Synaptic Adaptation to Repeated Hypoglycemia Depends on the Utilization of Monocarboxylates in Guinea Pig Hippocampal Slices

Hippocampal slices were prepared from guinea pig brains (Hartley, Shizuoka, Japan) by a standard technique (11). All animals were treated according to the guidelines for animal experimentation at Kobe University School of Medicine. Guinea pigs (14–28 days old) were anesthetized with sodium pentobarbital and decapitated. The hippocampi were removed from the skull and sliced rapidly into 300- to 400-μm-thick transverse sections. The slices were placed in an incubation chamber containing standard medium (in mmol/l: 125 NaCl, 4 KCl, 1.24 KH₂PO₄, 1.3 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 10 glucose) and bubbled with 95% O₂ and 5% CO₂ at 35°C. At the time of study, hippocampal slices were transferred individually to a submersion recording chamber, which was perfused with the standard medium at a flow rate of 4 ml/min.

Extracellular recordings of the field postsynaptic population spike (PS) were obtained in the granule cell layer of the hippocampal dentate gyrus with 2 mol/l NaCl glass electrodes (resistance: 1–5 MΩ). Evoked synaptic responses were elicited with 0.1-msec constant-current pulses through a bipolar electrode placed in the perforant pathway. After establishing a stable baseline for at least 20 min as well as a control input-output curve, we monitored synaptic responses by applying single stimuli to the perforant pathway every 10 s at intensity sufficient to elicit a 60–70% maximal PS. To induce glucose deprivation, we removed glucose from the standard medium. To test the effect of lactate or pyruvate during glucose deprivation, we replaced glucose in the medium with 10 mmol/l Na lactate or 10 mmol/l Na pyruvate. Adding sodium lactate or sodium pyruvate did not influence the pH of the circulation medium. To exclude irreparable damage to the slices, we reintroduced standard medium containing 10 mmol/l glucose after test-medium treatment, and recovery of the PS amplitude was measured. All experimental media were bubbled with 95% O₂ and 5% CO₂, and the temperature of the media was maintained at 35°C throughout the experiments.

After preincubation in the standard O₂-saturated medium for at least 20 min, hippocampal slices were incubated in test medium containing 0–10 mmol/l glucose. Two hippocampal sections were homogenized immediately in ice-cold 0.5 mol/l perchloric acid with 1 mmol/l EDTA and centrifuged for 15 min at 2,000 rpm. Pellets were solubilized in 1 mol/l NaOH and used for protein assay according to the Lowry method (12). Supernatants were neutralized with 2 mol/l KHCO₃, centrifuged, and used directly for assay of ATP, phosphocreatine (PCr), and glucose. ATP, PCr, and glucose concentrations were determined by sensitive enzymatic methods that analyze NADPH production (11).

For glycogen assay, hippocampal slices incubated in test perfusion medium were homogenized in 150 μl of ice-cold 30 mmol/l HCl and heated at 100°C for 3 min, which promptly halts glycogen metabolism. Hippocampal homogenates were stored at −20°C until assays were performed. Glycogen assays were performed by the method of Passonneau and Lauderdale (13), in which 1 mol/l HCl completely hydrolyzes glycogen to glucose. Briefly, 50 μl of the suspension was removed and added to 50 μl of 0.5 mol/l NaOH for protein assay. The remainder was divided into 70- and 20-μl fractions (fractions A and B, respectively). Fraction A was added to 630 μl of 1 mol/l HCl and heated for 3 h at 100°C, and fraction B was not. Glucose in both fractions was quantified by the glucose-6-phosphate dehydrogenase/NADP fluorescence method (11). The glucose in fraction B, which reflects the endogenous glucose in hippocampal slices, was subtracted from the glucose in fraction A, which reflects the sum of endogenous glucose and glucose derived from glycogen hydrolysis. Standards were prepared from glucose or from bovine liver glycogen after desiccation at 120°C. It was found that desiccated glycogen hydrolyzed with HCl yields almost exactly the predicted amount of glucose (data not shown).

