Aerobic Glycolysis in Schizophrenia: Developmental Rescue or Energetic Breakdown?

Abstract

Aerobic glycolysis (AG) refers to the preferential use of glucose through glycolysis and not oxidative phosphorylation (OxPhos) despite the presence of oxygen. Originally described in cancer cells as the Warburg effect, AG is now recognized as a broader physiological mechanism extending beyond cancer biology. This process is less efficient than OxPhos in terms of ATP yield, but supports biosynthesis, neural plasticity, oxidative stress reduction, and synaptogenesis under metabolically demanding conditions. Building on this physiological role and current findings, we propose that in schizophrenia (SZ), AG remains elevated in adulthood, likely reflecting a compensatory response to reduced brain biomass and mitochondrial dysfunction. Neuroimaging, spectroscopy, and postmortem studies link the presence of AG in the brain to impaired OxPhos, reductive stress and defective neuron-glia coupling. Oligodendrocyte dysfunction and white matter damage also additionally compromise energy homeostasis and connectivity. Though AG supports repair, its persistent activation may destabilize synaptic structures and function. This Perspective proposes that AG in SZ represents a mismatch between developmental demands and adult metabolic function, compensatory in early stages but ultimately (mal)adaptive. Understanding when, where and why AG persists may reveal new entry points for restoring energetic balance in vulnerable brain circuits.

Introduction

Aerobic glycolysis (AG) is a metabolic process in which cells convert glucose into lactate or other molecules rather than fully supplying oxidative phosphorylation (OxPhos) in the presence of oxygen, a stoichiometrically inefficient process yielding only 2 ATP per glucose molecule. This process increases glucose uptake and lactate secretion to meet biosynthetic and redox demands. Originally identified in cancer cells [1], AG is currently understood to play a role in various physiological processes, such as embryonic development and neuronal plasticity. Recent findings implicate that AG may also play a significant role in the pathophysiology of schizophrenia (SZ), a disorder characterized by widespread brain mitochondrial impairment and metabolic reprogramming [2]. Some investigators have recently pointed out that the presence of oxygen during physiologic reliance on glycolysis is not the critical parameter and therefore AG is an inaccurate term. Hyperglycolysis has been proposed in its place [3, 4]. Nonetheless, we use the term AG in this article in keeping with historical practice in the literature.

In SZ, mitochondrial defects such as downregulated pyruvate dehydrogenase (PDH) activity [5] and disrupted NAD⁺/NADH [6, 7] ratios seem to drive a metabolic shift towards glycolysis. PDH downregulation reduces the flow of pyruvate into the TCA cycle, favoring glycolysis as an alternative energy source. Thus, AG can be viewed in this context as a metabolic adaptation to sustain ATP production despite OxPhos defects. Mitochondrial dysfunction also contributes to reductive stress, reflected by a decreased NAD⁺/NADH ratio, which further impairs ATP synthesis [8, 9]. Together, these energetic challenges promote increased reliance on AG to meet biosynthetic and energetic demands. Compounding this, altered glutamate dynamics, including increased astrocytic lactate production, further tie AG to neuronal activity and energy metabolism in SZ [10].

Other literature has clearly shown that AG is not merely a compensatory response to energy deficits but in some contexts it plays a physiologic role. In the brain, under metabolic stress conditions, AG supports biosynthesis, oxidative stress management, and synaptic plasticity [11]. However, its effectiveness as a repair mechanism appears to decline with sustained damage. For example, in neurodegenerative conditions such as Alzheimer’s Disease (AD), the youthful topographic pattern of AG is maintained during asymptomatic progressive amyloid deposition, but subsequently is disrupted with cognitive decline [12].

We review the available evidence in SZ and propose that metabolic impairments may elevate AG rates in adulthood.

This perspective raises two critical questions: (1) what specific factors initiate and sustain the shift to AG in SZ, and (2) whether AG in SZ is (purely) pathological or it retains protective and adaptive roles in specific contexts. Based on the available evidence, we hypothesize that AG in SZ represents a developmentally misaligned compensatory mechanism, potentially insufficient during critical developmental periods yet persistently elevated / heightened in adulthood, contributing to both metabolic and circuit-level pathology.

