The effects of a ketogenic diet on ATP concentrations and thenumber of hippocampal mitochondria in Aldh5a1−/− mice
Kirk Nylen, Jose Luis Perez Velazquez, Venus Sayed, K. Michael Gibson, W.M.Burnham, and O. Carter Snead III
Summary
BACKGROUND—Succinic semialdehyde dehydrogenase (SSADH) deficiency is an inborn error of GABA metabolism characterized clinically by ataxia, psychomotor retardation and seizures. Amouse model of SSADH deficiency, the Aldh5a1−/− mouse, has been used to study thepathophysiology and treatment of this disorder. Recent work from our group has shown that the ketogenic diet (KD) is effective in normalizing the Aldh5a1−/− phenotype, although the mechanism of the effect remains unclear.
METHODS—Here, we examine the effects of a KD on the number of hippocampal mitochondria as well as on ATP levels in hippocampus. Electron microscopy was performed to determine the number of mitochondria in the hippocampus of Aldh5a1−/− mice. Adenosine triphosphate (ATP)levels were measured in hippocampal extracts.
RESULTS—Our results show that the KD increases the number of mitochondria in Aldh5a1−/−mice. We also show that Aldh5a1−/− mice have significant reductions in hippocampal ATP levels ascompared to controls, and that the KD restores ATP in mutant mice to normal levels.
CONCLUSIONS & GENERAL SIGNIFICANCE—Taken together, our data suggest that the KD’s actions on brain mitochondria may play a role in the diet’s ability to treat murine SSADH deficiency.
Introduction
Succinic semialdehyde dehydrogenase (SSADH) deficiency is a rare, autosomal recessive disorder of γ-aminobutyric acid (GABA) biotransformation[1,2]. A murine analog of SSADH deficiency, the Aldh5a1−/− mouse, was developed to study the pathophysiology and treatment of this disorder[3]. Aldh5a1−/− mice exhibit developmental delay, ataxia and a seizure disorder that progresses from absence seizures (~post-natal day, P, 15) to lethal status epilepticus (~P25)[3,4]. Recent work from our laboratory showed that the ketogenic diet (KD) is effective inprolonging the lifespan of Aldh5a1−/− mice[5]. Our studies showed that the KD normalizes perturbations to GABAergic systems, but that these effects may not fully explain the beneficial effects of the KD in this model. Recent trends in the literature have caused us to turn our attention towards the role of mitochondria in the KD’s mechanism of action in Aldh5a1−/−mice.
Mitochondria provide the majority of energy for cellular function. The energy comes fromadenosine triphosphate (ATP), which is produced by the Krebs cycle and the electron transport chain in mitochondria, through the oxidation of fats, carbohydrates and proteins[6]. Sauer and colleagues found a significant, hippocampal-specific impairment of mitochondrial function in Aldh5a1−/− mice, as compared to wildtype mice[7]. Other deficits have been found that implicate mitochondrial function. Gibson and colleagues reported a significant decreasein glutathione levels and increased apoptotic cell death in the hippocampus of Aldh5a1−/− mice, which is consistent with diminished mitochondrial function[8]. Hogema et al. likewise showed significant levels of gliosis in the hippocampi of Aldh5a1−/− mice[3]. Both apoptosis and gliosiscan be caused by reactive oxygen species, which are produced by unhealthy mitochondria[9].The KD has been shown to cause mitochondrial biogenesis in normal rats, resulting in a 50% increase in the total number of mitochondria and a corresponding significant up-regulation of mitochondrial-associated mRNA[10]. Masino et al.[11] have also found that the KD causes asignificant increase in brain ATP levels. These studies concluded that changes in brain energy metabolism might underlie the KD’s actions. Recently, Aldh5a1−/− mice have been shown to have a significantly impaired ability to oxidize glucose as compared to wildtype mice[12].Given the disruptions in hippocampal mitochondrial function reported in Aldh5a1−/− mice andthe reported beneficial effects of the KD on mitochondria, it was hypothesized that the KD would ameliorate the diminished mitochondrial function in Aldh5a1−/− mice. The present study, therefore, used electron microscopy to quantify the number of mitochondria inhippocampal CA1 pyramidal neurons of KD fed Aldh5a1−/− mice, as well as CD fedAldh5a1−/− mice and Aldh5a1+/+ mice.
ATP levels were measured in hippocampal tissue from the above-mentioned groups since a net increase in the number of mitochondria may not equate to a net increase in the production of ATP.
