|Home | About | Journals | Submit | Contact Us | Français|
Vitamins are essential constituents of our diet that have long been known to influence the immune system. Vitamins A and D have received particular attention in recent years as these vitamins have been shown to have an unexpected and crucial effect on the immune response. We present and discuss our current understanding of the essential roles of vitamins in modulating a broad range of immune processes, such as lymphocyte activation and proliferation, T-helper-cell differentiation, tissue-specific lymphocyte homing, the production of specific antibody isotypes and regulation of the immune response. Finally, we discuss the clinical potential of vitamin A and D metabolites for modulating tissue-specific immune responses and for preventing and/or treating inflammation and autoimmunity.
“A vitamin is a substance that makes you ill if you don’t eat it.” (Albert Szent-Gyorgyi, Nobel Prize in Physiology or Medicine, 1937).
The statement by Albert Szent-Gyorgyi epitomizes the impact of vitamins on the body’s vital organs, including the immune system. Vitamins (vital amines) are organic compounds that are required in trace amounts in the diet because they cannot be synthesized in sufficient quantities by an organism1. Vitamins and their metabolites are essential for a large number of physiological processes, fulfilling diverse functions as hormones and antioxidants, as regulators of tissue growth and differentiation, in embryonic development and in calcium metabolism, among others1.
In addition, vitamins have a role in the immune system, which extends to both innate and adaptive immune responses. Although some vitamins, such as vitamins C and E and members of the B complex, can act in a relatively nonspecific manner in the immune system (for example, as antioxidants)2–4, other vitamins, such as vitamins A and D, can influence the immune response in highly specific ways (see Supplementary information S1 (table)). Here we review the most important effects of vitamins on the immune system, with special emphasis on vitamins A and D, which have received particular attention owing to recent discoveries of their multi-faceted interactions with the immune system. Vitamins A and D are notably distinct from other vitamins in that their respective bioactive metabolites, retinoic acid and 1,25-dihydroxyvitamin D3 (1,25(OH)2VD3), have hormone-like properties. Both of these metabolites are synthesized from their vitamin precursors by different tissues and cells in the body and exert their effects on target cells remotely by binding to nuclear-hormone receptors.
Vitamin D3 (VD3), the most physiologically relevant form of vitamin D, is synthesized in the skin from 7-dehydrocholesterol5, a process which depends on sunlight, specifically ultraviolet B radiation (wavelengths of 270–300 nm). Alternatively, it can be acquired in the diet or in vitamin supplements5 (FIG. 1a). VD3 is then converted in the liver to 25-dihydroxyvitamin D3 (25(OH)VD3), which is the main circulating form of VD3. Finally, 25(OH)VD3 is metabolized in the kidneys to 1,25(OH)2VD3, the most physiologically active VD3 metabolite5. In addition to being processed in the liver and the kidneys, VD3 can also be metabolized by cells of the immune system5,6 (FIG. 1a). In this way, 1,25(OH)2VD3 is concentrated locally in those lymphoid microenvironments that contain physiologically high concentrations of VD3, thereby increasing its specific action and also limiting potentially undesirable systemic effects, such as hypercalcaemia and increased bone resorption7.
Activated T cells (and probably also B cells) can only perform the final step of converting 25(OH)VD3 to 1,25(OH)2VD3 (REFS 6,8). However, macrophages and some dendritic cells (DCs), such as monocyte-derived DCs and dermal DCs, express the two sets of enzymes needed to convert VD3 into 1,25(OH)2VD3 (REFS 6,7,9).
Finally, the enzyme 24-hydroxylase, which is most abundant in the kidney and intestine10, catabolizes 1,25(OH)2VD3 to its inactive metabolite, calcitroic acid, which is then excreted in the bile.
Vitamin A is obtained from the diet either as all-trans-retinol, retinyl esters or β-carotene11,12 (FIG. 1b). All-trans-retinol is esterified to retinyl esters and stored in the liver, mostly in the stellate cells11,12. In the tissues, all-trans-retinol and β-carotene are oxidized to all-trans-retinal by alcohol dehydrogenases or short chain dehydrogenase reductases, which are ubiquitously expressed enzymes11,12. All-trans-retinal is then oxidized to all-trans-retinoic acid through an irreversible reaction catalysed by retinal dehydrogenases (RALDHs), the expression of which is tightly controlled. A related metabolite of all-trans-retinal, 9-cis-retinoic acid, can be formed either by spontaneous isomerization of all-trans-retinoic acid or through oxidation of 9-cis-retinal by RALDH13. However, 9-cis-retinoic acid has not been shown to be synthesized in vivo13.
In adult mammals, RALDH can be found in some gut-associated cells, including intestinal epithelial cells (IECs)14,15 and gut-associated DCs, such as DCs from Peyer’s patches and mesenteric lymph nodes15,16. Interestingly, DCs from Peyer’s patches express RALDH-1 mRNA and protein, whereas DCs from mesenteric lymph nodes express mRNA encoding RALDH-2. Although the functional relevance of this differential RALDH isoform expression is currently unclear, it indicates that there might be more than one pathway or environmental stimulus that renders DCs capable of synthesizing retinoic acid from vitamin A. In addition, IECs also express RALDH-1 and can metabolize vitamin A to retinoic acid in vitro14, which indicates that they are another potential source of retinoic acid in the gut mucosa. The relative contributions and in vivo relevance of these different sources of retinoic acid in the gut are yet to be determined.
