under normal circumstances, the human brain is dependent on glucose as its sole metabolic fuel source. Certain situations promote a significant change in the use of brain fuel substrates. Prolonged starvation in obese subjects, for instance, results in an increase in brain ketone uptake, partially restoring energy balance (25). Systemic lactate concentration is known to increase as plasma epinephrine levels rise during insulin-induced hypoglycemia (12) and brain lactate concentration increases during fasting (26). Both of these changes in lactate concentration could suggest a role for lactate as alternate brain fuel under certain conditions. Lactate infusion to physiological (22, 23) or supraphysiological (34) concentrations during insulin-induced hypoglycemia results in attenuated symptomatic, physiological, and counterregulatory hormone responses in both diabetic (23) and nondiabetic (22, 23, 34) subjects. During insulin-induced acute hypoglycemia in nondiabetic dogs, Nemoto et al. (24) demonstrated a 17% decrease in cerebral metabolic rate of oxygen utilization (CMRo₂), which returned to control levels with lactate loading to a supraphysiological concentration. Hernandez et al. (15) demonstrated an increase in cerebral lactate with preservation of a normal CMRo₂ during insulin-induced hypoglycemia in newborn dogs.

We have previously demonstrated (6) that, during acute hypoglycemia to 3 mmol/l, normal human subjects experience a 24% reduction in brain glucose uptake (BGU). Although it is known that the human brain has the catabolic enzymes necessary to oxidize a number of compounds (10), it is unknown whether during acute hypoglycemia the brain can utilize these substances to a sufficient degree to make up for the energy deficit caused by the reduction in BGU.

In this study, we examined the significance of lactate, alanine, and leucine in brain metabolism during hypoglycemia. Alanine was chosen for study because it is a known precursor to pyruvate in gluconeogenesis and it has an efficient system of transport across the blood-brain barrier (19). Prior work with cell preparations has shown that, under certain conditions, there is a metabolic coupling between glial cells and neurons, in which the glial cells supply alanine as a primary fuel source for adjacent neurons (10).

Leucine was chosen for study because it has a high-energy yield, a highly efficient system for transport across the blood-brain barrier, a significant cerebral uptake in both diabetic and nondiabetic subjects at euglycemia (27, 8, 14), and a rise in systemic concentration during hypoglycemia (16). However, hypoglycemia has recently been shown not to be a stimulus for increasing peripheral leucine production (4), and Wahren et al. (35), using a stable isotope technique that is theoretically independent of cerebral blood flow, suggested that net cerebral exchange for any individual amino acid, or lactate, failed to rise in normal humans during acute hypoglycemia. Conversely, Brosnan et al. (8) demonstrated a significant cerebral uptake of leucine in diabetic, hyperglycemic rats but normal leucine uptake in nondiabetic, euglycemic rats. Grill et al. (14) similarly demonstrated a significant cerebral uptake of leucine in both diabetic and nondiabetic humans during both hyperglycemia and euglycemia. Thus controversy exists as to whether changes in blood glucose effect brain leucine uptake (BLeuU).

The overarching hypothesis of this study was to test whether or not brain uptake of lactate, alanine, and/or leucine would increase during acute hypoglycemia at a rate sufficient to compensate for acutely reduced BGU. We find that none of these potential alternate fuels can account for more than 25% of the loss of ATP induced by reducing the systemic glucose concentration to 3.0 mmol/l in normal humans.

MATERIALS AND METHODS

Subjects.

Seventeen normal human volunteers (11 men, 6 women) participated in the study. Eleven were non-Hispanic whites, two were Hispanic, two were Native American, one was African American, and one was Asian American. The nature, purpose, and possible risks of the study were carefully explained to the subjects before they agreed to participate. The Human Research and Review Committee of the University of New Mexico (UNM) School of Medicine approved the protocols and informed consents. All subjects were recruited from flyers posted on campus at UNM, and all reported that they had no chronic or acute medical problems and were taking no regular medications. All underwent a history and physical examination ∼1–2 wk before the study, at which time a screening blood was collected for a chemistry panel, complete blood count, liver function tests, cholesterol level, and a quantitative β-hCG for women.

Experimental protocol.

All subjects were admitted to the UNM General Clinical Research Center the evening before the study. Subjects fasted for 8 h before the start of the study and continued fasting until the end. At 0600, an 18-gauge peripheral intravenous line was placed in or near the right antecubital fossa to subsequently deliver insulin and dextrose. At 0630, a 21-gauge radial arterial line was inserted under local lidocaine anesthesia. This was followed by passage of a retrograde internal jugular catheter to the level of the right jugular bulb. Our group and others have previously used roentgenography to demonstrate proper placement of the cannula at the jugular bulb by this technique (6, 28).

