Indispensable Amino Acid–Deficient Diets Induce Seizures in Ketogenic Diet–Fed Rodents, Demonstrating a Role for Amino Acid Balance in Dietary Treatments for Epilepsy

Abstract

Background

Low protein amounts are used in ketogenic diets (KDs), where an essential (indispensable) amino acid (IAA) can become limiting. Because the chemically sensitive, seizurogenic, anterior piriform cortex (APC) is excited by IAA limitation, an imbalanced KD could exacerbate seizure activity.

Objective

We questioned whether dietary IAA depletion worsens seizure activity in rodents fed KDs.

Methods

In a series of 6 trials, male rats or gerbils of both sexes (6–8/group) were given either control diets (CDs) appropriate for each trial, a KD, or a threonine-devoid (ThrDev) diet for ≥7 d, and tested for seizures using various stimuli. Microchip analysis of rat APCs was also used to determine if changes in transcripts for structures relevant to seizurogenesis are affected by a ThrDev diet. Glutamate release was measured in microdialysis samples from APCs during the first meal after 7 d on a CD or a ThrDev diet.

Results

Adult rats showed increased susceptibility to seizures in both chemical (58%) and electroshock (doubled) testing after 7 d on a ThrDev diet compared with CD (each trial, P ≤ 0.05). Seizure-prone Mongolian gerbils had fewer seizures after receiving a KD, but exacerbated seizures (68%) after 1 meal of KD minus Thr (KD–T compared with CD, P < 0.05). In kindled rats fed KD–T, both counts (19%) and severities (77%) of seizures were significantly elevated (KD–T compared with CD, P < 0.05).

Gene transcript changes were consistent with enhanced seizure susceptibility (7–21 net-fold increases, P = 0.045–0.001) and glutamate release into the APC was increased acutely (4-fold at 20 min, 2.6-fold at 60 min, P < 0.05) after 7 d on a ThrDev diet.

Conclusion

Seizure severity in rats and gerbils was reduced after KDs and exacerbated by ThrDev, both in KD- and CD-fed animals, consistent with the mechanistic studies. We suggest that a complete protein profile in KDs may improve IAA balance in the APC, thereby lowering the risk of seizures.

Introduction

The ketogenic diet (KD) has been used in the treatment of seizure disorders for decades. It was introduced in the early 1900s as an improvement over starvation, which had been considered beneficial for centuries. Yet, starvation is not useful in the long term. The rationale for using a KD was that, like starvation, it induces ketosis (1). Considerable effort has gone into determining the physiologic basis for the effects of the KD [e.g., (2–4)]. Nevertheless, the KD is not always successful; many patients continue to have seizures on the KD protocol.

Another feature of starvation, less often mentioned, is that in addition to causing ketosis, protein breakdown releases free indispensable amino acids (IAAs). We suggest that the very low amounts of protein allowed in some of the KD protocols could be problematic. In low-protein feeding, a modest imbalance of IAA can activate (5) the chemically sensitive anterior piriform cortex (APC) (6). Using electrophysiologic methods, we have shown that the APC is activated by IAA depletion both in vivo and in vitro (5, 7). We have also shown that IAA deficiencies of Thr or Ile (or His in young, 100- to 120-g rats) can exacerbate chemically induced seizures (8). Could it be that the beneficial effects of starvation in seizure disorders are due to the restored balance of plasma IAAs, rather than, or in addition to, ketosis? Recently, KD protocols have included more protein, e.g., the Modified Atkins Diet (MAD) and others, reviewed by Martin et al. (9). Nevertheless, little has been published about the involvement (or even the content) of IAAs in KD protocols.

Three separate lines of evidence support the hypothesis that IAA balance may be important in seizure control. First, consistent with our work in rats tested with chemostimulants (8), others have shown that ingestion of an IAA-deficient food, as well as protein malnutrition, can increase seizure susceptibility and abnormal neural activity (10, 11). Moreover, inborn errors of AA metabolism can be associated with seizures (12).

Second, in the brain, the result of IAA imbalance is excitation of the ventrorostral area of the APC (APCvr), i.e., the chemosensory area for IAA deficiency (13–15), which is also known as the “area tempestas” (16). The APCvr, the most seizurogenic brain area, is known for its role in limbic seizures, and has a paucity of inhibitory components (6). Critically, after a Thr-devoid (ThrDev) meal, important proteins, including the potassium chloride co-transporter (KCC2), crucial for γ-aminobutyric acid (GABA)-ergic inhibition in neurons (17), are diminished in the cells of the APCvr (18). With less inhibition, positive feedback loops involving excitatory glutamatergic neurons are easily potentiated, resulting in increased activity in the glutamatergic output cells of the APCvr and their projection sites (19).

At the biochemical level, we have shown that an IAA-deficient meal (using the “aminoprivic” model where animals are pre-fed a low-protein diet and food restricted overnight) reduces the limiting IAA concentration in the APC, increasing concentrations of uncharged tRNA, and activating the general amino acid control kinase 2 (GCN2) cascade (20, 21). This results in inhibition of global protein synthesis and the loss of proteins associated with the GABAergic system, including KCC2, in the APC (18). Diminution of the GABAergic system leads to increased seizures in susceptible individuals and explains the chemosensitivity of the APCvr to IAA-deficient diets.