Locally produced 1,25(OH)2VD3 can act on immune cells in an autocrine or paracrine manner. On complexing with 1,25(OH)2VD3, the nuclear vitamin D receptor (VDR) heterodimerizes with nuclear receptors of the retinoic X receptor (RXR) family — which has three main isoforms: α, β and γ — and binds to VD3 response elements (VDREs) in the promoters of VD3-responsive genes (FIG. 1c).
Similarly, retinoic acid exerts its multiple effects by binding to nuclear receptors of the retinoic acid receptor (RAR) family, which also has three main isoforms: α, β and γ. These form RAR–RXR heterodimers, which interact with retinoic acid response elements (RAREs) within the promoters of retinoic acid-responsive genes11,12. RAR proteins are ubiquitously expressed and are also upregulated by retinoic acid11,12. As mentioned above, RXR proteins can also pair with VDR proteins or form RXR–RXR homodimers, which are specific receptors for 9-cis-retinoic acid but not for all-trans retinoic acid (hereafter referred to as retinoic acid) (although 9-cis-retinoic acid can also signal through RAR–RXR heterodimers). In addition, RXR proteins are partners for other nuclear receptors, such as thyroid hormone receptor, peroxisome proliferator-activated receptor (PPAR) and liver X receptor, among others17. Therefore, it is possible that, given their common RXR nuclear binding partners, some ligands, such as 1,25(OH)2VD3 and retinoic acid, might antagonize each other’s effects18.
Notably, retinoic acid can also bind and signal through the PPARβ (also known as PPARδ) nuclear receptor19. Whether signalling occurs through RAR or PPARβ depends on the ratio of cellular retinoic acid-binding proteins (CRABPs) to fatty acid-binding protein 5 (FABP5), which ultimately determines the partitioning of retinoic acid between the two types of receptors19. In this model, a high CRABP:FABP5 ratio promotes RAR signalling by CRABP-mediated ‘channelling’ of retinoic acid to RAR, which results in growth inhibition and apoptosis in some cell lines, whereas a low ratio favours FABP5-mediated delivery of retinoic acid to PPARβ and survival in the same cells19. However, although this is a potentially attractive mechanism for regulating retinoic acid responses, its significance in the immune system is yet to be determined.
The influence of VD3 metabolites in the immune system, particularly of 1,25(OH)2 VD3, has been known for more than 20 years20,21. In vitro, 1,25(OH)2VD3 exerts a marked inhibitory effect on adaptive immune cells (FIG. 2). It inhibits T-cell proliferation20,21, the expression of interleukin-2 (IL-2)21–23 and interferon-γ (IFNγ) mrNA and protein in T cells24,25, and CD8 T-cell-mediated cytotoxicity26. The decrease in the production of IL-2 and IFNγ by 1,25(OH)2VD3 is partially mediated by binding of the VDR–RXR complex to the VDRE in the promoters of genes encoding IL-2 (REF. 27) and IFNγ (REF. 28). The anti-proliferative effect could be explained, at least in part, by the decrease in IL-2 production, as proliferation is partially rescued by adding exogenous IL-2 (REFS 21,22,29). These inhibitory effects of 1,25(OH)2VD3 are most pronounced in the memory T-cell compartment30, which is concomitant with the higher expression of VDR in effector and memory T cells compared with naive T cells31. Moreover, 1,25(OH)2VD3 enhances nonspecific T-cell suppressor activity, as measured by the ability of 1,25(OH)2VD3-treated T cells to suppress primary mixed-lymphocyte reactions and cytotoxic T-cell responses26.
Overall, the net result of 1,25(OH)2VD3 action on T cells is to block the induction of T-helper-1 (TH1)-cell cytokines, particularly IFNγ, while promoting TH2-cell responses, an effect mediated both indirectly by decreasing IFNγ production and directly by enhancing IL-4 production7. The activity of 1,25(OH)2 VD3 on effector T-cell differentiation is further enhanced by its effect on antigen-presenting DCs, in which it suppresses the synthesis of IL-12, a cytokine that promotes TH1-cell responses32,33. Furthermore, 1,25(OH)2VD3 also inhibits TH17-cell responses, probably owing in part to its capacity to inhibit IL-6 and IL-23 production34,35, and induces the reciprocal differentiation and/or expansion of forkhead box protein 3 (FOXP3)+ regulatory T (TReg) cells35–37.
In addition to its inhibitory effects on T cells, 1,25(OH)2VD3 decreases B-cell proliferation, plasma-cell differentiation and IgG secretion8,20 (FIG. 2). It has been suggested that the effect of 1,25(OH)2VD3 on B cells might be indirectly mediated through the effect it has on antigen-presenting-cell (APC) function and/or T-cell help38. Indeed, there are conflicting reports concerning the expression of VDR by B cells8,31,39, leaving it unclear whether 1,25(OH)2VD3 can act directly on B cells.
