Induction of Treg by monocyte-derived DC modulated by vitamin D3 or dexamethasone: Differential role for PD-L1


  • Wendy W. J. Unger,

    Corresponding author
    1. Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands
    • Department of Molecular Cell Biology and Immunology, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • Sandra Laban,

    1. Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Fleur S. Kleijwegt,

    1. Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Arno R. van der Slik,

    1. Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Bart O. Roep

    Corresponding author
    1. Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands
    • Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, E3-Q, P. O. Box 9600, 2300RC Leiden, The Netherlands Fax: +31-71-5265267
    Search for more papers by this author


Specific therapy with modulated DC may restore immunological tolerance, thereby obviating the need for chronic immunosuppression in transplantation or autoimmunity. In this study we compared the tolerizing capacity of dexamethasone (Dex)- and 1α,25-dihydroxyvitamin D3 (VD3)-modulated DC. Treatment of monocytes with either VD3 or Dex resulted in DC with stable, semi-mature phenotypes compared with standard DC, with intermediate levels of co-stimulatory and MHC class II molecules, which remained unaltered after subsequent pro-inflammatory stimulation. IL-12p70 secretion was lost by VD3- and Dex-DC, whereas IL-10 secretion was unaffected. VD3-DC distinctly produced large amounts of TNF-α. Both VD3- and Dex-DC possessed the capacity to convert CD4 T cells into IL-10-secreting Treg potently suppressing the proliferation of responder T cells. However, only Treg induced by VD3-DC exhibited antigen specificity. VD3-, but not Dex-, DC expressed significant high levels of PD-L1 (programmed death-1 ligand), upon activation. Blockade of PD-L1 during priming redirected T cells to produce IFN-γ instead of IL-10 and abolished acquisition of regulatory capacity. Our findings demonstrate that both VD3- and Dex-DC possess durable but differential tolerogenic features, acting via different mechanisms. Both are potentially useful to specifically down-regulate unwanted immune responses and induce immune tolerance. These modulated DC appear suitable as adjuvant in antigen-specific clinical vaccination intervention strategies.


Restoring immunological tolerance is the “holy grail” in the fields of autoimmunity and transplantation. Current applied therapies, which include immunosuppressive drugs, do not target the cause of the disease and, in addition, are associated with considerable non-specific side effects. Therefore, it is desirable to design therapies that specifically target the immunopathogenesis.

DC are important modulators of T-cell activity 1. Depending on the type of pathogen encountered and the profile of co-stimulatory and T-cell polarizing molecules, DC drive the development of either pro-inflammatory Th type 1 (Th1), type 2 (Th2) and type 17 (Th17) cells or protective Treg 2–4. In conjunction with this, human monocyte-derived DC (moDC) can be differentiated into Th1-, Th2-, Th17- or Treg-promoting DC in vitro. Priming of moDC with microbial compounds or tissue-derived factors such as IFN-γ, prostaglandin E2 (PGE2) or IL-23 and IL-1 results in enhanced expression of MHC class II and co-stimulatory molecules and drives the development of effector Th1, Th2 and Th17 cells 5–7. It is now clear that certain immunosuppressive drugs and anti-inflammatory agents induce DC with tolerogenic properties 8–15. For example, DC treated with either dexamethasone (Dex) or the active form of vitamin D, 1α,25-dihydroxyvitamin D3 (VD3), arrest DC in a semi-mature state and prevent the up-regulation of co-stimulatory molecules and the secretion of pro-inflammatory cytokines, such as IL-12.

Treatment of murine immature bone marrow-derived DC with Dex prior to LPS-induced activation results in modulated DC that inhibit the proliferation of allo-specific T cells in both primary and secondary MLR 8, 16. Similar observations have been made with human moDC treated with Dex 10, 17. In addition, Dex-DC possess durable immaturity 10 as high IL-10 and little IL-12 production was maintained up to 5 days after removal of Dex. In vivo, application of Dex-DC prior to allograft transplantation significantly prolonged the survival of the grafts in an antigen-dependent fashion in mice 16, 18. Priming with Dex-DC leads to a decrease in the number of IFN-γ-producing cells, while increasing the number of IL-10 producing cells both in vivo and in vitro, which could point to the induction of Tr1-like Treg 16, 19. In line with this, an up-regulation of FoxP3 expression marking Treg in mice was recently observed in lymph node cells upon in vivo treatment with Dex-DC 18. However, conclusive data on functional involvement of Treg induced by Dex-DC are lacking.

Similar to Dex-DC, VD3-DC inhibit the proliferation of naive and committed T cells20–22, but, additionally, VD3-DC also change the cytokine responses of committed human autoreactive T cells 22. Besides preventing proliferation of allo-reactive T cells, VD3-DC also inhibit unwanted T-cell responses by selectively inducing apoptosis in a part of the responsive T cells 23. In contrast to Dex-DC, VD3-DC have been demonstrated to convert naive T cells into Treg, although several rounds of priming and boosting with VD3-DC were required 24. No data on in vivo function have been reported so far. In addition, it is unclear whether VD3-DC possess durable tolerogenic properties similar to Dex-DC.

