments were performed using male mice. All results are shown as mean ± SEM. The unpaired Student t test (two-tailed P value) and two-way ANOVA, followed by Šidák multiple comparison tests, were used for statistical analysis. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

To verify the role of PGs in the CRRs, we performed hyperinsulinemic-hypoglycemic clamp experiments in mice. In the hypoglycemic clamp, a constant rate of insulin (1 mU/kg/min) was infused into the circulation, while glucose was infused as needed to maintain blood glucose levels at 50 mg/dL (Fig. 4D). Ibuprofen injection into the VMH increased glucose infusion rate (GIR) (Fig. 4E), suggesting that ibuprofen inhibits glucose production during hypoglycemia, and thus, more glucose was needed to maintain at 50 mg/dL. Then, we measured CRRs-related hormones in the blood at 120 min of the clamp experiment. Ibuprofen injection significantly decreased glucagon levels but not corticosterone, epinephrine, and NE (Fig. 4F–I). These data suggest that hypothalamic PGs are necessary for glucagon secretion in response to hypoglycemia to induce glucose production.

To measure the neuronal activity in the hypothalamic GI neurons that regulate CRRs, we measured cFos immunohistochemistry. Insulin (i.p.) significantly increased the number of cFos-positive cells in the ventrolateral part of the VMH (vlVMH), while oral injection of ibuprofen suppressed it (Fig. 5A and B). The number of cFos-positive cells was not altered by ibuprofen injection in the dorsomedial (dm) and central (c) part of the VMH and ARC (Fig. 5A–C). 2DG is known to introduce glucose deprivation in tissues. Therefore, we examined the cFos-positive cell number in 2DG-injected mice (Fig. 5D). 2DG injection did not change the cFos expression in the dmVMH, cVMH, and vlVMH (Fig. 5E) and ARC (Fig. 5F). Given that ERα is known to be involved in CRRs within the vlVMH (32), we investigated whether the neurons, whose insulin-induced activity was suppressed by ibuprofen, also express ERα. We quantified cFos-ERα double-positive cells (Fig. 5G). While the number of these double-positive cells in the VMH did not differ between the saline and ibuprofen groups (Fig. 5H), ibuprofen administration increased insulin-induced cFos–Erα-positive cell numbers in the ARC (Fig. 5I). To investigate which type of PG regulates GI neuronal activity under physiological conditions, we administered vehicle, PGE2, or PGF2α intracerebroventricularly and examined cFos expression in the hypothalamus (Fig. 5J). In the VMH, PGF2α significantly increased the number of cFos-positive cells in cVMH and tended to increase in vlVMH (Fig. 5K), whereas cFos immunoreactivity was comparable among treatments in the ARC (Fig. 5I).

Figure 5.

Ibuprofen decreases insulin-induced cFos-positive cells specifically in vlVMH. A: Representative micrographs of cFos immunoreactivity in the VMH and ARC. White lines indicate the boundary of the nucleus in the hypothalamus. Quantification of cFos cell number in dmVMH, cVMH, and vlVMH part of the VMH (B) and ARC (C) in the mice that received vehicle (p.o.) and vehicle (i.p.) (Vehi + Vehi), ibuprofen (p.o.) and vehicle (i.p.) (Ibu + Vehi), vehicle (p.o.) and insulin (i.p.) (Vehi + Ins), ibuprofen (p.o.) and insulin (i.p.) (Ibu + Ins). D: Representative micrographs of cFos immunoreactivity in the VMH and ARC. Quantification of cFos cell number in dmVMH, cVMH, and vlVMH (E) and ARC (F) in the mice that received i.p. injection of saline or 2DG. G: Representative micrographs of cFos-ERα immunoreactivity in the VMH and ARC. Quantification of cFos-ERα double-positive cells in VMH (H) and ARC (I). J: Representative micrographs of cFos immunoreactivity in the mice received intracerebroventricular saline, PGE2, or PGF2α at −60 min of perfusion. Quantification of cFos-positive cells in VMH (K) and ARC (L). Scale bars: 100 µm. All experiments were performed using male mice. All results are shown as mean ± SEM. One-way ANOVA, followed by Tukey multiple comparison tests, and two-way ANOVA, followed by Šidák multiple comparison tests, were used for statistical analysis. *P < 0.05; **P < 0.01.

