1α,25-dihydroxyvitamin D3 acts via transforming growth factor-β to up-regulate expression of immunosuppressive CD73 on human CD4+ Foxp3– T cells

Vitamin D deficiency is associated with increased incidence and severity of various immune-mediated diseases. Active vitamin D (1α,25-dihydroxyvitamin D3; 1,25(OH)2D3) up-regulates CD4+ T-cell expression of the purine ectonucleotidase CD39, a molecule that is associated with the generation of anti-inflammatory adenosine. Here we aimed to investigate the direct impact of 1,25(OH)2D3 on expression of the downstream ecto-5′-nucleotidase CD73 by human CD4 T cells, and components of the transforming growth factor-β (TGF-β) pathway, which have been implicated in the modulation of CD73 by murine T cells. At 10−8 to 10−7 m, 1,25(OH)2D3 significantly increased expression of CD73 on peripheral human CD4+ T cells. Although 1,25(OH)2D3 did not affect the mRNA expression of latent TGF-β1, 1,25(OH)2D3 did up-regulate expression of TGF-β-associated molecules [latency-associated peptide (LAP), glycophorin A repetitions predominant (GARP), GP96, neuropilin-1, thrombospondin-1 and αv integrin] which is likely to have contributed to the observed enhancement in TGF-β bioactivity. CD73 was highly co-expressed with LAP and GARP following 1,25(OH)2D3 treatment, but unexpectedly, each of these cell surface molecules was expressed primarily on CD4+ Foxp3– T cells, rather than CD4+ Foxp3+ T cells. Notably, neutralization of TGF-β significantly impaired 1,25(OH)2D3-mediated induction of CD73. Collectively, we show that 1,25(OH)2D3 enhances expression of CD73 on CD4+ Foxp3– T cells in a process that is at least partially TGF-β-dependent. These data reveal an additional contributing mechanism by which vitamin D may be protective in immune-mediated disease.


Introduction
Vitamin D insufficiency has been negatively associated with multiple immune-mediated diseases which is believed to be attributed to its numerous immunomodulatory properties (recently reviewed in refs 1,2). We and others have shown that the frequency of Foxp3 + regulatory T cells can be increased by either high concentrations (10 À6 M) of the active form of vitamin D (1a,25-dihydroxyvitamin D3; 1,25(OH) 2 D3), or lower doses (10 À7 to À8 M) in the presence of high T-cell stimulation. 3,4 Between 10 À7 and 10 À8 M 1,25(OH) 2 D3 can also increase the frequency of a population of regulatory CD4 + IL-10 + Foxp3 -T cells. 3 Importantly these experimental data are supported by in vivo evidence whereby serum vitamin D status is positively associated with the frequency of Foxp3 + T cells in the periphery and airways, as well as levels of IL-10 in the airways. 3,[5][6][7] The downstream effects of 1,25(OH) 2 D3 are significantly influenced by the cytokine milieu, such that the combination of 1,25(OH) 2  transforming growth factor-b 1 (TGF-b 1 ) induces Foxp3 expression to a greater extent than either compound alone. 8 Regulatory T cells employ various immunoregulatory mechanisms including the secretion of immunosuppressive cytokines such as interleukin-10 (IL-10) and TGF-b, as well as modulating dendritic cell behaviour and granzyme deployment. 9 Of particular relevance here, cells can take up extracellular adenosine by two main mechanisms, either through specific adenosine transporters or through binding to one of four G-protein-coupled P1 purinergic receptors (A1, A2A, A2B and A3). Binding to the purinergic receptors can result in suppression of T-cell receptor signalling and induction of various immunomodulatory effects as reviewed elsewhere. [10][11][12] In contrast, ATP is sensed by P2X and P2Y purinergic receptors and generally promotes a pro-inflammatory environment. 13 The cell surface ecto-enzymes CD39 (ectonucleoside triphosphate dephosphorylase) and CD73 (ecto-5 0 -nucleotidase) regulate levels of ATP and adenosine, and therefore inflammation, by dephosphorylating ATP into ADP and AMP (CD39), and then into adenosine (CD73). 10,11 We have previously shown that 1,25 (OH) 2 D3 can modulate this pathway by increasing expression of CD39 on peripheral blood human CD4 + T cells, which contributes to dampening IL-17A secretion. 14 A multifaceted cytokine, TGF-b is involved in processes such as differentiation of T helper type 17 cells and Fox-p3 + regulatory T cells, proliferation of CD4 + T cells and innate immune cell chemotaxis. 15 Humans express three TGF-b isoforms (TGFb 1-3 ), all of which are synthesized in a latent form bound to latency-associated peptide (LAP) either alone or with latent TGF-b binding protein. Latent TGF-b can be expressed on the cell surface by the transmembrane orphan toll-like receptor glycophorin A repetitions predominant (GARP), which itself requires the chaperone heat-shock protein GP96 for correct conformational folding. 16 The active form of TGF-b is only released following cleavage of LAP or conformational remodelling by a range of molecules including neuropilin-1, thrombospondin-1 and various a v integrins. 17,18 The active moiety can then bind to a heterodimer of type I and II TGF-b receptors to mediate downstream effects. 15,19 Cross-talk between the TGF-b and adenosine pathways is indicated by the fact that treating mouse CD4 + T cells with TGF-b increases expression of CD73, and in CD73 knockout mice, experimental induction of TGF-b mRNA was blocked. 20,21 Furthermore, the ability of CD4 + CD73 + T cells to delay the development of diabetes in a mouse model was impaired by anti-TGF-b, indicating a key protective role for this cytokine. 22 1,25(OH) 2 D3 is known to enhance the frequency of human CD4 + CD39 + T cells; however, whether it controls the expression of the related ecto-5 0 -nucleotidase CD73 is unknown. We further investigated the role of TGF-b in this process given existing evidence for cross-talk between the purine and TGF-b pathways in mice.

