| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Establish the Regulatory Environment for the Control of Diabetogenic T Cells in the Nonobese Diabetic Mouse1
,¶,||
* Diabetes Research Center, Cincinnati Children’s Research Foundation, and
Division of Endocrinology,
Division of Rheumatology,
Division of Developmental Biology,
¶ Division of Ophthalmology, and
|| Division of Molecular Immunology, Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH 45229
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
cells leads to the acute onset of hyperglycemia. The nonobese diabetic mouse model of human T1DM reveals that T cells capable of inducing diabetes can escape normal central tolerance, and can cause T1DM if left unchecked. However, several regulatory T cell subsets can temper autoaggressive T cells, although it remains undetermined when and how, and by which subset, homeostatic control of diabetogenic T cells is normally achieved in vivo. Using a cotransfer model, we find that NKT cells efficiently dampen the action of diabetogenic CD4+ T cells, and do so in an indirect manner by modifying the host environment. Moreover, the NKT cell-containing population modifies the host via production of IFN-
that is necessary for driving the inhibition of diabetogenic T cells in vivo. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
cell self Ags seem required (2, 3, 4, 5). What controls or restrains the activation of autoreactive T cells to self Ags in individuals or mice that do not develop T1DM remain unsolved. In addition, what extent the breakdown of such immunoregulation plays in these individuals or mice that do develop T1DM also remains largely unresolved. Studies in NOD mice show dramatic differences in several regulatory T cell compartments, namely CD4+CD25+ (Treg), CD4+CD25–, and NKT cells (reviewed in Refs.1 and 6). The Foxp3-expressing Treg have been shown to play an important role in regulating T cell immune responses to self and foreign Ags, both in vivo and in vitro (6, 7, 8). Although the precise mechanism(s) is (are) enigmatic, it is clear that these T cells can modulate dendritic cell (DC) subpopulations in the host and can modify the activation and expansion of Ag-stimulated T cells (9, 10, 11). NOD mice appear to harbor reduced Treg numbers, and their modulation has a profound impact on T1DM in the NOD mouse (1).
However, Treg do not account for all of the regulatory function found in NOD mice. A number of studies have identified a critical role for NKT cells in modulating both adaptive and innate immune responses to self Ags (12, 13, 14). Like their Treg counterparts, these NKT cells have been reported to be "defective" in NOD mice (15). NKT cells, unlike other T cells, recognize glycolipid Ags presented by the CD1 molecule. Mice lacking CD1 or its covalently associated L chain, 2-microglobulin (2M), fail to develop NKT cells (16, 17). NKT cells can rapidly make either IL-4 or IFN-, without de novo mRNA synthesis (18), upon treatment with the superagonist 
-galactosylceramide (-GalCer) (19). NKT and DC subpopulations interact, presumably via CD1-TCR and CD40-CD40L, to alter each other’s function and maturation (11).
When NOD mice are treated with -GalCer, both NKT cell numbers and activity increase, while the incidence of spontaneous diabetes declines (13, 20, 21, 22, 23). Furthermore, when invariant NKT cells are artificially enhanced via transgenic expression of their V14J18 TCR- chain, the enrichment in invariant NKT cell production correlates with a subsequent decline in diabetes incidence (24), and the reduction of mature Th1, islet-responsive T cells (25). Activated NKT cells may inhibit T1DM through the recruitment of tolerogenic DC to the pancreatic lymph nodes (PLN) via unidentified soluble mediators (26).
When viewed in toto, there is clear and compelling evidence for T cell-mediated immunoregulation of diabetogenic T cells in the NOD mouse. In addition, both classical Treg and NKT cells can effect regulation, and Ag presentation by DC. But these studies also reveal that the nature of the T cell regulation is context dependent, and it remains an open question which regulatory population acts under native conditions in vivo. Moreover, the molecular mechanism(s) of action remain(s) poorly defined, with the notable exception of an Ab-blocking study suggesting that IL-4, -10, and -13 and TGF- are not required (27). Thus, the responsible regulatory cytokine and its target cell remain unresolved.