Together, these data suggest that Dex- or VD3-DC could provide a means to treat autoimmune or transplant patients. To this end, it is essential to establish whether modulated DC maintain their tolerogenic properties when encountering proinflammatory signals in vivo (e.g. during an allograft reaction). In addition, an important avenue to pursue for the suppression of detrimental responses in these patients is to obtain long-lasting tissue-specific tolerance, which may be achieved via the induction of antigen-specific Treg. Therefore, in this study we compared DC generated in the presence of VD3 versus Dex to evaluate their potential usage for vaccination of transplant and/or autoimmune patients. We addressed the stability of the modulated phenotype by re-exposing the DC to other stimuli, the expression of molecules involved in T-cell polarization and migration as well as their potential to induce antigen-specific Treg.


Dex- and VD3-derived DC display features of tolerogenic DC

First, we analyzed the phenotype and T-cell stimulatory capacity of both VD3- and Dex-DC and compared these with non-treated (NT)-DC to assess whether the results obtained are specific for modulated DC. Both Dex- and VD3-DC displayed a semi-mature phenotype (Fig. 1A). In both Dex- and VD3-DC, LPS-induced up-regulation of co-stimulatory and MHC class II molecules was hampered. However, immature VD3-DC displayed higher levels of CD86 than standard and Dex-DC. Although addition of LPS resulted in highly increased expression of CD86 on NT-DC, it evoked a minor increment of CD86 on Dex-DC (Fig. 1A). By contrast, the expression on VD3-DC remained unchanged, with half of the cells exhibiting intermediate levels of CD86 and the other half expressing similar high levels as activated standard DC.

Figure 1.

Dex- and VD3-treated moDC display tolerogenic features. Monocytes were isolated from PBMC and cultured for 6 days with IL-4 and GM-CSF to obtain immature NT-DC. Dex and VD3-DC were obtained by continuous presence of 10−6 M Dex or 10−8 M VD3 in the moDC cultures, respectively. Cells were activated overnight with 100 ng LPS. (A) The expression of the monocyte and DC markers CD14, CD1a and DC-SIGN, as well as the expression of MHC class II and co-stimulatory molecules was analyzed using flow cytometry before (filled histograms) and after (solid lines) LPS activation. Dashed lines represent isotype control mAb staining. Results are shown from a representative experiment out of 15 independent experiments performed. Cytokines released by DC after LPS activation was measured from the supernatant using Luminex. (B) The ratio of IL-10 over IL-12p70 and (C) TNF-α released by NT- (open circles), Dex- (gray circles) and VD3-DC (closed circles). (D) The T-cell stimulatory capacity of all DC types was tested using an MLR. Allogeneic CD4+ T cells (105) were cultured with LPS-activated NT- (open circles), Dex- (gray circles) and VD3-DC (closed circles) at indicated ratios. After 4 days cultures were pulsed overnight with [3H]-thymidine and incorporation was determined. Data represent the mean±SD of triplicates. Results are depicted from a representative experiment out of six. * indicates a significant difference between modulated DC and NT, standard DC (p<0.001).

Typically, both modulated DC types retain the expression of CD14 upon their differentiation from monocytes. Dex-DC co-express CD1a and CD14, which conceivably resulted from later addition of Dex to the monocyte culture than VD3 (day 3 versus day 0). In our hands, addition of Dex on day 0 resulted in significant cell death, which was avoided when either VD3 was added on day 0 or Dex on day 3. The expression of DC-SIGN by all types indicates that they are true DC, rather than monocytes.

Both modulated DC subtypes showed increased IL-10 production, whereas the secretion of IL-12p70 was prevented, resulting in an anti-inflammatory cytokine profile (Fig. 1B). When analyzing other cytokines secreted by the activated DC types, VD3-DC, but not NT- and Dex-DC, secreted high amounts of TNF-α (Fig. 1C). Nevertheless, VD3- and Dex-DC displayed a reduced T-cell stimulatory capacity in a primary allogeneic MLR compared with activated NT-DC that effectively primed CD4 T-cell proliferation even at very low DC:T cell ratio (Fig. 1D).

All experiments were also performed using DC activated by CD40 ligation for 24 h, yielding similar results (data not shown).

Addition of Dex or VD3 during DC differentiation yields irreversibly modulated DC

A potential risk of using these in vitro generated Dex- and VD3-DC for intervention strategies in vivo is that they may switch to a T-cell activating phenotype when encountering pro-inflammatory signals (e.g. from auto- or allo-reactive T cells) in vivo. To address this issue, Dex- and VD3-DC were generated and activated with LPS for 24 h. Thereafter, cells were extensively washed, rested for 5 days and then re-activated by co-culture with murine fibroblasts expressing CD40L. We examined their surface marker and cytokine profile as well as their immunostimulatory capacity. Modulated DC retained the characteristic CD14 expression, whereas CD1a remained absent (Fig. 2A). In addition, the differences observed after the first activation between NT- and the modulated Dex- and VD3-DC with regard to the expression of co-stimulatory and MHC class II molecules persisted after the second activation. Similarly, the anti-inflammatory cytokine profile, reflected by a high IL-10/IL-12p70 ratio (Fig. 2B) as well as their reduced T-cell stimulatory capacity (Fig. 2C) were resistant to re-stimulation, indicating that both modulated DC types possess a durable, stable tolerogenic phenotype.

Figure 2.

