Neurodegenerative diseases

Neurodegenerative diseases (NDDs) represent the exceptional therapeutic challenge of our time. Despite decades of intensive research, effective treatment and prevention strategies remain elusive. Understanding the primary initiating factors of sporadic NDDs is crucial in searching for a cure. While major NDDs vary greatly in affected brain areas and etiology, they share one common characteristic: all are characterized by energy (glucose) hypometabolism and oxidative stress. Numerous clinical and animal model studies have shown brain hypometabolism associated with oxidative stress to be an early (in some cases, the earliest) biomarker for most NDDs [1–4]. It is precisely this pathological combination that is likely the main initiating cause of the subsequent disease-associated detrimental cascades, and stopping it could finally prove to be an effective preventative strategy. The critical question here is, what is the exact trigger and source of this early pathology? This review proposes that the initiating oxidative stress (iOS) in the brain resulting specifically from the hyperactivation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX) is the primary initiating factor of major NDDs [3, 5–7].

Risk factors for NDDs

The risk factors for most NDDs overlap and lead to oxidative stress and energy deficiency [8–10]. Analysis of the preceding disease determinants reveals that disrupted glucose metabolism associated with oxidative stress and neuroinflammation is the expected end state of most, if not all, risk factors before the NDDs initiation [3]. As the result of analogous risk factors [8], major sporadic NDDs reveal similar detrimental hallmarks in the early stages, suggesting that these pathologies share common pathways initially. It is reasonable to propose therefore that uncovering the primary cause of these abnormalities might give a clue to efficient disease prevention.

Glucose hypometabolism

Normally, glucose is the primary fuel source in brain cells and the major substrate for endogenous antioxidant defense systems [11]. The underlying cause of glucose hypometabolism has been unclear until recently. Accumulating evidence renders oxidative stress a primary reason for glycolysis inhibition. Indeed, oxidative stress, which is defined as an imbalance between the cellular production of reactive oxygen species (ROS) and the cellular antioxidant system’s ability to neutralize them readily, may cause deleterious modifications as it can change DNA structure, resulting in modification of proteins and lipids, activation of several stress-induced transcription factors, and production of proinflammatory and anti-inflammatory cytokines [12–14]. In particular, it is known that ROS can suppress glycolysis by inhibiting multiple glycolytic enzymes, including pyruvate kinase, phosphofructokinase, and glyceraldehyde-3-phosphate dehydrogenase [15–17]. Notably, methylglyoxal is a component of glycolytic reactions [18, 19] and may be cytotoxic due to its ability to generate advanced glycation end products (AGEs) which increase oxidative stress and have been implicated in stroke, diabetes, and NDDs such as Parkinson’s disease (PD) and Alzheimer’s disease (AD) [20–25].

Oxidative stress associated with deficient glucose metabolism may trigger acquired epilepsy, sporadic AD and PD [3]. Butterfield and Halliwell [16] presumed that beta-amyloid 42 (Aβ₄₂) oligomer-induced oxidative stress impairs glucose metabolism ultimately causing mild cognitive impairment (MCI) and AD. Our animal experiments showed that: 1) exogenous hydrogen peroxide (H₂O₂) inhibits glucose consumption [26]; 2) oxidative stress triggers seizures, while its prevention results in attenuation of epileptiform activity in vivo [27]; 3) Aβ_1-42-induced oxidative stress causes brain glucose hypometabolism and network dysfunction [7].

ROS overproduction

Strong experimental evidence indicates a significant role of oxidative stress in the initiation of glucose hypometabolism. The essential question is the origin of this oxidative stress. Indeed, if ROS overproduction is due to mitochondrial dysfunction, as often postulated in the literature, prevention of ROS generation would be highly problematic. At least 11 sites of ROS production have been identified in mitochondria [28, 29]. Though it was reasonably proposed that oxidative damage in neurodegeneration should be prevented via the direct inhibition of ROS production from specific sources, rather than via scavengers [30], such a goal in mitochondria is very tough to reach. In support of this notion, mitochondrially targeted antioxidants failed in clinical trials [1, 30–37]. We also demonstrated in brain slices that potent exogenous antioxidants failed to impede fast ROS release during network activity [26]. Notwithstanding, accumulating evidence indicates the major role of NOX activity in iOS during NDD onset.

