Dendritic cell (DC) differentiation is abnormal in type 1 diabetes mellitus (T1DM). However, the nature of the relationship between this abnormality and disease pathogenesis is unknown. We studied the LPS response in monocytes and monocyte-derived DCs isolated from T1DM patients and from non-T1DM controls. In T1DM patients, late LPS-mediated nuclear DNA binding by RelA, p50, c-Rel, and RelB was impaired as compared with type 2 DM, rheumatoid arthritis, and healthy subjects, associated with impaired DC CD40 and MHC class I induction but normal cytokine production. In TIDM monocytes, RelA and RelB were constitutively activated, and the src homology 2 domain-containing protein tyrosine phosphatase (SHP-1), a negative regulator of NF-κB, was overexpressed. Addition of sodium stibogluconate, a SHP-1 inhibitor, to DCs differentiating from monocyte precursors restored their capacity to respond to LPS in ∼60% of patients. The monocyte and DC NF-κB response to LPS is thus a novel phenotypic and likely pathogenetic marker for human T1DM. SHP-1 is at least one NF-κB regulatory mechanism which might be induced as a result of abnormal inflammatory signaling responses in T1DM monocytes.
Type 1 diabetes mellitus (T1DM)4 is an organ-specific autoimmune disease characterized by inflammatory infiltration of the pancreatic islet and destruction of the pancreatic β cells by autoreactive T cells. Dendritic cells (DCs) are important APCs that are instrumental to the initiation and perpetuation of autoimmune diseases such as T1DM (1, 2). In the generation of tolerance, DCs participate in thymic-negative selection of autoreactive T cells and presentation of self Ag in the periphery. In infection and stress, DCs mature rapidly when presented with pathogen- and stress-associated molecular patterns (3), associated with the ability to stimulate Ag-specific pathogenic immunity, and activation of the NF-κB pathway.
NF-κB is a transcription factor family comprising p50/p105, p52/p100, RelA, RelB, and c-Rel. In resting cells, before activation, RelA, p50 c-Rel, and RelB form complexes with either IκBβ and NF-κB essential modulator or p100, where they are sequestered in the cytoplasm. Upon activation, the inhibitors are phosphorylated, ubiquitinated, and degraded in the proteasome after interaction with IκB kinases (IKK)-β and -γ, or IKKα and NF-κB-inducing kinase. Active NF-κB molecules translocate to the nucleus where they are transcriptionally active. p100 is an inhibitory precursor of p52. Upon activation, p52 is derived from p100 by phosphorylation and proteosomal processing. This process leads to the release of p52/RelB dimers for nuclear translocation (“alternate” NF-κB pathway) (4, 5).
Stimuli such as TNF, LPS, and CD154 signal through the adaptor molecule TNFR-associated factor 6 (TRAF6) and activate NF-κB, thus allowing translocation of released NF-κB dimers to the nucleus; these dimers then bind to nuclear DNA (6). TRAF6 functions as a ubiquitin ligase, recruiting kinases into a multimeric complex after cell activation, leading to phosphorylation of IκB, p100, and p38 MAPK (7). Activated NF-κB promotes the expression of many genes, particularly mediators of the innate immune response (4). RelB has been most directly associated with DC differentiation and functional maturation, although the other subunits of the NF-κB complex also contribute to the process. RelB is detected in the nucleus of mature interdigitating DCs of the lymphoid organs in inflamed tissues and in activated DCs in vitro (8, 9). RelB can act as either a transcriptional activator or repressor of its target genes. Only few genes specifically activated or repressed by RelB are known. RelB has been shown to activate the transcription of the MHC class I gene. Furthermore, RelB-deficient DCs specifically lack expression of CD40 (10).