Dex-DC and VD3-DC possess a stable tolerogenic phenotype. Immature NT-, Dex- and VD3-DC were generated as described in the legend to Fig. 1. After activation with LPS overnight, cells were harvested and, after extensive washing, cultured in the presence of GM-CSF. After 5 days, cells were re-activated via CD40 triggering by incubating 0.5×106 DC with 0.2×106 CD154 expressing L cells. (A) Before (filled histograms) and after (solid lines) activation, the expression of CD14, CD1a, CD80, CD86 and MHC class II was analyzed. Dashed lines represent isotype control mAb staining. (B) The ratio of IL-10 over IL-12p70 released by NT- (white bars), Dex- (gray bars) and VD3-DC (black bars) upon secondary activation was measured from the supernatant using Luminex and (C) the T-cell stimulatory capacity of all DC types was tested using an MLR as described in the Legend to Fig. 1. Results are shown from a representative experiment out of three independent experiments performed. * indicates a significant difference between modulated DC and NT, standard DC (p<0.001).

Modulated DC are equipped to migrate to inflamed tissues

To be effective in silencing reactive T cells in vivo, it is of key importance that modulated DC are capable of reaching lymphoid tissues or the inflamed site. We therefore examined the expression of several chemokine receptors. Immature Dex-, VD3- and standard NT-DC expressed significant levels of CCR4 and CCR5, but no CCR7 (data not shown and Fig. 3A). Upon activation, CCR4 and CCR5 (data not shown) were down-regulated, whereas CCR7, necessary to enter secondary lymphoid tissues, became up-regulated by all DC, albeit at reduced levels by the modulated DC types (Fig. 3A). Besides CCR5, CXCR3 and CXCR4 are also involved in migration to inflamed tissues. CXCR3 was only expressed on the surface of the modulated DC upon activation (Fig. 3). CXCR4 expression could not be detected on any DC type, either immature or mature (data not shown). DC migration studies confirmed the migratory capacity to CCL21 (ligand for CCR7) for all DC types (Fig. 3B). Only Dex-DC migrated significantly towards CXCL10/IP10 (ligand for CXCR3). Inactivated NT-, Dex- or VD3-DC did not migrate to either CCL21 or CXCL10 (data not shown).

Figure 3.

Modulated DC are equipped to migrate to inflamed and secondary lymphoid tissues. (A) The expression of CXCR3 and CCR7 was determined on NT-, Dex- and VD3-DC before (filled histograms) and after (solid lines) activation with 100 ng LPS. Dashed lines represent isotype control mAb staining. (B) Subsequently, the DC were tested for CCL21- and CXCL10-triggered migration in transwell assays. DC were added to the upper chambers at 2×105/100 μL medium. Either medium (white bars), 300 ng/mL CCL21 (gray bars) or 300 ng/mL CXCL10 (black bars) was present in the lower compartment. After 4 h of incubation, DC in lower wells were collected and their numbers were counted by FACS. Results shown are the mean+SEM of two independent experiments performed. * indicates a significant difference between migration towards specific chemokine and non-specific migration (p<0.05).

Only VD3-DC induce robust antigen-specific Treg

It has been described that CD4+ T cells that were repetitively primed by VD3- or immature DC became hyporesponsive and acquired suppressive capacity 24–26. We adapted this protocol to determine the effects of repetitive stimulation by either Dex- or VD3-DC on CD4+CD25 T cells. T cells were cultured with DC in a 10:1 ratio and after one or two rounds of stimulation their ability to suppress the proliferation of autologous CD4+ responder T cells on challenge with allogeneic mature DC was tested. The mature DC were generated from the same donor as the Dex- and VD3-DC, which allowed us to test antigen specificity. Proliferation of CD4+ T cells in a primary MLR was unaffected by T cells that had been primed with either Dex- or VD3-DC for one round (data not shown). Importantly, addition of T cells that had been primed with Dex- or VD3-DC for two rounds suppressed the response of CD4+ T cells to allogeneic mature DC (Fig. 4A, left panel: CFSE profiles; and right panel: black bars). The proliferation of responder T cells was suppressed by 75% in the presence of Treg-VD3 (p<0.00008), or 60% in the presence of Treg-Dex (p<0.0003). Addition of resting CD4+ T cells or T cells primed with NT-DC did not significantly affect the proliferation of the responder cells.

Figure 4.