To measure insulin-induced VMH neuronal activity with higher temporal resolution, we performed fiber photometry (Fig. 6A–C). VMH excitation induced by insulin was suppressed in the ibuprofen-treated group between 50 and 60 min (Fig. 6D).

Figure 6.

Prior ibuprofen treatment decreased insulin-induced VMH neuronal activity in the fiber photometry. A: Schematic illustration of AAV injection in the VMH and schedule of the experiment in male mice. In the recording day, mice received p.o. vehicle or ibuprofen 30 min prior to i.p. injection of insulin. B: A representative image of the expression of AAV1-syn-jGCaMP8s-WPRE in the VMH and optic fiber insertion (dotted line). Scale bar: 100 µm. C: Representative z-scored Ca signal (ΔF/F) from photometry recording in a mouse that received p.o. vehicle (black) or ibuprofen (red), respectively. Dotted line indicates i.p. injection of insulin. D: A significant decrease was seen in 10-min averaged z-scored Ca signals during 50–60 min by ibuprofen treatment. All experiments were performed using male mice. Two-way ANOVA, followed by Šidák multiple comparison tests, were used for statistical analysis. *P < 0.05.

To characterize the insulin-activated neurons, we performed double immunofluorescence staining for VGluT2 (Fig. 7A) or GAD67/65 (Fig. 7D) together with cFos in the hypothalamus. The numbers of cFos-VGluT2 and cFos-GAD67/65 double-positive cells in both the VMH and ARC were comparable between the saline- and insulin-injected groups (Fig. 7B, C, E, and F), and the proportions of double-positive cells among cFos-positive cells were also not significantly different (Fig. 7G and H).

Figure 7.

Insulin-induced cFos-positive cells colocalize with ERα, VGluT2, and GAD67/65, whereas only the cFos-ERα double-positive ratio increased. A: Representative micrographs of cFos and VGluT2 immunoreactivity in the VMH and ARC. White lines indicate the boundary of the nucleus in the hypothalamus. Quantification of cFos-VGluT2 double-positive cell number in dmVMH, cVMH, and vlVMH part of the VMH (B) and ARC nucleus (C) in the mice that received i.p. injection of saline or insulin (1 unit/kg). D: Representative micrographs of cFos and GAD67/65 immunoreactivity in the VMH and ARC. Quantification of cFos-GAD67/65 double-positive cell number in dmVMH, cVMH, and vlVMH (E) and ARC (F) in the mice that received i.p. injection of saline or insulin (1 unit/kg). Rate of cFos-positive cells colocalized with VGluT2 and GAD67/65 in VMH (G) and ARC (H). Scale bars: 100 µm. All experiments were performed using male mice. All results are shown as mean ± SEM. One-way ANOVA, followed by Tukey multiple comparison tests, were used for statistical analysis. No significance in P > 0.05.

