Impact of dietary ketosis on volatile anesthesia toxicity in a model of Leigh syndrome

Genetic mitochondrial diseases impact over 1 in 4000 individuals, most often presenting in infancy or early childhood. Seizures are major clinical sequelae in some mitochondrial diseases including Leigh syndrome, the most common pediatric presentation of mitochondrial disease. Dietary ketosis has been used to manage seizures in mitochondrial disease patients. Mitochondrial disease patients often require surgical interventions, leading to anesthetic exposures. Anesthetics have been shown to be toxic in the setting of mitochondrial disease, but the impact of a ketogenic diet on anesthetic toxicities in this setting has not been studied.


| INTRODUC TI ON
Genetic mitochondrial diseases are estimated to impact over 1 in 4000 individuals. 1,2Mitochondrial diseases are a large, clinically heterogeneous, group of disorders linked by the fact that their underlying genetic causes impact gene products involved in normal mitochondrial function.At present, over 400 unique genetic alterations have been linked to clinical genetic mitochondrial disease. 3tochondrial diseases most often present in infancy or early childhood.Together, they account for the most prevalent cause of heritable metabolic diseases and one of the most frequent causes of heritable neurologic diseases.
Striking hypersensitivity to volatile anesthetics (VAs) (compounds including isoflurane, sevoflurane, halothane, and desflurane) is a well-documented feature of certain forms of mitochondrial disease, in particular to patients with defects in mitochondrial electron transport chain Complex I (ETC CI). 4,5[8] Leigh syndrome (LS) is a severe manifestation of mitochondrial dysfunction and the most common pediatric presentation of genetic mitochondrial disease.Over 110 distinct genes have been identified as causes of LS, including multiple subunits of the ETC.
The Ndufs4(−/−) (homozygous loss of function, knockout) mouse is a well-established animal model of this pediatric disease; the gene encodes the Ndufs4 subunit of mitochondrial ETC CI.Loss of this subunit is a cause of LS in humans and results in LS symptoms in mice.Progressive, symmetric, neuroinflammatory lesions in specific brain regions, including the brainstem and cerebellum are defining clinical characteristics of LS in both humans and the Ndufs4(−/−) model. 9Seizures are also an important clinical feature of LS; up to 60% of LS patients may suffer from epileptic seizures, which significantly impact quality of life and can be extremely difficult to manage. 9,10Ndufs4(−/−) animals are also extremely hypersensitive to anesthesia with VAs (MAC ~1/3 that of wild type animals).Ndufs4(−/−) mice are born free from any overt disease, which is also typical of LS patients.In LS, symptoms typically appear by ~2 years of age, while disease onset occurs around postnatal day 37 in the mouse model.
We have recently shown that exposure to VAs is acutely toxic in Ndufs4(−/−) mice at ages after the onset of disease, but not earlier in life, and that toxicity is linked to the presence of the neuroinflammatory CNS lesions characteristic of LS. 11,12 This toxicity is distinct from anesthetic hypersensitivity, as MAC changes little with disease onset.Toxic sequelae in these animals include hyperlactatemia and hyperglycemia, weight loss, and, at higher concentrations or in longer exposures, acute mortality.
A ketogenic diet has been used clinically in mitochondrial disease patients both to manage seizures and in experimental attempts to intervene in other symptoms.Available evidence seems to support the notion that dietary ketosis is beneficial for mitochondrial disease related epilepsies, with mitochondrial ETC CI patients appearing to show the greatest benefits. 13,14Dietary ketosis also appears to reduce seizure incidence in the Ndufs4(−/−) mouse model of Leigh syndrome. 15However, studies to date do not seem to support the notion that overall disease progression is altered by a ketogenic diet in either the Ndufs4(−/−) model or patients, suggesting that benefits may be limited to seizure control. 14,15ile a ketogenic diet is being used for clinical care of mitochondrial disease patients, potential interactions between VAs and a ketogenic diet in the setting of genetic mitochondrial disease have not been tested.Given that mitochondrial disease patients can require a range of surgical interventions from muscle biopsy and dental care to epilepsy surgery, scoliosis repair, and heart valve repair, 5,[16][17][18][19][20][21] probing interactions between clinical interventions and anesthesia is considered prudent.
Here, we assess the impact of dietary ketosis on anesthesia tolerance in the Ndufs4(−/−) mouse model of the pediatric mitochondrial disease LS.We have previously established that complications of VA exposure are strongly linked to the CNS lesions associated with disease progression.Given that the presence of CNS lesions is a defining clinical feature of LS, we focus here on exploring the interactions between VA exposure and ketogenic diet at an age where lesions are present.To accomplish this, we exposed animals raised on a normal or ketogenic diet to brief anesthesia with isoflurane at the postnatal age of P50.This correlates with ages after LS symptom onset in children.All experiments contain approximately equal numbers of male and female mice of each genotype.As previously reported, the Ndufs4 deletion is recessive, and heterozygosity results in no reported phenotypes and no detectable defects in ETC CI activity.Accordingly, "control" cohorts include both Ndufs4(+/−) and Ndufs4(+/+) mice.Twenty-six Ndufs4(−/−) and 21 Ndufs4(control) mice were assigned to ketogenic and control diets for these studies, totaling 94 experimental animals.Replicate numbers are detailed in each figure, and all raw data provided as a supplemental file.Ndufs4(−/−) animals were housed with control littermates for warmth and stimulation in all studies.Mice were weighed and health assessed a minimum of three times per week.Control chow was wetted in cages housing Ndufs4(−/−) mice displaying neurological symptoms (ketogenic diet is already soft).Humane euthanasia criteria included 20% loss of body weight from maximum or the acute presentation of severe motility or neurologic symptoms perceived to impair access to food or water (immobility, prostrate posture, or otherwise moribund in appearance).

