Dual function of polycomb group proteins in differentiated murine T helper (CD4+) cells

Background Following antigen recognition, naive T helper (Th; CD4+) cells can differentiate toward one of several effector lineages such as Th1 and Th2; each expressing distinctive transcriptional profiles of cytokine genes. These cytokines eventually instruct the strategy of the immune response. In our search for factors that propagate the transcriptional programs of differentiated Th cells, we previously found that Polycomb group (PcG) proteins, which are known as epigenetic regulators that maintain repressive chromatin states, bind differentially the signature cytokine genes. Unexpectedly, their binding to the Ifng (Interferon-g) in Th1 cells and Il4 (Interleukin-4) in Th2 cells, was correlated with transcriptional activation. Therefore, in this study we aimed to determine the functional role of PcG proteins in the regulation of the expression of the signature cytokine genes. Methods PcG proteins were knocked down in primary and established murine Th cells using transduction of lentiviruses encoding short hairpin RNAs (shRNAs) directed to Mel-18, Ezh2, Eed and Ring1A, representative of two different PcG complexes. The chromatin structure and the binding activity of PcG proteins and transcription factors at the Ifng promoter were assessed by chromatin immunoprecipitation (ChIP) assays. Results Downregulation of PcG proteins was consistent with their function as positive regulators of the signature cytokine genes in primary and established Th1 and Th2 cells. Moreover, the PcG protein Mel-18 was necessary to recruit the Th1-lineage specifying transcription factor T-bet, and the T cell receptor (TCR)-inducible transcription factor NFAT1 to the Ifng promoter in Th1 cells. Nevertheless, our results suggest that PcG proteins can function also as conventional transcriptional repressors in Th cells of their known target the Hoxa7 gene. Conclusions Our data support a model whereby the non-differentially expressed PcG proteins are recruited in a Th-lineage specific manner to their target genes to enforce the maintenance of specific transcriptional programs as transcriptional repressors or activators. Although our results suggest a direct effect of PcG proteins in the regulation of cytokine gene expression, indirect functions cannot be excluded.


