Passive transfer of flt-3L-derived dendritic cells delays diabetes development in NOD mice and associates with early production of interleukin (IL)-4 and IL-10 in the spleen of recipient mice

Abstract

CD11c⁺/CD11b⁺dendritic cells (DC) with high levels of major histocompatibility complex (MHC) class II and co-stimulatory molecules have been derived from spleen cells cultured with granulocyte-macrophage colony stimulating factor (GM-CSF) + flt-3L + interleukin (IL)-6 (flt-3L-DC). Investigating in vivo the function of DC in non-obese diabetic mice (NOD), we showed that a single injection of this in vitro-derived subset of DC prevents the development of diabetes into prediabetic female mice. In contrast, DC derived from bone marrow cells cultured with GM-CSF + IL-4 [bone marrow (BM)-DC] induced no protection. Moreover, protection against diabetes following injection of flt-3L-DC was associated with IL-4 and IL-10 production in the spleen and the pancreatic lymph nodes of recipient mice, indicating that this DC population is able to polarize the immune response towards a Th2 pathway. As we shown previously, NOD BM-DC exhibit an enhanced capacity to produce IL-12p70 in response to lipopolysaccharide (LPS) and anti-CD40 stimulation compared to BM-DC from control mice. In contrast, NOD flt-3L-DC, as their control mouse counterpart, produced no IL-12p70 to these stimuli. Our findings show that a subset of DC, characterized by a mature phenotype and the absence of IL-12p70 production can be derived from NOD mouse spleen favouring IL-4 and IL-10 regulatory responses and protection from diabetes development.

INTRODUCTION

Insulin-dependent diabetes mellitus is a T cell-mediated autoimmune disease characterized by the destruction of pancreatic insulin-secreting β-cells. In the non-obese diabetic (NOD) mouse, defects in the regulatory T cell network that maintains peripheral tolerance to autoantigens possibly contribute to disease development [1–5]. Because of their key role in initiating immune responses, dendritic cells (DC) are expected to be involved in the autoimmune process either by failure to generate regulatory T cells and/or by inappropriate activation of effector T cells [6–9]. In addition, DC have been described to be an early component of the islet infiltration in the NOD mouse [10,11].

In 1992, Clare-Salzler et al. reported the use of DC in preventing the development of diabetes in NOD mice [12]. In this study, DC derived from pancreatic lymph nodes were able to induce protection, while DC derived from other lymphoid organs were not. According to the authors, the protection was due to tolerogenic presentation of islet antigens by transferred DC. However, the phenotype and the pattern of cytokine production of these DC were not described, due probably to the lack of specific reagents to characterize DC or to the small number of cells collected from pancreatic lymph nodes. A protective effect of NOD DC against the development of diabetes has also been reported following injection of ex-vivo IFN-γ-stimulated or human-γ-globulin-pulsed splenic DC or bone marrow-derived DC, suggesting that a protective effect can depend on the maturation/activation state rather than the tissue origin of DC [13–15].

Flt-3L induces a dramatic increase of all DC subsets when injected in vivo in the mouse [16,17] and, in vitro, this cytokine is able to promote the development of high numbers of functionally mature DC [18]. Therefore, we investigated the features of NOD DC derived in vitro in the presence of flt-3L. Herein, using spleen cells cultured during 3 weeks with granulocyte-macrophage colony stimulating factor (GM-CSF) + flt-3L + interleukin (IL)-6 (flt-3L-DC), we isolated DC with a high potential to modulate diabetes development following a single injection into young prediabetic NOD female mice.

MATERIALS AND METHODS

Mice

NOD (H-2^g7) mice were bred in our animal facilities under specific pathogen-free conditions and were checked every 6 months for bacterial, viral and parasitic infections. The spontaneous incidence of diabetes in our colony reaches 85% in females and 45% in males by postnatal week 35. All experiments were performed in 7–8-week-old NOD female mice, except for transfer experiments (see below). DBA/2 (H-2^d), C57BL/6 (H-2^b) and CBA (H-2^k) mice were purchased from Iffa Credo (l’Arbresle, France).