We have now modified the well-established BDC2.5 TCR model (28) to isolate, under physiological conditions, the regulatory and effector cell populations from each other and from host APC, thereby allowing us to study their interactions in isolation. T cells from the BDC2.5 TCR (BDC2.5/NOD) mice effectively transfer rapid onset autoimmune diabetes, thus demonstrating that these mice contain a significant population of diabetogenic effector T cells (28, 29). In their natural state, these BDC2.5 T cells are not capable of inducing a high incidence of spontaneous diabetes (28, 30), yet, when the selective introduction of the scid, Rag1 or Rag2, or Tcra mutations are used to engineer NOD mice that produce only T cells with BDC2.5 TCR, rapid diabetes ensues (28, 29, 30, 31). These data suggest that the rate-limiting event in realizing the diabetogenic potential of the BDC2.5/NOD mice is not in their transgenic TCR-encoded specificity per se, but rather the presence of other T cells that act to alter the course of diabetes in these mice. We hypothesized, therefore, that a distinct regulatory T cell population keeps the BDC2.5 T cells "in check." Indeed, when BDC2.5 T cells are cotransferred with normal NOD splenocytes from young (4- to 6-wk-old) mice, no diabetes ensues. We find that the loss of only the small subset of NOD splenocytes containing NKT cells, results in the complete abrogation of immune regulation of BDC2.5 T cells in vivo. How these NKT cells act to regulate the BDC2.5 T cells is not yet fully understood at a mechanistic level; however, it is clear that IFN- is a necessary soluble participant in the regulatory process at a molecular level, because IFN- must be made by the NKT cell-containing splenic subset for regulation to occur. Moreover, IFN- made by the regulatory cell population does not act directly on the diabetogenic T cells, but rather appears to alter the host, to establish a profound inhibition of diabetes in NOD mice, suggesting that IFN- may act to directly alter host APC function.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
BDC2.5 TCR transgenic mice (28) are maintained on NOD and NOD.scid breeding stocks. NOD and NOD.scid mice were originally purchased from The Jackson Laboratory. NOD.GFP mice, which express GFP, were a gift from Dr. R. Tisch, (University of North Carolina, Chapel Hill, NC). NOD mice deficient in IFN- (NOD.Ifng), IL-4 (NOD.Il4), IL-10 (NOD.Il10), IL-4 and IL-10 (NOD.Il4.Il10), IFN-R -chain (NOD.Ifngrb) and 2M (NOD.B2m) were a gift from Dr. D. Serreze (The Jackson Laboratory, Bar Harbor, ME) or purchased from The Jackson Laboratory (32). Both recipient and donor mice were 6–8 wk of age. All mice were housed and bred in the American Association for the Accreditation of Laboratory Animal Care-approved, specific pathogen-free facility at Cincinnati Children’s Research Foundation (CCRF). All experiments were performed under approval of the CCRF Institutional Animal Care and Use Committee with care given to ensure minimal pain and suffering.
Antibodies
mAbs to CD4 (clone GK1.5), CD8 (clone 53.6.7), CD11b (clone M1/70), CD45R (B220; clone RA3-6B2), CD25 (clone 7D4), V4 (clone KT4), and TCR- (clone H57-597) were purchased from BD Pharmingen. The anti-BDC2.5 Ab, BDC2.5 (33), was a gift from Dr. O. Kanagawa (Washington University, St. Louis, MO).
Quantification of NKT cells using CD1d-tetramers
NKT cells can be directly assessed using recombinant CD1d-tetramers loaded with the NKT superagonist, -GalCer. PE-conjugated CD1d-tetramers, a gift from Dr. M. Kronenberg (La Jolla Institute for Allergy and Immunology, La Jolla, CA), were produced and used as described (34, 35). Control staining was performed using CD1d-tetramers without -GalCer (vehicle control), although recent structural evidence by Giabbai et al. (36) has shown that the "empty" vehicle control tetramers are actually loaded with phosphatidylcholine. NKT cells were enumerated using CD4-FITC and CD1d-tetramers and size-gating for lymphocytes by flow cytometry.
Isolation of diabetogenic T cells
Spleens from 14- to 18-day-old BDC2.5./NOD.scid mice were harvested and processed to a single-cell suspension by dispersion with a ground glass homogenizer in HBSS buffer containing 3% BSA followed by removal of RBC by NH4Cl lysis procedure. The number of diabetogenic cells in the suspension was determined by flow cytometry on a FACScan instrument (BD Biosciences) by measuring the percentage of V4 TCR+ and CD4+ cells.
