MRI spectroscopic and tractography studies indicate consequences of long-term ketogenic diet

Under normal circumstances, energy required for proper brain functioning in adults derives from glucose generated in carbohydrate metabolism (Vannucci and Vannucci 2000). When carbohydrates are eliminated from the diet, energy can be obtained from fatty acid oxidation. This process generates the amount of acetyl-co molecules sufficient for the synthesis of acetylacetic acid in the liver. It can be either spontaneously transformed into acetone or enzymatically transformed into beta-hydroxybutyric acid (ketone bodies, bHb). This reaction can run in both directions in the presence of beta-hydroxybutyric acid dehydrogenase (Bailey et al. 2005). Ketone bodies generated in this way are an alternative energy source for metabolism. The status in which the synthesis of ketone bodies is accelerated is called ketosis. Ketone bodies are transported to the circulation and then to extrahepatic tissues where they are used as ATP precursors. They can also reach the brain easily crossing the blood–brain barrier; acetone via simple diffusion, beta-hydroxybutyric acid and acetylacetic acid through transporters, the expression of which is connected with ketosis level (Bailey et al. 2005). Upon crossing the blood brain barrier, ketones are transported across cell plasma membranes via monocarboxylate transporters 1 and 2 in astrocytes and neurons, respectively (Vijay and Morris 2014).

Guzman and Blazquez (2001) presumed that ketone bodies might have a potential to arise also in astrocytes and then be transported into neurons. This astrocyte–neuron transportation hypothesis seems to be very interesting and worth studying in detail.

A diet with significantly reduced protein and carbohydrate is commonly called ketogenic (KD) and has been described by many authors as having protective effects. To date, several mechanisms of the ketogenic diet have been proposed, despite the fact that along with the increase of knowledge on this subject, the number of hypotheses has also been growing. It was also demonstrated that as short as a 3-week feeding of rats with KD gave rise to elevation of glutathione, a free radical scavenger. It was most probably associated with a simultaneous increase of glutamate cysteine ligase (GCL), an enzyme that limits the GSH synthesis rate (Jarrett et al. 2008).

Ketones may also affect the activity of sirtuin 3, which is the major mitochondrial deacetylase (Rardin et al. 2013). Due to its enhanced activity, the ROS production diminishes, thereby preventing from oxidative stress episodes. The ketogenic diet suppressed dissociation of a proapoptotic factor Bad from the chaperon protein 14-3-3 which was demonstrated to occur in the kainate model of seizures (Noh et al. 2005). Other studies showed that the ketogenic diet increased the level of the inhibitory neurotransmitter GABA and raised the expression of glutamic acid decarboxylase (GAD), an enzyme participating in GABA synthesis (Cheng et al. 2004; Dahlin et al. 2005; Yudkoff et al. 2005; Gama et al. 2015; Melo et al. 2006).

The neuroprotective potential of ketogenic diet related to epileptic seizures can be effective not only by changing levels of neurotransmitters, but also by independent modification of the neuronal membrane potential. After entering the TCA cycle, ketone bodies can increase ATP levels by affecting the ATP-dependent potassium channel and leading, therefore, to increased polarization of the neuronal membrane (Koppel and Swerdlow, 2017).

Recently, KD is frequently applied for the treatment of neurodegenerative diseases such as epilepsy, migraine, brain damage and Alzheimer’s or Parkinson's diseases (AD, PD), autism, sleep disorders, amyotrophic lateral sclerosis (ALS), multiple sclerosis (MS), pain, depression and even cancer (Stafstrom and Rho 2012) or to lose body weight. The reasons leading to the use of different dietary treatments as attempts to cure the diseases or alleviate their symptoms are both a lack of effective pharmacological treatment, and/or great needs for more natural methods without commonly occurring adverse side effects. Furthermore, it is believed that normally functioning neurometabolic paths can be modified by an appropriately selected diet (Stafstrom and Rho 2012).

However, the ameliorating KD effects were documented using mostly animal models or clinical trials on individuals whose metabolism was modified, or even disturbed by different damages or ongoing disease processes. Unlike those studies, our previous research on normal animals (Gzieło et al. 2019), which were not subjected to any previous experimental procedure, showed that KD led to morphological transformation of glial cells and quantitative increases of calretinin-containing interneurons. Using the same experimental design, our research is, therefore, aimed at effects of long-term, 4-month-lasting KD on brain anatomy (structural MRI scans), subtle changes in the brain metabolism (MRI spectroscopy), and intracerebral connections (MRI-based tractography). To our best knowledge, such studies have not been conducted so far. Only brain glucose and ketone uptake were tested in 4- or 21-month-old animals after 14-day-lasting KD (Roya et al. 2012). These studies showed only that the region-to-whole brain ratio of cerebral metabolic rate of glucose was 37–41% lower in the cortex and 40–45% lower in the cerebellum compared to cerebral metabolic rate of cerebral metabolic rate of acetoacetate in 4 month- and 21 month-old rats (Roya et al. 2012). In the study by Zhang et al. (2018a, b), NAA, Cho, CR contents were assessed with MRI spectroscopy in animals fed a ketogenic diet after TBI. In addition, studies by Huang et al. (2019) showed that KD applied for 3 months in mice did not cause changes in behavior when tested. But when the threshold for pentylenetetrazole-induced seizures was tested, it turned out to be higher in animals on KD.