Glucose-6-phosphate dehydrogenase, hexokinase, creatine kinase, ADP, and NADP were purchased from Boehringer-Mannheim (Mannheim, Germany). Sodium lactate, sodium pyruvate, and α-cyano-4-hydroxycinnamic acid (4-CIN) were obtained from Sigma (St. Louis, MO), and all other chemicals were from Wako (Tokyo, Japan) or Nacalai (Kyoto, Japan). 4-CIN was dissolved in DMSO and applied to the perfusion medium at a final concentration of 200 μmol/l. 3-Isobutyl-1-methylxanthine (IBMX) was dissolved in 3.5 mol/l KOH and was applied to the extracellular solution at a concentration of 500 μmol/l.

Data are presented as means ± SE. Statistical analysis was performed by nonpaired t test. Treatment differences were considered significant at P < 0.05.

Similar to previously reported findings (11,14), a low glucose concentration in the extracellular solution produced a slowly developing decrease in synaptic response in the granule cell layer of guinea pig hippocampal slices (Fig. 1). Removal of glucose from the standard medium produced a complete elimination of the PS within 30 min, with a 50% reduction in amplitude by 10.6 min (Fig. 1A). Incubation in 1.0 or 0.5 mmol/l glucose-containing solutions for 30 min suppressed the PS amplitude to 76.6 ± 4.2 or 50.3 ± 5.0% of the control PS amplitude, respectively. In contrast, slices perfused with standard medium containing 10 mmol/l glucose maintained normal evoked activity.

To determine whether the decay in synaptic transmission in response to glucose deprivation is accompanied by a reduction of high-energy phosphates, we investigated the preservation of ATP and PCr contents in hippocampal slices. Control ATP and PCr concentrations were 13.3 ± 0.6 and 29.2 ± 1.9 mmol/kg protein, respectively. Incubation in glucose-depleted solution for 30 min decreased the ATP content to 85.9 ± 3.7% of the control level (P < 0.05); (Fig. 1B). PCr was decreased to 84.5 ± 5.0% of the control level (data not shown; P < 0.05). Because a reduction in ATP and PCr has been estimated to occur throughout hippocampal slices, we determined the specific changes in ATP and PCr in the dentate gyrus (15). The dentate gyrus was dissected from hippocampal slices under a stereoscope and incubated in glucose-depleted solution for 30 min. ATP and PCr in the dissected dentate gyrus were decreased to 88.8 ± 5.9 and 84.7 ± 4.9% of the control levels, respectively (15), in good agreement with the ATP and PCr preservation levels determined in intact hippocampal slices. These results indicate that glucose deprivation produces complete inhibition of synaptic activity in the hippocampal dentate gyrus, accompanied by a significant reduction in ATP and PCr.

We examined the effect of repeated glucose depletion on synaptic activity. After 30 min of glucose deprivation, hippocampal sections were incubated in standard medium containing 10 mmol/l glucose, resulting in recovery of the PS amplitude to 129.0 ± 5.9% of the control level (Fig. 2A). After stabilization of PS amplitude in standard medium, hippocampal slices were reintroduced to low-glucose medium. The repeated glucose depletion gradually inhibited PS amplitudes over 80 min, with a 50% inhibition by 36.5 min (Fig. 2A). However, in hippocampal slices subjected repeatedly to low glucose levels (1 or 0.5 mmol/l), PS amplitudes were maintained for >90 min (Fig. 2A).

Preservation of high-energy phosphates during repeated glucose deprivation is shown in Fig. 2B. After transient glucose deprivation, ATP concentration recovered to 96.7 ± 3.2% of the control level after a 30-min incubation in standard medium. Incubation in secondary glucose-free solution for 30, 60, and 90 min resulted in unchanged ATP contents: 87.7 ± 5.4, 93.4 ± 6.1, and 108.1 ± 7.3% of the control ATP concentration, respectively (Fig. 2B). PCr concentrations were also maintained during secondary glucose deprivation, ranging from 85.3 ± 6.5 to 95.0 ± 1.6% of the control PCr level (data not shown). Incubation in 0.5 mmol/l glucose solution did not decrease ATP levels (Fig. 2B); rather, ATP levels were increased after 60 min (120.8 ± 5.9%; P < 0.05). Similarly, PCr concentrations in hippocampal slices incubated in 0.5 mmol/l glucose solution did not decrease; rather, they increased after 60 min (data not shown). Incubation in 10 mmol/l glucose solution increased both ATP (P < 0.01) (Fig. 2B) and PCr concentrations (data not shown) after transient glucose deprivation. These results indicate that synaptic activity becomes resistant to repeated hypoglycemia, (i.e., hypoglycemic synaptic adaptation occurs) and suggest that hippocampal neurons utilize an alternative endogenous fuel to support synaptic function during periods of hypoglycemia.