Interestingly, 1,25(OH)2VD3 inhibits mitogen-stimulated IgG production by B cells from patients with inactive systemic lupus erythematosus (SLE), but not the spontaneous IgG production by cells from patients with active SLE40. Thus, it is possible that fully differentiated memory B cells and/or antibody-secreting cells (ASCs) are refractory to 1,25(OH)2VD3-mediated inhibition. In addition, serum levels of 1,25(OH)2VD3 are significantly decreased in patients with SLE, especially during active disease8. So, it is tempting to speculate that a decreased level of 1,25(OH)2VD3 could have an exacerbating role in SLE pathogenesis by releasing a basal physiologic or ‘tonic’ brake in humoral immunity.
Cells of the innate immune system can also be inhibited by 1,25(OH)2VD3 (FIG. 2), which is known to inhibit the differentiation, maturation and immunostimulatory capacity of DCs by decreasing the expression of MHC class II molecules and of CD40, CD80 and CD86 (REFS 7,9,33,41). Furthermore, VDR-deficient mice have increased numbers of mature DCs in skin-draining lymph nodes41. In addition, 1,25(OH)2VD3 decreases the synthesis of IL-12 (REFS 32,33) and simultaneously increases the production of IL-10 by DCs33. The net result is a decrease in TH1-cell responses and probably an induction of IL-10-producing T regulatory type 1 (TR1) cells7.
Although 1,25(OH)2VD3 primarily has inhibitory effects on the adaptive immune response, some of its effects on innate immune cells are stimulatory. For example, 1,25(OH)2VD3 can stimulate human monocyte proliferation in vitro42 and has been shown to increase the production of both IL-1 and the bactericidal peptide cathelicidin by monocytes and macrophages5,23.
Unexpectedly, in spite of the potentially important role of 1,25(OH)2VD3 in maintaining immune homeostasis, VDR-deficient mice have a normal composition of immune-cell populations and reject allogeneic and xenogeneic transplants at the same rate as wild-type mice43. However, to carefully dissect the in vivo effects of VD3 metabolites on the immune response, it will be necessary to study the effect of VD3 in animal models in which VDR deficiency is restricted to T cells, B cells and myeloid cells.
Vitamin A metabolites can also affect some aspects of the adaptive immune response (FIG. 3). Retinoic acid enhances cytotoxicity44 and T-cell proliferation45, the latter probably mediated, at least in part, by enhancing IL-2 secretion and signalling in T cells45. Consistent with an in vivo role for vitamin A in T-cell function, vitamin A-deficient mice have defects in TH-cell activity46. A possible mechanism for this observation is that in the setting of vitamin A deficiency, retinoic acid does not compete with 1,25(OH)2VD3 for their common nuclear binding partner RXR and, therefore, the inhibitory effects of 1,25(OH)2VD3 on T-cell function (including TH-cell activity) are not offset by retinoic acid.
Retinoic acid can inhibit B-cell proliferation47,48, although it has also been found to enhance B-cell activation under some conditions49,50. In addition, retinoic acid inhibits B-cell apoptosis. These effects are mediated through binding of vitamin A metabolites to RAR receptors51.
Notably, it has been reported that a distinct set of vitamin A metabolites classified as retro-retinoids can also affect general lymphocyte functions such as B-cell proliferation52 and T-cell activation and proliferation52. 14-hydroxy-retroretinol (14HRR) has a positive effect on proliferation, whereas anhydroretinol blocks B-cell proliferation and induces apoptosis in T cells53. Retroretinoids do not signal through RAR or RXR and, as 14HRR and anhydroretinol can antagonize each other’s effects, it has been suggested that they might compete for a common, as-yet unknown receptor53.
Retinoic acid can also modulate antigen presentation by exerting direct effects on DC function. For example, retinoic acid increases the expression of matrix metalloproteinases, thereby increasing the migration of tumour-infiltrating DCs to the draining lymph nodes, which have the potential to boost tumour-specific T-cell responses54. In addition, in the presence of inflammatory stimuli, such as tumour-necrosis factor (TNF), retinoic acid enhances DC maturation and antigen-presenting capacity, both of which are effects mediated by RXR receptors55. However, it should be noted that DCs pre-treated with retinoic acid can apparently store this metabolite50, which when released could ultimately act directly on T cells and/or other cells and contribute to the final outcome of an immune response.
Vitamin A metabolites also modulate more specific functional aspects of the immune response, such as the TH1–TH2-cell balance and the differentiation of TReg cells and TH17 cells (FIG. 3a). Vitamin A deficiency correlates with decreased TH2-cell responses56 and, conversely, vitamin A supplementation blocks the production of TH1-cell cytokines in vitro and in vivo57,58. These effects of vitamin A on TH1–TH2-cell differentiation are mediated by retinoic acid. In fact, retinoic acid promotes TH2-cell differentiation by inducing IL4 gene expression59. Moreover, retinoic acid blocks the expression of the TH1-cell master regulator T-bet and induces TH2-cell-promoting transcription factors, such as GATA3 (GATA-binding protein 3), macrophage-activating factor (MAF) and signal transducer and activator of transcription 6 (STAT6)57,60. Interestingly, vitamin A supplementation was correlated with an increase in disease severity in a mouse model of asthma, whereas vitamin A deficiency had the opposite effect, which was associated with a decrease in TH2-cell cytokines61. It has been proposed that retinoic acid exerts its TH2-cell-promoting effect indirectly through the modulation of APCs62. However, retinoic acid can also act directly on T cells to induce TH2-cell differentiation57,60 through RAR proteins57.