Background
When naive Th cells encounter an antigen for the first time, they can differentiate into the effector lineages Th1, Th2 and Th17 that differentially express cytokine genes [1][2][3]. The Th1 and Th2 lineages are characterized by the expression of the signature cytokines IFNγ and IL-4, respectively. IFNγ exerts protective functions in microbial infections and is observed clinically in cases of autoimmune diseases. IL-4 is strongly apparent in parasitic infections, and is associated with allergic reactions. The polarization of Th cells is most efficiently promoted by the cytokine milieu; IL-12 strongly potentiates the differentiation toward the Th1 lineage and IL-4 toward Th2. Although the activities of the polarizing cytokines ultimately lead to distinct Th phenotypes, the differentiation processes have similar features such as the expression of lineage-specifying transcription factors. The lineage-specifying transcription factors, T-bet in Th1 and GATA3 in Th2 cells, function as master regulators that establish the appropriate gene expression profiles for one lineage and oppose the alternative fates. Differential pattern of cytokine gene expression is also associated with selective recruitment of TCR-inducible transcription factors such as NFAT [4]. All of these activities are accompanied by major epigenetic changes [1][2][3] that are probably involved in the heritability of the transcriptional programs of differentiated Th cells. However, epigenetic regulation of Th cells preserves a certain level of plasticity that may allow adaptation to new immunological challenges [3,[5][6][7][8][9][10][11][12].
Which factors are necessary to maintain the cellular memory of differentiated Th lineages? Several studies showed that the maintenance of the Ifng and Il4 transcriptional patterns in Th cells is not mediated exclusively by the lineage-specifying transcription factors [13][14][15][16][17]. Therefore, the lineage-specifying transcription factors may induce the expression of downstream specific factors or alternatively, facilitate restricted binding of a generally expressed epigenetic machinery. We previously showed that PcG proteins, whose role in maintaining gene silencing during embryogenesis is well known [18][19][20][21][22][23][24], bind the cytokine genes [25]. However, their binding activity was associated unconventionally with gene expression not silencing; in each lineage they were associated with the signature cytokine gene Ifng in Th1 cells and Il4 in Th2 cells [25].
The PcG and trithorax group proteins were first identified in Drosophila, as transcriptional regulators of the homeotic (Hox) genes during development. In contrast to the PcG proteins, the trithorax group proteins were characterized by their ability to maintain active transcription. The PcG proteins form two major complexes, PcG repressive complex 1 (PRC1), which contains the core proteins M33, Bmi-1, Mel-18, Ring1A, and Ring1B, and PRC2, with the core proteins Suz12, Ezh2, and Eed. However, biochemical purification demonstrated a significant variety in the content of these complexes [21,24,26]. The mechanisms by which PcG proteins mediate the epigenetic inheritance of transcriptional silencing are not fully understood [27], but they are known to be associated with histone modifications: Ring1B is histone H2A ubiquitin E3 ligase, and Ezh2 is histone methyltransferase that preferentially trimethylates histone H3 on lysine 27 (H3K27me3). The function of the PcG proteins also entails non-catalytic activities [26,[28][29][30], such as chromatin compensation, long-range intrachromosomal interactions [31][32][33][34][35] and repression of transcriptional elongation [36]. Genome-wide binding profiles in Drosophila, murine and human embryonic stem cells have demonstrated that the PcG proteins have additional targets to the Hox genes, most of them are transcriptional regulators of development [37][38][39][40][41][42]. These genes are predominantly inactive in embryonic stem cells, and many are marked bivalently by both the PcG repressive mark H3K27me3 and the trithorax group permissive mark H3K4me3 [21,28,43]. Under this status, premature differentiation is prevented, and the pluripotency of the embryonic stem cells is maintained. Developmental signals eventually induce or stably silence the expression of these genes. The PcG proteins are also crucial during hematopoiesis [24,44,45], and their dysregulation has been linked with various human cancers, apparently due to the altered transcription of PcG-target genes that control cell proliferation and differentiation [24,46].
Considering the known function of the PcG proteins as transcriptional repressors, our previous results demonstrating the binding pattern of PcG proteins at the cytokine promoters in Th cells [25], raised the feasible scenario in which PcG proteins restrain overexpression of the active/accessible cytokine genes. However, our knockdown experiments are more compatible with the idea that the PcG proteins Mel-18, Ezh2, Eed and Ring1A can function as transcriptional activators of Ifng in Th1 and Il4 in Th2 cells, rather than transcriptional repressors. Moreover, in Th1 cells, Mel-18 was required for the recruitment of the crucial transcription factors T-bet and NFAT1 to the Ifng promoter. Nevertheless, our results also demonstrated that PcG proteins function as repressors of Hoxa7 in both Th1 and Th2 lineages, and of Tbx21 (which encodes T-bet) in Th2 cells. All together, our data propose a bi-functional role for the PcG proteins in differentiated Th cells.

Ethic statement
The studies have been reviewed and approved by the Inspection Committee on the Constitution of the Animal Experimentation at the Technion.

Mice
Female BALB/c mice were purchased from Harlan Biotech, Israel, and maintained under pathogen-free conditions in the animal facility of the Faculty of Medicine, Technion-Israel Institute of Technology.

RNA interference
The knockdown was accomplished with lentiviral shRNA (MISSION, Sigma). The lentiviral particles were produced by the calcium chloride-mediated transfection of HEK-293T cells. The supernatants were collected 24 hours post-transfection for 8 hrs and used immediately for transductions. For naïve Th-cell transduction, CD4+ cells were isolated and incubated in 6-well plates coated with anti-hamster antibodies, viruses, polybrene (8 μg/ ml), and anti-CD3 and CD28 antibodies under skewing conditions for 16-18 hours. The medium was then replaced with fresh skewing medium, and 24 hours later, the medium was replaced again with selection medium, containing puromycin (8 μg/ml, Sigma) for 3 more days. The D5 and D10 cells were stimulated with antigen and transduced 24 hours later with the lentiviral particles for 16-18 hours, then the medium was replaced with medium for selection (IL-2 and puromycin) for 2-5 weeks. The following shRNA sequences were used:

RNA extraction and Real-time PCR
Total RNA was extracted, reverse-transcribed, and amplified. Melt curves were run to ensure amplification of a single product. The ratio between the transcripts was calculated as: (

Flow Cytometry
Intracellular staining was performed using the BD Cytofix/Cytoperm kit, according to the manufacturer's instructions. The cells were stained with anti-Mel-18 (sc-10744, Santa Cruz) antibodies.

Cytokine ELISA
The ELISA kits purchased from BioLegend were used.

