Regulation of Interferon-γ During Innate and Adaptive Immune Responses - 5. SIGNALING PATHWAYS IN THE DIFFERENTIATION OF CD4 AND CD8 T CELLS

5. SIGNALING PATHWAYS IN THE DIFFERENTIATION OF CD4 AND CD8 T CELLS

As individual CD4 T cells commit to the Th1 or Th2 effector lineage to the exclusion of the other lineage, they have provided a well-utilized model in which to study cellular and molecular events involved in cell fate choices. Much less is known regarding the Th17 fate choice, although knowledge in this area is rapidly accruing. (For a review of Th17 differentiation, see Harrington et al., 2006.) Th2 differentiation has also been reviewed (Ansel et al., 2003, 2006; Barbulescu et al., 1998; Lee et al., 2006; Szabo et al., 2003).

 

Herein, we review the signaling pathways and cellular events that lead to Th1 differentiation and IFN-γ  production, and draw parallels to CD8 T cell effector development and IFN-γ secretion.

 

5.1. Naive T cells require antigen stimulation for proliferation and effector commitment

While NK and NKT cells are able to secrete IFN-γ within hours of their stimulation, differentiation of naive CD4 and CD8 T cells into efficient IFN-γ secreting cells requires several round of proliferation. While they proliferate, environmental signals influence the expression and activation status of specific receptors, downstream signaling molecules, and transcription factors, which in turn allow these T cells to express IFN-γ , remodel the Ifng locus, and commit to the Th1 CD4, or CD8 CTL effector lineage (Murphy and Reiner, 2002; Reiner, 2001; Reiner and Seder, 1999).

 

For naı¨ve CD8 T cells, signals delivered from the TCR and costimulatory molecules induce differentiation into a fully committed CTL that is capable of IFN-γ secretion and direct killing of infected or transformed cells expressing the cognate MHC class I:peptide complex. The directed commitment of CD8 T cells to become IFN-γ producing CTLs is thought to result from the constitutive expression of the T-bet paralog Eomesodermin (Eomes) by naive CD8 T cells (Pearce et al., 2003).

 

By contrast, naive CD4 T cells are more plastic. When stimulated via their TCR and costimulatory molecules, naive CD4 T cells transcribe and produce substantial amount of IL-2 and also produce small amounts of IFN-γ and IL-4 (Grogan and Locksley, 2002), which may poise them to adopt multiple effector phenotypes.

 

The initial coactivation of the Ifng, IL-4, and IL-2 cytokine genes in naive T cells occurs through the concerted activation of constitutively expressed, and rapidly activated transcription factors, including nuclear factor of activated T cells (NFATs), nuclear factor kB (NFkB), and activator protein-1 (AP-1).

 

This coactivation may be facilitated by the juxtaposition of the genes encoding these cytokines in the nuclei of naive T cells, even though these genes are located on different chromosomes (Spilianakis et al., 2005). Since their differentiation into Th1, Th2, or Th17 effector lineages determines whether the ensuing immune response will be appropriate for specific types of pathogens, it is appropriate that the dominant factor governing these cell fate choices is the cytokine milieu produced by APCs, which use TLRs and other cell surface and cytosolic recognition systems to deduce the nature of the microbial threat and instruct T cell fate choice accordingly (Pulendran and Ahmed, 2006).

 

Other factors involved in lineage choice are the nature of the APC and the magnitude of stimulation via the TCR and costimulatory molecules, with stronger and longer signaling, generally favoring Th1 development. However, the cytokine milieu present during CD4 differentiation remains the best-characterized and most vital influence on naive CD4 T cell priming and early differentiation

 

 

5.2. T cells require cytokine signals for sustained IFN- γ expression

Cytokines play an important role in IFN-γ induction, maintenance, and Th1 differentiation in CD4 T cells.

 

While IFN-γ  production appears to be relatively independent of cytokine signals in NKT cells, the importance of IL-12, IL-18, and IFN-γ  itself have been well documented in CD4 and CD8 T cells as in NK cells, and many of the downstream signaling pathways are shared among these cell types. The ability of other cytokines, such as type I IFNs, to support IFN-γ  expression varies among these cell types, as described below.

 

Th1 development is heavily influenced by IFN-γ  produced by NK cells and by IL-12 and IL-18 produced by DCs and other APCs. Binding of IFN-γ to its receptor on the surface of CD4 T cells leads to STAT1 phosphorylation and nuclear translocation, which in concert with signals from the TCR and CD28 costimulatory molecules, induces T-bet (Lighvani et al., 2001).

 

The induction of IFN-γ transcription and commitment to the Th1 lineage by T-bet is facilitated by its ability to induce two additional transcription factors, Hlx (Mullen et al., 2002) and Runx3 (Djuretic et al., 2007). Runx3 binds cooperatively with T-bet at the Ifng promoter to induce its transcription and to the Il4 silencer to extinguish its expression (Djuretic et al., 2007). Hlx facilitates Ifng transcription and chromatin accessibility at the Ifng promoter (Mullen et al., 2002) and helps to induce the expression of the IL-12Rb2 chain (Afkarian et al., 2002). Binding of the IL-12 p35 and p40 heterodimer to its receptor (IL-12Rb1 and IL-12Rb2) in turn activates Jak2/Tyk2 and induces signals via phosphorylation and nuclear translocation of STAT4 (Trinchieri et al., 2003), one effect of which is to induce expression of the IL-18 receptor. IL-12 also activates p38 MAPK.

 

Loss of p38a activity in T cells, either by use of a pharmacological inhibitor or p38a-deficient cells, blocks IFN-γ production in response to cytokine stimulation but not TCR stimulation (Berenson et al., 2006b). The combined signals from IFN-γ , IL-12, and IL-18 maximize expansion and optimal activation of Th1 cells (Grogan and Locksley, 2002; Ho and Glimcher, 2002; Murphy and Reiner, 2002), facilitating their permanent commitment to the Th1 lineage. IFN-γ acts in a positive feedback loop to facilitate its own expression by T cells, as it does in NK cells. In response to activation of naive CD4 T cells via the TCR and costimulatory molecules, the IFN-γ receptor is recruited to the immunologic synapse; this recruitment decreases significantly in the presence of IL-4 and is weaker in naive CD4 T cells from Th2- biased BALB/c mice compared to C57BL/6 mice (Maldonado et al., 2004).

 

The focused recruitment of IFN-γ receptors to the synapse where IFN-γ is being secreted results in an autocrine, positive feedback loop facilitating IFN-γ production and Th1 lineage commitment (Maldonado et al., 2004). While IFN-γ is not required for the induction of Th1 cells, it plays a critical role in suppressing the IL-4-producing potential of Th1 cells (Zhang et al., 2001) and in maintaining the expression of T-bet during CD4 Th1 effector commitment (Afkarian et al., 2002). Type I IFNs have been implicated in the phosphorylation of STAT4 and Th1 differentiation by human CD4 T cells (Cho et al., 1996; Rogge et al., 1997; Tyler et al., 2007). When combined with IL-18 stimulation, IFN-a/b can drive acute IFN-γ secretion by human CD4 T cells (AthieMorales et al., 2004; Brinkmann et al., 1993; Matikainen et al., 2001; Parronchi et al., 1992; Sareneva et al., 1998, 2000).

 

In human T and NK cells, IFN-a stimulation upregulates MyD88 mRNA and synergizes with IL-12 to induce the IL-18 receptor complex, sensitizing these cells to low concentrations of IL-18 (Sareneva et al., 2000). The role of type I IFNs in murine CD4 Th1 development, however, is less clear, as STAT4 phosphorylation in response to IFN-a/b stimulation does not occur as efficiently in mouse T cells as it does in human T cells (Berenson et al., 2004a; Farrar and Murphy, 2000; Farrar et al., 2000a,b; Freudenberg et al., 2002; Persky et al., 2005; Rogge et al., 1997, 1998; Szabo et al., 1997). As a result, mouse CD4 T cells are unable to secrete substantial amounts of IFN-γ or differentiate into Th1 effector cells in response to IFN-a/b stimulation (Berenson et al., 2004a; Persky et al., 2005; Rogge et al., 1998; Wenner et al., 1996). Thus, IL-12R but not type I IFNs can induce the sustained STAT4 activation and IFN-γ production signaling required for Th1 development (Berenson et al., 2006a).

