Immunity often depends on proper cell fate choice by helper T lymphocytes. A naive cell, with minimal expression of IFN-γ and IL-4, must give rise to progeny expressing high levels of either one, but not both, of those cytokines to defend against protozoan and helminthic pathogens, respectively. In the present study, we show that inactivation of the Mbd2 gene, which links DNA methylation and repressed chromatin, results in enhanced resistance to the protozoan parasite Leishmania major but impaired immunity to the intestinal helminth Trichuris muris. Helper T cells from methyl CpG-binding domain protein-2-deficient mice exhibit exuberant patterns of cytokine expression despite appropriate silencing of genes encoding the lineage-specifying factors T-bet and GATA-3. These results suggest that gene silencing can facilitate the ability of a progenitor cell to give rise to appropriately differentiated daughter cells in vivo. These findings also point to novel pathways that could participate in genetic control of resistance to infection and autoimmunity.
Lineage decisions occur frequently in development of multicelled organisms. The immune response to microbes also depends upon proper fate choice by Ag-specific helper T cells (1, 2). Successful clearance of protozoa, such as Leishmania major, requires Th1 cells that secrete large amounts of IFN-γ (3). Resistance to L. major is antagonized by Type 2 cytokines, including IL-4 and IL-10. Expulsion of intestinal helminths, such as Trichuris muris, requires Th2 cells that secrete IL-4 and IL-13 (4, 5). Resistance to T. muris is antagonized by IFN-γ produced by Th1 cells (6).
Proper maturation of Th1 and Th2 cells is regulated by the positive and negative actions of a variety of extrinsic factors, including Ags, cytokines, and other receptor-mediated signaling events (1, 2). The discovery of T-bet (7, 8, 9) and GATA-3 (10, 11, 12, 13) as apparent master controllers of the Th1 and Th2 fate has helped to establish the paradigm that distinct patterns of lineage-specific gene expression are sculpted by transcription factor regulatory networks (1, 2). However, mounting evidence suggests that covalent modification of histones and DNA, as well as the structural characteristics of chromosomes, may interact with conventional gene regulatory networks (14, 15, 16). Despite the attractiveness of chromatin-constrained models of gene expression and cell differentiation, there has been limited evidence that bona fide mammalian cell fate decisions are indeed controlled by epigenetic effects and gene silencing pathways.
Methyl CpG-binding domain protein-2 (MBD2)4 is a protein that links CpG methylation to repressive chromatin structure by recruiting complexes containing histone deacetylases and ATP-dependent nucleosome remodeling activities to methylated DNA (17). We previously offered in vitro evidence to suggest that MBD2-dependent gene silencing is necessary to facilitate temporally and spatially restricted patterns of gene expression (18). In the present study, we report the outcome of challenges of MBD2-deficient mice with two well-characterized parasitic pathogens that typically elicit polarized helper T cell responses. In the absence of efficient gene silencing, the cytokine gene expression of responding T cells is excessive, leading to profound changes in the susceptibility and resistance of Mbd2 mutant mice to infectious diseases. These results support an epigenetic model of helper T cell differentiation and suggest that a matrix of gene silencing, not simply the choice of the correct trans-activator, might be an essential component of lineage specification.
Materials and Methods
Mbd2 mutant mice (B6.129-Mbd2tm1.Bird) were generated as described previously (17). Experiments were performed with littermate cohorts from either C57BL/6 (B6) × 129sv (129), N6 B6, or N10 BALB/c heterozygote matings. All animal work was in accord with guidelines of the University of Pennsylvania.
For protozoan challenge, mice were inoculated in the left hind footpad with either 2 × 106 or 2 × 107 stationary phase promastigote parasites of L. major (WHOM/IL/80/Friedlin). Footpad measurements, quantitative parasite cultures, Ag-specific restimulation, and ELISA were performed as described previously (19). For helminthic challenges, mice were infected orally with 150 T. muris eggs, as described previously (4, 5, 6). Parasite-specific Ab isotypes were measured as described previously (4, 5, 6). Cecum and adjacent colon were harvested to determine worm burdens at day 31 postinfection. One-centimeter segments of mid-cecum were removed and fixed for 24 h in 10% neutral buffered formalin. Paraffin-embedded tissue (5-μm sections) were stained with H&E or Alcian blue-Periodic Acid Schiff (mucin stain) for detection of intestinal goblet cells.