Materials and methods
Cell isolation and culture Ethical approval was granted by Guy's Hospital Ethics Committee (09/H0804/77) and full written informed consent was obtained from all donors. CD4 + cells were isolated from peripheral venous blood as previously described using Dynabeads (Invitrogen, Paisley, UK). 3 Then, 1 9 10 6 cells/ml were cultured in RPMI-1640 containing 10% fetal calf serum, 2 mM L-glutamine and 50 lg/ml gentamycin, and stimulated with plate-bound anti-CD3 (1 lg/ml; OKT-3) and 50 IU/ml recombinant human IL-2 (Eurocetus, Harefield, UK) in the presence or absence of 1,25(OH) 2 D3 (BIOMOL Research Labs, Exeter, UK). For cultures going beyond 7 days, cells were counted and re-stimulated at 1 9 10 6 cells/ml on day 7 with fresh medium containing drugs. Where relevant, anti-TGF-b 1-3 or an isotype control (5 lg/ml; R&D Systems, Abingdon, UK) was added to the cultures on days 0 and 7. Viability did not significantly differ between culture conditions (data not shown).

ATP assay
After 7 days of culture alone or in the presence of 10 À7 M 1,25(OH) 2 D3, 25,000 CD4 + T cells were seeded in 100 ll culture medium with 500 lM ATP. Consumption of ATP was assessed 90 minutes later using a CellTiter-Glo Luminescent Cell Viability Assay (Promega, Southampton, UK) in accordance with the manufacturer's instructions. A standard curve was generated immediately before running the assay by dissolving ATP disodium salt hydrate (Sigma-Aldrich) in cell culture medium (100 pM to 100 mM).

Suppression assay
CD4 + T cells were cultured with 10 À7 M 1,25(OH) 2 D3 for 7 days as described, then live cells (7-aminoactinomycin-D-negative) were sorted based on CD73 expression using a BD FACSAria (BD Biosciences, Oxford, UK). CD73or CD73 + cells were subsequently cultured at a ratio of 0Á25 : 1 with freshly isolated autologous CellTrace violetlabelled CD4 + T cells for a further 7 days. Cell proliferation was assessed by the loss in fluorescence intensity of CellTrace Violet on an NxT attune (Life Technologies, Paisley, UK).

TGF-b bioassay
Bioactivity of TGF-b was assessed using mink lung epithelial cells transfected with a construct containing the 5 0 end of the human plasminogen activator inhibitor-1 gene fused to the firefly luciferase reporter gene (kindly donated by Prof. Daniel B Rifkin; New York University). 23 Then, 3 9 10 5 mink lung epithelial cells/ml were incubated for 14 hr at 37°C in 5% CO 2 with cell culture supernatants. Culture media were aspirated, cells were lysed and then luciferase activity (which corresponds to TGF-b activity) was determined using a Biotium Firefly Luciferase Assay Kit (Promega, Southampton, UK) in accordance with the manufacturer's instructions. Serumfree RPMI-1640 was used for these experiments due to high levels of TGF-b found in fetal calf serum.

Data analysis
Flow cytometry data were analysed using FLOWJO (Treestar Inc., Ashland, OR; version 10) and cumulative data analysis was performed in GRAPHPAD PRISM version 6Á00 for Windows (Graphpad Software Inc., San Diego, USA). Data were assessed for Gaussian distribution and the appropriate statistical test was then performed as described in the figure legends. Data are shown as mean AE standard error of mean.