Mitochondrial dysfunction

It is very important to identify the principal source of oxidative stress in the onset of disease and find out whether mitochondrial dysfunction parallels the decrease in glucose consumption, or whether the glycolysis impairment induced by some specific source of oxidative stress precedes mitochondrial dysfunction. It is generally accepted that under physiological conditions mitochondria are the major source of ROS production (up to 90%) in the brain cells [38, 39], which imposed the conclusion of mitochondria-biased oxidative stress in many reports. However, although the brain is presumed to have a weak antioxidant defense [36, 40–42], this conclusion is not valid for mitochondria which possess a highly efficient system for antioxidant defense (consisting of several detoxifying enzymes such as glutathione, catalase, and others) [43–45], which normally neutralizes ROS as soon as they are generated. ROS are produced at various sites in mitochondria. Still, most of them are generated as by-products [superoxide anion (O₂^–)] of the electron transport chain during the oxidative phosphorylation process following the dismutation of O₂^– to H₂O₂ by copper and zinc superoxide dismutases in the intermembrane space and manganese superoxide dismutase in the matrix [30, 46, 47]. H₂O₂ removal can be two to three times faster than H₂O₂ production in rodent brain mitochondria [44, 45]. Therefore, the physiological emission of ROS from mitochondria is negligible considering oxidative stress [44], while may implement a signaling function [48, 49]. Moreover, due to its powerful scavenging potentials, mitochondria can neutralize penetrating cytoplasmic ROS and serve as their sink [28, 38, 44].

During pathology onset, deteriorated glycolysis would eventually lead to mitochondrial impairment, overproduction of ROS, and an increase in their emission to the cytoplasm in the subsequent stages of the disease, as has been shown in multiple studies [31, 34, 41, 50]. In addition, shifting the redox state balance towards oxidative stress may impair several mitochondrial proteins leading to dysfunction in the production of ATP and energy starvation [30, 47]. Indeed, it is generally accepted that the primary reason for mitochondrial dysfunction during NDDs is oxidative stress. It is highly unlikely, however, that the origin of such oxidative stress is mitochondria themself due to their potent antioxidant protection and high intrinsic resistance to oxidative stress [28, 51, 52]. More likely is the involvement of extra-mitochondrial ROS accumulation via the activity of other sources, e.g., NOX. This iOS may lead to damage/dysfunction of mitochondria, but its primary target would be cytoplasmic processes such as glycolysis.

Mitochondrial dysfunction occurs in all significant sporadic NDDs [53, 54], however, it is currently unknown whether oxidative mitochondrial damage occurs early in disease progression or is caused by secondary manifestations of the disease pathophysiology. Many authors agree that reduced energy metabolism and oxidative damage are at the center of NDD pathogenesis [1–4, 13, 55, 56]. Interestingly, although impaired glucose metabolism is one of the earliest features of the AD brain, the previous studies reported that early in AD, the cerebral metabolic rate of oxygen was not altered or was changed disproportionally to the prominent decrease in glucose utilization [57–59]. It was hypothesized that unaltered oxygen utilization and normal carbon dioxide (CO₂) production may indicate undisturbed substrate oxidation in mitochondria [57]. Moreover, other early studies that used the arterio-venous difference method showed that brain ketone uptake is still normal in moderately advanced AD [60, 61], while ketone catabolism is entirely mitochondrial. Recent studies using positron emission tomography (PET) ketone tracer, carbon-11-labeled acetoacetate (¹¹C-acetoacetate), reported that brain metabolism of ketones is unchanged in MCI and early AD [57, 62–66] supporting the previous assumption that oxidative phosphorylation may still be normal in the AD onset. This suggests that brain hypometabolism in prodromal AD may be specific to glucose and the primary site of metabolic abnormalities is glycolysis [63] but does not include dysfunctional mitochondrial oxidative phosphorylation. Indeed, it is highly problematic to explain the normal brain ketone metabolism unless suggesting that the enzymes of mitochondrial oxidative phosphorylation continue to function relatively normally, at least early in AD.