CD4 T cells differentiate into IL-10-secreting antigen-specific Treg or non-antigen-specific Treg upon priming with VD3- or Dex-DC. Immature NT-, Dex- and VD3-DC (donor A) were generated as described in the legend to Fig. 1. DC were activated and co-cultured with haplo-identical CD4+ T cells (donor B) at a 1:10 ratio. After a second round of priming with modulated DC, the suppressive capacity and cytokine production of the obtained Treg was analyzed. (A) The suppressive capacity of Treg-Dex or Treg-VD3 was examined by adding them to CFSE-labeled responder CD4+ T cells (donor A) that were primed by allogeneic DC (donor B) at a 1:1 ratio. CD4+ T cells that have been primed with activated NT-DC were added as control for crowding. CFSE profiles of gated responder cells (left panel) and the percent inhibition of responder T-cell proliferation (right panel, black bars) are shown. To assess specificity of the induced Treg, activated third party DC (right panel, hatched bars) were used in the primary MLR. Results refer to mean±standard error from one representative out of six separate experiments (each performed in triplicate). (B) Treg-VD3 were tested for specificity in a primary MLR in which either specific DC (black bars); third party DC (hatched bars) or both specific and third party DC at a 1:1 ratio (gray bars) were present. In control cultures CD4+CD25 T cells were added. After 4 days, cells were stained for CD4 and proliferation of CFSE-labeled responder T cells was analyzed on a flow cytometer. Data are representative of two independent experiments. (C) Treg-Dex and Treg-VD3 were incubated overnight with specific, LPS-activated DC at a 10:1 ratio or left untreated. Antigen-specific IFN-γ (gray bars) and IL-10 (black bars) production was examined using a cytokine secretion assay. The mean percentage of cytokine-positive CD4+ T cells was determined using flow cytometry. Results shown are the mean percentage±SEM of cytokine-positive cells of triplicates. (D) Suppression of responder T-cell proliferation when Treg were separated from responder T cells by Transwell inserts (hatched bars) were compared with assays that are not separated by inserts (black bars) at the suppressor to responder ratio of 1:1. LPS-activated DC were present in both the upper and lower chambers of the transwell insert at a ratio of 1:10 DC to T cell. Results of one representative experiment are shown. * indicates a significant difference between modulated DC and NT, standard DC (p<0.01).

Next, we assessed whether the suppressive capacity of Treg-Dex and Treg-VD3 cells was maintained when using mature DC of a third party donor in the primary MLR. Suppression of responder T-cell proliferation by Treg-Dex was still significant despite the presence of third party DC (p<0.003; Fig. 4A, right panel: hatched bars). By contrast, the suppressive capacity of Treg-VD3 cells was almost completely abolished as proliferation of responder T cells was less different from those in the presence of control activated CD4+ T cells (p<0.05). However, when both specific DC and third party DC were present at a 1:1 ratio in the primary MLR, Treg-VD3 were able to suppress responder T-cell proliferation (Fig. 4B). In light of our finding that Treg-Dex suppressed responder T-cell proliferation in an antigen-non-specific fashion, we assessed whether Treg-Dex, rather than Treg-VD3, produced IL-10 that is known to have broad, non-specific immunomodulatory effects. Indeed, upon specific activation CD4+ Treg-Dex produced IL-10 (Fig. 4C, black bars). However, IL-10-producing cells were also present in Treg-VD3. Both CD4+ Treg populations also contained IFN-γ-producing cells (Fig. 4C, grey bars). The effector T-cell population hardly contained IL-10-producing cells, whereas the number of IFN-γ-producing cells was not significantly different from those present in both Treg populations.

To determine whether Treg-Dex and Treg-VD3 mediated suppression via anti-inflammatory cytokines, we performed suppression assays using transwells. Treg-Dex- and Treg-VD3-mediated suppression involves cell–cell contact, as the suppressive capacity was lost when they were separated from responder T cells (Fig. 4D).

VD3-DC express high levels of PD-L1

It is conceivable that a complex series of events occurring during the interaction between T cells and the Dex- or VD3-DC mediates the induction of Treg. To investigate this, the expression of inhibitory T-cell polarizing molecules and factors were examined.

We first tested for the contribution of IDO, an enzyme that degrades the essential T-cell growth factor tryptophan, shown to be involved in allograft-specific tolerance 27. The amount of the tryptophan catabolite kynurenine generated is a correlate of IDO activity. Therefore, DC were cultured in the presence of 1 mM tryptophan and, after activation by LPS, the supernatants were analyzed for the amount of tryptophan and kynurenine. No significant differences in the amount of kynurenine were detectable in the supernatant of NT-DC versus modulated DC, implying that the induction of hyporesponsiveness by the modulated DC is not mediated via IDO (Table 1).

Table 1. IDO activity is similar in modulated and standard DCa)
Kynurenine (μM) in supernatant after stimulation
 24 h48 h24 h48 h
  • a)

    a) Immature NT-, Dex- and VD3-DC were plated at 0.5×106 cells/mL in the presence of 1 uM tryptophan and activated with either LPS or CD40L-expressing L-cells. Sups were harvested 24 or 48 h later and the amounts of tryptophan and kynurenine were measured using HPLC.


Analysis of the expression of B7 family members that are known to negatively regulate T-cell responses 28–31 revealed that activated VD3-DC expressed the highest levels of PD-L1 (Fig. 5A), whereas these levels did not differ between Dex-DC and NT-DC. Expression of PD-L2, B7-H3, B7-H4 and ICOS-L was similar in all three DC types (Fig. 5A and data not shown). In addition, analysis of the TNF-R family members OX40L and 4-1BBL revealed equal expression of OX40L by all DC types, whereas 4-1BBL was expressed at lower levels by VD3-DC than NT- and Dex-DC. In line with previous findings on VD3-DC 24, we observed a high expression of the inhibitory receptor Ig-like transcript (ILT)-3 on VD3-DC. By contrast, the expression of ILT-3 on Dex-DC did not differ from the expression on NT-DC (Fig. 5A).

Figure 5.