1. Evidence Supporting Elevated AG in SZ

Pyruvate Dehydrogenase (PDH) Downregulation

PDH complex is a regulator between glycolysis and OxPhos, and its activation is known to reduce glycolytic activity, converting pyruvate into acetyl-CoA for entry into the tricarboxylic acid (TCA) cycle [13, 14]. In postmortem brain tissues from individuals with SZ, PDH levels are reduced in the striatum [5], indicating a shift toward glycolysis and away from OxPhos.

Reductive Stress and Decreased NAD⁺/NADH Ratio

In SZ, there is a notable decline in the NAD⁺/NADH ratio [6, 7], particularly in the early course of illness. This decrease is primarily due to elevated NADH levels coupled with a slight reduction in NAD^+. Since glycolysis builds up NADH and OxPhos consumes this metabolite, this pattern is consistent with a transition from OxPhos to glycolysis [10].

PDH, the enzyme responsible for the initial stage of pyruvate oxidation, irreversibly converts pyruvate to acetyl-CoA. Its activity is inhibited when the NAD⁺/NADH ratio is reduced [15]. Conversely, inhibiting glycolysis affects PDH upregulation. An elevated mitochondrial membrane potential leads to a decreased NAD⁺/NADH ratio by limiting NAD⁺ production through mitochondrial electron transport. Furthermore, introducing alternative NAD⁺ regeneration pathways into cells significantly diminishes AG while leaving their proliferation rate unchanged. These observations imply that cells resort to AG as the NAD⁺ demand for oxidative processes surpasses the ATP requirements, suggesting AG serves as a compensatory mechanism when mitochondrial respiration alone cannot maintain sufficient NAD⁺ levels [16].

There is also a significant relationship between lactate production and mitochondrial NADH shuttle activity. When glycolysis exceeds the capacity of mitochondrial NADH shuttles, proliferating cells primarily convert glucose into lactate. These findings indicate that AG may also result from the saturation of NADH shuttles [17]. In patients with SZ, a reduced NAD⁺/NADH ratio [6] and OxPhos defects [18] are frequently observed. As a result, increased NADH due to OxPhos defects may exceed the capacity of NADH shuttles, leading to a shift toward AG.

Glutamatergic Activation of AG

In SZ, there is a notable alteration in brain glutamate concentration. Specifically, patients with first-episode psychosis (FEP) exhibit increased glutamate levels, while in chronic SZ, these levels begin to decrease [10]. Astrocytic uptake of glutamate during synaptic activity stimulates AG. This glutamate-driven stimulation of AG is mediated by sodium-dependent signaling cascades that lower the astrocytic ATP/ADP ratio, enhance oxygen consumption, and accelerate glycolytic pyruvate production. Although oxidative metabolism contributes to this response, glutamate’s net effect primarily promotes AG [19]. This points to the important function of glutamate in the regulation of energy metabolism in astrocytes and shows convergent evidence of elevated AG in SZ.

Lactate as a Biomarker and Driver of AG

Additional evidence for AG in SZ comes from studies of lactate and pH in this condition. In SZ, regional lactate levels increase as the disease progresses over time [10, 20]. This has been confirmed with MR Spectroscopy at 7 Tesla in the anterior cingulate cortex of SZ patients compared to controls [21].

Sustained upregulation of glycolysis leads to the accumulation of lactate and thus may serve as a reasonable biomarker of AG in brain. While levels of lactate are typically within the normal range during prodromal and initial phases of SZ, they elevate in chronic disease, possibly through persistent mitochondrial dysfunction and a compensatory increase in AG over time [20].

Beyond its metabolic role, lactate also acts as a signaling molecule. It binds to G-protein coupled receptors such as GPR81 on astrocytes and can trigger cAMP production, which amplifies glycolytic activity and promotes further lactate synthesis [22] [23]. Fluctuations in extracellular glutamate levels can further stimulate this loop, reinforcing lactate driven AG in glial cells.

Lactate has also been implicated in memory formation and synaptic plasticity. Its availability affects neuronal excitability, and blocking of lactate transport impairs memory consolidation in animal models [24, 25].