Measurement of cerebral blood flow.

For measurement of cerebral blood flow (CBF), subjects rested in a 1-ft³ plastic tent with a high flow of medical air flowing through it to prevent humidity from building up and to allow rapid introduction of the tracer gas. A mixture of 9% N₂O and 21% oxygen was then delivered for 25 min while simultaneous arterial and venous blood samples for N₂O were collected. We have noted that the shorter time frames suggested by Kety and Schmidt (18) may fall short of allowing equilibrium to be established in some subjects. Samples for arterial and venous N₂O concentrations were collected at 30-s intervals for 6 min and then every 2 min until 25 min.

Brain glucose uptake (BGU) is the product of the CBF and the arteriovenous difference (A-V)_diff for glucose across the brain. Utilization of other brain substrates were similarly determined. As with other substrates, brain oxygen uptake (CMRo₂) was calculated by multiplying the CBF by the (A-V)_diff for oxygen content across the brain. CBF was determined using Fick methodology, utilizing the equation developed by Kety and Schmidt (18): CBF = 100V_uS/(∫₀^u [A-V]dt). The arterial concentration rises to equilibrium values more rapidly than the venous compartment, and the area between the two curves is inversely related to the CBF. V_u in the equation is the venous equilibrium concentration, and S is the partition coefficient for N₂O between the blood and the brain at equilibrium. S is known to be 1.0 in humans (17). A Statistical Analysis Software (SAS) program (32) was written to perform multiple iterations to fit the best sigmoidal curve to each data set as previously described (7). The program forces convergence at the final equilibrium point. The coefficients derived from the iterative process were used to calculate theoretical y-axis values at 0.1-min intervals, and the area between these finite increments was calculated so that sequential trapezoidal summation yielded the total area between the curves.

An hour after all lines had been inserted, baseline blood samples were collected from the radial artery and internal jugular lines for lactate, alanine, and leucine concentrations as well as oxygen saturation. CBF was then measured to permit calculation of brain uptake of lactate, alanine, leucine, and glucose. Over the next 30 min, the subject's plasma glucose was lowered to ∼3.0 mmol/l by continuous infusion of regular human insulin (1.5 mU·kg⁻¹·min⁻¹). A variable-rate 20% dextrose infusion was utilized to maintain target plasma glucose level. Arterial plasma glucose was measured every 7 min on a Beckman glucose analyzer (Beckman Instruments, Fullerton, CA), and the investigators adjusted the dextrose infusion rate to maintain target concentrations. The target plasma glucose level was maintained for 90 min to allow glucose and lactate concentrations in the brain to reach equilibrium with systemic levels. Time spent at our target glucose level of 3.0 mmol/l is similar to that in previous human studies (12, 22–34). Neither blood insulin nor glucose counterregulatory hormone levels were measured.

After the induction of sustained hypoglycemia, CBF and cross-brain substrate differences were again measured. The glucose concentration was then recovered to normal, all lines were removed, and the patient was given a meal before being discharged.

Estimation of energy lost and gained by changes in substrate uptake.

Energy lost or gained by glucose and lactate in the brain was calculated in micromoles of ATP per 100 grams of brain per minute. Glucose oxidation through glycolysis, the tricarboxylic acid (TCA) cycle, and other associated oxidative phosphorylation yields 30 μmol ATP/μmol glucose (33). Lactate oxidation, via conversion to pyruvate and processing through the TCA cycle, yields 17 μmol ATP/μmol lactate, as reported by Magistretti and Pellerin (21). This ratio is more current than the value of 15 μmol ATP/μmol lactate reported by Albritton (2).

Analytical methods.

Plasma glucose was measured by a glucose oxidase technique (Beckman Instruments). N₂O content was measured with an infrared spectrophotometer designed to detect trace amounts of N₂O in parts per million reported in single whole integers (Transverse Medicinal Monitors, Saline, MI), after the technique of Robertson et al. (31). Plasma lactate was measured using the technique of Lloyd et al. (20). Alanine and leucine concentrations were measured from derivatives of t-butyldimethylsilyation followed by gas chromatographic-mass spectrometric analysis using electron impact ionization (29). The internal standard used was norleucine. The samples were run on a Hewlett Packard Mass Selective Detector no. 5972 with an HP 5890 gas chromatograph and a 7673A automatic sampler (Hewlett-Packard, Wilmington, DE). Oxygen content (%saturation and hemoglobin concentration) was measured using an IL 1400 (Instrumentation Laboratories, Lexingtion, MA).

Statistical analysis.

Comparisons of parameters of interest were made with a two-sided paired t-test. Data sets not normally distributed were analyzed with the nonparametric Wilcoxin signed rank test.

RESULTS

All results, including results in the figures, are reported as means ± SE. Mean arterial blood glucose levels were 5.4 ± 0.1 (euglycemia) and 3.0 ± 0.1 mmol/l (hypoglycemia). Figure 1 shows the values obtained for CBF, (A-V)_diff for glucose, and BGU during euglycemia and hypoglycemia. CBF was comparable at both glucose concentrations: 57.9 ± 2.9 vs. 58.0 ± 3.8 ml·100 g of brain⁻¹·min⁻¹, euglycemia vs. hypoglycemia (P = 0.98). (A-V)_diff for glucose decreased significantly, from 0.62 ± 0.03 to 0.46 ± 0.02 mmol·100 g⁻¹·min⁻¹ (P < 0.001), and BGU decreased significantly from 36.3 ± 2.6 to 26.6 ± 2.1 μmol·100 g⁻¹·min⁻¹ (P < 0.001). Values for CMRo₂ can be seen in Fig. 2. Both (A-V)_diff for O₂ and CMRo₂ displayed a slight downward trend but did not change significantly. (A-V)_diff for O₂ values were 2.8 ± 0.2 vs. 2.6 ± 0.2 mmol·100 g⁻¹·min⁻¹ (P = 0.13), and CMRo₂ values were 164 ± 17 vs. 151 ± 13 μmol·100 g⁻¹·min⁻¹ (P = 0.32), euglycemia vs. hypoglycemia. At the same time, as seen in Fig. 3, mean brain lactate uptake (BLU) changed from a net release of −1.8 ± 0.6 to a net uptake of 2.5 ± 1.2 μmol·100 g⁻¹·min⁻¹ (P < 0.001), whereas mean brain leucine uptake (BLeuU) decreased from a net uptake of 5.5 ± 1.0 to 1.9 ± 0.4 μmol·100 g⁻¹·min⁻¹ (P < 0.003), and mean brain alanine uptake (BAU) tended to decrease from 2.1 ± 1.8 to −1.9 ± 1.6 μmol·100 g⁻¹·min⁻¹, although this was not statistically significant (P < 0.08). Arterial lactate concentration rose from 0.85 ± 0.09 to 1.90 ± 0.15 mmol/l (P < 0.001), whereas the arterial concentrations of both alanine and leucine decreased from 0.32 ± 0.02 to 0.26 ± 0.02 mmol/l (P < 0.009) and from 0.11 ± 0.01 to 0.03 ± 0.003 mmol/l (P < 0.001), respectively.

DISCUSSION

In agreement with our prior investigations (6), in the face of a 27% decline in the uptake of glucose by brain, we observed no significant change in CMRo₂ during hypoglycemia in normal humans, although CMRo₂ did display a slight downward trend from 164 ± 17 to 151 ± 13 μmol·100 g⁻¹·min⁻¹, euglycemia to hypoglycemia, (P = 0.32). Our values of 164 and 151 μmol·100 g⁻¹·min⁻¹ fall within the range of normal values seen in the literature (7, 13, 15, 24). Two basic possibilities exist to explain these facts: 1) fuels other than glucose were being oxidized as provision of glucose from the circulation fell, or 2) there was a change in the rate of glucose oxidation (either aerobic or anaerobic). The brain relies on anaerobic glycolysis, as ∼15% of brain glucose is converted to lactate and does not enter the TCA cycle (30). The potential exists that more pyruvate could be diverted to the more energy-yielding (and oxygen-utilizing) TCA cycle as hypoglycemia develops. Fox et al. (13) demonstrated a dramatic increase in BGU assessed by 18-fluorodeoxyglucose positron emission tomography with visual stimulation. In the face of this dramatic increase in glucose utilization, a minimal increase in CMRo₂ occurred, indicating an increase in anaerobic glycolysis. As such, one might envision a decrease in BGU during hypoglycemia being associated with a fall in anaerobic glycolysis and a rise in TCA oxidation. We did not measure changes in rates of glucose oxidation (either aerobic or anaerobic); therefore, we cannot be certain whether or not a change in the percentage of each possible metabolic process for glucose utilization may have occurred. The brain uptake of at least one of the substrates we measured did increase during hypoglycemia.