PD-L1 is crucial for generation of Treg by VD3-DC. (A) Immature NT-, Dex- and VD3-DC were generated as described in the legend to Fig. 1. After 24 h activation with 100 ng LPS, the expression of PD-L1, PD-L2, B7-H4, 4-1BBL, OX40L and ILT-3 was examined on NT- (black dotted lines), Dex- (gray dashed lines) and VD3-DC (black bold lines) using flow cytometry. (B) The ratio of PD-L1/CD86 or PD-L2/CD86 for NT-(white bars), Dex- (gray bars) and VD3-(black bars) DC was determined by dividing the MFI of PD-L1 expression by the MFI of CD86 expression. Results shown are the mean from 10 independent experiments. (C–E) VD3-DC were co-cultured with haplo-identical CD4+ cells for 4 days in the presence or absence of anti-PD-L1 mAb or isotype control. (C) Specific IFN-γ (gray bars) and IL-10 (black bars) production was assessed after overnight priming with LPS-activated DC using the cytokine secretion assay, (D) the expression of the Th1-differentiation factor T-bet was determined using quantitative RT-PCR. Experiments were performed in triplicate; mean±standard error from two separate experiments are shown. (E) Suppressive activity of the T cells was examined by addition to CFSE-labeled CD4+ T cells primed with allogeneic DC. After 4 days of culture, the proliferation of the responder cells was determined using flow cytometry. CFSE profiles of responder cells in the presence of ctrl T cells or T cells primed with VD3-DC in the presence of anti-PD-L1 mAb or isotype control mAb are shown in the right panel. From these plots, percent inhibition of responder T-cell proliferation was calculated (left panel). A representative experiment out of three performed is shown.

Role of PD-L1 in Treg induction by VD3-DC

A high ratio of PD-L1/CD86 has been correlated with reduced levels of T-cell allostimulatory ability 32. Indeed VD3-DC exhibit higher PD-L1/CD86 ratios than NT-DC and Dex-DC (Fig. 5B). We did not observe differences in PD-L2/CD86 ratios, which is in concurrence with the similar expression of PD-L2 on all DC subsets. In addition, PD-L1 has been demonstrated by a number of studies to be involved in tolerance induction 33–36. Since 4-1BBL, of which signaling has shown to suppress pathogenic CD4 T-cell responses in several autoimmune disease models 37–39, is expressed at reduced levels by VD3-DC and ILT-3 on VD3-DC has been demonstrated to be dispensable for Treg induction, we examined whether PD-L1/PD1 interactions are involved in the conversion of resting CD4+ T cells into Treg by VD3-DC. Importantly, the higher surface expression of PD-L1 on VD3-DC is retained even after a second inflammatory stimulation (data not shown).

To this end, T cells were primed by VD3-DC in the presence of blocking PD-L1 Ab for two rounds. To avoid confounding influences of Fc-receptor triggering and non-specific interference, parallel cultures were incubated with isotype control Ab. PD-L1 blockade drove T cells to effector phenotype, as revealed from high levels of IFN-γ and low levels of IL-10 (Fig. 5C) as well as an increase in the mRNA levels of T-bet, the transcription factor that directs Th1-lineage commitment (Fig. 5D). Concordantly, these T cells were incapable of suppressing the response of CD4+ T cells to allogeneic mature DC (Fig. 5E), whereas those primed in the presence of a control Ab effectively suppressed the proliferation of the responder-cell population. Blocking PD-L1 during T-cell priming by Dex-DC did not affect Treg induction (data not shown).

In sum, the induction of Treg by VD3-DC is crucially dependent on signaling via PD-L1 as blocking of this co-stimulatory receptor during T-cell priming with VD3-DC abolished Treg generation.


We investigated the differences between Dex- and VD3-DC in terms of stability, expression of migration and T-cell polarization molecules, and their functional ability to induce Treg from resting CD4+ T cells. Detailed phenotypic analysis and comparison with standard NT-DC showed that addition of Dex or VD3 during differentiation of DC yields DC with stable and durable tolerogenic features. Subtle differences between VD3- and Dex-DC were identified, with VD3-DC exhibiting a higher expression of CD86 and PD-L1 and secreting higher amounts of TNF-α compared with Dex-DC. Resting CD4+ T cells that were primed by either Dex- and VD3-DC differentiated into IL-10-producing Tr1-like Treg, which were capable of suppressing the proliferation of responder T cells in a primary MLR. However, only Treg induced by VD3-DC exhibited antigen specificity. Finally, we revealed a crucial role for PD-L1 in the induction of Treg by VD3-DC. Blockade of PD-L1 during priming redirected T cells to produce IFN-γ instead of IL-10 and abolished acquisition of regulatory capacity.

These data confirm and extend previous studies 8, 14, 15, 17, 20, 22 showing that addition of either Dex or VD3 during monocyte differentiation into DC yielded modulated DC with a semi-mature phenotype and reporting higher expression of CD86, PD-L1 and CCR7 by VD3-DC than Dex-DC, as well as secretion of high amounts of TNF-α and IL-6 (Chamorro et al., submitted) by VD3-DC compared with both Dex- and NT-DC.