Generation of DC

DC were derived from splenic progenitors as described [18]. Briefly, after ammonium chloride treatment to remove erythrocytes, unfractionated nucleated splenocytes were cultured in complete Iscove's modified Dulbecco's medium (IMDM) (supplemented with 12·5% heat-inactivated fetal calf serum (FCS), 100 U/ml penicillin, 100 µg/ml streptomycin, essential amino acids and sodium pyruvate) with recombinant human flt-3L (Tebu, France), murine rGM-CSF (R&D Systems, UK) and recombinant human IL-6 (TEBU). Splenocytes were cultured for 6 days with recombinant GM-CSF (rGM-CSF) 1 ng/ml + flt-3L 50 ng/ml + IL-6 25 ng/ml at 7·5 × 10⁵ cells/ml. Beyond day 6, cells were reseeded every 4–5 days at 3–6 × 10⁵ cells/ml, with rGM-CSF 1 ng/ml + flt-3L 30 ng/ml. DC were recovered after 3 weeks of culture and referred to in the text as flt-3L-DC.

Bone marrow (BM)-DC were obtained as described [19,20] with slight modifications. Briefly, bone marrow cell suspensions from 7–8-week-old-female mice were depleted from lymphocytes with a cocktail of antibodies (anti-mouse CD4, clone RL172, antimouse CD8, clone TIB 105 and anti-mouse B cell, clone B220, kindly provided by Dr Laurence Zitvogel), and rabbit complement (Sigma, St Louis, MO, USA). After overnight culture at 37°C in complete RPMI medium (RPMI-1640 supplemented with 10% FCS, 2 mm l-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, essential amino acids and sodium pyruvate), non-adherent cells were recovered and cultured for 7 days in complete RPMI medium with 1 ng/ml murine rGM-CSF (R&D Systems) in combination with 100 U/ml rIL-4 (TEBU, Le Perray en Yveline, France). Both BM-DC and flt-3L-DC were purified with CD11c magnetic beads (Miltenyi, Germany). The average purity of DC (CD11c⁺ cells) was >95% and viability (propidium iodide negative) was >95%.

Cytokines produced by activated DC were detected in the culture supernatants using enzyme-linked immunosorbent assay (ELISA) (see below). BM-DC and flt-3L-DC (2 × 10⁵ cells/ml) were stimulated with 1 µg/ml lipopolysacharide (LPS) (Escherichia coli, Sigma, St Louis, MO, USA) and 1 µg/ml agonist anti-CD40 antibody (clone 3/23, Pharmingen, San Diego, CA, USA) for 16 or 48 h.

Generation of B LPS

Total spleen cells from 7–8-week-old female NOD mice were cultured for 48 h in complete RPMI medium with 1 µg/ml LPS. The activated B cells were recovered on density gradient (Ficoll-Paque, Amersham Pharmacia Biotech, France).

Transfer protocol

BM-DC and flt-3L-DC (purity > 95%) were washed three times in PBS and injected i.v. into 5-week-old NOD females (7–15 mice/group). Each mouse received a single injection of 5 × 10⁵ cells. Age- and sex-matched NOD mice receiving a single injection of splenocytes from 4-week-old NOD female mice were used as negative control. Mice were monitored for glycosuria twice a week (Glucotest, Boehringer Mannheim, France). Glycosuric mice were checked for fasting glycaemia by using test strips and a colourimetric assay (Glucotrend, Boehringer Mannheim). Diabetes was diagnosed when fasting glycaemia was over 3 g/l twice in a 24-h interval. Kaplan–Meier actuarial graph survival and log-rank tests were used for statistical analysis of diabetes development.

In vitro assays

Cells were isolated from the spleen and from the pancreatic, mesenteric and peripheral lymph nodes (LN) of control, BM-DC- or flt-3L-DC-treated mice (3–5 mice/group) on day 3 and/or day 7 post-transfer. In some experiments, an additional control group of mice receiving a single i.v. injection of B LPS (5 × 10⁵ cells, purity > 95%) was used. Cells (1·5 × 10⁶/ml) were cultured for 3 days in 96-well culture plates in triplicate in complete RPMI-1640 supplemented with 5% FCS without stimulation or in culture wells coated with 3 µg/ml anti-CD3 antibody (clone 2C11, our own production). IL-4, IL-10 and IFN-γ production was assayed in culture supernatants using ELISA.

