Vol. 14, No. 14, pp. 1693-1711, July 15, 2000

REVIEW
Lineage commitment in the immune system: the T helper lymphocyte grows up

¹ Department of Immunology and Infectious Diseases, Harvard School of Public Health, Boston, Massachusetts 02115-6017 USA; ² Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115 USA; ³ Department of Pathology, Washington University School of Medicine, St. Louis, Missouri 63110 USA

	Article

Top Article References

Cells of the immune system provide particularly fruitful subjects for the study of lineage commitment. Both T and B lymphocytes undergo complicated patterns of differentiation from uncommitted, nonfunctional precursor cells to highly sophisticated effector cells. The development of the helper T lymphocyte is one of the most elegant examples of this. A little over a decade ago, Mosmann and Coffman (1989) discovered that naive mouse CD4⁺ T helper lymphocytes, upon receiving an antigenic stimulus, differentiate into two distinct subsets defined both by their function and by unique cytokine profiles. These subsets, T helper 1 (Th1) and T helper 2 (Th2) (Mosmann et al. 1986; Mosmann and Coffman 1989; Paul and Seder 1994; O'Garra 1998; Rengarajan and Glimcher 2000), are responsible for cell-mediated/inflammatory immunity and humoral responses, respectively (Fig. 1). This division of labor fits nicely with previous demonstrations that an organism tends to mount either a cell-mediated or humoral response, but not both, in response to pathogens. The function of T helper cells can largely be explained by the cytokines they secrete. Cytokines (or lymphokines) are small hormone-like polypeptides that have pleiotrophic biological activities in several cell types. Resting T cells do not transcribe cytokine genes, but they are rapidly induced upon coactivation through the T-cell receptor (TCR) and costimulatory receptors (Lenschow et al. 1996). Much progress has been made in identifying the signaling pathways and transcription factors that control Th1 and Th2 differentiation as shown schematically (Fig. 2a). This review will summarize what is currently known about the signals that regulate lineage commitment in T helper cells with a special focus on three subset-specific transcription factors, T-bet, GATA-3, and c-Maf, responsible for lineage commitment (Fig. 2b).

View larger version (28K): [in this window] [in a new window]   Figure 1.   Signals that influence TH differentiation.

View larger version (76K): [in this window] [in a new window]   Figure 2.   (A)  Transcription factors that influence Th differentiation. (B) Tissue-specific factors that regulate CD4+ T helper cell differentiation.

Th cell cytokines, phenotype, and function

The functional differences between the Th subsets can largely be explained by the activities of the subset-specific cytokines produced. The hallmark cytokine of Th1 cells is interferon- (IFN), and Th1 cells also produce IL-2, TNF, and LT, cytokines that mediate delayed type hypersensitivity responses and macrophage activation. The signature cytokine of Th2 cells is interleukin-4 (IL-4), and Th2 cells also secrete IL-5, IL-9, IL-10, and IL-13, cytokines that provide help to B cells and are critical in the allergic response (Arthur and Mason 1986; Paliard et al. 1988; Mosmann and Coffman 1989; Paul and Seder 1994). This paradigm extends to other species including human where clearcut human Th1 and Th2 clones have been generated (Romagnani 1992, 1994). However, simultaneous production of IL-2, IL-4, and IFN can be observed in human helper cells (Paliard et al. 1988).