Observations in T1DM patients suggest dysregulation in the T cell response to Ag presentation by DCs. Monocytes isolated from T1DM patients or first-degree relatives and differentiated to DCs ex vivo and stimulated with either tri-iodothyronine or GM-CSF and IL-4 exhibited reduced B7 expression and a reduced capacity to cluster with, and activate, T cells, in comparison to monocyte-derived DCs from healthy controls (HC) (11, 12, 13). Given the role of NF-κB in DC differentiation, we further investigated DC dysfunction in T1DM by assessing NF-κB function in monocyte-derived DCs and circulating monocytes (14).
Materials and Methods
Cell preparation and culture
Peripheral blood was collected from 60 T1DM, 10 type 2 diabetic (T2DM), 10 rheumatoid arthritis (RA) patients, and 50 HC into heparinized tubes. Patients with T1DM included 60 insulin-dependent patients (age range 7–72 years, disease duration 1–60 years, mean age 33 years, 45% female) who had neither concomitant infection, nor severe hyperglycemia at the time of venesection. HbA1c levels averaged 7.4–7.6% in T1DM and T2DM patients. The diagnosis of T1DM was based on clinical and biochemical parameters, including age, weight loss, episodes of ketoacidosis, islet autoantibody status, and autoimmune diathesis at presentation. The diagnosis of T2DM was established according to standard criteria (lack of insulin dependence, common coexistence of other features of metabolic dysfunction, and clinical response to oral hypoglycemic agents). None of the T1DM or T2DM patients received immunosuppressive medication although they were treated with other necessary medications, such as antihypertensives, diuretics, and lipid-lowering agents. RA patients were treated with disease-modifying drugs, such as methotrexate, sulfasalazine, and plaquenil. The study was approved by the Human Ethics Committees of the Princess Alexandra and Mater Children’s Hospitals. PBMC were prepared as described (6). For DCs, monocytes were purified from PBMC using CD14 microbeads (Miltenyi Biotec), then cultured in 24-well plates in X-VIVO 20 (BioWhittaker) with 800 U/ml recombinant human GM-CSF and 600 U/ml recombinant human IL-4 (Schering-Plough) for 72 h. In some experiments, 10 μg/ml sodium stibogluconate was added at the commencement of DC culture. LPS (100 ng/ml) was added to some wells for the last 24 h of the culture period.
RelB DNA binding
Nuclear extracts were prepared according to the iso-osmotic/Nonidet P-40 method (6, 8, 15) from DCs cultured for 72 h, or from PBMC, to which LPS or sodium stibogluconate were added for the last 24 h. Extracts were preserved at −70°C and protein concentration was estimated using a micro BCA Protein Assay kit (Pierce). DNA binding by NF-κB was measured with a chemiluminescent ELISA-based kit (E-Z detect transcription factor kits for NF-κB p50 or RelA/p65; Pierce). DNA binding by c-fos was measured with the E-Z detect c-fos transcription factor kit (Pierce), both as described (6). NF-κB family members were detected with rabbit Abs against RelA (sc-372), RelB (sc-226), c-Rel (sc-70), p50 (sc-7178) (Santa Cruz Biotechnology), or p52 (Upstate Biotechnology). Chemiluminescence was detected with a luminometer (Berthold Detection Systems); luminometric data were expressed as photon units or converted to fold difference after LPS treatment, relative to control samples without LPS.
Human monocyte-derived DCs were cultured for 72 h, stained with mAb, and analyzed using a FACSCalibur flow cytometer. Cells were stained with the following mAbs: CD1a-FITC, CD86-FITC, isotype IgG1-PE (Cymbus Biotechnology), CD11c-FITC, HLA-A,B,C-FITC, CD83-PE, isotype IgG1-FITC (BD Pharmingen), CD40-FITC, HLA-DR-FITC, CD80-FITC, mouse IgG2a-FITC (Biolegend).