Taken together, the evidence suggests that IAA imbalance or deficiency may contribute to increased seizurogenicity in seizure disorders. This would underscore the importance of balanced IAA nutrition for seizure patients, particularly where the KD has not been successful in reducing seizures.

Here we studied the effects of IAAs and the KD on seizure activity in rodents. Rats and gerbils were given an appropriate control diet (CD), a KD, or a ThrDev diet, or combinations of these diets, and then tested, to learn if IAAs may play a role in the dietary treatments of seizure disorders.

Methods

Maintenance and feeding of experimental animals

Rats

Young adult male Sprague-Dawley rats (Bantin and Kingman, Freemont, CA) weighing 150–350 g were used, except as indicated. Animal protocols were approved by the UC Davis Institutional Animal Care and Use Committee. Rats were housed individually, in hanging wire cages, in an artificially illuminated 22 ± 2°C vivarium with a 12 h:12 h light-dark cycle and offered water and powdered diets as described below.

Gerbils

Under protocols approved by the Vassar Animal Care and Use Committee, age- and gender-matched male and female gerbils (raised and pretested in the Vassar gerbil colony) determined to be seizure sensitive (n = 64 of the 92 gerbils tested) were maintained in a temperature- and humidity-controlled colony room under a 12 h:12 h light-dark schedule, with weekly bedding changes and daily food checks. Water was provided ad libitum.

Diets

The diets (Table 1) have been described in detail previously (22–24). Briefly, the diets were made of semipurified ingredients with casein, or purified L-AAs, as the sole protein source. The CDs were basal control diet (CD_Bas) with Thr, corrected [replete with all IAAs (CD_COR)], control diet with 10% casein (CD_Cas10), or control diet with 14% casein (CD_Cas14). Diets that were devoid of Thr (as a representative IAA) were prepared by removing Thr from the appropriate diet, i.e., ThrDev = CD_COR minus Thr; KD–T = KD minus Thr. The ThrDev diet is a controlled dietary model reliably producing an IAA imbalance (25). All diets contained the necessary vitamins and minerals and 5% (or, for the KDs, 70%) corn oil or vegetable shortening (Crisco) as the fat source with starch:sucrose (2:l, wt:wt) as the carbohydrate. Carbohydrate was proportionately reduced when AAs or fat were added.

TABLE 1.

Composition of diets used in the rat and gerbil experiments¹

	CDs				KDs		ThrDev Diets
Ingredients, g/kg diet	CDBas	CDCOR	CDCas10	CDCas14	KDBas	KDCas	ThrDev diet	KD–T
Dispensable AA mix2	80.8	80.8	—	—	80.8	—	80.8	80.8
Indispensable AA mix3	43.7	142.3	—	—	43.7	—	142.3	43.7
Thr	2.0	10	—	—	2.0	—	0	0
Casein + 0.3 Met	—	—	103	143	—	103	—	—
Sucrose	254.1	218.6	262	248.7	37.5	45.3	222	38.2
Corn starch	508.4	427.2	524	497.3	75	90.7	443.9	76.3
Vitamin mix4	10	10	10	10	10	10	10	10
Salt mix4	50	50	50	50	50	50	50	50
Corn oil	50	50	50	50	0	700	50	0
Vegetable shortening5	0	0	0	0	700	0	0	700
50% choline chloride	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0
Total	1000	1000	1000	1000	1000	1000	1000	1000

¹For details of specific diets used in any trial, see methods for that trial. In diets where amino acids, protein, or fat were changed, the difference was made up by adjusting the carbohydrates proportionately, sucrose:corn starch, 2:1. CD_Bas, basal diet; CD_{Cas 10}, 10% casein diet; CD_{Cas 14}, 14% casein diet; CD_COR, fully replete diet; IAA, indispensable amino acid; KD, ketogenic diet; KD_Bas, basal ketogenic diet plus 70% fat; KD_Cas, ketogenic diet in which the protein fraction is 10% casein and the fat is increased to 70%; KD–T, ketogenic diet without Thr; ThrDev, Thr-devoid.

²Dispensable amino acid mix, grams of amino acid per 1000 g diet: Ala, 3.5; Arg, 8.3; Asp, 10.0; Glu, 30.0; Gly, 10.0; Pro, 10.0; Ser, 3.5, plus sodium acetate, 3.8.

³IAA mixes for CD_COR included all the IAA except Thr (L-amino acid, grams per 1000 g diet): Cys, 6.0; His, 8.0; Ile, 14.0; Leu, 20.0; Lys-HCl, 24.0; Met, 9.0; Phe, 14.5; Trp, 3.8; Tyr, 8.5; Val, 15.0; sodium acetate, 10.8; proportionately reduced for CD_Bas, KD_Bas, and KD–T.

⁴Described previously in Hammer et al. (24).

⁵Crisco brand vegetable shortening.

Seizure testing

Seizures were induced with either chemical stimulation using intraperitoneal injections of the GABA antagonist (26) pentelenetetrazole (PTZ: 15, 30 or 40 mg/kg; Sigma-Aldrich) as described previously (8), or the maximal electroshock seizure (MES) protocol.