A host of surface antigens that are unique to, or preferentially expressed on, Th1 or Th2 cells have recently been identified. Th1-preferentially expressed genes include the IFN receptor- chain, IL-12 receptor  chain, IL-18 receptor, the P-selectin glycoprotein ligand-1, and the CXCR3 and CCR5 chemokine receptors (Davis et al. 1984; Bach et al. 1995; Borges et al. 1997; Szabo et al. 1997a; Bonecchi et al. 1998; Sallusto et al. 1998; Xu et al. 1998b). Markers preferentially expressed on Th2 cells include a novel IL-1-like molecule T1/ST2 and the chemokine receptors CCR3 (eotaxin receptor), CCR4, and CCR8 (Bonecchi et al. 1998; D'Ambrosio et al. 1998; Lohning et al. 1998; Sallusto et al. 1998; Xu et al. 1998a; Zingoni et al. 1998; Coyle et al. 1999; Hoshino et al. 1999; Townsend et al. 2000). T1/ST2 is a stable marker for Th2 cells in vivo that may be functionally important in the generation of Th2 responses. Antibodies against T1/ST2 attenuated eosinophilic inflammation in airways and suppressed Th2 cytokines in vivo (Lohning et al. 1998), and mice lacking T1/ST2 had impaired Th2 responses to Schistosoma mansoni. The preferential expression of these markers on effector Th1 and Th2 cells may be the consequence of ligand-induced events or of the cytokine milieu because surface antigens that will distinguish Th cells in the early stages of Th1 or Th2 differentiation have not yet been found. The best way to phenotypically distinguish the two subsets is by their unique cytokine profiles as determined by cellular staining for IFN and IL-4 or by a powerful new technique that allows detection of surface IFN and IL-4 (Ouyang et al. 2000). We and others have also generated transgenic "knock-in" mice at the IL-4 locus (Riviere et al. 1998; Ho et al. 1999) that may prove useful in marking early Th2 cells.

Factors that influence lineage commitment

After differentiation and emigration from the thymus to the peripheral immune organs, CD4⁺ T helper cells are termed naive T helper precursor (Thp) cells. Thp cells are functionally immature and capable of secreting only IL-2. Much has been learned about the signals that drive these naive Thp cells, which secrete only IL-2, to become Th1 or Th2 effector cells. Th1 and Th2 cells appear to derive from a common precursor that expresses the IL-4 gene (Kamogawa et al. 1993). Thp activation and differentiation in the periphery requires at least two separate signals. The first signal is delivered by the TCR/CD3, after its interaction with antigen/MHC on antigen presenting cells (APCs). The second signal is produced by a number of costimulatory or accessory molecules typified by the CD28/B7, OX40, and LFA-1/ICAM receptor-ligand pairs. Whether an immune response will be dominated by Th1 versus Th2 CD4⁺ cells is clearly influenced by the nature of these two signals (Seder and Paul 1994). However, neither of these two signals is as potent a determinant of Th cell fate as the cytokine milieu itself.

APC, antigen, costimulation  Signal one is delivered by interaction of the TCR on a naive Thp cell with MHC/antigen on an APC such as a dendritic cell. The recent identification of phenotypically distinct subsets of dendritic cells has revealed unique functions for these subsets in the development of Th cells (Maldonado-Lopez et al. 1999; Pulendran et al. 1999; Rissoan et al. 1999; Smith and Fazekas de St. Groth 1999). In the mouse, CD8⁺ lymphoid-like dendritic cells produce IL-12 and preferentially stimulate Th1 differentiation. Recently, a T-cell secreted factor called Eta1/osteopontin has been shown to drive Th1 differentiation by inducing IL-12 and inhibiting IL-10 secretion from APC (Ashkar et al. 2000). Eta1-deficient mice fail to mount DTH responses to pathogens such as herpes simplex keratitis and Listeria monocytogenes. A second subset of myeloid-like CD8a DC have been shown to stimulate Th2 differentiation. The mechanism by which this occurs remains unclear because the APC counterpart cytokine of IL-12 for Th2 differentiation is not known, although IL-6 has been suggested as a candidate (Rincon et al. 1997). An intriguing possibility is that specific chemokines may polarize Thp cells, as evidenced by a recent report that MCP-1 deficient mice fail to mount Th2 responses (Gu et al. 2000).