These observations support the notion that AG is elevated in SZ, but they do not necessarily point to lactate as a pathological metabolite. We propose that elevated lactate in chronic SZ likely reflects sustained AG, with accumulating mitochondrial impairments such as reduced PDH activity [5] and complex I/IV defects [18] that further limit oxidative capacity. Beyond indicating impaired OxPhos, lactate also serves as a signaling molecule facilitating AG in astrocytes [23]. (Figure 1).

Figure 1.

Evidence pointing to aerobic glycolysis (AG) in schizophrenia (SZ), summarizing contributing mechanisms (left) and glucose metabolic pathways toward lactate and oxidative phosphorylation (OxPhos) (right).

Gene Expression Alterations in Glycolytic Enzymes

Gene expression studies have revealed changes in the levels of several major glycolytic enzymes in postmortem brain tissues from individuals with SZ, such as glyceraldehyde-3phosphate dehydrogenase, pyruvate kinase, hexokinase, enolase, fructose-bisphosphate aldolase, and phosphoglycerate mutase [26, 27]. These findings are consistent with evidence of increased expression of hexokinase-2, a rate-limiting glucose-phosphorylating enzyme [28]. In addition, genetic differences in the regulation of glucose metabolism may predispose one to SZ, suggesting an association between glucose metabolism and SZ [29].

PET Studies

PET studies have provided additional support for dynamic changes in brain metabolism across the stages of SZ. In typical individuals, the cerebral metabolic rate of glucose consumption (CMRglc) has been shown to be elevated during early childhood [30]. Though data in children are limited, this developmental increase in CMRglc appears to be accompanied by a smaller relative increase in oxygen consumption [11, 31, 32]. In adults, CMRglc persists at higher levels compared to CMRO2 in specific regions [33]. With aging, brain glucose use may decrease faster than oxygen consumption, leading to a decrease of AG in some older adults, particularly in prefrontal regions associated with developmental neoteny [12, 34–36].

In SZ, PET meta-analyses suggest that regional brain AG may be disrupted. Specifically, patients with chronic SZ exhibit reduced CMRglc, particularly in frontal regions, a phenomenon termed “hypofrontality” [37]. In contrast, findings from individuals experiencing FEP are more variable, with some studies even reporting elevated frontal metabolism [37]. This heterogeneity suggests that AG related glucose metabolism may remain intact, or in some cases upregulated during the early phases of SZ, but tends to decline with disease progression. Meta-analytic FDG-PET data suggest that frontal hypometabolism is more apparent in medicated than in drugfree cohorts and tends to be greater in chronic SZ than first-episode patients [37].

Antipsychotics have been shown to impair mitochondrial function by reducing Complex I activity and ATP production [38], preclinical studies demonstrate impairment of central glucose metabolism [39]. For instance, haloperidol and olanzapine impair glucose sensing by blocking glucose-induced suppression of endogenous glucose production [40]. Clozapine induces astrocyte-dependent cortical hypometabolism in ex-vivo analyses [41], yet paradoxically promotes glycolysis and myelin lipid synthesis in cultured oligodendrocytes, whereas haloperidol shows opposite effects [42]. These findings suggest that antipsychotic effects on brain metabolism may be agent-specific and cell-type dependent, though their relevance to AG dynamics in SZ requires further investigation.

However, a key limitation is that most studies report CMRglc alone, without simultaneous assessment of the cerebral metabolic rate of oxygen consumption (CMRO2). Since AG is inferred from by disproportionately high glucose utilization relative to oxygen consumption, the absence of CMRO2 data prevents definitive conclusions about changes in AG per se.

2. Is AG a Pathologic or Physiologic Mechanism?

Association with Development

During the early postnatal period, AG emerges as the main biosynthetic pathway, supplying proteins and lipids essential for axonal growth, synapse formation, myelination, and cell proliferation [11, 43]. In the brains of preterm infants, 90% of glucose is directed toward AG [44, 45]. This pathway may partially persist into adulthood, potentially continuing to support the production of molecules for necessary synaptic modifications and the maintenance of neuronal structures [11] as well as metabolic homeostasis [46].

AG rates differ by region in the resting healthy young adult human brain. For instance, around 25% of glucose usage is non-oxidative in the medial prefrontal cortex, whereas in the cerebellum, AG accounts for only 2% of glucose consumption [33].