| Animal diets
Breeders and experimental mice were fed PicoLab Lab Diets 5053 and 5058, respectively.Mice fed a ketogenic diet (Envigo, Teklad TD.96335) were gradually acclimated starting at weaning (P21) using the following protocol (as in our previous work 15 ): 3 days (starting at weaning, postnatal day 21) on a 50/50 mix (by weight) of ketogenic diet and ground normal mouse diet (PicoLab Diet 5058), followed by 3 days of 75/25 ketogenic/normal, then finally to 85/15 ketogenic/ normal for the duration of the study.

| Anesthesia
Isoflurane (Patterson Veterinary, 14 043 070 406) was provided at concentrations indicated using a routinely calibrated isoflurane vaporizer (Summit Anesthesia Solutions) at a flow rate of 1.5-2 L/ min with an in-line humidifier (see Figure 1).Isoflurane concentration was monitored using an in-line volatile anesthetic sensor (BC Biomedical AA-8000 analyzer).100% O 2 or medical air were used as carrier gas as specified in individual experiments.The plexiglass exposure chamber and humidifier were pre-warmed to and held at 38°C throughout exposures using a circulating water heating pad (Adroit Medical, HTP-1500).Mice were fed ad-lib prior to and after exposures.
As detailed in our prior studies, 8,12 0.4% isoflurane was used to anesthetize Ndufs4(−/−) mice, while 1.25% isoflurane, an equipotent concentration, was used to anesthetize control animals.Note: (−)keto (−)iso data appear in, 12 as the experiments were performed concurrently to reduce overall animal use.

| Postanesthesia monitoring
Following anesthesia, mice were placed into a clean cage sitting on a water heating pad and monitored for seizures and emergence from anesthesia.Observation was concluded when mice were deemed alert by the researcher or 20 min had elapsed, whichever was longer for a given mouse.
Loss of righting reflex (LORR) is a well-established method for measuring anesthetic state in mice and is thought to be equivalent to loss of consciousness.Righting reflex measures allow for assessment of animal anesthetic depth without disruption of gas concentration or flow through the plexiglass anesthesia chamber.
The isoflurane concentrations we used in these studies (0.4% for Ndufs4(−/−) and 1.25% for control animals) are above MAC; moreover, LORR occurs at lower concentrations than loss of response to tail clamp or pedal withdrawal, so any toxicities observed at concentrations causing LORR are relevant to higher anesthetic concentrations (see 27,28 ).
To assess LORR, animals were tilted on their side, with righting reflex considered intact if the animal regained a position where all four paws face the ground within 10 s.
Respiratory rate was assessed by counting breaths during a 15-30 s interval.
Peripheral blood oxygen saturation, SpO 2 , and heart rate were monitored by pulse-oximetry using a Kent Scientific MouseSTAT Pulse Oximeter and Heart Rate module attached to a PhysioSuite monitor.These monitors utilize paw-pad pulse-oximeters.Values are impacted by movement and tend to be variable in alert/unanesthetized animals.Accordingly, high variance in unanesthetized data and early in anesthetic exposures (when animals are not yet fully anesthetized) is expected.Nevertheless, we saw no notable changes in heart rate or SpO 2 during the three exposures, consistent with our prior findings using this paradigm in animals fed a control diet. 12See our Data S1 tab SpO 2 and heart rate for data.