The 5-point Racine scale (27) was used to evaluate seizures as follows: 0.5, jaw clonus; 1, myoclonic jerks of the forelimbs; 2, mild forelimb clonus with or without facial clonus lasting ≥5 s; 3, severe forelimb clonus lasting ≥15 s; 4, rearing in addition to severe forelimb clonus; 5, rearing and falling in addition to severe forelimb clonus. Latency (time between drug injection and first seizure) and duration of the seizures were also recorded in some trials.

Statistical analysis

Data were analyzed by 1- or 2-way ANOVA. Post hoc testing, after a significant ANOVA, used Fisher's protected least-significant differences test for comparisons of means among >2 groups. Multiple regression was used to compare diet, and days on diet, with seizure level, latency, and percentage of animals seizing, using the SAS statistical program. We also used Student's 2-tailed t test if 2 groups were tested and for preplanned comparisons, or the chi-square test. Where appropriate, corrections for multiple comparisons (Bonferroni) or repeated measures were applied. For comparing the 2 gerbil studies, Pearson's correlation was used. Nonparametric tests included the t test for rank and the Wilcoxon signed rank test as recommended by Snedecor and Cochran (28). Trend lines yielding slope and R² values for glutamate in Trial 6 were estimated using Excel (Microsoft). Values are expressed as mean ± SE. Significance was assumed at P ≤ 0.05.

Trials

Trial 1: Chemical induction

PTZ was used to evaluate dosage and protocol, and to replicate our previous work (8) showing that seizures in response to chemostimulants are more severe with IAA deficiency. A previous paired-feeding trial in rats (8) showed that food intake depression did not alter the seizure result; this was not repeated here. All rats (n = 99) were conditioned on CD_Bas for ≥7 d prior to introduction of the experimental diets. Then the rats (n = 6–10/group) were fed either ThrDev diet or CD_COR for 3, 5, 7 (repeated twice), or 12 d. At the end of the assigned feeding period, each rat was tested for seizure susceptibility with PTZ (40 mg/kg). Seizures were scored on the Racine scale described above. After PTZ injection, the time to onset of seizure activity (Racine level 0.5 or 1) and the severity (Racine level 1–5) of seizures were recorded for ≥30 min. Data were analyzed by multiple regression using SAS, as described above. In this trial, rats were given PTZ only once to eliminate any kindling effect due to PTZ (29), and therefor the results of this trial are solely due to the effects of the diet over time.

Trial 2: MES testing

MES was employed as a nondrug alternative to assess seizure susceptibility. Classical MES testing for tonic hindlimb extension (HE) was administered as described (30). Rats (n = 6/group) were randomly assigned to the CD_Bas and ThrDev diet groups. Both groups were fed CD_Bas on days 1–8; 4 MES tests were administered during this period. The ThrDev group was then fed the ThrDev diet on days 9–23 and the CD_Bas group continued on CD_Bas. Both groups were given 4 MES tests in this period. Lastly, both groups ate the higher-protein CD_COR for days 24–26 (2 MES tests). The study was repeated with a second cohort of naive rats using the same numbers and treatments, except that the limiting IAA was His (a semi-IAA in adults, an IAA for young animals) and there were fewer MES tests. Data were analyzed by ANOVA, 2-factor with replication (Excel).

Trial 3: Gerbil seizure model

The Mongolian gerbil is genetically predisposed to epileptiform seizures, and has been shown to be a good experimental model (31, 32). Seizure sensitivity develops spontaneously and, in susceptible individuals, seizures can be triggered by innocuous external stimuli, such as being moved to a fresh cage, without any drug or shock treatment.

In the first study, to test for sensitivity to IAA deficiency, pairs of gerbils (total n = 64) of similar weight and sex were randomly assigned to treatment groups (n = 8/group). They were placed on CD_Cas10 (Table 1) for 1 wk. They were next fed the CD_Bas for 1 wk, fasted for 5 h, and then fed a single meal of either ThrDev diet or CD_Bas. After 15 min the gerbils were placed into a novel testing arena, a 38-L glass aquarium. The gerbils’ behavior was videotaped for 30 min and any seizure activity was scored on the Racine scale by observers blind to the dietary treatment. Data were subjected to 2-way ANOVA for seizure levels according to sex and diet.

In the second study to test for sensitivity to KD, young (2–3-mo-old, 8/group) or 2-y-old male seizure-sensitive gerbils (5/group) were fed KD_Cas or CD_Cas10 for 3.5 wk, prior to recording seizure activity. The KD_Cas gerbils were then returned to the CD_Cas10 ad libitum for 2 wk and seizure susceptibility was again tested for duration of the KD effect. Data were analyzed as for the first study and the 2 study results were compared by Pearson's correlation for consistency.

Trial 4A: KD and IAA deficiency in rats

To determine if IAA deficiency would still exacerbate seizures after eating KD, we used 2 diet groups (n = 2–3/group): KD_Bas and CD_Cas14. The rats were fed CD_Cas14 for 20 d and then assigned to KD_Bas or CD_Cas14. On day 21 both groups were fasted for 24 h. On day 22 the assigned diets were given. On days 23 and 30, both groups were tested with PTZ (15 mg/kg). Over 30 min, latency to and number of Racine level 1 seizures were recorded. On day 35, after 2 wk on KD or CD_Cas14, rats in both groups were fasted for 5 h and given a single meal of their respective diet. After 1.5 h the rats were injected with a higher dose of PTZ (30 mg/kg), and latency and number of level 1 seizures were recorded. This protocol was repeated after 2 d but this time the KD_Bas diet was devoid of Thr (KD–T). After 1.5 h on KD–T or the CD_Cas14, they were given PTZ injections (30 mg/kg) and seizures recorded as before. Values are means ± SEs for n = 3 or means for n = 2.