For intracellular staining, PBMC were fixed with 2% paraformaldehyde and permeabilized with 90% cold methanol for 30 min. Cells were washed with perm/wash buffer (BD Biosciences), then stained with either: anti-phospho-p38 MAPK (T180/Y182)-Alexa Fluor 647 (BD Pharmingen); anti-src homology 2 domain-containing protein tyrosine phosphatase (SHP-1; Upstate Biotechnology); or anti-phospho-IκBα (Santa Cruz Biotechnology). Cells stained with anti-SHP-1 or anti-phospho-IκBα were stained with biotinylated goat anti-rabbit Ig and then with streptavidin-FITC. Isotype controls were included for every sample. Cells from diabetic patients and HC were included in every run; the same flow cytometry settings were maintained throughout. Some samples were run repeatedly on consecutive days to confirm that the result did not change with time. The intra-assay coefficient of variation was 1%. Delta mean fluorescence intensity (ΔMFI) was calculated using the CellQuest Pro software by subtraction of the isotype-matched control MFI from the sample specific marker MFI.
Sodium stibogluconate (Calbiochem) was dissolved at 1 mg/ml in water using the MiniBeadbeater (Biospec Products). The p38 MAPK inhibitor SB230580 (Promega) was used at 10 μg/ml.
Cytokines in DC or cultured PBMC supernatants were measured by ELISA. IL-10, IL-1β, IL-12p70, and TNF were measured with OPEIA kits (BD Pharmingen), Max Set Standard kits (BioLegend), or the Cytometric Bead Array (CBA) Human Inflammation kit (BD Biosciences).
Total RNA was extracted from 106 DCs using TRIzol (Invitrogen Life Technologies). First-strand cDNA was synthesized from RNA by addition of random primers (Promega), dNTPs, RNase Out, and SuperScript III Reverse Transcriptase (Invitrogen Life Technologies) and used as a template for real-time PCR. GAPDH and NF-κB primers and cycling conditions were as described (6). Standard curves were generated for GAPDH and NF-κB using LPS-stimulated DC cDNA with serial dilution. mRNA levels were measured using the threshold cycle and corresponding standard curves. The relative amount of mRNA for each NF-κB subunit was calculated by normalization to the level of GAPDH. Relative expression of LPS-induced NF-κB genes in the 2- and 24-h time periods was expressed as a fold increase over the relative expression of NF-κB in untreated cells.
Differences were analyzed using paired or unpaired Student t tests, or for non-Gaussian data sets the Mann-Whitney U test (two data sets), ANOVA (more than two data sets). For sensitivity and specificity, receiver-operator curves were generated (Graph Pad Prism version 4 software).
NF-κB activity induced by LPS and expression of cell surface markers in T1DM DC
We demonstrated constitutively low nuclear NF-κB expression in monocyte-derived DCs cultured for 72 h from healthy donors (6). Furthermore, such DCs efficiently translocated NF-κB to the nucleus after activation with TNF, LPS, or CD154 (6). In monocyte-derived DCs, LPS activates both classical and alternate NF-κB pathways (unlike in B cells where LPS and TNF predominantly activate the classical pathway) (4, 16, 17). Because DCs from T1DM patients differentiate abnormally and NF-κB is required for the normal differentiation of DCs, we first examined the ability of NF-κB to bind its consensus DNA-binding motif and therefore to initiate transcription. DNA-binding activity due to the classical and alternate NF-κB pathway was analyzed by ELISA, and measured with a sensitive luminometric technique. Nuclear extracts were prepared from DCs generated from peripheral blood (PB) monocytes drawn from T1DM patients and HC. The DCs were derived by incubating PB monocytes for 72 h in serum-free medium containing GM-CSF and IL-4. Half the DCs received LPS for the final 24 h of incubation to induce differentiation. Although DNA binding by nuclear RelB, RelA, c-Rel, and p50 increased in response to LPS for DCs derived from HC, such DNA binding did not increase and often decreased in response to LPS in DCs derived from T1DM patients. An example of NF-κB DNA binding with or without LPS in DCs derived from a HC and from a T1DM patient is shown in Fig. 1,a. Fig. 1,b shows the fold change in LPS-stimulated NF-κB binding of DNA in nuclear extracts from DCs derived from T1DM and from HC groups. In T1DM patients, RelB, RelA, p50, and c-Rel LPS-induced DNA binding was significantly lower than that for HC DCs; p52 binding was low, and equivalent in each group probably reflecting inefficient p52 activation by LPS (Fig. 1,b). As reduced responsiveness to LPS in DCs might be a characteristic of cells derived from a hyperglycemic environment, or might occur generally in patients with autoimmune disease, we compared RelB DNA binding in LPS-exposed DC derived from patients with T2DM and RA. LPS-induced RelB DNA binding in DCs derived from these patients did not differ from that found in DCs derived from HC (Fig. 1 b), indicating that the NF-κB dysfunction is T1DM DC specific. Although hyperglycemia has previously been associated with NF-κB activation as a result of advanced glycation end products (18), this is unlikely to account for the LPS-induced NF-κB repression observed here. Moreover, NF-κB repression was not observed in our patients with T2DM, who had similar glycemic control and blood glucose levels at the time of venesection to the patients with T1DM.
