Vitamin D and diabetes
- First Online:
- Cite this article as:
- Mathieu, C., Gysemans, C., Giulietti, A. et al. Diabetologia (2005) 48: 1247. doi:10.1007/s00125-005-1802-7
- 7k Downloads
Vitamin D deficiency predisposes individuals to type 1 and type 2 diabetes, and receptors for its activated form—1α,25-dihydroxyvitamin D3—have been identified in both beta cells and immune cells. Vitamin D deficiency has been shown to impair insulin synthesis and secretion in humans and in animal models of diabetes, suggesting a role in the development of type 2 diabetes. Furthermore, epidemiological studies suggest a link between vitamin D deficiency in early life and the later onset of type 1 diabetes. In some populations, type 1 diabetes is associated with certain polymorphisms within the vitamin D receptor gene. In studies in nonobese diabetic mice, pharmacological doses of 1α,25-dihydroxyvitamin D3, or its structural analogues, have been shown to delay the onset of diabetes, mainly through immune modulation. Vitamin D deficiency may, therefore, be involved in the pathogenesis of both forms of diabetes, and a better understanding of the mechanisms involved could lead to the development of preventive strategies.
Keywords1,25(OH)2D3 Autoimmunity Beta cell Diabetes NOD mouse Prevention Vitamin D Vitamin D deficiency Vitamin D receptor polymorphism
nuclear factor of activated T cell
retinoid X receptor
vitamin D receptor
vitamin D response element
The discovery of receptors for 1α,25-dihydroxyvitamin D3 (1,25(OH)2D3), the activated form of vitamin D, in tissues with no direct role in calcium and bone metabolism (e.g. pancreatic beta cells and cells of the immune system) has broadened our view of the physiological role of this molecule [1, 2]. An increased prevalence of type 2 diabetes has been described in vitamin D-deficient individuals [3, 4, 5], and insulin synthesis and secretion have been shown to be impaired in beta cells from vitamin D-deficient animals. Glucose tolerance is restored when vitamin D levels return to normal.
The identification of receptors for 1,25(OH)2D3 in cells of the immune system led to experiments in animal models of type 1 diabetes in which the administration of high doses of 1,25(OH)2D3 was shown to prevent type 1 diabetes [3, 6], mainly through immune regulation. It has been demonstrated that 1,25(OH)2D3 is one of the most powerful blockers of dendritic cell differentiation and that it directly blocks IL-12 secretion . Lymphocyte proliferation is inhibited and regulator cell development is enhanced .
This review provides an overview of the data available on the role of vitamin D in type 1 and type 2 diabetes, and discusses possible applications of the molecule or its synthetic analogues [9, 10] in clinical disease. The terminology used in many papers to describe vitamin D and its metabolites is confusing, with misnomers leading to misunderstanding and over-interpretation of data. In this review the term vitamin D refers to the product that is in food (vitamins D2 and D3) and is synthesised in the skin under the influence of UVB radiation (vitamin D3), whereas the metabolically active molecule is referred to as 1,25(OH)2D3.
Vitamin D and its metabolism
In the 17th century, Whistler (in 1645) and Glisson (in 1650) described rickets as a bone disease of young children (for review see ), but it was only in 1923 that Goldblatt and Soames identified and characterised vitamin D . Most vertebrates synthesise vitamin D in their skin under the influence of UV light [1, 13]. An ‘efficient’ sun exposure—exposure of the face and hands to the sun for 2 h/week—is probably sufficient to maintain normal levels. Food supplementation is required during pregnancy and lactation and for newborns and young children (especially in dark-skinned children living in northern countries). Vitamin D can be obtained from dietary sources of vegetable (vitamin D2, also known as ergocalciferol) or animal origin (vitamin D3, also known as cholecalciferol). The best food sources are fatty fish or their liver oils; however, small amounts are also found in butter, cream and egg yolk. Human and cows’ milk are poor sources of vitamin D. In many parts of the world, especially North America, fluid and dried milk, as well as some margarines, butter and cereals are supplemented with vitamin D. However, the real vitamin D content is frequently quite different from the labelling standard and often insufficient to reach the daily requirements for vitamin D (400–600 IU/day). Skimmed milk, in particular, frequently has no detectable vitamin D.