We therefore hypothesized that Aldh5a1−/− mice would have significantly reduced number and function of mitochondria in the hippocampus. Also, in keeping with previous reports[10] we hypothesized that KD fed mutants would show a restored number of mitochondria, and a significant elevation of ATP levels.
Materials and Methods
Subjects
Aldh5a1−/− and Aldh5a1+/+ mice served as subjects for the present experiments. All subjects were housed in a pathogen free environment with controlled lighting (12h light/12h dark, lightson at 7am). Water was available to all subjects ad libitum. All experimental protocols wereapproved by the Hospital for Sick Children Laboratory Animal Services Committee.
Experiments were carried out in accordance with the guidelines of the Canadian Council onAnimal Care.
Diets
A 4:1 ketogenic diet (KD) served as the KD in all experiments. Standard laboratory mouse chow served as the control diet (CD) in all experiments[In other papers it is explained that the mouse chow contains 80-85% carbs)]. All diets were available to the micead libitum.
The composition of the 4:1 KD has been published elsewhere[5]. The 4:1 KD was composedof four parts fat to one part of combined carbohydrate and protein (by weight), to provide aclassic KD. The classic KD is the most often used clinically. Powder containing protein andmicronutrients such as mineral and vitamins was obtained from Harlan Teklad (Madison, WI;TD.03490) and was stored at 4 Celsius. Fats in the form of unsalted butter, lard and canola oilwere added to the powder[5]. The KD, once made, was stored at 4° Celsius.
Normal laboratory rodent chow (Purina, #5001) served as the control diet for all experiments[5].
All diets were introduced to subjects on postnatal day 12. The reason for administering the KD this early was to ensure that any non-suckling feeding that took place was still ketogenic innature.
Tissue Preparation for Electron MicroscopyFor the electron microscopy experiment, subjects were divided into three groups: CD fedAldh5a1+/+ (N=3), CD fed Aldh5a1−/− (N=3) and KD fed Aldh5a1−/−(N=3). Between 22–25days of age, mice were anesthetized with 0.01mg/kg sodium pentobarbital (injected i.p.). Uponreaching a surgical plane of anesthesia, mice were perfused transcardially with 0.1M phosphatebuffer for 5 minutes at a rate of 5ml/min using a varistatic infusion pump (Model 72-315-000Manostat™, Barnant Company, Barrington, IL, USA). This was followed by a 10 minuteperfusion with 2% glutaraldehyde in 0.1M phosphate buffer (pH 7.4) at the same rate. Uponcompletion of the perfusion, brains were extracted and post-fixed by submersion in theglutaraldehyde fixative solution for at least two days.
Transverse coronal sections were subsequently cut at 50μm using a Vibrotome (Series 1000;Technical Products International; Ellisville, MD). Hippocampi were then dissected out of thesesections and washed 3 times with 0.1M sodium cacodylate buffer, pH 7.3. Three sections weretaken from each brain. The tissue was then treated with 1% osmium tetroxide and 1.25%potassium ferrocyanide in cacodylate buffer for 1.5 hours at room temperature. It was thenwashed 3 times with cacodylate buffer. The hippocampal sections were then dehydratedthrough a graded series of ethanol (EtOH) as follows: 70% EtOH (2 × 10 minutes), 90% EtOH(2 × 10 minutes) and 100% EtOH (3 × 10 minutes). Samples were then infiltrated with EPON™resin (Hexion Inc, Houston TX, USA) as follows: 100% propylene oxide (3 × 10 minutes), 1:1EPON™ to propylene oxide for 2 hours, 3:1 EPON™ to propylene oxide for 2 hours, 100%EPON™ overnight and finally, 100% EPON™ for 4 hours. Samples were then placed in freshEPON™ which was polymerized overnight in a 70°C oven.
Sections were further cut at 80nm using an UltraCut Leica EM FCS system (LeicaMircosystems; Wetzlar, Germany) and collected on copper grids and stained with uranylacetate and lead citrate.
Electron Microscopy
10 CA1 pyramidal cell bodies were imaged from each subject, with three subjects per group. Sections were examined using a FEI Tecnai G2 F20 transmission electron Microscope (FEICompany, Hillsboro, Oregon). The CA1 pyramidal cell region of the hippocampus was located.Pictures of somatic mitochondria were obtained at a magnification of 6900× to 8500×.
Analysis of Mitochondrial Counts
The number of mitochondria per micrograph was blindly counted by two independent researchers. The mean of the two researchers’ counts was used for subsequent analyses. ImageJ(v. 1.38X, National Institutes of Health, USA) image analysis software was used to determine the density of mitochondria in the soma, as well as the total area of the soma—excluding the nucleus, which does not contain mitochondria—that was occupied by mitochondria. This calculation gave the percent-area that the mitochondria occupied in the cell body.