Several types of T cells that have dominant immunomodulatory effects have been described. The best characterized are TReg cells that express the transcription factor FOXP3. Although transforming growth factor-β (TGFβ) drives the generation of induced TReg cells in peripheral tissues63, it was recently demonstrated by several groups that this process can be significantly enhanced by retinoic acid64,65 (FIG. 3a). In addition, DCs from the gut-associated lymphoid tissue (GALT) or small intestinal lamina propria also enhance TReg-cell differentiation in a retinoic acid-dependent manner16,66. Notably, one recent study indicated that macrophages and not DCs are responsible for inducing TReg cells in the intestinal lamina propria67, and that DCs are mainly involved in the induction of TH17 cells at this site67,68. The reasons for these seemingly discrepant results are unclear. In addition to inducing FOXP3, retinoic acid also upregulates gut-homing receptors on TReg cells, targeting these cells to the gut mucosa65,66. The relative contribution of in situ-generated gut TReg cells to both oral and peripheral tolerance is yet to be determined.
The differentiation of TReg cells and TH17 cells is reciprocally regulated by cytokine signals69. Exposure of activated CD4+ T cells to TGFβ alone induces TReg cells, whereas the combination of TGFβ with IL-6, IL-1β and IL-23 or IL-21 blocks FOXP3 induction and induces TH17-cell differentiation69,70. However, exposure of CD4+ T cells to retinoic acid together with TGFβ and IL-6 negates the TH17-cell-promoting effect of IL-6 by enhancing the induction of TReg cells and blocking the induction of retinoic-acid-receptor-related orphan receptor-γt (RORγt), a key transcription factor for TH17-cell differentiation64,71. It should be noted that this effect is only observed when retinoic acid is used over a certain threshold concentration68. In fact, low concentrations of retinoic acid seem to be necessary for TH17-cell differentiation68. So, retinoic acid has a dual role in maintaining immunological tolerance: it favours the induction of TReg cells and can simultaneously either block or enhance TH17-cell differentiation, depending on its concentration.
An important feature of activated B cells is their capacity to undergo immunoglobulin class-switching and to give rise to different antibody isotypes. TH1- and TH2-cell cytokines differentially influence antibody class-switching: the TH1-cell cytokine IFNγ promotes switching to IgG2a and IgG3, whereas the TH2-cell cytokine IL-4 induces the production of IgG1 and IgE and suppresses the generation of IgG2b and IgG3 (REF. 72). DCs can also modulate B-cell activation and antibody class-switching73, which could occur indirectly by influencing TH-cell differentiation74. However, DCs can also directly promote the induction of specific immunoglobulin isotypes. GALT-resident DCs efficiently induce the generation of IgA+ ASCs when cultured with activated B cells in vitro75,76. Several cytokines and other bioactive factors are involved in the capacity of GALT-resident DCs to induce IgA+ ASCs, including TGFβ1 (REFS 16,67,74,77), IL-6 (REFS 75,76), APRIL (a proliferation-inducing ligand)78 and nitric oxide79.
GALT- and lamina propria-resident DCs also contribute to the generation of IgA+ ASCs by a mechanism depending, at least in part, on retinoic acid68,76 (FIG. 3b). The presence of retinoic acid efficiently induces IgA secretion by lipopolysaccharide (LPS)-activated splenocytes80 or by LPS-stimulated B cells cultured with spleen DCs68. However, the presence of retinoic acid during the activation of purified B cells is not sufficient to impart these effects68,76. Retinoic acid-induced IgA secretion requires either IL-5 (REFS 76,80) or IL-6 (REFS 75,76), which are known to possess a general adjuvant role in IgA production73,81,82. In fact, both retinoic acid and IL-6 are required for GALT-resident DCs to induce optimal IgA production by mouse and human B cells in vitro75,76. Vitamin A-depleted mice have decreased numbers of IgA+ ASCs in the small bowel lamina propria76,83, consistent with an in vivo role for retinoic acid in gut mucosal IgA responses. In addition, oral administration of a RAR agonist significantly increases serum IgA levels in rats84. Interestingly, exposure to retinoic acid together with IL-6 or IL-5 is not sufficient to induce IgA+ ASCs from in vitro-activated naive B cells in the absence of DCs76. Therefore, it is likely that DCs provide some necessary B-cell survival and differentiation signals that are not tissue-specific.
It was recently shown that inducible nitric oxide synthase (iNOS; also known as NOS2A) and nitric oxide are crucial for the generation of IgA+ ASCs79. Interestingly, the promoter of the iNOS gene contains a RARE that is directly activated by retinoic acid bound to its nuclear RARα–RXR heterodimeric receptor85. Moreover, intraperitoneal administration of either retinoic acid or a RAR agonist enhances LPS-induced iNOS expression in several organs and increases plasma levels of nitrate and nitrite in rats86. Therefore, retinoic acid might also indirectly contribute to IgA secretion by inducing iNOS and nitric oxide expression.