In vitro differentiation of helper T cells
Splenocytes from naive mice were depleted of CD8+ cells and stimulated as described previously (18, 20). Briefly, 2 × 106 cells/ml were stimulated using soluble anti-CD3 mAb, anti-CD28 mAb, and rIL-2. Experimental Th1 conditions included mitogenic stimulation plus rIL-12 and anti-IL-4 mAb. Experimental Th2 conditions included mitogenic stimulation plus rIL-4 and anti-IL-12 mAb. CFSE labeling was performed as described previously (18, 21). Retroviral constructs encoding T-bet and dominant-negative (DN) T-bet were constructed and transduced as described previously (18, 20). Analysis of intracellular cytokines was performed as described previously (18, 21). Flow cytometric analyses in this article include only CD4+ events. Gates indicate specific staining of IFN-γ- or IL-4-positive cells as determined by fluorochrome-conjugated isotype control Abs. RT-PCR was performed using primers and probes described previously (18, 20). Northern blotting was performed as described previously (22).
Results and Discussion
MBD2 restricts IFN-γ induction during helper T cell differentiation in vivo
We previously showed that mice deficient in MBD2, a molecule proposed to link DNA methylation and repressed chromatin structure, have disordered helper T cell differentiation in vitro (18). MBD2-deficient helper T cells exhibit a higher frequency of IL-4- and IFN-γ-expressing cells under all polarizing conditions (18). Because we found ample evidence for ectopic expression of IL-4 in MBD2-deficient Th1 cells, we assessed the ability of MBD2-deficient mice to resist challenge with L. major, an intracellular protozoan parasite of macrophages. Successful eradication of Leishmania is contingent on development of IFN-γ-expressing helper T cells (3). In contrast, IL-4 plays a deleterious role by blocking the ability of IFN-γ to activate macrophages.
We challenged genetically resistant (B6 × 129) littermate Mbd2+/+, Mbd2+/−, and Mbd2−/− mice with hind footpad inoculation of a curable dose of 2 × 106 stationary phase promastigote parasites of L. major. Although genetically susceptible BALB/c mice are unable to control parasite replication using this inoculum, all genetically resistant mice, regardless of Mbd2 status, were able to control the infection and resolve their footpad lesion (data not shown). When we challenged mice with a 10-fold higher dose of virulent parasites (2 × 107 promastigotes/footpad), we found that both BALB/c and wild-type B6 × 129 (Mbd2+/+) mice are unable to control parasite replication due to the overwhelming inoculum of parasites (Fig. 1 A).
Unexpectedly, we found that both Mbd2+/− and Mbd2−/− mice are robustly resistant to the normally incurable challenge (Fig. 1,A). Identical results were obtained using littermate Mbd2+/+, Mbd2+/−, and Mbd2−/− mice that were generated from the intercross of sixth generation B6 Mbd2+/− mice (data not shown). MBD2-deficient mice (heterozygote and homozygote) controlled the local replication of the parasite in their footpad lesions, as assessed by quantitative culture (Fig. 1,B), and exhibited less dissemination of the parasite to visceral organs (data not shown). Consistent with their increased resistance to infection, cells recovered from the draining lymph nodes of Mbd2+/− and Mbd2−/− mice secreted more IFN-γ than cells from Mbd2+/+ mice when restimulated in vitro with soluble Leishmania Ags (Fig. 1 C). Hyperresistance of B6 × 129 and B6 Mbd2−/− mice was completely abrogated by treatment with neutralizing anti-IFN-γ Abs, resulting in an explosive footpad lesion of greater severity than infected wild-type mice (data not shown). Thus, hemizygous or homozygous loss of Mbd2 results in an exuberant Th1 response and enhances host resistance against an intracellular protozoan parasite. MBD2 has not been reported to be expressed in a monoallelic fashion, suggesting that minimally dysfunctional alleles of Mbd2 might exert significant immunological phenotypes. However, the Mbd2 gene is 44 cM from the centromere of mouse chromosome 18, which is some distance from a putative L. major resistance locus that is situated between 16 and 24 cM from the centromere (23) and distinct from the chromosomal locations of the other known resistance loci (24, 25, 26, 27).