NF-κB expression is regulated transcriptionally and posttranslationally (19). Transcription of NF-κB family members in response to LPS was examined by quantitative real-time PCR. All mRNA levels were initially normalized to GAPDH. NF-κB transcription induced after 2 or 24 h exposure to LPS was expressed as the fold increase over the NF-κB mRNA level in untreated DCs. There were no significant differences between T1DM DCs and HC in the induction of RelA, RelB, c-Rel, p100, or p105 mRNA by LPS for 2 or 24 h (Fig. 1,c). When NF-κB DNA-binding values for LPS-untreated DCs from HC and T1DM subjects were compared, we found no significant difference for any of the NF-κB subunit (Fig. 1,d). Therefore, to further characterize the LPS-induced NF-κB response, DCs derived from HC subjects or T1DM patients were treated for 0, 2, 6, or 24 h with LPS and the time-dependent DNA binding of RelA, p50, c-Rel, and RelB was determined. RelB, RelA, c-Rel, and p50 from T1DM DCs generally showed an increase in DNA binding between 2 and 6 h after LPS stimulation, consistent with the observed normal NF-κB transcription. This was followed by a decrease in the DNA binding of each subunit. In HC DCs by contrast, DNA binding by NF-κB increased between 1 and 24 h after LPS addition. Representative individual plots are shown in Fig. 1 e. Taken together, the data demonstrate a specific repression of NF-κB nuclear activity in T1DM DC after addition of LPS, likely due to posttranslational NF-κB regulatory mechanisms.
After 72 h of differentiation ex vivo, monocyte-derived DCs from T1DM or control subjects expressed CD11b, low levels of surface CD40, HLA-DR, MHC class I, CD86, and CD80. None of the DCs expressed CD14 or CD83, consistent with an immature DC phenotype (Fig. 2,a and data not shown). LPS stimulation of DCs from HC, T2DM, and RA patients for the final 24 h of culture up-regulated expression of HLA-DR, CD86, CD80, CD83, CD40, and MHC class I. T1DM DCs stimulated with LPS in the final 24 h of culture also up-regulated expression of HLA-DR, CD86, CD80, and CD83, but expressed significantly fewer cell surface CD40 and MHC class I molecules. A similar DC phenotype was observed after incubation of T1DM DCs with either CD154 or TNF (data not shown). Of interest, there was no significant difference in the production of IL-12p70, IL-1β, TNF, and IL-10 by resting or LPS-stimulated DCs derived from HC, T1DM, T2DM, or RA patients (Fig. 2,b). Furthermore, LPS stimulated similar levels of HLA-DR, CD80, CD86, and CD83 expression in DCs in all disease groups and in the HC (Fig. 2 a). These data demonstrate a specific reduction in LPS-induced CD40 and MHC class I expression in T1DM DCs, with no effect on induction of other costimulators, or of cytokine production.
p38 MAPK activity and cytokine production by T1DM DCs
Because NF-κB and p38 MAPK both contribute to DC activation and cytokine production (6, 20, 21), the p38 MAPK pathway was investigated by comparison of LPS-induced phospho-p38 MAPK expression and c-fos nuclear DNA binding in patients with T1DM and in HC. Equivalent amounts of intracellular phospho-p38 expression were induced 15–30 min after stimulation of both HC and T1DM DCs with LPS (Fig. 3,a). In contrast, c-fos nuclear DNA binding to the specific c-fos DNA motif was reduced 24 h after stimulation of T1DM DCs with LPS, as compared with the binding by nuclei of LPS-stimulated HC DCs (Fig. 3,b). To determine the contribution of p38 MAPK signaling to the up-regulation of cytokines in response to LPS, DCs were cultured for 24 h with LPS in the presence or absence of a p38 MAPK inhibitor (SB230580). This inhibitor partially blocked the secretion of cytokines in the LPS-treated control DCs and the T1DM DCs (Fig. 3,c). These data suggest that p38 MAPK signaling only partly accounts for the stimulatory effect of LPS on cytokine expression in the face of reduced NF-κB activation in T1DM DCs. In addition, the early NF-κB activation in response to LPS may have been sufficient to activate the transcription of cytokines. Indeed, production of cytokines was significantly up-regulated 6 h after stimulation with LPS in both T1DM and control DCs (Fig. 3,d). Furthermore, the cytokine level after 24 h of LPS stimulation was similar to that after 6 h of LPS stimulation in both T1DM and control DCs (Fig. 3 c). Taken together, these data suggest that the NF-κB and p38 MAPK pathways in T1DM DCs are intact and can be activated at an early stage by LPS, despite later repression of both pathways in T1DM DCs. This activation is sufficient to support the expression of proinflammatory cytokines and some costimulatory molecules.
Abnormal NF-κB function in circulating monocytes isolated from T1DM patients
NF-κB activity is also required for DC differentiation from bone marrow progenitors or blood monocyte precursors. This means that abnormal NF-κB activity in the precursors can alter the survival, phenotype, and function of the DC product (6, 21, 22, 23, 24). Therefore, we assessed the binding of NF-κB to DNA in PB monocytes with or without LPS stimulation for 24 h. These cells were derived from T1DM subjects or HC. We found that LPS-induced DNA binding by NF-κB was also reduced in monocytes derived from T1DM patients as compared with HC (Fig. 4,a). When NF-κB binding values for LPS-untreated monocytes from T1DM and HC subjects were compared (Fig. 4,b), RelB and RelA DNA binding was significantly higher in T1DM subjects (p < 0.05). Induction of phospho-IκBα in PB monocytes after 30 min of LPS treatment was also impaired in T1DM, as compared with the HC (Fig. 4 c). These data indicate that RelA and RelB nuclear DNA binding is constitutively elevated and LPS-induced NF-κB activation is impaired in T1DM monocytes. LPS-induced NF-κB activation is similarly impaired in T1DM monocyte-derived DCs relative to HC. This did not vary according to age, sex-or duration of diabetes. Receiver-operator characteristic curves were generated to determine a cutoff fold RelB or RelA response to LPS, separating healthy and T1DM subjects. The RelB cutoff predicted T1DM with 86% sensitivity and 79% specificity (p < 0.001 for area under the receiver-operator characteristic curve compared with line of identity), and the RelA cutoff with 79% sensitivity and 92% specificity (p < 0.001).