Mel-18 regulates transcriptional patterns in primary Th1 and Th2 cells
Since, as we have previously shown, Mel-18 binds the Ifng promoter in Th1 cells and Il4 promoter in Th2 cells in association with gene expression [25], we wanted to examine its functional role in the regulation of these cytokine genes. Freshly isolated CD4 + T cells (naive) were stimulated under Th1-or Th2-skewing conditions; Th1 cells expressed Ifng mRNA and Th2 cells expressed Il4 mRNA after stimulation, indicating that the cells were adequately differentiated and stimulated (data not shown). Mel-18 mRNA level increased in differentiating Th1 and Th2 cells, peaking on the second day ( Figure   1A). The expression was basically unchanged following re-stimulation with PMA and ionomycin (P+I), which mimics TCR stimulation. Although the expression was higher in Th2 than in Th1 cells, the pattern was similar, supporting the idea that Mel-18 has parallel functions in both lineages.
Naive cells were stimulated under Th1-or Th2-skewing conditions, and simultaneously transduced with lentivirus encoding either one of two different shRNAs directed to Mel-18 or a control scrambled sequence ( Figure 1B and 1C). In Th1 cells the level of Mel-18 mRNA was downregulated to~35-45% ( Figure 1B); Mel-18 protein level was also reduced ( Figure 1D).
Downregulation of Mel-18 resulted in a decreased amount of Ifng mRNA ( Figure 1B), as well as of IFNγ protein ( Figure 1E), indicating that Mel-18 is a positive regulator of Ifng. Robust expression of Ifng requires the activity of transcription factors downstream to the TCR and cytokine receptors. However, the decline in Ifng expression probably did not result from changes in the expression levels of the mRNA or protein of the Th1lineage specifying transcription factor T-bet, as these levels were essentially unaffected ( Figure 1B and 1F). Similarly, the amounts of NFAT1 (NFATc2) mRNA and protein were almost unchanged ( Figure 1B and 1F). The mRNA level of another cytokine Tnfa was also reduced, as well as of Rad50, which is expressed in both Th1 and Th2 cells from the Il4 locus. The expression of Noxa, encoding a pro-apoptotic protein that is repressed in Th cells by the PcG protein Bmi-1, [50], was almost unchanged. In contrast, the expression of Hoxa7, a known PcG target gene during development, which is also involved in T-cell leukemia [51], was significantly upregulated.
In Th2 cells, the expression of Mel-18 was knocked down to~20-40%, and consequently the amount of Il4 mRNA was reduced by half ( Figure 1C). The decrease in the levels of Mel-18 and IL-4 was confirmed ( Figure 1D and 1E). These results indicate that in Th2 cells, as in Th1 cells, Mel-18 positively regulates the expression of the hallmark cytokine gene. Mel-18 was also necessary for the expression of Il10 and two other Th2 cytokine genes that are expressed from the Il4 locus, Il5 and Il13 (the later was reduced significantly only with one of the Mel-18 shRNAs). The levels of the two transcripts of the Th2-lineage specifying transcription factor Gata3 (containing the alternative 1a and 1b exons [52]) were similar to the control, except that one of the shRNAs led to a decreased Gata3(1a) level. We did observe a reduced level of GATA3 protein ( Figure 1F), thus it is possible that Mel-18, like Bmi-1 [53], is involved in the stabilization of GATA3. The mRNA amounts of NFAT1, Rad-50, and Noxa were comparable to the control. However, the expression of Hoxa7 was increased and Th2 (right) cells as described in Figure 1B, except the cells were re-stimulated with αCD3 and αCD28 antibodies. Differences between knockdown and control with p values ≤ 0.05 are indicated with an asterisk. The results in Figure 1 are the mean of two to five independent experiments +S.D., except in panels 1D and 1F, which are representative experiments.
following Mel-18 knockdown, even more than in Th1 cells.
Similar results demonstrating the positive regulation of cytokine genes and negative regulation of Hoxa7 by Mel-18 were obtained when the cells were re-stimulated with anti-CD3 and anti-CD28 antibodies ( Figure 1G). To determine whether Mel-18 is essential only for the initiation of cytokine gene expression or also functions in differentiating Th1 and Th2 cells, the cells were transduced with the lentiviral shRNAs 48 hours after the first stimulation; the results were comparable (data not shown). Given our previous data showing the inducible and selective binding pattern of Mel-18 at the signature cytokine gene loci in Th1 and Th2 cells [25], the above results suggest that Mel-18 can function unconventionally as transcriptional activator of Ifng in Th1 cells and of Il4 in Th2 cells. Other cytokine genes, such as Tnfa, Il5, and Il10, may also be direct targets of Mel-18.