The proximal renal tubule is the principal site of 1α-hydroxylation, although high levels of 1α-hydroxylase mRNA have also been found in human keratinocytes , dendritic cells  and macrophages . This extra-renal production of 1,25(OH)2D3 is regulated in a completely different manner. For example, production in macrophages is resistant to stimulation by parathyroid hormone but may be directly stimulated by immune stimuli such as IFN-γ and lipopolysaccharide .
Another hydroxylation enzyme, 24-hydroxylase, initiates the catabolic cascade of 25-hydroxyvitamin D3 and 1,25(OH)2D3 . In the circulation, all metabolites of vitamin D are bound to a carrier protein known as vitamin D-binding protein .
Molecular action of 1α,25-dihydroxyvitamin D3
Only the 1,25(OH)2D3 form of vitamin D is metabolically active, and this molecule exerts its effects by activating the nuclear vitamin D receptor (VDR). The VDR is a member of the nuclear receptor super family of ligand-activated transcription factors, which also includes the thyroid hormone receptor, the retinoic acid receptor and the peroxisome proliferator-activated receptor. The structure of the VDR has only very recently been described and the protein is yet to be fully characterised . In humans the gene encoding the VDR is located on chromosome 12cen–q12 and shows extensive polymorphism, including a mononucleotide (A)n repeat polymorphism in the 3′ untranslated region and four RFLPs: FokI within exon 2, BsmI and ApaI within successive introns between exon 7 and exon 9, and TaqI within exon 9 (for review see ).
The effects of 1,25(OH)2D3 on tissues other than calcium and bone can only be observed at concentrations (10−10 mol/l) that exceed the physiological levels needed for maintenance of calcium and bone homeostasis by a factor of 100–1,000. The fact that cells such as macrophages can secrete 1,25(OH)2D3 suggests that such high concentrations of 1,25(OH)2D3 may be attained locally within specific target areas (e.g. sites of inflammation).
The administration of 1,25(OH)2D3 in humans to exploit its ‘non-calcaemic’ effects is not realistic, since the doses needed to reach the high levels locally within target organs would lead to severe systemic hypercalcaemia and other (bone) side effects. Structural analogues containing side chain modifications or, more recently, CD-ring modifications have been synthesised that share the ‘non-classical’ effects of 1,25(OH)2D3 but have lesser effects on calcium and bone [25, 26, 27]. At present, more than 2,000 analogues have been patented, and some analogues have already reached the clinic (e.g. calcipotriol for psoriasis). A better understanding of the differential effects of these analogues on various tissues could permit targeted drug design and the development of analogues with an even greater dissociation of calcaemic and non-calcaemic effects.
Vitamin D and type 2 diabetes
Vitamin D deficiency was linked to IGT and type 2 diabetes in humans many years ago [5, 28]. These observations were confirmed in animal models, which demonstrated that pancreatic insulin secretion is inhibited by vitamin D deficiency . Several reports have ascribed an active role to vitamin D in the functional regulation of the endocrine pancreas, particularly the beta cells. Not only are receptors for 1,25(OH)2D3 found in beta cells , but the effector part of the vitamin D pathway is also present in the form of vitamin D-dependent calcium-binding protein, also known as calbindin-D28k . The expression of calbindin-D28K has been shown to protect beta cells from cytokine-mediated cell death .
Several studies have demonstrated a link between VDR gene polymorphisms and type 2 diabetes, although the findings differ from one population to another. A study in Bangladeshi Asians demonstrated that the ApaI RFLP influences insulin secretion in response to glucose , while associations between the VDR ApaI RFLP and higher fasting plasma glucose levels and glucose intolerance were observed in a community-based study of older adults without known diabetes . More recently, genotyping for TaqI, ApaI, BsmI and FokI RFLPs revealed that the BsmI RFLP is associated with high fasting glucose levels in young males with low physical activity .
Effects of vitamin D and its metabolites on insulin synthesis and secretion
Vitamin D deficiency leads to the impaired secretion of insulin—but not other islet hormones—in both animal models and humans, and induces glucose intolerance [29, 36, 37], while replenishment with vitamin D rectifies the abnormalities [38, 39, 40]. This impairment is primarily caused by the direct effect of vitamin D deficiency on the beta cell, but other effects of vitamin D deficiency, such as impaired food intake and hypocalcaemia, may also play a role. Data from VDR knock-out mice are conflicting, with some groups reporting IGT  and others reporting no impairment in glucose metabolism . In these studies the genetic background of the mouse in which the VDR knock-out is introduced seems to be of critical importance.