 

In CD8 T cells, which are less dependent on IL-12 for differentiation than CD4 T cells (Carter and Murphy, 1999), IFN-a appears to be sufficient for clonal expansion and gain of cytotoxic effector function on primary stimulation, but is still considerably less effective than IL-12 in facilitating their differentiation into efficient producers of IFN-γ (Curtsinger et al., 2005). These differences in the ability of IFN-a/b to induce STAT4 phosphorylation likely result from differences between human and murine T cells in the recruitment of STAT4 to the IFNAR. In humans, STAT4 recruitment to IFNAR is STAT2 dependent. However, the murine Stat2 gene contains a carboxy-terminal minisatellite repeat not found in the human gene (Farrar et al., 2000a; Park et al., 1999; Paulson et al., 1999).

 

Expression of mouse STAT2 in human STAT2/ T cells failed to restore IFNa/bdependent STAT4 phosphorylation, demonstrating that mouse STAT2 is not functional in human T cells. A chimeric murine N-terminal/human C-terminal STAT2 gene did restore STAT4 phosphorylation in STAT2- deficient human fibroblasts (Farrar et al., 2000a,b), but was unable to support Th1 development or IFN-γ production in response to type I IFNs in transgenic mice (Persky et al., 2005). The N-terminal domain of STAT4 was shown to interact with the cytoplasmic domain of the human IFNAR2 subunit but not murine IFNAR2 (Tyler et al., 2007), suggesting that differences in both IFNAR and STAT2 contribute to the impaired STAT4 phosphorylation in murine T cells. Together, these data suggest that in murine T cells type I IFNs are unable to induce an effective and sustained STAT4 signal that is necessary for full commitment to IFN-γ production and Th1 differentiation. While IL-12 is a potent regulator of Th1 immunity, two closely related cytokine family members, IL-23 and IL-27, have been found to play different roles in the development of CD4 effector lineages. Similar to IL-12, IL-23 is produced by activated DCs and its receptor complex is similarly upregulated on NK cells and activated/memory CD4 T cells, DCs, and bone marrow-derived macrophages (Trinchieri et al., 2003).

 

Early studies implicated IL-23 in later stages of Th1 development or maintenance of IFN-γ production (Oppmann et al., 2000), but IL-23 has recently been shown to be involved in the survival and expansion of the Th17 lineage of CD4 effector T cells (Dong, 2006; Weaver et al., 2006). IL-23 shares the p40 subunit with IL-12, which together with a unique p19 subunit signals through the IL-23 receptor containing the IL-12Rb1 and IL-23R chains to induce activation of STAT3/STAT4 heterodimers, as compared to the STAT4 homodimer that is induced in response to binding of IL-12 to its receptor. IL-27 synergizes with IL-12 and IL-18 to induce IFN-γ production, contributes to Th1 differentiation, and can induce the proliferation of Th1 effector cells. IL-27 is a heterodimer expressed by virally infected cells, activated macrophages, and DCs, and is composed of a p28 subunit and an Epstein–Barr virus-induced gene 3 (EBI3) subunit. The receptor for IL-27, composed of GP130 and T cell cytokine receptor (TCCR)/WSX-1 subunits, has strong homology to that of IL-12Rb2 and is expressed primarily by NK cells and resting CD4 T cells. IL-27 activates both STAT1 and STAT3 proteins; the activation of STAT1 in turn induces T-bet expression, thereby facilitating IFN-γ production and the differentiation of naive CD4 T cells into IL-12-responsive, Th1 effectors (Pflanz et al., 2002). Consequently, mice with deficiencies in IL-27 signaling show diminished CD4 IFN-γ production following primary immunization with keyhole limpet hemocyanin protein (Yoshida et al., 2001) and increased susceptibility to L. monocytogenes and Leishmania major (Chen et al., 2000; Yoshida et al., 2001). However, while the tempo with which Th1 responses develop in vivo is often delayed in IL-27 receptor-deficient mice, given time, these mice ultimately mount Th1 responses in response to chronic infection (Hunter, 2005), likely as a result of alternative pathways to induce T-bet and to facilitate Th1 differentiation (e.g., via IFN-γ and IL-12). By contrast to its redundant role in Th1 development, IL-27 may be essential for repressing the development of Th17 effectors (Hunter, 2005).

 

5.3. Other factors influencing Th1 lineage commitment

Notch signaling, an evolutionarily conserved pathway involved in cell fate choices, also affects T cell differentiation and Th1 and Th2 effector commitment. In mammals, there are four Notch receptors: Notch1, Notch2, Notch3, and Notch4, and five canonical Notch ligands that can be divided into the Jagged ligand family, containing Jagged1 and Jagged2, and the Delta ligand family, containing Delta-like 1 (DLL1), DLL3, and DLL4 (Baron, 2003). Notch ligands are expressed by APCs and, as described above, TLR receptor recognition of pathogens by APCs likely leads to differences in the expression of cytokines and accessory molecules, such as Notch ligands, that provide early instructional signals to differentiating CD4 T cells. Notch signaling is central to several of the cell lineage choices made during T cell development, including T versus B cell commitment, ab versus gd T cell choice, and whether to develop into a CD4 or CD8 functionally mature T cell following selection (reviewed in Osborne and Minter, 2006). A role for Notch receptors in T cell activation, proliferation, and cytokine production has been established (Adler et al., 2003; Palaga et al., 2003). Use of pharmacological inhibitors that block Notch upregulation in response to TCR stimulation in mature CD4 and CD8 T cells significantly impaired T cell proliferation and IFN-g production (Palaga et al., 2003). Expression of Delta proteins by APCs during in vitro stimulation of naive CD4 T cells favors differentiation into a Th1 cell fate, whereas Th2 development is favored when APCs express the alternate Jagged family of Notch ligands (Amsen et al., 2004). These effects are largely independent of cytokine signaling, suggesting that Notch signaling may have a direct effect on the expression of IL-4 and IFN-γ. Inhibition of Notch function using a g-secretase inhibitor, which prevents cleavage of the cytoplasmic portion of the Notch receptor, blocked T-bet expression and Th1 commitment in Th1 conditions, but had no effect on developing Th2 cells (Minter et al., 2005). TCR stimulation of naive CD4 T cells in the presence of a DLL1-Fc fusion protein was both necessary and sufficient to significantly increase the expression T-bet and the number of cells secreting IFN-γ (Maekawa et al., 2003). Furthermore, naive CD4 T cells stimulated in Th2 conditions in the presence of DLL1-Fc produced decreased amounts of IL-4. Pretreatment of BALB/c mice with Delta1 prior to L. major infection supported the development of a Th1-based immune response, reducing the footpad swelling and promoting IFN-γ secretion by CD4 T cells. In good agreement with these data, pretreatment of mice with the g-secretase inhibitor delayed the onset of disease in the EAE mouse model of multiple sclerosis, as well as reduced the severity of disease when continuously administered to affected mice (Minter et al., 2005).

 

5.4. TGF-b and IL-6 negatively regulate  IFN-γ production and Th1 development

Similar to NK cells, TGF-b is a potent antagonist of  IFN-γ secretion and Th1 development by CD4 T cells. Mice deficient in TGF-b1 or its receptor develop widespread autoimmunity characterized by uncontrolled CD4 T cell activation and excess  IFN-γ production (Boivin et al., 1995; Gorelik and Flavell, 2000; Kulkarni et al., 1993; Lucas et al., 2000; Shull et al., 1992). TGF-b blocks Th1 differentiation by inhibiting expression of factors necessary for Th1 development: T-bet, STAT4, IL-12Rb2, and  IFN-γ itself. The SMAD3/4 heterodimers activated by TGF-bR signaling directly inhibit expression of the Th1 master regulator T-bet; conversely, the T-bet paralog, Eomes, which is expressed in NK cells and CD8 T cells but not in CD4 T cells, is not downregulated by SMAD3/4, allowing these cells to maintain some level of  IFN-γ production even in the presence of TGF-b. Studies in transgenic mice expressing a dominant-negative form of the TGF-bRII under the control of the CD4 promoter in T cells (Gorelik and Flavell, 2000), indicate that TGF-b signaling in CD4 and NK cells serves to reduce the number and function of Th1 effector cells in mice infected with L. major. Thus, while normal BALB/c mice are susceptible to L. major, dnTGF-bRII transgenic BALB/c mice generate greater numbers of antigen-specific  IFN-γ -producing Th1 effector CD4 T cells and exhibit decreased footpad swelling and parasite numbers (Gorelik and Flavell, 2000; Laouar et al., 2005). Ablation of TGF-bRII in T cells inhibits NKT cell development and CD8 T cell maturation in the thymus. The remaining peripheral T cells are phenotypically activated Th1 effector CD4 T cells that express T-bet, whereas the frequency of regulatory T cells, which require TGF-b for development, is reduced (Li et al., 2006a). Together these data implicate TGF-b signaling in various aspects of NKT and T cell development,  IFN-γ production, and homeostasis. IL-6 is produced by several cell types including macrophages, DCs, and B cells and is involved in the differentiation of B cells, macrophages, regulatory T cells, and Th17 CD4 effector T cells (Diehl et al., 2000). IL-6 also inhibits Th1 polarization by facilitating NFAT-dependent activation of the Il4 gene in naive CD4 T cells and by potentiating expression of the suppressor of cytokine signaling-1 (SOCS-1). SOCS-1 blocks signaling from the  IFN-γR by preventing the phosphorylation and subsequent activation of STAT1 (Diehl et al., 2000).