In vivo application in patients faces strong, repetitive pro-inflammatory triggering. It is, therefore, of great importance that the VD3- or Dex-DC do not divert into DC with strong immunostimulatory capacities upon injection. In the present study we show that DC generated in the presence of either Dex or VD3 retain a stable immunomodulatory phenotype and function, even upon repetitive triggering to mimic activation signals that potentially might deviate their modulated phenotype and function in vivo. It is equally important that activating stimuli other than LPS yield similar results. We mimicked in vivo DC activation by activated T cells using CD40 triggering. Indeed, similar results showing preservation of the anti-inflammatory phenotype were obtained.

Our observation that VD3-DC secrete large amounts of TNF-α is not necessarily a concern for using these DC as clinical intervention therapy. We have clearly demonstrated that these inhibit the proliferation of autoreactive T cells 22 and can prime CD4+ T cells such that they differentiate into Treg (this paper). Although this is established in a controlled in vitro system, it has been shown that TNF-α can actually be beneficial in vivo as injection of recombinant TNF-α in adult NOD mice resulted in a delayed onset and a decreased incidence of the disease 40. By contrast, blocking TNF-α, which is beneficial in rheumatoid arthritis, 41 exacerbated disease in NOD mice 40 and in systemic lupus erythematosus patients 42, 43.

Repetitive priming of CD4+ T cells with either immature or IL-10- or VD3-modulated DC results in the generation of IL-10-producing Treg 24–26. We confirm and extend these studies by showing that repetitive priming of resting CD4+ T cells with VD3- or Dex-DC results in the induction of Treg. Two rounds of priming with either modulated DC sufficed to obtain robust Treg. These Treg potently suppressed the proliferation of allo-specific CD4+CD25 T cells activated by activated, immunostimulatory DC via a cell-contact-dependent mechanism. T-cell populations obtained after priming with the modulated DC also consisted of IFN-γ-producing T cells. Although this might represent a subpopulation of effector T cells, we favor the interpretation that IFN-γ is secreted by the Treg themselves, which might be beneficial in propagating tolerance. In a skin-allograft model, antigen-specific activation of Treg evoked a rapid and transient IFN-γ production, which was essential for local control of the immune response 44. Notably, Tr1 cells defined by their ability to produce high levels of IL-10 and TGF-β produce normal levels of IFN-γ 45.

An important finding of our study is that Treg induced by VD3-DC exhibit a strong antigen-specific component. This suggests that frequent and specific interactions between T cells and the VD3-DC have to occur. Induction of antigen-specific Treg is an advantage for clinical application of VD3-DC with the aim of specific, rather than general immune suppression, to avoid collateral damage by infection or tumors. Alternatively, these VD3-DC may be loaded with known epitopes of β-cell-specific autoreactive T cells to treat T1D patients. Here, the VD3-DC elicit dual action: direct inhibition of autoreactive T cells (proliferation and cytokine production)22 and selective induction of apoptosis 23, and indirect suppression via induction of antigen-specific Treg. The absence of specificity of Treg-Dex does not exclude their clinical application. These might be applied in situations when the specific antigens are unknown, in an autoimmune setting where epitope spreading has occurred. Repetitive injections of Dex-DC may invoke antigen specificity in vivo and this is subject of future studies.

The expression of the suitable migration markers as well as the ability of Dex-DC to migrate towards CCL21 and CXCL10 (IP10) implies that they might accomplish a binary function in vivo: CCR7 expression enables their migration to lymphoid organs, where conversion of naive T cells into Treg will occur, while CXCR3 expression may guide these cells to the inflammatory lesion in the pancreas to counteract autoreactive T cells 46. Distressed pancreatic islets have been shown to produce CXCL10 46. Indeed, blockade of IP10 in mice abrogated autoimmune diabetes development by interfering with trafficking of autoreactive cells to the pancreas 46, while human islet autoreactive T cells expressing CXRC3 injected in NOD/scid mice homed to pancreatic tissue after in vivo activation by islet antigen 47. By contrast, VD3-DC showing migratory activity to CCL21, will likely end up in secondary lymphoid organs following injection, where they can divert naive T cells into antigen-specific Treg. The modulated DC did not migrate without activation by LPS, which is in line with a recent study reporting the necessity of LPS activation of tolerogenic DC to promote their migratory activity and to induce tolerogenic characteristics 48.

It is now becoming clear that the tolerogenicity of Dex- and VD3-DC is not the result of mere inhibition of changes brought about by differentiation and activation. Recent publications demonstrate that both VD3 and Dex also enhance the expression of several target genes 49, 50. Using micro-array analysis, it has been shown that VD3 regulates a large set of its targets in moDC cultures autonomously and not via inhibition of differentiation and maturation, leading to the tolerogenic state. Similar properties may be attributed to Dex. In addition, both VD3 and Dex exert regulatory effects on the NF-κB pathway, resulting in the reduced production of IL-12 as well as the expression of MHC class II and co-stimulatory molecules. However, Dex and VD3 regulate NF-κB activity at different levels 50, which might explain differences seen between DC treated by either compound. Treatment of tolerogenic myeloid DC with VD3 resulted in increased CCL22 and decreased CCL17 production. By contrast, Dex treatment resulted in the reciprokal effect on myeloid DC. This complex difference of NF-κB regulation between Dex and VD3 may also explain the specific increased expression of PD-L1 on VD3 DC. PD-L1 on VD3-DC is crucial for the acquisition of Treg function by CD4+ T cells, as blockade of PD-L1 on VD3-DC during T-cell priming resulted in the generation of Th1-like T cells incapable of suppressing alloreactive T-cell proliferation and producing IL-10. Several studies report an association of high PD-L1 expression combined with low expression of positive co-stimulatory molecules with tolerogenic function of DC 51–53. Tolerogenic plasmacytoid DC in the peripheral blood of clinically tolerant liver transplant patients, who were off immunosuppression, exhibit a higher PD-L1/CD86 ratio than those in non-tolerant patients, which has been correlated with elevated numbers of CD4+CD25hi Treg in these patients 51. Very recently, cytomegalovirus was shown to evade the immune system by targeting DC to maintain high levels of PD-L1 while down-regulating positive co-stimulatory molecules and MHC on their surface 54. This prevented naive T cells from acquiring effector function and arrested the antiviral T-cell response. Instead, these T cells became anergic.