More convincing support for a beneficial effect of 1,25(OH)2D3 on beta cell function has emerged from studies on islets isolated from normal animals in which insulin synthesis and release were provoked by glucose challenge in the presence of high doses of 1,25(OH)2D3 .
The mechanism by which 1,25(OH)2D3 might act on insulin secretion is suggested by the significant rise in cytosolic Ca2+ levels observed following 1,25(OH)2D3-stimulated secretion of insulin by islet cells. Controversy remains as to whether an influx of external Ca2+ via voltage-dependent Ca2+ channels is solely responsible for this rise, or whether the mobilisation of Ca2+ from intracellular organelles and the activation of release-potentiating systems via protein kinase C and protein kinase A pathways are also involved [44, 45, 46].
Interventions with vitamin D and its metabolites in vivo: clinical implications for type 2 diabetes
The restoration of vitamin D reserves in vitamin D-deficient patients has been shown to improve glucose tolerance . The amount of vitamin D administered to these patients with rickets was, however, subject to wide variation (from the daily oral administration of 2,000 IU  to a single intramuscular injection of 100,000 IU ), making it difficult to draw clear conclusions and establish guidelines. Nonetheless, it is clear that vitamin D deficiency should be treated, and that improved glucose tolerance is a desirable side effect. In contrast, repletion of 1,25(OH)2D3 in the relatively 1,25(OH)2D3-deficient state of uraemia only partially reverses glucose intolerance, probably because insulin resistance is also involved in this condition .
Vitamin D in the aetiology of type 2 diabetes: published studies in humans
Type of subjects studied (age at study)
Vitamin D intake via supplementation 2,000 IU/day (50 μg) for 1 month
Vitamin D-deficient women (adulthood)
Improved glucose tolerance and improved beta cell function
Vitamin D intake via supplementation 2,000 IU/day for 6 months
Vitamin D-deficient subjects (adulthood)
Improved insulin secretion
Vitamin D intake via single i.m. injection 100,000 IU/day
Vitamin D-deficient subjects (adulthood)
Improved insulin and C-peptide responses
1,25(OH)2D3 treatment 0.5 μg/day for 21 days (or +500 mg Ca2+)
Uraemic women (adulthood)
Improved first-phase insulin secretion and insulin sensitivity
1,25(OH)2D3 treatment 1 μg/day for 4 days
Type 2 diabetic women (adulthood)
Improved insulin and C-peptide responses to Sustacal (Mead Johnson, Evansville, IN, USA)
Hypovitaminosis (<5 ng/ml)
Type 2 diabetic women (adulthood)
Decreased 25-hydroxyvitamin D3 levels decreased beta cell function
Vitamin D intake via supplementation 1,332 IU/day for 1 month
Type 2 diabetic women (adulthood)
Improved first-phase insulin secretion
Vitamin D intake via single i.m. injection 300,000 IU/day
Type 2 diabetic men and women (adulthood)
Increased insulin resistance
Vitamin D and type 1 diabetes
Several epidemiological studies have described an intriguing correlation between geographical latitude and the incidence of type 1 diabetes, and an inverse correlation between monthly hours of sunshine and the incidence of diabetes. A seasonal pattern of disease onset has also been described for type 1 diabetes , once again suggesting an inverse correlation between sunlight and the disease. Vitamin D is an obvious candidate as a mediator of this sunshine effect.