 

5.5. IFN-γ production by memory T cells in response to cytokine stimulation

Formation of a long-lived pool of memory T cells that are capable of a more rapid and robust immune response on reencounter with pathogens forms the basis of vaccination and protective immunity.

 

The ability of memory T cells to mount an effective secondary response is due in part to the increase in memory T cell precursor frequency compared to naive T cell precursor frequencies. In addition, relative to naive T cells, memory T cells have a lower activation threshold, enter the cell cycle more rapidly, and are poised for rapid and potent secretion of  IFN-γ (reviewed by Sprent and Surh, 2002).

 

Effector memory T cells, which primarily reside in nonlymphoid tissues, are poised to provide immediate immunity on antigenic reexposure. These cells constitutively express mRNA for Ifng and other effector molecules (Bachmann et al., 1999; Cho et al., 1999; Stetson et al., 2003; Zimmermann et al., 1999), allowing them to secrete  IFN-γ within hours of activation (Cho et al., 1999).

 

The efficiency with which memory CD8 T cells protect against reinfection has been demonstrated by studies showing that the transfer of memory CD8 CTL into  IFN-γ-deficient hosts that were subsequently infected with wild-type Listeria monocytogenes provided greater protection than transferred NK cells (Berg et al., 2005). Generation of memory CD8 T cells capable of responding in a recall response is dependent on the presence of CD4 T cells and IL-2 during the primary response (Sun et al., 2004a), whereas the subsequent maintenance of these cells is dependent, like NK cells, on IL-15 (Nishimura et al., 2000; Schluns et al., 2002).

 

The clonal expansion and long-term survival of memory CD8 T cells is further influenced by IL-12-dependent STAT4 activation during priming, which results in increased expression of Bcl3 and Bcl2-related genes (Li et al., 2006c). Unlike naive T cells, which require sustained antigen stimulation, costimulation, and cytokine signals to differentiate into effector cells capable of sustained robust expression of  IFN-γ, memory T cells can respond to cytokine or antigen stimulation in the absence of costimulation. Stimulation of human memory CD8 T cells with IL-12 and IL-18 is sufficient to induce  IFN-γ secretion (Berg et al., 2003; Smeltz, 2007), which can be substantially increased by addition of IL-15 (Smeltz, 2007). The ability of memory T cells to secrete  IFN-γ in response to cytokine stimulation in amounts similar to those induced by antigen stimulation is largely due to the fact that effector and memory CD8 T cells express more IL-12Rb2, IL-18Ra, and IL-18Rb than naive CD8 T cells (Berg et al., 2003; Raue et al., 2004).

 

6. TRANSCRIPTION FACTORS DOWNSTREAM OF THE TCR, ACTIVATING NK RECEPTORS, AND CYTOKINE RECEPTORS

Signaling cascades relay information from receptors on the plasma membrane through the cytoplasm and into the nucleus by inducing the expression, posttranslational modification, and/or nuclear translocation of transcription factors.

 

Transcription factors may activate expression by recruiting RNA polymerase-containing complexes to target genes, by recruiting protein complexes that alter chromatin structure such that the binding of other transcriptional activators is facilitated, or by a combination of these mechanisms; the converse is true for transcriptional repressors.

 

6.1. Factors downstream of the TCR, costimulatory, and activating NK receptors

A number of ubiquitously expressed transcription factors are involved in Ifng transcription-induced downstream of the TCR and costimulatory molecules. These include members of the NFAT, NFkB, AP-1, ATF-CREB, C/EBP, and Ets families (reviewed by Lin and Weiss, 2001; Murphy et al., 2000; Szabo et al., 2003). Nearly all of these factors are also involved in Ifng transcription in NK cells (Glimcher et al., 2004; Vivier et al., 2004; Zompi and Colucci, 2005). While T cells require antigen stimulation to express Ifng message, resting NK cells express low amounts of this transcript constitutively (Stetson et al., 2003).

 