Increased SHP-1 expression by T1DM PBMC
Constitutive NF-κB activation followed by LPS-induced repression suggested overactivation of NF-κB regulatory pathways in T1DM monocytes and DCs. SHP-1 inhibits NF-κB translocation (25), potentially by interaction with TRAF6. SHP-1 is a classical nonreceptor protein tyrosine phosphatase, transcriptionally activated by NF-κB that negatively regulates cell signaling and is abundant in hemopoietic cells, e.g., monocytes and lymphocytes (26). The motheaten viable mutant mouse strain (mev/mev) expresses a catalytically defective SHP-1 protein and was used to show that SHP-1 is a negative regulator of immune cell function (27, 28). Increased NF-κB activation and high levels of MHC class I have been observed in mev/mev mice, as well as in SHP-1-deficient cells (29, 30). We observed that LPS-induced CD40 expression is increased in mev/mev DCs relative to wild-type DCs. Conversely, LPS-induced CD86 expression is equivalent in DCs derived from both mev/mev and wild-type mice (our unpublished data), hence changes in mev/mev DCs were the reverse of changes seen in T1DM DCs following activation by LPS. To evaluate a potential contribution by SHP-1 to the T1DM phenotype, we quantitated relative expression by intracellular staining and flow cytometric analysis of PBMC. Expression of SHP-1 by resting PB monocytes was significantly higher in ∼60% of T1DM patients as compared with SHP-1 expression in resting PB monocytes from HC (Fig. 5,a). Prolonged cell culture decreased SHP-1 expression, in that SHP-1 expression was low in both control and T1DM monocyte-derived DCs that were cultured for 48 h in GM-CSF and IL-4. Expression of SHP-1 by PB monocytes is indicative of SHP-1 expression by lymphocytes in both diabetic patients and HC. This implies a similar process regulating SHP-1 expression in PB monocytes and in lymphocytes of T1DM subjects (Fig. 5 b). SHP-1 expression level did not vary with age or sex in T1DM patients or HC. These data indicate that PB monocytes in ∼60% of the T1DM subjects expressed SHP-1 at higher levels than found in PB monocytes from HC.
Overexpression of SHP-1 by monocytes inhibits NF-κB activity and DC differentiation in some T1DM patients
A dysfunctional NF-κB response to LPS was evident in both monocyte-derived DC cultures and PB monocytes. However, SHP-1 expression was low in monocyte-derived DCs from T1DM patients but high in PB monocytes. We hypothesized that overexpression of SHP-1 in monocytes might inhibit their NF-κB-dependent differentiation to DCs. To address this, we assessed the effect of the SHP-1 inhibitor sodium stibogluconate (15) on the NF-κB response in differentiating DCs in the presence and absence of LPS. Sodium stibogluconate was added to PB monocytes and the cells were cultured for 48 h with GM-CSF and IL-4. We found that sodium stibogluconate promoted DC differentiation in T1DM monocyte-derived DCs cultures. LPS treatment of these cells stimulated cell surface CD40 expression and DNA binding by RelA, p50, and RelB (Fig. 5,c, upper panel). These effects were noted in cells derived from ∼60% of T1DM donors. However, cells derived from the remaining donors and from HC did not respond to sodium stibogluconate (Fig. 5 c, lower panel, and data not shown). The capacity of an individual to respond to sodium stibogluconate appeared to be associated with elevated untreated monocyte RelA and RelB levels and a suppressed LPS-induced NF-κB response, rather than the absolute level of SHP-1 expression (data not shown). Taken together, these data indicate that NF-κB function is abnormal in patients with T1DM. PB monocytes demonstrate constitutive RelA and RelB DNA-binding activity, elevated SHP-1 expression, and reduced capacity for LPS-induced NF-κB activation. After differentiation from this dysfunctional precursor population, the capacity of monocyte-derived DCs for LPS-induced NF-κB and c-fos activation is also impaired, characteristically at late time points after LPS. SHP-1 overexpression contributes to suppression of NF-κB activity in a subset of the T1DM population. These data are consistent with a model in which disease-associated NF-κB activation promotes expression of NF-κB regulators which modulate the response to LPS.
