Vitamin D protects against immobilization-induced muscle atrophy via neural crest-derived cells in mice

Vitamin D deficiency is a recognized risk factor for sarcopenia development, but mechanisms underlying this outcome are unclear. Here, we show that low vitamin D status worsens immobilization-induced muscle atrophy in mice. Mice globally lacking vitamin D receptor (VDR) exhibited more severe muscle atrophy following limb immobilization than controls. Moreover, immobilization-induced muscle atrophy was worse in neural crest-specific than in skeletal muscle-specific VDR-deficient mice. Tnfα expression was significantly higher in immobilized muscle of VDR-deficient relative to control mice, and was significantly elevated in neural crest-specific but not muscle-specific VDR-deficient mice. Furthermore, muscle atrophy induced by limb immobilization in low vitamin D mice was significantly inhibited in Tnfα-deficient mice. We conclude that vitamin D antagonizes immobilization-induced muscle atrophy via VDR expressed in neural crest-derived cells.


Results
Vitamin D deficiency worsens immobilization-induced muscle atrophy. To establish vitamin D-deficient conditions, we fed 6-week-old wild-type (WT) mice a low (L) or standard (S) vitamin D diet for 4 weeks (Fig. 1) and then monitored serum 25(OH)D levels (Fig. 1a). Mice fed the L diet showed significantly reduced serum 25(OH)D levels compared with mice fed the standard vitamin D diet (Fig. 1a). We then created an immobilization-induced muscle atrophy model in lower extremities by stapling of hind limbs, as previously described 28 . Specifically, at 6-weeks of age, we began feeding WT mice the L or S diet, three weeks later immobilized the hind limb, and a week after that sacrificed animals and analyzed muscle volume in gastrocnemius and quadriceps. Volume of both muscles decreased following immobilization, and that effect was significantly more severe in mice fed a low (L4) rather than standard (S4) diet (Fig. 1b, c). Histological analysis showed that cross sectional area (CSA) of gastrocnemius muscle in mice fed the L diet was significantly smaller than in mice fed the S diet ( Fig. 1d-f). Moreover, expression of the atrogenes Atrogin-1 and MuRF1, both E3 ubiquitin ligases, was significantly higher in gastrocnemius muscle of L versus S mice (Fig. 1g, h). Increased levels of Smad2 and Smad3 proteins, upstreams of the atrogenes, are reportedly required for immobilization-induced muscle atrophy 28 . Indeed, Smad2/3 protein expression increased following immobilization in gastrocnemius muscle compared with non-stapled side ( Fig. 1i-m).
Refeeding a standard diet or administration of a vitamin D agonist rescues muscle atrophy phenotypes in limb-immobilized vitamin D-deficient mice. We next varied the protocol used to establish vitamin D-deficient mice by feeding 6-week-old WT mice either the S diet for 4 or 6 weeks (S4 or S6) or the L diet for 2 weeks followed by the S diet for either 2 or 4 more weeks (L2S2 or L2S4) (Figs. S1a, b and 2a-e) and then monitored serum 25(OH)D levels. Serum 25(OH)D levels of L2S2 group were equivalent to S4 group (Fig. 2a). On the other hand, when we applied this protocol and immobilized the hind limb at week 9 (S4 or L2S2) (Fig. S1a) or 11 (S6 or L2S4) (Fig. S1b), we observed rescue of immobilization-induced phenotypes in both muscles based on muscle weight in the L2S4 but not the L2S2 group ( Fig. 2b-e). Moreover, administration of the vitamin D analogue, eldecalcitol, at 8-weeks of age did not alter volume of either muscle in an immobilization model in mice fed the S diet only (Figs. S1c and 2f and g). However, we observed significant rescue of immobilization-induced muscle atrophy in gastrocnemius and quadriceps of 10-week old mice that had been administered eldecalcitol for two weeks and fed the L diet starting at 6 weeks of age ( Fig. 2h and i). CSA in gastrocnemius muscle of eldecalcitol-treated mice was also significantly larger than that in vehicle-treated mice ( Fig. 2j-l). Moreover, higher Atrogin-1 and MuRF1 expression in immobilized gastrocnemius was significantly inhibited by eldecalcitol treatment for 2 weeks (Fig. 2m and n). Similarly, elevated pSmad2/3 levels seen in immobilized gastrocnemius were attenuated by eldecalcitol treatment (Fig. 2o-s). Finally, eldecalcitol treatment for 2 weeks significantly blocked upregulation of tumor necrosis factor alpha (Tnfα) promoted by immobilization ( Fig. 2t-v).