Thus, the role of these transcription factors in Ifng induction in NK cells may be somewhat different than in T cells. In T cells, cyclic AMP-response element binding protein (CREB) is rapidly activated in response to TCR stimulation (Barton et al., 1996). CREB may also be activated in response to increases in cAMP (reviewed by Kuo and Leiden, 1999). CREB binds as a homodimer or heterodimer in conjunction with activating transcription factor (ATF) proteins. In resting T cells, CREB is maintained in an inactive unphosphorylated state that can bind DNA. Following antigen stimulation, CREB becomes phosphorylated on serine 133, allowing it to interact with an essential coactivator and histone acetyltransferase, CREB binding protein (CBP). The CREB–CBP complex can then recruit and activate the basal transcriptional machinery. By contrast to this positive role, CREB binding to the IFNG promoter (Table 2.1) appears to inhibit transcription, apparently by inhibiting the binding of Jun/ATF2 complexes (Penix et al., 1996), perhaps contributing to the known inhibitory effect of cAMP on  IFN-γ production. Activation of T cells through the TCR and CD28 leads to the activation of the MAPKs Erk1 and Erk2, which induces the transcription factor Elk, which in turn upregulates expression of the transcription factor c-Fos (Dong et al., 2002; Lin and Weiss, 2001). In addition, the c-Jun N-terminal kinases 1 and 2 (JNK1, JNK2) and p38 MAPKs are induced resulting in activation of c-Jun and ATF2, respectively, which results in the formation of AP-1, a heterodimer of c-Fos and c-Jun, and of c-Jun/ATF2 heterodimers. Binding of these heterodimers to the IFNG promoter helps to activate transcription (Penix et al., 1996; Sica et al., 1997; Sweetser et al., 1998). Consistent with the importance of ATF2, pharmacological inhibitors of the MAPK p38 or dominant-negative forms of this kinase inhibit IFN-γ production by T and NK cells (Berenson et al., 2004b; Rincon et al., 1998). CCAAT/enhancer binding proteins (C/EBPs) are a family of basic leucine zipper transcription factors (Poli, 1998). Two C/EBP isoforms have been described in lymphocytes: C/EBPg and C/EBPb. While most members of the C/EBP family are relatively limited in their tissue distribution, C/EBPg is unique in that it is ubiquitously expressed and constitutively active (Roman et al., 1990). C/EBPg can interact with other transcription factors containing a leucine zipper region, and positively or negatively influence their function. C/EBPg-deficient NK cells and splenocytes produce 90% less IFN-γ in response to IL-12 or IL-18 than control cells (Kaisho et al., 1999). The reduced  IFN-γ secretion is not a result of defects in the signaling pathways downstream of IL-12 or IL-18 receptors. Although the precise mechanism by which C/EBPg facilitates IFN-γ production was not demonstrated in these studies, these two transcription factors synergistically activate the IFNG promoter and augment IFN-γ production, even in cells in which a critical cytosine in this regulatory element is methylated (Tong et al., 2005). However, in the absence of T-bet, C/EBPb does not enhance and may inhibit transcription of Ifng (Berberich-Siebelt et al., 2000; Tong et al., 2005), suggesting that the function of C/EBP proteins in Ifng regulation is context dependent. Multiple signaling pathways including the TCR, CD28, and IL-18 receptors converge on the NFkB/Rel family of transcription factors. The five NFkB family members include RelA (p65), RelB, and c-Rel, which can transactivate target genes when dimerized, and NFkB1 (p50) and NFkB2 (p52), which do not contain transactivation domains and on their own can repress activation. NFkB proteins are maintained in an inactive form in the cytoplasm by interaction with IkB proteins. T cell activation induces IkB phosphorylation and degradation, which allows NFkB dimers to translocate to the nucleus (Ghosh et al., 1998). CD4 T cells that are unable to activate NFkB have decreased Th1 responses and IFN-γ production (Aronica et al., 1999; Corn et al., 2003; Tato et al., 2003). Furthermore, the number of CD4 T cells elicited in response to infection by the Th1 pathogen Toxoplasm gondii in transgenic mice expressing a degradation-resistant IkB-a is severely decreased, and is associated with the inability of these cells to proliferate and produce IFN-γ . Cytolytic function and IFN-g secretion of NK cells from these mice is also severely impaired (Tato et al., 2003). RelB-deficient CD4 T cells have impaired T-bet and STAT4 expression resulting in very low IFN-γ secretion (Corn et al., 2005). Consistent with this, RelB-deficient mice are more susceptible to infection with T. gondii (Caamano et al., 1999), and have impaired NK cell cytotoxicity and IFN-γ secretion, despite normal IL-12 secretion by APCs (Caamano et al., 1999). c-Rel-deficient NK cells also produce less IFN-γ in response to stimulation with IL-12 and IL-2 (Tato et al., 2006). In contrast, p50/ NK cells proliferate and secrete more IFN-γ in response to IL-12 and IL-2 or T. gondii infection, suggesting that it may function as a repressor of IFN-γ. Consistent with this possibility, chromatin immunoprecipitation (ChIP) assays show that p50 is bound to the murine Ifng promoter in resting NK cells and binding diminishes in response to IL-12 plus IL-18 (Tato et al., 2006). It is possible that this loss of p50 reflects its replacement by other NFkB proteins that contain activating domains, but if so, it is not c-Rel, because this factor was not detected at the Ifng promoter by ChIP in stimulated murine NK cells (Tato et al., 2006). There are other sites in the IFNG promoter and first intron to which NFkB proteins can bind, but whether this occurs in situ is not known (Sica et al., 1992, 1997). Furthermore, p50, p52, and RelB proteins are required for development of mature NKT cells (Matsuda and Gapin, 2005; Sivakumar et al., 2003). There are five members of the NFAT family of transcription factors, which are commonly referred to as NFAT1 (NFATp or NFATc2), NFAT2 (NFATc or NFATc1), NFAT3 (NFATc4), NFAT4 (NFATx or NFATc3), and NFAT5, of which NFAT1, NFAT2, and NFAT3 are expressed in lymphocytes (Macian, 2005). NFAT proteins are maintained in an inactive, hyperphosphorylated form within the cytoplasm of resting cells. Following TCR stimulation, sustained elevations in cytosolic calcium activate the phosphatase calcineurin, which dephosphorylates NFATs, resulting in their rapid nuclear import and increased DNA binding affinity. NFAT signaling processes are attenuated by the action of several kinases that phosphorylate NFATs, decreasing their DNA binding, and resulting in their rapid export to the cytoplasm (Rao et al., 1997). NFAT proteins are capable of binding DNA as homodimers or heterodimers and interact with several other transcription factors, including AP-1 (Hogan et al., 2003), making it difficult to assign a simple relationship between the activation of a specific cytokine gene and individual NFAT family members. NFAT proteins can bind to multiple sites within the IFNG promoter (Sica et al., 1997; Sweetser et al., 1998) and to the upstream enhancer CNS1/IfngCNS-6 (Table 2.1) (Lee et al., 2004), but the extent to which the effects of NFAT on Ifng expression are mediated through these regions of the gene is not clear. Mice with deficiencies in individual NFAT members have relatively mild immunophenotypes (Macian, 2005). Although NFAT1 and NFAT2 are expressed both by Th1 and Th2 CD4 cells and can induce transcription of either Ifng or Il4 (Monticelli and Rao, 2002; Ranger et al., 1998), NFAT1- deficient CD4 T cells have more sustained IL-4 production following stimulation, resulting in mild Th2-skewing, diminished-Th1 differentiation, and decreased expression of IFN-γ (Hodge et al., 1996; Kiani et al., 1997; Monticelli and Rao, 2002). Enforced expression of a constitutively active form of NFAT2 resulted in increased  IFN-γ production, strong Th1 skewing, and incomplete ability to commit to the Th2 effector lineage, even in the presence of IL-4 and blocking antibodies to IL-12 and IFN-γ (Porter and Clipstone, 2002). Conversely, NFAT2-deficient T cells are impaired in their ability to produce IL-4 and other Th2 cytokines (Monticelli and Rao, 2002; Ranger et al., 1998). The immediate phase of human NK cell activation through CD16 (FcgR3) is associated with NFAT1 activation and binding to the Ifng promoter (Aramburu et al., 1995). Within 2 h, expression of NFAT2 is increased, though its role in IFN-γ regulation is less clear (Aramburu et al., 1995), as is the role of NFAT in IFN-γ production by NKT cells (Wang et al., 2006). Multiple signaling pathways, including MAPKs, PI3 kinases, and calcium signaling, are able to activate the winged helix–loop–helix transcription factors of the Ets family. Ets family transcription factors are widely expressed and are important for developmental and cell fate choices (Yordy and Muise-Helmericks, 2000). Ets-1 is the active form in resting T cells, in which it can regulate transcription from Ets-dependent regulatory elements. Following T cell activation, Ets-1 becomes phosphorylated on four serine residues, which abrogates DNA binding, resulting in its rapid degradation (Yordy and Muise-Helmericks, 2000). Ets1/Rag-2/ chimeric mice have defects in thymocyte development and numbers of peripheral lymphocytes, implicating Ets-1 in T cell development (Barton et al., 1998; Bories et al., 1995; Muthusamy et al., 1995). Furthermore, while the proportion of peripheral CD4 and CD8 T cells are largely normal in Ets1/Rag-2/ chimeric mice, they show proliferative defects and increased apoptosis (Bories et al., 1995; Muthusamy et al., 1995) and greatly impaired cytokine expression (Grenningloh et al., 2005). Ets-1 contributes to increased STAT4 and IL-12Rb2 expression in developing Th1 cells, and is recruited to the Ifng promoter where it collaborates with T-bet to enhance IFN-γ expression (Grenningloh et al., 2005). Ets-1-deficient CD4 T cells proliferate normally in response to TCR stimulation but have dramatically impaired IFN-γ and IL-2 secretion. By contrast, the effect of Ets-1 deficiency on Th2 cytokine production is more limited, suggesting that Ets-1 is primarily a Th1-specific transcription factor. In good agreement with these data, while wild-type CD4 T ells induced colitis following transfer into severe combined immunodeficiency (SCID) recipients, Ets-1-deficient CD4 T cells did not (Grenningloh et al., 2005). Unlike CD4 T cells, which require Ets-1 transcription to activate the Ifng promoter and upregulate IL-12Rb2 and STAT4 during Th1 development, NK and NKT cells require Ets-1 for their development, proliferation, and survival (Bories et al., 1995; Muthusamy et al., 1995; Walunas et al., 2000). Ets-1-deficient mice show a drastic reduction of splenic NK cells, and those that do develop are severely impaired in their cytolytic activity and ability to secrete IFN-γ . In addition to Ets-1, the Ets related molecule (ERM), and the interferon response factor (IRF)-1 have been implicated in Th1 development or IFNg production by CD4 T cells. These factors are induced in response to IL12 in CD4 T cells (Ouyang et al., 1999). However, ERM only modestly enhances Ifng expression (Ouyang et al., 1999), whereas IRF-1-deficient mice have impaired Th1 responses in vivo (Lohoff et al., 1997). It is unclear if IRF-1 acts downstream of IL-12 to enhance IFN-γ production in T cells in addition to its role in facilitating IL-12 production by APCs. Tec kinases are a family of tyrosine kinases that are activated downstream of Src family kinases in response to signals from the TCR. TEC kinases in turn phosphorylate PLC-g, a second messenger necessary for sustained intracellular calcium release and activation of NFAT transcription factors (Schwartzberg et al., 2005). The primary TEC kinases in T lymphocytes and NK cells are Rlk (TXK in humans) and Itk, which contribute to Th1 and Th2 responses, respectively. Itk and Rlk are crucial for the proper expression of Eomes and the subsequent development of conventional CD8 T cells (Atherly et al., 2006; Berg, 2007; Broussard et al., 2006). On stimulation, TXK can translocate to the nucleus, bind to the human IFNG promoter (Kashiwakura et al., 1999; Takeba et al., 2002), and enhance the expression of IFNG promoter-driven reporters (Kashiwakura et al., 1999; Takeba et al., 2002). In naive mouse CD4 T cells, Rlk expression is extinguished on activation of naive cells but is reexpressed when these cells differentiate into Th1 effectors (Miller et al., 2004). However, Rlkdeficient mice have only minor defects in IFN-γ production or in their response to the Th1-inducing pathogen, T. gondii (Schaeffer et al., 1999, 2001), although overexpression of Rlk in vivo is associated with Th1 skewing (Takeno et al., 2004). The extent to which the effects of TXK/Rlk are mediated by interaction with the IFNG promoter, rather than by the activation of NFAT, are unclear.