Mutation of Mbd2 partially counteracts genetic susceptibility to L. major
We also challenged genetically susceptible BALB/c Mbd2+/+, Mbd2+/−, and Mbd2−/− mice with the lower dose of 2 × 106 stationary phase promastigote parasites of L. major. The inability of the BALB/c strain to control L. major, as well as its predisposition to enhanced Th2 differentiation in vitro, is a complex genetic trait, involving multiple, independently segregating loci (3, 23, 24, 25, 26, 27). We found that loss of both Mbd2 alleles on the BALB/c background confers substantial protection against L. major with regard to footpad lesion size (Fig. 2,A) and control of parasite dissemination to the spleen (Fig. 2 B). In contrast to the situation in genetically resistant backgrounds, we found that BALB/c Mbd2+/− mice are substantially more susceptible than BALB/c Mbd2−/− mice.
Cells recovered from the draining lymph nodes of infected mice were restimulated in vitro with soluble Leishmania Ags and analyzed at the single-cell level for expression of the signature cytokines of Th1 and Th2 cells, IFN-γ and IL-4, respectively (Fig. 2 C). In genetically resistant B6 mice, we detected primarily single-positive, IFN-γ-expressing, Ag-specific Th1-like cells. In susceptible BALB/c Mbd2+/+ mice, we detected significantly more Ag-specific, single-positive, IL-4-expressing, Th2-like cells. In BALB/c Mbd2−/− mice, we found an increase in IFN-γ-expressing, Ag-specific, Th1-like cells. Therefore, mutation of Mbd2 can partially ameliorate the genetic susceptibility of BALB/c mice to L. major.
MBD2 is required for immunity to T. muris
A well-characterized parasitic infestation that elicits a polarized Th2 response in many strains of mice is infection with the gut-dwelling helminth T. muris. Immunity to T. muris is dependent on Th2 cells that secrete IL-4 and IL-13 (4, 5), while resistance to infection is antagonized by IFN-γ (4, 6). Wild-type BALB/c mice and B6 × 129 Mbd2+/+, Mbd2+/−, and Mbd2−/− littermate mice were orally infected with embryonated Trichuris eggs to assess their ability to expel adult worms. Mbd2−/− mice had an increase in the number of parasitic worms recovered from their ceca compared with Mbd2+/+ and Mbd2+/− mice at day 15 (data not shown) and day 31 (Fig. 3 A) postinfection. At later time points, all genotypes with the exception of Mbd2−/− mice cleared the parasites. Thus, deficiency in Mbd2 results in chronic parasitic infestation with the helminth T. muris.
The enhanced susceptibility of Mbd2−/− mice is not likely due to an inability to mount a Th2 response against T. muris. Indeed, the serum levels of parasite-specific, Th2-associated IgG1 Ab were equivalent or higher in Mbd2 mutant mice compared with Mbd2+/+ control mice (Fig. 3,B). However, Mbd2+/− and Mbd2−/− mice exhibited higher levels of parasite-specific, IFN-γ-associated IgG2a Ab (Fig. 3,B). Mbd2−/− mice also developed intestinal pathology consistent with excessive IFN-γ production (6). The protective response of Mbd2+/+ mice was accompanied by Th2 cytokine-dependent goblet cell responses and mucin secretion (Fig. 3,C). By contrast, Mbd2−/− mice exhibited severe colonic inflammation, lymphatic dilation, edema, mucosal thickening, and numerous intraluminal parasites (Fig. 3,C; 5×). Extensive transmural inflammatory infiltrate in the intestinal mucosa of infected Mbd2−/− mice was associated with marked crypt hyperplasia and loss of goblet cells and mucin responses (Fig. 3 C; 20×). The pathology of infected Mbd2−/− mice is similar to Th1 cytokine-associated intestinal inflammatory responses reported in models of inflammatory bowel disease (6). Despite the systemic evidence for a mixed Th1/Th2 response in MBD2-deficient mice, the net effect on target intestinal tissue was an aborted Th2 response, with Th1-associated pathology and impaired immunity.