| Blood point-of-care data
Blood metabolites (glucose, βHB, and lactate) were collected using point-of-care meters and the minimally invasive tail-prick method.
We have previously validated the accuracy of these point-of-care meters (see 15 ).

| Experimental design and statistical analyses
Controls were spread chronologically throughout the experiments.
Animals were randomly assigned to treatment groups.All exposures were performed at approximately the same time of day (within a 2 h window) to avoid variance that differences in circadian cycle might introduce.
All statistical analyses were performed using GraphPad Prism as detailed in figure legends.Where parametric tests were employed, normality was first assessed using the Shapiro-Wilk test.
For nonparametric LORR data comparisons were made using the nonparametric Kruskal-Wallis one-way ANOVA test with Dunn's multiple comparisons test for all pairwise comparisons.
Parametric tests utilized unpaired two-tailed t-test (for pairwise comparisons) and ordinary one-way ANOVA with Tukey's multiple testing corrected pairwise comparisons (for multiple comparisons).Details regarding statistical tests and experimental design are provided in figure legends.In addition, all raw data and statistical analyses are provided in the supplemental data file.[31][32]

| Assessing the interaction of ketogenic diet and VA exposure in mitochondrial disease
To characterize any interactions between VA exposure and a ketogenic diet in the setting of mitochondrial disease, Ndufs4(−/−) and control animals were raised on a ketogenic diet from weaning.As in our prior work, the diet composition was altered gradually during the early postweaning period (Figure 1A, see Section 2).Consistent with our prior findings, 12 this regimen led to dietary ketosis in the Ndufs4(−/−) model: in ad-libitum fed mice at postnatal day 50 (P50), blood beta-hydroxybutyrate was significantly elevated in animals fed the ketogenic diet compared to those control chow, while average blood glucose concentration was significantly reduced (Figure 1B,C).Blood lactate levels were not altered by dietary ketosis (Figure 1D).
On the ketogenic diet regimen, control animals show a milder ketosis compared to Ndufs4(−/−) mice and a slight increase in blood glucose at baseline 12,15 (Figure 1E-G).P50 animals were anesthetized with isoflurane for 30-min.
Isoflurane was provided in 100% oxygen, currently considered standard of care for anesthesia in murine animal research, with the combined anesthetic gas passing through a warmed humidifier prior to entry into the plexiglass anesthetic chamber (Figure 2A, see Section 2).Ndufs4(−/−) and control mice were anesthetized with 0.4% and 1.25% isoflurane, respectively, which are approximately equipotent for providing anesthesia in Ndufs4(−/−) and control mice (see, 12 Figure 2B,D, Section 2).Ketogenic diet did not alter the loss of righting in control or Ndufs4(−/−) animals at these concentrations, which are above MAC 8,12 and sufficient for anesthesia.As previously reported, 12 Ndufs4(−/−) animals show significant respiratory rate depression by 30-min of exposure to 0.4% isoflurane, compared to the small (not significant in our small cohort) effects of 30-min exposure to 1.25% isoflurane in controls (Figure 2C,E).Dietary ketosis had no impact on respiratory rate in Ndufs4(−/−) or control animals.
Blood glucose levels were significantly increased following a 30min exposure to 0.4% isoflurane in both the control fed and ketogenic diet fed Ndufs4(−/−) cohorts, though the effect was partly (but nonsignificantly) attenuated in the ketogenic diet group compared to the control diet group (Figure 3E,F).Control mouse blood glucose levels were not elevated under any of the tested conditions (Figure 3G,H).
Increased blood lactate can occur in genetic mitochondrial disease and is an important clinical sequela.We have previously shown that anesthesia with 0.4% isoflurane leads to a significant raise in blood lactate in Ndufs4(−/−) mice. 11,12Consistent with these prior findings, 30-min exposures to 0.4% isoflurane led to significant increases in blood lactate in Ndufs4(−/−) mice, regardless of diet (Figure 3I,J).This increase was significantly higher in Ndufs4(−/−) mice fed a ketogenic diet (Figure 3J), despite the ketogenic diet not impacting baseline blood lactate (see Figure 2).1.25% isoflurane increased blood lactate in control mice, and the effect was greater in ketogenic diet fed versus control fed animals (Figure 3K,L).
However, the lactate in controls was substantially lower compared to Ndufs4(−/−) animals regardless of diet; in ketogenic diet treated animals, blood lactate at the end of isoflurane anesthesia was around 5 mM in control mice exposed to 1.25% isoflurane versus 10 mM in Ndufs4(−/−) mice exposed to 0.4% isoflurane (Figure 3I-L).
Two-way ANOVA suggests the higher increase in blood lactate in the ketogenic diet fed anesthetized group results from a sum of effects from diet and anesthetic exposure, rather than an interaction between the two (see Section 4).