Vitamin D in the aetiology of type 1 diabetes: published studies in humans
Type of subjects studied (age at study)
Vitamin D intake via supplementation for the first year of life
Infancy (<5 years)
Reduced risk of childhood-onset type 1 diabetes (OR=0.83)
Childhood (5–9 years)
Reduced risk of childhood-onset type 1 diabetes (OR=0.81)
Childhood (10–14 years)
Reduced risk of childhood-onset type 1 diabetes (OR=0.47)
Vitamin D intake via use of cod-liver oil 10 μg for ≥5×/week for the first year of life
Childhood (<15 years)
Reduced risk of childhood-onset type 1 diabetes (OR=0.74)
Vitamin D intake via supplementation ≥2,000 IU/day during the first year of life
Infancy till early adulthood (1–31 years)
Reduced risk of childhood-onset type 1 diabetes (OR=0.12)
Rickets during the first of life
Infancy till early adulthood (1–31 years)
Increased risk of childhood-onset type 1 diabetes (OR=2.6)
Vitamin D intake via food continuous variable IU during pregnancy
Infancy (<5 years)
Reduced risk of insulin auto-antibodies in offspring (OR=0.49)
It remains to be determined whether these observations are the result of supplementation of vitamin D to supraphysiological levels, or are simply the result of the prevention of vitamin D deficiency. Observations in animal models suggest the latter, since regular supplements of vitamin D in neonatal and early life offered no protection against type 1 diabetes in nonobese diabetic (NOD) mice or in BioBreeding (BB) rats [56, 57], whereas the prevalence of diabetes is doubled in NOD mice rendered vitamin D-deficient in early life [58, 59]. The results of genetic studies investigating a possible relationship between VDR polymorphisms and type 1 diabetes are highly confusing: a clear correlation exist in some populations [60, 61, 62, 63, 64, 65], whereas no correlation can be found in others .
Immunomodulatory effects of 1α,25-dihydroxyvitamin D3 and its analogues
The identification of VDRs on almost all cells of the immune system, especially antigen-presenting cells (macrophages and dendritic cells) and activated T lymphocytes, prompted the investigation of 1,25(OH)2D3 as a potential immunomodulator (for reviews see [57, 67]). Immune cells—activated macrophages and dendritic cells in particular—contain the enzyme 1α-hydroxylase, which is necessary for the final activating step in the conversion of vitamin D3 to the metabolically active molecule, and are therefore able to synthesise and secrete 1,25(OH)2D3 [17, 68]. The 1α-hydroxylase present in immune cells is identical to the renal enzyme, but regulation of its expression and activity is different. Whereas the renal enzyme is principally under the control of calcaemic and bone signals (such as parathyroid hormone and 1,25(OH)2D3 itself), the macrophage enzyme is primarily regulated by immune signals, with IFN-γ and Toll-like receptor agonists being powerful stimulators. It is of interest that a defect in the upregulation of 1α-hydroxylase in response to immune stimuli has been reported in NOD mice .
Several T cell cytokines are direct targets of 1,25(OH)2D3 and its analogues. The secretion of IL-2 is suppressed through the inhibition of nuclear factor of activated T cells (NFAT) complex formation by ligand-bound VDR–RXR heterodimers through binding to the distal NFAT binding site in the human IL-2 promoter . Another key T cell cytokine, IFN-γ, is downregulated by 1,25(OH)2D3 . The promoter region of the gene encoding this protein contains a classical VDRE (DR3 type), and binding of the ligand-bound VDR–RXR heterodimers causes direct suppression of transcription .
Taken together, these observations suggest a physiological role for 1,25(OH)2D3 in the immune system, with a tightly regulated secretion of 1,25(OH)2D3 by macrophages and dendritic cells upon immune stimulation on the one hand and a direct inhibitory effect of the molecule on antigen presentation and T cell proliferation and cytokine secretion on the other hand. These immune effects are typically mediated through the VDR. Therefore, the fact that VDR knock-out mice have no apparent immune abnormalities suggests the redundancy of 1,25(OH)2D3 as a signal in the immune system .
