1. BHB supplementation but not ketogenic diet was well tolerated by Grin1KD mice
Although we ensured >10g body weight before initiating ketogenic diet (Keto), 8 out of 16 Grin1KD mice receiving ketogenic diet were found dead or met endpoint (>20% weight loss with or without lesions on face, limbs, or chest) before completing the study. This was not observed in WT or Grin1KD mice receiving control diet (Figure 1A). All animals receiving BHB-supplemented water showed tolerance to the treatment (no health concerns). The weekly consumption of BHB-supplemented water was significantly more than that of control water (Figure 1B, p<0.01), especially after two and three weeks of treatment (p<0.05). The effect of genotype on BHB preference was not tested since mice of both genotypes were housed together.
Consistent with our previous observations, Grin1KD mice had significantly lower body weights than WT mice. (Figure S1, p<0.05). In studies with BHB, sex, age, and treatment were all significant factors on body weight. BHB improved Grin1KD body weights in both male and female mice (Figure S1C-D, males: p=0.0548, female: p=0.0104). A trend of increased body weight in male Grin1KD mice was also observed with ketogenic diet in comparison to control diet (Figure S1A, p=0.0804). However, Grin1KD body weight was still significantly lower than wildtype mice in adulthood (p’s<0.05, data not shown).
2. Fasting blood ketone was elevated after short-term ketogenic diet or BHB treatment but not after chronic exposure
As expected, increased blood ketone was observed in WT (2.08±0.26mmol/L) and Grin1KD (3.73±0.24mmol/L) mice receiving ketogenic diet for one week, in comparison with those on control diet (WT control: 1.10±0.13mmol/L, Grin1KD control: 1.91±0.23mmol/L, p’s<0.05, Figure 2A). However, after nine weeks of ketogenic diet treatment, elevated blood ketone level was only observed in Grin1KD mice (2.43±0.20mmol/L in Keto-treated vs. 1.00±0.09mmol/L in control, p<0.05) but not in WT mice (1.11±0.17mmol/L in Keto-treated vs. 0.85±0.69mmol/L in control, p>0.05, Figure 2B).
After one week of BHB treatment, we found significantly increased blood ketone levels in both WT (0.57±0.06mmol/L) and Grin1KD (0.98±0.10mmol/L) mice compared to untreated mice of the same genotype (WT: 0.18±0.04mmol/L, p<0.01; Grin1KD: 0.11±0.03mmol/L, p<0.0001, Figure 2C). As with the ketogenic diet, the elevation in blood ketone level was no longer detected with nine weeks of BHB treatment in both genotypes (untreated WT: 0.15±0.03mmol/L, BHB-treated WT: 0.14±0.03mmol/L; untreated Grin1KD: 0.36±0.09mmol/L, BHB-treated Grin1KD: 0.14±0.03mmol/L; Figure 2D). Post hoc comparisons revealed increased blood ketones in untreated Grin1KD mice compared to WT (p<0.01), which was “normalized” after BHB treatment (Grin1KD non-treated vs. treated: p<0.01, BHB-treated WT vs. Grin1KD: ns).
3. BHB supplementation and ketogenic diet showed modest behavioral benefits on Grin1KD mice
As expected, hyperactivity was observed in untreated Grin1KD mice in comparison with WT. Hyperlocomotion was reduced in both Keto-treated (Figure 3A-B) and BHB-treated (Figure 3C-D) Grin1KD mice in comparison with untreated Grin1KD mice. Neither ketogenic diet nor BHB supplementation affected the locomotor activity of WT mice (Figure 3A-D).
Sensorimotor gating was assessed by measuring the prepulse inhibition of acoustic startle response (PPI). Consistent with previous studies, reduced PPI was observed in Grin1KD mice at baseline. BHB supplementation but not ketogenic diet showed a modestly ameliorating effect on PPI impairment in Grin1KD mice (Figure 3E-F). Analysis of the acoustic startle response (ASR) detected the elevated ASR in Grin1KD treated by control diets but both Keto and BHB had no effect on ASR (Figure S2).