VDR-deficient mice exhibit muscle atrophy without immobilization. We next analyzed muscle atrophy in global Vdr knockout (Vdr KO) and WT mice, and immobilized the hind limb at week 9 (Fig. 3). VDR deficient mice were smaller than WT mice at 10 weeks of age when the mice were sacrificed (Fig. 3a, b), and muscle weight per body weight of both gastrocnemius and quadriceps in the non-stapled side was significantly lower in VDR KO relative to WT mice (Fig. 3c, d). Grip power was also significantly lower in VDR KO relative to WT mice (Fig. 3e).
Immobilization-induced atrophy of both gastrocnemius and quadriceps muscle was comparable in VDR KO and WT mice (Fig. 3c, d). However, Atrogin-1 and MuRF1 expression following immobilization was significantly higher in gastrocnemius of VDR KO compared to WT mice (Fig. 3f, g). Furthermore, induction of Tnfα or interleukin-1 beta (IL-1β) in gastrocnemius muscle following immobilization was significantly higher in VDR KO compared to WT mice ( Fig. 3h- 32 . To determine which cells are critical in regulating immobilization-induced muscle atrophy via VDR, we established 2 lines of conditional VDR knockout mice. In one, we crossed VDR flox/flox mice with neural crest cell-specific P0 Cre mice to yield P0 Cre;VDR flox/flox mice (or, sVDR cKO). In the other, we established skeletal musclespecific conditional VDR KO mice using the muscle creatine kinase promoter to generate Ckmm Cre; VDR flox/flox mice (or, mVDR cKO). Immunohistochemical analysis of both lines revealed loss of VDR protein expression in P0-positive Schwann cells (Fig. S2a) and laminin-positive muscle fibers (Fig. S2b) in respective sVDR cKO and mVDR cKO mice fed an S diet. Then, when mice reached 9 weeks of age, we applied the immobilization protocol to both sVDR and mVDR cKO animals and 1 week later evaluated muscle phenotypes in cKO and corresponding control mice. Immobilization-induced atrophy in both gastrocnemius and quadriceps was significantly more severe in sVDR cKO than in control mice (VDR flox/flox ) fed the S diet ( Fig. 4a, b). Histological and CSA analysis confirmed that immobilization-induced atrophy was significantly more severe in sVDR cKO than in control mice ( Fig. 4c-e). By contrast, immobilization-induced gastrocnemius and quadriceps atrophy was comparable Frequency distribution of fiber area of gastrocnemius muscles in the stapled side of S4 and L4 groups (x axis, fiber area; y axis, % of cross-sectional area (CSA) of muscle fiber; data represent mean % CSA ± SD). (f) Relative mean cross-sectional areas (CSA) of gastrocnemius muscles on control and stapled sides of S4 and L4 groups.
(g-h) Relative Atrogin-1 (g) and MuRF1 (h) expression in control and stapled gastrocnemius muscles of S4 and L4 groups based on quantitative realtime PCR. (i) Western blot of Smad2 and 3 protein in control and stapled gastrocnemius muscles of S4 and L4 groups. Representative images are shown. (j-m) Quantitation of levels of phosphorylated Smad2 (j), total Smad2 (k), phosphorylated Smad3 (l) and total Smad3 (m) per Gapdh shown as means ± SD relative to control side of S4 and L4 groups (S4, n = 3; L4, n = 3). (b, c and f) Means ± SD relative to control side of S4 group are shown. (g and h) Shown is mean indicated expression relative to Gapdh ± SD relative to control side of S4 group. Statistical analysis was done by Student's t-test (*P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant).
