ketogenic diet (KD) is an established nonpharmacologic treatment for patients
with intractable epilepsy (1,2) and was designed to
mimic the biochemical changes seen upon fasting, specifically the formation of
ketone bodies: β-hydroxybutyrate (BHB), acetoacetate (ACA), and to a lesser
extent, acetone. Despite decades of clinical experience with the KD, the
mechanisms underlying its anticonvulsant activity remain poorly understood.
have observed that ketosis is necessary but not always sufficient for seizure
control with the KD. It is well known that seizure control gradually increases
within the first few weeks of initiating a KD, as serum ketone levels steadily
increase, but then is abruptly lost when ketosis is broken, usually through
ingestion of carbohydrate (3). In addition, seizure
control appears to correlate with blood BHB levels (4), although this relation is
not consistent (5). Taken together, these
observations suggest a direct role for ketone bodies in limiting seizure
investigate whether ketone bodies are directly anticonvulsant, we tested the
effects of ACA, acetone, and both stereoisomers, d-(−)- and l-(+), of BHB on
sensory-evoked seizures in Frings audiogenic seizure-susceptible mice. We found
that ACA, acetone, and the l-(+)-isomer of BHB had anticonvulsant activity in the
however, subsequent experiments established that the activity of L-(+)-BHB was
due to dibenzylamine (DBA), a chemical contaminant.
Frings audiogenic seizure-susceptible mice were
obtained from an in-house colony maintained by the Anticonvulsant Screening
Program at the University of Utah. Before testing, all mice were maintained on
a 12-h light/dark cycle, and food and water were available ad libitum according
to the guidelines outlined in the NIH Guide for Care and Use of Laboratory
Animals. All animal procedures were approved by the Institutional Animal Care
and Use Committee at the University of Utah.
The Frings audiogenic seizure-susceptible
mouse is a
model of sensory-evoked reflex seizures (6). The seizure phenotype is characterized by a sequence of wild
running, loss of righting reflex, tonic flexion, and ultimately tonic extension
in response to a high-intensity sound stimulus. One-month-old female Frings
mice were injected intraperitoneally (i.p.) with varying doses (1–30 mmol/kg)
of ACA, acetone, l-(+)-BHB, d-(−)-BHB,
or the racemate, dl-BHB. Fifteen minutes after
i.p. administration of the test compound, individual mice were placed into a
round, plexiglas sound chamber and exposed to 110-dB, 11-KHz sound for a total
of 20 s or until a full tonic extension seizure was elicited (6).
were prepared in the following manner. Stock solutions (1 mg/ml) of DBA,
D-(−)-BHB, L-(+)-BHB, and the racemate, DL-BHB, were prepared by using methanol
(Fisher Scientific, Pittsburgh, PA, U.S.A.) as the solvent. These samples were
further diluted to 1 μg/ml and stored overnight at 2–8°C. The following day,
0.1 ml of each of these solutions was transferred to an autosampler vial
equipped with a conical low-volume insert, and evaporated under N2 gas. Each sample was
re-dissolved in 50 μl of ethyl acetate (which had been dried with magnesium
sulfate) and then reevaporated. They were again dissolved in 50 μl of ethyl
acetate to which 50 μl of a derivatizing agent, perfluoroacetic anhydride, was
added. Vials were then capped and heated to 70°C for 60 min. Gas
chromatography–mass spectrometry (GC-MS) analysis was performed with a Hewlett
Packard 5890 Series II Gas Chromatograph (equipped with an electron-capture
detector, model 7673 Injector, and a HP 3396 Series II Integrator) as
previously described (7).
DBA (C14H15N; MW, 197.28), ethyl
acetate, and perfluoroacetic anhydride were purchased through Aldrich
(Milwaukee, WI, U.S.A.), Fisher Scientific (Pittsburgh, PA, U.S.A.), and Fluka
(Ronkonkoma, NY, U.S.A.), respectively. Differing lots of ACA, acetone, d-(−)-BHB, l-(+)-BHB, and dl-BHB, were obtained from
both Sigma Chemical Co. (St. Louis, MO, U.S.A.) and Aldrich.
mmol/kg i.p. injection of ACA, eight of eight mice were protected against
sound-induced tonic extension. Acetone also was protective in eight of eight
mice administered 10 mmol/kg acetone. With 30 mmol/kg l-(+)-BHB,
seven of eight mice injected were protected against sound-induced tonic
extension, whereas none of eight mice given a similar dose of the d-(−)-isomer
were protected 15 min after injection. Summary data obtained with ACA, acetone,
are presented in Table 1.
Table 1. Summary
anticonvulsant efficacy data in Frings audiogenic seizure-susceptible mice
At a 30
mmol/kg dose, both stereoisomers of
BHB produced signs of acute toxicity characterized by mild gait abnormality,
lack of exploratory activity coupled with blank stares, and watery stools. All
animals recovered fully from sound-induced seizures and BHB toxicity within 2
h. Conversely, ACA and acetone doses of <20 mmol/kg were well tolerated.
did not protect Frings mice against sound-induced seizures, raising concerns
about the effects seen with l-(+)-BHB.