Immunostaining and flow cytometric analysis

Cells were harvested and washed twice in PBS 3% FCS. After incubation with unlabelled anti-FcRIIγ antibody (clone 2·4G2, PharMingen, San Diego, CA, USA) to avoid non-specific binding to Fc-receptors, DC were double-stained with biotinylated or FITC-anti-CD11c antibody (clone HL3, Pharmingen) and one of the following monoclonal antibodies. PE-anti-CD11b (clone M1/70·15, Caltag), -anti-CD80 (clone RMMP-2, Caltag), -anti-CD86 (clone RMMP-1, Caltag), -anti-CD40 (clone 3/23, Caltag), -anti-CD8α (clone CT-CD8α, Caltag), anti-H-2K^d (clone SF1-1·1, Pharmingen), FITC-anti-mouse macrophage F4/80 antigen (clone F4/80, Caltag), -anti-Gr-1 (clone RB6–8C5, Caltag), -anti-B220 (clone RA3–6B2, Caltag) and biotinylated anti-I-A^b&d (clone 28–16–8S, Caltag, Burlingame, CA, USA) or anti-I-A^k&g7 (clone 10-3-6, Pharmingen) Ab, + PE-Streptavidin (Caltag) were used. Control isotype rat IgG2a, IgG2b, hamster IgG, mouse IgG2a, IgG2b for flow cytometry were purchased from Tebu (France). Cell viability was determined using propidium iodide. Flow cytometry analysis were performed on a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA, USA) using CellQuest software.

Elisa

IL-12p70 and IL-4 production was determined using a mouse IL-12p70 ELISA kit (Pharmingen, San Diego, CA, USA) and a mouse IL-4 ELISA kit (R&D Systems, UK) according to the manufacturer's recommendations. IL-10 and IFN-γ production was determined as follows. Briefly, 50 µl of standard recombinant cytokine (R&D Systems) or culture DC supernatant were incubated in anti-mouse IL-10 (5 µg/ml) (clone JES5–2A5, our own production) or in anti-mouse IFN-γ (2 µg/ml) (clone AN18, our own production) coated wells for 2 h and then incubated for 1 h with 100 µl of the corresponding biotinylated antibody (anti-IL-10 antibody (clone SXC-1, Pharmingen) or anti-IFN-γ antibody (clone R4·6A2, our own production)) combined with HRP-streptavidin (Amdex, Amersham Pharmacia Biotech), and revealed with OPD substrate solution (Sigma) (490 nm). Absorbance was measured using an ELISA reader (MRX Microplate Reader, Dynatech, USA).

RESULTS

flt-3L-DC protect NOD mice from diabetes development

NOD DC were derived from total splenocytes cultured for 21 days with GM-CSF + flt-3L + IL-6 (flt-3L-DC) and compared to DC derived from bone marrow progenitors cultured with GM-CSF + IL-4 (BM-DC). Both culture conditions led to the development of CD11c⁺/CD11b⁺ DC, which were major histocompatibility complex (MHC) class II⁺/CD80⁺/CD86⁺/CD40⁺/CD8α^–/ Gr-1^–/B220^– (Fig. 1), indicating that flt-3L-DC as BM-DC belong to the classical myeloid-related DC subset. However, whereas BM-DC contained immature and mature DC as shown by heterogeneous levels of I-A, CD80, CD86 and CD40 molecules on cell surface (Fig. 1), flt-3L-DC consisted of DC with a high homogeneous level of MHC class II and co-stimulatory molecules. Serial reseeding used in this culture system might favour the accumulation of DC with a mature phenotype. To study the role of flt-3L-DC in diabetes development, transfer experiments were performed in 5-week-old prediabetic NOD females. As shown in Fig. 2, a single injection of 5 × 10⁵ flt-3L-DC induced both a delay and a reduction in diabetes incidence. In contrast, no significant modulation of diabetes development was observed following a single injection of BM-DC, indicating that the flt-3L-DC population displayed a unique high protective potential against diabetes development.

Fig. 1.