Social behavior was measured with a modified “three-chamber” test, where the study was conducted with a single arena. Video tracking software measured mouse interactions with novel mice constrained to a wire cage in the arena, or with an empty cage providing a non-social stimulus. Given that Grin1KD mice are hyperactive, average duration of time per visit was analyzed and reported for both genotypes. WT mice showed the expected preference for socially unfamiliar mouse over non-social neutral object, spending more time per visit near the stranger mouse (Figure 4A, p’s<0.05). In contrast, Grin1KD mice spent a similar amount of time per visit exploring the stranger mouse and the empty cup. The time per visit near the stranger mouse was significantly reduced in Grin1KD mice in comparison to WT mice (Figure 4B, p<0.001). There was no significant effect of the ketogenic diet on sociability in either genotype (Figure 4A-B, p’s>0.05). BHB treatment did not improve social motivation in Grin1KD mice; both treated and untreated Grin1KD mice spent significantly less time per visit around the stranger mouse in comparison to WT mice (Figure 4D, p<0.05). Interestingly, BHB treatment reduced social motivation in WT mice. BHB treated WT mice showed no preference for social over non-social interactions, spending a reduced amount time investigating the stranger mouse in comparison to non-treated WT mice (Figure 4C-D). In conclusion, BHB supplementation had no beneficial effect on the social motivation of Grin1KD mice but did impair the social motivation of WT mice.
To address potential effects ketogenic diet and BHB on cognition, we tested working memory in a Y-maze assay and short-term spatial novelty detection in a displaced object recognition test. In the Y-maze assay, the percentage of spontaneous alternations was reduced in Grin1KD mice in comparison to WT mice (Figure 5B, D, p’s<0.0001), indicating impaired working memory. Neither the ketogenic diet nor BHB supplementation had a significant effect on working memory in Grin1KD or WT mice (Figure 5) regardless of ketogenic diet’s effect on hyperactivity of Grin1KD mice in Y-maze (Supplementary Figure S3).
Similar results were observed in the test of the short-term spatial memory, assessed by the displaced-object recognition paradigm. In this task, WT mice displayed a preference to explore the displaced object over the non-displaced object while the Grin1KD mice did not (Figure 5A, Genotype: F (1,36) = 8.396, p=0.0064). This observed deficit in spatial recognition was not improved by ketogenic diet. In the cohorts used to study BHB supplementation, similar results were observed in the untreated WT and Grin1KD mice (Figure 5C, Genotype: F (1,35) = 6.885, p=0.0128). BHB supplementation did not improve spatial recognition.
4. BHB but not ketogenic diet improved myelination in corpus callosum of Grin1KD mice.
Our previous study showed volumetric deficits in white matter of Grin1KD mice (30), but the reason for the deficits was not understood. Since BHB has been reported to improve myelination in a mouse model of multiple sclerosis (31), we explored the possibility that the benefits of BHB diet on Grin1KD mice might be related to changes in white matter.
Myelin integrity was assessed by electron microscopy of the corpus callosum (Figure 6A). We measured the percentage of axons that were myelinated, the thickness of the myelin sheath (G-ratio), and the incidence of abnormal myelination (decompaction, fragmentation, separation, hypermyelination, aberrant outfolding and invagination). In the ketogenic diet study, we observed no difference in the percentage of myelinated axons or the thickness of myelin sheath between Grin1KD and WT mice. Furthermore, there was no effect of the ketogenic diet in either genotype (Figure 6B; Figure S4A-B). Surprisingly, the ketogenic diet significantly increased the percentage of abnormal myelination in both genotypes (Figure 6C, Keto effect: F (1,14) = 4.790, p=0.0461). Decompaction was the primary form of myelin abnormality observed in ketogenic diet-treated mice (Figure 6D, A).
In the cohort of mice used for the BHB study, two-way ANOVA detected a main effect of genotype: F (1,13) = 4.452, p=0.0548 and genotype´ BHB diet interaction: F (1,13) = 4.031, p=0.0659] on the percentage of myelinated axons. The control Grin1KD mice had a significant reduction in the percentage of axons that were myelinated as compared to control WT mice (Figure 7A-B, p<0.05). BHB supplementation showed a modest trend towards increased the percentage of myelinated axons in Grin1KD mice (Figure 7B). Similarly, no difference in myelination thickness was observed between WT and Grin1KD mice and BHB had no effect in either genotype (Figure S4C-D). Importantly, unlike the ketogenic diet, BHB supplementation did not increase the percentage of abnormally myelinated axons in mice of both genotypes (Figure 7C-D).