1,25(OH) 2 D3 up-regulates CD4 + T-cell expression of CD73
It has previously been shown that human CD4 + T cells cultured in the presence of 10 À7 M 1,25(OH) 2 D3 express elevated levels of CD39. 14 Here this observation is extended to reveal that expression of downstream ecto-5 0nucleotidase CD73 is also significantly up-regulated at the gene (Fig. 1a) and protein level (% CD73 + cells and mean fluorescence intensity; Fig. 1b and c) following 7 or 14 days culture with 1,25(OH) 2 D3. Furthermore, a trend towards lower ATP levels 90 min after spiking cells with 500 lM ATP was observed in 1,25(OH) 2 D3-treated CD4 + T cells (P = 0Á06; see Supplementary material, Fig. S1A). CD73 + cells were also shown to be regulatory because they suppressed the proliferation of autologous CellTrace Violet-labelled CD4 + T cells (ratio of 0Á25 : 1) to a greater extent than CD73cells (lower percentage of cells divided (P < 0Á05) and division index (P = 0Á09); see Supplementary material, Figure S1B,C).

1,25(OH) 2 D3 up-regulates expression of TGFbassociated molecules
To investigate whether enhancement of CD73 expression was TGF-b dependent, we first assessed the impact of 1,25(OH) 2 D3 on components of the TGF-b pathway. After 7 and 14 days in culture alone or in the presence of the indicated concentration of 1,25(OH) 2 D3, CD4 + T-cell expression of latent tgfb2 and tgfb3 was typically undetectable by quantitative RT-PCR (data not shown). Although latent tgfb1 was highly expressed at the mRNA level, its expression was not modulated by 1,25(OH) 2 D3 treatment (Fig. 2a). In contrast, lrrc32 (gene for GARP) mRNA was significantly up-regulated by 1,25(OH) 2 D3 with a trend (P = 0Á13) towards increased expression of gp96, a chaperone for GARP, by day 14. At the protein level, 1,25(OH) 2 D3 treatment significantly increased overall expression of GARP and LAP as well as the percentage of cells co-expressing both of these molecules (Fig. 2b-d).
In order to generate active TGF-b, LAP must be cleaved. Within the immune cell compartment, this is reported to be primarily carried out by the cell surface expressed molecules thrombospondin-1, neuropilin-1 and a v integrin, 17,18 each of which were also up-regulated by 1,25(OH) 2 D3 treatment at the mRNA level, most notably after 14 days in culture (P = 0Á09, P < 0Á05 and P < 0Á05, respectively; Fig. 2e, and see Supplementary material, Fig. S2).
1,25(OH) 2 D3-mediated up-regulation of CD73, LAP and GARP occurs primarily on CD4 + Foxp3 -T cells Foxp3 is conventionally seen as the lineage defining transcription factor of natural and induced regulatory T cells, but Foxp3suppressor cells do exist. 24 As the purinergic and TGF-b pathways have been linked to the function of regulatory T cells, 9 the interaction between intranuclear Foxp3 expression and surface expression of CD73, LAP and GARP was assessed. In agreement with previous findings, no significant increase in Foxp3 expression was observed in the presence of 10 À7 M 1,25(OH) 2 D3 (Fig. 3). 3 Furthermore, 1,25(OH) 2 D3-mediated up-regulation of CD73 (Fig. 3a, b), GARP (Fig. 3c, d) and LAP (Fig. 3e, f) was principally seen on CD4 + Foxp3 -T cells as opposed to CD4 + Foxp3 + cells.
Cross-talk between CD73 and TGF-b in 1,25 (OH) 2 D3-treated CD4 + T cells Mouse studies have shown that cross-talk can occur between the adenosine and TGF-b pathways, 20,22 but whether this is true in humans and any impact of 1,25 (OH) 2 D3 remains to be ascertained. Indeed, cell surface staining showed a significant increase in the percentage of cells co-expressing CD73 with TGF-b-associated molecules LAP, GARP and neuropilin -1 following 1,25(OH) 2 D3 treatment (Fig. 4).
Furthermore, TGF-b bioactivity was elevated in supernatants harvested after 7 days of culture with 1,25 (OH) 2 D3 (P = 0Á09), as determined using mink lung epithelial cells and a luciferase assay (Fig. 5a). These observations were extended by culturing CD4 + T cells in the presence of a neutralizing TGF-b antibody or the relevant isotype control. Anti-TGF-b impaired the 1,25 (OH) 2 D3-mediated induction of CD73 at the mRNA level (P = 0Á08; Fig. 5a) and in both the percentage of CD4 + CD73 + T cells and the mean fluorescence intensity of CD73 (P < 0Á05; Fig. 5b).