Cytokines  The cytokines themselves play the most critical role in T helper cell polarization. The two critical cytokines that control Th1 and Th2 differentiation are IL-12 and IL-4, respectively. These two cytokines enhance the generation of their own Th subset and simultaneously inhibit the generation of the opposing subset (Scott 1991; Maggi et al. 1992; Parronchi et al. 1992; Hsieh et al. 1993; Macatonia et al. 1993; Manetti et al. 1993; Powrie and Coffman 1993; Seder et al. 1993; Trinchieri 1993; Wu et al. 1993), events that likely occur at a precursor stage (Thp) because once established, Th1 and Th2 cells cannot revert (Scott 1991; Murphy et al. 1996). The requirement for these two cytokines has been demonstrated unequivocally by the phenotype of mice that lack these cytokines, cytokine receptors, or the effector molecules downstream of the receptors. IL-12, secreted by APC, activates the Stat4 signaling pathway, and mice lacking IL-12 or Stat4 do not have Th1 cells (Kaplan et al. 1996a; Magram et al. 1996; Thierfelder et al. 1996). Two other cytokines that influence Th1 development are IL-18, whose receptor is related to the IL-1 receptor family (Cerretti et al. 1992; Okamura et al. 1995; Gu et al. 1997), and IFN (Scott 1991; Meraz et al. 1996). IFN activates the Stat1 pathway (Meraz et al. 1996). Mice lacking IL-18 or Stat1 have defective in vivo Th1 responses (Meraz et al. 1996; Takeda et al. 1998). In contrast, mice that lack IL-4, IL-4 receptor, or Stat6, the downstream signaling molecule for the IL-4 receptor, fail to develop Th2 cells in response to most stimuli (Kuhn et al. 1991; Kopf et al. 1993; Kaplan et al. 1996b; Shimoda et al. 1996; Takeda et al. 1996; Noben-Trauth et al. 1997). The biologic function of another cytokine, IL-13, partially overlaps with IL-4 because, in some instances, IL-13 drives Th2 development and IgE synthesis in an IL-4-independent fashion (Minty et al. 1993; Punnonen et al. 1993; Barner et al. 1998; Cohn et al. 1998; Emson et al. 1998; McKenzie et al. 1998). IL-13 is especially important in the asthmatic response (Wills-Karp et al. 1998).

Development of the Th1 lineage

Signaling pathways  Signaling pathways emanating from both the TCR and the IL-12 receptor drive Th1 differentiation at least in part by inducing the expression of the Th1-restricted transcription factor T-bet described below (Fig. 3). The p38 MAP kinase pathway regulates IFN production in CD4⁺ T cells (Dong et al. 1998; Rincon et al. 1998; Yang et al. 1998) because p38 MAPK inhibitors and a dominant-negative p38 transgene reduced Th1 responses (Rincon et al. 1998). Targeted disruption of MKK3 (Lu et al. 1999) also diminished Th1 responses, but primarily by decreasing IL-12 production. The Jun NH₂-terminal kinase (JNK) pathway has also been implicated in Th1 development or function. A substantial reduction in IFN production was observed in JNK2-deficient T cells (Yang et al. 1998), although the decreased IFN production may be secondary to impaired expression of IL-12R2 in JNK2-deficient T cells rather than a direct requirement for JNK2 in IFN gene transcription.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Transcription factors that control Th1 differentiation.

Role of Stat4 in Th1 development  Th1 development involves signaling through Stat4, cloned by homology to other STATs (Zhong et al. 1994) but later found to be dedicated to IL-12 signaling in the mouse (Jacobson et al. 1995). Stat4 activation was first correlated with IFN expression by recognizing the loss of IL-12-induced Stat4 activation in Th2 cells (Szabo et al. 1995), and its role in Th1 development later proven using Stat4-deficient mice (Kaplan et al. 1996a; Thierfelder et al. 1996). Clearly not all IFN production by T cells is Stat4-dependent (Kaplan et al. 1998; Carter and Murphy 1999). Stat6/Stat4 double-deficient T cells produce some IFN (Kaplan et al. 1998), and CD8⁺ T cells are largely Stat4-independent for TCR-induced IFN production, in contrast to CD4⁺ T cells (Carter and Murphy 1999). Although Stat4 is critical in the generation of Th1 cells, its role in directly controlling IFN gene transcription is unclear. Stat4 may act on the IFN gene via cooperative binding to nonconsensus low-affinity STAT sites in the IFN gene promoter and first intron (Xu et al. 1996), but TCR-activated Th1 cells produce IFN without Stat4 activation (Ouyang et al. 1999; Yang et al. 1999). In addition, IFN-induced Stat4 activation in human T cells is not sufficient to activate IFN transcription (Rogge et al. 1998). Although IL-12 activates Stat1 and Stat3 as well as Stat4, only Stat4 is required for Th1 differentiation. Explanations for this selectivity might include induction of unique Stat4-activated targets or involve selective interactions with Stat4-interacting coactivators. Several STAT factors have been shown to interact with the general transcriptional coactivators p300/CBP (Bhattacharya et al. 1996; Zhang et al. 1996; Korzus et al. 1998).