Preparation of regulatory T cell populations
Splenocytes from donor mice (4–8 wk of age) were processed to single-cell suspensions as described above for diabetogenic cells. If further purified, cells were isolated by enrichment for T cells by nylon wool depletion of B cells followed by labeling with FITC-conjugated Abs to CD45R (B220), CD11b and CD4 or CD8, and the negative population was collected, analyzed for purity by flow cytometry, enumerated, and transferred to host mice. In experiments to isolate the CD25+CD4+ cells, the splenocytes were costained with CD25-PE and the negative population (representing the CD4+CD25– cells) and the PE-positive population (CD4+CD25+ cells) were both collected. The CD25+ population in 6- to 8-wk-old NOD spleen is 0.5–1.0% of the total cell population. The number of cells cotransferred were as follows: 6–10 x 106/mouse for CD4+CD25– cells and 6–8 x 105/mouse for CD4+CD25+ cells, numbers equivalent to each subset of unfractionated cells in prior experiments.
Transfer model
The transfer model for diabetes was used as described (37), with the following modifications: splenocytes were transferred to recipient NOD.scid mice (6–8 wk of age) by the i.p. transfer of 1.0 x 105 BDC2.5 diabetogenic T cells with or without cotransferred regulatory cells. All transfers were normalized to those found in unfractionated splenocytes containing in 1 x 107 CD4+ T cells. Recipient mice were then followed for the onset of diabetes as determined by a standard one-step glucometer. Diabetes was dated from the first of 2 consecutive days of blood glucose levels of 
250 mg/dl (13.6 mM).
Pancreatic influx studies
Separate groups of NOD.scid mice were injected with BDC2.5 T cells alone (1 x 105) or in combination with GFP.NOD splenocytes as regulatory cells (1 x 107 CD4+ T cells). Pancreata were removed at the indicated days posttransfer and processed in pools of two pancreata per analysis. The pancreata were crushed with the rubber tip of a syringe plunger then further dispersed by drawing the suspension sequentially through 16-, 18-, and 20-ga hypodermic needles. The resulting cell homogenate was then filtered through a 40-µm filter basket to remove tissue debris followed by isolation of the mononuclear cell population by Percoll-gradient centrifugation. The percentage of diabetogenic cells was determined by flow cytometry by staining the enriched cell population with the BDC2.5 or V4 mAb. Total numbers of diabetogenic and regulatory cell populations were determined by multiplying the total number of cells by the percentage of GFP+ cells for the regulatory cell population or by percent BDC2.5 diabetogenic T cells.
Insulitis and diabetes
The incidence of diabetes and the enumeration and grading of insulitis were performed as described in Ref.28 . Multiple sections, 10 µm apart, were analyzed for each mouse; 35–50 islets per mouse were graded for insulitis.
Statistics
Statistical comparisons were performed using GraphPad Prism (version 4) software. Analysis inhibition of infiltration of BDC2.5 T cells and NOD.GFP splenocytes in Fig. 5 were performed as Student’s t test, unpaired datasets.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
We and others (28, 29, 30, 31) observe that the BDC2.5/NOD mouse fails to develop autoimmune diabetes as quickly or with the same penetrance as BDC2.5 TCR transgenic mice harboring reduced T cell complexity. That is to say, the BDC2.5/NOD.scid, BDC2.5/NOD.Tcra–/–, and BDC2.5/NOD.Rag1–/– mice, all of which have no or dramatically reduced T cell diversity, other than those expressing the BDC2.5 TCR and CD4 coreceptor, exhibit markedly accelerated diabetes. On a cell-to-cell basis, T cells expressing high levels of the BDC2.5 TCR, as revealed by the anti-BDC2.5 clonotypic (BDC2.5) mAb, show similar responsiveness to islet cell Ag in culture, regardless of the source of the BDC2.5 T cells themselves (33). Interestingly, these BDC2.5 TCRhigh T cells seem to develop early in the NOD mouse, but wane with age (33). This correlates well with the early nondestructive insulitis seen in these mice and the subsequent failure of these mice to progress to diabetes as more diverse T cells develop with time.