Precursors of retinoids in the diet (which include β-carotene and retinyl esters) are absorbed from the gut lumen and can be metabolized to produce retinoic acid in IECs14 and in DCs that reside near the germinal centres in Peyer’s patches15. The proximity of Peyer’s patches to the gut bacterial flora and to other potential sources of retinoic acid and cytokines, such as IECs, could help to explain the predominance of IgA class-switching that occurs in Peyer’s patches compared with mesenteric lymph nodes87. Consistent with a role for vitamin A in gut IgA production, rats or mice depleted of vitamin A have decreased levels of total IgA in intestinal lavages and decreased mucosal antigen-specific IgA responses, which correlates with decreased protection against infections and oral bacterial toxins56,88,89. Furthermore, vitamin A supplementation prevents the decline in IgA levels observed in malnourished mice58. In addition to a direct effect on ASCs, it should be considered that vitamin A deficiency could also decrease IgA secretion in the gut by reducing the expression of the polymeric immunoglobulin receptor, leading to a decrease in the secretion of dimeric IgA to the gut lumen88,89. Moreover, although vitamin A-deficient mice have decreased numbers of IgA+ ASCs in the small bowel76,83, their serum IgA levels are normal76, which indicates that retinoids are not absolutely required for generating IgA+ ASCs in other mucosal compartments.
Although naive lymphocytes migrate mainly through secondary lymphoid organs, effector and memory lymphocytes acquire ‘traffic’ molecules that endow them with the capacity to migrate to select extralymphoid tissues and to sites of inflammation. Of the extralymphoid compartments, the gastrointestinal mucosa and the skin are the two main body surfaces exposed to environmental antigens and are also the two paradigmatic tissues for which tissue-specific adhesion and chemoattractant receptors (also known as homing receptors) have been characterized in detail. For example, effector and memory lymphocytes migrating to the small bowel require expression of the α4β7-integrin and CC-chemokine receptor 9 (CCR9), whereas those migrating to the skin rely on the expression of ligands for E- and P-selectin and CCR4 or CCR10 (REF. 73) (FIG. 4a). It has been demonstrated that the lymphoid microenvironment in which lymphocytes are activated determines the set of homing receptors that they acquire — for example, T cells activated in skin-draining lymph nodes acquire skin-homing receptors whereas those activated in the GALT acquire gut-homing receptors90. In the lymphoid microenvironment, DCs are essential for efficient T-cell activation91. Several groups have shown that DCs from Peyer’s patches and mesenteric lymph nodes are sufficient to induce the expression of α4β7-integrin and CCR9 and therefore imprint gut-homing capacity on activated mouse T cells92–96 and mouse and human B cells68,76, whereas lymphocytes activated by mouse DCs from peripheral lymph nodes preferentially acquire skin-homing receptors95,96.
How do GALT-resident DCs imprint lymphocytes with a gut-homing phenotype? More than 25 years ago it was shown that rats suffering from both protein-caloric and vitamin A deficiencies exhibited impaired migration of recently activated mesenteric lymphocytes to the small intestinal mucosa97. Protein-caloric malnutrition without vitamin A deficiency did not affect lymphocyte migration97. Adoptive transfer experiments showed that impaired lymphocyte migration was observed only when donor lymphocytes were from protein-caloric-deficient and vitamin A-deficient rats and not when wild-type cells were transferred to protein-caloric-deficient and vitamin A-deficient recipients97. This suggests that vitamin A deficiency mainly affected lymphocyte migratory capacity, but not the target tissues97. More recently, it was described that vitamin A-depleted rats had a marked decrease in the number of IgA+ ASCs and CD4+ T cells in the ileum83. The molecular basis for these observations was recently determined in a study that showed that mice depleted of vitamin A had decreased numbers of effector and memory T cells in the gut mucosa, but not elsewhere15. The vitamin A metabolite retinoic acid was sufficient to induce the expression of α4β7-integrin and CCR9 by activated T cells, even in the absence of DCs15. Blocking retinoic acid receptors of the RAR family significantly decreased the induction of α4β7-integrin expression by T cells by GALT-resident DCs, which shows that retinoic acid is essential for the gut-imprinting capacity of the DCs15. Consistently, GALT-resident DCs, unlike DCs from other tissues, express RALDH enzymes, which are essential for retinoic acid biosynthesis15. Together, these results indicate that retinoic acid is pivotal for the imprinting of gut-homing T cells.
Although vitamin A deficiency decreases the number of T and B cells in the small bowel lamina propria15,76,83,97, it does not affect lymphocyte migration to the colon97. Analogously, GALT-resident DCs imprint T and B cells with homing capacity for the small bowel, but they do not induce colon-homing T cells93. Therefore, retinoic acid is neither necessary nor sufficient to imprint colon-homing lymphocytes. The molecular signals that are responsible for lymphocyte homing to the colon and the reasons why T-cell migration to this compartment is controlled differently from homing to the small bowel are still to be determined.
Regarding the migration of ASCs, it has been proposed that CCR10 might have a role in the homing of IgA+ ASCs to the colon, mammary glands and probably to other mucosal compartments73 (FIG. 4b). However, it is currently unclear how CCR10 expression is induced by ASCs. Recent reports indicate that IgA+ ASCs might acquire CCR10 expression in colonic patches or in iliac lymph nodes following rectal immunization98 and that the expression of this receptor can also be induced by 1,25(OH)2VD3 in human ASCs39. However, 1,25(OH)2VD3 does not induce CCR10 expression in murine ASCs in vitro and VDR-deficient mice have normal numbers of CCR10+ IgA+ ASCs39, which indicates that 1,25(OH)2VD3 might not be necessary for the induction of CCR10 expression by B cells in vivo, at least in mice.