Phenotype of NOD flt-3L-DC. Bulk culture of flt-3L-DC (a) and BM-DC (b) or culture of B LPS from 8-week-old-NOD female mice were double-stained with anti-CD11c (a,b) or anti-B220 (c) antibodies and the indicated antibodies and analysed by flow cytometry. Histograms represent I-A, CD80, CD86, CD40, CD11b, Gr-1, B220, K^d and CD8α expression among CD11c⁺ gated cells (a,b) or I-A, CD80, CD86, CD40 expression among the B220⁺ gated cells.

Fig. 2.

Transfer of flt-3L-DC protects NOD mice against diabetes development. Five-week-old NOD female mice received a single i.v. injection of 5 × 10⁵ purified (purity > 95%) BM-DC (P = 0·2387, n = 8) or flt-3L-DC (P = 0·0015, n = 7). Splenocytes from 4-week-old NOD females were used as negative control (n = 8). Kaplan–Meier actuarial graph survival and log-rank test were used for statistical analysis of diabetes development. Data shown are representative of four independent experiments (seven to 15 mice/group).

Cell proliferation in spleen and pancreatic LN from DC-treated mice

To investigate the mechanisms of protection induced by flt-3L-DC, spleen and LN from control and DC-treated mice were analysed on day 3 post-injection. Cell proliferation was measured in vitro without stimulation (autoproliferation). On day 3 post-injection (Fig. 3), cells isolated from the spleen from both flt-3L-DC- and BM-DC-treated NOD mice exhibited a spontaneous proliferation in vitro, indicating that DC transfer led to increased ex-vivo autologous MLR. Increased autologous MLR was also detected in the pancreatic LN of mice treated with flt-3L-DC. Autoproliferation was not detected in peripheral and mesenteric LN whatever the time at which the cells were tested (Fig. 4a,b). Increased autologous MLR was also detected in DBA/2 (Fig. 3) recipient mice following injection of BM-DC or flt-3L-DC with, however, a lower intensity. On day 7 post-injection (Fig. 4), NOD cells were no longer proliferating in the pancreatic LN and proliferation was reduced by an order of magnitude of 10 in the spleen of flt-3L-DC-treated mice, indicating that this increased autologous MLR was transient. The decreased cell proliferation observed on day 7 post-transfer was not due to T cell-induced cell death or anergy, because cells from all organs tested in all groups of mice responded equally to anti-CD3 (Fig. 4c). In conclusion, passive transfer of both BM-DC and flt-3L-DC were able to induce an early and transient increase of the ex-vivo autologous MLR in the NOD mice.

Fig. 3.

Ex-vivo autologous MLR from DC-treated mice. Cells from individual DC (BM-DC and flt-3L-DC)-treated or control mice (three mice/group) were isolated from the spleen and the pancreatic LN on day 3 post-injection and cultured (3 × 10⁵ cells/0·2 ml) for 3 days in RPMI 5% FCS. Cell proliferation was measured by [³H]-Tdr incorporation during the last 16 h of culture. Data represent cpm/culture × 10⁻³ (mean ± s.d.).

Fig. 4.

Transient increase of autologous MLR following flt-3L-DC injection. Cells from individual flt-3L-DC-treated or control mice (three mice/group) were isolated on day 3 (a) or day 7 (b,c) post-injection and cultured (3 × 10⁵ cells/0·2 ml) for 3 days in RPMI 5% FCS. On day 7, cells were also cultured in the presence of anti-CD3 antibody (1 µg/ml) (c). Cell proliferation was measured by [³H]-Tdr incorporation during the last 16 h of culture. Data represent cpm/culture (mean ± s.d.).