Hypothalamic PGs Deteriorate Glucose Production after RH

To generate the RH mouse model, 2.5 units/kg of insulin was injected i.p. to induce hypoglycemia (40 mg/dL) once a day for 5 days (Fig. 8A, D, and G). To inhibit the production of PGs in the hypothalamus, we injected AAV that expresses shRNA against cPLA2 (shPLA2) into the VMH (Fig. 8A and B), since cPLA2 is activated by insulin injection (Fig. 1K). AAV-infected cells were found in the vlVMH and ARC (Fig. 8B). cPLA2 mRNA in the hypothalamus was significantly decreased (Fig. 8C). On day 6, the hypoglycemic clamp was performed (Fig. 8A and E). In the control group, GIR was significantly increased in RH mice (Fig. 8F), suggesting that RH attenuated glucose production. The shPLA2 group also received RH treatment for 5 days, and the hypoglycemic clamp was performed (Fig. 8G–I). GIR in the shPLA2 without RH was similar to that of control RH mice and ibuprofen-injected mice (Figs. 4E and 8F), confirming that hypothalamic PG production enhances CRRs in the non-RH condition. However, the shPLA2 inhibited the increase in GIR after RH (Fig. 8I). RH decreased plasma glucagon and epinephrine at 120 min in the hypoglycemic clamp (Fig. 8J and K). RH did not change plasma norepinephrine, corticosterone, and growth hormone (Fig. 8L–N). shPLA2 did not improve glucagon, epinephrine, NE, corticosterone, or growth hormone levels (Fig. 8J–N). After RH, the shPLA2 group had a higher glucose production after glucagon injection than shSCRM group (Fig. 8O), while NE-induced glucose production was comparable between groups (Fig. 8P).　

Figure 8.

shRNA against cPLA2 in the hypothalamus improves RH-induced impairment of glucose production. A: A schematic illustrating AAV injection into the VMH and experimental timeline of AAV injection, RH, and hypoglycemic clamp. B: Representative image of RFP expression after AAV injection. C: mRNA expression of cPLA2 in the hypothalamus. D and G: Daily blood glucose levels during 5 days of recurrent hypoglycemia. E and H: Blood glucose levels during hypoglycemic clamp. Insulin infusion was started at t = 0 min. F and I: GIR to maintain hypoglycemia at 50 mg/dL during clamp and average GIR in the last 40 min. Concentration of serum hormones, glucagon (J), epinephrine (K), NE (L), corticosterone (M), and growth hormone (N), corrected at 120 min of the clamp experiment. Glucagon (O) and NE (P) sensitivities in shPLA2- or shSCRM-injected mice after the RH. All experiments were performed using male mice. All results are shown as mean ± SEM. The unpaired Student t test (two-tailed P value), one-way ANOVA, followed by Tukey multiple comparison tests, and two-way ANOVA, followed by Šidák multiple comparison tests, were used for statistical analysis. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Discussion

In this study, we found that 1) insulin, and partially insulin-induced hypoglycemia, increase PG production using AA from phospholipids; 2) the PG production in the VMH is necessary for the neuronal activation in the vlVMH, glucagon secretion from the pancreas, and an increase in hepatic glucose production during acute hypoglycemia; 3) RH attenuates glucagon secretion and glucose production; and 4) cPLA2 in the hypothalamus is critical for attenuating glucose production after RH by changing glucagon sensitivity.

NSAIDs, including salicylates, were used to treat patients with diabetes in the early 1900s (33). NSAIDs cause gastrointestinal bleeding (33) and severe hypoglycemia when used with other blood glucose-lowering agents (34). Hence, they are no longer used as antidiabetes drugs. The mechanism of the hypoglycemic effect of NSAIDs has been studied only in the peripheral tissues (26,27). While NSAIDs are the most commonly used drug for analgesics due to their inhibition of PG production in the brain, research on their effect on lowering blood glucose levels in the central nervous system is limited. Intracerebroventricular injection of PGD2, PGE1, PGE2, or PGF2α increases the glucose concentration in the hepatic vein (35). Central PGF2α has the most substantial effect but is mediated by epinephrine secretion from the adrenal medulla, not glucagon (35). We found that PGF2α led to an increase in cFos-positive neurons within the cVMH and a tendency for similar increases in the vlVMH. Given that ibuprofen suppressed insulin-induced cFos expression in the vlVMH, it is plausible that PGF2α plays a role in mediating CRR-related activation within the VMH. Other PGs, such as 11-β-13,14-dihydro-15-keto-PGF2α and 20-hydroxy-PGF2α, which were increased by insulin in the hypothalamus in this study, could also have been involved.