Mel-18 has a dual function in Th cells
We next aimed to test whether Hoxa7 is a direct target of Mel-18 in Th cells. The Hoxa7 mRNA level was high in normal naive cells and decreased after first stimulation and again after re-stimulation ( Figure 2A). This pattern was the opposite to the pattern of cytokine gene expression. The expression of Hoxa7 was higher in Th2 cells than in Th1 cells, but the dynamic was similar. ChIP experiments confirmed that Mel-18 bound directly to the promoter region of Hoxa7 in Th2 cells ( Figure  2B), and to a lesser extent, in Th1 cells (data not shown). The binding was induced following re-stimulation, as at the cytokine genes.
Since Mel-18 positively regulates the expression of cytokine genes and negatively of Hoxa7, it may function as both a transcriptional activator and repressor in Th cells, in a gene-specific manner. We previously observed some low binding activity of the PcG proteins at the opposing hallmark cytokine genes [25], and we asked whether Mel-18 represses their expression. The knockdown of Mel-18 concomitant with the first stimulation did not elevate the expression of Ifng in Th2 cells, which stayed low (data not shown). In contrast, the level of Tbx21 was increased in Th2 cells by~1.5-2-fold following downregulation of Mel-18 ( Figure 2C). Mel-18 bound to the Tbx21 promoter in 1 hr re-stimulated Th1 cells, but its binding activity was much stronger in restimulated Th2 cells ( Figure 2D). Thus, the expression of Tbx21 in Th2 cells was possibly repressed directly by Mel-18. All together these results suggest that Mel-18 plays a bi-functional role in Th cells.
The levels of Il4, Il5, and Il13 and of both Gata3 transcripts, were almost unaffected in Th1 cells (data not shown). In contrast to its binding pattern at Tbx21, Mel-18 was bound to the Gata3 promoters in correlation with gene expression, in Th2 and not in Th1 cells ( Figure 2E). These results suggest that Mel-18 is not involved in the transcriptional silencing of Gata3 in Th1 cells, but it is possible that a more severe downregulation of Mel-18 is necessary to reveal its effect on Gata3 expression in Th2 cells, as in Mel-18 deficient mice [54].
Mel-18 knockdown resulted in~20-50% of live cells compared to the control ( Figure 2F), thus, in Th cells Mel-18 most likely regulates the expression of additional genes regulating proliferation or cell survival.
Eed and Ring1A positively regulate the expression of the hallmark cytokine genes in Th cells We next investigated the function of the PRC2 protein, Ezh2, which also binds to cytokine gene promoters selectively in Th cells [25]. The knockdown of Ezh2 in Th2 cells resulted in decreased expression of Il4, however, in Th1 cells the expression of Ifng, was unchanged (data not shown). At present, we do not know whether the knockdown was insufficient to reveal an effect, or whether Ezh2 is unnecessary for the transcription of Ifng in developing Th1 cells. Alternatively, Ezh2 function is partially redundant with Ezh1 [55,56].
To determine whether other PRC2 proteins regulate cytokine gene expression in Th1 cells, we repeated the experiments using shRNAs directed to Eed. Eed, as other PcG proteins, binds differentially to Il4 and Ifng promoters, in correlation with gene expression [25]. Eed is necessary for the histone lysine methyltransferase activity of Ezh2, and also for the propagation of the H3K27me3 mark [57,58]. The pattern of Eed mRNA expression in Th cells ( Figure 3A) was similar to that of Mel-18 mRNA ( Figure 1A).
Three different shRNAs knocked Eed down to~40-50% in Th1 cells and reduced the expression of Ifng to the same extent; in contrast, the expression of Tbx21 was unchanged, and that of Hoxa7 increased ( Figure 3B, left panel). Similarly, in Th2 cells, three different shRNAs knocked Eed down and consequently the expression of Il4 was decreased, but not that of Gata3 (1b), at least with two of the shRNAs ( Figure 3B, right panel). The expression of Hoxa7 was always higher following Eed knockdown, but the results in some cases are not considered statistically significant because of large differences in the extent of the upregulation (For example, downregulation of Eed with shRNA (Ee3), increased the expression of Hoxa7 mRNA by 1.51, 4.28, 5.34, and 1.55 in four different experiments). The decreased expression of Eed protein was confirmed (Figure 3C), as well as of IFNγ and IL-4 ( Figure 3D). These results indicate that PRC2 components are necessary for the expression of the signature cytokine genes in developing Th1 and Th2 cells.  The PRC1 complex contains also Ring1A, which binds the cytokine genes in correlation with their expression [25]. Ring1B catalyzes the ubiquitination of histone H2A (H2AK119ub1), while both Ring1A and Ring1B contribute to this activity in vivo [59][60][61]. This modification can repress gene expression by maintaining RNA polymerase II at poised configuration [36]. The pattern of Ring1A expression resembled those of the other PcG proteins, although its relative level was higher in naive cells ( Figure 4A). The knockdown of Ring1A in Th1 and Th2 cells with each one of three different shRNAs downregulated its expression approximately by half ( Figure 4B). Accordingly, the expression of Ifng in Th1 and Il4 in Th2 cells was reduced in a similar degree, but that of the lineage-specifying transcription factors was not. The expression of Hoxa7 was increased in both Th1 and Th2 cells with most of the shRNAs. Decreased amounts of the protein Ring1A and of the proteins IFNγ and IL-4 was also observed (Figure 4 C,D). The knockdown of Eed or Ring1A, like that of Mel-18, reduced the number of live cells (data not shown). Under these conditions, none of the signature cytokines or tissuespecifying transcription factors was derepressed in the opposing lineage following the knockdown of Eed or Ring1A (data not shown). Taken together, our results show that PcG proteins from two different complexes positively regulate the expression of the hallmark cytokine genes in Th1 and Th2 cells.