Early production of IL-4 and IL-10 in NOD mice following flt-3L-DC injection

Cytokine produced in the autologous MLR was analysed on day 3 post-injection. Spleen and LN from control and DC-treated DBA/2 and NOD mice were tested for IFN-γ, IL-4 and IL-10 production (Table 1). Transfer of BM-DC led to high amounts of IFN-γ and low amounts of IL-10 in both DBA/2 and NOD mice. In contrast, in cultures derived from flt-3L-DC-treated NOD mice, higher amounts of IL-10 and lower amounts of IFN-γ were produced, leading to a shift of the IFN-γ/IL-10 ratio. Although less pronounced, this shift was also observed in DBA/2 cultures. In addition, a significant production of IL-4 was detected in all cultures of flt-3L-DC-treated mice (DBA/2 and NOD), whereas no IL-4 was detected in cultures from control, BM-DC- and B LPS-treated mice. In the pancreatic LN of NOD mice, cytokine production was rarely detected. When cytokines were produced, the IFN-γ/IL-10 ratio in culture derived from BM-DC and flt-3L-DC-treated NOD mice underwent the same shift as observed in the spleen (Table 1). Neither cytokine production (Table 1) nor cell proliferation (not shown) were detected in recipient mice following injection of LPS-activated B cells expressing high level of MHC class II and co-stimulatory molecules (Fig. 1c). These results indicate that the production of cytokines following injection of syngeneic antigen-presenting cells was specific to the DC and not due to only increased antigenic presentation. These data show that only flt-3L-DC were able to induce a shifts towards IL-4 and IL-10 production early after cell transfer in NOD mice.

Table 1.

Cytokine production following DC transfer

	Spleen				Pancreatic lymph node
Recipient mice	IFN-γ ng/ml	IL-10 ng/ml	IL-4 ng/ml	IFN-γ/IL-10 ratio	IFN-γ ng/ml	IL-10 ng/ml	IL-4 ng/ml
NOD control 1	0·03 ± 0·01	<0·1	<0·03	–	<0·02	<0·1	<0·03
NOD control 2	0·12 ± 0·02	<0·1	<0·03	–	0·80 ± 0·10	<0·1	<0·03
NOD control 3	<0·02	<0·1	<0·03	–	<0·02	<0·1	<0·03
NOD BM-DC 1	3·50 ± 0·10	0·50 ± 0·04	<0·03	7	<0·02	<0·1	<0·03
NOD BM-DC 2	4·00 ± 0·30	0·70 ± 0·05	<0·03	6	2·50 ± 0·10	1·00 ± 0·20	0·24 ± 0·08
NOD BM-DC 3	2·60 ± 0·40	0·97 ± 0·03	<0·03	3	<0·02	<0·10	<0·03
NOD flt-3L-DC 1	3·60 ± 0·10	4·13 ± 0·08	0·20 ± 0·02	0·9	<0·02	<0·10	<0·03
NOD flt-3L-DC 2	1·60 ± 0·20	6·10 ± 0·10	0·24 ± 0·10	0·3	0·20 ± 0·10	1·20 ± 0·30	0·07 ± 0·04
NOD flt-3L-DC 3	n.d.	n.d.	0·21 ± 0·07	–	n.d.	n.d.	0·40 ± 0·09
NOD flt-3L-DC 4	0·60 ± 0·07	2·6 ± 0·04	0·17 ± 0·04	0·2	<0·02	<0·10	<0·03
NOD flt-3L-DC 5	0·70 ± 0·10	1·40 ± 0·05	0·10 ± 0·02	0·5	<0·02	<0·10	<0·03
NOD LPS 1	0·10 ± 0·02	<0·1	<0·03	–	<0·02	<0·1	<0·03
NOD LPS 2	<0·02	<0·1	<0·03	–	<0·02	<0·1	<0·03
NOD LPS 3	<0·02	<0·1	<0·03	–	<0·02	<0·1	<0·03
DBA/2 control 1	<0·02	<0·1	<0·03	–
DBA/2 control 2	<0·02	<0·1	<0·03	–
DBA/2 control 3	<0·02	<0·1	<0·03	–
DBA/2 BM-DC 1	2·40 ± 0·06	0·10 ± 0·05	<0·03	24
DBA/2 BM-DC 2	2·60 ± 0·10	0·37 ± 0·08	<0·03	7
DBA/2 BM-DC 3	0·10 ± 0·10	<0·1	<0·03	–
DBA/2 flt-3L-DC 1	2·10 ± 0·10	0·95 ± 0·20	0·07 ± 0·01	2·2
DBA/2 flt-3L-DC 2	1·80 ± 0·20	0·80 ± 0·05	0·06 ± 0·02	2·2
DBA/2 flt-3L-DC 3	n.d.	n.d.	0·07 ± 0·02	–