Scientific RepoRtS | (2020) 10:12242 | https://doi.org/10.1038/s41598-020-69021-y www.nature.com/scientificreports/ Figure 2. Refeeding a normal vitamin D diet or treatment with a vitamin D analogue rescues atrophy phenotypes following muscle immobilization. (a-c) 6-week-old female WT mice were fed the L diet for 2 weeks and then the S diet for 2 weeks (L2S2 group), or fed the S diet for 4 weeks (S4 group). Hind limbs were stapled at 9-weeks of age during the diet regime. Mice were sacrificed 1 week after stapling, and Sera were collected (Each group n = 5). (a) Serum 25(OH)D levels in indicated groups were analyzed by radioimmunoassay. (b, c) Weights of gastrocnemius (b) and quadriceps (c) muscles adjusted to body weight relative to those of control sides of the S4 and L2S2 groups. (d, e) 6-week-old female WT mice were fed the L diet for 2 weeks and then the S diet for 4 weeks (L2S4 group), or fed the S diet for 6 weeks (S6 group). Left hind limbs were stapled at 11-weeks of age and mice were sacrificed 1 week after stapling (each group n = 5). Wet weights of gastrocnemius (d) and quadriceps (e) muscles adjusted to body weight of S6 and L2S4 groups relative to control sides of the S6 group. (f-v) 6-week-old female WT mice were fed the S (f and g) or L (h-v) diet, and treated with 3.5 ng ED71 (ED group) or vehicle (Veh group) twice per week starting at 8 weeks of age. Left hind limbs stapled at 9 weeks of age and mice were sacrificed a week later (each group n = 5). www.nature.com/scientificreports/ in mVDR cKO and corresponding control mice (Fig. 4f-j). Atrogin-1 and MuRF1 expression in immobilized gastrocnemius was also significantly higher in sVDR cKO than in control mice (Fig. 4k, l), but those levels were comparable in mVDR and control mice (Fig. 4m, n). Neural crest-derived cells give rise to various tissue types, among them adrenal glands, which regulate muscle homeostasis 33 . Adrenal grands produce the glucocorticoid cortisol, which promotes muscle atrophy 34 . However, serum cortisol levels were comparable in sVDR cKO and control mice (Fig. S3a). Neural crest cells also contribute to formation of the parathyroid gland 35 ; however, serum levels of parathyroid hormone were equivalent in sVDR cKO and control mice (Fig. S3b). Moreover, ED71 treatment did not rescue immobilization-induced muscle atrophy in either gastrocnemius or quadriceps muscle of sVDR mice (Fig. S4). These results suggest that VDR expressed in neural crest-derived cells, antagonizes immobilization-induced muscle atrophy in the presence of active vitamin D analogues. Indeed, when we established an immobilization model in WT mice that had been subjected to denervation of gastrocnemius muscle by cutting the sciatic nerve at 9 weeks of age, we observed comparable levels of muscle atrophy and atrogene expression in model mice fed the L or S diets (Fig. 5a-c) or in vitamin D-deficient mice treated with ED71 or vehicle (Fig. 5d-f). These results suggest overall that sciatic nerve loss impairs the ability of vitamin D to antagonize immobilization-induced muscle atrophy, likely via effects on Schwann cells. mVDR or sVDR cKO mice fed the S diet were applied the immobilization protocol at nine weeks of age, and one week later, expression of the inflammatory cytokines Tnfα and IL-1β in immobilized gastrocnemius muscle was analyzed (Fig. 6a-f). We found that that expression was significantly higher in sVDR cKO than control mice (Fig. 6a-c) but differences in these levels were not significant in mVDR cKO versus control muscle (Fig. 6d-f). Finally, Tnfα −/− (TNFα KO) mice fed the L diet from 6 weeks of age were subjected to limb immobilization at 9 weeks old. One week later, Tnfα KO mice showed a partial but significant rescue of atrophy of gastrocnemius or quadriceps muscle (Fig. 6g, h).

Discussion
Vitamin D plays diverse roles in calcium, bone and muscle homeostasis as well as in IGF-1 induction 36 . Most IGF-1 is secreted from the liver, and serum 25 (OH) D levels correlate positively with serum IGF-1 levels 37 . Vitamin D also stimulates local IGF-1 production 38 , and indeed, here we show induction of IGF-1 expression in sciatic nerve from control (VDR flox/flox ) mice, an effect absent in sVDR cKO mice (Fig. S5). We previously showed that IGF-1 treatment suppressed Smad2/3 activation in muscle cells in vitro 28 . Others have reported that IGF-1 suppresses inflammatory cytokine expression in liver and in atherosclerotic plaque of aortic sinuses 39,40 . Thus decreased local IGF-1 levels due to VDR loss in neural crest-derived cells likely promote local inflammation, worsening immobilization-induced muscle atrophy in sVDR cKO mice. In mammals, synthesis of most circulating vitamin D is stimulated in skin by sun exposure, but some is also consumed in the diet 41,42 . Elderly people tend to spend longer periods indoors, limiting sun exposure and potentially leading to vitamin D deficiency. Accordingly, vitamin D deficiency is frequently seen in the elderly 43,44 . However, young people are equally vulnerable to low vitamin D status [45][46][47] , making vitamin D deficiency a general concern.
Here, we use a mouse model to show that muscle atrophy seen following limb immobilization is more severe under vitamin D-deficient conditions (Fig. 7). Our findings overall suggest that immobilization promotes local inflammatory cytokine expression in skeletal muscles via neural crest-derived cells under vitamin D deficient conditions, leading to severe muscle atrophy (Fig. 7). Our findings suggest that maintaining appropriate vitamin D levels is crucial to protect muscle from significant atrophy.