| Dietary ketosis worsens anesthetic outcomes in the Ndufs4(−/−) mouse model of LS
We have previously shown that exposure to 0.4% isoflurane leads to acute and chronic toxicities and accelerates mortality in the Ndufs4(−/−) mouse model of LS. 12 To determine whether dietary ketosis alters anesthesia outcomes, we exposed animals on each diet to 30-min of 0.4% isoflurane anesthesia once daily for three consecutive days.Exposure to 0.4% isoflurane led to significant weight loss compared to control (oxygen) exposed Ndufs4(−/−) animals, as previously reported, while animals on a ketogenic diet displayed slightly (but not significantly) increased weight loss in this paradigm (Figure 4A).Survival of Ndufs4(−/−) animals fed a ketogenic diet was congruent with prior survival data for ketogenic diet treated Ndufs4(−/−) mice up until P50. 15At P50, Ndufs4(−/−) animals were randomly assigned to (−) isoflurane or 0.4% isoflurane exposure groups (Figure 4B).Survival curves of non-ketogenic and ketogenic diet fed Ndufs4(−/−) cohorts exposed to mock anesthesia exposures (carrier gas only) overlap.Exposure to 0.4% isoflurane shortened survival in both control diet fed (as previously shown, 15 ) and ketogenic diet fed mice (Figure 4C).Critically for focus here, Ndufs4(−/−) mice exposed to 0.4% isoflurane show significantly reduced survival if fed a ketogenic diet compared to a non-ketogenic diet (Figure 4C).
In Ndufs4(−/−) mice fed a normal diet, the primary cause of death was euthanasia due to severe neurologic disease or reaching a predetermined weight cut-off (20% loss from maximum, see Section 2).
The cause of death distribution was shifted by both isoflurane and ketogenic diet.50% of Ndufs4(−/−) mice on a ketogenic diet that were exposed to isoflurane died during the immediate post-recovery from anesthesia (Figure 4D,F).Overall mortality among isoflurane exposed animals during the P50-53 period (the anesthetic treatment paradigm period) was significantly increased in mice fed the ketogenic diet but the ketogenic diet did not impact death during the P50-53 period in the mock anesthesia (carrier gas only) groups (Figure 4D,E).