We next investigated putative metabolic substrates supporting synaptic potentials and levels of high-energy phosphates during repeated hypoglycemia. In the brain, glycogen is synthesized from glucose-6-phosphate, localized exclusively in glial cells, and utilized as the energy source for neuronal excitability and during glucose insufficiency (16–18). It is plausible that hippocampal slices recovering from transient glucose depletion absorb excess exogenous glucose and synthesize astrocytic glycogen, which is, in turn, utilized as the substrate to maintain synaptic activity during the subsequent hypoglycemia (Fig. 3A). To test this possibility, we measured glycogen and glucose in hippocampal slices during repeated hypoglycemia. Control glycogen and glucose concentrations were 224.9 ± 16.8 mmol glucosyl units/kg protein and 101.9 ± 6.6 mmol/kg protein, respectively. As shown in Fig. 3B, hippocampal glycogen promptly fell to a low level after initial glucose deprivation, and repeated glucose deprivation rapidly decreased glycogen to an almost identical level (P < 0.01) (Fig. 3B). In the presence of 0.5 mmol/l glucose in secondary hypoglycemic perfusion medium, glycogen fell to a similar low level (P < 0.01). It is not surprising that hippocampal glycogen concentration remained low after 30 min of glucose depletion, rather than dropping to 0, because glucose controls glycogen levels by binding to and inactivating the enzyme responsible for glycogen breakdown, phosphorylase A (17). Cultured astrocytes are not able to mobilize glycogen stores completely in the absence of glucose (17). Indeed, glucose levels were diminished to nearly 0 within 30 min during the initial and the secondary glucose deprivation periods (P < 0.01) (Fig. 3C). Glucose concentration in slices incubated in secondary hypoglycemic solution containing 0.5 mmol/l glucose also decreased almost entirely within 30 min (P < 0.01). These results indicate that glycogenolysis does not support hypoglycemic synaptic adaptation.

In the adult brain, monocarboxylates, including lactate, pyruvate, and β-hydroxybutyrate, can be used to produce high-energy phosphates (14,15,18–20) and protect against neuronal damage after hypoxic or ischemic insult (21,22). In the immature brain, monocarboxylates can be used to fuel synaptic transmission during periods of glucose deprivation (23). Therefore, we investigated whether monocarboxylates can be used to sustain synaptic activity in response to repeated hypoglycemia (Fig. 4B), using monocarboxylate transporter (MCT) inhibitors 4-CIN, IBMX, and phloretin (21). 4-CIN prevents monocarboxylate utilization by inhibiting the mitochondrial pyruvate carrier as well as by inhibiting the plasma membrane MCT. First, we tested the effect of 200 μmol/l 4-CIN on 10 mmol/l glucose-supported PS (Fig. 4A). Whereas 0.1% DMSO used for dissolving of 4-CIN did not affect PS amplitude (Fig. 4A, column 1), application of 200 μmol/l 4-CIN suppressed PS amplitude by 43.0 ± 6.8% (column 2) (P < 0.01). In contrast, 10 mmol/l glucose-supported PS recovered from transient glucose deprivation was suppressed by 22.4 ± 3.9% after the application of 200 μmol/l 4-CIN (P < 0.01) (Fig. 4C). After 10 mmol/l glucose-supported PS was stabilized in the presence of 4-CIN, extracellular glucose was removed (Fig. 4C). Secondary glucose depletion induced a depression in PS amplitude, which exhibited a similar time course to that observed in response to the initial glucose deprivation. Subsequent addition of 10 mmol/l glucose in the perfusion medium partially restored the PS amplitude.