These findings led us to propose that the BDC2.5 TCRhigh T cells that develop in the NOD mice are controlled or regulated by other T cell subsets. Loss of these regulatory cells by the introduction of the scid, Tcra, or Rag1 mutations (29, 30, 31) allows BDC2.5 T cells to operate unfettered without natural controls and induce rapid insulitis and diabetes.
Alternatively, it can be argued that the BDC2.5 T cells in the BDC2.5/NOD mouse fail to develop diabetes owing to poor allelic exclusion of the transgenic TCR- chain. In this model, the transgenic BDC2.5 TCR- chain, which does exclude the expression of endogenous TCR- chains, pairs with endogenous TCR- molecules, thereby diluting out the transgenic BDC2.5 TCR- receptor. Studies by Kanagawa et al. (33) suggest that T cells from the BDC2.5/NOD mouse do coexpress additional TCR- pairings, but the expression of the BDC2.5 TCR-, especially early on, is quite strong. Using the anti-BDC2.5 mAb developed by Kanagawa and coworkers, we have found similar results. Sixteen- to 18-day-old BDC2.5/NOD and BDC2.5/NOD.scid mice do have differing levels of BDC2.5 TCR-, but the numbers of BDC2.5 TCRhigh cells is, in fact, larger on an absolute-cell-number basis in BDC2.5/NOD mice than in BDC2.5/NOD.scid mice of similar age (Fig. 1A). Yet only the BDC2.5/NOD.scid mice go on to develop robust diabetes. Therefore, based on our own work and that of others, we cannot find compelling evidence for a TCR density-based argument to explain the stark difference in disease outcome between these two NOD sublines.
|
CD4+, CD25– T cells regulate BDC2.5 T cells
To determine which cellular constituent in the NOD spleen is responsible for controlling the diabetogenic potential of the BDC2.5 T cells, we undertook a set of fractionation or ablation studies. We know from the rapid diabetes seen in BDC2.5/NOD.Tcra–/– mice that B cells and TCR-
+ T cells are not responsible for the immunoregulation (Fig. 2). Moreover, in similar transfer studies where NK cells are depleted with anti-asialo-GM1, no marked change in the tempo or severity of diabetes is observed (Fig. 3). It is worthwhile noting that similar to the results of Gonzalez et al. (38), both the DX5+ and DX5– populations of T cells contained regulatory function (data not shown). Importantly, however, when we enrich for CD4+ or CD8+ T cells from the splenic population of NOD mice by negative (nonactivating) cell sorting, it became clear that the regulatory T cells reside in the CD4+ but not the CD8+ subset (Fig. 4A). In addition, the double-negative T cell compartment, although present (4–6%) in both CD4+- and CD8+-enriched subsets, appeared not to participate in the regulatory process, or at least not without the help of CD4+ T cells (Fig. 4A).
|
|
|
NKT cells act as regulatory cells
The protective CD4+CD25– T cell population contains several different subsets that have been shown to participate in T cell regulation. One of these, NKT cells, is totally dependent on the class Ib CD1d molecule and its L chain partner 2M for their development (17, 19). Subsets of NKT cells coexpress CD4 and would, therefore, reside in the CD4+CD25– subpopulation of regulatory T cells found in Fig. 4C. Because mice lacking CD1 or 2M, fail to develop NKT cells, we cotransferred BDC2.5 T cells along with CD4+ cells from NOD.B2m–/– mice to assess regulation. As seen in Fig. 5, unlike T cells from 2M+/+ NOD mice, T cells from NOD.B2m–/– mice could no longer inhibit diabetes transferred by BDC2.5 T cells. Although mice deficient in the B2m gene, also lack CD8+ T cells, CD8+ T cells did not act to regulate BDC2.5 T cells (Fig. 4A). NOD.B2m–/– mice also lack other T cell subsets dependent on class I and class Ib MHC, but to date, we have not found any evidence for any of these T cells in regulation of BDC2.5 T cells (Fig. 4A and data not shown). Moreover, we found that NOD.B2m–/– splenocytes did not induce graft-vs-host disease in NOD.scid mice nor were they cleared by NOD.scid NK cells, which are notoriously poor in NOD.scid mice (39). Therefore, we conclude that the main regulatory activity resides in the CD4+, 2M-dependent T cell subset, most likely NKT cells (see Fig. 8).