These data suggest that PD-L1/PD-1 interactions impair the generation of effector T cells in favor of Treg formation. We indeed observed complete reversal of function and cytokine profile from Treg and IL-10 secretion to effector T cells producing IFN-γ, when blocking PD-L1 in co-cultures with VD3-DC and CD4+ T cells. It remains unclear what the underlying mechanism is. One explanation may be reverse signaling by PD-L1 into DC. Triggering PD-L1 on DC using sPD-1-Ig resulted in decreased expression of the positive co-stimulatory molecules CD80, CD86 and CD40 and increased IL-10 production 55. We propose that ligation of PD-L1 on VD3-DC triggers local secretion of high amounts of IL-10 that affect activation and surface marker expression of contacting T cells.

In sum, our findings demonstrate that both VD3- and Dex-DC possess durable, but distinct tolerogenic features, acting via different mechanisms. Modulated DC are potentially useful to specifically down-regulate unwanted immune responses and induce immune tolerance in transplantation or autoimmunity.

Materials and methods

Generation of human DC

Human HLA-DR3+ or -DR4+ peripheral blood mononuclear cells were isolated by Ficoll gradient from HLA-typed buffy coats, obtained from healthy blood donors (n=25).

Monocytes were purified using CD14-MicroBeads according to supplier's protocol (Miltenyi Biotech, Amsterdam, The Netherlands) and seeded in 6-well tissue culture plates (Corning Costar, Cambridge, MA, USA) at a density of 2×106 cells per well and cultured for 6 days in RPMI-1640 medium supplemented with 8% FCS (Greiner, Wemmel, Belgium), recombinant human IL-4 (500 U/mL, Invitrogen, Breda, The Netherlands), and recombinant human GM-CSF (800 U/mL; Invitrogen). On day 3, culture medium including supplements was refreshed. On day 6, the resulting immature DC were activated either by the addition of 100 ng/mL LPS (Sigma Aldrich Chemie, Zwijndrecht, The Netherlands) or via CD40 triggering by incubating 0.5×106 DC with 0.2×106 CD154-expressing L cells 56. After 24 or 48 h, activated DC were harvested for further analysis.

Addition of VD3 and Dex to DC cultures

VD3, a generous gift from Dr. M. Sno (Solvay Pharmaceuticals, Weesp, the Netherlands), was added at the beginning of the culture at a final concentration of 10−8 M and refreshed at day 3. Dex (Sigma Aldrich) was added on day 3 of culture at a final concentration of 10−6 M. Treatment of moDC cultures with Dex or VD3 resulted in a lower yield of moDC. However, the viability of the recovered moDC was not affected and similar for all treatments used.

Ab and flow cytometry

The following fluorochrome-conjugated Ab for flow cytometry were used: PE-conjugated Ab: anti-CD25 (clone MA251), anti-HLA-DR (clone G46-6), anti-CD80 (clone L307.4), anti-CD86 (clone IT2.2), anti-CD1a (clone HI-149), anti-CD14 (clone M5E2), anti-CD83 (clone HB15e), anti-CD40 (clone 5C3), anti-CD274 (PD-L1; clone MIH1), anti-CD273 (PD-L2; clone MIH18), anti-CCR4 (clone 1G1), anti-CCR5 (clone T21/8) and anti-CXCR3 (clone 1C6). PerCp-conjugated anti-CCR7 (clone 3D12) mAb, Allophycocyanin-conjugated anti-CD4 (clone SK3) mAb.

All Ab were obtained from BD Pharmingen (San Diego, CA, USA), except for anti-PD-L1 and PD-L2 Ab, which were obtained from e-Bioscience (Emelca Bioscience, Breda, The Netherlands). Unconjugated blocking anti-PD-L1 (clone MIH1) and mouse IgG1,κ isotype control Ab were obtained from e-Bioscience. Flow cytometric staining was analyzed on a FACS Calibur (Becton Dickinson).

Cytokine analysis

Culture supernatants were analyzed for the presence of cytokines using either the Cytometric Bead Array (human Inflammation kit, BD Pharmingen) or Luminex bead array (Biorad, Veenendaal, The Netherlands). Direct ex vivo cytokine secretion was analyzed using IFN-γ and IL-10 capture and detection Ab (Miltenyi Biotech) according to the manufacturer's instructions.

