The study of brain energy metabolism has taken second place to that of muscle ever since the dawn of this field of research. Consequently, each new discovery made using muscle tissue that advanced our understanding of the biochemistry of energy metabolic processes was attempted to be duplicated in brain tissue. It was only when the brain\'s high energy needs were recognized that researchers realized its vulnerability to any mishap in its energy supplies and that this vulnerability may play a role in various brain disorders. Understanding of the mechanisms by which the brain deals with energy shortage is of utmost importance in shedding light on the fundamentals of brain disorders and their potential treatment. To achieve such understanding, accurate measurement of brain energy metabolic rates is necessary. This chapter summarizes the history of the current knowledge of the biochemical processes responsible for the production of adenosine triphosphate (ATP) in the brain. It briefly reviews the various techniques used to measure cerebral metabolic rates of oxygen (CMRO2) and glucose (CMRglucose), and elaborates on the potential of measuring the cerebral metabolic rate of lactate (CMRlactate) to improve our understanding of brain energy metabolism.
- BOLD fMRI
- cerebral metabolic rate
- paradigm shift
1. A short review of brain energy metabolism research
1.1 The first eight decades (1900–1980)
Most human cells produce adenosine triphosphate (ATP) via two, mostly interconnected biochemical pathways, glycolysis and mitochondrial oxidative phosphorylation. Erythrocytes (red blood cells, RBCs) produce their ATP via the glycolytic pathway alone, since they lack mitochondria. Under anaerobic conditions, the less efficient glycolytic pathway is the main source of ATP supply, since without oxygen the oxidative phosphorylation pathway cannot be maintained. Throughout the first half of the twentieth century the majority of the researchers in the field of energy metabolism made muscle their tissue of choice for the study of energy metabolism. While muscle was believed to require a great deal of energy to perform its work, the brain was assumed to be a low consumer of energy, as indicated by the following quote: “
1.2 The last four decades (1980–2018)
As indicated above, questions as to the correctness of the old description of glycolysis began to appear in the late 1980s. The lactic acidosis hypothesis of delayed neuronal damage  had a strong following at the time. This hypothesis postulated that lactate accumulation in the ischemic brain is the cause of delayed neuronal damage, damage observed after a recovery from the original ischemic insult had occurred. The popularity of this hypothesis was so strong that any insinuation that lactate could be anything but a menacing factor in cerebral ischemia aroused great skepticism. Consequently, when Fox et al.  published a study, which indicated that neural activation does not require an increase in energy supply and is supported by a mere non-oxidative glucose utilization (“anaerobic” glycolysis), they met with both doubt and a degree of cynicism. Almost simultaneously, Schurr et al.  published their findings demonstrating that neuronal function
To answer this question, one must consider the mounting evidence supporting a paradigm shift in our comprehension of the glycolytic pathway . The shift entails redrawing the glycolytic pathway as one consisting of 11 steps, beginning with glucose as its substrate and ending with lactate as its end-product, independent of the presence or absence of oxygen. From its inception, distinguishing between aerobic and anaerobic glycolysis was based not on specific evidence that the two pathways exist and produce two different products, pyruvate and lactate, respectively. That separation was an attempt by the pathway’s elucidators to somehow fit it into, what they concluded, is an outcome that produces pyruvate as its main end-product since they accepted Kreb’s suggestion that this monocarboxylate is the substrate of the TCA cycle. It must have been relatively easy to accept that suggestion considering lactate’s negative reputation . Hence, the glycolytic pathway should be considered one, uninterrupted chain of biochemical reactions that begins with glucose and ends with lactate (Figure 1). Accordingly, its last reaction (number 11), the reduction of pyruvate to lactate by the cytosolic lactate dehydrogenase (cLDH), plays a crucial role in keeping this pathway’s cyclical nature operational as it provides a continuous supply of NAD+. If pyruvate was the glycolytic end-product, NAD+ would have to be imported from other sources and locations, a proposition that has offered a somewhat shaky resolution (see  and references within). This, of course, is not the only factor that justifies a paradigm shift. There are numerous studies published over the past two decades demonstrating the presence of lactate dehydrogenase in mitochondria (mLDH), an enzyme that converts lactate to pyruvate [26, 27, 28, 29, 30, 31, 32]. Brooks et al.  also demonstrated the presence of monocarboxylate transporter 1 (MCT1) in mitochondria, the transporter that is responsible for the transport of lactate along its gradient from the cytosol to the mitochondrion. Havel et al. showed that in blood and in other tissues the ratio lactate/pyruvate is >10, a value that is not consistent with the assumption that pyruvate is the glycolytic end-product . Moreover, the proposal that aerobic glycolysis ends with pyruvate does not meet the known standard free-energy (∆G0’) change of the reaction pyruvate → lactate, which is −6.0 kcal/mol, a value indicating that this reaction should proceed independently of the presence or absence of oxygen. In other words, glycolysis, whether aerobic or anaerobic, should always end up with lactate. Figure 2A demonstrates the free energy change profile of aerobic glycolysis that ends with the reaction phosphoenolpyruvate → pyruvate, although the potential free-energy change of the conversion pyruvate → lactate (Figure 2B, anaerobic glycolysis) determines that glycolysis should end with lactate regardless of the oxygenation condition (Figure 2C). Last but not least, the reaction equilibrium of cLDH is tilted heavily in the direction of lactate production, which makes it unlikely for lactate to be converted back to pyruvate by that cytosolic enzyme. In contrast, the reaction equilibrium of mLDH tilts in the direction of lactate oxidation to pyruvate [35, 36].
The above points support the proposed paradigm shift in the glycolytic pathway , where lactate, not pyruvate, is its end-product and the oxidative mitochondrial substrate for the TCA cycle. Accordingly, is measuring the cerebral metabolic rates of oxygen () and glucose (CMRglucose) sufficient in providing an accurate picture of brain energy metabolism during rest or activation, in health or disease? If lactate is an oxidative energy substrate, should not CMRlactate also be measured in order to have a more complete account of cerebral energy metabolism? How would the measurement of CMRlactate contribute to our understanding of the brain’s ability to handle its energy demands under those conditions?
2. Measurement of cerebral energy metabolic rates
At the basis of each technology designed to measure the rate of brain energy metabolism is the idea that measuring the consumption rate of the main two substrates of glycolysis and mitochondrial respiration, glucose and oxygen (O2), should provide a complete picture of the brain’s energy use. Theoretically, under normal conditions, each glucose molecule that enters the glycolytic pathway requires six molecules of oxygen to be fully oxidized via the mitochondrial TCA cycle and the electron transport chain. Thus, simultaneous measurements of glucose and oxygen consumption during rest or activation supposedly produces accurate estimate of the energy needs for the brain region under observation. However, the ratio /CMRglucose values calculated are often significantly smaller than the expected 6/1. Such discrepancies have attributed to other glucose-consuming reactions not accompanies with oxygen consumption. Consequently, it has been a common understanding that a value of /CMRglucose < 6 indicates that a partial non-oxidative glucose consumption. The smaller the value of /CMRglucose, the greater is the non-oxidative consumption of glucose. This understanding makes sense when one assumes that a fully coupled glycolytic-mitochondrial respiratory apparatus should produce a /CMRglucose value of 6 and an uncoupled apparatus (non-oxidative) should produce a /CMRglucose value of ~0.
As indicated above, myriad techniques and technologies have been developed during the past six decades to measure both and CMRglucose. To measure cerebral energy metabolism
In principle, CMR can be expressed as: CMR = CBF (A − V).