Regions of the brain exhibiting high levels of AG also show upregulation of genes associated with synapse formation and growth [31]. On the other hand, areas with elevated oxidative metabolism demonstrate higher expression of genes associated with mitochondrial function and synaptic communication [31].

Have Similar Mechanisms Been Investigated in Other Conditions?

Cancer cells use AG to prioritize biosynthesis, providing building blocks for biomass creation [47, 48]. In the brain, reactivation of this pathway may represent an adaptive repurposing of a biosynthetic pathway.

In individuals with cognitive impairment, the characteristic pattern of AG seen in young brains deteriorates. Whereas early amyloid deposition before cognitive impairment appears to preserve a youthful topography of AG, interestingly, a higher burden of white matter hyperintensities is linked to a loss of the youthful AG pattern in gray matter [12]. In patients with small vessel cerebrovascular disease, glycolysis is elevated in both perilesional tissues and the undamaged white matter compared to age-matched controls [49]. Together, these findings suggest that glycolysis may rise in reaction to early localized damage until the tissue and function become significantly compromised, at which point it decreases below normal levels. These findings suggest that AG may serve as an early compensatory and repair mechanism but fails to sustain function once damage exceeds a threshold.

Studies of pyruvate kinase M (PKM) isoforms show that in induced neurons from individuals with Alzheimer’s disease, a switch from PKM1 to PKM2 drives metabolic reprogramming toward AG, increasing neuronal vulnerability to apoptosis. Inhibiting PKM2 restores normal neuronal characteristics [50]. These findings may highlight a potential therapeutic strategy to restore metabolic balance in AD and other neurodegenerative diseases while also raising concerns about the role of AG, as its dysregulation could contribute to neuronal dysfunction and increased susceptibility to cell death.

AG involvement has also been proposed in affective disorders. Bipolar disorder, for instance, shows mitochondrial dysfunction, elevated lactate and altered glutamate metabolism, suggesting AG involvement [3]. Lithium treatment has been reported to increase leukocyte mitochondrial complex I activity during depressive episodes in bipolar disorder [51]. Moreover, proton MRS studies have shown increased brain lactate during depressive episodes in bipolar disorder, which decreased following lithium monotherapy [52]. Similarly, in major depressive disorder (MDD), impairments in glucose metabolism, including glycolysis and OxPhos have been described [53].

3. The Role of AG in Brain Function and Connectivity

Association with Neuronal Energy Metabolism

Neurons operate under strict energy limitations, as the brain functions near its maximum capacity in terms of energy consumption [54]. Once this limit is reached, there is little room for further energy generation. The brain cannot exceed this threshold due to inherent physiological constraints, and attempts to increase energy supply from the rest of the body are also limited [54]. This finite energy ceiling forces neurons to rely on AG not for proliferation, but as the only viable mechanism to meet acute energy demands without surpassing their metabolic capacity. Neurons may engage AG to buffer against metabolic stress, ensuring survival and supporting plasticity when OxPhos is insufficient to meet demands. This process highlights the adaptive role of AG in the brain—not as a driver of growth, but as a critical mechanism for maintaining cellular homeostasis. [46, 54]

Myelin

The field’s understanding of myelin’s role has evolved from being seen as a passive membrane sheath to being recognized as a metabolically active organelle of the glia that offers both insulation of axons and protection of neurons [55, 56]. An unresolved question is how myelin loss leads to axon degeneration in various white matter diseases. One hypothesis is that myelin may physically protect axons from autoreactive T cells [57]. Demyelinated axons consume significantly more energy than myelinated axons in performing the same functions over equivalent distances. Myelinated axons do not directly connect with extracellular metabolites, but exist within an environment regulated by surrounding glial cells, which supply essential metabolic support [58].

Patients with SZ exhibit concurrent hypomyelination and intra-axonal abnormalities that correlate with worse clinical outcomes [59] and widespread white matter (WM) abnormalities have also been confirmed by consortium studies [60] and postmortem showing downregulation of myelin and oligodendrocyte related genes [61].

Myelin lipids serve as temporary metabolic reserves during energetic stress, as shown by transient reductions after marathon running[56]. Metabolic impairments observed across aging and neurodegenerative disorders suggests that myelin loss may be a key, yet overlooked, driver of brain energy deficits [55]. For instance, in multiple sclerosis, demyelinated axons become metabolically undernourished as they lose access to oligodendrocyte-derived lipid substrates. [55, 62].