In animal experiments, mitochondria were reported not to be the main source of ROS overproduction in AD models [30, 52] and during seizure activity [67, 68]. We also did not observe the change in oxygen consumption either during epileptiform network hyperactivity [26] or under the application of Aβ_1-42 [7, 69] suggesting maintained mitochondrial functioning, while a significant reduction in glucose utilization was detected in all cases.

Finally, supporting the proposition of minor mitochondria contribution to the iOS triggering major central nervous system diseases, mitochondrially targeted antioxidant therapies have been tested in clinical trials but failed to reveal evident benefits [1, 30–37].

Contribution of NOX to iOS

NOXs have been known for a long time to be responsible for the respiratory burst in phagocytes [70]. This unique enzyme family has the only biological function of ROS generation. The extensive expression of NOX isoforms has been discovered rather recently in a variety of brain cells [71] with NOX2 and NOX4 being the most prominent isoforms detected in neurons, microglia, and astrocytes [72, 73].

NOXs are multi-subunit enzymes, comprising membrane subunits and cytosolic subunits. Under the resting condition, NOX is normally dormant and the cytosolic components remain dispersed in the cytosol. Still, upon activation, which requires specific agonists, e.g., in neurons, NOX activation requires N-methyl-D-aspartate receptor stimulation [74], cytosolic components translocate to the membrane and assemble to the functioning complex [75]. Interestingly, NOX-generated ROS appeared as a major source of oxidative stress in NDDs, including AD, PD, and amyotrophic lateral sclerosis (ALS) [5, 76–78], as well as in acquired epilepsy and stroke [68, 79–81].

In animal experiments, we demonstrated that spontaneous seizure-like events in brain slices were initiated by NOX activation, while NOX inhibition prevented their generation [27]. Moreover, inhibition of NOX in vivo suppressed epileptiform activity in several seizure models [27]. In addition, Aβ_1-42 was found to be an agonist of NOX [5, 82, 83] and we demonstrated both in slices and in vivo that Aβ-NOX-induced oxidative stress resulted in prominent glucose hypometabolism [7, 84]. Importantly, recent studies suggested alpha-synuclein-induced activation of NOX [83]. Several recent reviews summarized current progress regarding the crucial role of NOX enzymes in NDDs like PD, AD, Huntington’s disease, multiple sclerosis, and ALS, and in acute neurological disorders such as stroke, spinal cord injury, traumatic brain injury, and related cerebrovascular diseases [5, 73, 76, 83, 85].

How to counteract iOS induced by NOX in humans

The activated NOXs generate superoxide in phagocytes providing a major role in the human immune response [70]. This fact justifies the obvious conclusion that non-selective inhibition of NOXs is not a suitable option in developing NOX-targeting treatment. Therefore, only a selective inhibition of NOX family members may be considered to avoid harmful side effects. Targeting NOX activity without any off-target effects was recently impossible because of the lack of isoform-specific inhibitors. About thirty NOX inhibitors have been analyzed in recent comprehensive reviews [6, 86, 87], but only a few selective antagonists have been synthesized, specifically for NOX2 and NOX4 isoforms, at present, and only one (GKT137831, a specific inhibitor of NOX1 and NOX4) is in human clinical trials (phase 2) for pulmonary fibrosis and cirrhosis [6]. Several other promising inhibitors have been recently developed, such as NOS31 for NOX1, CPP11G/CPP11H and GSK2795039 for NOX2, and GLX7013114 for NOX4. Altogether, as NOX isoforms are well identified and studied, the elaboration of efficient selective inhibitors is hopefully a question of the nearest future.