 

6.2. STATs are activated in response to cytokine receptor signaling

STAT1 is activated by the binding of IFN-γ , IL-27, and type I IFNs to their receptors and appears to indirectly influence IFN-γ expression predominantly by potentiating the expression of the transcription factor, T-bet (Afkarian et al., 2002; Lighvani et al., 2001). STAT1 is the only STAT family member activated by IFN-γ , whereas IL-27 and type I IFNs also activate STAT3 (Darnell et al., 1994; Velichko et al., 2002). Despite the ability of multiple cytokine receptors to activate STAT1, only IFN-γR signaling is sufficient to induce sustained expression of T-bet. Doing so creates a positive feedback loop, allowing IFN-γ produced by Th1 cells to provide an autocrine signal back to the cell, thereby facilitating stable Th1 commitment (Lighvani et al., 2001). Given the prominent role of STAT1 in the induction of T-bet, it is surprising that Th1 responses develop in STAT1/ mice infected with lymphocytic choriomeningitis virus (LCMV) or T. gondii. This finding may reflect the copious amounts of type I IFNs or IL-12/IL-27 produced, respectively, in response to these infections (Lieberman et al., 2004; Nguyen et al., 2000). In the latter infection, STAT1-deficient CD4 and CD8 T cells expressed T-bet, albeit at reduced levels, which in the presence of high amounts of IL-12 were sufficient to induce functional Th1 immunity. Furthermore, in mice infected with murine cytomegalovirus (MCMV), STAT1 was required for NK cell expansion and cytotoxicity but not for IFN-γ production (Nguyen et al., 2002a). In contrast to STAT1, STAT4 is critical for Th1 lineage commitment and sustained Th1 responses in vivo and for antigen-independent induction of IFN-γ by IL-12 (Carter and Murphy, 1999; Kaplan et al., 1996; Mullen et al., 2001). However, STAT4/ T cells can produce low levels of IFN-γ in response to activation via the TCR (Afkarian et al., 2002), consistent with a role for STAT4 in amplifying rather than initiating IFN-γ production. Consistent with these in vitro findings, IFN-γ producing CD4 T cell responses develop, albeit more weakly, in STAT4- and IL12-deficient mice in response to infection with L. monocytogenes and certain viruses (Brombacher et al., 1999; Nguyen et al., 2002b; Oxenius et al., 1999; Schijns et al., 1998). Sustained Th1 immunity to T. gondii or L. major requires continual expression of IL-12p40; and Th1 immunity can be boosted in IL-12p35- and IL-12p40-deficient mice by provision of IL-12 in vivo (Park et al., 2000; Yap et al., 2000). Together, these data suggest that ablation either of IFN-γ signaling via STAT1 or of IL-12 signaling via STAT4 impairs, but does not abolish, IFN-γ production by CD4 T cells, but that both of these transcription factors are necessary for robust and sustained IFN-γ production in response to infection in vivo. By contrast to the importance of STAT4 for antigen-dependent and antigen-independent IFN-γ production by CD4 T cells, STAT4 is required in CD8 T cells only for antigen-independent induction of IFN-γ by IL-12 (Carter and Murphy, 1999). And while STAT4/ mice show deficiencies in NK cell cytolysis of target cells (Kaplan et al., 1996; Thierfelder et al., 1996) and are susceptible to infection by T. gondii (Cai et al., 2000), their NK cells do secrete IFN-γ , albeit at reduced levels compared to controls, in response to IL-2 and IL-18 in vitro (Cai et al., 2000) and infection with MCMV in vivo (Nguyen et al., 2002a). These findings suggest that additional STATs are able to compensate in part for lack of IL-12 signals. Signaling through the IL-2 and IL-15 receptors primarily activates STAT5a and STAT5b, which facilitate NK cell survival, cytolytic activity, and IFN-γ production (Fujii et al., 1998; Ma et al., 2006). The effects of IL-2 and IL-15 on IFN-γ in NK cells may be mediated in part by binding of STAT5a and STAT5b to the upstream regulatory element IfngCNS-6/ IFNGCNS-4 kb (Table 2.1), respectively, since disruption of this STAT5 binding site abrogated IL-2-dependent stimulation of IFNG reporter constructs in human NK-92 cells in vitro (Bream et al., 2004). This site is also responsive to STAT5 signaling downstream of CD2, a receptor expressed on the surface of human NK cells and T cells (Gonsky et al., 2004). Exposure to IL-2 has also been reported to activate STAT4 and the MAPKs MKK/ERK in human NK cells (Wang et al., 1999; Yu et al., 2000), consistent with the ability of IL-2 to prime for or induce low-level IFN-γ secretion. By contrast, another group identified a STAT5adependent mechanism of Th1 inhibition involving the induction of SOCS-3 (Takatori et al., 2005), which potently inhibited IL-12-induced STAT4 activation in this study. Ablation of STAT5a resulted in increased numbers of Th1 effector cells, particularly in Th2 conditions. Activation of STAT3 in T and NK cells occurs in response to a variety of cytokines, including IL-6, IL-12, IL-2, and IL-27 (Hibbert et al., 2003; Jacobson et al., 1995; Lucas et al., 2003; Sun et al., 2004b), of which IL-27 is the most potent (Hibbert et al., 2003). However, STAT3/ mice have no defect in Th1 differentiation, either in IFN-γ production or T-bet expression in CD4 T cells, but do have defects in Th17 development (Mathur et al., 2007; Yang et al., 2007).

 

6.3. T-box family members are crucial for IFN-γ secretion

T-bet, encoded by the Tbx21 gene, is the ‘‘master regulator’’ of Th1 development. Enforced expression of T-bet in the EL4 T cell line or in primary naive T cells markedly facilitates activation-induced IFN-g expression, and in developing Th2 cells impairs the expression of IL-4 and IL-5 while activating IFN-γ expression (Szabo et al., 2000). Conversely, T-bet-deficient CD4 T cells produce little or no IFN-g even when cultured in strong Th1 conditions. In addition, T-bet-deficient mice on a normally resistant C57BL/6 background are unable to mount an effective Th1 immune response to immunization with protein antigens in the presence of strong, Th1-inducing adjuvants or to infection with L. major or LCMV. These mice instead develop Th2 responses characterized by the production of IL-4 and IL-5 (Szabo et al., 2002) These findings suggest that in addition to inducing Th1 development, T-bet is necessary to block Th2 development, which it does so by binding with Runx3 to the Il4 silencer and by blocking the interaction of GATA-3 with its target DNA sequences in the Il4 30 enhancer (Djuretic et al., 2007; Usui et al., 2006). The interaction of T-bet with GATA-3 requires phosphorylation of Tyr525 on T-bet by the kinase Itk. However, phosphorylation of Tyr525 is not essential for T-bet to inhibit Th2 differentiation or to induce Th1 differentiation (Hwang et al., 2005). T-bet acts directly on the Ifng gene to facilitate its expression and does so by binding to multiple sites within the Ifng promoter and within other distal regulatory elements located upstream and downstream of the gene (Table 2.1) (Chang and Aune, 2005; Cho et al., 2003; Hatton et al., 2006; Lee et al., 2004; Shnyreva et al., 2004). T-bet works in concert with at least two other factors to facilitate IFN-g expression and Th1 immunity: Runx proteins and Hlx. Runx3 and Hlx are direct targets of T-bet, which induce their expression, enabling them to work cooperatively in a feed-forward loop that reinforces Th1 commitment (Singh and Pongubala, 2006). Runx transcription factors are involved in many cellular differentiation processes and have been implicated in CD4 expression and Th1 commitment in T cells. Overexpression of Runx1 in naive CD4 T cells induces Th1 differentiation, whereas expression of a dominant-negative Runx1 favors Th2 differentiation (Komine et al., 2003). Runx3, which is highly expressed in Th1 and CD8 T cells, interacts with T-bet and binds cooperatively with it to the Ifng promoter to activate its transcription, and to the Il4 silencer, to repress its transcription (Djuretic et al., 2007).