PcG proteins positively regulate the expression of cytokine genes in established Th cells
Since the PcG proteins are well known for their role in maintaining epigenetic states (although in association with gene silencing), we examined their effect on cytokine gene expression in established Th cells, using the murine D5 (Th1) and D10 (Th2) clones. Mel-18 bound to Ifng and Il4 genes in D5 and D10 cells in correlation with their expression, as in primary Th1 and Th2 cells ( Figure 5A). In D5 cells, Mel-18 mRNA was knocked down to~50% ( Figure 5B, left panel), and Mel-18 protein was reduced as well ( Figure 5C). The level of the Ifng mRNA varied; it was reduced to about half in some experiments and unchanged in others (for example, the results for shRNA (M1) were: 0.62, 0.53, 0.52, 1.07, 1.19, 0.97). The knockdown was probably not efficient enough to consistently reduce the mRNA levels, especially since D5 cells express high levels of Mel-18 mRNA ( Figure 5D). The levels of the Tbx21 and NFAT1 mRNAs were basically unaffected. The D5 cells expressed very low amounts of Hoxa7 mRNA (data not shown). In D10 cells, the expression of Mel-18 mRNA and protein were reduced by~50% ( Figure 5B, right panel and Figure 5C), and consequently, the Il4 mRNA level was diminished. The level of Gata3(1a) was reduced moderately with two shRNAs, and those of Gata3(1b) and NFAT were almost unchanged. In contrast to primary Th2 cells, the level of Hoxa7 was similar to the control, suggesting that in committed cells the expression of this gene is repressed by other, probably more permanent mechanisms. The binding of Ezh2 at the cytokine promoters in stimulated D5 and D10 cells was also differential (Figure 6A), although not as strong as of Mel-18. Knockdown of Ezh2 mRNA ( Figure 6B) and consequently of Ezh2 protein ( Figure 6C), resulted in lower amounts of Ifng and Il4 mRNAs in D5 and D10 cells, respectively.  Figure 3 were performed for Ring1A.
The downregulation of Ifng expression following Ezh2 knockdown in D5 cells was in contrast to the results obtained with primary Th1 cells, and perhaps reflects the low Ezh2 mRNA level in these cells ( Figure 6D), which may have caused the knockdown to have a greater impact. Alternatively, the difference may be related to the differentiation stage or to the possibility that Ezh2 is necessary to propagate the epigenetic state during cell cycle, and therefore a longer-term experiment was required to observe an effect. All together, these results demonstrate that PcG proteins are necessary for the expression of the signature cytokine genes, even in committed Th cells.