To our best knowledge, this is the first study on effects of long‐term feeding of KD in young adult rats, which were not subjected to any previous experimental procedure simulating a neurodegenerative disease. The study has shown that the 4-month-lasting ketogenic diet changed the metabolism and structure of the brain in normal, healthy animals. First, the changes were observed in the blood. As in other studies, after 1 week of KD consumption, the level of ketone bodies significantly increased in peripheral blood, which was the natural evidence of the treatment application. Also, in blood collected from the heart immediately before perfusion, the level of bHb was significantly higher (Gzieło et al. 2019). The glucose level remained normal, which was in agreement with Zhang et al. (2018a, b) who explained this fact as the result of adaptation to a long stay on a carbohydrate-free diet. Importantly, the morphotic blood elements were not affected, which suggests that these parameters were not modified when this diet was used in individuals suffering, for example, from epilepsy.

Since ketone bodies formed during the ketogenic diet can penetrate BBB, different parts of the central nervous system can also be affected but at different degrees. Our research was focused on changes in concentration of the most important brain metabolites occurring in the neocortex and hippocampal formation after 4 month-lasting KD. Level of the metabolites was measured with magnetic resonance spectroscopy frequently used not only for experimental approach but also in clinical diagnostic tests. Most of the metabolite levels did not change, which suggests that ketone bodies are rather not involved in all metabolic pathways within the nervous tissue.

The changes that we observed in both the cortex and hippocampal formation affected, among others, the levels of Glu and Gln which were significantly higher after 4-month-lasting KD than after normal diet. Glu is the most common excitatory neurotransmitter in the nervous system. Its occurrence and action are closely related to Gln via the Glu–Gln cycle. After being released from synapses, Glu is taken up by adjoining astrocytes, transformed into Gln, and then transported back to the neuron (Danbolt et al. 2016). That is why it was not surprising in our research that the increase in Glu after KD was accompanied by an increase in Gln level. The energetic cost of Glu uptake is very high—at 1.5 ATP molecules per one Glu molecule transported. Therefore, the large involvement of astrocytes in this process protects neurons from metabolic overload (Danbolt et al. 2016). For this uptake, KD provides more energy than a normal diet because metabolized fat delivers more energy than glucose (Włodarek 2019). Even though Glu is a precursor of GABA, in our study, the level of GABA did not undergo a corresponding KD-induced change. Also, no significant change was detected in the ratio of Glu to GABA concentration both in relation to the stage before KD introduction in the same animal group and to the control group fed with normal diet. Since GABA is the most important inhibitory neurotransmitter, increases in its level are postulated to determine epileptic seizure inhibition in response to KD (Yudkoff et al. 2008; Greene et al. 2003; Maalouf et al. 2009). In addition, Glu can act via ionotropic IGluRs receptors and enhance GABA secretion by inhibitory neurons (Mahmoud et al. 2019). However, KD does not always lead to increases in GABA content. As in our own study, KD-induced elevation of GABA did not occurr in the mouse forebrain and cerebellum (Yudkoff et al. 2001), mouse whole brain (Yudkoff et al. 2005), mouse neocortex (Melo et al. 2006) or in the whole rat brain (Hartman et al. 2007). Supposedly, the lack of changes in GABA concentration in our study resulted from the fact that we were focused on KD effects in normal animals but not in those under pathological conditions which can influence molecular interactions differently. This problem needs further investigations.

Another postulated mechanism of KD action is a decrease in oxygen free radical production. Such an antioxidant effect may be caused by increased levels of glutathione, which was also observed here. This is most likely associated with an increase in glutathione peroxidase activity. It is suggested that higher activity of this enzyme, induced by KD, may be protective against neurodegenerative changes (Ziegler et al. 2003). In addition, it is believed that most GSH-dependent detoxification processes occur in astrocytes (Fernandez-Fernandez et al. 2012), which confirms the involvement of these cells in the protective KD effects.

Our analyzes also showed a decrease in creatine concentration in the cerebral cortex. In our previous research using synchrotron radiation, an increase in the number of creatine deposits was detected in the hippocampal formation following 30-day-lasting KD (Skoczen et al. 2017). This discrepancy is probably due to different creatine content in particular brain structures. Creatine in the brain and its phosphorylated form acts as an energy reservoir necessary, among others, to maintain membrane potential (Kurosawa and Hamaoka 2017). Probably, during 4 months on KD, ketone bodies became the source of this energy, which is why in the cortex a decrease in the creatine content was observed while in the hippocampal formation it remained stable.