Type I interferons in Th1 responses  Although the control of human and mouse Th differentiation is quite similar, one important difference has emerged. In human T cells, the cytokine IFN, not IL-12, controls Th1 development and IFN production (Parronchi et al. 1992; Brinkmann et al. 1993; Rogge et al. 1997). This is in contrast to the inability of IFN to induce Th1 development in the mouse (Wenner et al. 1996; Rogge et al. 1998). Both IL-12 and IFN activate Stat4 and induce Th1 development in human T cells, whereas only IL-12 exerts these effects in murine T cells (Rogge et al. 1998). Although IL-12 receptor expression appears to be modulated in the mouse, being lost in Th2 cells (Xu et al. 1998b), the constitutive expression of IFN receptors on human T cells provides an attractive explanation for the observation that human T cells show less exclusive polarization of IFN and IL-4 than mouse T cells.

IL-1 family cytokines and signaling in Th1 development  IL-12 and IL-4 are considered the dominant factors in Th1 and Th2 development, but other factors, especially the cytokine IL-18, also contribute. Differential actions of IL-1 on T-cell subsets were recognized early on (Lichtman et al. 1988; Taylor-Robinson and Phillips 1994; Shibuya et al. 1998), and IL-1R-deficient mice display enhanced Th2 responses in vivo (Satoskar et al. 1998). More recently, IL-18, an IL-1-related factor (Ghayur et al. 1997; Gu et al. 1997; Fantuzzi et al. 1998), was shown to be a selective activator of IFN in Th1, but not Th2, cells. IL-1 and IL-18 signaling pathways share similar components. Both IL-1 and IL-18 activate IRAK (Robinson et al. 1997), NF-B in Th1 cells (Matsumoto et al. 1997; Robinson et al. 1997), and TRAF6 (Kojima et al. 1998). MyD88-deficient mice not only lose IL-1-induced function but also IL-18-mediated functions (Adachi et al. 1998). IRAK-deficient mice have defective IL-18-mediated natural killer (NK) and Th1-type responses to pathogens in vivo (Kanakaraj et al. 1999). In addition, potential IL-18 activation of the MAP kinase pathway has been reported (Tsuji-Takayama et al. 1997).

Transcriptional regulation of Th1-specific cytokines

IL-2  IL-2 is expressed exclusively in T lymphocytes and is the cytokine produced earliest during Th differentiation. It is the major cytokine produced by the naive Thp cell. Differentiated effector Th1, but not Th2, cells also produce IL-2 albeit at lower levels than primary, stimulated Thp cells. IL-2 expression is primarily controlled at the transcriptional level, although costimulation via CD28 regulates IL-2 expression post-transcriptionally, by increasing IL-2 mRNA stability (Umlauf et al. 1995). TCR engagement activates MAP kinase and Ca²⁺-dependent signaling pathways, both essential for IL-2 expression, and leads to the activation of several transcription factors including members of the nuclear factor of activated T cells (NFAT) and activator protein 1 (AP-1) families (Durand et al. 1988; Shaw et al. 1988; Marx 1993; Northrop et al. 1994; Jain et al. 1995). In resting T cells, NFAT proteins reside in the cytoplasm in a phosphorylated state. Upon stimulation they are rapidly dephosphorylated by the phosphatase calcineurin and translocate into the nucleus where, in combination with AP-1 proteins, they activate the transcription of multiple cytokine genes, including IL-2 (Jain et al. 1995). The repression of cytokine gene transcription by the immunosuppressants cyclosporin A (CsA) and FK506 is explained by their blockade of calcineurin leading to inhibition of Ca²⁺-mediated, and hence NFAT-dependent, pathways (Emmel et al. 1989), as well as their blockade of JNK activation (Shibasaki et al. 1996).