Mel-18 is required for the binding of NFAT1 and T-bet to the Ifng promoter
To examine the mechanisms underlying PcG function, we next characterized the state of the Ifng promoter in Mel-18-knockdown cells. These experiments were done using D5 cells, since their proliferation was less sensitive to the downregulation of Mel-18 after 3-4 weeks of selection, compared with D10 cells ( Figure 7A). The

D10
Relative mRNA expression knockdown of Mel-18 in D5 cells with shRNA(M1) decreased the Ifng expression following re-stimulation with P+I, similar to but more consistent than that observed following re-stimulation with anti-CD3 and anti-CD28 antibodies (Figures 7B and 5B). A ChIP experiment demonstrated that, as expected, the downregulation of Mel-18 diminished significantly its binding to the Ifng promoter, but as a result also the binding of the transcription factors NFAT1 and T-bet ( Figure 7C). This was probably a specific effect and not a general change in the locus accessibility, because the binding of the PcG protein YY1 was unaffected. The binding of Ezh2 was not significantly reduced either. This also did not reflect changes in the level of NFAT1 or T-bet ( Figure 7D), as although the level of T-bet mRNA decreased moderately ( Figure 7B), almost no change was detectable at the protein level. The binding of Mel-18 at the Ifng promoter was abolished in D5 cells in the presence of Cyclosporin A (CsA), which impairs the translocation of NFAT to the nucleus ( Figure 7E). We previously observed similar results in primary Th cells [25]. In contrast, the binding activity of Mel-18 at the Hoxa7 promoter region was unaffected. These results suggest that the binding activities of the PcG proteins and NFAT are mutually dependent, and also that the mechanisms for the recruitment of the PcG proteins differ, depending on their function.  Figure 6 are the mean of three to six independent experiments +S.D., except panels 6C, which is a representative experiment.
The knockdown of Mel-18 resulted in a reduced acetylation of histone 3 (H3K9/14Ac), which is a transcriptional permissive mark ( Figure 7C). It is not clear yet whether this was the cause or the consequence of the decreased binding activity of T-bet and NFAT. In contrast, H3K4me3, which is also a permissive mark, was stronger. We did not recognize significant changes in H3K27me3, which was lower at the Ifng promoter than at the Il4 promoter ( Figure 7C and data not shown). We did not observe a detectable level of H2AK119ub1 before or after Mel-18 knockdown (data not shown). Taking together these results suggest that the activity of the PcG proteins as transcriptional activators does not necessarily involve their classical histone modifications.

