High dose vitamin D significantly improves grip strength compared to normal vitamin D
To examine calciometabolic-independent effects of vitamin D on muscle, we used the vitamin D receptor knockout rescue diet (VDRKR diet) and varied vitamin D in wild-type (C57BL/6) male mice. We used three levels of added vitamin D in the diet: 0 IU/kg (no-D), 2000 IU/kg (normal-D) and 10,000 IU/kg (high-D). After 4 weeks, these diets allowed us to achieve target serum 25(OH)D of less than 5 ng/mL, between 20–30 ng/mL and above 30 ng/mL (experimental schematic in Fig. 2a).
To determine whether high-D improves muscle function in a dose-responsive fashion above the bottom of the normal range, we measured grip strength as previously described17. Normal-D significantly improved grip strength over no-D. High-D more dramatically increased grip strength over normal-D (Fig. 2b, ***p < 0.001 for each by Tukey’s Post-hoc test). While the improvement in going from no-D to normal-D is expected, the more substantial improvement in going from normal-D to high-D is novel.
High dose vitamin D increases lean mass, but normal vitamin D does not
To examine whether the high-dose vitamin D stimulated increase in strength reflected changes in muscle mass, we measured mouse body composition via NMR as described previously17. Normal-D did not increase lean mass over low-D but high-D increased lean mass and decreased fat mass relative to normal-D without altering weight (Fig. 2c for lean mass, *: p < 0.05 by ANOVA with Tukey post-tests; Fig. 2d for fat mass, *: p < 0.05 by ANOVA with Tukey post-tests; Fig. 2e for weight, p > 0.05 by ANOVA). Thus, the high-dose vitamin D stimulated increase in strength is mediated in part by increased calorie allocation to build muscle mass instead of fat mass (physiologic schematic in Fig. 2f).
Increased vitamin D inhibits myostatin production
Recent work in muscle-specific VDR knockout mice as well as in mouse models of vitamin D deficiency reveal that normal vitamin D concentrations facilitate VDR-dependent decreases in myostatin mRNA in muscle11. To examine the possibility that vitamin D dose alters circulating myostatin to regulate the proportion of muscle mass and fat mass, we measured serum myostatin in all three diet groups. Normal D decreased serum myostatin significantly relative to no-D (Fig. 3, no-D vs normal-D, p < 0.05, ANOVA with Tukey post-tests). High-D did not further alter serum myostatin relative to normal-D (Fig. 3, normal-D vs high-D, p > 0.05, ANOVA with Tukey post-tests). This first decrease in myostatin (Fig. 3a) occurred without any change in fat free mass (Fig. 2c) suggesting that normal-D decreased average production of myostatin by muscle mass. The unchanged average myostatin in the high-D diet compared to normal-D occurred in the context of increased average lean mass (Fig. 2b), suggesting that high-D further decreased average myostatin production per muscle mass. Overall, these results are consistent with increasing doses of dietary vitamin D inhibiting myostatin production (Fig. 3b).
Normalizing vitamin D improves leptin generation per fat mass, but high-dose vitamin D increases leptin sensitivity
Vitamin D receptor deficient mice are hypoleptinemic and are largely deficient in white fat, the type of fat largely responsible for leptin secretion. To examine the possibility that vitamin D dose alters leptin signaling to regulate the proportion of muscle mass and fat mass, we measured serum leptin in all three diet groups of wild-type mice. Normal-D increased serum leptin significantly relative to no-D (Fig. 4a, normal-D vs no-D). High-D decreased leptin significantly relative to normal-D (Fig. 4a, high-D vs normal-D). To better understand how this complex pattern related to leptin production and sensitivity, we examined the relationship between serum leptin, fat mass, and vitamin D treatment group using linear regression. Within a vitamin D treatment group, total fat mass determined serum leptin (Fig. 4b, for no-D r = 0.98 (F < 0.01 for slope being significantly different than zero), for normal-D r = 0.96 (F < 0.01 for slope being significantly different than zero) and for no-D r = 0.85 (F < 0.05, for slope being significantly different than zero)). However the slope of this relationship was different between groups (* p < 0.05 for normal-D and high-D vs no-D by ANOVA with Tukey post-tests). This graph reveals how the relationship between leptin and fat mass shifted significantly and meaningfully between groups (Graph in Fig. 4b, schematic in Fig. 4c). The slopes in this graph (Fig. 4b) are representative of leptin produced per fat mass. Raising vitamin D from low to normal significantly increased the slope of the curve for leptin vs fat mass (red arrow in Fig. 4b, red lettering in Fig. 4c): that is, normalizing vitamin D increased the amount of leptin produced per fat mass (red arrow in Fig. 4b). Raising vitamin D from normal-D to high-D did not further change the slope of this relationship. However, increasing vitamin D from normal-D vs high-D shifted the distribution on this line (blue arrow in Fig. 4b, blue lettering in Fig. 4c). That is, further raising vitamin D from normal to high did not alter the slope of the curve for leptin vs fat mass but instead shifted the distribution of fat mass down the curve. This is a shift consistent with high dose vitamin D increasing sensitivity to leptin.