View larger version (27K): [in this window] [in a new window]   Figure 4.   Regulating regions of cytokine genes.

IFN-  Th1 cells, CD8⁺ cells, and NK cells are the major secretors of IFN that is induced within 24 hr after stimulation of naive cells (Lederer et al. 1996). Very little is known about the regulatory regions of the IFN gene. For example, the location of elements that direct its tissue-specific expression has not been established in vitro or in vivo although early studies designated 8.6 kb as conferring T cell-specific expression (Agarwal and Rao 1998a). The Th1-specific regions within this sequence have not yet been identified. Reporter constructs containing 500 bp or 3 kb of upstream sequence are active but are expressed in both Th1 and Th2 cells (Young et al. 1994).

T-bet, a transcription factor that controls IFN transcription and Th1 lineage commitment  The approach that several laboratories took to unravel the genetic programs that controlled Th lineage commitment was to isolate the factors that accounted for the Th-subset selective expression of IFN and IL-4, reasoning that such factors should be important in specifying lineage fate. This search resulted in the isolation of one factor, T-bet, that is Th1 specific and controls IFN gene transcription, and two Th2-specific factors, GATA-3 and c-Maf that control IL-4 gene transcription.

Development of the Th2 lineage

Signaling pathways  Two major signaling pathways drive Th2 differentiation: TCR and IL-4 receptor (Fig. 5). As discussed below, signals emanating from the TCR activate the transcription of two Th2-specific factors, c-Maf and GATA-3, and also regulate the expression and the activation of the NFAT and AP-1 transcription factors. Upon signaling through the IL-4 receptor, Stat6 is recruited to bind to the cytoplasmic domain of the Il-4 receptor where it is then phosphorylated by Jak1 and Jak3 kinases leading to Stat6 dimerization and translocation to the nucleus (Hoey and Grusby 1999). Although it is unlikely that Stat6 itself directly controls IL-4 gene transcription to any significant degree (Lederer et al. 1996), it does activate a number of IL-4 target genes such as germ-line IgE gene transcription and switch recombination (Linehan et al. 1998), class II MHC, IL-4R, CD23 (Kaplan et al. 1996b; Shimoda et al. 1996; Takeda et al. 1996), and GATA-3 (Ouyang et al. 1998). Ectopic expression of an inducible, activated Stat6 into developing Th1 cells induces c-Maf and GATA-3 with subsequent expression of IL-4 (Kurata et al. 1999). There is controversy about whether the JNK kinase pathway, another signaling pathway, regulates Th differentiation. Mice that lack JNK1 or JNK2 have been reported by one group to have increased Th2 and defective Th1 differentiation. Enhanced Th2 and normal Th1 responses were found in JNK1-deficient T cells (Dong et al. 1998), when T cells were activated under neutral in vitro conditions (Dong et al. 1998; Yang et al. 1998). However, the Th2 skewing observed in JNK2^/ mice was not observed in an independently generated strain of JNK2-deficient mice (Sabapathy et al. 1999).

View larger version (34K): [in this window] [in a new window]   Figure 5.   Transcription factors that control TH2 differentiation.