Tissue Preparation for ATP Assay
For the ATP experiment, subjects were divided into three groups: CD fed Aldh5a1+/+ (N=13),CD fed Aldh5a1−/− (N=11) and KD fed Aldh5a1−/− (N=12). Between P22–25, mice from eachgroup (one at a time) were injected with 0.01mg/kg sodium pentobarbital (injected i.p.). Upona surgical plane of anesthesia, each subject was decapitated and its brain was extracted.Between 5–20μg of tissue was extracted from the left and right hippocampi. Upon removal,the tissue was immediately flash frozen by submersion in liquid nitrogen. Samples were storedat −80° until the assay was performed.
ATP Calibration Curves
Immediately before running the assay, samples were removed from the freezer and thawed oncrushed ice. ATP levels were determined using a commercially available ATP-Glo ™Bioluminometric Cell Viability Assay kit (#30020-1, Biotium Inc., Hayward, CA, USA).ATP calibration curves were generated according to kit. The assay uses firefly luciferase inthe presence of ATP, which oxidizes D-luciferin resulting in the emission of light. Lightemission levels were measured using a luminometer (Turner Designs, Inc., Sunnyvale, CA).A series of ten-fold titrations from 100 ρmoles (picomoles) to 0.01 ρmoles of ATP wereprepared in 100μL of distilled water (DH2O) for each sample in a 1.5mL microfuge tube.100μL of ATP-Glo™ detection cocktail was then added to each microfuge tube containing theindicated amount of ATP. Each microfuge tube was agitated to ensure thorough mixing beforethe tube was placed in the luminometer. Light emission was integrated over 10 seconds withno pre-read delay. A sensitivity setting of 31% was used.
Quantification of ATP Production
After the calibration curves were run, ATP levels from the tissue samples were quantified. Firefly luciferase was added to the luciferin –containing ATP-Glo™ assay solution in a ratioof 1μL to 100μL (25μL luciferase for 2.5 mL of the ATP-Glo™ assay solution). The ATPGlo™ Detection Cocktail was prepared immediately before each use according to themanufacturer’s directions.
The luminometer was always adjusted to the settings obtained when running the standard samples. As such, the luminometer was set with a delay time of 0 seconds and an integrationtime of 10 seconds. The sensitivity setting was 31%. Samples were run, one-at-a-time, in thesame order that they were prepared. One hundred μL of ATP-Glo™ Detection Cocktail wasadded to each sample. Each tube was manually agitating before the tube was placed in theluminometer and measurement initiated. The relative luminescence activity was recorded andthe next sample was then prepared. Relative luminescence was translated into ATP concentration using the calibration curves constructed earlier.
Results
Mitochondrial Density
Figure 1 shows representative electron micrographs from CA1 pyramidal cells ofAldh5a1+/+ mice fed a CD (a), Aldh5a1−/− mice fed a CD (b) and Aldh5a1−/− mice fed a KD(c). Using electron microscopy we counted the number of mitochondria present in the somata of CA1 pyramidal cells from the brains of CD fed Aldh5a1+/+ mice, CD fed Aldh5a1−/− miceand KD fed Aldh5a1−/− mice.
Figure 2a shows the mean (±s.d.) number of mitochondria per 10μm2 in electron micrographstaken from CD fed Aldh5a1+/+ mice, CD fed Aldh5a1−/− mice and KD fed Aldh5a1−/− mice.This gives a measure of mitochondrial density. As indicated, the mean number of mitochondria was lowest in CD fed wildtype mice, intermediate in the CD fed mutants and the highest in the KD fed mutants. A one-way ANOVA was used to compare group means. A significant difference was detected among the groups (F=4.569, p=0.019). Tukey’s post-hoc analyses revealed that KD fed Aldh5a1−/− mice (mean±s.d; 2.58±0.52) had a significantly higher density of mitochondria than Aldh5a1+/+ mice (2.03±0.24; p<0.05). The CD fed Aldh5a1−/− mice did not differ significantly from either of the other groups (p>0.05).