In SZ, widespread WM pathology may similarly deprive axons of insulation and myelinderived energy reserves. Under these conditions, AG may provide the rapid and adaptable energy supply.

Oligodendrocytes

Oligodendrocytes, responsible for producing myelin in the central nervous system, preserve the long-term stability of axons [63]. Nave and colleagues demonstrated that inhibiting mitochondrial complex 4 in mice resulted in demyelination and severe neuropathy in the peripheral nervous system. Interestingly, the same manipulation did not lead to glial cell death or demyelination in the central nervous system as shown by electron microscopy and g-ratio measurements. In vivo magnetic resonance spectroscopy revealed that brain lactate levels were higher in mutant mice compared to controls, suggesting that lactate is processed locally rather than removed through drainage. Nave and colleagues concluded from these experiments that mature oligodendrocytes rely on AG for survival [64].

AG supports oligodendrocyte survival and neuronal communication, particularly in thin, information-efficient axons that lack mitochondria. This glycolytic activity ensures timely and localized generation of ATP, enabling axons to maintain high firing rates while remaining thin and information-efficient. The hypothesis of efficiency tradeoff posits that maximizing informational efficiency is a higher priority for the brain than the local ATP production per glucose molecule. [65].

Abnormalities in both myelination and axon diameter are observed in SZ [59, 66]. As Nave et al. [64] have demonstrated, oligodendrocytes can sustain normal myelination via AG despite the presence of OxPhos related defects.

White matter damage is prevalent in SZ, with a higher incidence of degenerating myelin sheaths. These changes are linked to degenerative alterations in oligodendrocytes [67].

Impaired myelination and altered g-ratio significantly impact brain architecture in SZ, and individuals with SZ exhibit a markedly reduced pericapillary oligodendrocyte number compared to healthy individuals. Additionally, the oligodendroglial cell density is diminished in layer VI and the surrounding Brodmann area white matter [68, 69]. Another study demonstrates that the cell density of oligodendroglial cells is notably decreased in layer VI of individuals with SZ by 25%, in those with bipolar disorder by 29%, and in individuals with major depression by 19%, compared to healthy individuals. This reduction in oligodendroglial cells may contribute to neuronal atrophy observed in severe psychiatric disorders [70].

In disorders like SZ, where myelination and oligodendrocyte function are frequently compromised, the dependence on AG highlights its vital role in sustaining both the structural integrity and signaling efficiency of neural networks.

Synaptogenesis

Brain abnormalities follow a trajectory from first episode to chronic SZ. In early psychosis, structural and metabolic alterations are modest: gray matter loss is limited, frontal gyrification may be increased, and anterior cingulate metabolism is preserved or elevated [37]. By contrast, chronic stages show widespread cortical thinning, marked frontal hypometabolism, and reduced gyrification. The anterior cingulate cortex illustrates this shift, transitioning from preserved function to profound metabolic impairment [71]. These temporal changes appear most pronounced in association cortices, prefrontal and temporal regions that normally sustain high AG [72].

Studies using PET combined with fMRI have shown that AG colocalizes with functional network hubs in prefrontal and parietal association cortices, regions that also anchor major white matter tracts [31, 33]. Longitudinal imaging further revealed that age-related AG reductions are closely associated with cognitive impairments [12]. PET imaging studies also show a strong correlation between areas of high AG and the lengths of axonal projections, suggesting its crucial role in establishing effective axonal connections [73]. In the context of SZ, it is known that there is a disruption of cortical association networks, particularly in the prefrontal cortex and lateral parietal cortex, providing a potential link to the AG observations [74] and suggesting that metabolic vulnerability in these regions may contribute to impaired connectivity and cognition.

Brain regions with the highest AG levels, such as the PFC and lateral parietal cortex, show elevated gene expression associated with synapse formation and growth [33]. In the neotenous adult brain, where juvenile traits are retained into adulthood, AG remains active and is closely linked with the genes involved in synaptic plasticity and formation [31]. Emerging data in vitro and animal models support this view [75, 76]. This suggests that AG continues to support synaptogenesis in the adult brain. Several molecular mechanisms may underlie AG’s contribution to synaptogenesis. Astrocyte-derived lactate induces synaptic plasticity genes such as Arc and BDNF [77]. AG also facilitates membrane lipid synthesis required for dendritic growth [75]. Additionally, diversion of glycolytic intermediates into the pentose phosphate pathway provides NADPH and nucleotides to support synaptogenesis [78]. Its effectiveness in facilitating fast axonal transport from the soma to the axon terminal suggests that it may also support axonal elongation [43].