Transcriptional regulation of Th2-specific cytokines

IL-4  The expression of IL-4 is restricted to a subset of CD4 (Th2) and CD8 (Tc2) T cells, NK T cells, T cells, mast cells, basophils, and eosinophils (Abbas et al. 1996). Naive Thp cells express IL-4 mRNA by 24 hr after stimulation via the TCR, and levels peak at 48 hr (Lederer et al. 1996). Studies in transgenic mice established that 3 kb of the IL-4 promoter conferred expression that mirrored endogenous IL-4 expression (Todd et al. 1993). Later studies that generated a series of IL-4 promoter transgenics demonstrated that the 741- to +60-bp region was preferentially expressed in Th2 cells. A trimer of the region spanning an NFAT-AP-1 site (88 to 61) was moderately Th2 specific. As for the IL-2 promoter, however, expression was low relative to endogenous IL-4 suggesting the presence of enhancer elements outside of the promoter (Wenner et al. 1997).

NFAT  NFAT was first identified as a TCR-inducible factor that regulated the IL-2 gene (Durand et al. 1988; Shaw et al. 1988; Crabtree 1989; Rao et al. 1997). NFAT proteins also regulate the promoters of multiple other cytokine genes expressed in T cells, including IL-4, GM-CSF, IL-3 and TNF (Miyatake et al. 1991; Chuvpilo et al. 1993; Goldfeld et al. 1993; Masuda et al. 1993; Rooney et al. 1994, 1995a; Cockerill et al. 1995). Isolation of the genes encoding these proteins has yielded four NFAT family members called NFATc1 (also called NFATc, NFAT2), NFATc2 (also called NFATp, NFAT1), NFATc3 (NFAT4 and NFATx), and the nonlymphoid NFATc4 (NFAT3) that are highly homologous within a region distantly related to the Rel domain (McCaffrey et al. 1993; Northrop et al. 1994; Hoey et al. 1995; Masuda et al. 1995). NFAT family members differ markedly in amino- and carboxy-terminal regions (Luo et al. 1996), and this, together with their differing tissue distribution and response to stimuli (Hoey et al. 1995; Ranger et al. 1998b), suggested that their functions might differ, as indeed proved to be the case.

GATA-3 and c-Maf: Th2-specific transcription factors

Substantial progress has been made in identifying the transcription factors responsible for the tissue-specific expression of the IL-4 gene in Th2 cells (Szabo et al. 1997b; Glimcher and Singh 1999; Glimcher et al. 1999) (Figs. 4 and 5). Two Th2-specific factors, the c-Maf proto-oncogene and the GATA-3 zinc finger factor, have been identified (Ho et al. 1991, 1996; Ting et al. 1996; Zhang et al. 1997; Zheng and Flavell 1997). Non-Th2-specific factors such as Stat6, NFAT proteins (Durand et al. 1988; Shaw et al. 1988; Crabtree 1989; McCaffrey et al. 1993; Northrop et al. 1994; Hoey et al. 1995; Masuda et al. 1995; Rao et al. 1997), and a novel nuclear antigen, NFAT-interacting protein 45 kD (NIP45) (Hodge et al. 1996a), also help to regulate IL-4 production.

GATA-3  The zinc finger protein GATA-3 was initially cloned as a T-cell-specific transcription factor that bound to the Ta3 element of the TCR gene enhancer (Ho et al. 1991). Deletion of the GATA-3 gene produces lethal abnormalities of the central nervous and hematopoietic systems (Pandolfi et al. 1995), whereas mice lacking GATA-3 in the lymphoid system do not generate T cells because GATA-3-deficient thymocytes arrest at the immature double-negative stage (Ting et al. 1996; Hendriks et al. 1999). A key discovery by two groups was the selective expression of GATA-3 in Th2 cells (Zhang et al. 1997; Zheng and Flavell 1997). Ectopic expression of a GATA-3 transgene led to increased levels of Th2 cytokines, whereas a dominant-negative GATA-3 transgene inhibited Th2 differentiation (Zheng and Flavell 1997; Zhang et al. 1999). GATA-3 directly controls IL-5 gene expression through binding to and transactivating elements in the region 70 to 59 (Siegel et al. 1995; Zhang et al. 1997, 1998a; Lee et al. 1998). GATA-3 probably does not directly bind to and transactivate the IL-4 promoter (Zhang et al. 1997, 1998a). Rather, several genomic regions within the IL-4/IL-13 locus have GATA-3-dependent enhancer activity (Ouyang et al. 1998) suggesting GATA-3 may augment expression of IL-4 or IL-13 via interactions at sites distant from the proximal-promoter.