Mitochondrial Area
Figure 2b shows the mean (±s.d.) somatic area occupied by mitochondria (calculated as apercentage of total somatic area) in CD fed Aldh5a1+/+ mice, CD fed Aldh5a1−/− mice and KD fed Aldh5a1−/− mice. The mean somatic area was lowest in CD fed wildtype mice, similarly low in the CD fed mutants and significantly elevated in the KD fed mutants. A one-way ANOVA detected a significant difference among the groups (F=6.626, p=0.0046). Tukey’spost-hoc analyses showed that mitochondria occupy a significantly larger area of the soma in KD fed mutant mice (mean±s.d.; 8.00±2.25) than CD fed wildtype mice (6.03±0.55; p<0.05).The difference between KD fed mutant mice and CD fed mutant mice approached significance(p=0.06), but was not statistically significant. There was no statistical difference between CD fed mutant mice and CD fed wildtype mice.
ATP Quantification in Hippocampal Tissue
Figure 3 shows the mean (±s.d.) hippocampal ATP levels (expressed as picomoles ATP perdecigram of tissue) in CD fed Aldh5a1+/+ mice, CD fed Aldh5a1−/− mice and KD fedAldh5a1−/− mice. As indicated, hippocampal ATP levels are high in CD fed wildtype mice andKD fed mutant mice, intermediate in KD fed wildtype mice and low in CD fed mutant mice.A one-way ANOVA showed a significant different between the groups (F=3.90, p=0.03). ATukey’s post-hoc revealed that CD fed mutants had significantly lower ATP levels than CD fed wildtype mice and KD fed mutant mice (p<0.05). Therefore, CD fed mutants had significantly decreased hippocampal ATP levels whereas KD fed mutants had normal levels of hippocampal ATP.
Discussion
The present experiments were designed to explore whether KD-induced changes to mitochondria in Aldh5a1−/− mice play a role in the diet’s mechanism of action in these mutantmice.
The present study found that the KD does act to increase the number of mitochondria in theCA1 pyramidal cells of Aldh5a1−/− mice. The KD also normalizes the deficits in hippocampal ATP levels that are seen in CD fed Aldh5a1−/− mice.
Mitochondrial Profiles
Experiments on mitochondrial profiles were designed to determine whether Aldh5a1−/− micehad normal mitochondrial numbers and size as compared to wildtype controls. We hypothesized that Aldh5a1−/− mice would have significantly fewer hippocampal mitochondria and that the KD would elevate mitochondrial numbers in Aldh5a1−/− mice.
Mitochondria Number—Interestingly, mitochondrial number is not lower in CD fedmutants, as hypothesized. Sauer et al. showed that CD fed mutants have impaired hippocampal mitochondrial function[7]. Our data suggest that this impairment is not caused by a decrease in the number of mitochondria.
The present study confirms the findings of Bough et al. by showing that the KD increases the number of hippocampal mitochondria[10].
Mitochondrial Size—Although the above study showed that KD fed mutants had significantly more mitochondria, it was not clear whether these mitochondria were normal in size. Therefore, the next study was to compare the percentage of the cell body that is occupied by mitochondria. We found that approximately 6% of the soma was occupied by mitochondriain CD fed wildtype mice and CD fed mutant mice. This jumped, however, to approximately 8% in KD fed mutant mice.
These data show that the KD-induced increase in mitochondrial number corresponds to an increase in the area of the soma occupied by mitochondria.
Mitochondrial ATP Levels in Hippocampus
This experiment was performed to test the hypotheses that CD fed mutant mice would have lower ATP levels than wildtype mice, and that the KD would restore ATP levels in the mutant mice. We found that hippocampal ATP levels are high in CD fed wildtype mice and KD fed mutant mice, and low in CD fed mutant mice. An ANOVA revealed that hippocampal ATP levels were significantly lower in CD fed mutant mice as compared to CD fed wildtype mice and KD fed mutants.
Sauer et al. showed that hippocampal neurons from CD fed Aldh5a1−/− mice have significantlyimpaired mitochondrial function[7]. Specifically, they identified deficiencies in complex I-IVof the electron transport chain, which is essential for the aerobic production of ATP. Our data extend the findings of Sauer et al.[7] to show that CD fed mutants have significantly lower hippocampal ATP levels. Significantly reduced hippocampal ATP may play a role in thephenotype of Aldh5a1−/− mice. This possibility is discussed further below.
Previous reports have suggested that the KD elevates ATP levels in the brain[11,13], while other groups have failed to show such an elevation[10], but instead found a significant increase in other high-energy bonds as evidenced by a significance increase in the phosphocreatine-tocreatineratio[10].