|
To determine the cellular mechanism by which immunoregulation of BDC2.5 T cells occurred, it is necessary to determine when and where the NKT cells reside during the regulatory phase. Therefore, we asked whether the regulating cells are required before, at, or after the time of BDC2.5 T cell activation. We performed cotransfer experiments in which NKT cell-containing NOD splenocytes were transferred from 3 days before to 3 days after the cotransfer of BDC2.5 T cells. As seen in Fig. 6, the introduction of NKT cell-containing NOD splenocytes to recipient mice between 3 days before to the time of BDC2.5 T cell transfer resulted in little or no diabetes. In contrast, if BDC2.5 T cells were given a 1- to 3-day "head start" they escaped control and diabetes swiftly followed.
|
1%, cannot be seen as merely coincidental. The natural 40- to 50-fold enrichment of NKT cells in the target organ and primary efferent lymph nodes of NOD mice undergoing acute immunoregulation strongly suggests that the NKT cells actively regulate diabetogenic BDC2.5 T cells in vivo.
|
|
production during regulation is necessary to control diabetogenic T cellsSeveral studies suggest that NKT cells constrain the action of self Ag-reactive T cells and curb autoimmunity by host APC conditioning—most notably by altering DC maturation, via unknown soluble mediator(s) (26, 27). In previous studies, blocking Abs were used (27). These studies have the disadvantage that many cytokines are pleiotropic and have different, often contrasting effects, on different target cell types.
By making chimeric mice with targeted deficits in cytokine action, we tested the requirement for IL-4, IL-10, and IFN- in the regulatory process. We cotransferred BDC2.5 T cells into NOD.scid mice along with splenocytes from IFN--, IL-4-, IL-10-, and IL-4/10-deficient NOD mice. The deficiency in IL-4, IL-10 singly or IL-4 and IL-10 together had no adverse effect on the regulation of BDC2.5 T cells (Fig. 9A). Therefore, these cytokines, whether produced by NKT cells or any other cell types within the transferred splenocytes population, were not necessary for the observed immunoregulation. Of interest, however, was the absolute requirement for IFN-. When NKT cell-containing splenic T cells from NOD.Ifng–/– mice were used, the regulation of BDC2.5 T cells was totally lost (Fig. 9A). The molecular action of IFN- could be at several levels. Its loss could result in poor or absent NKT cell function (autocrine regulation via IFN-), could impact the direct regulation of BDC2.5 T cells, or alter the host environment, for example, the conditioning of host APC (paracrine regulation via IFN-). Therefore, to determine the molecular and cellular consequence of IFN- deficiency, we cotransferred splenocytes from IFN-R--chain (Ifngrb)-deficient NOD mice, along with wild-type BDC2.5 T cells into NOD.scid recipients. The regulation of BDC2.5 was unaffected, as seen in Fig. 9B. Thus, splenocytes from NOD.Ifngrb–/– mice retain full regulatory competence. Thus, we conclude that the NKT cells do not themselves require IFN- for development, but that IFN- acts on either the host or the diabetogenic BDC2.5 T cells themselves, as part of the regulatory control mechanism.