In vitro-generated tissue-tropic lymphocytes retain marked plasticity; skin-homing T cells can be converted to gut-homing T cells and vice versa if they are re-stimulated with or without GALT-resident DCs, respectively95,96. Similarly, previously activated B cells can be re-educated and acquire or lose gut-homing potential when they are restimulated with or without retinoic acid, respectively76. The decrease in α4β7-integrin and CCR9 expression observed in T and B cells that are activated in the absence of retinoic acid might be a default differentiation mechanism. However, it is also possible that other factors can actively contribute to the downregulation of gut-homing-receptor expression by lymphocytes. As both RAR and VDR must form heterodimers with RXR to signal, it is possible that 1,25(OH)2VD3 could actively antagonize the effects of retinoic acid by competing for the same nuclear partner18. Consistent with this possibility, 1,25(OH)2VD3 blocks the retinoic acid-induced upregulation of gut-homing receptors on human T cells6,99. However, whether this retinoic acid antagonism by 1,25(OH)2VD3 has a regulatory role in the imprinting of gut-homing lymphocytes in vivo remains unknown.
In addition to imprinting a gut-homing phenotype on lymphocytes, retinoic acid and GALT-resident DCs also block the upregulation of the skin-homing receptors CCR4 and ligands for P- and E-selectin by T cells15,95. Therefore, acquisition of a skin-homing phenotype might be the default pathway for T-cell activation in the absence of retinoic acid or when RAR signalling is blocked100. Nonetheless, as VD3 as well as 1,25(OH)2VD3 can be synthesized in the skin101, it is conceivable that, like retinoic acid in the gut, 1,25(OH)2VD3 might have a reciprocal role in imprinting lymphocyte homing to the skin. In agreement with this possibility, it was recently shown that 1,25(OH)2VD3 synergizes with IL-12 to induce the expression of skin-associated CCR10 by human T cells6. However, it was also shown that 1,25(OH)2VD3 actually blocks the upregulation of ligands for E-selectin6,99 and the expression of fucosyltransferase-VII (REF. 99), an enzyme essential for the synthesis of selectin ligands102. This was correlated with decreased homing of T cells to the inflamed skin in a model of contact hypersensitivity induced by oxazolone99. Although these data indicate that 1,25(OH)2VD3 might block skin-homing, it should be noted that these experiments were carried out without IL-12 supplementation, which could potentially counteract the negative effect of 1,25(OH)2VD3 on the expression of E-selectin ligands and skin-homing. It is also possible that 1,25(OH)2VD3 induces CCR10 expression by T cells after they have homed to the skin to increase their retention in this tissue. In fact, because keratinocytes express the CCR10 ligand CC-chemokine ligand 27 (CCL27), it has been proposed that CCR10 upregulation might promote T-cell trafficking to and/or retention in the epidermis6.
Finally, 1,25(OH)2VD3 might also affect leukocyte migration by blocking chemokine synthesis at effector sites. For instance, 1,25(OH)2VD3 decreased the expression of CCL2, CCL3, CXCl10 and subsequent monocyte infiltration in experimental autoimmune encephalomyelitis (EAE)103. Similarly, a 1,25(OH)2VD3 analogue decreased the production of the chemokines CCL2, CCL5, CXCl10 and consequent TH1-cell infiltration in non-obese diabetic (NOD) mice, a model of type 1 diabetes104.
It has been known for more than 30 years that some vitamins with antioxidant properties, including vitamin A, vitamin B6 (pyridoxine), vitamin C (ascorbic acid) and particularly vitamin E, have protective effects on animal models of atherosclerosis and ischaemia-reperfusion injury (IRI)2–4. Vitamin E collectively refers to eight related compounds (tocopherols and tocotrienols), of which α-tocopherol has the greatest bioavailability and is the best characterized105. Vitamin E decreases the release of reactive oxygen species by monocytes106 and the expression of CD11b and very late antigen 4 (VLA4), thereby decreasing monocyte adhesion to the endothelium106. Vitamin E also blocks the release of pro-inflammatory cytokines, including IL-1, IL-6, TNF and the chemokine IL-8, by monocytes and macrophages107,108. Moreover, vitamin E prevents the upregulation of the adhesion molecules vascular cell-adhesion molecule 1 (VCAM1) and intercellular adhesion molecule 1 (ICAM1) on the endothelium induced by oxidized low-density lipo protein (LDL)109 and IL-1β110, as well as the upregulation of E-selectin and some chemokines108. Reactive oxygen species activate the nuclear factor-κB (NF-κB) pathway106, which initiates many pro-inflammatory events. Therefore, the therapeutic antioxidant effect of these vitamins could be explained, at least in part, by their capacity to decrease NF-κB activation.
Vitamin E can also act directly on T cells by decreasing IFNγ production111 and CD95L (also known as FASL)112 expression, thereby helping to decrease inflammation and immune-mediated tissue damage. These effects on macrophages and T cells are believed to be important for the protective effect for vitamin E in animal models of atherosclerosis108,113 and IRI114,115. Consistent with a potential physiological role for vitamin E in preventing atherosclerosis, hyperlipidemic mice that are deficient in α-tocopherol transfer protein, which is important for transporting α-tocopherol and for preventing its degradation, have more severe atherosclerosis116.