Beta cell protection by 1α,25-dihydroxyvitamin D3 and its analogues
Direct beta cell damage by cytokines and other inflammatory agents is an important step in the pathogenesis of type 1 diabetes. The inhibition of beta cell function (insulin synthesis and insulin secretion) induced by IL-1β or IFN-γ in vitro is prevented by 1,25(OH)2D3 and its analogues, MC903 and KH1060 [82, 83]. In contrast, Mauricio et al observed no effect of 1,25(OH)2D3 on IL-1β-induced beta cell dysfunction . This discrepancy may be due to the fact that an incubation time of 24 h was used in the latter study, whereas incubation periods ranging from 48 to 72 h were used in the studies describing protection. In addition to the alteration of the effects of cytokines on beta cell function, 1,25(OH)2D3 blocks the induction of surface markers by these cytokines [83, 85, 86]. The upregulation of expression of MHC-II molecules and intercellular adhesion molecule-1 on rat islets after exposure to IFN-γ was markedly decreased by co-incubation with 1,25(OH)2D3 or an analogue (ZXY1106) . More recently, Riachy et al demonstrated that 1,25(OH)2D3 might preserve the insulin content of human islets and prevent MHC-I expression, IL-6 production and NO release . Moreover, it was reported that 1,25(OH)2D3 induced and maintained high levels of A20, an anti-apoptotic protein, in rat RINm5F cells and human islets after exposure to inflammatory cytokines . We recently performed a comparative study of different cell systems (whole rat islets, FACS-purified beta cells and INS-1E cells) and did not observe direct protection by 1,25(OH)2D3 against cytokine-induced beta cell death, but demonstrated decreased expression of chemokines by beta cells treated with 1,25(OH)2D3 .
In summary, 1,25(OH)2D3 protects beta cells from cytokine-induced beta cell dysfunction but cannot protect them from direct cytokine-induced cell death (Fig. 3).
Effects of 1α,25-dihydroxyvitamin D3 and its analogues in animal models of type 1 diabetes
The chronic administration of pharmacological doses of 1,25(OH)2D3 reduces the incidence of both insulitis and diabetes in NOD mice [6, 88]. When 1,25(OH)2D3 was administered lifelong from weaning in NOD mice, the incidence of insulitis was reduced from 80% in control mice to 58% in 1,25(OH)2D3-treated mice, and the incidence of diabetes was reduced from 56 to 8%. These findings have been confirmed in other studies . Extensive immunological screening, including FACS analysis of major lymphocyte subsets, evaluation of T cell proliferation and natural killer cell function, did not reveal any major changes in the treated mice compared with the control mice . The reported restoration of deficient suppressor cell function of NOD mice in response to treatment with 1,25(OH)2D3 represents a major finding . Adorini et al demonstrated that the regulatory T cell induced by 1,25(OH)2D3 or its analogues is most likely a CD4CD25 cell . However, it remains to be determined whether this restoration of suppressor cells is the main mechanism involved in the protection afforded by 1,25(OH)2D3 against diabetes. Protection against insulitis was also observed, suggesting interference with the induction of autoimmunity itself. This is consistent with the observation that 1,25(OH)2D3 protects NOD mice against cyclophosphamide-induced diabetes .
The molecular basis of protection by 1,25(OH)2D3 seems to involve the reshaping of the immune repertoire, with elimination of effector cells, although the direct protective effects of 1,25(OH)2D3 on the beta cell might also play a role in disease prevention. It has been shown that there is a shift in the production of T cell cytokines from predominantly Th1 (IL-2, IFN-γ) in control mice to Th2 (IL-4, IL-10) in mice treated with 1,25(OH)2D3 or an analogue of the molecule . The shift appeared to be antigen specific and is probably caused by the direct interference of 1,25(OH)2D3 or its analogues by the antigen-presenting dendritic cells. Indeed, 1,25(OH)2D3 induces a reshaping of dendritic cells towards tolerogenic cells [69, 70, 77]. We have also demonstrated that dendritic cells generated in the presence of 1,25(OH)2D3 or an analogue can re-direct already committed T cell clones derived from a type 1 diabetic patient towards non-proliferation [8, 91]. In the NOD mouse, the reshaping of the immune system happens centrally, in the thymus, where treatment with 1,25(OH)2D3 restores the sensitivity of T lymphocytes towards apoptosis-inducing signals, thereby enhancing the elimination of autoimmune effector cells .
In marked contrast to the NOD mouse, no significant difference in the incidence of diabetes was observed between control BB rats (33%) and BB rats treated with 1,25(OH)2D3 from weaning up to 120 days (24%) . This finding confirms the basic differences in disease pathogenesis in the two available animal models for type 1 diabetes, and indicates that caution is warranted when transferring findings from either of these models to the human situation.