The homeobox gene Hlx is expressed by Th1 cells and works synergistically with T-bet to increase the frequency of IFN-g-producing cells and amount of IFN-g produced, as well as to enhance the expression of T-bet and IL-12Rb2 (Mullen et al., 2002). Consistent with this notion, Hlx transgenic mice express less IL-4Ra and generate increased frequencies of IFN-γ -producing cells under Th2-polarizing conditions than controls (Zheng et al., 2004). In contrast, Hlx haploinsufficient CD4 T cells have elevated expression of IL-4Ra and a Th2 bias (Mikhalkevich et al., 2006). While T-bet is sufficient for IFN-γ induction in CD4 T cells, IFN-g production in CD8 T cells is induced through the concerted action of T-bet and its paralog, Eomes; together they collaborate to assure proper effector differentiation and maintenance of memory CD8 T cells (Pearce et al., 2003). Eomes is expressed in naive and memory CD8 T cells and is important for induction of IFN-γ in these cells. As a consequence, IFN-γ production in response to TCR stimulation is normal in T-bet-deficient CD8 T cells (Szabo et al., 2002). By contrast, IFN-γ expression by CD8 T cells is greatly impaired in the absence of Eomes, and this can be further exacerbated by loss of T-bet, suggesting independent but overlapping roles of these two transcription factors in IFN-γ production (Intlekofer et al., 2005). Similar to CD4 Th1 cells, T-bet expression in CD8 T cells is increased by IL-12 in response to infection, whereas clearance of infection and loss of IL-12 signals increases Eomes expression (Intlekofer et al., 2005; Sullivan et al., 2003). Optimal expansion of effector CD8 T cells requires IL-12-dependent induction of T-bet, as IL-12Ra–/– mice maintain high levels of Eomes through primary infection and have reduced numbers of effector cells (Takemoto et al., 2006). Although T-bet deficiency does not affect cytokine production by CD8 T cells in response to in vitro stimulation with mitogens, antigen-specific IFN-γ production by T-betdeficient CTLs following LCMV infection is impaired, suggesting T-bet functions to regulate IFN-γ production or to sustain IFN-γ -producing effector CD8 T cells in vivo (Sullivan et al., 2003). Perhaps consistent with the latter possibility, Eomes, which is highly expressed by memory CD8 T cell populations in both humans and mice, induces expression of the high-affinity IL-15Ra chain (Intlekofer et al., 2005). Together, these data indicate that in CD8 T cells, expression of genes associated with effector function and migration are initiated by Eomes and augmented by T-bet, whereas Eomes functions to induce genes associated with self-renewal and homeostasis. NK cells are poised to rapidly produce IFN-g immediately after activation, which is enabled by their constitutive expression of Ifng mRNA (Stetson et al., 2003; Tato et al., 2004). Consistent with this finding, T-bet and Eomes are constitutively expressed in NK cells (Szabo et al., 2002), and the Ifng promoter in these cells displays epigenetic marks indicative of transcriptionally permissive chromatin (Tato et al., 2004). In the absence of T-bet, NK cells have dramatically decreased IFN-γ production and cytolytic activity (Szabo et al., 2002). Surprisingly, T-bet and Eomes are also required to maintain the expression of receptors necessary for NK cell development (Intlekofer et al., 2005; Townsend et al., 2004). Eomes is required for expression of the IL-15Ra, Eomes heterozygous mice show substantially reduced frequencies of NK cells compared to wild-type mice, and this defect can be further exacerbated by the loss of one T-bet allele (Intlekofer et al., 2005). While CD8 and NK cells express both Eomes and T-bet, only T-bet is expressed by NKT cells and is crucial to their development. T-bet-deficient mice are nearly devoid of NKT cells (Intlekoferet al., 2005; Townsend et al., 2004), likely due to the lack of IL-2R/IL-15Rb (CD122), which is required for IL-15-induced proliferation and survival of developing NKT cells (Matsuda and Gapin, 2005). Ectopic expression of T-bet in developing NKT cells results in increased expression of select genes associated with their maturation, but expression of IFN-γ , granzyme B and perforin are only modestly enhanced (Matsuda et al., 2006). It is unknown how T-bet is induced during NKT cell differentiation, as IFN-γ /, IFN-γ R/, and STAT1/ mice show no defects in NKT cell development or survival (Townsend et al., 2004).

 

 

7. EPIGENETIC PROCESSES GOVERN PLASTICITY OF CELL FATE CHOICES AND HELP TO IDENTIFY DISTAL REGULATORY ELEMENTS

In principle, the transcription factors that function as ‘‘master regulators’’ of T and NK cell effector function could be both necessary and sufficient for the initiation and faithful propagation of their respective effector lineages. In practice, transcription factors must bind to their recognition sites within regulatory elements and then recruit general transcriptional activators or repressors to regulate gene transcription. The ability of NK and NKT cells to rapidly produce substantial amounts of IFN-γ on stimulation derives from their constitutive expression of Eomes and T-bet (NK cells) or of T-bet alone (NKT cells), and the fact that the regulatory elements to which these and other transcription factors must bind in the Ifng locus are contained within accessible chromatin, thereby facilitating activation of Ifng transcription (Bream et al., 2004; Chang and Aune, 2005; Stetson et al., 2003). This is not the case for naive CD4 T cells, which can produce only small amounts of IFN-γ immediately following TCR stimulation and require multiple rounds of cell division under appropriate conditions to gain the capacity to produce IFN-γ efficiently (Mullen et al., 2001). This lag reflects the time required for these cells to induce the expression of T-bet, for T-bet to induce the expression of its partners Runx3 and Hlx, and for these factors together to induce Ifng expression, to modify the chromatin of the Ifng locus to make it more accessible, and to induce IL-12Rb2 and IL-18R. Naive CD8 T cells are more efficient in gaining the capacity to produce substantial amounts of IFN-γ than naive CD4 T cells, reflecting, at least in part, their constitutive expression of Eomes. Thus, the ability of T-bet and Eomes to induce IFN-γ expression and enforce the Th1 and CTL effector fates appears to derive not only from the ability of these transcription factors to initiate the transcription of Ifng and other target genes, but also from the ability of these transcription factors to induce epigenetic modifications within the Ifng gene to assure heritability of expression thereafter. Eukaryotic cells face the challenge of fitting a genome consisting of several billion nucleotides into a nucleus only microns in diameter, while maintaining spatial organization and accessibility to factors that govern the transcription, repair, and replication processes. This is achieved by the association of DNA with proteins in chromatin, the basic unit of which is the nucleosome, consisting of an octamer of histone proteins with two copies each of H2A, H2B, H3, and H4, around which 150 bp of DNA is wrapped (Felsenfeld and Groudine, 2003). Differences in nucleosome composition, relative position and interactions with DNA, posttranslational modifications to histone tails, and methylation of cytosines within CpG dinucleotides of the DNA itself encode information without affecting the underlying DNA sequence, and constitute the epigenetic code of that cell. The epigenetic code influences transcription factor binding, transcription initiation, and progression, and thus, along with transcription factor abundance and activity, determines in a given cell whether a gene is or can be expressed and to what degree. Unlike the fixed information encoded by DNA sequence, epigenetic information is plastic but potentially heritable, and thus can be propagated from parental cells to daughter cells in a manner conducive to maintaining the overall program of gene expression that characterizes a specific cell type. These epigenetic modifications can be detected by several techniques, and because they are often targeted to regulatory elements, these techniques are one means by which to search for cis-regulatory elements (Nardone et al., 2004; Wilson and Merkenschlager, 2006). Hypersensitivity of specific sequences to digestion by low DNase I concentrations indicates specific sites where nucleosomes have been displaced or their conformation altered, either reflecting or facilitating the binding of transcription factors to important regulatory elements. Modification of histone tails can provide a signal for the binding of protein complexes associated with transcriptional activation or silencing. Among the histone modifications associated with permissiveness to gene transcription are acetylation of lysines of histone H3 (AcH3) or H4 (AcH4) and di- or trimethylation of histone H3 lysine 4 (H3-K4me2 or H3-K4me3). In contrast, di- or trimethylation of H3 lysine 27 (H3-K27me2/me3) or lysine 9 (H3-K9me2/me3) are repressive marks characteristic of Polycomb-mediated repression and heterochromatic silencing, respectively (Kouzarides, 2007). DNA methylation is mediated by DNA methyltransferases and is linked to the formation of repressive chromatin through the recruitment of proteins that mediate transcriptionally silent histone modifications and ATP-dependent nucleosomal remodeling (Vire et al., 2006). These epigenetic modifications often are found together in varying degrees; the extent of gene expression or potential for gene expression is influenced by the combined contribution of multiple epigenetic processes working in concert with transcription factors and regulatory elements to activate or inhibit gene expression. Changes in cytokine expression during Th1 and Th2 commitment by CD4 T cells are associated with changes in the epigenetic modifications at the Th2 cytokine locus and at the Ifng gene (Ansel et al., 2003, 2006; Lee et al., 2006; Wilson and Merkenschlager, 2006). In general, both the Il4/Il13/Il5 and Ifng loci are poised in naive CD4 cells containing epigenetic modifications indicative both of transcriptional repression and permissiveness. During Th1 development, the Ifng locus gains transcriptionally favorable histone modifications, loses CpG methylation, and thereby gains accessibility to binding of transcriptional activators, while the Il4/Il13/Il5 locus gains repressive modifications including H3-K27me3. Selected areas of the Ifng locus maintain accessibility in naive and Th2 cells, suggesting that these regions may play a role in preserving locus plasticity in these cell types.