Perspective

Therefore, the disorders mentioned above differ markedly in their etiology but they share common pathologies in brain function—oxidative stress and glucose hypometabolism, which principally define the disease onset and pathogenesis. This knowledge provides a potential opportunity to elaborate on a treatment counteracting oxidative stress and glucose hypometabolism and thus be efficient in curing/preventing the diseases. Unfortunately, at present, there is a lack of clinically available pharmaceuticals able to counteract the disease-initiating pathologies. On the other hand, lifestyle interventions have emerged in the spotlight and an increasing body of evidence suggests that dietary means can influence pathophysiological features of NDDs and therefore could alter the course of disease neurological conditions. To tackle this problem, various nutrition interventions have been suggested including dietary supplements and dietary restrictions. Former regiments supposed the intake of specific nutrients, while the latter restricted particular nutrient(s) (e.g., carbohydrates) or time-restricted eating patterns, which can result in marked long-term changes in brain metabolism and functions such as neurogenesis and synaptic plasticity, oxidative stress and inflammation or epigenetic regulations as well as gut microbiota changes [9, 88–93].

Based on the assumption that the efficient diet has to counteract both brain oxidative stress and glucose hypometabolism, which presumably trigger the onset of NDDs, the only available strategy which satisfies these requirements is based upon a metabolic shifting from carbohydrates to fat utilization resulting in ketosis and a substantial increase of ketones level in the blood. The classical example of this approach is the ketogenic diet (KD), the only widely clinically accepted metabolic treatment for epilepsy [94]. The KD provides an alternative fuel substrate for mitochondria and thus supports brain energy production especially when glucose metabolism is impaired as in the case of NDDs. The clinical KD incorporates a 4:1 ratio of fat to protein plus carbohydrate that results in an increase in ketone blood level from < 100 μmol/L (typically observed in diets with unrestricted carbohydrate utilization) up to > 3 mmol/L [95]. In the brain, ketones bypass the glycolytic pathway directly entering mitochondria, thus constituting an even more efficient energy source than glucose and promoting mitochondrial oxidative metabolism [96–98]. Importantly, as mentioned above, the brain mitochondrial metabolism of ketones is unchanged in the prodromal stage of AD [57, 62–66] despite significantly impaired glucose utilization, indicating that ketones are able indeed to substitute glucose in energy production at least early in the disease. In addition, to avoid oxidative stress, brain cells possess a cytoplasmic antioxidant system utilizing the glucose pentose-phosphate-pathway [99] where enzymes of the glutathione system are used for the neutralization of H₂O₂ [99–101]. The cellular redox state is also controlled by specific gene transcription factors [e.g., nuclear factor erythroid 2-related factor 2 (Nrf2)] which regulate more than 200 genes including those containing an antioxidant response element (ARE) in their promoter and activation of the Nrf2-ARE signaling enhances the expression of enzymes involved in antioxidant defense in pentose-phosphate-pathway [102]. However, during acute oxidative stress, the glutathione system activation occurs in seconds while the onset of transcriptional responses takes hours [99, 103]. The pentose-phosphate-pathway has a large reserve capacity for upregulation and when amplified by oxidative stress it may reach up to 30% of glucose utilization compared to 3–7% in physiological conditions [99, 100, 103]. Therefore, the efficiency of this cytoplasmic antioxidant system depends directly on glucose availability and the glucose-sparing effect of the KD may be crucial for the operation of cellular antioxidant defense during NDDs [11, 84]. Moreover, it has been reported that KD induces initially mild oxidative stress and is related to the systemic activation of the Nrf2 pathway [95, 104].

The clinical KD settings are presumably the most efficient, however, due to severe nutrition restrictions, the diet possesses long-term adverse effects such as uric acidemia, hypocitraturia, hypercalciuria, aciduria, decreased bone mineral density, anemia, neuropathy, and also includes frequent gastrointestinal disturbances such as constipation, abdominal pain, emesis, and gastroesophageal reflux disease ​​[98]. In addition, clinical studies show that maintaining the KD can be challenging for adult patients due to poor tolerance and lack of motivation [92]. Therefore, a reasonable diet for adults has to integrate the main benefits of the KD but be free of its disturbing pitfalls.