Relationship Between Mitochondrial Data and Seizures in Aldh5a1−/− Mice
The observation that Aldh5a1+/+ mice have similar a number of mitochondria as Aldh5a1−/−fed a CD, and at the same time the mutants have lower levels of ATP, supports the previous observations of mitochondrial impairment in the mutants[7]. It was shown in other studies that mitochondrial dysfunctions and oxidative stress are closely related to epileptiform activity.Thus, in vivo kindling was reported to increase oxidative stress leading to neurodegenerationin the hippocampi of rats[14]. Increased free radical production was also reported in other invivo seizure models[15], and in vitro studies demonstrated a close correlation between paroxysmal events and free radical formation as well as intracellular calcium accumulation,which causes mitochondrial dysfunction[16]. In addition, oxidative stress is also known to enhance the propensity to paroxysmal activity by altering the intrinsic membrane properties of neurons[17], hence there seems to be a clear relationship between oxidative stress/mitochondrial dysfunction and epileptiform activity[18][In the next paper I'll quote from, there is much more about stress and the possible benefits from the KD.] Considering these observations, it is therefore possible that Aldh5a1−/− have mitochondrial alterations, which may enhance the propensity of seizures, which in turn cause more oxidative stress and mitochondrial impairment.
Role of mitochondria in the KD’s actions in Aldh5a1−/− mice
Lacking the SSADH enzyme results in numerous abnormalities that contribute to the phenotypeof Aldh5a1−/− mice. GABA and GHB are highly elevated and almost certainly play a role inthe absence seizures[4], decreased TBPS binding[5,19] and reduced mIPSC activity[5] inAldh5a1−/− mice. Elevation of these neuroactive compounds, however, might not explain the tonic-clonic seizures[4] (whose onset is not blocked by the suppression of absence seizures using ethosuximide), ataxia, and weight loss[5].
We hypothesize that Aldh5a1−/− mice experience a significant energy deficit as a result of the inability to efficiently utilize glucose as an energy substrate[12]. [Well, this might apply to humans as well, since although we do seem to have the ability to utilize glucose for energy, we haven't evolved to live from sugar, and we end up paying a high price with the normal diet! ] This idea is further supportedby data showing that Aldh5a1−/− fed a CD are ketotic (even when infused with glucose)[5,12], and that the onset of weight loss and seizures corresponds to the time of weaning inAldh5a1−/− pups[3]. Also, upon autopsy, Aldh5a1−/− mice have little-to-no body fat, suggesting that they are oxidizing their fat stores for energy, even while on a high carbohydrate diet[5]. Our working hypothesis is that the KD’s beneficial effects in Aldh5a1−/− mice are a result of the diet’s ability to offer an alternate oxidizable energy substrate, i.e., fat. This yields asignificant increase in ketone bodies, which can be oxidized for energy production in place of glucose. This restores the amount of energy available to the mice, allowing processes that were perturbed, due to inadequate energy, to begin functioning more normally. The present data support this idea as KD fed mutants had significantly more mitochondria. Further, CD fed mutants had significantly less hippocampal ATP, whereas KD fed mutants had normal hippocampal ATP levels.
This hypothesis is not without its problems, however, as it does not explain why Aldh5a1−/−mice still progress in their disease state—albeit significantly more slowly—when supplied with an alternative energy source through administration of the KD. One possible explanation relates to the fact that ketones can only account for 30–60% of total brain energy[20,21]. This is due to the rate-limiting step in ketone body utilization in the brain, which is the transport of acetoacetate and beta-hydroxybutyrate into the brain via the monocarboxyllic transporter[20].If mutant mice are not using glucose efficiently, and ketones can only account for 30–60% of total brain energy, then perhaps brain function begins to break down after prolonged exposur eto inadequate levels of energy substrate. This may explain why mutant mice fed a KD ultimately succumb to the same fate, albeit significantly later in life, than CD fed Aldh5a1−/− mice. [This suggests that genetic diseases might be prevented from progressing when it is not possible to find a cure. In some cases, though, it might actually restore the mitochondrial function enough to allow the body to compensate for the deficit and heal?]
Summary
The present study found that the KD does act to increase the number of mitochondria in theCA1 pyramidal cells of Aldh5a1−/− mice. The KD also normalizes the deficits in hippocampal ATP levels that are seen in CD fed Aldh5a1−/− mice. Taken together, the KD’s beneficial effects in Aldh5a1−/− mice may be mediated, in part, through the diet’s actions on mitochondria.
I sometimes wish my body could send me an alert making me aware what feels right or not to my health. The fact that doesn´t help much either is as I made the food reintroduction after two months, as described in the Ultra mind solution, I wasn´t able to detect any discomfort. Also, I never suffered constipation, congestion but some gas when starting the diet that immediately subsided and don´t have to lose weight either. So either I´m the typical silent yeast sufferer or something else is going on. So I guess I´ll start all over again staying at least six months away from any body stressor food. 