It was recently shown that vitamin C prevents oxidative damage during ischaemia reperfusion in rats117 and humans4. Notably, it was shown that vitamin C but not vitamin E prevented leukocyte adhesion to the microvascular endothelium in hamster models of oxidative endothelial stress induced by cigarette smoke or oxidized LDL118,119. Differences in lipophilicity might potentially have an impact in the distribution and/or location of vitamins C and E and partially account for the differential effect of these vitamins. Therefore, it is possible that combined supplementation of vitamins C and E could offer synergistic benefits. The antioxidant and/or anti-atherogenic role for other vitamins, such as vitamin B6 and vitamin K, is less well documented and somewhat controversial, with evidence both in favour120,121 and against122 a protective role for these vitamins in atherosclerosis.
Finally, it should be noted that many of the encouraging results in animal models have not been consistently translated into a significant therapeutic benefit in controlled clinical trials of vitamin supplementation for the prevention of cardiovascular diseases and IRI121. Therefore, it remains to be determined whether antioxidant vitamins will prove to be useful for the treatment and/or prevention of these ailments in humans.
Given that 1,25(OH)2VD3 has a physiological protective role in dampening or limiting potentially pathogenic immune responses at the cellular level, one might predict that interfering with its effects could predispose to hypersensitivity or autoimmunity7. Consistent with this idea, levels of serum 1,25(OH)2VD3 are often decreased in patients with type I diabetes123 and SLE8,124, and 1,25(OH)2VD3 levels are inversely correlated with disease activity in patients with rheumatoid arthritis125. Moreover, VD3 deficiency accelerates intestinal inflammation in IL-10-deficient mice, which develop inflammatory bowel disease126. VD3 deficiency might also predispose to type I diabetes7,127. However, although children with rickets (typically caused by insufficient dietary supply of vitamin D) have a higher incidence of diabetes than VD3-sufficient children7, they are also more susceptible to infection7,128, which suggests that VD3 might also be important for protective immune responses. This effect could be partially mediated by the capacity of 1,25(OH)2VD3 to increase the bactericidal capacity of macrophages5,23.
Given its immunomodulatory properties, 1,25(OH)2VD3 or its analogues might be clinically useful for the treatment of inflammatory and auto immune diseases. Administration of 1,25(OH)2VD3 or an analogue prevented proteinuria and prolonged life span in a mouse model of experimental SLE129,130. In addition, 1,25(OH)2VD3 prevented EAE in mice103,131,132, an effect that was dependent on IL-10 and IL-10 receptor signalling132. As 1,25(OH)2VD3 in combination with glucocorticoids induces IL-10-producing TR1 cells133, it is possible that 1,25(OH)2VD3 might exert its therapeutic effect, at least in part, through the generation of TR1 cells. However, induction of FOXP3+ TReg cells might also have an important role35–37. Other models of experimental autoimmunity in which 1,25(OH)2VD3 has shown a therapeutic benefit include insulitis in NOD mice127, prostatitis34 and rheumatoid arthritis7,134. 1,25(OH)2VD3 can also block cutaneous contact hypersensitivity99, an effect that could be mediated in part by blocking the induction of skin-homing-receptor expression by lymphocytes99. Consistent with its anti-inflammatory role, a 1,25(OH)2VD3 analogue has been successfully used as a therapy for psoriasis135,136. However, it should be pointed out that topical skin application of 1,25(OH)2VD3 also has the potential to trigger allergic dermatitis by increasing TH2 cell-mediated responses137.
It has been proposed that 1,25(OH)2VD3 could also be used as an adjuvant in immunomodulatory therapy in transplantation. 1,25(OH)2VD3 and a 1,25(OH)2VD3 analogue prolonged the survival of mouse cardiac allografts138,139, and decreased the rates of allograft rejection and increased survival in a rat model of liver transplantation140 and in fully mismatched mouse pancreatic islet transplants141. In addition, a 1,25(OH)2VD3 analogue provided significant protection from graft-versus-host disease (GVHD) in rats142. Moreover, 1,25(OH)2VD3 and a 1,25(OH)2VD3 analogue significantly prevented chronic allograft rejection in a rat model of renal transplantation143 and delayed chronic allograft rejection in a mouse model of aortic transplantation139.
Interestingly, polymorphisms in VDR are associated with a higher incidence of GVHD in patients who undergo bone-marrow transplantation144, which indicates that 1,25(OH)2VD3 might also have a role in suppressing alloreactive immune responses in humans. In agreement with this possibility, 1,25(OH)2VD3 supplementation has been shown to have a beneficial effect by improving allograft function of human renal transplants145, which is especially relevant considering that renal insufficiency is associated with decreased 1,25(OH)2VD3 synthesis146,147. Importantly, although 1,25(OH)2VD3 helps to prevent transplant rejection, it does not seem to interfere significantly with protective immune responses against pathogens148. Therefore, it is possible that once 1,25(OH)2VD3 induces a ‘homeostatic’ immunomodulatory threshold, it does not exert further immunosuppression.
Although 1,25(OH)2VD3 could be a potentially useful immunomodulatory agent for clinical use, it can cause some serious adverse effects, in particular the induction of hypercalcaemia and bone resorption7. Therefore, multiple drug development efforts are aimed at finding 1,25(OH)2VD3 analogues that exert immunomodulation without causing significant hypercalcaemia7. Indeed, long-term administration of a 1,25(OH)2VD3 analogue efficiently decreased serum levels of IL-2 and IgG in mice without significant adverse side effects149, and some 1,25(OH)2VD3 analogues have been successfully used in autoimmune disease models, such as EAE, to decrease the doses of conventional immunosuppressive drugs7. These results indicate that 1,25(OH)2VD3 analogues may be an effective and safer alternative to 1,25(OH)2VD3 for immune modulation.