| DISCUSS ION
Here, we have tested whether dietary ketosis changes responses to isoflurane exposure in the setting of genetic mitochondrial disease using the Ndufs4(−/−) mouse model of LS.
4][15] Given these data, our findings here further suggest that use of a ketogenic diet in mitochondrial disease should be carefully considered.These results may also have some bearing on the practice of fasting mitochondrial patients prior to anesthetic exposure, as a fasting induced ketogenic state may similarly impact anesthesia.Recent reviews of the management of mitochondrial disease F I G U R E 4 Anesthesia toxicity in the mouse model of Leigh syndrome is worsened by ketogenic diet.(A) Change in weight over the 3-day exposure period (see Section 2).Two-way ANOVA: anesthesia exposure effect *p < .05,diet effect not significant, interaction nonsignificant.Pairwise comparisons *p < .05by Tukey's multiple testing corrected multiple comparisons with all possible comparisons made.n = 11, 11, 8, and 4 for (−)keto/(−)iso, (−)keto/(+)iso, (+)keto/(−)iso, and (+)keto/(+)iso, respectively (only animals surviving to at least P51).(B) Survival of Ndufs4(−/−) mice treated with ketogenic diet in this study compared to previously published data (see 15 ), showing survival curves up until P50, the start of anesthetic exposures.Survival to this age is consistent with prior cohorts.n = 16 (historic data) and 11 (cohort in this manuscript).(C) Survival of ketogenic diet and normal (non-ketogenic) diet fed Ndufs4(−/−) mice exposed to 0.4% isoflurane or mock conditions as detailed (see Figure 1, Section 2).Isoflurane exposed ketogenic diet fed animals had significantly reduced survival compared to normal (non-ketogenic) diet fed mice exposed to isoflurane (*p = .0121by Gehan-Breslow-Wilcoxon test, blue versus green lines), while ketogenic diet did not reduce survival in mock exposed animals (pink versus orange lines).0.4% isoflurane exposure significantly reduced survival in both ketogenic diet and normal (non-ketogenic) diet fed mice, as previously demonstrated (*p = .0108and *p = .0546by Gehan-Breslow-Wilcoxon test, green versus orange and pink versus blue, respectively).n = 10, 11, 8, and 8 for (−)keto/(−)iso, (−)keto/(+) iso, (+)keto/(−)iso, and (+)keto/(+)iso, respectively.(D) Fraction of mock (no isoflurane) exposed animals dying during or after the exposure paradigm period (P50-53).(E) Fraction of 0.4% isoflurane exposed animals dying during or after the exposure paradigm period (P50-53), *p = .0198by two-sided Fisher's exact test.(D-E) n's as in C. (F) Distribution of causes of death among animals in the survival studies (A-E).Full numerical breakdown provided in Data S1.
patients recommend against prolonged fasting in these patients in order to avoid acidosis and ketosis. 4,33,34However, while the consensus recommendation from the Mitochondrial Medicine Society notes that IV glucose might be avoided in patients on a ketogenic diet for seizure control, no recommendations are provided regarding the management of anesthesia in those patients, or potential interactions between the diet and anesthesia exposures.
A ketogenic diet alters overall metabolic status, lowering steady-state blood glucose and raising blood ketones.We originally hypothesized that dietary ketosis would attenuate some of the toxicities of anesthesia, in particular hyperlactemia and hyperglycemia.In contrast to these expectations, dietary ketosis exacerbated metabolic dysfunction, as noted.Together with disease progression overall, the impact of dietary ketosis in the setting of mitochondrial disease has proven hard to predict, emphasizing the importance of preclinical studies and cautious clinical studies.
Our findings here also emphasize the potential for interactions between therapeutic dietary ketosis and other clinical interventions.
While the work here focused on anesthesia, mitochondrial disease patients are often treated with high doses of putative mitochondrial metabolism supporting vitamins, supplements such coenzyme Q (CoQ), and antioxidants, in an ad-hoc mixture colloquially referred to as the "mito cocktail" (e.g., see 35 ).Our observations warrant a more cautious approach to such forms of supplementation given the difficulty in predicting drug-diet-disease interactions in the setting of mitochondrial disease.In particular, any intervention which might impact metabolic state should be met with heightened metabolic monitoring for increased lactate and acidosis during anesthesia.
We have previously shown that carrier gas oxygen concentration and immune-targeting interventions are two potent means by which to mitigate volatile anesthesia toxicity in the Ndufs4(−/−) model of mitochondrial disease. 12Future studies using this model may provide insight into opportunities for attenuating anesthetic toxicities and improving outcomes in mitochondrial disease patients managed with dietary ketosis, and perhaps in the setting of dietary ketosis more broadly.
As dietary ketosis is viewed as an effective strategy for management of seizures, identifying strategies to mitigate harms remains important.In addition, given the frequency of surgical intervention in the clinical management of patients with mitochondrial disease, it is important to consider possible interactions between anesthetic exposures and other dietary, supplement, and pharmacologic interventions.We anticipate that such studies will contribute to better informed clinical management of mitochondrial disease patients.
Anesthesia hypersensitivity in the setting of genetic mitochondrial dysfunction is now well documented, in particular among patients with ETC CI defects. 5Our findings here, and in our 2023 study, 12 support the need for published comprehensive guidelines for anesthesia in mitochondrial disease patients which consider anesthetic agent, carrier gas, fasting and perioperative fluids, etc. Protocols for the management of anesthesia in pediatric mitochondrial disease patients at Seattle Children's Hospital emphasize IV glucose, minimal fasting, EEG monitoring, and oversight by a mitochondrial disease anesthesia team who shepherd these patients successfully through the operating room.Our observations indicate that guidelines for management of mitochondrial disease patients also consider potential interactions between experimental therapies, such as a ketogenic diet, and anesthetics.