MBD2 limits IFN-γ expression but is not required to repress T-bet expression
Despite the disorganized phenotype of cytokine expression in the absence of MBD2 in vitro (18) and in vivo, we found that mutant cells were capable of restricting expression of the forbidden trans-activators. As assessed by Northern blotting, both Mbd2+/+ and Mbd2−/− cells cultured in vitro can induce T-bet and repress GATA-3 in Th1 conditions (Fig. 4,A). Similarly, both Mbd2+/+ and Mbd2−/− cells are able to induce GATA-3 and repress T-bet in Th2 conditions. The possible induction of substantial amounts of ectopic T-bet in developing Th2 cells from Mbd2 mutant mice was further excluded using real-time PCR (Fig. 4 B).
Taken together, these results suggest that developing Mbd2−/− Th2 cells can misexpress “T-bet-less” IFN-γ. Such findings parallel our previous suggestion that developing Mbd2−/− Th1 cells misexpress “GATA-less” IL-4 (18). The competitive relationship between T-bet and MBD2 could be further visualized by genetic manipulation of each factor. Provision of exogenous T-bet, using retroviral transduction, induces higher levels of IFN-γ in the absence than in the presence of Mbd2 (Fig. 4,C). Additionally, antagonism of T-bet with a DN factor prevents IFN-γ induction with lesser efficiency in the absence of Mbd2 (Fig. 4 D). Moreover, preliminary analysis of compound mutant mice also indicates that loss of Mbd2 can partially rescue the Th1 defect caused by loss of T-bet (A. S. Hutchins and S. L. Reiner, unpublished results). Thus, T-bet and MBD2 are in a genetic competition to activate and repress IFN-γ, respectively. This competition may also extend to the STAT4 pathway (28) because the absence of MBD2 also reduces the requirement for IL-12 to induce IFN-γ expression. We have not yet been able to detect MBD2 binding to the Ifng gene body or promoter, although it was detected at the Il4 locus (18). This could indicate that MBD2 binding to the Ifng gene is ephemeral, that it occurs at a more distant cis-acting sequence, or that derepression of a trans-acting factor other than T-bet might account for the apparent “T-bet-less” IFN-γ expression. Whether the Ifng gene is a direct or indirect target, one role of MBD2 seems to be to limit IFN-γ expression in vivo.
We now show that the loss of MBD2, a component of DNA methylation-mediated gene silencing, perturbs the normal restraints on effector cytokine expression in maturing helper T cells in vivo. Using small molecule inhibitors of maintenance methylation or histone deacetylases (21, 29), as well as gene inactivation of Mbd2 (18) or the maintenance methyltransferase Dnmt1 (30, 31), it was suggested previously that helper T cell differentiation was constrained by DNA methylation-mediated silencing. Prior studies emphasized the misexpression of “GATA-less” IL-4 in Th1 cells and CD8+ T cells in vitro (18, 30). Unexpectedly, we found that the dominant derangement of MBD2-deficient mice is an excessive IFN-γ-mediated response against both protozoan and helminthic invaders. This is unlikely the result of defective counterregulation because we found that MBD2-deficient cells also overexpress the regulatory cytokine IL-10 and that Mbd2 mutant mice have normal numbers of FoxP3+CD25+CD4+ regulatory T cells (A. S. Hutchins and S. L. Reiner, unpublished results). Even highly enriched naive T cells from MBD2-deficient mice cultured in vitro in the absence of regulatory T cells and APCs exhibit excessive effector cytokine expression compared with MBD2-proficient cells (A. S. Hutchins and S. L. Reiner, unpublished results), suggesting a T cell-autonomous component to the phenotypes observed in vivo.
These findings provide formal in vivo evidence that gene silencing organizes signaling and transcription factor networks important for the maturation of cytokine expression in helper T cells. The unexpectedly significant protection against L. major afforded by inactivation of only one allele of Mbd2 also raises a novel possibility regarding the genetics of disease resistance to infectious and autoimmune processes. Future determination of candidate loci might rightly include gene silencing pathways, in addition to traditional regulators of subset-specific cytokines.
We are grateful to J. Northrup and S. Longworth for excellent assistance.
The authors have no financial conflict of interest.
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 is supported by National Institutes of Health Grant AI053827 and the Abramson Family (to S.L.R.).
Abbreviation used in this paper: MBD2, methyl CpG-binding domain protein-2; DN, dominant negative.