 

8. TRANSCRIPTIONAL REGULATORY ELEMENTS WITHIN THE IFNG GENE

We will begin by reviewing the promoter and intronic regulatory elements, the transcription factors binding to them, their functions, and the epigenetic modifications to these regions that are typical of naive and effector T cells and NK cells. Then in the following sections, we will discuss newer information on the structure of the extended Ifng locus, distal regulatory elements within the locus, and what is known regarding their functions.

 

8.1. Regulatory elements within the Ifng promoter and gene

The Ifng gene is composed of four exons spanning 5.5 kb in humans and rodents, upstream of which is its core 600 bp promoter. The binding of transcription factors and the subsequent epigenetic and chromatin changes within the Ifng gene have been studied in some detail. These studies together with a number of functional assays have defined the Ifng promoter and two intronic enhancers (Fig. 2.1). The Ifng promoter directs Th1-specific expression of transgenic reporter constructs in vivo (Young et al., 1989; Zhu et al., 2001), which is consistent with the presence of multiple binding sites for T-bet in the promoter (Cho et al., 2003) and the demonstration of T-bet binding to the promoter in Th1 cells by ChIP (Beima et al., 2006; Chang and Aune, 2005; Hatton et al., 2006; Shnyreva et al., 2004). In transgenic reporter assays, addition of introns 1 and 3 enhances expression from the Ifng promoter but abolishes Th1-specificity by as yet unknown mechanisms (Young et al., 1989; Zhu et al., 2001). In addition to T-bet, a number of other transcription factors can bind to the Ifng promoter and/or introns 1 and 3 and transactivate reporter constructs containing these elements, including AP-1, ATF-2/c-Jun, Ets-1, NFkB, NFAT, STATs, and T-bet (reviewed in Murphy et al., 2000) (Barbulescu et al., 1998; Glimcher et al., 2004; Persky et al., 2005; Szabo et al., 2003). Conversely, CREB can bind to the promoter and inhibit expression (Penix et al., 1996), as can GATA-3; repression by GATA-3 may be dependent on its ability to interact with and inhibit the function of T-bet and STAT4 (Hwang et al., 2005; Kaminuma et al., 2004) and/or to recruit the H3K27-methyltransferase and Polycomb protein EZH2 to Ifng (Chang and Aune, 2007). YY1, a ubiquitously expressed transcription factor has been reported to inhibit or facilitate expression driven by the Ifng promoter (Sweetser et al., 1998; Ye et al., 1996). YY1 and other transcription factors are also likely to mediate chromatin remodeling at the Ifng promoter. As noted above, naive T cells are poised to express low levels of Ifng mRNA shortly after activation (Grogan and Locksley, 2002; Mullen et al., 2002). In these cells, the Ifng promoter and introns lack transcriptionally favorable or repressive histone marks, and CpG dinucleotides in the introns are heavily methylated (Agarwal and Rao, 1998; Avni et al., 2002; Fields et al., 2002; Mullen et al., 2002). Naive T cells also lack the DNase I hypersensitive sites in the Ifng promoter (HS1) and introns 1 (HSII) and 3 (HSIII), which are characteristic of Th1 cells (Agarwal and Rao, 1998). However, the Ifng promoter is demethylated in naive T cells and is juxtaposed to the Th2 cytokine locus, perhaps creating a hub that allows Ifng and Il4 to compete for limiting amounts of transcription activators present in naive T cells prior to lineage specification (Spilianakis et al., 2005). Initial activation of naive CD4 T cells results in the acetylation of histones H3 and H4 in the Ifng promoter under Th0, Th1, and Th2 conditions, but sustained acetylation and extension of acetylation further downstream into the gene is Th1 specific and requires STAT4 and T-bet (Avni et al., 2002; Chang and Aune, 2005; Fields et al., 2002; Mullen et al., 2001). Similarly, T-bet in concert with Hlx induces Th1-specific DNA demethylation that extends from the promoter into the intronic regions of the gene and also induces DNase HSII and HSIII in introns 1 and 3 (Agarwal and Rao, 1998). These actions are likely mediated in part by the binding of T-bet to a T-box half-site immediately upstream of a C/EBP-AP1-ATF site; together these transcription factors cause the dissociation of the mSin3a repressor, which is associated with histone deacetylase activity (Tong et al., 2005). As noted above, Runx3 also binds cooperatively with T-bet to the Ifng promoter (Djuretic et al., 2007). Th1 differentiation also results in nucleosome remodeling at the Ifng promoter, as demonstrated by the formation of DNase HSI. This remodeling utilizes the ATP-dependent nucleosome remodeling complex Swi-SNF, including the subunit Brg-1, which is recruited to the Ifng promoter in a STAT-4 dependent, Th1-dependent manner and likely facilitates transcription factor binding and/or the transit of RNA polymerase containing complexes required for efficient transcription (Zhang and Boothby, 2006). Th2 cells, as well as cells that are incapable of producing IFN-γ, such as fibroblasts and hepatocytes, exhibit extensive CpG methylation within 200 bp of the Ifng promoter (Schoenborn et al., 2007; Winders et al., 2004). However, enforced expression of T-bet into terminally differentiated Th2 cells induces expression of IFN-γ (Szabo et al., 2000). This may be due, in part, to the ability of T-bet to bind to the Ifng promoter even when it is methylated (Tong et al., 2005). These data demonstrate a mechanism by which forced expression of T-bet in Th2 cells can override repressive epigenetic modification to induce Ifng expression. Consistent with the constitutive transcription of Ifng by NK cells and their ability to rapidly produce large amounts of this cytokine, the Ifng gene in NK cells displays permissive epigenetic marks similar to those found in Th1 cells; DNase hypersensitive sites are present, histones are acetylated, and intronic as well as promoter CpGs are demethylated (Chang and Aune, 2007; Tato et al., 2004).

 