It is worth mentioning a popular dietary regimen, caloric restriction, which reduces food intake without incurring malnutrition. In fact, both the KD and caloric restriction diets are thought to act primarily through the same mechanisms as both diets result in an overall reduction in caloric intake and an increased level of circulating ketone bodies [93, 105, 106]. Indeed, the caloric restriction regimen represents a specific case of intermittent fasting, which is discussed below. Moreover, the diet characteristics largely intertwine, e.g., intermittent fasting results in intermittent ketosis, known for its appetite suppression, which results in turn in voluntary calorie reduction. Meanwhile, intermittent fasting has more metabolic benefits than permanent calorie-restriction [93, 106].

Fasting: the first known treatment of neurological diseases

The ability to survive the seasonal and diurnal periods of fasting is evolutionarily acquired and thus is supported by appropriate homeostatic mechanisms. As a result, an altered metabolic state occurs—ketosis. Fasting as a treatment for epilepsy was first reported in 1910 [107]. In both animals and humans, fasting prevents and treats metabolic syndrome, a major risk factor for many neurological diseases. Fasting improves cognition, stalls age-related cognitive decline, slows neurodegeneration, reduces brain damage, enhances recovery after stroke, and mitigates the pathological and clinical features of epilepsy and multiple sclerosis in animal models [108]. The КD appeared as a metabolic imitation of fasting [109] (which was considered hard to implement) since both conditions induce ketosis [92]. However, adherence to the KD has also often been reported as difficult with noticeable side effects [98].

The neuro- and metabo-protective effects of ketogenic regimens are of great practical importance given the epidemics of metabolic and NDDs, particularly, in protecting the brain from hypometabolism [9, 110, 111]. On the other hand, there is recent evidence of a number of exclusive functions of glucose that must be kept in mind in the regiments seriously restricting carbohydrate content: the synthesis of glycogen and nucleotides, antioxidant protection, rapid generation of ATP, and the production of pyruvate [11].

Long-term fasting (from many days to weeks) is hard to implement and is not free of severe adverse effects [110]. A convenient form of the fast regimen, intermittent fasting, includes eating patterns in which individuals follow recurrently the prolonged time periods (16–48 h) with no food intake, while normal food intake between these periods. Intermittent fasting evokes evolutionarily conserved, adaptive cellular responses that are integrated into the body in a manner that improves glucose management, increases stress resistance, and suppresses inflammation [106]. Several clinical trials are underway to test the effects, efficacy, and safety of intermittent fasting in patients with NDDs [112, 113]. Indeed, the neuroprotective effects of intermittent fasting observed in preclinical studies and clinical trials suggest that this regimen has broad-spectrum benefits for many health conditions and could represent a promising treatment option, especially in the earliest stages of the disease [106, 114].

As mentioned above, in the fed state in humans, the blood levels of ketone bodies are low but rise within 8–12 h after the onset of fasting, reaching levels above 2 mmol/L by 24 h [115, 116]. During the intermittent fasting regimen, repeated exposure to fasting periods prompts cells to the lasting adaptive stress responses resulting in increased antioxidant defense and down-regulation of inflammation [106]. Interestingly, following the administration of individuals with “ketogenic drinks” containing medium-chain triglycerides, the blood concentration of ketones increased rapidly but returned to the low basal level during a few hours [117, 118]. An analogous situation occurred at one-week intermittent fasting when blood ketone levels markedly increased during two fasting days but the next day returned back to the baseline [119]. In contrast, during the recurrent intermittent fasting procedure, the baseline ketone levels were significantly increased [120, 121] suggesting that fundamental adaptive metabolic systemic alterations occurred including ketone production in the liver. This is in support of the proposition reported previously [106] that “periodic flipping of the metabolic switch not only provides the ketones that are necessary to fuel cells during the fasting period but also elicits highly orchestrated systemic and cellular responses that carry over into the fed state to bolster mental and physical performance, as well as disease resistance”.

Time-restricted eating is the most convenient protocol

Regimens of intermittent fasting include 16–48 h periods with no food intake repeatedly [122], however, it has been noted that some people are unable or unwilling to adhere to a scheme with infrequent but longer fasting periods [110, 123] and it is not recommended to AD and PD patients with a risk of malnutrition [111]. Therefore, a type of intermittent fasting, time-restricted eating (or feeding in animal experiments), which leads to the benefits of total fasting without obligatory reduction of nutrient intake appears to be the most convenient protocol [124].