Discussion
While the role of the PcG proteins during development has been extensively studied, there is little information about their functions in differentiated cells. We used the RNAi approach to study the functional role of the PcG proteins in Th cells since the knockout of PRC2 members causes embryonic lethality [62]. And although the PRC1 members, excluding Ring1b, display more redundancy, they have a pivotal role during hematopoiesis [24,44,45].
Mel-18 deficient mice have severe proliferative defects in lymphoid cells resulting in hypoplasia of spleen and thymus [44,63,64], and have less than 5% of the thymocytes of wild-type mice [65]. The differentiation of Mel-18-deficient Th2 cells is impaired, but not of Th1 [54]. It is possible that the absence of Mel-18 during T cell development interferes selectively with Th2 differentiation. The expression of Gata3 is reduced in cells derived from Mel-18-deficient mice [54], and -apart from the fact that GATA3 is important for T cell development in the thymus [66] -its early presence may be required for  Figure 7B. c-jun is used as loading control. Cyt; cytosolic, Nuc; nuclear. (E) ChIP experiment assessing the binding of Mel-18 at the Ifng and Hoxa7 promoters in resting and 1-hour-restimulated D5 cells in the presence or absence of CsA. The binding activity at the Ifng promoter following stimulation in the absence of CsA was set as 1. The results in Figure 7 are the mean of three to five independent experiments +S.D., except panel 7D, which is a representative experiment.
normal subsequent Th differentiation. GATA3, for example, might be necessary for the initiation of the intrachromosomal conformation at the Il4 locus [67]. The negative regulation of Tbx21 by Mel-18 in Th2 cells, as we showed, may also explain the modest enhancement in the expression of IFNγ in Th cells derived from Mel-18-deficient mice.
T cells in general response moderately to the RNAi procedure [68], and since the knockdown is partial, our results probably underestimate the importance of PcG protein functions in Th cells. Nevertheless, our data suggest that PcG proteins from two different complexes can function as transcriptional activators of the signature cytokine genes in primary Th1 and Th2 cells. We showed similar results also in primary Th17 cells [69]. And Mel-18 and Ezh2 function as positive regulators of cytokine genes in established Th1 and Th2 cell lines. Therefore, PcG proteins may act as a general Thmachinery that induces the expression or maintains the permissive epigenetic state of the effector cytokine genes. The PcG proteins also perform the opposite activity in Th cells, acting as transcriptional repressors of Hoxa7 in both lineages, and of Tbx21 in Th2 cells. We also cannot exclude the possibility that a stronger downregulation of PcG proteins is necessary to detect negative regulation of the opposing cytokine genes. The PcG probably have many other targets in Th cells and the type of PcG activity might depend on the epigenetic context, the proteins available at the target genes, and differential posttranslational modifications. But it is also possible that discrete PcG complexes or isoforms are involved.
Since we observed fewer live cells following the knockdown of Mel-18, Eed, or Ring1A, additional targets of these PcG proteins in Th cells are probably genes involved in proliferation or survival. It was shown that the PcG proteins Bmi-1 and Ring1B modulate apoptosis of Th2 cells through regulation of Bim and Noxa genes, respectively [50,70]. However, the lower cell numbers is probably not the reason for the downregulation in the expression of the cytokine genes based on several reasons: (i) The expression of the mRNAs of NFAT and the lineage specifying transcription factors was in general normal, indicating that the effect on the cytokine genes was specific and did not reflect a general impairment in lineage differentiation or gene expression; (ii) There was no correlation between the potential of specific shRNAs to reduce the expression of cytokines and their effect on cell numbers; shRNA(M2) directed to Mel-18 reduced the cell numbers more efficiently than shRNA(M1), but both presented similar efficiencies in downregulating cytokine gene expression; (iii) knock down of Ezh2 in Th17 cells [69], and with some shRNAs in Th1 and Th2 cells (data not shown) did not result in diminished cell numbers, but yet downregulated cytokine gene expression.
Because the PcG proteins are well known as transcriptional silencers, one possible explanation for our results is that the PcG proteins positively regulate the expression of the cytokine genes indirectly, by repressing a repressor(s). However, it is more likely that the PcG proteins function as genuine transcriptional activators of cytokine genes for the following reasons: (i) The PcG proteins bind directly, differentially, and inducibly to the active cytokine genes [25]. It is less plausible that such a strong correlation is unrelated to their function. Moreover, their binding pattern was dynamic following stimulation, and different PcG proteins bind regulatory elements of the same gene with differential relative intensity [25]. Also, the PcG proteins do not bind exclusively active genes; for example, Mel-18 was associated with Tbx21 in Th2 stronger than in Th1 cells, in association with gene repression. All together these results strengthen the idea that the binding activity is specific and does not reflect irrelevant 'stickiness' to accessible DNA; (ii) Several studies in Drosophila and in mammalian embryonic stem cells reported that 10-20% of the PcG targets were transcriptionally active; therefore, the binding of PcG proteins does not necessarily lead to silencing (reviewed in [43]). Moreover, in a human colon cell line, Suz12 and Ezh2 were associated with the promoters of several genes that were downregulated upon Suz12 depletion [71]. In addition, Ezh2 (but not other PcG proteins) was found as a transcriptional activator of c-Myc and cyclin D1 in a breast cancer cell line [72]. The dual function of the PcG proteins may be a dominant feature of more-committed cells, but possibly exists in other developmental stages as well.