Additionally, the clinical
relevance of l-(+)-BHB
effects was raised because it is the d-(−)-isomer that is
biologically produced (8). Doepner
et al. (9) reported
block of transient outward K+ current in
myocardial mouse cells was due to DBA, which they identified with UV spectroscopy; however, no direct
demonstration of this was provided. Therefore, we used GC-MS to determine
whether commercial lots of the l-(+)-isomer of BHB in fact
In the GC-MS scan mode, the derivitized DBA resulted in
at 7.19 min with M/Z (mass/charge) ions of 91.10 and 202.10 as the most
abundant ions. As shown in Fig.
the select ion monitoring (SIM) at M/Z 202.10, the peak at 7.29 min was not
present in either the dl-racemate or the d-(−)-BHB standards. However,
there was evidence of DBA in the l-(+)-BHB. Using standard
solutions of DBA, we estimated that there was not >0.01% DBA contamination
of the d-(−)-BHB
compounds, but that DBA composed ∼0.4% of the l-(+)-BHB. Thus
1–30 mmol of l-(+)-BHB
would contain roughly 40 μmol to 1.2 mmol of DBA. Dibenzylamine also was
identified in l-(+)-BHB
from different lots obtained through both Aldrich and Sigma.
these findings, we next examined the effects of DBA in Frings
audiogenic seizure-susceptible mice. Fifteen minutes after administering 200
μmol/kg DBA i.p., complete protection against sound-induced tonic extension was
observed in four of four mice. Over a concentration range of 50–200 μmol/kg
(four to 11 mice per dose), DBA produced a dose-dependent block of audiogenic
seizures (Table 1).
The principal finding of this study is that the ketones ACA and acetone,
but not BHB, exhibited anticonvulsant efficacy in Frings audiogenic
seizure-susceptible mice. Our data confirm Keith's observation (10) that
ACA blocked seizures induced by thujone, a convulsant constituent found in many
essential oils and an antagonist of γ-aminobutyric acid (GABA)A receptors (11). We also
demonstrated that the active component in l-(+)-BHB is likely to be DBA for the following
reasons: (a) the racemate, dl-BHB, does not contain DBA and was inactive in the
Frings model; (b) although we cannot exclude a pharmacodynamic interaction
between the d- and l-enantiomers, d-(−)-BHB
was not effective in blocking audiogenic seizures; and (c) Doepner et al. (9) reported
a similar experience with these ketones in electrophysiologic experiments conducted
in mouse myocardial cells.
Frings audiogenic seizure-susceptible mice were chosen to test for the
potential anticonvulsant properties of ketones because of the promising role
these mice have played in antiepileptic drug (AED) screening (6). In
addition, protection against these audiogenic seizures is believed to be
nondiscriminatory with respect to the differing clinical efficacies of known
AEDs (i.e., agents effective against both partial and generalized seizures are
all active against sound-induced seizures in Frings mice) (6).
Given the strong (but not universal) correlation between blood ketone
levels and seizure control with the KD (4), it is
important to determine if ketones can directly modulate neuronal excitability
and synchronization. In cellular electrophysiologic experiments, BHB and ACA
did not directly alter synaptic or inhibitory synaptic transmission in the
hippocampus (12) and
did not directly interact with GABAA,
ionotropic glutamate receptors, or voltage-gated sodium channels in cultured
neocortical neurons (Rho et al., unpublished data).
The less prevalent ketone bodies, ACA and acetone, clearly possess
anticonvulsant properties, but the precise mechanisms underlying these actions
are unclear. One possibility is that cerebral acetone may contribute to the
anticonvulsant actions of the KD, possibly through nonspecific general
anesthetic actions or through uncoupling of gap junctions (P. Carlen, personal
communication, 2001). Recently, it was observed that patients successfully
treated with a KD exhibit elevated levels of acetone in the brain (13).
Acetoacetate can enter the brain via monocarboxylic transporters (MCTs)
and can then be spontaneously decarboxylated to acetone. Further, brain
accumulation of acetone from the periphery may be enhanced because of the high
lipid solubility of this species. Elevated brain levels of BHB such as that
seen in a fasting-induced ketotic state (14) also
may indirectly contribute to an anticonvulsant effect, because BHB can be
subsequently converted to ACA by β-hydroxybutyrate dehydrogenase.
Interestingly, a ketogenic diet has recently been shown to increase MCT levels
in rat brain (15).
Our experience with DBA emphasizes once again the pitfalls of ascribing
specific biologic activity to chemical preparations containing active trace
contaminants (9). This
cautionary note may be relevant because many published studies addressing the
effects of BHB in a variety of in vivo and in vitro models do not indicate
which enantiomer was used. Nevertheless, although DBA possesses anticonvulsant
properties (and might warrant further preclinical development), DBA also has
been identified as a potent irritant and is listed in the Environmental
Protection Agency inventory under the Toxic Substances Control Act (TSCA).
Therefore, its future as a clinically useful anticonvulsant seems doubtful.
An additional issue merits comment: the potential biologic relevance
It is generally accepted that d-(−)-BHB, the stereoselective product of d-3-hydroxybutyrate
dehydrogenase, is the clinically significant (i.e., physiologic)
stereoisomer (8). l-(+)-BHB
has not yet been directly assayed in either animals or humans, yet there is
evidence that this isomer could be biologically produced, either in liver
and/or brain (16–18).
Whether l-(+)-BHB is
relevant to the KD is unknown.
In summary, we demonstrated that the ketone bodies ACA and acetone are
directly anticonvulsant in vivo, whereas the more prominent species, BHB, is
not. Interestingly, these metabolic products of BHB appear more potent as AEDs
than BHB itself (Table 1). Given
these findings, we propose that the anticonvulsant efficacy of the KD may be
due in part to the accumulation of acetone in the brain, a hypothesis supported
by previous published data (13). How
acetone exerts such an action remains unclear.
Acknowledgment: We thank the following
individuals: Eric Kantor for technical assistance and William Howald of the
Mass Spectrometry Facility (University of Washington School of Pharmacy). This
work was supported by NIH grant K08 NS01974 (J.M.R.) and the Anticonvulsant Drug
Development Program, The University of Utah (H.S.W.).