|
acts on the host but not on the diabetogenic T cells
Having concluded that the regulating population need not respond to but must make IFN-, we next introduced the IFN-R (Ifngrb) null mutation on the diabetogenic BDC2.5 T cells, to test whether their regulation requires direct IFN- responsiveness. When we transferred sort-purified, Ifngrb–/– BDC2.5 T cells alone to wild-type (Ifngrb+/+) NOD.scid recipients, the Ifngrb–/– BDC2.5 T cells produced rapid transfer of diabetes, thereby establishing that the diabetogenic potential of BDC2.5 T cells that cannot respond to IFN- (Fig. 10A). Moreover, observations by one of us (B. Wang, unpublished results) have shown that both Ifngrb–/– and Ifngrb+/+ T cells can transfer diabetes to NOD.scid recipient mice lacking IFN-R (NOD.scid.Ifngrb–/–), suggesting that IFN- is not needed for diabetes to develop in NOD mice. Yet, when wild-type (IFN--producing) NKT cell-containing splenocytes were cotransferred along with the BDC2.5/NOD.Ifngrb–/– T cells into wild-type NOD.scid recipients, regulatory control of diabetes was still observed (Fig. 10A). Therefore, having shown that IFN- responsiveness is not needed by the regulatory cells, this demonstration that BDC2.5 T cells that fail to respond to IFN- can still be regulated suggests that the effect of IFN- on BDC2.5 is not direct, but rather via modification of the IFN--responsive host. To demonstrate that only regulatory T cell-derived IFN- is necessary to control diabetes, purified BDC2.5 T cell from NOD.Ifng–/– mice were cotransferred along with wild-type NKT cell-containing splenocytes into either IFN--deficient or wild-type recipient mice. Although the overall diabetes transfer was not as robust, the regulation of BDC2.5/NOD.Ifng–/– T cells by IFN-+/+ cells was identical, regardless of whether the recipient made IFN- (Fig. 10B). These results, therefore, suggest that the sole sufficient source of IFN- necessary at the time of regulatory control is from the NKT cell-containing splenocytes and that the host, but not the diabetogenic T cells, must respond to IFN-.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
cell Ag have a rapid course of disease. In contrast, NOD mice that contain a more diverse TCR repertoire that includes both enhanced numbers of islet-reactive T cells as well as other T cells, have a more modest incidence of diabetes. We hypothesized that the striking difference in diabetes incidence observed is the result of T cell-mediated immunoregulation of the deleterious diabetogenic T cells in vivo. Using an adoptive transfer model of T1DM in which a controlled number of diabetogenic BDC2.5 T cells were transferred into NOD.scid recipient mice, we have identified a subset of T cells capable of regulating the diabetogenic potential of BDC2.5 T cells. This T cell subset is a CD4+, CD1d-restricted, and 2M-dependent NKT cell subset, as shown by several lines of evidence: 1) inclusion in the CD4 but not CD8-depleted subset of T cells; 2) absence in NOD.B2m-deficient mice; and 3) reaction with CD1-tetramers in the pancreas and PLN of mice undergoing regulation. In addition, we find that these cells must be present at or before the activation of the BDC2.5 T cells in vivo, that these NKT cells can only exert regulatory function if they derive from IFN--producing mice, and if the host is capable of responding to IFN-. Moreover, the BDC2.5 T cells need not make nor respond to IFN- to produce diabetes, if left unchecked, or to be regulated.
The apparent ability of NKT cells to regulate diabetogenic T cells shown here, stands in substantial agreement with reports showing that manipulation of NKT cells can profoundly influence the rate and severity of diabetes in NOD mice (13, 20, 21, 23). Lehuen and colleagues (25) have shown that NKT cells inhibit the differentiation of effector T cells in NOD mice and that, as NKT cells numbers rise in NOD mice, either in response to -GalCer or transgene-mediated enrichment, the corresponding incidence of diabetes falls (40). How the NKT cells impart immunoregulation to the BDC2.5 T cell in vivo remains incompletely understood; for example, do they act alone or in concert with other cells? Do both NKT cells and other splenocytes need to make IFN-, or can NKT cells simply "go it alone"? Several findings, presented here and elsewhere (26), suggest that the action of the NKT cells upon the diabetogenic T cells is indirect via the alteration or "conditioning" of the host. We find, for example, that cotransfer of NKT-containing subsets with BDC2.5 T cells results in poor accumulation of diabetogenic T cells in and around the islets of recipient mice, suggesting that the diabetogenic T cells fail to become sufficiently activated to home to or settle within the target organ. BDC2.5 T cells can be found in other lymphoid compartments, suggesting that they are not being directly destroyed by NKT cells.