where (A − V) is the difference in concentration in arterial and cerebral venous blood, CBF is the rate of cerebral blood flow in volume of blood per unit time, a CMR (cerebral metabolic rate) is the steady state of utilization or production of a substance by the brain . This equation is the foundation on which quantitative CMR studies
In this respect, Hu and Wilson  studied the coupling of a temporary local energy pool to neuronal activity in the rat brain (Figure 3). They were the first to combine the use of three separate rapid response sensors (microelectrodes) to measure tissue oxygen, glucose and lactate concentrations. The investigators placed them in the dentate gyrus of the rat hippocampus, observing how they fluctuate in response to 10 consecutive electrical stimulations of the perforant pathway (each stimulus lasted 5 s and applied every 2 min). Their results were analyzed by Schurr and Gozal  (Figure 3). A literature search shows that Hu and Wilson’s interpretation of their findings has its supporters [15, 42, 43, 44, 45, 46, 47] and detractors [48, 49, 50]. The former group argued that these findings are strengthening the concept that lactate is the energy substrate that is utilized oxidatively upon neuronal activation. The latter group disagreed with this conclusion. Schurr  further analyzed the results of Hu and Wilson  beyond an earlier analysis . The more recent analysis was prompted for two reasons. First, two decades have passed since the publication of the paper by Hu and Wilson , a period in which numerous studies added much support to the idea that lactate is a mitochondrial oxidative energy substrate. Second, many other studies on cerebral energy metabolism continue to conclude that neural activity is supported by “anaerobic” glycolysis and not by oxidative utilization of glucose, while ignoring the possibility that such activity may be supported by oxidative utilization of lactate.
3. Lactate cerebral metabolic rate and the importance of its measurement
When the results of the study by Hu and Wilson  were analyzed before , the analysis showed that upon a series of 10 stimulation of the rat hippocampal perforant pathway a steady glucose consumption was accompanied by a gradual increase in lactate consumption. Considering the conclusion of Fox et al.  that aerobic glycolytic ATP production is sufficient to answer the energy needs of activated neural tissue, one could assume that it should be sufficient to provide the energy needs of the stimulated hippocampal dentate gyrus. In addition, this analysis points out that if the conclusion of Fox et al.  is correct, the energy needs of the activated dentate gyrus declined with each stimulation or stayed the same at a very low level of ATP production (0.8–0.3 mM). However, if lactate oxidative consumption is postulated to be responsible for the ATP production that sustains the energy needs of the stimulated tissue, the increased lactate consumption with each consecutive stimulation signals a concomitant increased ATP production. The calculation shows that the response to the first stimulation produced 3 mM ATP, while the response to the last stimulation produced almost 11 mM (Figure 4).
The more recent analysis  also indicates that the increased levels of tissue lactate following each stimulation  could not be produced from the glycolytically metabolized glucose (Figure 5). Hence, this additional lactate had to be originated from other sources i.e., the surrounding tissue or glycogen stores . As was shown by Hu and Wilson (Figure 3), a larger amount of lactate was consumed during each consecutive stimulation, while a smaller amount of glucose was consumed. Moreover, following each stimulation, except the first one, the lactate level measured exceeded the level expected from the amount of glucose consumed glycolytically i.e., two moles of lactate from one mole of glucose. Following the second stimulation, the tissue ratio of lactate to glucose was 3.95 and by the 10th stimulation this ratio increased to 8.33 (Figure 5). Meanwhile, oxygen tissue levels dipped and rose as expected during and after each stimulation, respectively, signaling that the electrical stimulation evoked an oxidative consumption of substrate. Initially, glucose and lactate were oxidatively consumed at equal amounts however, from the second stimulus onward more lactate than glucose was consumed (Figure 5). The oxygen level as measured by Hu and Wilson  fluctuated with a dip upon stimulation and a sharp rise upon its cessation (Figure 3). The fast rise can be interpreted as evidence that ample oxygen was available if and when needed. This rise also indicates that the tissue was well oxygenated during the duration of the experiment. Considering that one mole of lactate consumes three moles of oxygen for its full oxidation as compared to six moles of oxygen consumed by glucose for its full oxidation, if lactate, rather than glucose, is the main oxidative energy substrate during neural tissue activation, the expected ratio :CMRlactate should not exceed 3:1. Therefore, it is reasonable to presume that during neural activation, when lactate oxidation is a major supplier of the ATP necessary to support said activation, the ratio :CMRglucose should be considerably lower than 6:1. Obviously, most, if not all, studies aimed at measuring cerebral metabolic rates postulate that the ratio :CMRglucose measured or calculated should approach 6 . The conclusions of Fox et al.  are in complete disagreement with the measurements and calculations of Hyder et al. . While the former concludes an almost complete uncoupling between glucose and oxygen consumption by activated neural tissue, the latter asserts the maintenance of full coupling between glucose and oxygen consumption during neural activation. In both studies [21, 23] the interpretation of the results is based entirely on the original, dogmatic paradigm of glycolysis according to which aerobic glycolysis ends with pyruvate, the assumed substrate of the mitochondrial TCA cycle.
