T. L. Delovitch, Laboratory of Autoimmune Diabetes, Robarts Research Institute, University of Western Ontario, 100 Perth Drive, London, ON, Canada N6A 5K8. E-mail: firstname.lastname@example.org
Protection from type 1 diabetes (T1D), a T helper type 1 (Th1)-mediated disease, is achievable in non-obese diabetic (NOD) mice by treatment with α-galactosylceramide (α-GalCer) glycolipids that stimulate CD1d-restricted invariant natural killer T (iNK T) cells. While we have reported previously that the C20:2 N-acyl variant of α-GalCer elicits a Th2-biased cytokine response and protects NOD mice from T1D more effectively than a form of α-GalCer that induces mixed Th1 and Th2 responses, it remained to determine whether this protection is accompanied by heightened anti-inflammatory responses. We show that treatment of NOD mice with C20:2 diminished the activation of ‘inflammatory’ interleukin (IL)-12 producing CD11chighCD8+ myeloid dendritic cells (mDCs) and augmented the function of ‘tolerogenic’ DCs more effectively than treatment with the prototypical iNKT cell activator KRN7000 (α-GalCer C26:0) that induces Th1- and Th2-type responses. These findings correlate with a reduced capacity of C20:2 to sustain the early transactivation of T, B and NK cells. They may also explain our observation that C20:2 activated iNK T cells depend less than KRN7000 activated iNK T cells upon regulation by regulatory T cells for cytokine secretion and protection from T1D. The enhanced anti-inflammatory properties of C20:2 relative to KRN7000 suggest that C20:2 should be evaluated further as a drug to induce iNK T cell-mediated protection from T1D in humans.
Invariant natural killer T (iNK T) cells comprise a subset of innate lymphocytes that express a semi-invariant T cell receptor (TCR) consisting of Vα14 paired preferentially to Vβ8·2, Vβ7 or Vβ2 in mouse and Vα24 paired to Vβ11 in humans. The defining characteristics of iNK T cells are their ability to recognize a lipid or glycolipid antigen presented on an antigen-presenting cell (APC) by a major histocompatibility complex (MHC) class I-like molecule CD1d [1–3]. All type 1 ‘classical’ iNK T cells recognize and react to the prototypic synthetic glycolipid KRN7000, which is a form of α-galactosyl-ceramide (α-GalCer) C26:0 based on structurally similar immunostimulatory compounds derived originally from a marine sponge . A hallmark of iNK T cells is their ability to produce rapidly and secrete large amounts of both T helper type 1 (Th1) [interferon (IFN)-γ] and Th2 [interleukin (IL)-4] cytokines within hours after activation with KRN7000 [2,3]. Although α-GalCer is most probably not a naturally occurring iNK T cell ligand in mammals, it has been used experimentally to explore the immunomodulatory properties of iNK T cells in many preclinical models of autoimmune disease.
The pathogenesis of type 1 diabetes (T1D), an autoimmune disease characterized by the T cell-mediated destruction of insulin-producing pancreatic islet β cells, arises in part from functional deficiencies in regulatory T cell (Treg) populations, including CD4+CD25+forkhead box P3 (Foxp3+) (Treg) cells and iNK T cells [5,6]. Studies conducted in the NOD mouse model of experimental T1D  have shown that many therapies that prevent T1D depend upon their ability to activate Tregs and/or iNK T cells [5,6]. Although several investigators reported that activation of iNK T cells with KRN7000 can protect NOD mice from developing T1D [8–10], the mechanism(s) of iNK T cell-mediated protection against T1D is not well understood. Current evidence suggests that therapeutic effects elicited by activation of iNK T cells elicited upon KRN7000 treatment of NOD mice occur in an IL-4-dependent manner  and are associated with a Th2-like environment [8,9] that promotes the recruitment of tolerogenic dendritic cells (DCs) to the draining pancreatic lymph nodes (PLN) [12–14]. We have demonstrated that iNK T-mediated protection from T1D induced by KRN7000 requires the activity of CD4+CD25+ T cells that are enriched for CD4+CD25+Foxp3+ Tregs. Young (4–5 weeks old) NOD mice treated in vivo first with an anti-CD25 monoclonal antibody (mAb) to inactivate Tregs and then with KRN7000 to activate iNK T cells still developed T1D spontaneously. Importantly, we found that KRN7000 induces the surface expression of CD25 by only about one-third of activated iNK T cells in wild-type NOD mice , indicating that anti-CD25 treatment does not inactivate the majority of activated iNK T cells. These findings support the notion that Tregs may be required to down-regulate the activity of iNK T cells stimulated by synthetic α-GalCer ligands.