8.2. Identification of candidate distal regulatory elements in the Ifng locus

In addition to the promoter, proper expression of mammalian genes is dependent on other regulatory elements that may be located in the introns or in upstream and downstream flanking sequences. These regulatory elements may include enhancers, silencers, boundary elements, and locus control regions, which can influence locus accessibility and the initiation and maintenance of gene expression. Such elements are most commonly located within 50–75 kb on either side of the gene they regulate, but may be located up to hundreds of kb away. Thus, to fully understand the regulation of a gene such as Ifng one must identify each of its regulatory elements, characterize the function of these elements, and identify and understand the actions of the transcription factors that bind to them. As described above, Ifng is regulated in part through its promoter and intronic regulatory elements, but these regions are not sufficient for proper control (Young et al., 1989; Zhu et al., 2001) (Soutto et al., 2002; Young et al., 1989). By contrast, a 191 kb bacterial artificial chromosome (BAC) containing the human IFNG gene and 90 kb of upstream and downstream flanking sequences resulted in high level, CD8- and Th1-specific IFN-γ production (Soutto et al., 2002), suggesting that distal transcriptional regulatory elements are required for proper expression, the essential elements are present within this extended region, and elements from the human IFNG locus function properly in mice. To identify additional regulatory elements, our lab and other laboratories have utilized evolutionary conservation to distinguish two conserved noncoding sequences (CNS) 6 kb upstream (known as CNS1 or IfngCNS-6) and 18–20 kb downstream (referred to as CNS2 or IfngCNSþ 18–20) from the mouse Ifng promoter (Fig. 2.1); the region analogous to IfngCNS-6 in mice is located 4 kb upstream in humans (IFNGCNS-4 kb) (Bream et al., 2004). These CNS elements were subsequently shown to contain Th1-restricted permissive histone marks and DNase hypersensitive sites and to enhance IFN-γ expression in reporter assays (Bream et al., 2004; Lee et al., 2004; Shnyreva et al., 2004). Another group subsequently identified additional CNSs, some of which were enriched in acetylated histones in Th1 effectors and/or NK cells, extending >50 kb upstream and downstream of Ifng (Chang and Aune, 2005), suggesting that additional regulatory elements were present within this region. Unlike the Th2 cytokine locus, which consists of three coordinately expressed cytokine genes and a housekeeping gene, the extended Ifng locus contains up to two alternately expressed cytokine genes, a housekeeping gene, and, in rodents, a series of genomic aberrations. In humans and rodents, the Il22 and Mdm1 genes are located upstream of Ifng in the same transcriptional orientation (Fig. 2.2). Mdm1, a housekeeping gene, is expressed ubiquitously. Il22 encodes a proinflammatory member of the IL-10 cytokine family (Wolk et al., 2004), which is most highly expressed by activated Th17 CD4 T cells (Liang et al., 2006; Zheng et al., 2007). The distance between Il22 and Ifng is much greater in the mouse than human genome due to the presence of a complex set of structural rearrangements and segmental duplications located between these genes in mice. This region includes the Iltifb paralog, an inverted duplication of the Il22 gene in which a portion of the promoter has been lost precluding its expression (Dumoutier et al., 2000), and is flanked on either side by six highly conserved short tandem sequence duplications. Iltifb is present in C57BL/6 and 129 strain mice but not in the BALB/c or DBA/2 strains nor in the rat. The gene encoding IL26, also a member of the IL-10 cytokine family, is located between IL22 and IFNG in humans. Orthologs of IL26 are found in all vertebrates for which information is available but is not present in rodents (Igawa et al., 2006), suggesting that the locus order found in humans (MDM1 ! IL22 ! IL26 ! IFNG) is ancestral. Downstream of Ifng, no known coding genes are found for >500 kb, but a noncoding antisense transcript, Tmevpg1, extends to within 100 kb of the Ifng start site. Tmevpg1 is expressed by na¨ıve CD4, CD8, and NK cells and is downregulated on activation, suggesting a potential role in Ifng gene regulation (Vigneau et al., 2001, 2003). It is unknown if there are differences in Tmevpg1 expression among mouse strains or Th1, Th2, or Th17 cells. Despite the striking differences in genomic structure in the Ifng locus between rodents and humans, the pattern of Ifng expression is substantially similar in these species, suggesting that most, if not all, regulatory elements needed for proper expression are proximal to the region where synteny between rodents and other vertebrate species is lost. Based in part on this notion, several groups including ours have utilized bioinformatics and various experimental approaches to identify candidate regulatory elements extending over 120 kb surrounding the Ifng locus (Chang and Aune, 2005; Hatton et al., 2006; Schoenborn et al., 2007). Eight CNS elements were identified based on their exhibiting >70% homology between human and C57BL/6 mouse for 100 or more base pairs. Seven of these eight CNSs exhibit differential epigenetic modifications in naive, Th1, Th2, and NK cells that suggested they might participate in the cis-regulation of Ifng (Chang and Aune, 2005, 2007; Hatton et al., 2006; Schoenborn et al., 2007). In naive CD4 and CD8 T cells, these CNSs (like the Ifng promoter and introns) lack the transcriptionally favorable histone modifications H3-K4me2 and Ac H4 with two exceptions—low levels of H3-K4me2 are present at two conserved regions termed IfngCNS-34 and IfngCNS-22 based on their distance in kilobases from the Ifng promoter (Fig. 2.3) (Hatton et al., 2006; Schoenborn et al., 2007). Conversely, naive CD4 T cells have moderate levels of the repressive histone modification H3-K27me3 in the region between IfngCNSþ 29 and IfngCNSþ 46 (Schoenborn et al., 2007). Interestingly, in naive CD4 T cells these four CNSs lack CpG methylation, and two of them, IfngCNS-34 and IfngCNSþ 46, have weak HS sites (hs-35 and hsþ 49/53) nearby but not directly associated with them, as mapped by a new high-resolution method (Schoenborn et al., 2007). Together, the modifications found at these CNSs in naive T cells suggest that they may play an early role in the initiation of Ifng activation or repression during CD4 T cell functional differentiation. Th1 differentiation is associated with the acquisition of DNase HS sites at IfngCNS-22, IfngCNS-6, IfngCNSþ 18, and IfngCNSþ 29, and discrete peaks of enrichment for H3-K4me2, AcH3, and AcH4 at these CNSs and at IfngCNS-54 (Fig. 2.3) (Hatton et al., 2006; Lee et al., 2004; Schoenborn et al., 2007; Shnyreva et al., 2004). IfngCNS-54, IfngCNS-6, and IfngCNSþ 18–20, which have methylated CpGs in naive T cells, become demethylated during Th1 differentiation in parallel with a complete loss of H3-K27me3 in the Ifng locus (Schoenborn et al., 2007). Collectively, these changes result in heightened transcriptional accessibility within the Ifng locus. Perhaps counter intuitively, the Ifng promoter and regions upstream and downstream of the gene, with the exception of the region from IfngCNS-34 and IfngCNS-22, appear to gain and retain, at least for several days, the typically repressive H3K9me2 mark (Chang and Aune, 2007), which here is associated with active transcription as it sometimes is (Vakoc et al., 2005). The gain in histone acetylation is largely STAT4- and T-bet-dependent (Chang and Aune, 2005), and the gain of H3-K4me2 and loss of H3-K27me3 are largely T-bet-dependent (our unpublished observations). Similarly, as naive CD8 T cells differentiate into CTL effectors they gain histone acetylation and H3-K4me2 at conserved elements (Zhou et al., PNAS 2004 and our unpublished observations). Consistent with their constitutive expression of Ifng mRNA, freshly isolated NK cells have modest amounts of AcH4 not only within the Ifng gene but also near IfngCNS-22 and between IfngCNSþ 29 and IfngCNSþ 46 (Chang and Aune, 2005). AcH4 increases at these sites and AcH4 is acquired at IfngCNS-6 and at 50 kb downstream of the gene when NK cells are cultured in IL-2 (Chang and Aune, 2005). In these studies, histone modifications were not examined at high resolution throughout the Ifng locus; thus these modifications may be present at additional sites as well. Together, these data suggest that in IFN-γ-producing Th1 and CD8 effector T cells and in NK cells, multiple regulatory elements are made more accessible as a means to facilitate Ifng expression. This is likely to be true in NKT cells as well, but, to our knowledge, studies to test this possibility have not been done. Conversely, Th2 effector CD4 T cells must silence the expression of Ifng, which they do both by altering the expression of key transcription factors and the epigenetic modifications in the Ifng locus. In Th2 CD4 T cells, the Ifng locus is marked by CpG methylation of the Ifng promoter and absence of H3-K4me2, with the exception of modest enrichment at IfngCNS-34 and IfngCNS-22 similar to that seen in naive CD4 T cells (Fig. 2.3) (Hatton et al., 2006; Schoenborn et al., 2007). Furthermore, the Ifng locus in Th2 CD4 T cells is extensively marked by the repressive histone modifications H3-K27me2/me3 (Chang and Aune, 2007; Schoenborn et al., 2007). While T-bet is crucial for inducing IFN-γ expression, GATA-3 acts to silence Ifng. In developing Th2 effector T cells, GATA-3 can bind to the promoter and intron 1 of Ifng, as well as IfngCNS-54, and forced expression of GATA-3 in developing Th1 cells inhibits Ifng transcription and results in the recruitment of the H3K27 methyltransferase EZH2 to the Ifng locus and the acquisition of H3-K27me2 and loss of the H3-K9me2, such that H3-K9me2 is only transiently present in the Ifng locus in Th2 conditions (Chang and Aune, 2007). Since the expression of Ifng is markedly repressed by day 7–8 in Th2 cells, the finding that four distal elements (IfngCNS-34, IfngCNS-22, IfngCNSþ 29, IfngCNSþ 46) remain completely demethylated during Th2 differentiation suggests that CpG methylation of these elements is not required for the initial repression of Ifng, though it cannot be excluded that these elements may gain CpG methylation during extended culture. Remarkably, Th2 cells acquire a number of HS sites, including HSIII in intron 3 and HS-22 within IfngCNS-22 that are also found in Th1 but not in naive CD4 T cells, as well as four strong Th2-specific HS sites located at HS-40, HS-35, HSþ 8, and HSþ 26 (Fig. 2.3). In addition, HSþ 49 is strongest in Th2 cells, intermediate in naive CD4 T cells and just above background in Th1 cells, and a number of weaker sites located throughout the Ifng locus are found exclusively in Th2 effector T cells. While the HS sites found in Th1 cells are located within conserved sequences, nearly all of the Th2-specific HS sites are adjacent to conserved sequences. The location of these Th2-specific HS sites adjacent to CNSs may reflect nucleosome sliding or displacement from nearby clustered regulatory elements that are employed differentially in Th1 cells versus Th2 and naive CD4 T cells.