While BOLD fMRI estimates yielded a :CMRglucose ratio value of 6:1, the 15O measurements produced a ratio value of 0.4:1. These completely opposing outcomes make one wonder whether or not the measurements performed using these two methods and the calculated values of they produced truly reflect the changes in the consumption of molecular oxygen upon neural activation. Could the direct measurements of , CMRglucose and CMRlactate done by Hu and Wilson  along with the indirect ones made by Fox et al.  and Hyder et al. , be reconciled such that a better picture of cerebral metabolic rates of activated neural tissue can be visualized? It is widely agreed that over 90% of the normal brain’s energy production originates from glucose oxidation [21, 51]. The normal glucose concentration in the brain is ~2 mM and its normal lactate concentration is about half of that of glucose. Thus, it is safe to postulate that the normal resting brain is supplied with ample amounts of oxygen to continuously oxidize more than 90% of the brain glucose. However, glucose supplies to the normal brain are limited (only 40% of normal blood glucose level). Consequently, the increased rate of CBF along with the increased consumption of glucose upon activation [21, 23, 52] should supply all the oxygen necessary to match the increased demand, in contrast to the limited supplies of glucose. Low resolution techniques for the measurements of oxygen concentrations are unable to detect local fluctuations accurately if at all, which could explain why Fox et al.  reached the conclusion regarding the very low oxygen consumption during neural activation. Nonetheless, their conclusion that the energy demands of activated neural tissue are being met through glycolytic ATP production is most likely incorrect. In other words, undetectable or slightly detectable dip in tissue oxygen level upon activation is not necessarily an indication that oxygen is not consumed. The higher resolution of oxygen measurement afforded by polarography exemplifies the fact that local oxygen levels dipped upon stimulation and overshot upon its cessation (; Figures 3 and 4). Although local fluctuations in tissue oxygen levels were evident, its overall tissue concentration did not significantly change and may even have risen somewhat above its baseline level. In contrast, both glucose and lactate levels were changed significantly from their baseline levels [24, 36, 41] (Figures 3–5). The fluctuations between lactate and oxygen were highly synchronized, indicating that lactate is being oxidized upon tissue activation. During the 20 min following the 10th stimulation, the tissue level of both oxygen and glucose climbed above the baseline level, while the high level of lactate gradually declined (; Figure 3). These shifts seem to indicate that upon cessation of stimulation, as the tissue is recovering from activation and high energy demands, lactate becomes the preferred oxidative energy substrate, sparing glucose. That the cerebral tissue would prefer lactate over glucose, especially when the former is abundant, is reasonable, considering the fact that lactate oxidative mitochondria. Consequently, the increased rate of CBF along with the increased consumption of glucose upon activation [21, 23, 52] should supply all the oxygen necessary to match the increased demand, in contrast to the limited supplies of glucose. Low resolution techniques for the measurements of oxygen concentrations are unable to detect local fluctuations accurately if at all, which could explain why Fox et al.  reached the conclusion regarding the very low oxygen consumption during neural activation. Nonetheless, their conclusion that the energy demands of activated neural tissue are being met through glycolytic ATP production is most likely incorrect. In other words, undetectable or slightly detectable dip in tissue oxygen level upon activation is not necessarily an indication that oxygen is not consumed. The higher resolution of oxygen measurement afforded by polarography exemplifies the fact that local oxygen levels dipped upon stimulation and overshot upon its cessation (Figures 3 and 4 and ). Although local fluctuations in tissue oxygen levels were evident, its overall tissue concentration did not significantly change and may even have risen somewhat above its baseline level. In contrast, both glucose and lactate levels were changed significantly from their baseline levels (Figures 3–5 and [24, 36, 41]). The fluctuations between lactate and oxygen were highly synchronized, indicating that lactate is being oxidized upon tissue activation. During the 20 min following the 10th stimulation, the tissue level of both oxygen and glucose climbed above the baseline level, while the high level of lactate gradually declined (Figure 3 and ). These shifts seem to indicate that upon cessation of stimulation, as the tissue is recovering from activation and high energy demands, lactate becomes the preferred oxidative energy substrate, sparing glucose. That the cerebral tissue would prefer lactate over glucose, especially when the former is abundant, is reasonable, considering the fact that lactate oxidative utilization, in contrast to glucose, does not involve ATP investment prior to its utilization by mitochondria.