DCs present Ag during thymic selection, determine the repertoire of potentially self-reactive and immunoregulatory T cells, and present self and foreign Ags in the periphery (31). Therefore, pathways controlling the function of DCs impact the T cell repertoire as well as the peripheral immune response that controls infection and prevents autoimmune disease. NF-κB and p38 MAPK pathways are the major controls for DC differentiation and maturation (22, 32). RelA and p50 are required for DC survival and the classical NF-κB pathway. p38 MAPK activation is required for induction of costimulatory molecules and production of proinflammatory cytokines. Furthermore, RelB is required for the development of monocytes and monocyte-derived DCs, as well as functional DC maturation (21, 22, 23). RelB also plays a critical role in the development of medullary thymic epithelial cells, and thus in the regulation of thymic-negative selection. Indeed, negative selection of autoreactive T cells is reduced in RelB−/− mice. This leads to severe systemic autoimmune inflammation, including of the exocrine and endocrine pancreas (33). RelB is also a key regulator of NF-κB (34). Particularly after LPS, PMA, or TNF signaling, RelB can hetero-dimerize with RelA and prevent NF-κB binding (35, 36). Classical and alternate NF-κB pathways also contribute to the development of natural T regulatory cells (31, 37, 38).
In the current study, immature DCs were derived from monocytes ex vivo for 2 days in the absence of serum. This was followed by treatment with LPS to induce activation (6). Our experimental approach differs from those reported in previous publications in which DC differentiation or APC function was examined in T1DM patients or first-degree relatives (39, 40). Previous protocols used longer culture periods or different activation stimuli. Conversely, our current DC differentiation protocol was designed to optimize NF-κB activation in immature DCs. To exclude the possibility that these variations in DC culture conditions accounted for our data, we also assessed constitutive and LPS-induced NF-κB activity in primary PB monocytes. The data indicate that monocytes demonstrate higher constitutive RelA and RelB DNA-binding activity, elevated SHP-1 expression, and reduced capacity for LPS-induced NF-κB activation. After differentiation from this dysfunctional precursor population, the capacity of monocyte-derived DCs for LPS-induced NF-κB and c-fos activation is also impaired, characteristically at late time points after LPS (Fig. 6). LPS-induced NF-κB transcription was intact. This is consistent with a posttranslational abnormality of NF-κB regulation. LPS-induced cytokines and costimulatory molecules on the DC cell surface were expressed at similar levels in T1DM and healthy subjects, with the exception of MHC class I and CD40. Reduced capacity of T1DM DCs to stimulate allogeneic T cells was observed previously (40). The importance of the expression of the CD40 receptor by APC for T cell cross-talk, leading to appropriate levels of costimulation, has been well-documented (14, 41, 42). The consistency of the dysfunction within the T1DM population strongly implicates abnormal NF-κB monocyte and DC signaling as a necessary component of T1DM pathogenesis in humans. Indeed, the monocyte RelA or RelB response to LPS showed high sensitivity and specificity for separation of T1DM subjects from HC.
RelB normally translocates late to the DC nucleus after exchange of NF-κB dimers derived from the classical pathway (43). Hence, the NF-κB repression which we have observed to occur late after LPS signaling would be anticipated to have maximum impact on genes dependent on RelB nuclear dimer exchange, or genes requiring prolonged NF-κB signaling. Reduced expression of MHC class I and CD40 by T1DM DCs may have been due to the long duration of NF-κB transcriptional activity required to induce these genes (6). Both RelA and RelB contribute to CD40 expression by myeloid cells (23, 44). Furthermore, RelB activity is associated with induction of MHC class I (45). NF-κB is required for DC differentiation and survival and our data are consistent with previous evidence of abnormal blood DC differentiation in T1DM patients, and with a recent study showing reduced DC counts (suggesting poor DC survival) in the PB of children with recent-onset T1DM (11, 22, 40, 46).