Various biological function of vitamin D have been reported 36 , but overall, reports of manifestations of vitamin D or VDR deficiency in humans are limited development of rickets or alopecia. VDR-deficient mice die after weaning, a lethality reportedly rescued by feeding a high calcium diet 13,48 suggesting that vitamin D-VDR signaling is required to maintain calcium homeostasis. However, the impact of vitamin D activity on preventing falls is not as striking as its impact on inhibiting rickets development or regulating calcium homeostasis. Thus, prevention of falls is likely provided by a combination of several minor biological functions of vitamin D.
Falls in the elderly frequently occur due to impaired balance, and administration of an active vitamin D analogue reportedly improves balance in the elderly [49][50][51] . Indeed, administration of an active vitamin D analogue to osteoporosis patients was shown to significantly reduce the frequency of forearm fractures 52,53 , most of which occur by falls. Myoblasts and myotubes are known to express VDR and CYP27B1 (also called 1α-hydroxylase), which converts 25(OH)D to the active vitamin D3, 1,25(OH) 2 D 3 24,54,55 . These results suggest that active vitamin D analogues directly act on muscle cells via VDR, or that muscle cells can convert inactive 25(OH)D to active 1,25(OH) 2 D 3 by CYP27B1 activity in an autocrine or paracrine manner. Indeed, an active vitamin D analogue reportedly promotes muscle cell differentiation 56 . VDR signaling also functions in regulating neuromuscular maintenance and enhances locomotor ability 38 . However, our current study strongly suggests that muscle atrophy due to vitamin D deficiency has its origins in neural crest-derived rather than skeletal muscle cells.
Vitamin D promotes expression of IGF-1 57 , which acts as an anabolic factor and an inhibitor of catabolic signals in muscles 28,58,59 . Thus, IGF-1 plays a pivotal role in maintaining muscle homeostasis. Serum IGF-1 levels reportedly decrease with age in humans 60 , and reducing IGF-1 levels in adult mice promotes muscle atrophy and reduces muscle power, phenotypes seen in sarcopenia patients 61 . Peripheral nerve dysfunction also causes falls 62 , and proper function of peripheral nerves is determined by neurons and Schwann cells, which are required for peripheral nerve myelination. Schwann cells are known targets of vitamin D, and IGF-1 expression is stimulated in Schwann cells by vitamin D 21 . Our data suggests that vitamin D inhibits immobilization-induced muscle atrophy in part by inhibiting local inflammation and that vitamin D may prevent falls by a combination of these functions.
Vitamin D levels are regulated by various pathways. Vitamin D whether synthesized in skin or taken through the diet is converted into 25(OH)D in liver and stored 63 , g and j) or Vdr flox/flox mice relative to those in control side of Vdr flox/flox mice. (k-n) Mean expression relative to Gapdh ± SD of sVDR cKO (k and l), mVDR cKO (m, n) or Vdr flox/flox mice relative to Vdr flox/flox mice is shown. Statistical analysis was done by Student's t-test (*P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant). (a, b, f, g, m and n 66 . Overall, either loss of or excessive activation of these axes deregulates serum vitamin D levels.
In summary, we show that low vitamin D status worsens muscle atrophy. The increasing number of sarcopenia patients and individuals showing low vitamin D status is a global concern, and these conditions are frequently associated with each other. Our data suggests that maintaining vitamin D status at an appropriate level before injury or decline in physical activity is likely crucial to prevent deterioration and muscle atrophy.

Materials and methods
Mice. C57BL/6 (WT) mice were purchased from Sankyo laboratory.   Atrogin-1 (b, e) and MuRF1 (c and f) in control (sham) and denervated gastrocnemius muscles of S4 or L4 group with vehicle injection (b, c), and L4 with vehicle or ED71 injection (e, f). Mean ± SD relative to sham side of S4 (a-c) or Veh (d-f) group are shown. Statistical analysis was done by Student's t-test (*P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant).  (a-f) sVDR (neural-crest-specific) or mVDR (skeletal-muscle-specific) KO or control (Vdr flox/flox ) female mice were fed the S diet, their hind limbs stapled at 9 weeks of age, and mice were sacrificed 1 week after stapling for analysis of expression of Tnfa (a, d), Il-b (b, e) and Il-6 (c, f) relative to Gapdh in gastrocnemius muscles of indicated genotypes. (g, h) 6-week-old female TNFα KO (Tnfα -/-) or WT mice were fed the L diet and subjected to the same protocol as described above (each group n = 5). Wet weights of gastrocnemius (g) and quadriceps (h) muscles adjusted to body weight in control and stapled sides of TNFα KO and WT mice were determined relative to values on the WT control side. (a-h) Mean indicated parameters in sVDR cKO, mVDR cKO, Vdr flox/flox (a-f), TNFα KO or WT (g and h) mice ± SD relative to those on the stapled side of Vdr flox/flox (a-f) of WT (g and h) mice are shown. Statistical analysis was done by Student's t-test (*P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant).