Ketone body 3‐hydroxybutyrate mimics calorie restriction via the Nrf2 activator, fumarate, in the retina

Summary Calorie restriction (CR) being the most robust dietary intervention provides various health benefits. D‐3‐hydroxybutyrate (3HB), a major physiological ketone, has been proposed as an important endogenous molecule for CR. To investigate the role of 3HB in CR, we investigated potential shared mechanisms underlying increased retinal 3HB induced by CR and exogenously applied 3HB without CR to protect against ischemic retinal degeneration. The repeated elevation of retinal 3HB, with or without CR, suppressed retinal degeneration. Metabolomic analysis showed that the antioxidant pentose phosphate pathway and its limiting enzyme, glucose‐6‐phosphate dehydrogenase (G6PD), were concomitantly preserved. Importantly, the upregulation of nuclear factor erythroid 2 p45‐related factor 2 (Nrf2), a regulator of G6PD, and elevation of the tricarboxylic acid cycle's Nrf2 activator, fumarate, were also shared. Together, our findings suggest that CR provides retinal antioxidative defense by 3HB through the antioxidant Nrf2 pathway via modification of a tricarboxylic acid cycle intermediate during 3HB metabolism.

mitochondrial hydroxymethylglutaryl-CoA synthase (mHMG-CoAS), is regulated through deacetylation by Sirt3 (Shimazu et al., 2010) and mTOR (Sengupta et al., 2010), master regulators of energy metabolic homeostasis signaling, and these activities have been linked to ameliorating diseases associated with oxidative stress. A more recent finding also showed that 3HB is an endogenous histone diacetyl-lase inhibitor that helps cells resist oxidative stress (Shimazu et al., 2013). Furthermore, endogenous 3HB inhibits the activation of the NLRP3 inflammasome in macrophages, which is linked to chronic anti-inflammatory effects (Youm et al., 2015). However, the overlapping mechanism between CR and 3HB has not yet been Several studies have shown the beneficial effect of CR intervention for retinal degeneration in rodents. CR delays the age-related degeneration of the retina, including RGC (Li, Sun, & Wang, 2003). CR attenuates the decrease in RGC activity caused by intraocular pressure elevation (Kong et al., 2012) and protects against ischemia/ reperfusion stress-induced RGC reduction (Kawai et al., 2001). A ketogenic diet and high-fat/low-carbohydrate and protein diet cause the elevation of endogenous 3HB and attenuate age-related damage to RGC in rats (Zarnowski et al., 2015).
Here, we investigated the role of 3HB in the ameliorative effect of CR on retinal degeneration. Retinal degeneration was evaluated using an optic nerve and central retinal blood vessels transection (ONVT) model which simulates the acute retinal ischemic degeneration. Animals receiving CR were compared to animals receiving exogenous 3HB without CR. We focused on metabolic changes under these conditions. This study demonstrates not only the potential effect of CR on retinal degeneration, but also the novel role of 3HB in CR.

| RESULTS
2.1 | Intermittent fasting and exogenous, repetitive application of 3HB attenuate retinal degeneration First, to verify the validity of the experimental conditions, changes in body weight, total calorie intake, serum biochemicals, and blood 3HB were compared between three groups:(i) allowed food and water ad libitum (AL), (ii) intermittent fasting (IF), and (iii) repeated administration of 3HB under the AL condition (3HB-r).
Linear elevation of the body weight was observed in the AL group, and it increased by about 25% compared with the initial value on day 7 (Figure 1a). In the IF group, a decrease followed by an increase was observed repeatedly, corresponding to the daily fasting and feeding treatments, respectively. From days 2 to 7, the body weight significantly decreased to about 80%-90% compared with that in the AL group. In the 3HB-r group, body weights were linearly elevated similar to that in the AL group. The total calorie intake in the IF group significantly decreased to approximately 60% of that of the AL group (Figure 1b). In the 3HB-r group, the calorie intake was almost equivalent to that of the AL group.
Changes in the 3HB concentration in the vitreous fluid, forming a fluid border with the retina, were measured during IF and after 3HB subcutaneous administration (Fig. S1, and Data S1). 3HB increased gradually for 12 hr after food restriction, and then a marked rise was noted in the subsequent 12 hr. The concentrations of 3HB at 12 hr and 18-24 hr were four-and 10-fold higher, respectively. The resumption of feeding markedly reduced 3HB to the same level as the initial value within 2 hr. The concentration and pattern of increase and decrease in response to fasting and subsequent feeding of the vitreous were the same as in the serum. 3HB subcutaneous administration resulted in an increase in the vitreous 3HB concentration, which peaked within 30 min, being 15-fold higher than the initial value.
Vitreous 3HB concentration then gradually decreased to the same level as the initial value within 3 hr (Fig. S1b). The concentration and pattern of increase and decrease in response to 3HB injection were the same in the vitreous fluid and serum. The peak 3HB concentration was similar to that observed in the IF group. Thus, the kinetics of the changes in 3HB concentration were suitable to simulate IF conditions. Serum biochemical analysis indicating changes in levels of glucose and lipid-related factors is shown in Table S1. In the IF group, concentrations of glucose, total lipids (TL), and triglycerides (TG) decreased to 50%, 60%, and 20%, respectively, compared with those in the AL group. These changes may be due to the interruption of carbohydrate and lipid absorption by fasting. Nonesterified fatty acids (NEFAs) increased to 170% when compared with that in the AL group. This increase was a lipolytic response of adipose tissue to carbohydrate deprivation by fasting. Released NEFA is used as a source for ketogenesis through beta oxidation of fatty acid in hepatic mitochondria as an alternative energy supplementation. These glucose and lipid metabolism changes were consistent with a previous CR report (Cahill & Veech, 2003). In the 3HB-r group, the glucose level was similar to that of the AL group. Concentrations of TL, TG, and NEFA decreased to 80%, 70%, and 80%, respectively. These decreases may be associated with the decrease in food intake in the 3HB-r group, which was approximately 90% of that of the AL group (Table S2). Altogether, these results suggest that 3HB-r, namely the elevation of 3HB without CR, minimally affected the metabolic state in relation to ketogenesis.   We also analyzed retinal cell viability in freshly isolated retina 4 hr after ONVT ( Figure 1f). The viability decreased to approximately 35%, 80%, and 75% compared with that in the sham-operated eye in the AL, IF, and 3HB-r group, respectively. Significant preservation of retinal cell viability was observed in the IF and 3HBr groups compared to that in the AL group.
Induction of edematous thickening in INL to GCL, swelling of ganglion cells, and decrease in retinal viability in freshly isolated retina after ONVT were not suppressed by single fasting (SF) and 3HB-s injection ( Fig. S2a, b, and c).
These results using IF, 3HB-r with SF, and 3HB-s suggest that repeated, prominent elevation of 3HB plays a role in retinal protection by IF. For the downregulated cluster, six metabolites were associated with the pentose phosphate pathway (PPP). The pentose phosphate pathway is a cytosolic process that is an alternative pathway for glucose oxidation. The role of PPP is to provide pentose phosphates for nucleotide synthesis and to generate reducing agents for many biosynthetic pathways and antioxidant defense (Eggleston & Krebs, 1974).

|
The changes in PPP and its downstream pathways as well as metabolites from metabolomic analysis in AL, IF, and 3HB-r are summarized in Figure 2d. Glucose-6-phosphate (G6P), a starting metabolite of PPP, and the following PPP elements: fructose-6phosphate (F6P), fructose-1,6-bisphosphate (F1,6P), and 6-phosphogluconate, were identified in this cluster. NADPH is a cofactor involved in reductive biosynthesis generated from the first two steps of PPP. Reduced glutathione is generated at the expense of as a metabolic fuel has been well recognized, although its involvement was excluded because SF and 3HB-s had no protective effects.

| PPP upregulation by elevation of 3HB
suppresses ROS generation after ONVT PPP generates NADPH for the detoxification of reactive oxygen species (ROS; Li et al., 2014). ROS generation from freshly isolated retinas and presence of an oxidative stress marker Ne-(hexanoyl) lysine (HEL) were detected in retinal cross-sections.
Reactive oxygen species generation 30 min after ONVT was approximately 1.5-, 1.1-, and 1.1-fold compared with that in the sham-operated eye in the AL, IF, and 3HB-r groups, respectively.

Significant suppression of ROS generation was observed in the IF
and 3HB-r groups compared to that in the AL group.
In the AL group, HEL expression was observed predominantly in GCL and IPL after ONVT (Figure 3b, upper). In the IF and 3HB-r groups, HEL was not detected in any of the layers. Consistent with metabolomic data, this finding suggested that the preserved retinal PPP presented ROS detoxification properties after ONVT. 2.4 | Prior to ONVT, G6PD, a rate-limiting enzyme of PPP as well as an upstream transcriptional factor of Nrf2, was upregulated by IF and 3HB-r As antioxidant PPP was preserved by IF and 3HB-r, but not by SF or 3HB-s, after ONVT, we hypothesized that the mechanisms for the preservation of PPP are established prior to ONVT. G6PD is a ratelimiting enzyme of PPP which activates this pathway in response to the breakdown of NADPH (Ben-Yoseph et al., 1996). In addition, G6PD is one of the biological protective factors orchestrated by Nrf2 (Mitsuishi et al., 2012), which is a transcriptional factor that ubiquitously responds to various physiological and pathological conditions.
The nuclear levels of Nrf2 and G6PD were evaluated prior to ONVT. In the IF and 3HB-r groups, significant increases in G6PD activity (

| DISCUSSION
Ketone body metabolism is a metabolic pathway based on ketone body production with a release to the blood circulation mostly from the liver (ketogenesis) and utilization in extrahepatic tissues as an energy source (ketolysis). Investigation of the pleiotropic effects of 3HB on health benefits mainly focused on the ketogenesis pathway.
In the present study, we demonstrate that upregulation of ketolysis in the retina, coupled with activated ketogenesis, plays a role in CRinduced retinal antioxidative defense.
Ketogenesis is activated in concert with gluconeogenesis to provide an energy alternative to fatty acids during fasting (Garber, Menzel, Boden, & Owen, 1974). The prominent synthesis of 3HB through ketogenesis is one of the characteristic metabolic changes due to nutrient deprivation such as CR. In this state, the plasma 3HB level rises over four orders of magnitude, despite other substrates, free fatty acids, glycerol, and amino acids from gluconeogenesis being buffered within a far smaller range (Cahill, 1970). The urinary excretion level of 3HB is synchronized with that of plasma, suggesting that synthesis is excessive to deal with the cellular energy demand (Owen & Cahill, 1973) and it has been recognized as a waste material. However, little attention has been directed to clarify the relationship between the prominent synthesis of 3HB and health benefits.
In the present study, a repeated prominent increase of 3HB was Ketolysis is a degradation process of 3HB into acetyl-CoA through the following three phases (Krebs, 1970) with the second phase of ketolysis (Krebs, 1970 Despite providing novel insights, this study presents some limitations. Our results were limited to the short-term effects of CR in acute-phase retinal degeneration using young and middle-aged rats.
Further studies investigating the effect of extended CR on chronic retinal degeneration with aging will provide insights into the role of the association between 3HB, fumarate, and Nrf2 in aging-related retinal diseases.
In the present study, we propose a novel mechanism for the antioxidative effect of endogenous 3HB during CR ( Figure 6). Calorie restriction-initiated ketogenesis, a prominent 3HB synthesis pathway in hepatic tissue, increases the 3HB concentration in blood and peripheral tissues. This increase adaptively upregulates the ketolytic pathway through its rate-limiting enzyme SCOT, and is accompanied by an acceleration of the TCA cycle flux with an increase in fumarate, an antioxidative Nrf2 activator and TCA cycle intermediate.

Nuclear translocation of Nrf2 upregulates its various downstream antioxidative factors.
Ketolysis is a well-recognized cellular metabolic defense mechanism against energy deprivation, which is fundamental to peripheral tissues. Nrf2 is ubiquitously expressed in important cell defense and survival pathways under various physiological and pathological conditions. Therefore, it can be assumed that the mechanism underlying the antioxidant effect of 3HB participates in not only retinal protection, but also the ubiquitous, wide-ranging health benefits of CR.

Investigations of dietary intervention that focus on alterations in
ketone body metabolism, especially on the 3HB-utilizing process, ketolysis, may be a straightforward CR mimic strategy for various aging-associated disorders, including retinal degeneration with vision loss such as glaucoma.

| Experimental design
The following five groups were created: ad libitum (AL), provided access to food every other day (intermittent fasting, IF), 24-hr fasting (single fasting, SF), 7 days of twice-daily subcutaneous injections of 3HB with ad libitum (3HB-r), and single-dose subcutaneous injection of 3HB with ad libitum (3HB-s). Body weight was recorded daily.

| Calorie intake
Food intake was recorded daily (AL and 3HB-r) or every other day (IF) during the experiment. Total calorie intake was calculated as the sum of food-derived calorie intake with (3HB-r) or without (AL and IF) exogenously injected 3HB-derived calorie intake -given calorie intake values of 3.59 kcal/g derived from food and 4.96 kcal/g from 3HB.

| Calorie restriction
For IF, rats were provided with unlimited access to food every other day for 7 days. Animals were used for experiments 24 hr after the last fasting started. For SF, after the acclimation period, AL treatment was continued for 6 days. Subsequently, rats had only access to water for 24 hr and were then used for experiments.

| 3HB administration
Sodium D-3-hydroxybutyrate (3HB, Ophtecs Corporation, Ltd., Hyogo, Japan) was used. It was dissolved in saline and injected into the dorsal skin of rats at 1,000 mg/kg.

| Optic nerve and central retinal blood vessel transection
The rats were systemically anesthetized with pentobarbital (64.8 mg/kg) and the ocular surface was locally anesthetized with benoxil eye drops (Santen Pharmaceutical, Osaka, Japan). Then, the

| Retinal viability
Freshly isolated neural retinas were immersed in a 96-well culture plate containing 100 ll of serum-free neurobasal medium and 10 ll of WST-8 (Cell Counting Kit-8, Dojindo, Kumamoto, Japan) per well, and then incubated (37°C, 5% CO2) for 1 hr immediately after euthanization. After incubation, the optical density of the culture supernatant was measured at 450 nm using a microplate reader (SpectraMax Paradigm; Molecular Devices, Sunnyvale, CA, USA).

| Metabolomic analysis
A total of 10 groups, AL, SF, IF, 3HB-r, and 3HB-s with or without transection, underwent analysis. Retinas harvested 1 hr after ONVT were used. Two retinas were pooled for each measurement. Analyses of a total of 109 cations and 91 anions by capillary electrophoresis-mass spectrometry were carried out as previously described (Soga & Heiger, 2000;Soga et al., 2009Soga et al., , 2003 with some modifications. Z-scored averages of three samples (n = 3) were used for the construction of a clustered heatmap using R software (http://www.rproject.org/) with the function "heatmap.2".

| Western blot analysis
After euthanization, eyeballs were harvested and isolated retinas were homogenized with RIPA buffer (50 mM Tris/HCl pH 7.5, 150 mM NaCl, 1.0% Igepal CA-630) containing protease inhibitor cocktail (Complete Mini; Roche Diagnostics, Basel, Switzerland), and 10 lM MG132 (Peptides International, Louisville, KY, USA). Samples were centrifuged at 4°C, 20,400 g, for 5 min, and the protein concentration was determined by the DC protein assay kit (Bio-Rad, Hercules, CA, USA). Then, the same volume of 29 Laemmli sample buffer was added and 5% b-mercaptoethanol. After boiling, the sample was separated by SDS-PAGE. After electrophoresis, the proteins were transferred onto a PVDF membrane. The membranes were incubated with 5% skim milk or 5% BSA + 0.1% Tween-20 in TBS, and then with primary antibodies. The primary antibodies were as follows: anti-G6PD antibody (Atlas antibodies AB

| Translocation of nuclear Nrf2 in the retina
For the Western blot analysis, retinal nuclear protein was extracted from freshly isolated retina using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce Biotechnology, Rockford, IL, USA) according to the manufacturer's instructions. The extracted retinal nuclear proteins were applied to Western blotting analysis as described above. Anti-Nrf2 antibody (MBL, Nagoya, Japan) at 1:1,000 and anti-TBP antibody (Cell Signaling Technology) at 1:1,000 were used as primary antibodies. The ratio of the band intensity of Nrf2 to that of TBP was calculated.
For the localization analysis by immunohistochemical staining, rats were euthanized with an overdose of pentobarbital sodium (120 mg/ kg) prior to ONVT. The above-mentioned immunostaining protocol was applied as Nrf2 staining method. Anti-Nrf2 antibody (Abcam, Cambridge, UK) was used as the primary antibody at 1:100. The colocalization of Hoechst-stained nuclei and immunostained Nrf2 was observed by using a confocal fluorescence microscope (LSM 710).
4.14 | Measurement of glucose-6-phosphate dehydrogenase activity Total glucose-6-phosphate dehydrogenase (G6PD) activity levels per retinal weight were measured using the Glucose-6-Phosphate Dehydrogenase Assay Kit (Sigma-Aldrich) according to the manufacturer's instructions.

| Measurement of fumarate
The total fumarate content per retina was measured using the Fumarate Assay Kit (Sigma-Aldrich) according to the manufacturer's instructions.

| Measurement of retinal oxygen consumption rate
Immediately after euthanization, their eyeballs were immediately excised and the retina was isolated in ice-cold saline. The retina was spread flat, and hollowed out with 1.5-mm biopsy punch (Kai Industries, Gifu, Japan) and used for measurement of retinal basal OCR. Basal OCR was measured by XF 24-3 flux analyzer (Seahorse Biosciences, Billerica, MA, USA). The retinal basal OCR was measured in 0, 5, and 10 mM 3HB in saline for 10 min, followed by saline for 20 min as the baseline. All measurements were performed at 37°C. The percentage to the baseline value of each 3HB concentration was calculated.
All experimental protocols, including calorie restriction, 3HB administration, and ONVT, were performed following the protocol used for the rat study.

| Statistical analysis
We used Student's t-test for comparison of the two groups and the Dunnett's test for multiple comparisons. Differences between measured variables were considered significant if the resultant p-value was .05 or less. Analysis was performed using JMP version 12.0.1 (SAS Institute, Cary, NC, USA).

ACKNOWLEDG MENTS
We thank Dr. Takashi Kojima (Department of Ophthalmology, Keio University School of Medicine, Shinjuku, Tokyo, Japan) for generously providing Nrf2 knockout mice. We thank Koji Obayashi and Sachiko Shimoda for their assistance. This work was partly supported by