In humans, the timing of meals is adjusted to the light-darkness cycle rather than to genetically determined metabolic cycles, thus the prolongation of the light phase due to artificial illumination makes the calorie-restriction protocols challenging to comply with [125]. In experiments with free-running circadian rhythms in isolated environments, the sleep-wake cycle governed the meal timing, and the duration of both increased due to the exclusion of culturally determined [113, 126].

Time-restricted eating is a dietary regime that gathers all calorie intake independent from their macronutrient composition into 6–10 h during the active phase of the day [127]. Excluding the use of nutritional supplements, this is the only feasible way of compensation for nutritional deficiencies, unavoidable in any restrictive regime [127]. Intermittent ketosis protocol allows one balanced meal daily, complemented by 1–2 ketogenic meals [128]. Intermittent fasting protocols include alternate-day fasting and fasting for two days per week [129]. Time-restricted eating achieved by consistently reducing daily meal counts is more feasible than intermittent fasting. Time-restricted eating improves glucose control and insulin sensitivity and reduces oxidative stress [130–132]. It includes a direct antiepileptic calorie-restriction effect since it results in a voluntary reduction in energy intake and exerts long-term neuroprotective effects [124, 133].

There is an additional option to make the time-restricted eating regimen more feasible and palatable. Indeed, both chronic and periodic carbohydrate restriction in high-fat diets metabolically wise mimics fasting [134–140]. By the same token, a high-fat/low-carbohydrate breakfast mimics the metabolic features of time-restricted eating. Indeed, eating a very high-fat breakfast improved cognition [106, 141–144] supposedly by prolonging the overnight fast effects since Freemantle et al. [145] showed that a ketogenic breakfast does not interrupt overnight ketosis. Consequently, both the “ham and egg” ketogenic breakfast [146] and skipping breakfast results in a metabolic condition that can be termed intermittent ketosis. In terms of meal timing, skipping breakfast represents a type of intermittent fasting [138, 147, 148] leading to intermittent ketosis known for its appetite suppression effect [149] resulting in voluntary calorie reduction (e.g., [150]). Breakfast skipping and exclusion of late eating results in a reduction of voluntary calorie intake [151, 152], which is important since calorie-restriction has been shown to have profound metabolic benefits including neuroprotective, anti-ageing, and anti-inflammatory [110, 153]. Furthermore, Mattson et al. [154] suggested intermittent fasting has more metabolic benefits than permanent calorie-restriction. Thus, skipping breakfast may be more beneficial than traditional restrictive dieting.

Various supplementary nutrients (unsaturated fatty acids, complex sugars, fibers, etc.) positively affecting brain functions have been described in the literature (see e.g., [88, 91, 155]). We would like to mention pyruvate specifically since it’s a unique array of neuroprotective properties and its ability to restore substantially glucose metabolism impaired by oxidative stress [26, 84, 156].

Conclusions

Fasting is the first known treatment of NDDs in history. The neurodegenerative processes trigger the four major pathologies: 1) oxidative stress; 2) hypometabolism of glucose leading to energy deficiency; 3) neuroinflammation; and 4) insulin resistance. Intermittent ketosis and its metabolic equivalents are efficient in preventing all of them excluding impaired glycolysis. Supplementation by oral pyruvate can be the missing part of the equation. Nowadays ketosis is considered a metabolic analogue of fasting [157, 158], it is well-known that the ketogenic regimen can reproduce the effects of fasting, and that ketogenic meals mimic the fasting period in time-restricted protocols. On the other hand, both chronic and periodic carbohydrate restrictions also produce effects characteristic of fasting. This unifying phenomenon can be termed “guided metabolic timing”. It can include skipping and eating ketogenic breakfast since they both result in intermittent ketosis. These regiments cause a reduction of voluntary calorie intake thus they are metabolically favorable. Thus, pyruvate and intermittent ketosis combined might counteract all four major neurodegenerative pathologies and provide a core for efficient disease treatment.