The VMH plays a vital role in the regulation of CRRs to hypoglycemia. A lesion of the VMH suppresses CRRs in response to systemic injection of 2DG (36). 2DG perfusion in the VMH via microdialysis increases glucagon, epinephrine, and NE secretion (37). In contrast, glucose perfusion in the VMH during the hypoglycemic clamp blocks increases in the secretion of CRR-related hormones (38). Synaptic glutamate release from the VMH-specific steroidogenic factor 1 (SF1) neuron regulates CRRs (39). Neuronal pathways from the parabrachial nucleus to the VMH (10) and from VMH to the bed nucleus of the stria terminalis regulate CRRs (40). Therefore, it is considered that the VMH has a critical role in detecting low blood glucose levels and changing CRRs. In our previous study (17), the knockdown of cPLA2 in SF1 neurons did not change blood glucose levels after systemic 2DG injection compared with control mice. Therefore, neurons in the dmVMH, including SF1 neurons, may not be necessary for NSAID-related change in CRRs. Conversely, ERα-expressing neurons in the vlVMH are involved in the CRRs (32). Both GI-type ERα neurons projecting to the medioposterior ARC and GE-type ERα neurons projecting to the dorsal raphe nucleus regulate CRRs (32). We attempted to determine the neuron type (e.g., glutamatergic, γ-aminobutyric acidergic, or ERα) of the vlVMH, but were unable to identify it. Insulin and 2DG can both induce hypoglycemia or glucose deprivation. However, in our study, 2DG did not alter cFos expression in the VMH and ARC, unlike insulin, which increased cFos-positive cells in the vlVMH. This suggests that neuronal activity in the vlVMH is not solely regulated by glucose shortage but may also involve direct hypothalamic insulin signaling.

In our data, ibuprofen decreased VMH neuronal activity during hypoglycemia detected by fiber photometry. Although our experimental system could not confirm whether the GCaMP-expressing neurons in the VMH were GI, the observed suppression of neuronal activity by ibuprofen during insulin-induced hypoglycemia is consistent with our cFos data. However, the precise central mechanisms underlying this effect are still unclear.

Under healthy conditions, hypothalamic PGs are produced by the hyperglycemia to decrease blood glucose level (17). Thus, PGs are generated in response to both increases and decreases in blood glucose, contrary to our initial expectation that hypoglycemia would decrease PG levels. Indeed, when individuals with type 2 diabetes mellitus (T2DM) undergo a hypoglycemic clamp, an increase in inflammatory markers is observed (41), suggesting that hypoglycemia in humans also triggers PG production. There are subtle differences in the types of PGs produced in the hypothalamus during hyperglycemia versus hypoglycemia. Hyperglycemia increases 6-keto-PGF1α, PGD2, 13,14-dihydro-15-keto-PGF2α, and PGE2, whereas hypoglycemia increases PGF2α and 20-hydroxy-PGF2α. Our results also suggest that PGF2α is the important regulator of VMH activation. Notably, these PGs were measured from the whole hypothalamus, not only VMH. So, the specific types of PGs produced in the VMH remain unknown. These distinct PG profiles might play a crucial role in determining the appropriate physiological response—either lowering hyperglycemia or facilitating recovery from hypoglycemia.

RH is a critical incident in the treatment of diabetes, and it can result in hypoglycemia-associated autonomic failure. RH increases hexokinase activity and attenuates the activity of glucose-sensing neurons in the VMH (43). Indeed, we found that the RH decreased plasma glucagon, epinephrine, and glucose production during the hypoglycemic clamp. Expression of shPLA2 in the VMH improved the RH-induced attenuation of glucose production. However, shPLA2 did not change glucagon or epinephrine production. Instead, shPLA2 enhanced glucagon sensitivity to increase glucose levels. Interestingly, although a higher dose of glucagon (0.1 units/kg = 0.1 mg/kg) was administered compared with previous reports, 16 µg/kg (44,45) or 30 mg/kg (46), blood glucose levels showed minimal elevation in the control shSCRM group. Consistent with previous studies reporting that RH impairs glucagon-induced hepatic glucose production (47), these findings strongly suggest that cPLA2-knockdown in the VMH protects against RH-induced impairment of CRRs, particularly the central regulation of glucagon-induced gluconeogenesis.