However, despite this evidence linking AG to synaptic plasticity and neurite outgrowth, sufficient neuronal repair appears lacking in SZ. A potential explanation is that while new neurons and synapses may be generated, these processes must be correctly regulated within a specific structural and functional context. In the absence of appropriate circuit-level integration, repair efforts may be misdirected or incomplete. Furthermore, this aberrant synaptic plasticity may destabilize neural circuits [79–81].

Another explanation is that biosynthetic capacity may remain regionally insufficient or misdirected, failing to support effective repair. This highlights the need for studies investigating the “region x timing” concept: metabolic stress in a critical development period of life might serve as a predisposing factor for severe mental illnesses, especially if affecting specific areas of the brain. The nature and severity would be influenced by a combination of these factors: which regions are affected, intensity of the stress, and when it occurs. Investigation of these dynamics (for example using MRS to measure lactate levels in vivo) may shed light on how disrupted energy metabolism at vulnerable stages contributes to the onset and progression of psychiatric disorders.

4. Conclusion

AG appears to be elevated in the adult brain in early SZ, but its biological significance remains unresolved. The evidence reviewed in this Perspective suggests that AG may represent the intersection of developmental vulnerability, metabolic compensation, and pathological persistence. It is still unclear whether elevated AG reflects an adaptive response, a downstream effect of mitochondrial dysfunction, or a primary metabolic signature of the disorder itself. A key emerging theme is the misalignment between AG dynamics and neurodevelopmental timing. AG may be insufficient during early brain development yet persistently elevated in adulthood, (Figure 2) potentially inverting its typical trajectory. This temporal imbalance in AG rates could impair neural plasticity and contribute to disease progression.

Figure 2.

Conceptual model summarizing proposed changes in cerebral metabolic rates of glucose (CMRglc) and oxygen (CMRO₂) across healthy controls (HC), first-episode psychosis (FEP) and chronic schizophrenia (SZ).

Although no definitive thresholds exist to demarcate functional from dysfunctional AG, certain conditions (e.g. timing, location and chronicity of the disease) may tip the balance. In SZ, elevated AG may initially serve as an adaptive response to mitochondrial impairments or reduced brain biomass, helping sustain energy and biosynthetic needs. However, AG elevation in SZ may become maladaptive through multiple mechanisms. Although AG appears elevated in adulthood, this does not ensure effective repair. Regional or temporal capacity may remain insufficient, and biosynthetic outputs may be misdirected toward synaptic connections that fail to integrate functionally into neural circuits.

Several methodological limitations constrain interpretation of existing evidence. Most PET studies report CMRglc without simultaneous CMRO₂ measurement, preventing direct quantification of AG. Moreover, PET studies rely on assuming a lumped constant, which itself might be affected by development, aging, and disease. Longitudinal data tracking metabolic trajectories across neurodevelopment in individuals at clinical high risk (CHR) for psychosis are lacking. Most lactate studies are cross-sectional with small samples, and no direct measurements of AG exist during critical developmental periods. These gaps limit our ability to definitively establish when, where, and why AG becomes maladaptive in SZ. Addressing these gaps through longitudinal multimodal imaging could identify therapeutic windows for restoring metabolic balance in SZ and clarify how antipsychotics shape brain metabolism. From a clinical perspective, these findings do not support a single/simple metabolic intervention, but highlight the need for context/timing dependent modulation of AG to support its physiological roles during development while limiting its potentially maladaptive effects later in life.

If future studies confirm that AG persists abnormally and contributes to aberrant [79–81] synaptic plasticity, or if a loss of youthful AG patterns similar to those observed in Alzheimer’s disease [12] is evident, then targeting the underlying drivers of this imbalance, for example reducing lactate accumulation or alleviating reductive stress may hold therapeutic promise. However, until AG dynamics are more clearly delineated, such treatment proposals remain premature.