c-Maf  c-Maf was isolated using a yeast two hybrid screen with NFATc1 as bait and cDNAs from activated Th2 cells as library (Ho et al. 1996). c-Maf, the cellular homolog of the avian viral oncogene v-maf, is a member of the AP-1 family of basic region/leucine zipper factors and binds to a consensus site (MARE) in the proximal IL-4 promoter. The expression of c-Maf is limited to Th2 cells, and its own expression is induced by signals transmitted via the TCR. Ectopic expression of c-Maf in all cells tested including yeast potently activates the IL-4 promoter. In CD4⁺ T cell lines prepared from patients with atopic disease, c-Maf is expressed at very high levels in Th2 clones but minimally in Th0 clones that produce IFN and very little IL-4. Extremely strong evidence for the critical role of c-Maf in controlling IL-4 production has been gathered in vivo with the production of c-Maf overexpressor transgenics (Ho et al. 1998) and most convincingly with mice that lack c-Maf (Kim et al. 1999). c-Maf transgenic mice have an increased Th2 immune response in vivo and in vitro that can be ablated by backcrossing onto an IL-4-deficient background (Ho et al. 1998). Targeting of the c-Maf locus revealed severely impaired IL-4 production in the absence of c-Maf (Kim et al. 1999). These data provide very strong evidence that c-Maf is required for the production of IL-4 in vivo. However, the provision of c-Maf to mature effector Th1 cells cannot allow them to produce IL-4, suggesting that other factors such as Stat6 or GATA-3 may also be required.

DNA replication and monoallelic cytokine gene expression

The above discussion has emphasized signaling and transcriptional control of cytokine expression underlying a predominantly instructional model of T helper development. However, certain recent results have raised the issue of whether T helper commitment might not rely on stochastic or selective processes. Cytokine-driven responses in bulk populations do not necessarily distinguish between instructive or selective models of T cell differentiation. Recently, Coffman and Reiner (1999) have outlined three competing models of T helper development: instructive differentiation, selective differentiation, and a hybrid instructive/selective model. In the first model, all daughter cells of uncommitted progenitors respond to specific differentiating signals to acquire one particular developmental state. In the second, daughter cells of a progenitor adopt multiple development states in a random or stochastic manner, which later appears to polarize only by selective outgrowth of one cell type in responses to "polarizing" cytokines. Their hybrid model adds an effect of cytokines on altering the ratio in the initial developmental fates but retains the component of selective outgrowth in response to cytokines for achieving polarization.

Examining the stability or reversibility in bulk cultures reveals the potential for early reversibility and of heterogeneity but does not address issues of single cell fate determination (Murphy et al. 1996; Coffman et al. 1999). Experiments using a lineage ablation approach, in which the IL-4 promoter drove expression of thymidine kinase (Kamogawa et al. 1993; Minasi et al. 1993; Nakamura et al. 1997), suggested that precursors to both Th1 and Th2 cells could proceed through a stage where the IL-4 promoter was active. Favoring the early commitment to the Th2 pathway is the rapid loss of a functional IL-12 receptor (Szabo et al. 1997a). Early Th1 populations, in contrast, exhibit greater reversibility (Mocci and Coffman 1995, 1997; Szabo et al. 1997a) due perhaps to uncommitted precursors within the early Th1 population (Mocci and Coffman 1997).

The possibility of epigenetic regulation of cytokine expression has recently sparked a renewed interest in noninstructional models of T helper development. Monoallelic expression using polymorphic IL-2 alleles from F₁ mouse strains was revealed in both T cell clones and primary T cells to occur in a substantial fraction of T cells (Hollander et al. 1998), suggesting that alleles of IL-2 were not activated uniformly in all T cells. Similar findings for IL-4 alleles in Th2 cells were later reported (Bix and Locksley 1998), and these observations extended using reporter genes or knock-ins targeted to both the IL-2 (Naramura et al. 1998) and IL-4 loci (Riviere et al. 1998), allowing individual alleles to be examined in early populations of developing TCR transgenic T cells. In contrast to earlier findings, however (Hollander et al. 1998), biallelic expression of both IL-2 and IL-4 was observed as the predominant pattern in T-cell activation (Naramura et al. 1998; Riviere et al. 1998). The level of T-cell activation could also alter the distribution between monoallelic and biallelic expression (Riviere et al. 1998). These data favor a model in which activation of alleles is a stochastic process whose probability depends on TCR signal strength.