Previous report showed that nitric oxide production and activation of its receptor soluble guanylyl cyclase (sGC) are accompanied by increased reactive oxygen species, which suppress CRRs via S-nitrosylation of sGC in the VMH during hypoglycemia (42). Thus, the central mechanisms involved in RH-induced CRRs impairment are likely diverse. Although reactive oxygen species may have influenced the PG-mediated effects we investigated in this study, the details await future research.

Liu Z et al. (20) reported that RH changes adrenergic sensitivity in the liver and visceral fat in rats, but few papers have studied glucagon sensitivity. Glucagon sensitivity is not altered in type 1 diabetes mellitus (T1DM) (48). Metabolic-associated steatotic liver disease and metabolic-associated steatotic liver disease–related gestational diabetes mellitus impair glucagon sensitivity (49,50). Conversely, pancreatectomized patients increase hepatic glucagon sensitivity (51). In mice, Western diet–fed mice, high-fat diet–fed mice, and db/db mice do not change glucagon sensitivity (52). However, there are no reports of the brain modulating glucagon sensitivity.

Recently, the importance of glucagon in both T1DM and T2DM is apparent (53). In our study, intrahypothalamic injection of ibuprofen and experiencing RH decreased glucagon secretion during hypoglycemia. In T2DM patients, NSAIDs decrease blood glucose concentration and HbA_1c (54,55). In a retrospective cohort study, NSAIDs also reduced the risk of becoming T2DM (56). Thus, NSAIDs could potentially be immediately used for managing glucose fluctuations in both T1DM and T2DM. Since glucagon concentrations were not measured in these studies, the relationship between NSAIDs, glucagon, and T2DM is still unclear. Further study is needed to clarify the mechanisms by which NSAIDs improve T1DM and T2DM. The possibility that the brain regulates glucagon sensitivity may be an interesting research topic and has the potential to open a new field of glucagon research. Targeting cPLA2 specifically in the hypothalamus may be a more effective therapeutic approach during hyperglycemia, but it may cause hypoglycemia because it suppresses CRRs.

To assess ibuprofen’s central effects, we used intracerebroventricular injection. Although this method may appear to diverge from typical human clinical scenarios involving oral intake, ibuprofen is known to cross the blood-brain barrier (57). Furthermore, considering that ibuprofen can alter glucagon production in the pancreatic islet (58)—potentially confounding the interpretation of its central effects—intracerebroventricular administration was deemed appropriate for investigating its central role in regulating CRR.

The current study revealed the importance of PGs in the hypothalamus for the recovery from hypoglycemia. Hypothalamic PGs regulate the activity of GI neurons and enhance glucagon secretion. RH impairs glucagon and epinephrine secretion. Hypothalamic PGs attenuate glucose production after RH by changing glucagon sensitivity. More studies are needed to broaden our understanding of strategies to prevent RH.

This article contains supplementary material online at https://doi.org/10.2337/figshare.30086254.

Funding Statement

This work was supported by Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research (JSPS KAKENHI, grants 21H02352, 21K18275T, 23H00512, 25K02313, 25K19662), Japan Agency for Medical Research and Development (AMED-RPIME, grants JP21gm6510009h0001, JP22gm6510009h9901, 23gm6510009h9903, 24gm6510009h9904), JST CREST (JPMJCR21P1), the Takeda Science Foundation, the Uehara Memorial Foundation, Astellas Foundation for Research on Metabolic Disorders, Suzuken Memorial Foundation, Akiyama Life Science Foundation, Narishige Neuroscience Research Foundation, Japan Science and Technology Agency Support for Pioneering Research Initiated by the Next Generation (JST SPRING, JPMJSP2119), JASSO Novo Nordisk Foundation, Japan IDDM Network Foundation, and Daiichi Sankyo Foundation of Life Science.