The capacity of DCs to produce proinflammatory cytokines in response to upstream signaling by NF-κB may be important in promoting autoimmunity in T1DM. This may be especially important in the context of RelB or CD40 deficiency that can themselves promote autoimmunity with high levels of proinflammatory cytokines and chemokines (47, 48, 49, 50). It has been suggested that the monocyte-derived DC model may represent DC differentiation in vivo in the context of acute inflammation (51). Our data also suggest the intriguing possibility that disease-associated inflammation or environmental factors, or a genetically programmed abnormal response to inflammatory signals may trigger monocyte RelA and RelB activation, leading to a cascade of dysfunctional adaptive immunity in T1DM DCs. Potential factors impacting on monocyte response to inflammatory signals in T1DM include vitamin D deficiency and associations reported with a number of tyrosine phosphatase gene polymorphisms (52, 53, 54, 55, 56). Our human data are also of interest in view of previous reports of either NF-κB hyperactivation or dysregulation in the nonobese diabetic mouse model of T1DM (57, 58, 59), further implicating abnormal NF-κB responsiveness to inflammatory signals in the pathogenesis of T1DM.
Protein tyrosine phosphorylation is a ubiquitous signaling mechanism regulated by a balance between the action of kinases and phosphatases. SHP-1 plays a negative role in cellular signaling, including TCR signaling and T cell selection in the thymus, hence changes in SHP-1 expression can have profound effects on many systems (26, 60, 61, 62). SHP-1 is transcriptionally activated by NF-κB and degraded after tyrosine phosphorylation and ubiquitination in the proteasome (63). Our current data suggest that SHP-1 might be overexpressed in response to NF-κB activity in circulating monocytes. SHP-1 might then act as negative feedback on NF-κB activation in response to LPS in some T1DM patients. Alternative or additional pathways regulating NF-κB are also likely to be disturbed in the T1DM population. This is because SHP-1 was not overexpressed by monocytes from all patients and SHP-1 inhibition did not restore NF-κB activity in DCs derived from monocytes of all T1DM patients. Moreover, although T1DM DCs displayed a similar LPS-induced repression of NF-κB and c-fos activation, SHP-1 levels were equivalent in T1DM and healthy DCs. This suggests that alternative pathways, such as RelA/ RelB dimers, might be involved in the observed LPS-induced NF-κB regulation in DCs.
In T1DM there is an imbalance between autoreactivity and regulation of the immune system. This means that islet Ag-autoreactive T cells secrete proinflammatory cytokines generating autoreactivity. Dysfunctional regulatory T cells cannot keep this autoreactivity in check (64, 65). Consistent with the immunoregulatory abnormalities in DCs shown here, deficient T cell proliferation and impaired primary immune responses have been demonstrated in T1DM (13, 66). Indeed, anti-CD3 therapy in a phase II trial was often followed by an acute infectious mononucleosis response to EBV in patients with recent-onset T1DM. This suggests that anti-CD3 unmasked the dysregulation of viral immunity in T1DM (67). We hypothesize that the observed time-dependent NF-κB dysregulation associated with a focal abnormality of circulating monocytes, subtly impairs DC maturation, associated with impaired T cell costimulation. This could impact on the threshold for T cell activation during thymic selection and in the peripheral immune response (61), and development or activation of CD25+CD4+ regulatory T cells (37, 38), all affecting the peripheral response of T cells to self Ags and pathogens, and the maintenance of central and peripheral tolerance.
We thank Andrew Cotterill, David McIntyre, Tracey Baskerville, and Joyce Cotterill for assistance with clinical samples; Ian Frazer, Steven Gerondakis and William Burns for critical reading of the manuscript; and Matthew Brown and Herbert Strobl for helpful discussions.
Provisional patent pertaining to the subject matter filed by Thomas, Mollah, and Cardinal.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Grants 210237 and 351439 from the National Health and Medical Research Council, and by the Princess Alexandra Hospital Foundation. R.T. was supported by Arthritis Queensland and B.J.O. by a Queensland Government Smart State Fellowship.
Abbreviations used in this paper: T1DM, type 1 diabetes mellitus; DC, dendritic cell; IKK, IκB kinase; T2DM, type 2 diabetes mellitus; SHP-1, src homology 2 domain-containing protein tyrosine phosphatase; HC, healthy control; RA, rheumatoid arthritis; PB, peripheral blood; MFI, mean fluorescence intensity; TRAF6, TNFR-associated factor 6.