The effect of IBMX on hypoglycemic synaptic adaptation was then tested. IBMX was dissolved in 3.5 mmol/l KOH, and 500 μmol/l IBMX was applied to the perfusion solution (Fig. 4A). Addition of 3.5 mmol/l KOH in standard solution increased the PS amplitude by 158.8 ± 1.9% (P < 0.01) (Fig. 4A, column 3), and additional administration of 500 μmol/l IBMX increased the amplitude to 198.2 ± 15.8% of the control amplitude (P < 0.01) (Fig. 4A, column 4). In hippocampal slices recovered from transient glucose deprivation, 500 μmol/l IBMX in 3.5 mmol/l KOH increased the amplitude to 183.2 ± 11.1% of the control value (P < 0.01) (Fig. 4C). Secondary glucose depletion in the presence of 500 μmol/l IBMX produced a decay of PS amplitude in a time course similar to that observed in the presence of 200 μmol/l 4-CIN. We also tested the effect of 200 μmol/l phloretin and found it to decrease synaptic activity similar to 4-CIN during secondary glucose depletion (data not shown).

We further examined the effect of 4-CIN on PS maintained in secondary hypoglycemic medium containing 0.5 or 1.0 mmol/l glucose (Fig. 5). After transient glucose deprivation, administration of 200 μmol/l 4-CIN in 0.5 or 1.0 mmol/l glucose solution reversibly suppressed PS amplitudes entirely or by 42.0 ± 10.6%, respectively (Fig. 5B).

Finally, we investigated whether exogenous lactate or pyruvate substituted for glucose could sustain synaptic transmission in slices recovered from transient glucose deprivation (Fig. 6). As we reported previously (15,20), replacement of extracellular glucose with Na lactate or Na pyruvate did not maintain synaptic activity (Fig. 6B). However, in hippocampal slices recovered from transient hypoglycemia, subsequent replacement of glucose with 10 mmol/l Na lactate or 10 mmol/l Na pyruvate did result in the maintenance of synaptic activity (Fig. 6B). These results indicate that synaptic adaptation to repeated hypoglycemia depends on the utilization of monocarboxylates in guinea pig hippocampal neurons.

In the present study, we report 1) a unique synaptic phenomenon whereby synaptic transmission becomes resistant to repeated hypoglycemia, i.e., hypoglycemic synaptic adaptation occurs; 2) hypoglycemic synaptic adaptation is not dependent on glycogenolysis as a metabolic substrate; and 3) synaptic utilization of monocarboxylates sustains synaptic function during repeated hypoglycemia. Adaptation of brain function is frequently postulated to explain the deteriorated neuroendocrine and neurobehavioral responses to recurrent hypoglycemia (3–5). To our knowledge, however, this is the first report showing that synaptic function is able to adapt to repeated hypoglycemia in hippocampus dissociated from cerebral vessels.

In this study, we demonstrated hypoglycemic synaptic adaptation in the granule cell layer of hippocampal dentate gyrus from the guinea pig brain. The hippocampus is particularly sensitive to hypoglycemia, as is the neocortex and the striatum, which have not been thought to adapt to hypoglycemia. In contrast, the hypothalamus, which contains glucose-sensing areas, and the thalamus, brainstem, and cerebellum are more resistant to hypoglycemia. It should be noted that synaptic activity in hypoglycemia-sensitive tissues is served by a protective mechanism against repeated hypoglycemia. Similar observations were reported recently by Hasuo et al. (24), who used optical recordings. In rat brain sections containing lateral septum and hippocampal area CA1, the initial exposure to exogenous glucose depletion suppressed synaptic activity, whereas activity remained unchanged in response to repeated hypoglycemia. Our findings, together with the observations of Hasuo et al., indicate that hypoglycemic synaptic adaptation is not a phenomenon specific to the hippocampal dentate gyrus, but rather is a synaptic phenomenon distributed perhaps throughout the central nervous system.