Whether purified NKT cells can act alone or whether a more complex mixture of cells is needed remains to be determined. Recently, Bezbradica et al. (41) have suggested that a dynamic interaction among DC, B cells, and NKT cells can have physiologically distinct consequences on NKT cell activation and disease outcome. It would, thus, not be surprising if highly purified, undifferentiated NKT cells failed to act alone, and are greatly influenced by the microenvironment they occupy. In addition, others have shown that NKT cell-derived soluble factors may inhibit the complete maturation of DC subsets into professional APC, competent in both delivering Ag to the local draining lymph nodes in mice and in inducing full Ag presentation and T cell activation (26). Moreover, our data do not formerly exclude a role for DX5/CD49b+ T cells as potential regulatory cells, but our data, particularly that of the CD1d-tetramer staining, strongly suggest that the NKT cells and not DX5+ T cells mediate the control of the BDC2.5 T cells in vivo. Moreover, DX5 is a poor marker for NKT cells (42). We now believe that one such factor is IFN-, or that IFN- regulates the production or action of these factors. In addition, the ability of BDC2.5 T cells from NOD.Ifng–/– mice to transfer diabetes into both NOD.Ifng–/– and NOD.Ifng+/+ recipients suggests that IFN- does not act solely to up-regulate MHC class II or costimulatory molecules on the host APC for, if this were the case, one would expect IFN--deficient mice to fail to develop diabetes. This is clearly not the case here, which jibes with earlier reports that the effector phase of diabetes is IFN- independent (32, 43, 44). Our findings suggest that NKT cells cannot regulate the BDC2.5 T cells in an IFN-R-deficient host, because the action of NKT cells is via IFN--dependent modulation of host APC. We have found, as was published elsewhere, that BDC2.5 T cells do not transfer diabetes into NOD.Ifngrb–/– mice and that IFN- impacts diabetes in multiple ways (45); this has made a regulation experiment impractical using this model system, because regulation is not meaningful in the absence of diabetes transfer. However, it is important to note that other T cell transfer models that do not rely on BDC2.5 T cells do transfer diabetes into NOD.Ifngrb–/– mice and that regulation in these models is disrupted in the absence of IFN- (B.Wang, unpublished results).
We found no significant role for Treg in the peripheral control of BDC2.5 T cells, although Treg control had been seen in the past (46); that multiple levels of regulation should and do exist is not surprising given the complexity of the mammalian immune system. However, we cannot rule out some role for CD4+CD25+ Treg in our system. Given the monoclonal nature of the diabetogenic effector cells, BDC2.5 T cells, and the use of a transfer model, NKT cells may merely represent a more potent modulator of BDC2.5 T cell function that classical Treg. Moreover, this may represent a "numbers game" where the frequency of Ag-specific Treg are too low to affect regulation while NKT cells are available in numbers sufficient to produce regulatory control. In addition, we cannot formally exclude the possibility that CD4+CD25– cells convert to CD4+CD25+ Treg cells upon the induction of Foxp3, in vivo, upon the localized encounter of TGF-, which has been shown to facilitate Treg conversion (47).
How, when, or if, both NKT and Treg cells interact to produce immunoregulation and self tolerance also remain open questions. NKT and Treg cells may act at different ends of the same regulatory cascade, as some have proposed (14), or they may provide distinct, and at times redundant, checks on the immune response. Because our data suggest that NKT cells need to reside in the target organ and its pertinent draining lymph nodes at or before the activation (or reactivation) of autoreactive T cells, NKT cells seem to be a more likely candidate to be an initial check on the development of autoimmunity via the dampening of APC function by tissue-resident DC populations. Treg cells in contrast have clearly been shown to actively suppress even memory T cells from proliferating in response to Ag in vitro and in vivo (1, 7, 8). In this respect, the Treg may be seen more as an active brake on the system, whereas NKT cells may act more passively and indirectly, via host APC modulation, to inhibit the nucleation of an autoimmune response. Specific subpopulations of DC have distinct expression patterns of IFN-Rs, that can modulate during DC maturation (11, 48). Therefore, the localized production of IFN- by NKT cells may be expected to have fundamental consequences on specific DC subsets as they acquire effector functions.