Discussion
The frequency of regulatory T-cell populations is known to be enhanced by vitamin D both in vitro and in vivo. However, how these cells suppress in different environments and what markers identify them is not clear. 3,4 The focus of this work was to assess the immunomodulatory effects of 1,25(OH) 2 D3 on the induction of two complex immunoregulatory molecules, CD73 and TGF-b.
In the present study a significant up-regulation of CD73 expression in CD4 + T cells by 1,25(OH) 2 D3 was observed, complimenting the previously published finding that 1,25(OH) 2 D3 enhances surface expression of CD39. 14 Heightened expression of these two ecto-nucleotidases most likely contributes to the observed faster consumption of ATP by 1,25(OH) 2 D3-treated cells (Fig. S1A) and to the enhanced generation of downstream adenosine. 25 Studies have shown that CD73 is critical in mediating many immunosuppressive properties of cells in mice and man 26,27 and that adenosine can impair T-cell proliferation in humans. 28,29 The latter data correlate with the anti-proliferative properties of CD73 + cells that are reported both here and elsewhere. 30 As well as the traditional CD39 and CD73 pathway for ATP catabolism, a CD39-independent pathway involving CD38 and CD73 has been documented in human T cells. 31 As CD38 is known to be highly up-regulated on T cells in response to 1,25(OH) 2 D3, 3,32 we hypothesize that 1,25(OH) 2 D3 is further enhancing ATP breakdown and adenosine generation by acting via the CD39-dependent and -independent pathways. Since 10 À7 M 1,25(OH) 2 D3 has previously been shown to increase the frequency of CD4 + Foxp3 + cells in the presence of TGF-b, 8  , which together result in cells that have an enhanced capacity to both generate and present latent TGF-b. These data are in line with a study in patients with multiple sclerosis in whom vitamin D supplementation resulted in increased levels of LAP in the serum relative to those in the placebo group. 33 Although a whole host of molecules and environmental conditions such as heat and extreme pH have been implicated in activating TGF-b by cleaving LAP, a v integrins, neuropilin-1 and thrombospondin-1 are the primary mediators within the immune cell compartment; 18 here expression of each of these molecules, at least at the mRNA level, was shown to be higher in 1,25 (OH) 2 D3-treated cells, which most likely contributed to the enhanced TGF-b bioactivity observed.
In addition to its anti-inflammatory role, TGF-b is involved in tissue remodelling and therefore can have potentially negative consequences in diseases such as asthma. 34 However, 1,25(OH) 2 D3 has also been shown to inhibit the proliferation of airway smooth muscle cells. 35 Moreover, in utero vitamin D deficiency in mice leads to increased airway smooth muscle mass and airway resistance 36 and in children with severe asthma, lower levels of vitamin D were associated with increased airway smooth muscle mass. 37,38 As with many in vitro studies, the data presented here are observational. Actual levels of molecules as well as their expression relative to other mediators in vivo are critical; for example, lung remodelling in mice has been shown to result from an imbalance between TGF-b 1 and bone morphogenic protein 7 (BMP-7). 39 Whether TGF-b acts in a regulatory or tissue remodelling manner in various disease contexts is most likely dependent on the cytokine milieu as a whole and so warrants studying in vivo.
A major limitation in the study of regulatory T cells is a lack of specific cell surface markers, particularly in  humans. 40,41 Foxp3 is a widely accepted marker of many regulatory T-cell populations, but is intracellular so cells must be permeabilized in order to perform staining, which limits the isolation of pure and live cells. Since CD73, GARP and LAP are putative markers of regulatory T cells, [42][43][44] their relationship with Foxp3 was examined in our culture conditions. CD73 was expressed primarily on CD4 + Foxp3 -T cells and 1,25(OH) 2 D3-mediated induction of CD73, GARP and LAP was only clearly evident on CD4 + Foxp3 -T cells. In support of this, Foxp3 expression has been found to be independent of GARP expression in cells from HIV-positive patients. 45 Although not absolute markers, these data support the idea that CD73, GARP and LAP help to identify Foxp3regulatory cells.
Previous studies have shown that cross-talk between the TGF-b and adenosine pathways exists in mice, [20][21][22] and here we confirm this in humans too. A significant proportion of CD4 + T cells co-express CD73 with GARP, LAP and neuropilin-1 following 1,25(OH) 2 D3 treatment and by impairing TGF-b activity, the induction of CD73 by 1,25(OH) 2 D3 was reduced.
In summary, we show that 1,25(OH) 2 D3 can enhance expression of CD73 on CD4 + Foxp3 -T cells in part by up-regulating autocrine TGF-b bioactivity. These data help to further elucidate the mechanisms by which vitamin D may skew the immune system towards a more regulatory phenotype and consequently protect against various immune-mediated diseases. Further work to probe the functional consequences of these observations is warranted both in vitro and in vivo.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Figure S1. CD4 + CD73 + T cells suppress the proliferation of autologous CD4 + T cells to a greater extent than CD4 + CD73 -T cells. Figure S2. The effect of 1a,25-dihydroxyvitamin D3 (1,25(OH) 2 D3) treatment on the mRNA expression of transforming growth factor-b-associated molecules.