The above-mentioned changes in levels of metabolites associated with glial cells were accompanied by an increase in NAA found in the hippocampal formation. N-Acetylaspartate is one of the most common molecules in the brain. Its unique properties determine its extremely high concentration in the brain and its strong signal recorded with magnetic resonance spectroscopy (Setkowicz et al. 2015). Thus, it is represented by the largest peak in the spectrum of healthy brain tissue (Moffett et al. 2007) and its decrease is characteristic of many pathological conditions including, among others, different brain damages, ischemia, cancer, multiple sclerosis, or Alzheimer disease, (Danielsen and Ross 1999).

Despite the prevalence of NAA in the brain, its role remains not fully understood. It is thought to act as an osmolyte compensating anion deficiencies in neurons. It is considered as a precursor to NAAG dipeptide and the source of acetate necessary for the synthesis of myelin lipids. Supposedly, NAA supports the metabolism of neuronal mitochondria (Moffett et al. 2007). It is located in neurons. including their processes, at levels depending on specificity of a given neuronal population (Moffett et al. 2007). Besides that, strong interactions between glial cells and neurons mediated by NAA were detected. Baslow (2000) described this signaling pathway considering the fact that both NAA and NAAG are synthesized in neurons while the related catabolic enzymes are found in oligodendrocytes and astrocytes.

Therefore, NAA released from neurons may be a signal for oligodendrocytes and NAAG for astrocytes (Moffett et al. 2007). Then, after completing the appropriate catabolic reactions in target cells, the substrates can be reutilized by neurons for NAA and NAAG resynthesis. It can be concluded that the elevation of NAA and NAAG contents following KD is a natural functional consequence of presence of NAA as the source of acetoacetate for the production of myelin. The number of mitochondria and levels of their metabolism increase (Hartman et al. 2007), so it seems to lead to higher concentration of NAA, which is involved in energy processes in mitochondria. Thus, the supply of substrates necessary for the myelination process increases, affecting particular brain regions differently.

Following the high fat diet conditions in our investigation, the striatum, as the biggest brain region of interest, showed the most spectacular signs of increases in myelinated fibers. These increases were accompanied by declines in parameters for fibers bypassing the striatum and other ending in the internal capsule. It can be assumed that the two fiber components might, at least partially, overlap each other and form a part of connections with the neocortex. The striatum controls motor functions through brainstem structures and pathways passing through the pons. Thus, the relatively small increases in the pons volume and in the length of 10+ fibers terminating in it could have been associated with the increases found in the striatum. The hippocampal formation and fimbria–fornix, belonging to the same functional subsystem, show consistent increases in 3+ and 30+ fibers which may have functional effects.

All these results indicate that KD could evoke significant changes in intracerebral connections but, because of their preliminary and general character, particular fiber components cannot be recognized.

The effect of ketogenic diet on the intracerebral connectivity has not yet been studied. Our tractography measurements revealed the strongest significant increases related to the striatum, the area responsible for animal movement. Available data on the effect of ketogenic diet on locomotor activity is ambiguous (Murphy et al. 2005; Murphy and Burnham 2006; Thio et al. 2010), and those on the motor system in the normal brain are very rare. So far, KD effects have been reported in diseases with substantial motor dysfunctions and in corresponding rodent models of Alzheimer's and Parkinson’s diseases, amyotrophic lateral sclerosis, or spinal cord injury (Veyrat-Durebex et al. 2018). Little is known about the molecular mechanisms underlying observed improvements, especially when only a selective local impact on the striatum would be considered, including associated increases in the fiber quantity. The subject of our research was the impact of KD on the normal brain, and this means the absence of a pathological condition that can significantly modify the consequences of KD.

In KD-treated rats, an elevated expression of glutamate decarboxylase was detected in the striatum (Cheng et al. (2004). This could be indirect evidence of functional changes in this brain region since this enzyme is involved in GABA synthesis. In the present study, we found quantitative increases in fibers related not only with the striatum but also with the hippocampal formation and fornix. Selective KD effects on these regions, but not on the cerebral cortex, were reported by Zarnowski et al. (2012) but concerning increased levels of kynurenic acid (KA). Similarly to the present study, this regional effect was detected in normal animals without any extra pathological influences. However, Dabrowski et al. (2015) reported that, in the spinal cord, myelin was negatively affected by elevated KA concentration, so it could rather not promote increases in fiber quantities like those in our study. The very diversified results obtained for different regions of the central nervous system might also indicate possible differences between local responses to KD action (Cheng et al., 2004) or even between responses within the same region (Balietti et al. 2008). This might also depend on changes in a local functional status or the type of neuronal subsystems within a given structure. Therefore, these results need more detailed examinations.

The present study showed that KD applied to normal animals had a little effect on general brain anatomy, detected only as a volumetric increase of the pons. However, significant changes occurred in the pattern of intracerebral connections. They were most strongly related to the striatum and appeared to point to changes in the role of this region in the entire neural network of the brain. At the same time, as the result or cause of these changes, there were increases in the concentrations of metabolites being of critical functional importance for the nervous system such as NAA, glutamate or glutathione. Since the most significant changes were related to metabolites associated with glial cells, especially astrocytes, it can be assumed that these cells are the main target of KD and can also underlie its neuroprotective effects. Confirmation requires further research.