| CON CLUS ION
Our data reveal that dietary ketosis aggravates toxicities resulting from volatile anaesthetic exposure in the Ndufs4(−/−).In particular, hyperlactemia was significantly exacerbated, and dietary ketosis was associated with a significantly increased rate of mortality in Ndufs4(−/−) mice during the anaesthetic exposure paradigm.Mouse responses may differ from those in humans, but the findings detailed here suggest that volatile anaesthetic exposure in patients with genetic mitochondrial disease on a ketogenic diet should be undertaken with increased caution.
All experiments were approved by the Institute Animal Care and Use Committee at Seattle Children's Research Institute (Seattle, WA) under protocols IACUC00611 and IACUC00070.The Ndufs4 knockout (null, homozygous loss of function, (−/−)) mouse line (Jackson Laboratory strain #027058) was obtained from the Palmiter Laboratory, University of Washington, Seattle, USA.

F
I G U R E 1 Experimental paradigm for assessing interactions between ketogenic diet and anesthesia in the Leigh syndrome model.(A) A schematic detailing the timing of dietary intervention (ketogenic diet treatment), disease symptom onset, anesthetic exposure, and survival (in untreated Ndufs4(−/−) mice).Ketogenic diet is administered starting at weaning, with an initial ramp-up in percent (see Section 2).Symptom onset occurs at ~P37 in Ndufs4(−/−) animals, with progressive cachexia and visible signs of neurodegenerative changes, such as forelimb clasping, visible starting around this age.To assess anesthetic responses, animals were exposed to brief (30-min) anesthesia with isoflurane on P50, 51, and 52 (see Section 2 and 12 ).Median and maximum survival in untreated Ndufs4(−/−) mice is ~60 and 80 days of age.For reference-control animals (Ndufs4(+/−) and Ndufs4(+/+)) of the C57Bl/6 background live ~26-30 months.(B-G) Levels of blood metabolites β-hydroxybutyrate (BHB), glucose, and lactate at P50, prior to any anesthesia exposures.(B) Blood BHB at P50 in Ndufs4(−/−) mice fed a control (non-ketogenic) diet or a ketogenic diet.n = 22 and 15 for control and ketogenic diet, respectively.****p < .0001,unpaired two-tailed t-test.(C) Blood glucose at P50 in Ndufs4(−/−) mice fed a control (non-ketogenic) diet or a ketogenic diet.n = 22 and 15 for control and ketogenic diet, respectively.**p = .0075,unpaired two-tailed t-test.(D) Blood lactate at P50 in Ndufs4(−/−) mice fed a control (nonketogenic) diet or a ketogenic diet.n = 23 and 15 for control and ketogenic diet, respectively.Not significantly different, unpaired two-tailed t-test.(E) Blood BHB at P50 in control mice fed a control (non-ketogenic) diet or a ketogenic diet.n = 21 and 15 for control and ketogenic diet, respectively.****p < .0001,unpaired two-tailed t-test.(F) Blood glucose at P50 in control mice fed a control (non-ketogenic) diet or a ketogenic diet.n = 21 and 16 for control and ketogenic diet, respectively.**p = .0061,unpaired two-tailed t-test.(G) Blood lactate at P50 in control mice fed a control (non-ketogenic) diet or a ketogenic diet.n = 21 and 16 for control and ketogenic diet, respectively.See Data S1 for additional statistics and effect sizes.