Cells isolated from the spleen or the pancreatic LN from individual DC- (BM-DC and flt-3L-DC)-, B LPS-treated or control mice (three to five mice/group) were isolated on day 3 post-injection and cultured (7·5 × 10⁵ cells/0·5 ml) for 3 days in RPMI 5% FCS. Supernatants were assayed for IFN-γ sensitivity 20 pg/ml), IL-4 (sensitivity 30 pg/ml) and IL-10 (sensitivity 100 pg/ml) production using ELISA.

NOD flt-3L-DC do not produce IL-12p70

In order to elucidate why flt-3L-DC induced IL-4 and IL-10 production in NOD mice, whereas BM-DC did not, IL-10 and IL-12p70 production were analysed in the supernatant of BM-DC and flt-3L-DC stimulated with LPS and anti-CD40 antibody (Table 2). As we have shown previously [21], NOD BM-DC produced more IL-12p70 than DBA/2, C57BL/B6 and CBA BM-DC (Table 2) (P < 0·05). In contrast, no IL-12p70 was detected in either NOD or control flt-3L-DC supernatant. Neither BM-DC nor flt-3L-DC produced IL-10 when stimulated with LPS + anti-CD40 antibody (Table 2).

Table 2.

IL-12p70 and IL-10 production by NOD DC

	DBA/2		C57BL/6		CBA		NOD
Cytokine (pg/ml)	IL-12p70	IL-10	IL-12p70	IL-10	IL-12p70	IL-10	IL-12p70	IL-10
BM-DC
16 h	247 ± 196	≤100	n.d.	n.d.	n.d.	n.d.	2531 ± 163	≤100
48 h	156 ± 111	≤100	545 ± 132	n.d.	407 ± 153	n.d.	1433 ± 493*	≤100
flt-3L-DC
16 h	39 ± 12	≤100	n.d.	n.d.	n.d.	n.d.	34 ± 23	≤100
48 h	43 ± 50	≤100	n.d.	n.d.	n.d.	n.d.	29 ± 10	≤100

BM-DC and flt-3L-DC (2 × 10⁵ cells/ml) were stimulated for 16 or 48 h with 1 µg/ml LPS and 1 µg/ml anti-CD40 agonist antibody. Amounts of IL-12p70 and IL-10 were measured in culture supernatants using ELISA. Data represent amounts of cytokines (pg/ml, mean ± s.d.). IL-10 sensitivity, 100 pg/ml, IL-12p70 sensitivity, 20 pg/ml. Statistical analysis using the non parametric Mann–Whitney test reveals significant difference between NOD and control mice values for BM-DC

^*(P < 0·05).

Thus, the capacity of NOD flt-3L-DC to shift the cytokine production towards Th2 pathway appears to be associated with the absence of IL-12p70 production.

DISCUSSION

DC derived from splenic progenitors cultured with flt-3L, IL-6 and GM-CSF were injected into young prediabetic NOD mice and their potential to modulate diabetes development was investigated. We showed that a single injection of flt-3L-DC both delayed onset and reduced incidence of diabetes in recipient mice, whereas a single injection of BM-DC into prediabetic female failed to induce significant inhibition of diabetes.

The syngeneic (or autologous) MLR, which has been reported to be involved in the generation of regulatory/suppressive T cells, is decreased in NOD mice [5] as in other autoimmune strains [22]. In NOD mice, it has been proposed that the depressed syngeneic MLR results from a defect of antigen-presenting cells in activating suppressive T cells, rather than an absence of immunoregulatory T cell populations [5]. In agreement with this hypothesis, we showed that passive transfer of in vitro-derived syngeneic DC can restore the depressed syngeneic MLR in NOD mice. The fact that the level of the proliferative response was higher following injection of flt-3L-DC compared to BM-DC indicates that the autologous MLR does not correlate with the levels of IL-12p70 and probably relates to the higher level of MHC class II and co-stimulatory molecules on flt-3L-DC cell surface. Indeed, we have shown previously that flt-3L-DC are more efficient in stimulating T cells than BM-DC [21]. However, according to the poorly defined role of the autologous MLR in generation of regulatory processes, the relationship between this autoproliferation and the generation of a protective response against diabetes development is not clear and has to be investigated.