4. CMRs measurements and their possible implications in brain disorders
Energy metabolic interruptions are at the basis of several brain disorders and measuring CMRs of patients inflicted by such brain disorders can offer a potentially better diagnosis and treatment. Measurement of and CMRglucose have been performed regularly in numerous studies of cerebral ischemia in an effort to better understand the mechanisms of neuronal ischemic damage. CMRglucose measurement has been used in studying obsessive-compulsive disorder, mood disorder and depression, where the main aim is to follow changes in glucose metabolism in specific brain regions believed to be involved in these disorders. Other brain disorders where glucose metabolic rate has been measured in include amyotrophic lateral sclerosis, Alzheimer’s disease, epilepsy, Parkinson’s disease and Huntington’s disease. The purpose behind the measurement of glucose cerebral metabolic rate when investigating diseases and disorders is usually to identify brain regions that are involved in a given disorder or disease, not to investigate how energy metabolism is being affected by the disease or the disorder. Also, the energy metabolic rates of brain tumors have received great attention due to the unique energy requirements of these tumors. Nevertheless, cerebral ischemia and traumatic brain injury (TBI) are the two disorders for which measurements of and CMRglucose are most abundant. The results of these measurements prompted proposals both for treatments and mechanisms of neuronal damage due to these insults. The most heralded hypothesis attempting to explain delayed neuronal cerebral ischemic damage  known as the lactic acidosis hypothesis, postulated the accumulation of lactic acid as the cause of that damage. Consequently, physicians dealing with stroke patients were encouraged to control blood glucose levels in these patients, based on the assumption that the higher the glucose level during cerebral ischemia, the higher the level of lactic acid produced and the damage it causes. The lactic acidosis hypothesis was discarded, although the practice of controlling the blood glucose level of stroke patients remained. To this end, lactate was shown to support neuronal recovery post-ischemia
A paradigm shift of a scientific model should, naturally, result in reconsideration of hypotheses and concepts that have been formulated on its foundation prior to its shift. The understanding of cerebral metabolic rates of energy substrates during rest and activation of neural tissue, the use of the method best suited for the measurement of these rates and the interpretation of the results have always relied on two fundamental assumptions. First, cerebral energy metabolism includes the obligatory glycolytic breakdown of glucose to pyruvate and the utilization of the latter by the mitochondrial TCA cycle and the electron transport chain with oxygen as its final receptor. Second, the activation of cerebral tissue is sustained by an increase in ATP production and therefore an increase in the consumption of glucose and oxygen. Two seminal papers that were published almost simultaneously [16, 21] have forced biochemists, and especially neuroscientists, to reassess these two basic postulates. The paper by Fox et al.  has perplexed many with its conclusion that the energy requirements of activated neural tissue are minimal and can be fulfilled by the glycolytic pathway alone (glucose → lactate + 2ATP). The paper by Schurr et al.  provoked great skepticism upon demonstrating that neural tissue can function and be activated when lactate is its sole oxidative energy substrate (lactate + 3O2 + mitochondria → pyruvate → TCA cycle → 3CO2 + 3H2O + 17ATP). While the proposal that lactate is a suitable oxidative energy substrate had faced strong skepticism for many years, it has gained greater support over the past three decades. The proposal that glycolysis could be served as the sole supplier of energy for the activated neural tissue still divides scientists working in this field. By accepting the proposed paradigm shift of glycolysis  and its application in the interpretation of the results obtained by Fox et al. , Hyder et al. [23, 67], Hu and Wilson  and many others, a scenario can be drawn where lactate is supplementing most if not all the energy requirements of activated neural tissue. The data and the line of reasoning presented here strongly argue against the conclusion that these energy requirements are solely fulfilled by glycolysis. Future studies of activated cerebral metabolic rates should include, along with the measurements of and CMRglucose, the measurement of CMRlactate. Resolving the existing debated issues of cerebral energy metabolism is paramount for our better understanding the many brain diseases and disorders. Hopefully, this chapter provides a possible resolution of some of these issues.