The rate of BLuU increased whereas BLeuU fell, and BAU remained essentially unchanged. Quantitatively, the reduction in BGU of 9.7 μmol·100 g⁻¹·min⁻¹ observed in the current investigation translates into a fall in ATP generation of 291 μmol·100 g⁻¹·min⁻¹. On the other hand, uptake of lactate by the brain increased from a net loss of −1.8 to a net gain of 2.5 μmol/100 g/min. Such a change translates into an absolute increase in ATP of 74 μmol·100 g⁻¹·min⁻¹, or 25% of the energy deficit incurred from diminished BGU. Although the pharmacological elevation of lactate concentration has been shown to decrease physiological symptoms of hypoglycemia and attenuate the counterregulatory hormonal response, a normal, acute elevation of endogenous lactate production during hypoglycemia (and consequent increase in BLU) is inadequate to satisfy the loss in brain energy metabolism from reduced glucose uptake.

One limitation to our investigation is that the arteriovenous differences of any of the substrates were small and we could have missed a lesser change. Others, like Avogaro et al. (3), have generated large arterial concentrations, and consequently larger arteriovenous differences, which lead to dramatic increments in BLU. This led them to postulate that lactate could potentially replace glucose as the primary brain energy substrate during hypoglycemia. Such findings stimulated the first investigations in humans that demonstrated that infusion of exogenous lactate or ketones could partially attenuate the counterregulatory response associated with insulin-induced hypoglycemia and help prevent cerebral dysfunction in type 1 diabetes, presumably through increased lactate utilization for fuel (12, 22, 34).

Lactate is generated locally by glial cells and presented to neurons, serving as a primary neuronal fuel (10). Brain extracellular fluid lactate concentrations rise during hypoglycemia (1), suggesting that there is either increased lactate production or decreased lactate clearance. Given that neuronal tissue appears to utilize lactate, it would seem illogical to conclude that lactate uptake would decrease in the face of an energy deficit. Therefore, brain lactate utilization (or uptake from the circulation as we have shown) should be expected to increase during hypoglycemia.

The possibility remains that exogenous insulin may have influenced our results. We did not measure blood or brain insulin levels, and it is therefore difficult to speculate regarding any effect resulting from exogenous insulin administration. Certainly the brain has large numbers of insulin receptors, but their function is probably not related to glucose transport or utilization, although controversy still exists on this issue (5, 31).

One might speculate that brain glycogen could serve as a transient source of fuel during acute hypoglycemia and, if oxidized, would lead to stable oxygen utilization in the face of reduced brain glucose uptake from the circulation. Recent NMR spectroscopy data have suggested that the amount of brain glycogen available for mobilization for energy during hypoglycemia may be larger than previously believed (9), and this may contribute to the observed increase in brain extracellular lactate such as that seen by Abi-Saab et al. (1). Choi et al. (9) demonstrated depletion of brain glycogen during insulin-induced hypoglycemia in rats, concluding that brain glycogen may play a significant role in brain metabolism during acute hypoglycemia. Furthermore, they demonstrated a supercompensation of brain glycogen generation following a single episode of hypoglycemia, opening the door to the possibility that clinical recurrent hypoglycemia in patients with diabetes may result in greater and greater ability to compensate for subsequent hypoglycemia. Thus glycogen breakdown and combustion could also explain why we failed to observe a fall in CMRo₂. No human investigations that demonstrate utilization of brain glycogen exist.

We documented a significant decrease in BLeuU during hypoglycemia in normal humans. In addition to finding a decrease in BLeuU, we found a significant decrease in arterial leucine levels during hypoglycemia vs. euglycemia, a finding also noted by Battezzati et al. (4). Thus diminished substrate delivery from decreased peripheral production seems the likely explanation for the reduced rate of BLeuU. Although we observed no significant change in BAU during hypoglycemia, there was a tendency for BAU to decrease (P < 0.08), whereas arterial plasma alanine significantly decreased (P < 0.009). Davis et al. (11) studied levels of numerous endogenous compounds in dogs during hypoglycemia. Their results were similar to ours, with arterial blood alanine decreasing and arterial blood lactate increasing significantly during both peripheral and head insulin infusions.

In conclusion, we find that that none of the potential alternate fuels we evaluated could account for more than 25% of the loss of ATP induced by reducing the systemic glucose concentration to 3.0 mmol/l in normal humans. The applicability of our findings to the setting of humans with type 1 diabetes who experience recurrent bouts of hypoglycemia is limited. Improved uptake and utilization of alternate fuels in response to repeated episodes of neuroglycopenia would be metabolically advantageous; however, this hypothesis remains to be tested.

GRANTS

This research was supported by National Institute of Neurological Disorders and Stroke Grants 1 R29 NS-29972-01A1 and K24 NS-02097-05, by dedicated Health Research Funds of the University of New Mexico, and by a grant from General Clinical Research Program, Division of Research Resources 5 M01 RR-00997-28.

FOOTNOTES

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.