The main difference between BM-DC-induced and flt-3L-DC-induced response was the nature of the cytokines produced in the ex-vivo autologous MLR early after DC transfer. We showed that only flt-3L-DC induced IL-4 and IL-10 production following in vivo injection. This correlation between IL-4 and IL-10 early production and induction of diabetes resistance in prediabetic NOD mice suggests strongly that flt-3L-DC induces a Th2 regulatory response in vivo. Our data are in agreement with previous results, showing that protection against diabetes development can be achieved in NOD mice if the immune response is shifted to a Th2 response [23], whereas a Th1 polarized response after IL-12 injection has been shown to induce disease acceleration [24]. Our findings raise questions on the nature of the IL-4- and/or IL-10-producing cells stimulated following transfer of flt-3L-DC. IL-10 is a suppressive cytokine produced by a variety of cells including classical Th2 T cells and some CD4⁺ regulatory T cell subsets, such as Tr1 cells [25]. Thus, injection of flt-3L-DC could stimulate IL-10 production by CD4⁺ regulatory T cells in NOD mice. Recently, it has been reported that peptide-pulsed mature DC with poor IL-12 and IL-10 production are able to prevent EAE in mice through stimulation of antigen-specific IL-10-producing CD4⁺ T cells [26]. In addition to IL-10 secretion, flt-3L-DC injection leads to early production of IL-4 and IFN-γ. Thus, NK T cells [27] could be involved in the suppressive mechanisms which developed in the NOD mice after flt-3L-DC treatment. In vivo administration of anti-IL-10 and/or anti-IL-4 antibodies into flt-3L-DC-treated NOD mice will be further used to define the role of both cytokines in diabetes protection.

Our data could appear conflicting with those recently reported by Feili-Hariri et al. [28] showing that a protective Th2 response can be induced in young NOD mice following BM-DC transfer. However, the protective response was observed only when mice received several injections of high numbers of BM-DC, whereas a unique injection showed no effect on diabetes development [15], showing that DC with protective potential were sparse when derived from bone marrow progenitors cultured with GM-CSF with or without IL-4. Interestingly, in this study, the efficiency of the protection was greater with mature BM-DC than with immature BM-DC. In our study, although the maturational state of BM-DC and flt-3L-DC could not be strictly compared (DC were derived with two independent culture systems and from distinct cell origins), flt-3L-DC consisted of a homogeneous population of mature DC with high expression of MHC class II and co-stimulatory molecules, whereas BM-DC were heterogeneous. Altogether, these data suggest strongly that, in NOD mice, enhancement of the mature/immature DC ratio in vivo may favour protection against diabetes development. In agreement with this hypothesis several studies, including our recent report, showed that DC development is altered in NOD mice [21,29–36].

Increased production of IL-12 in both DC [21,30,37,38] and macrophages from NOD mice [31], as well as polymorphisms in the Il12b gene [36], have been reported and might favour Th1 response in this strain. While DC maturation is associated with an initial production of IL-12p70 followed by a progressive decrease at later stages of DC maturation [39,40], it remains unclear whether this defect is due to a global defect of DC maturation or is related to deficiency of one particular lineage of DC. However, we showed recently that defects in other cell types that have a role in triggering DC differentiation/maturation could also lead to DC defect in NOD mice [41]. In agreement with these results, the data reported here show that, in appropriate conditions favouring spontaneous maturation, mature DC that do not produce IL-12p70 and harbouring the capacity to induce protective response against the development of the autoimmune process can be derived from NOD mice.

We hypothesize that flt-3L-DC represent the in vitro counterparts of a subset, or a maturational state, of DC that is defective in NOD mice and cannot be derived easily from BM progenitors cultured with GM-CSF + IL-4. This subset, characterized by mature phenotype and no IL-12p70 production, may be critical in stimulating regulatory process as an immune deviation to Th2 response. Specific stimulation of this subset of DC may open new possibilities of therapeutic intervention to prevent diabetes onset.