In conclusion, we have shown that diabetogenic CD4+ T cells can be actively controlled by a CD4+, 2M-dependent T cell subset, most likely with NKT cells, which were shown to accumulate in high numbers at the site of regulation. The regulatory action of the NKT cells appears indirect, mediated via IFN- production. The diabetogenic T cells need not make or respond to IFN- for regulation to occur; however, the regulating NKT cell-containing population must make IFN-. Our data are most consistent with the model of NKT cells altering the activation of diabetogenic T cells by modifying the maturation and function of host (IFN--responsive) DC rendering them incapable of mediating the full activation of effector CD4+ T cells in the NOD mouse, thereby rendering the diabetogenic T cells functionally inert.
|
Acknowledgments |
|---|
and IFN-R PCR. We thank Drs. Archana Khurana and Mitchell Kronenberg from the La Jolla Institute for Allergy and Immunology for the generous gift of the CD1-tetramer and controls. We thank Drs. Christopher Karp, Fred Finkelman, and David Hildeman for critical review of the manuscript. |
Disclosures |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
|
Footnotes |
|---|
1 This work was supported by National Institutes of Health Grants R01 AI44416 and R01 DK62274 (to J.D.K.), an American Diabetes Association Career Development Award (to B.W.), and the Cincinnati Children’s Research Foundation Diabetes Campaign. J.A.C. is the recipient of a National Institute of Diabetes and Digestive and Kidney Diseases Mentored Research Scientist Development Award (K01 DK64836). 
2 Address correspondence and reprint requests to Dr. Jonathan D. Katz, Department of Pediatrics, University of Cincinnati College of Medicine, 3333 Burnet Avenue, Cincinnati, OH 45229. E-mail address: jonathan.katz{at}cchmc.org
3 Abbreviations used in this paper: NOD, nonobese diabetic; T1DM, type 1 diabetes mellitus; Treg, CD4+CD25+ T regulatory cell; DC, dendritic cell; 2M, 2-microglobulin; -GalCer, -galactosylceramide; PLN, pancreatic lymph node; MLN, mesenteric lymph node; CD1-Tet, CD1d-tetramer.
Received for publication September 22, 2005. Accepted for publication November 21, 2005.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
2-Microglobulin-deficient NOD mice do not develop insulitis or diabetes. Diabetes 43: 500-504. [Abstract]-galactosylceramide specifically stimulates V14+ NK T lymphocytes. J. Immunol. 161: 3271-3281. 14i natural killer T cells are resistant to cytokine polarization in vivo. Proc. Natl. Acad. Sci. USA 100: 8395-8400. -galactosylceramide demonstrates its immunopotentiating effect by inducing interleukin (IL)-12 production by dendritic cells and IL-12 receptor expression on NKT cells. J. Exp. Med. 189: 1121-1128. -galactosylceramide treatment prevents the onset and recurrence of autoimmune type 1 diabetes. Nat. Med. 7: 1057-1062. [Medline]14-J281 transgenic nonobese diabetic mice is associated with a Th2 shift circumscribed regionally to the islets and functionally to islet autoantigen. J. Immunol. 166: 3749-3756. cells. Immunity 17: 725-736. [Medline] Cell apoptosis in T cell-mediated autoimmune diabetes. Proc. Natl. Acad. Sci. USA 94: 213-218. receptor signaling is dispensable in the development of autoimmune type 1 diabetes in NOD mice. Diabetes 49: 2007-2011. [Abstract]
, n = 10) or along with NOD (•, n = 4) or 
, n = 16) splenocytes. Cumulative diabetes incidence (as percentage) from three independent experiments.
), compared with BDC2.5/NOD.Tcra+/– (
, n = 12); with unfractionated NOD splenic T cells inhibited diabetes (•, n = 6). The transfer of BDC2.5 T cells alone (
, n = 6); BDC2.5 plus CD4+CD25– (R3-gated) T cells (
, n = 10). Cumulative incidence of diabetes from three independent experiments.
50% infiltration), 
; peri-insulitis, 
, Diabetes incidence in mice receiving BDC2.5 T cells on day 0 and NKT cells on the day indicated is represented. 
, A significant decrease in cell numbers (p = 0.0004, day 3; p = 0.0013, day 10; p = 0.0013, day 11; p = 0.0001, day 21 by Student’s t test), compared with results in A. D, Numbers of BDC2.5 T cells after cotransfer with NOD.GFP splenocytes: days 3, 5, 7, and 10 (n = 4); day 14 (n = 5); day 21 (n = 6). 