Recently, Th cells were shown to exhibit more plasticity than it was previously appreciated [3,[5][6][7][8][9][10][11][12]. Th17 cells exhibit the highest degree of flexibility, even at a late developmental stage, in the absence of polarizing cytokines [73]. However, viral infection can reprogram also Th2 cells in vivo into a Th2-Th1 cells expressing GATA3 and T-bet simultaneously [74]. Whereas Tbx21 in Th1 cells and Gata3 in Th2 cells are marked selectively with the permissive H3K4me3 and not with the repressive H3K27me3, in the opposing lineage and in Th17 cells, these genes are marked bivalently [75], like the transcriptional regulator genes during embryonic development. This is considered as a "poised" gene status, with the potential for subsequent activation or silencing. Tbx21 was bound more strongly by Mel-18 in Th2 than Th1 cells, in correlation with gene repression. Indeed its expression was derepressed in Th2 cells following knockdown of Mel-18. These results suggest that the PcG proteins may regulate, as transcriptional activators or repressors, gene networks that specify the expression profiles in Th cells. Gata3 expression was not increased following the knockdown of Mel-18 in Th1 cells, and it is possible therefore that it is silenced by other, more stable, mechanisms, or that the knockdown strength was not sufficient to observe this effect.
We found that PcG proteins were necessary during early Th differentiation and in established Th cell lines. They were bound under resting conditions, but their binding activity was increased following TCR stimulation, in correlation with the inducible expression of the cytokines (here and [25]). Therefore, they might function as either: (i) acute transcription factors similar to NFAT or even as co-factors; (ii) epigenetic regulators that inducibly increase their binding activity to propagate the transcriptional status during cell duplication, which normally follows stimulation. They may cooperate with other chromatin remodeling factors such as BRG1, which is also recruited in a selective and partially inducible manner to the cytokine genes Ifng in Th1 [76] and Il4 in Th2 cells [77,78]. Also they may interact with trithorax group proteins such as MLL that was found to be essential for the ability of Th2 memory cells to express the Th2 cytokines [79].
In Th2 cells, the expression level of the GATA3 protein was decreased following Mel-18 knockdown. However, stabilization of the lineage specifying transcription factors cannot be the only mechanism by which the PcG proteins regulate cytokine gene expression. For example, the level of GATA3 was essentially unchanged following the knockdown of Ezh2 in Th2 cells (data not shown), but the expression of Il4 mRNA was reduced. Similarly, the expression of T-bet was unchanged in Th1 and D5 cells following the knockdown of Mel-18, but the expression of Ifng mRNA was downregulated.
What are the relative functional roles of the lineage specifying transcription factors, TCR-inducible transcription factors and the epigenetic machinery in the potential network that maintains the transcriptional programs in Th cells? The restricted recruitment of the generally expressed PcG proteins is probably regulated directly or indirectly by the selectively expressed lineage-specifying transcription factors. The targeting of the PcG proteins to the cytokine genes is also most likely NFAT-dependent since we found that impaired translocation of NFAT to the nucleus abrogates the binding of PcG proteins at the cytokine genes in primary [25] and established Th cell lines (here and data not shown). It is possible that NFAT is directly involved in the recruitment of these PcG proteins, and the same mechanisms that restrict the binding of NFAT [4], dictate the selective targeting of PcG proteins.
Conversely, we also showed that in an established Th1 cell line Mel-18 was necessary for the recruitment of T-bet and NFAT1 to the Ifng promoter. Therefore, the binding activities of the PcG proteins and of NFAT are mutually dependent. The binding activities of the PcG proteins and the lineage-specifying transcription factors can also be interrelated. The PcG proteins in general do not have sequence-specific binding sites, but they may stabilize the binding of these transcription factors on the chromatin. They may also have an indirect effect by affecting chromatin accessibility, although it must be specific since the association of YY1 was unchanged. Instead, PcG proteins can regulate the expression of factors that are necessary for the binding activity of T-bet and NFAT.
The lineage specifying transcription factors are not necessarily associated with the PcG proteins, but both types of factors can take functional turns in a way that the binding of each of them is a prerequisite for the binding of the other one. The lineage-specifying transcription factors may specify the binding sites for the general machinery, which on its turn facilitates the heritability of the epigenetic programs during mitosis, and by that assists the re-establishment of the specific factors. Absence of the specific factors that re-enforce the selective recruitment of the general maintenance machinery, may result in progressive dilution of epigenetic marks during mitosis, and consequently impaired the accessibility of the lineage specifying transcription factors to their target genes.

Conclusions
Our results show that PcG proteins have a dual function in Th cells as positive and negative regulators of gene expression. Our results also suggest that both activities can results from a direct effect. The lineage-dependent recruitment of the PcG proteins and consequently their restricted regulation of the effector cytokine genes have a potential for therapeutic applications. The expression of Ezh2 mRNA is significantly downregulated in patients with active systemic lupus erythematosus [80], which could be one of the reasons for their abnormal cytokine production.