Since the discovery of KRN7000, several investigators have generated other synthetic analogues of α-GalCer that polarize iNK T cells towards Th1- or Th2-type cytokine responses [16–18]. In the case of a Th1-mediated autoimmune disease such as T1D, the use of an analogue that enhances IL-4 production significantly while reducing simultaneously the synthesis of a strong proinflammatory cytokine such as IFN-γ represents a desirable therapeutic approach to the prevention of T1D [17,18]. One such analogue, termed α-GalCer C20:2 (referred to as C20:2 hereafter), in which the C26:0 N-acyl group of KRN7000 is replaced by an 11,14 cis unsaturated 20 carbon fatty acid, can potently activate iNK T cells and promote Th2-biased responses associated with improved clinical and immunological outcomes in the prevention of T1D in NOD mice when compared to the parent glycolipid α-GalCer [18,19]. Having shown that the activity of Tregs is required for iNK T cells activated with KRN7000 to protect NOD mice from T1D , it was of interest to determine whether this is also the case for C20:2-induced protection from T1D and, if so, the mechanisms involved. Interestingly, we report in this study that iNK T cells activated by C20:2 depend less on regulation by CD4+CD25+ Treg cells than do iNK T cells activated by KRN7000 for protection from T1D. This decreased dependency of C20:2-induced protection on Treg function correlated with its reduced capacity to sustain the transactivation of T cells, B cells and NK cells. In addition, treatment of NOD mice with C20:2 diminished the activation of ‘inflammatory’ IL-12-producing CD11chighCD8+ myeloid dendritic cells (mDCs) and augmented the function of ‘tolerogenic’ DCs more effectively than treatment with KRN7000. The latter findings may explain why C20:2-activated iNK T cells depend less than KRN7000-activated iNK T cells on regulation by Tregs for cytokine secretion and protection from T1D. Our results demonstrate that greater anti-inflammatory activity is associated with protection from T1D induced by C20:2 than that induced by KRN7000 and presumably other forms of a-GalCer that induce mixed Th1- and Th2-type cytokine responses. These data suggest that C20:2 should be evaluated further as a target drug for iNK T cell-mediated protection from T1D in humans.
Materials and methods
NOD/Del and C57BL/6 mice were bred in a specific pathogen-free barrier facility at the Robarts Research Institute (London, ON, Canada). The incidence of T1D in female NOD mice in our colony is 25–30% at 15 weeks of age and ≥75% by 25 weeks. All experimental mice were female and were maintained in a specific pathogen-free facility in the Animal Care and Veterinary Services at the University of Western Ontario according to institutional guidelines.
Antibody and reagents
KRN7000 (α-GalCer) and its proprietary vehicle control was kindly provided by Kirin Pharmaceutical Research Laboratories (Gunma, Japan), solubilized in water and injected intraperitoneally (i.p.) into mice (4 µg/dose) every other day for 3 weeks, as described previously . C20:2 was synthesized as described previously  and dissolved at 200 µg/ml in phosphate-buffered saline (PBS) containing 0·05% Tween 20, and injected i.p. (4 µg/mouse, in 250 µl of vehicle consisting of PBS plus 0·05% Tween 20) or vehicle (PBS plus 0·05% Tween 20). Fluorescein isothiocyanate (FITC)-conjugated anti-TCR-β (H57-597), anti-CD25 (7D4), anti-B220 (RA3-6B2), anti-Pan NK cells (DX5), anti-Siglec H (eBio440c); phycoerythrin (PE)-conjugated anti-CD69 (H1·2F3), anti-I-Ad (AMS-32·1), anti-CD86 (Gl-1), anti-CD40 (MR1), anti-IL-4 (11B11), anti-IFN-γ (XMG1·2), anti-IL-12p40/70 (C15·6); peridinin chlorophyll (PerCP)-conjugated anti-CD8α (53–2·1), anti-CD4 (RM4-5), anti-CD3ε (145-2C11); and APC-conjugated anti-CD11c (N418) mAbs were purchased from eBiosciences (San Diego, CA, USA) or BD Biosciences (Mississauga, ON, Canada). Fluorescently labelled tetrameric CD1d molecules loaded with α-GaCer (KRN7000) were prepared as described previously . The anti-CD25 (PC61) mAb used to inactivate CD4+CD25+ Treg cells in vivo was prepared in house from hybridomas (American Type Culture Collection no. TIB 222). RPMI-1640 tissue culture medium was supplemented with 10% heat-inactivated fetal calf serum (FCS), 10 mM HEPES buffer, 1 mM Na pyruvate, 2 mM L-glutamine, 100 units/ml penicillin, 0·1 mg/ml streptomycin and 0·05 mM 2-mercaptoethanol (ME) (all purchased from Invitrogen Life Technologies, Burlington, ON, Canada).
Monitoring of diabetes
Mice were monitored beginning at 12 weeks of age for hyperglycaemia by measurement of blood glucose levels (BGL) twice weekly using an Ascensia ELITE glucometer and strips (Bayer, Toronto, ON, Canada), as described previously . Mice were considered diabetic when two consecutive BGL readings of >11·1 mmol/l were obtained.
In vivo treatment with glycolipid
To monitor the spontaneous development of T1D, NOD mice (4–5 weeks old) were first injected intravenously (i.v.) with 500 µg of either anti-CD25 mAb (PC61) to inactivate CD25+ cells or isotype control immunoglobulin G (IgG), rested for 3 days, treated by i.p. injection with KRN7000 (4 µg/dose), C20:2 (4 µg/dose) or vehicle (control) every other day for 3 weeks and then monitored for the incidence of T1D, as reported previously .
Cell purification and flow cytometry
Single-cell lymphocyte suspensions were prepared from the spleen and PLN . Non-viable cells were excluded by electronic gating for all experiments. For intracellular cytokine staining, cell suspensions were cultured without further stimulation at 107 cells/ml in culture media containing Golgi Stop (BD Biosciences) for 3–4 h. Intracellular staining for IL-4, IFN-γ and IL-12 was performed using a BD Cytofix/Cytoperm buffer set (BD Biosciences), according to the manufacturer's protocol. Flow cytometry was performed using a fluorescence activated cell sorter (FACS)Calibur instrument (BD Biosciences), and the acquired data were analysed using FlowJo software (Tree Star Inc., Ashland, OR, USA).
In vitro cultures and enzyme-linked immunosorbent assay (ELISA)
Splenocytes obtained from various treatment groups of mice as indicated after the last α-GalCer treatment were cultured (5×106 cells/ml) for 72 h in the absence or presence (restimulation) of KRN7000, C20:2 or control (100 ng/ml, or as indicated). A standard sandwich ELISA was performed for mouse cytokines using paired antibody kits for IFN-γ, IL-13 (eBiosciences) or IL-4 (BD Biosciences). For signal detection, streptavidin–horseradish peroxidase (HRP) conjugate and development solution from BD OptiEIA Reagent Set A (BD Biosciences) were used. For DC cultures, NOD mice were treated i.p. with glycolipid according to our multi-low-dose protocol (4 µg/dose every other day for 3 weeks), as reported previously . At 1 week after the last dose, spleen and PLN suspensions were pooled and CD11c+ microbeads (Miltenyi Biotec Inc., Auburn, CA, USA) were used to sort CD11c+ DCs, according to the manufacturer's instructions for each treatment group. Concurrently, CD4+ peptide-primed T cells were isolated (CD4+ T cell-negative selection kit; Miltenyi Biotec) from NOD mice injected (i.p.) 10 days previously with either vehicle control (PBS) or 100 µg of a B9-23 insulin peptide (Ins B9-23, kindly provided by Dr B. Singh, University of Western Ontario, London, ON, Canada) emulsified in incomplete Freund's adjuvant (IFA) (Sigma Aldrich, Oakville, ON, Canada) or dissolved in PBS. Peptide primed-CD4+ T cells were cultured with CD11c+ DC at a 20 : 1 ratio in the presence of the Ins B9-23 peptide for 72 h for the ELISA (2 × 106 T cells + 1 × 105 DC) and proliferation (5 × 105 T + 1 × 104 DC) assays, respectively. Culture supernatants were collected for IL-2, IFN-γ and IL-4 detection using a standard sandwich ELISA (eBiosciences), and cell proliferation was determined by [3H]-thymidine pulse (1 µCi/well; PerkinElmer, Woodbridge, ON, Canada) for the last 18 h of culture, harvested and read on a 1450 Microbeta counter (PerkinElmer).
Statistical significance of cell expansion as well as cytokine production and secretion assays were conducted using two-way analysis of variance (anova) comparisons with Bonferroni post-tests. The incidence of T1D was compared using a Mantel–Cox log rank test. In all experiments, differences were considered significant when P was less than 0·05.
C20:2 activated iN K T cells induce a Th2-biased cytokine response
Relative to KRN7000, the structure of C20:2 consists of a fatty acyl chain reduced from C26 to C20 with unsaturations at carbons 11 and 14 (Fig. 1a). Unique among Th2-biased glycolipid analogues, C20:2 has a relatively similar binding affinity to CD1d as α-GalCer (Kd = 1·83 µM versus Kd = 1·29 µM, respectively) , and based on staining with glycolipid-loaded CD1d tetramers has been shown to be recognized by the same population of iNK T cells as α-GalCer [18,19]. Accordingly, we analysed initially whether the proliferative capacity and cytokine secretion profile of iNK T cells from NOD mice stimulated in vitro by KRN7000 or C20:2 are similar. Indeed, the proliferative responses of NOD splenocytes stimulated by KRN7000 or C20:2 were found to be very similar (Fig. 1b). Supernatants from C20:2 activated splenocytes generally contained a higher concentration of IL-4 and lower concentration of IFN-γ than supernatants from KRN7000-stimulated splenocytes (Fig. 1c), consistent with the reported ability of C20:2 to induce a Th2-biased response . More importantly, the induction of a Th2-biased response by C20:2 was even more pronounced in vivo, as NOD mice treated first with isotype control IgG and then with C20:2 (IgG/C20:2) yielded a greater serum IL-4 response than did IgG/KRN7000-treated mice at 2 h post-treatment (Fig. 1d, left panel). Interestingly, the IgG/C20:2 serum IL-4 response was increased further ∼twofold at 2 h in anti-CD25/C20:2-treated mice, whereas no differences in IL-4 or IFN-γ responses were detected between anti-CD25 versus IgG-treated mice at 24 h. Note that the anti-CD25 mAb used here inactivates rather than depletes Tregs in NOD mice [15,22]. Considerably lower IL-4 responses were seen at 24 h post-treatment, and the responses obtained in IgG/KRN7000- and IgG/C20:2-treated mice were not significant. While only a very weak serum IFN-γ response was detected at 2 h in IgG/KRN7000- and IgG/C20:2-treated mice, a markedly higher serum IFN-γ response was observed at 24 h in IgG/KRN7000-treated mice than in IgG/C20:2-treated mice (Fig. 1d, right panel). Thus, C20:2 can induce a Th2-biased cytokine response by iNK T cells in NOD mice, and prior inactivation of Tregs by treatment of these mice with an anti-CD25 mAb can augment this Th2-type response further.
C20:2-activated iNK T cells are less dependent than KRN7000-activated iNK T cells on CD4+CD25+ Treg cells for protection against T1D
Given that Treg cells are required to regulate KRN7000-induced iNK T-mediated protection of NOD mice from T1D , and having shown that C20:2 elicits a Th2-biased iNK T cytokine response in NOD mice (Fig. 1), we next investigated whether C20:2-activated iNK T cells require the activity of Tregs for protection from T1D. Young NOD mice (4–5 weeks old) were administered a single dose of anti-CD25 mAb, rested for 3 days, and treated with a multi-low-dose protocol of glycolipid known to protect NOD mice from T1D . Mice treated with vehicle only developed T1D spontaneously beginning at 12–15 weeks of age in mice receiving anti-CD25 or control IgG, with more than 75% of mice becoming diabetic by 30 weeks of age. Similarly, mice that received anti-CD25 mAb and KRN7000 developed T1D beginning at 12 weeks of age and greater than 75% of mice were diabetic by 30 weeks of age (Fig. 2), which was consistent with our previous report . However, we were surprised to find that mice treated with anti-CD25 mAb and C20:2 were relatively protected from T1D, with only 35% of mice developing T1D by 30 weeks of age (Fig. 2). A similar level of protection was seen in mice treated with control IgG and either C20:2 or KRN7000. These results suggest that C20:2-activated iNK T cells depend less than KRN7000-activated iNK T cells on regulation by Tregs for protection against T1D.
KRN7000 and C20:2 differ in their capacity to expand iNK T cells in the spleen
The observation that KRN7000- and C20:2-activated iNK T cells differ in their level of requirement of Treg regulation for protection from T1D raised the possibility that these two glycolipid analogues may vary in their capacity to activate iNK T cells. To test this possibility, we administered KRN7000 or C20:2 to NOD mice in vivo and assayed the kinetics of iNK T cell expansion in the spleen and PLN, two sites of localization of activated iNK T cells that mediate protection from T1D. Using α-GalCer/CD1d tetramers to identify iNK T cells, we observed that both KRN7000 and C20:2 induce iNK T cell TCR down-regulation by 24 h after glycolipid administration, which is typical of iNK T cell activation in vivo[23,24]. A comparison of the frequency of iNK T cells in the spleen at 72 h and 2 h post-treatment revealed that whereas KRN7000 stimulated the proliferation and expansion (seven- to eightfold) of spleen iNK T cells (3·72 ± 0·7% versus 0·54 ± 0·1%), the frequency of C20:2-activated spleen iNK T cells remained at homeostatic levels (0·45 ± 0·04% versus 0·37 ± 0·1%) (Fig. 3a). A similar comparison of iNK T cell frequencies in the PLN at 72 h and 2 h post-treatment showed that both KRN7000 (2·63 ± 0·4% versus 0·31 ± 0·1%) and C20:2 (0·97 ± 0·1% versus 0·13 ± 0·1%) stimulated the proliferation and expansion (seven- to eightfold) of iNK T cells (Fig. 3b). A similar kinetics of expansion was also observed in NOD mice treated with anti-CD25 mAb prior to glycolipid administration (our unpublished observations). However, at 72 h after treatment, we observed differences in the absolute number of iNK T cells in the spleen between mice treated first with anti-CD25 or IgG and then with KRN7000 (Fig. 3a, right panel). This difference was not observed upon treatment with C20:2. The differences noted in the spleen for KRN7000-activated iNK T cells were not detected in the PLN at 72 h post-treatment (Fig. 3b, right panel). Thus, iNK T cells in the spleen but not PLN appear to be sensitive to a different level of expansion induced by KRN7000 versus C20:2. Different subsets of DCs interact with iNK T cells in the spleen and PLN, and may account for this variation in sensitivity to iNK T cell activation and expansion.
KRN7000 and C20:2 differ in their capacity to activate iNK T cells for cytokine secretion. Next, we analysed the ability of activated iNK T cells to secrete cytokines upon glycolipid restimulation. Because the greatest difference in iNK T cell expansion was observed at 72 h post-glycolipid treatment in vivo, splenocytes were restimulated at this time-point in vitro and their secretion of IFN-γ, IL-4 and IL-13 was analysed by ELISA. In support of a role of Tregs in the down-regulation of immune responses, we observed an increase in cytokine secretion by splenocytes from mice treated with anti-CD25 prior to glycolipid or vehicle by comparison to splenocytes from IgG-treated mice (Fig. 4a). A fourfold and fivefold increase in IFN-γ secretion over that of vehicle control was detected in mice treated with anti-CD25/KRN7000 and IgG/KRN7000, respectively. These increases in IFN-γ secretion were not evident in C20:2-treated mice (Fig. 4a) in which the basal levels of IFN-γ secretion detected were similar to those in vehicle control-treated mice. Thus, inactivation by anti-CD25 seems to inhibit the ability of Tregs to regulate iNK T cell secretion of IFN-γ after stimulation by KRN7000, but not C20:2.
Th2 cytokine responses also differ between KRN7000- and C20:2-activated iNK T cells. Consistent with the observation that C20:2 induces Th2-biased responses, we found that splenocytes from C20:2-treated mice, irrespective of previous Treg inactivation, produced greater amounts of IL-4 and IL-13 compared to splenocytes from KRN7000-treated mice (Fig. 4a). Interestingly, the greatest difference between C20:2- and KRN7000-induced secretion of Th2 cytokines was noted in mice treated with control IgG. Whereas Treg inactivation resulted in an increase in IFN-γ secretion by splenocytes from KRN7000- and IgG-treated mice, only a marginal increase in C20:2-induced Th2 cytokine secretion was triggered upon Treg inactivation. Thus, although C20:2-activated iNK T cells in the spleen do not expand appreciably at 72 h post-stimulation, these cells retain their ability to secrete Th2 cytokines. While Tregs appear to control the level of Th1 and Th2 cytokine secretion induced by KRN7000-activated iNK T cells, Tregs seem to exert less stringent control of cytokine secretion by C20:2-activated iNK T cells.
Glycolipid-induced iNK T cell cytokine bias probably originates from interactions with APCs and the transactivation of other immune cells, rather than biasing iNK T cells themselves, which are resistant to cytokine polarization in C57BL/6 mice . To test whether C20:2 polarizes cytokine expression directly by iNK T cells in vivo in NOD mice, we analysed C20:2-activated iNK T cells for their pattern of intracellular cytokine expression. Prior to iNK T TCR down-regulation at 2 h after glycolipid administration in vivo, iNK T cells in the spleen and PLN of NOD mice were stained directly for their intracellular accumulation of IFN-γ and IL-4 without further stimulation in vitro. Gated TCR+ and α-GalCer/CD1d tetramer+ iNK T cells from both C20:2- and KRN7000-treated splenocytes expressed intracellular IFN-γ and IL-4, predominantly in CD4+ iNK T cells (Fig. 4b). Small increases were noted in the percentages of CD4+ iNK T cells from mice treated with anti-CD25 versus IgG and then either KRN7000 or C20:2, with the greatest difference observed in IFN-γ accumulation between anti-CD25/KRN7000 (51·7%)- and IgG/KRN7000 (43·5%)-treated mice. Of note, activation of iNK T cells in the spleen and PLN from IgG/C20:2-treated mice contained more IL-4- and IFN-γ-producing cells than in the spleen and PLN from IgG/KRN7000-treated mice (Fig. 4b). At this early time-point, we did not observe any bias in cytokine secretion by NOD iNK T cells, as C20:2 and KRN7000 induced an equivalent level of accumulation of intracellular IL-4 and IFN-γ.
C20:2-activated iNK T cells elicit reduced bystander cell activation
As IFN-γ was detected in the serum of KRN7000-treated mice only at 24 h post-administration (Fig. 1d), interactions between iNK T cells and other cell types may influence the level of IFN-γ secretion at later times after iNK T cell activation. To compare the ability of C20:2 and KRN7000 to transactivate other immune cells, we assayed the activation of B cells and NK cells in the spleen at 6 h and 24 h post-glycolipid administration. Based on the percentages of B220+CD69+ cells detected, B cell activation was evident in both C20:2- and KRN7000-treated mice. At 6 h post-glycolipid administration, the level of B cell activation was similar between C20:2- and KRN7000-treated mice, irrespective of anti-CD25 treatment (Fig. 5a). Differences between C20:2 and KRN7000 were observed at 24 h post-glycolipid administration, when the percentage of CD69+ B cells activated by C20:2 was reduced about twofold in mice treated with anti-CD25 or IgG. Similar results were obtained for the transactivation of T cells (our unpublished observations). Analyses of NK cell activation profiles revealed that whereas C20:2 and KRN7000 both activate NK cells at an early time-point (6 h), only KRN7000 sustains this activation for a longer time, as the frequency of DX5+CD69+ NK cells was reduced by 1·6-fold and 2·5-fold, respectively, in anti-CD25/C20:2- and IgG/C20:2-treated mice at 24 h (Fig. 5a). Moreover, at 6 h after activation in vivo both C20:2 and KRN7000 induced an equivalent accumulation of IFN-γ in NK cells, while at 12 h post-treatment the percentage of NK+IFN-γ+ cells in C20:2-treated mice was reduced about twofold compared to that in KRN7000-treated mice (Fig. 5b). Thus, C20:2 and KRN7000 each induce immune cell transactivation, but only KRN7000-activated iNK T cells sustain this transactivation.
C20:2-activated iNK T cells influence DC function
DCs present glycolipid and translate iNK T activation signals to other immune cells , and KRN7000 but not C20:2 sustains B and NK cell transactivation. Accordingly, we investigated whether KRN7000 and C20:2 induce DC maturation and function differentially. CD11c+ DCs were assayed by flow cytometry for their surface expression of activation and maturation markers at early and late time-points after iNK T cell activation. At 6 h post-treatment, C20:2- and KRN7000-activated CD11c+ mDCs equivalently, as evaluated by the up-regulation of MHC class II, CD86 and CD40 expression, and the levels of mDC activation observed in IgG-treated mice were similar to those in anti-CD25-treated mice (Fig. 6a). At 24 h post-treatment, when KRN7000-activated iNK T cells up-regulate further the surface expression of co-stimulatory molecules on mDCs, a significant reduction in the expression was observed on mDCs in C20:2-treated mice. While mDCs from anti-CD25/C20:2- and anti-CD25/KRN7000-treated mice displayed a similar level of MHC class II expression, CD86 and CD40 expression was decreased on mDCs from anti-CD25/C20:2-treated mice (Fig. 6a). Thus, KRN7000 stimulation of iNK T cells elicits more robust mDC activation than C20:2 stimulation. This difference in mDC activation may account for the greater bystander activation of B, T and NK cells noted for KRN7000 relative to C20:2.
The latter notion is supported further by our finding that the intracellular expression of IL-12p70 at 6 h post-administration in CD11chighCD8+ mDCs was greater in anti-CD25/KRN7000-treated than anti-CD25/C20:2-treated mice (Fig. 6b). We detected very little IL-12p70 secretion in CD11chighCD8- mDCs (data not shown). These results are consistent with the increased level of serum IFN-γ observed in anti-CD25/KRN7000-treated mice (Fig. 1d). Thus, the decreased ability of C20:2-stimulated iNK T cells to sustain the transactivation of other immune cells may be due in part to the reduced activation of CD11chighCD8+ mDCs.
Previous studies have suggested that iNK T cell-mediated protection against T1D may be due to the recruitment of tolerogenic DCs that suppress rather than prime effector T cells [12–14]. More recent analyses of APCs that mediate the development of T1D demonstrate that plasmacytoid DCs (pDCs) may play a role in reducing inflammation and insulitis . Using Siglec H, a siglec-like molecule that binds specifically precursor pDCs  and CD11c as surface markers that distinguish between pDCs and conventional mDCs, we determined if the frequencies of these DC subsets are altered in mice treated with a therapeutic dose of glycolipid. Young NOD mice (4–5 weeks old) were administered a single dose of anti-CD25 mAb or control IgG, rested for 3 days, and then treated with a multi-low-dose protocol of KRN7000 or C20:2. One week after the last dose, spleen and PLN lymphocytes were harvested for flow cytometric analysis of DC subsets. No differences were detected in the frequency of Siglec H+CD11clow pDCs (data not shown) or CD11chighCD8–/+ mDCs in the spleens of mice treated with either glycolipid or control vehicle (Fig. 7a). However, the frequency of CD11chighCD8+ mDCs but not CD11chighCD8– mDCs was reduced in the PLN of C20:2-treated mice compared to the frequencies observed in KRN7000- and vehicle-treated mice (Fig. 7b). The greatest reduction was observed in mice pretreated with IgG rather than anti-CD25. Note that treatment with KRN7000 or C20:2 each increased the absolute number of CD11chighCD8– mDCs recruited to the PLN relative to control vehicle treatment. Thus, C20:2 activation of iNK T cells leads to the differential recruitment of CD11chighCD8– and CD11chighCD8+ mDCs to the PLN of treated mice, and Treg inactivation enhances the recruitment of both mDC subsets but only in mice treated with KRN7000.
To determine if glycolipid treatment alters the function of DCs, we assayed the capacity of CD11c+ DCs to stimulate antigen-primed CD4+ T cells. CD11c+ DCs from NOD mice treated with a multi-low-dose protocol of glycolipid were co-cultured with Ins B9-23 peptide primed CD4+ T cells and assayed for their capacity to stimulate a T cell recall response (Fig. 7c). CD11c+ DCs from NOD mice administered control vehicle elicited the robust activation of primed CD4+ T cells, as evidenced by the induction of cell proliferation as well as IL-2 and IFN-γ secretion in the presence of recall antigen (Fig. 7c). IL-4 secretion was not detected in any of the cultures (data not shown). In contrast, in vivo treatment of NOD mice with C20:2 was more effective than that of KRN7000 treatment in decreasing the stimulatory capacity of CD11c+ DCs, with the decreases noted for C20:2 being ≥twofold. These observations suggest that chronic activation of iNK T cells by synthetic glycolipids can alter DC antigen-presenting capacity, and that in vivo treatment of NOD mice with C20:2 appears to induce DC function with less stimulatory activity than treatment with KRN7000.
Immunomodulation of iNK T cells with glycolipids represents a potential therapeutic strategy for protection against autoimmune T1D. Notwithstanding, activation of iNK T cells by KRN7000, the prototypic form of α-GalCer that is used currently for iNK T cell analysis and immunomodulation, can be problematic. This glycolipid induces not only ‘protective’ Th2 cytokine responses but also strong proinflammatory Th1 cytokine responses, depending on the system under study . Moreover, as we reported, Treg inactivation by anti-CD25 treatment augments the ability of KRN7000 to induce IFN-γ responses that exacerbate rather than protect from T1D in NOD mice . Thus, the use of modified analogues of α-GalCer that reduce Th1 responses and amplify Th2 responses may be preferable due to their ability to direct iNK T cell functions towards more desirable outcomes. In this study, we analysed iNK T cell activation and function in response to a Th2-biased glycolipid analogue of α-GalCer termed C20:2. We confirmed that C20:2 reduces IFN-γ and enhances IL-4 and IL-13 secretion by iNK T cells in NOD mice and protects them from T1D . Interestingly, however, we found that C20:2-activated iNK T cells depend less on Treg activity than KRN7000-activated iNK T cells for protection. This may result from the reduced capacity of C20:2 to sustain the transactivation of T, B, NK and mature myeloid DCs. Previously, C20:2 was reported to provide improved clinical outcomes against T1D when compared to the α-GalCer-related compound C24:0, which induces a mixed Th1- and Th2-type cytokine response indistinguishable from KRN7000 . Thus, based on these anti-inflammatory functional properties of C20:2 and its capacity to bind to and activate human iNK T cells , our findings support further the idea that C20:2 should be evaluated as a candidate target drug to indue iNK T cell-mediated protection from T1D in humans.
The ability of Tregs to suppress autoimmunity and protect against autoimmune T1D  is currently receiving much attention due to its potential clinical benefit [29,30]. In this regard, it is important to consider that this beneficial effect of Tregs in T1D may be attributable in part to their ability to suppress the proinflammatory functions of activated iNK T cells [31,32]. Given that T1D patients may have functional differences in Treg activity , and that KRN7000-activated iNK T cells depend on Treg cells for protection against NOD T1D , we investigated the effect of Tregs on C20:2-activated iNK T cells during the spontaneous development of T1D. In contrast to the high incidence of T1D observed in anti-CD25/KRN7000-treated mice, anti-CD25/C20:2-treated mice were protected from T1D. Based primarily on two sets of findings, we reason that the different outcome noted between the KRN7000- and C20:2-treated mice arises from the differential ability of these glycolipids to activate iNK T cells rather than the ability of anti-CD25 treatment to inhibit iNK T cell activation. First, the effect of anti-CD25 on Treg inactivation exceeded that on iNK T cell activation, as iNK T cells expanded to a similar extent in mice treated with either anti-CD25 or control IgG. At 3 days post-KRN7000 treatment, spleen cellularity was elevated and this was accompanied by an increase in the absolute number of iNK T cells in the spleen of anti-CD25/KRN7000-treated mice. Secondly, glycolipid activation of iNK T cells in vivo and restimulation in vitro enhanced significantly cytokine secretion by iNK T cells from anti-CD25-treated compared to IgG-treated mice. In particular, KRN7000 experienced iNK T cells secreted much more IFN-γ than C20:2-activated iNK T cells, not only when left unstimulated but also after restimulation with glycolipid. Presumably, this increase in IFN-γ secretion contributed to the development of T1D. In contrast, iNK T cells from C20:2-treated mice retained their ability to secrete Th2 cytokines and were protected from T1D. Taken together, our findings indicate that the inactivation of Tregs augments the function of iNK T cells and are consistent with the idea that Tregs can down-regulate iNK T cell-mediated immune responses. Thus, anti-CD25 mAb treatment may ‘unmask’ the potentially harmful side effects of KRN7000-activated iNK T cells, whereas the effect of this treatment on C20:2-treated iNK T cells is minimal. This observation may be relevant clinically to prediabetic patients who may be deficient in the number and/or function of Tregs. Thus, an intervention protocol that treats prediabetic subjects with C20:2 and relies less on Treg activity may increase the benefit : risk ratio of these subjects due to its reduced ability to stimulate proinflammatory responses.
Several differences in the activation of iNK T cell-mediated responses by C20:2 and KRN7000 were observed. One of the most striking was the inability of C20:2 to sustain an increase in the level of IFN-γ in the sera of C20:2-treated mice even after Treg inactivation. As iNK T cells are resistant to polarization , this response was due probably to the inability of C20:2-activated iNK T cells to sustain the activation of a downstream target cell(s). NK cells are probable candidates, as they are major contributors to IFN-γ synthesis. Depletion of NK cells by treatment with anti-asialo-GM1 antibody can abrogate IFN-γ in serum induced by KRN7000, a mechanism that is dependent upon IL-12 signalling . Our results presented here demonstrate that C20:2 may not sustain the activation of NK cells and B cells as measured by their lower surface expression of CD69. We also found a reduction in the frequency of intracellular IFN-γ-expressing NK cells as late as 24 h after C20:2 treatment. Despite this reduced ability of C20:2 to sustain cellular transactivation, the initial activation of iNK T cells by C20:2 at an earlier time-point (6 h) was equivalent to that seen with KRN7000 and sufficient to induce early bursts of NK cell and B cell activity. No significant differences in CD69 expression were detected on NK cells and B cells from anti-CD25-treated versus IgG-treated mice. Therefore, the major effect of Treg inactivation by anti-CD25 appears to be on the modulation of cytokine secretion. Our findings suggest that that the initial burst of activation of iNK T cells stimulated by C20:2 is not deficient, but rather that C20:2-activated iNK T cells fail to sustain their initial level of iNK T cell activation and transactivation that leads to an increase in the serum concentration of IFN-γ.
The capacity of iNK T cells to interact and modulate DC activity by co-stimulation is important for the translation of iNK T cell signals to immune responses. The initial iNK T cell cytokine burst induced by KRN7000 occurs independently of CD40–CD40L and CD80/86–CD28 signalling , but the expansion and immunogenicity induced by KRN7000-activated iNK T cells is dependent on these signalling pathways [26,34]. iNK T cells activated from mice deficient in CD40–CD40L signalling do not undergo expansion and cannot induce T cell activation to third-party antigens such as ovalbumin (OVA) [26,34]. As reported for KRN7000 , we found that C20:2 can activate and mature CD11chigh mDC cells upon administration of glycolipid. Up-regulation of MHC class II, CD86, and CD40 were detected at 6 h post-treatment. However, as observed for NK and B cell transactivation, mDCs from C20:2-treated mice were less capable than mDCs from mKRN7000-treated mice in sustaining the surface expression of co-stimulatory molecules required for iNK T cell expansion. The latter observation may explain why C20:2-treated iNK T cells induced less IL-12 secretion by CD11chighCD8+ DCs, which may account for the reduced levels of serum IFN-γ detected in downstream responses. Interestingly, the activation of iNK T cells can induce the migration of CD11chigh DCs capable of ‘tolerizing’ autoreactive T cells in the PLN of NOD mice [12,13]. As the transfer of syngeneic PLN DC into young NOD mice protects against T1D [12,35], iNK T cell recruitment of DC may restore a deficiency in DC function inherent in NOD mice [35,36]. Our observations also indicate that glycolipid treatment modulates the frequency and function of DCs. Furthermore, we did not detect differences in the frequency of pDCs upon iNK T cell activation, although the function of these pDCs was not assayed. While glycolipid administration stimulated the recruitment of CD11chighCD8– mDCs to the PLN of NOD mice, as reported previously , the frequency of CD11chighCD8+ mDCs recruited to the PLN using our treatment regimen was lower in C20:2-treated than KRN7000-treated mice. Importantly, CD11chighCD8+ mDCs were found to be the major producers of IL-12. In addition, iNK T cell-mediated recruitment of DC is probably dependent upon the glycolipid treatment regimen. One weekly dose of C20:2 administered for 7 weeks resulted in an increased recruitment of DC subsets in the PLN of NOD mice at 30 weeks of age , whereas the current study indicates that treatment every other day for 3 weeks starting at 4 weeks of age resulted in a reduced frequency of CD11chighCD8+ mDC when analysed at 8–9 weeks of age. The difference in treatment regimen and/or age of the mice at time of killing may account for the apparent difference in DC subset proportions observed. None the less, both results suggest that glycolipid treatment modulates the frequency of recruited DC, but how this correlates with reduced T1D requires further experimentation and is beyond the scope of this study. Interestingly, Treg inactivation resulted in a significant increase in the recruitment of both CD11chighCD8– and CD11chighCD8+ mDCs in KRN7000-treated mice, suggesting that DC recruitment alone is insufficient to prevent T1D in the absence of functional Tregs. Thus, our observations using the current treatment regimen suggest that in vivo treatment of NOD mice with C20:2 diminish the activation of ‘inflammatory’ IL-12-producing CD11chighCD8+ mDCs and augment the function of ‘tolerogenic’ DCs more effectively than treatment with KRN7000. Moreover, the latter findings may explain why C20:2-activated iNK T cells depend less than KRN7000-activated iNK T cells on regulation by Tregs for cytokine secretion and protection from T1D.
Our study demonstrates that Tregs play either a major or more minor role in the control of iNK T cell function that varies according to the durability of the activation signal elicited by a given glycolipid. Both the KRN7000 and C20:2 forms of α-GalCer induce a strong early signal in iNK T cells. The signal induced by KRN7000 sustains iNK T cell activation for a sufficiently long time (several days) to induce robust downstream responses, and Tregs are required for iNK T cell-mediated protection from T1D. Conversely, the signal induced by C20:2 in iNK T cells is not sustained beyond 6 h, reduced downstream responses are stimulated and iNK T cells are less dependent on the activity of Tregs for iNK T cell protection from T1D. These results are consistent with a model in which the extent of Treg suppression required for a given immune response correlates directly with the level of T effector cell activation . Accordingly, iNK T cells activated by KRN7000 promote a more ‘proinflammatory’ response that appears to increase the requirement for Treg activity to return iNK T cells to a homeostatic level of activation (Fig. S1). Alternatively, activation of iNK T cells by C20:2 yields a more anti-inflammatory response that is less dependent on Tregs for the maintenance of iNK T cell homeostasis. INK T cell activation by a Th2-biased ligand such as C20:2 may provide a better outcome and a ‘safer’ therapeutic alternative due to its decreased propensity to elicit proinflammatory responses. Thus, our study further underscores the importance of selective iNK T cell modulation for the immunotherapy of T1D.
We thank the staff of the Robarts Barrier Animal Facility for their assistance with the breeding and maintenance of our mice. This work was supported by grants from the Juvenile Diabetes Research Foundation International (24-2007-388), Canadian Institutes of Health Research (MOP 64386) and Ontario Research and Development Challenge Fund to T. L. Delovitch; National Institutes of Health Grants RO1 AI45889 and RO1 AI064424 to S. A. Porcelli; and the Medical Council and The Wellcome Trust (084923/B/08/7) to G. S. Besra. During these studies, D. Ly was the recipient of a Canadian Diabetes Association Doctoral Student Award; R. Tohn and H. Blumenfeld were the recipients of a Schulich Graduate Enhancement Scholarship Award; B. Rubin was supported by the Centre National de la Recherche Scientifique, CNRS, France; S. A. Porcelli was the recipient of an Irma T. Hirschl Career Scientist Award; G. S. Besra was supported by a Personal Research Chair from Mr James Badrick, Royal Society Wolfson Research Merit Award and Lister Institute–Jenner Research Fellowship; and T. L. Delovitch was the Sheldon H. Weinstein Professor in Diabetes at the Robarts Research Institute and University of Western Ontario.
All the authors declare there are no conflicts of interest.