High dose dietary vitamin D significantly increases energy expenditure without altering activity level or intake
Previously, another group reported that acute paraventricular hypothalamic injection of 1,25D (the active form of vitamin D) decreases appetite in mice1. They interpreted this result to indicate that 1,25D increased leptin sensitivity1. While some actions of vitamin D may be acute, most actions of vitamin D are mediated by slower transcriptional changes. Thus, it is not clear that this acute response represents a physiologic effect of vitamin D. Furthermore, leptin has actions on both appetite and energy expenditure, but leptin action is generally thought to alter energy expenditure more than appetite. We wanted to better understand the apparent increase in leptin sensitivity described above (Fig. 4) due to high-dose dietary vitamin D. To further examine this relationship, we measured intake and energy expenditure.
High dose dietary vitamin D had no effect on mass-adjusted intake (Fig. 5a and Fig. 5b, unadjusted intake was also not significantly different between groups). High dose dietary vitamin D had no effect on activity level (data not shown). However, high dose dietary vitamin D significantly increased fat free mass adjusted energy expenditure (Fig. 5c and 5d, * p < 0.05 by t-test). Raw-energy expenditure (unadjusted for fat free mass) was also significantly different between groups (* p < 0.05 by t-test, data not shown)) without altering activity level. Thus high-dose dietary vitamin D signals to differentially allocate calories for use by muscle instead of storage as fat without altering weight by increasing leptin secretion and leptin sensitivity.
Dietary vitamin D rescues growth factor deficit in the clinic
We replenished dietary vitamin D for 2 months (50000 IU weekly for 4 weeks and 4000 IU/d subsequently), and we repeated these labs (Table 1). As we expected, repleting vitamin D normalized serum 25(OH)D and calcium. Surprisingly, repleting vitamin D improved IGF-1 to just above the middle of the normal range for age (z-score 0.05). The initially normal prealbumin and albumin indicated that this young adolescent had appropriate overall energy and protein nutrition and that her low IGF-1 at that time was not related to calorie or protein-calorie availability. In the context of our other data supporting the model that vitamin D conveys energy stores centrally to modulate calorie allocation (Fig. 5e), this result, novelly, suggests that vitamin D also conveys nutrient availability to enable growth.
High dose dietary vitamin D significantly increases linear growth in mice
Our results in mice to this point suggested to us a model where myostatin was not simply playing a homeostatic role at muscle alone, but was also conveying nutrient needs centrally (Fig. 1b for overall model). That is, myostatin conveys not how much energy is being used but what the energy needs are likely to be. If we use the metaphor of the muscles as an engine then myostatin conveys not just how much gas the engine is consuming minute to minute, but how large the engine is. This model would predict that the high-D mediated inhibition of myostatin and increase in leptin signaling would not only increase energy expenditure but would also facilitate increased growth. To examine this possibility, we measured length in anesthetized mice from our diet groups. As we hypothesized based on this model, high-D increased nose-to-tail and nose-to-rump length in mice (Fig. 6b-c; p < 0.05 for each by t-test). This result is consistent with vitamin D facilitating increased central energy sensing across modalities (e.g. for allocation to muscle as well as for use in linear growth).
A genetic predisposition to higher Vitamin D increases final height in humans
Based on our results in mice, and the interesting result we observed in clinic in an adolescent with early rickets, we theorized that vitamin D might play a similar role in humans to facilitate energy sensing. To examine the possibility that vitamin D might modulate height by regulating calorie allocation in a clinically significant way we wanted to look whether vitamin D influenced height in humans. There are a variety of recent studies that attempt to address this question18,19. Interpretation of this work is bedeviled by issues around selection of treatment group (age, vitamin D status, season, etc) as well as duration and dose of treatment. One approach that we have used in other contexts to avoid these issues is Mendelian randomization20. We used the most recent genome wide association study (GWAS) to generate instrumental variables for serum 25(OH)D21. We applied these results to the most recent and comprehensive GWAS of height22. Using this approach we found that our instrumental variables identified significant positive relationships between SNPs that are associated with increased vitamin D and height; (Fig. 6d: Manousakis beta = 0.19, P = 2.09E-3). This result supports the notion that 25(OH)D may influence final height via conveying energy status.
Raising vitamin D from normal to high-normal in zebrafish embryos increases length
Based on our intriguing results in mice and in humans, we wondered if the role vitamin D plays to convey nutrient status reflects that calories are more plentiful in late summer and fall when vitamin D is generated and stored. This model would predict that the role of vitamin D to convey nutrient status would be evolutionarily conserved in non-mammalian species. To address this hypothesis, we measured growth in zebrafish over the first five days of life (schematic in Fig. 7a) when dosed with normal (25 ng/mL – four typical 5 day old larvae in Fig. 7b) compared with high-normal (50 ng/mL - four typical 5 day old zygotes in Fig. 7c) concentrations of 25(OH)D. We found that high-normal 25(OH)D significantly increased length in 5 day old zebrafish relative to normal (Fig. 7d: 50 ng/mL vs 25 ng/mL **** p < 0.0001 by unpaired t-test, n = 20 for each treatment). Other researchers have previously described that developmental vitamin D deficiency increases fat storage and decreases leanness and length relative to normal vitamin D in zebrafish at older time points. Similar to our results in mice, this result suggests that there are metabolic and growth benefits to raising vitamin D to high-normal relative to low-normal. Overall, these results provide support for the model that 25(OH)D is significantly associated with growth via conveying energy status.