Given the crucial role of retinoic acid in imprinting a gut-homing capacity on T and B cells15,76, as well as its potential to promote the differentiation of IgA+ ASCs76,80, it is not surprising that vitamin A deficiency is associated with impaired intestinal immune responses56,88,89 and increased mortality associated with gastrointestinal and respiratory infections150. Conversely, vitamin A supplementation correlates with a significant decrease in diarrhoea and mortality in HIV-infected or malnourished children76,151,152. Therefore, retinoic acid or RAR agonists could be used for targeting T- and B-cell responses to the gut mucosa for vaccination purposes.
In addition, given that retinoic acid can potentiate the TGFβ-mediated induction of TReg cells while antagonizing the differentiation of pro-inflammatory TH17 cells64, treatment with retinoic acid together with TGFβ could be a useful strategy to generate TReg cells for treating inflammatory pathologies affecting not only the intestine, but also peripheral tissues. In fact, retinoic acid and RAR-agonists have been successfully used in some models of autoimmune inflammation, such as EAE153,154, adjuvant arthritis155,156 and experimental nephritis157. Retinoids have also been successfully used to treat psoriasis158 and they are effective in treating and/or preventing contact dermatitis in mice and humans159,160. In these models, therapeutic effects partially correlated with the induction of TH2-cell responses153 and decreased expression of α4β7-integrin on effector T cells157, but the role of retinoic acid in the induction of TReg cells and the inhibition of TH17 cells in these settings has yet to be assessed.
Although 1,25(OH)2VD3 clearly exerts immunomodulatory activity in vitro and in vivo, its relative physiological role in maintaining immune tolerance and in shaping immune responses is still unclear. Moreover, as retinoic acid and 1,25(OH)2VD3 can potentially antagonize each other’s effects6,18,99, it will be important to dissect the interplay between 1,25(OH)2VD3, retinoic acid and other mechanisms of immunomodulation in vivo.
1,25(OH)2VD3 can upregulate CCR10 on human T cells and ASCs6,39 while blocking the expression of skin- and gut-homing receptors6,99. However, the in vivo relevance of the effects of 1,25(OH)2VD3 on CCR10 expression by T cells that are infiltrating the skin and by IgA+ ASCs that are migrating to the gut lamina propria remains to be determined. In addition, although GALT-resident DCs enhance IgA secretion and induce gut-homing effector lymphocytes ex vivo, it will be important to determine the relative in vivo contribution of DCs versus other potential sources of retinoic acid in the gut (such as IECs), and to determine whether there are retinoic acid-independent mechanisms of imprinting a gut-homing phenotype. Moreover, as GALT-resident DCs can also enhance the differentiation of TGFβ-induced TReg cells, it will be necessary to determine the in vivo scenarios in which GALT-resident DCs and retinoic acid promote either effector or suppressive T-cell responses. Along these same lines, it will be important to study the contribution of these in situ-generated gut TReg cells compared with their systemic counterparts in maintaining immune tolerance at intestinal and extra-intestinal sites.
Aside from the antioxidant effects of vitamins C and E that have been demonstrated in animal models of cardiovascular disease and IRI, there is a lack of published information on the impact of these and other vitamins, such as vitamin B6 and K120–122, on the adaptive immune system and in other inflammatory settings, such as autoimmune diseases. Whether these vitamins will offer a therapeutic benefit in the settings of human cardiovascular diseases, IRI and other pathologies remains unclear.
Although many open questions remain, there is promise that vitamin A and D metabolites or their analogues have the potential to be used in clinical settings for therapeutic benefit. In particular, it will be important to assess the impact of using 1,25(OH)2VD3 analogues as an adjuvant immunomodulatory therapy in the setting of autoimmune diseases and in transplant recipients. It will also be important to determine the net effects of retinoic acid or synthetic RAR-agonists, especially in the intestine, where these agents appear to have a role in enhancing immune responses. The capacity of vitamin A metabolites to foster gut-homing T cells might improve strategies of mucosal vaccination or aid in decreasing pathogenic immunity by potentiating the induction of TReg cells.
We thank S. Davis for editorial assistance and E. Villablanca for critical reading of this manuscript. J.R.M. is grateful to I. Ramos for constant support. J.R.M. is supported by grants from the Crohn’s & Colitis Foundation of America, the Cancer Research Institute, the Howard M. Goodman Fellowship and the Center for the Study of IBD (DK 43351). M.I. is supported by the Grants-in-Aid from Japan Society for the Promotion of Science and the Ministry of Education, Culture, Sports, Science and Technology, the Naito Foundation and the Uehara Memorial Foundation. U.H.v.A. is supported by National Institutes of Health grants AI061663, AI069259, AI072252, HL56949 and AR42689.
Ulrich H. von Adrian’s homepage: http://www.cbr.med.harvard.edu/labs/vonandrian
Rodrigo Mora’s homepage: http://www.massgeneral.org/gastroenterology/research/research_mora.html
See online article: S1 (table)