A major obstacle to the administration of 1,25(OH)2D3 in humans is its important effects on calcium and bone metabolism. However, in view of the therapeutic potential, structural analogues have been designed and synthesised with the aim of achieving a dissociation between calcaemic and immune effects. Several analogues developed in different chemical laboratories have been successfully tested in the NOD mouse . The mechanism of protection against insulitis and diabetes appears to be similar to that of 1,25(OH)2D3. Effects of the analogues on dendritic cell phenotype, regulator cell induction and beta cell protection have been described [8, 77, 83].
The optimal analogue should provide a combination of beta cell protection, immune modulation and low calcaemic effects. Although several analogues have provided promising results , long-term safety data will have to be collated before long-term interventions in individuals at high risk of type 1 diabetes are initiated. A critical question as regards the applicability of analogues in the human situation is whether these 1,25(OH)2D3 analogues are able to arrest progression to clinically overt diabetes if administered when active beta cell destruction is already present—the situation in pre-diabetic subjects in whom immune intervention is considered . Casteels et al demonstrated that, when combined with a short induction course of a classical immunosuppressant such as cyclosporine A, some of these analogues can arrest the progression of the disease when administered after autoimmunity has started . This approach of combining a short induction course of a classical immunosuppressant with non-hypercalcaemic analogues of 1,25(OH)2D3 is promising and might open new perspectives for the prevention of autoimmune diabetes in humans.
The capacity of some of the analogues of 1,25(OH)2D3 to prevent disease recurrence after islet transplantation has been evaluated in spontaneously diabetic NOD mice. The most convincing results were obtained using a combination of KH1060 and cyclosporine A . A similar synergism has been reported between 1,25(OH)2D3 or an analogue and mycophenolate mofetil . One novel approach involved the combination of analogues of 1,25(OH)2D3 with other physiological immune modulators, such as recombinant IFN-β . In this study, the combination of recombinant IFN-β and ZXY1106 markedly delayed the recurrence of autoimmune diabetes following islet transplantation .
Solid evidence exists that vitamin D deficiency is detrimental to beta cell function, leads to glucose intolerance in animal models and humans, and predisposes to type 2 diabetes. Vitamin D deficiency in early life predisposes NOD mice and humans to the later development of autoimmune diabetes. A major practical conclusion that can be drawn from the studies conducted on vitamin D and diabetes to date is that vitamin D deficiency is undesirable, not only for calcium and bone, but also for glucose metabolism.
The metabolically active form of vitamin D, 1,25(OH)2D3, and its analogues have been shown to have effects on the major players involved in the pathogenesis of type 1 and type 2 diabetes. Beta cell function has been shown to be improved by 1,25(OH)2D3 in vitro and in vivo, and the avoidance of vitamin D deficiency is essential for normal beta cell function. In NOD mice, 1,25(OH)2D3 protects against insulitis, diabetes and disease recurrence after islet transplantation, primarily through immunomodulatory effects.
The use of 1,25(OH)2D3 for the prevention or cure of diabetes is limited by its hypercalcaemic and bone remodelling effects, as its protective effects are only observed in response to supra-physiological doses. Future applications of this therapy in human diabetes are conceivable, since the calcaemic and immunomodulatory effects of 1,25(OH)2D3 can be dissociated by structural analogues of the molecule.
It is likely that these analogues will find a use as beta-cell-protective and -stimulating agents as adjuncts to the current treatment for type 2 diabetes. These agents could play a major role in preventative strategies for type 1 diabetes in humans, because of their profile as both beta cell-protective and immuno-active drugs. More information is required concerning their mechanism of action and long-term safety before they can be routinely administered in humans.
The authors would like to thank E. van Etten for drawing Fig. 3. They would also like to thank the Belgium Program on Interuniversity Poles of Attraction initiated by the Belgian State (IUAP P5/17), the Catholic University of Leuven (Geconcerteerde Onderzoeksacties [GOA] 2004/10), the Flemish Research Foundation (FWO G.0084.02 and G.0233.04), the Juvenile Diabetes Research Foundation Centre for Prevention of Beta Cell Destruction in Europe (grant number 4-2002-457), and the European Community concerted Sixth Framework Program (grant number TONECA 503245) for support. C. Mathieu has a clinical research fellowship (FWO), C. Gysemans a postdoctoral fellowship (FWO), and A. Giulietti a doctoral scholarship from the Catholic University of Leuven (Interfaculty Council for Development Co-operation).