Trial 4B: Kindling

The objective of this study was to learn whether we could kindle rats such that they would provide a seizure-prone rat model, consistent with the genetically seizure-prone gerbils used in Trial 3. Here rats were kindled with 6 doses of PTZ and then fed KD_Bas for 20 d to determine if the KD would protect the kindled rats from the deleterious effects of a ThrDev diet.

Using a similar protocol to Trial 4A, 18 rats (n = 9/group) were pre-fed the CD_Cas10 on days 1–40. For kindling (on days 26–40) the rats were given 6 doses of PTZ (35 mg/kg; 3 times/wk), during the last 2 wk on the CD_Cas10. On day 41 all the rats were fasted for 24 h.

After kindling and KD (KD_Bas) feeding (during days 42–62), all rats were given a test PTZ injection and “baseline” seizure scores were recorded. Kindling was shown by comparing the 9 control rats given 6 PTZ injections to a similar control group (same strain, age, weight, sex, and supplier) from Trial 1, which had received only one PTZ injection. Kindling was highly successful in the kindled (6-PTZ-injections) group, compared with the single-PTZ-injection group (main effects: P < 0.001); post hoc tests for seizure levels showed that the seizures at each of the Racine levels 1, 3, and 5 were significantly greater (P < 0.05) in the kindled rats.

Finally, 3 d after the baseline tests (day 65), the rats, which had now been both kindled and fed KD, were divided into KD–T and CD_COR groups. After 1.5-h access to the test diets, they were retested with PTZ, to affirm the preliminary results with KD–T in Trial 4A. Results were evaluated with the Wilcoxon signed rank test as recommended (28).

Trial 5: Gene expression

Gene expression was evaluated in samples taken from the rat APC after 2 h or 7 d on the ThrDev as in Trials 1 and 2, and compared with control samples taken at the same times from CD_Bas-fed rats. Routine housing, feeding, brain dissection protocols, and mRNA extraction were as reported previously (33, 34). After 7 d on the CD_Bas, rats (n = 42) were divided into 3 groups to be fed either ThrDev diet, CD_COR, or CD_Bas. The diet groups were divided again: half of each diet group (n = 7) was assigned to be studied either after 2 h on the first day, or after 7 d. After the 2-h feeding period on the assigned day, rats were decapitated and their brains removed. The APC was quickly dissected, homogenized in TRIzol (Gibco-BRL), and extracted according to the manufacturer's protocol. Microarray procedures and data analysis were conducted by Genome Systems Inc. (Incyte Pharmaceuticals, Inc.). mRNA in the APC tissues from ThrDev and control rats was labeled with either Cy3 or Cy5 and hybridized to the Affymetrix rat brain chip (Rat Neuro U34), which contains sequences from 1300 known genes in the rat brain. Results for ThrDev compared with CD_Bas at 2 h on day 1 and on day 7 were reanalyzed using the April 2015 revision of the Affymetrix annotation tables and subjected to Gene Ontology analysis (GO Ontology database, released 6 August 2015). For the Gene Ontology analysis, the Bonferroni correction for multiple comparisons was used.

Trial 6: Microdialysis in the APC

Glutamate is the major excitatory neurotransmitter in mammalian brain, and the chief transmitter in the principal output neurons of the APC (35); its release is an indicator of activation in the APC's primary output cells. It has long been known that because IAA imbalance induces protein degradation, the non-IAAs, including glutamate, are increased in the plasma (36). To determine if glutamate release is activated with ThrDev feeding as expected, we used microdialysis of the APC to measure glutamate levels in the extracellular dialysate before and during the first meal of day 8 on the ThrDev diet. Six rats, n = 3/group, bearing cannulas directed toward the APC, were conditioned as described above, and then placed on CD_COR or ThrDev for 7 d, for consistency with the seizure-testing studies above. Before the first meal on day 8, each rat had a dialysis probe inserted into the previously placed cannula aimed at the APC; baseline samples were collected as explained in detail previously (37). After the baseline samples and at the onset of the first meal (t = 0), samples of dialysate were collected at t = 20 min and every 40 min thereafter for 140 min, for a total of 2 baseline (pre-meal) and 5 experimental samples. Glutamate values were obtained by HPLC with electrochemical detection as reported previously (38). Differences between ThrDev diet and CD_COR at 20 and 60 min into the meal, based on previous meal pattern analysis in rats with these diets (39), were determined by Student's t test for preplanned comparisons.

Results

Trial 1: Chemical induction

There were no differences between diet groups after 3 d on the experimental diets (Figure 1). However, after 5 d on the ThrDev diet, the rats seized sooner, i.e., latencies to the first seizure were shorter (Figure 1A, P < 0.05) and seizures tended to be more severe in the ThrDev group compared with the CD_COR group (Figure 1B), as expected (8). Three trials were performed on day 7; the results were statistically similar, and therefore the data for these 3 trials were combined. We found that after 7 d on the ThrDev diet, rats had significantly shorter seizure latency (P = 0.0017) and more severe seizures (P = 0.0031) than the CD_COR group (Figure 1A, B). The results at day 12 were directionally similar, but did not reach significance. Paired feeding was not done here, as explained above and in the Discussion, but the ThrDev diet did produce the expected weight loss (25). The rats fed the ThrDev diet for 12 d showed a 36% weight loss compared with the weight gain in the CD_COR group (P < 0.001), showing appropriate diet preparation.

FIGURE 1.

Effect of ThrDev diet on the latency to and severity of PTZ-induced seizures in naïve male rats tested with PTZ after 3, 5, 7, or 12 d (Trial 1). (A) Latency to first seizure (diet: *P < 0.05 on day 5; **P < 0.01 on day 7). (B) Maximal seizure score using the Racine scale (1–5; 5 most severe) (diet: **P < 0.01 on day 7). Values are means ± SEs (n = 7/group). CD_COR, corrected control diet, fully complete; PTZ, pentelenetetrazole; ThrDev, Thr-devoid.

Trial 2: MES testing

After 5 d on the ThrDev diet (day 14, Figure 2) there was a significant increase in the duration of the seizures (measured as HE in seconds) in the ThrDev group compared with CD_Bas (main effect P = 0.0001; effect of diet, P = 0.0004; effect of days, P = 0.003). Increased seizure severity, compared with the CD_Bas group, continued for the remainder of the ThrDev feeding period, as tested again on days 17 and 22. After the rats were placed on CD_COR the effects were quickly reversed and by day 26 HE scores for the ThrDev group, now eating CD_COR, did not differ from their scores on CD_Bas (P = 0.69). In the His trial (note that His is semi-essential in the adult rat), the effect of diet was directional, tending toward significance for increased duration of HE (s: HisDev compared with CD_Bas, P = 0.08).

FIGURE 2.

Effect of diet on MES-induced seizures in male rats (Trial 2), tested on days indicated on the x axis (diet: *P < 0.05). Seizures were quantified by the duration of hindlimb extension (in seconds). Both groups ate CD_Bas on days 1–9 and CD_COR on days 24–26. Values are means ± SEs (n = 8/group). Diets were CD_Bas, ThrDev diet, and CD_COR, given during the days indicated in horizontal bars below the chart. CD_Bas, basal control diet; CD_COR, corrected control diet, fully complete; MES, maximal electroshock seizure test; ThrDev, Thr-devoid.

Here we have shown, using very different stimuli, that a ThrDev diet produces increased seizure susceptibility (Figures 1 and 2). The rats used in these 2 studies were not otherwise predisposed to seizures and took 5–7 d on the ThrDev diet to develop seizure sensitivity. We next asked whether an animal model known to be seizure prone would respond more quickly to IAA deficiency.

Trial 3: Gerbil seizure model

Increased seizure severity was seen in seizure prone-gerbils that had a single meal of ThrDev (Figure 3). The percentage of ThrDev-fed gerbils having Racine level 1–4 seizures was slightly, but not significantly, greater than those fed CD_Bas (54% for CD_Bas compared with 63% for ThrDev, P > 0.05). However, the occurrence of the most severe seizures was significantly exacerbated by acute IAA deficiency. The percentage of gerbils with Racine level 5 seizures increased dramatically after the ThrDev meal (16% for CD_Bas compared with 50% for ThrDev, P < 0.05); there was no difference due to sex of the animals, and no interaction (P > 0.05 for both sex and interaction). These results demonstrate increased severity of seizures in a genetic animal model of epilepsy, after a single IAA-deficient meal.

FIGURE 3.

Effect of diet on seizure severity in seizure-prone Mongolian male and female gerbils after one meal of ThrDev diet, Trial 3. Seizures were stimulated by moving the animals to a new arena. Maximal seizure score using the Racine scale (1–5; 5 most severe). Values are percentages of gerbils seizing (n = 8/group, P < 0.05). CD_Bas, basal control diet; ThrDev, Thr-devoid.

In the second study, the 2- to 3-mo-old, but not 2-y-old, gerbils on the KD exhibited a significantly lower incidence and reduced severity of seizures compared with both isocalorically pair-fed and ad-libitum-fed CD_Cas10 controls (P < 0.05). In the gerbils, as seen before in the rat (8), there was no difference in the data for seizures between pair-fed and ad-libitum-fed controls (P > 0.05). After 2 wk on CD_Cas10, the gerbils' seizures returned to the higher control levels (P > 0.05). Pearson's correlation analysis showed that agreement between scores in these studies was excellent (r = 0.95, P < 0.001). This study provided evidence that a KD attenuates, but apparently does not cure, the severity or incidence of seizures in young, genetically seizure-prone gerbils.

Trial 4A: KD and IAA deficiency in rats

In this small study the KD_Bas diet appeared to increase the latency and decrease the number of seizures, compared with the CD_Cas14-fed rats, suggesting beneficial effects of KD. The responses were similar with single 15- and 30-mg/kg doses of PTZ (Figure 4A, B, left and middle bars). The effect of a single KD–T meal (right bar in Figure 4A, B) was to shorten the latency to the first seizure from >10 min to 1.5 min, and to increase the number of level 1 seizures from <10 to 28, indicating that a lack of Thr may exacerbate seizures even with a KD.

FIGURE 4.

Effect of diet on severity of PTZ-induced seizures in kindled male rats, Trial 4 (A, B) Preliminary trial: Trial 4A. (A) Latency to first seizure. (B) Number of Racine level 1 (mild) seizures in 30 min. Diets: CD, KD, and KD–T. Values are means ± SEs (n = 2–3/group). (C, D) KD + Kindling, Trial 4B. (C) Total seizures in 30 min. (D) Maximal Racine seizure score (ratio test: baseline, most severe = 5). Values are means ± SEs (n = 7–9/group). Values with different lowercase letters differ, P < 0.05. Diets are given on each horizontal axis. Diets: CD, CD_Bas or CD_COR, KD, and KD–T. CD, control diet with 14% casein; CD_Bas, basal control diet; CD_COR, corrected control diet, fully replete; KD, ketogenic diet; KD–T, ketogenic diet minus Thr; PTZ, pentelenetetrazole.

Trial 4B: Kindling

Upon testing (on day 65), the effects of the lack of Thr in the KD–T showed the expected food intake depression (2.4 ± 0.2 g for KD–T compared with 3.0 ± 0.2 g for CD_COR, P < 0.05). Only rats that had eaten ≥2 g of the KD–T diet were included in the analysis. Seizures were significantly increased in the KD–T group (P < 0.05; Wilcoxon rank test, Figure 4C) compared with their KD_Bas baseline (kindled) measurement. The kindled CD_COR group did not differ from their own baseline (P > 0.05; Figure 4C). Maximal seizure scores for the kindled KD–T group, as a ratio to that rat's baseline seizure score, were significantly greater than for the CD_COR group (P = 0.04, Figure 4D). In a correlation between food intake and total seizure score, based on the slopes of total seizure number compared with food intake (grams), the positive slope (+4.9) for KD–T differed by 15.5 units from a negative CD_COR slope (–10.6), suggesting that eating more of the higher-protein equivalent CD_COR diet may be protective against seizures. Ingesting a KD for 20 d may have blunted the effects of seizures 1.5 h after the animals ate their first meal of KD–T. Nonetheless, after eating a single meal of the devoid diet, there were significantly more severe seizures in the kindled KD–T-fed rats than in KD-fed kindled rats, as noted above.

Trial 5: Gene expression

Of the 1300 genes on the Rat Neuro U34 microchip (Affymetrix), 40 transcripts differed between CD_Bas and ThrDev group 2 h after a single meal (P = 0.02–0.002, Figure 5A, B). The changes at 2 h were predominantly negative, showing loss of these transcripts shortly after ingestion of the ThrDev diet, consistent with previous results (18). In contrast, by day 7, 34 transcripts were significantly upregulated; only 6 were downregulated (Figure 5C, D and Table 2). Genes coding for proteins involved in new synapse formation (structural), ion channels, transporters, signalling, and receptors were significantly altered after 7 d on a ThrDev diet; Gene Ontology groupings for these categories showed significance levels between P = 0.001 and P = 0.045 (Figure 5). Details for differences between CD_Bas and ThrDev diet on day 7 (or between CD_Bas and the ThrDev diet on day 1 at 2 h; see footnotes to Table 2) are given in Table 2.

FIGURE 5.

Effects of ThrDev diet on gene expression in the APC, Trial 5. The APCs of male rats, previously fed CD_Bas for ≥7 d, were collected after 2 h (A, B) or 7 d (C, D) eating ThrDev diet or CD_Bas. Changes with >2-fold differences between ThrDev diet and CD_Bas groups are shown in the GO categories indicated below each horizontal axis (categories: *P < 0.05, **P < 0.01, and ***P < 0.001). Values are fold changes from CD_Bas diet group (n = 7/group). Pie charts show relative abundance of transcript changes for 2 h (B) or 7 d (D); see also Table 2. APC, anterior piriform complex; CD_Bas, basal control diet; GO, Gene Ontology; ThrDev, Thr-devoid.

TABLE 2.

Gene transcripts in male rat APC significantly altered after 7-d access to CD_Bas or ThrDev diet, Trial 5¹

Accession ID	Fold difference	UniGene name	Gene symbol
Receptors, P = 0.009
AI170268	2	β-2 microglobulin (MHC receptor)	B2m
L08493	2.1	GABA-A receptor α-4 subunit	Gaba4
AB0161612	4	GABA-B R1	Gabbr1
AI2304042	5.1	Guanine nucleotide binding protein, β 1	Gnb1
U883242	2.5	Guanine nucleotide binding protein, β 1	Gnb1
AI228113	2.2	Neuronal pentraxin receptor	Nptxr
Net fold change	17.9
Transporters, P = 0.001
AF106563	2.3	ATP-binding cassette transporter	Abcb6
S68135	2.5	Glucose transporter 1	Glut1
AI146214	2.2	Vesicular monoamine transporter 2	Vmat2
S56141	2.8	Na-dependent neurotransmitter transporter	Slc6a15
Net fold change	9.8
Ion channels, P = 0.045
AI230614	2	ATPase Na+/K+ transporting β 1 polypeptide	Atp1b1
AF051527	2.3	Ca channel, V-dependent, α 1A	Cacna1a
M86621	4.4	Ca channel, voltage-dependent, α2/δ1	Cacna2d1
X628401	2.9	K V-gated channel, Shaw-related	Kcnc1
M84210	2.9	K V-gated channel, Shaw-related 3	Kcnc3
X83580	–4.7	K inwardly-rectifying channel, J4	Kcnj4
U926551	2.2	K V-gated channel, KQT-1	Kcnq1
M24852	2	Purkinje cell protein 4 (Ca2+ binding)	Pcp4
M222533	2.3	Na V-gated channel, type 1α	Scn1a
Net fold change	16.3
Structural changes, P = 0.017
M64780	2.2	Agrin	Agrn
M57664	–2.7	Creatine kinase	Ckb
AF0303582,3	3.3	Chemokine (C-X3-C motif) ligand 1	Cx3cl1
AI175948	2.3	Fibroblast growth factor receptor 4	Fgfr4
D00913	2.9	Intercellular adhesion molecule 1	Iam1
X065541	–2.2	Myelin-associated glycoprotein	Mag
AI2276081	–5.6	Microtubule-associated protein τ	Mapt
AI101255	5.1	Macrophage differentiation	Mmd
U30938	2.6	Microtubule-associated protein 2	Mtap2
H33836	2.1	Neural cell adhesion molecule 1	Ncam1
L14851	2.8	Neurexin 3	Nrxn3
AF000423	–3.8	Synaptotagmin 11	Syt11
AA8600302	6.5	β Tubulin	Tubb5
AA9435322	5.6	β Tubulin	Tubb5
Net fold change	21.1
Signaling, P = 0.009
AI013765	2.4	Arrestin, β 2, pseudogene	Arrb2-ps
E02315	2	Calmodulin	Cam
M161122	2.3	CaMKIIb	Camk2b
AI237836	3.2	Gnas complex GPCR αsubunit s	Gnas
AI2276602	–7.8	Guanine nucleotide-binding protein β polypeptide	Gnb1
E135412	2.6	Neuroglycan C	Ngc
AF044910	2.7	Survival motor neuron	Smn
Net fold change	7.4

¹Affected similarly after 2 h on ThrDev diet. APC, anterior piriform cortex; CD_Bas, basal diet; GABA, γ-aminobutyric acid; GPCR, G-protein-coupled receptor; ID, identifier; MHC, major histocompatibility complex; ThrDev, Thr-devoid; V, voltage.

²Significantly altered after 2 h on ThrDev diet, and significant difference noted between 2 h and 7 d.

³Also found in “eGENE” lists related to epilepsy.

Trial 6: Microdialysis in the APC

The data for glutamate release over time (Figure 6) showed best fit to a polynomial curve in the ThrDev group: y = 0.0002x³ – 0625x² + 43.6, R² = 0.70 (Excel Trendline function, Microsoft). The curve for the CD_COR group was described by a linear equation: y = 0.0079x + 40.729, R² = 0.53. Visual inspection of the graph for glutamate showed that levels of this excitatory neurotransmitter in the ThrDev rats were clearly elevated at 20 and 60 min relative to CD_COR, confirmed by P < 0.05 using the t test for planned comparisons; glutamate concentrations for both groups diminished to baseline concentrations thereafter as expected after a ThrDev meal (18). Excitatory neurotransmitter release into the APC during and shortly after a ThrDev meal is consistent with, and supports, the above studies showing potentiation of the APC circuitry. Taken together, increased glutamate and limited GABAergic function (18) likely contribute to the seizure susceptibility induced by the ThrDev diet.

FIGURE 6.

Effect of ThrDev diet on excitatory neurotransmitter (Glu) release in male rat APC after 7 d (Trial 6). Values are Glu concentrations in samples from in vivo microdialysates of APC (percentages of baseline taken just prior to onset of feeding at time t = 0). Dashed line is polynomial trendline (Excel) for ThrDev diet data. Details and equations for the lines are given in the text. Values are means ± SEs (n = 3/group). APC, anterior piriform complex; CD_COR, corrected control diet, fully replete; Glu, glutamate; ThrDev, Thr-devoid.

Discussion

Here we have confirmed and extended our previous report that IAA deficiency increases seizure susceptibility (8). We provide details as to the timing and insight into the mechanisms for the role of IAAs in seizure susceptibility, and the role for IAA balance in successful KD treatment of seizures.

Neurochemical changes

Increased seizure susceptibility due to Thr deficiency in the short term likely results from the loss of GABA-related inhibitory elements (18) due to the inhibition of protein synthesis that prevents their replacement. These changes start to occur early in Thr deficiency (18). In kindled rats or seizure-prone gerbils, we saw that a single IAA-devoid meal, importantly including the ketogenic KD–T, can cause, or worsen, seizures, as in the results for kindled rats after a single meal of KD–T in Trial 4B, and the seizure-prone gerbils in Trial 3. In patients who are clearly seizure prone such a meal could also be harmful. This is consistent with the concept that increased seizure susceptibility, resulting from IAA imbalance, is likely related to the reduced inhibitory control in the APC, i.e., loss of proteins associated with Cl^– transport and GABAergic function in <2 h as we reported previously (18), and the inability to replace those proteins while protein synthesis is blocked by the GCN2 system (20).

Along with GABA as the inhibitory system, glutamate is the primary excitatory neurotransmitter in the brain. With the reduction of GABAergic inhibition in the APC (18), and the increased release of glutamate (Trial 6), the recurrent excitatory positive-feedback loops would have less inhibitory balance to maintain control within the circuitry of the APC, as clearly detailed by Doherty et al. (40).

Moreover, the increased release of glutamate (Figure 6) activates its receptors, resulting in Ca²⁺ influx into neurons via both N-methyl-d-aspartate receptors (41) and voltage-gated Ca²⁺ channels. This is consistent with an influx of Ca²⁺ into neurons after a single meal of ThrDev diet, as we have seen, resulting in phosphorylated CaMKII, involved in activating calcineurin (34, 42). These data are consistent with Ca²⁺ signaling in the mechanisms of epileptogenesis reviewed by McNamara et al. (43).

Structural changes

After 7 d on ThrDev diet, the structural changes shown in the transcriptional results of Trial 5 would promote increased excitability in otherwise normal rats. In the nonseizure-prone animals, dietary deficiency must be maintained for several days before the changes in the structures of the APC neural circuitry are sufficient to enhance the seizures produced by PTZ challenge. These changes in structural and functional genes are consistent with increased excitability in the neural circuitry of the APC, and these changes also correspond to important genes associated with epilepsy in human and rat databases. The combined effect of increased transcription of genes involved in synaptogenesis, ion channels, and receptors suggests that new synapses are likely being formed, existing synapses are being strengthened, and that these remodeled neurons will be excited more easily.

Dietary issues

Paired-feeding trials were done in rats previously (8), and in gerbils here (Trial 3). No differences due to reduced food intake in the pair-fed controls were noted in the seizure results. Thus, the increased seizures are not due to the well-known food intake depression seen in animals offered IAA-deficient or imbalanced diets (25). This should be expected, because starvation decreases seizure activity, while IAA-deficient diets increase seizure levels and duration, as we have seen before and in the present studies.

The literature on dietary treatments for epilepsy is replete with reports, with >600 cited up to 2013 (44) and 300 more, including 90 reviews, between 2014 and 2017 (PubMed update). In general, clinical studies report a benefit from KDs and MADs. Indeed, a recommendation to begin dietary treatment with a MAD has been proposed (45). Nevertheless, the profile of IAA is rarely, if ever, included in reports concerning these diets, although many common foods are deficient in ≥1 of the IAAs (e.g., a gelatin snack sweetened with aspartame, as reportedly recommended in the KD, would be high in phenylalanine and very low in tryptophan and thus severely imbalanced). Our results, which show increased seizure activity after a single ThrDev meal in seizure-susceptible gerbils and kindled rats, support the idea that such a snack could be harmful, particularly in refractory patients on the KD protocol.

In conclusion, seizure disorders afflict some 50 million people worldwide, according to the WHO (http://www.who.int/mediacentre/factsheets, February 2017). These disorders vary in onset, severity, and etiology, but have one common feature: recurrent synchronized discharges in neural circuits. Human epilepsies often have a genetic component, but for many patients with seizure disorders, other causes must be sought, such as metabolic or biochemical aberrations (43). Those with a genetic background as their chief etiology, exemplified here by our studies using the Mongolian gerbil, along with our kindled rats, a model for injury-induced epilepsy, may also be more sensitive to IAA disproportions than others. IAA deficiency (as in the ThrDev diet) provides another model for the study of seizures. This model contributes to understanding the long-term neural consequences of IAA deficiencies, such as in the protein deficiency disease kwashiorkor or protein-calorie malnutrition, in which both protein and calories are deficient. Since the 1950s, the incidence of seizure disorders has been reported to be markedly higher in Third World countries, where much of the population is at risk for protein-calorie malnutrition (46). These findings may be helpful not only to victims of seizures generalized from limbic foci but also for understanding and treating other disorders associated with seizures, especially if IAAs in the usual diets are imbalanced.

We suggest that the results of these studies support increased attention to the protein fraction of diets for seizure patients. This would include the use of balanced proteins in KDs, and evaluating IAA concentrations in such patients, with adjustments as indicated for diets in the protocols for seizure patients. In addition, we would like to see the inclusion of IAA profiles in publications on diets for use in epileptic patients.

Notes

Supported by NIH grants DK50347, NS37490, and NS043210 (to DWG) and DK35747 (to the CNRU, UC Davis).

Abbreviations used: AA, amino acid; APC, anterior piriform cortex; APCvr, ventrorostral APC; CD_Bas, basal control diet with Thr; CD_Cas10, control diet with 10% casein; CD_Cas14, control diet with 14% casein; CD_COR, control diet replete with all indispensable amino acids; GCN2, general amino acid control kinase 2; HE, hindlimb extension; IAA, indispensable amino acid; KCC2, potassium chloride co-transporter; KD, ketogenic diet; KD_Bas, basal ketogenic diet; KD_Cas, ketogenic diet with 10% casein; KD–T, ketogenic diet minus Thr; MAD, modified Atkins diet; MES, maximal electroshock seizure test; P-eIF2α, phosphorylated eukaryotic initiation factor 2 α; PTZ, pentelenetetrazole; ThrDev, threonine devoid.