Migration assays

DC were cultured overnight in the presence or absence of 100 ng LPS. The next day, cells were harvested and chemotaxis of moDC was measured by migration through a polycarbonate filter of 8 μm pore size in 24-well transwell chambers (Corning Costar). RPMI/5% FCS (600 μL) containing indicated doses of CCL21, IP10, RANTES (all obtained from Immunotools, Friesoythe, Germany) or medium alone as a control for spontaneous migration was added to the lower chamber; 2×105 DC (100 μL) were added to the upper chamber and were incubated for 4 h at 37°C. A 500-μL aliquot of the cells that migrated to the bottom chamber was taken and 50 000 Flow-Count Fluorospheres (Coulter, Miami, FL, USA) were added. For each sample 10 000 microbeads were acquired, facilitating calculation of the total number of migrated cells. Values represent the mean number of spontaneously and specifically migrated cells±SEM of two independent experiments.

T-cell differentiation in vitro

CD4+ T cells were isolated from PBMC using the negative selection kit (Dynal). To deplete CD4+CD25+ T cells, purified CD4+ T cells were stained with CD25 MicroBeads (Miltenyi Biotech) according to the supplier's protocol. A total of 1×106 purified CD4+CD25 T cells were co-cultured with modulated DC from a haplo-identical donor at a 1:10 ratio in IMDM supplemented with 10% human serum. After 5 days, T cells were recovered and rested for 2 days in the presence of IL-7 (10 ng/mL) and IL-15 (5 ng/mL). On day 7 after primary stimulation, the recovered CD4+ T cells were re-stimulated under the same conditions. Alternatively, CD4+ T cells were co-cultured with activated NT DC for 5 days, recovered and rested for 2 days. T cells stimulated repeatedly with Dex- or VD3-DC are referred to as Treg-Dex and Treg-VD3 cells, respectively. Those stimulated with NT-DC are referred to as control T cells and used in suppression assays to control for crowding and nutrient consumption.

Suppression assay

To assess whether CD4+ Treg were generated after two rounds of culture with modulated DC, we examined the suppressive capacity. Therefore, naive CD4+CD25 T cells (donor A) were labeled with 1 μM CFSE and cultured in the presence of activated NT-DC (donor B) at a 1:10 ratio in 96-well round bottomed plates, coated with 0.1 μg/mL anti-human CD3 mAb (clone UCHT1, BD Bioscience). Treg (donor A) were added at a 1:1 ratio to responder T cells. Specificity of Treg was assessed by using mature NT-DC from a non-related third party (donor C). To investigate if the Treg-mediated suppression of responder T-cell proliferation is mediated through soluble factor(s), transwell experiments were performed in 24-well dual chamber plates (6.5 mm diameter, polycarbonate membranes with pore size of 0.3 μm, Costar Corning, NY, USA). A total of 105 CFSE-labeled responder T cells (donor A) were placed in the lower chambers together with activated NT-DC (donor B) at a 1:10 ratio. To the upper chamber either Treg-Dex, Treg-VD3 or control activated T cells (105, donor A) together with activated NT-DC (donor B) at a 1:10 ratio were added. After 4 days, cells were recovered, stained for CD4 and CFSE dilution of CD4+ responder cells was analyzed on a FACS Calibur. Just prior to the analysis, 10 000 Flow-Count Fluorospheres were added. For each sample 5000 microbeads were acquired, facilitating comparison between samples.

Quantification of cell division

Analysis of cell division was conducted as described 57, with the number of undivided cohorts obtained by dividing the number of events at each cell division (n) by 2n. The proliferative index was obtained by dividing the total cellular events at 0–x divisions by the total number of undivided cohorts for each condition. Thereafter, the proliferative index of responder T cells in the absence of any other T-cell population was set at 100% to calculate the mean percentage of proliferation of responder T cells in the presence of Treg or control cells.

Quantitative PCR

RNA was extracted using RNABee according to the manufacturer's instructions. First strand cDNA was synthesized from 1 μg total RNA using Superscript II (Invitrogen, Paisley, UK) and oligo-dT. Real-time quantitative PCR was performed using the iCycler (Biorad, UK) based on specific primers and general fluorescence detection with SYBR Green. 18-S rRNA and GAPDH were used to control for sample loading and to allow for normalization between samples. Specific primers for T-bet, 18-S and GAPDH were derived from Sigma Aldrich.

IDO enzymatic activity

To assess IDO activity, the amount of the trypthophan catabolite L-kynurenine in culture supernatants was determined using a previously described HPLC method 58. Immature DC were incubated with or without 100 ng LPS in the presence of 1 mM L-Tryp for 18 h. Sups were harvested and proteins were denatured by addition of acetonitrile. Subsequently, supernatants were freeze-dried, pellets were dissolved in mQ and 100 μL was injected onto a C-18-column.

Statistical analysis

Data are expressed as the mean±SD, unless stated otherwise. Results of experiments were analyzed either using the non-parametric Mann–Whitney U test or unpaired Student's t test. Values were considered to be significantly different when p<0.05.


The authors thank Dr. D. Roelen and Dr. F. Ossendorp for critically reading the manuscript. This work has been supported in part by a JDRF grant (7-2005-877) for W.W.J.U. and S.L. and a VICI award (ZonMW) for B.O.R. and F.S.K.

Conflict of interest: The authors declare no financial or commercial conflict of interest.