Previous in vitro experiments have shown that hypoxia-induced inhibition of synaptic transmission is associated with a 15–25% decline in ATP in brain slices (25,26). In our experiments, initial glucose deprivation decreased ATP and PCr concentrations by ∼15%, together with a complete decay of synaptic activity (Fig. 1). Our data indicate that synaptic function in the hippocampal dentate gyrus usually depends on exogenous glucose as an energetic substrate. However, in response to secondary glucose deprivation, high-energy phosphates of ATP and PCr were preserved during >120 min of observation, indicating that hippocampal neurons switch to an alternative endogenous fuel source during hypoglycemia. Glycogen or glucose does not seem to be used as an endogenous substrate, as demonstrated in Fig. 3. Other readily available substrates could be provided by citric acid cycle intermediates or free amino acids. However, the majority of these substrate pools are used during the first 5 min of severe hypoglycemia (27,28). Previous studies provided evidence that severe hypoglycemia induces a significant breakdown of phospholipids, concomitant with an increase in free fatty acids (28,29). The brain contains the enzyme systems required for metabolism of a range of different carbohydrate and lipid substrates (28). Substrates for these systems may serve as yet-undiscovered metabolic fuels during hypoglycemic synaptic adaptation.

Despite the sustained level of high-energy phosphates, synaptic activity was gradually eliminated during secondary glucose deprivation. In contrast, addition of 0.5 mmol/l glucose in the circulating medium after transient glucose deprivation sustained synaptic function and preserved high-energy phosphate levels (Fig. 2B). Alternative endogenous substrates were presumably used for production of high-energy phosphates. However, alteration of structural components, in addition to the absence of exogenous glucose, may result in the loss of synaptic activity.

Our results indicate that neurons use monocarboxylate as a main energetic substrate to sustain synaptic function during repeated hypoglycemia. We and others have reported that replacement of exogenous glucose with lactate, pyruvate, or β-hydroxybutyrate fails to maintain synaptic function (14,15,19,20,30,31). However, monocarboxylate (lactate or pyruvate) was able to maintain synaptic activity during secondary hypoglycemia (Figs. 4–6). These results suggest that temporary glucose deprivation triggers the utilization of monocarboxylates to sustain synaptic activity during hypoglycemic synaptic adaptation.

The concept of metabolic coupling between astrocytes and neurons has been investigated for decades and explains the metabolic compartmentation that occurs between astrocytes and neurons. In the honeybee retina, it has been demonstrated that glucose taken up by astrocytes from capillaries is converted glycolytically to lactate, which is then released into the extracellular space to be utilized by photoreceptor neurons during photostimulation (32). In mammalian neuronal tissues, there is accumulating evidence to support the hypothesis of metabolic compartmentation (21,32,33). The specific subcellular distribution of MCTs and lactate dehydrogenase (LDH) isoforms (LDH₁ and LDH₅) between neurons and astrocytes supports the molecular aspects of the astrocyte-neuron monocarboxylate shuttle (32). In the adult rat brain, eight MCT isoforms have been identified with differential affinities for monocarboxylates (21). LDH₅, which is present in tissues that produce lactate glycolytically, is localized predominantly to astrocytes, and LDH₁, which is enriched in tissues that use lactate as a substrate, is distributed mainly in neurons. These observations are consistent with the concept of lactate production by astrocytes and lactate utilization by neurons. The physiological prevalence of the astrocyte-neuron monocarboxylate shuttle may contribute to hypoglycemic synaptic adaptation.

Clinically, hypoglycemic unawareness is a complication of diabetes, especially in aged patients with cognitive deficits (1,2,34). In hypoglycemia unawareness, previous episodes of hypoglycemia cause delayed and diminished symptomatic and hormonal responses to subsequent hypoglycemia. Similar defects of neurohormonal responses to hypoglycemia are induced by intravenous infusion of lactate during hypoglycemia in nondiabetic individuals (35,36). These results suggest the possible involvement of lactate in hypoglycemic unawareness. However, the neural basis of brain adaptation to recurrent hypoglycemia is still unclear. The possible contribution of hypoglycemic synaptic adaptation to brain adaptation in response to recurrent hypoglycemia and the molecular mechanism underlying the activation of an astrocyte-neuron monocarboxylate shuttle after transient glucose deprivation will be addressed in future studies.

We thank Dr. Y. Okada, Professor Emeritus (Kobe University School of Medicine) for helpful discussions and technical assistance.