Reiner and colleagues have proposed that cell cycle is a more important regulator of T helper development than instructive cytokine signals based on distinct requirements for numbers of cell division in initiating cytokine expression (Bird et al. 1998). Naive T cells were immediately capable of producing IL-2, but IFN and IL-4 production correlated with the number of cell divisions after activation, IFN requiring only one to two cell divisions, but IL-4, four divisions. Similar results reported for cytokines that are not differentially expressed among Th1 and Th2 cells, such as IL-3 (Gett and Hodgkin 1998), implied that these differences may not control Th1/Th2 commitment. Furthermore, a very recent analysis (Richter et al. 1999) provided evidence against a strict cell-division model and in favor of an instructional model in which the epigenetic changes induced by TCR and IL-4R signaling are required for cytokine gene expression. This study found that naive T cells were capable of initiating IL-4 production after priming even when cell division was blocked as long as entry into the S phase of the cell cycle was allowed. Once again, the required epigenetic changes appeared to be induced by cytokines, consistent with an instructional model of development.

Role of chromatin structure

Genomic DNA is packaged into nucleosomes, which then form higher order chromatin structures. Chromatin remodeling is accompanied by changes in nucleosomal positioning, an event that is ATP dependent. It has been suggested that cytokine gene loci, including IL-2, IL-4, IL-12, IL-13, IFN, and GM-CSF, undergo changes in chromatin structure that allow access to gene-specific transcription factors, likely mediated in part through acetylation of histones. Distal control regions such as enhancers and LCRs have been shown to regulate accessibility to gene loci by regulating chromatin structure (Ernst and Smale 1995). The location of regions that control these changes have been identified in cytokine genes using both assessment of CpG methylation status and DNase I hypersensitivity assays since transcriptionally active chromatin is hypomethylated and is more accessible to nucleases (Agarwal and Rao 1998a; Kadonaga 1998; Takemoto et al. 1998; Agarwal et al. 1999). The most complete information on the role of chromatin remodeling in the regulation of cytokine genes is available for two cytokines, IL-4 and IFN-, and is summarized below and reviewed more thoroughly by Rengarajan and Glimcher (2000).

Regulation of the IL-4 and IFN- loci  Several observations suggested that the IL-4 gene was regulated at the level of chromatin structure. First, in vitro analyses had indicated that only 157 bp of the proximal IL-4 promoter was required for tissue-specific expression in Th2 cells. Nevertheless, in vivo, the proximal 800 bp of the promoter was required to confer significant Th2-selective expression of an IL-4 promoter-reporter transgene (Wenner et al. 1997). However, neither 800 bp nor up to 3 kb of sequence were sufficient to achieve expression equivalent to the endogenous IL-4 gene, suggesting the presence of additional elements for optimal expression (Todd et al. 1993). Second, differentiated, effector Th2 cells produce IL-4 more rapidly and at higher levels than naive Thp cells, implying that the IL-4 regulatory regions are more accessible or "poised" in effector, but not in naive, T cells (Reiner and Seder 1999).

Evidence for chromatin remodeling  In both mouse and human chromosomes, the Th2 cytokine genes IL-4, IL-5, and IL-13 are clustered together within 150 kb, consistent with the notion that these genes compose a single chromosomal locus that may be controlled by long-range modulation of chromatin structure. Rao and colleagues (Agarwal and Rao 1998b) provided evidence in favor of this model by using both methylation and DNase I hypersensitivity assays to identify regions in the IL-4/IL-13 locus that responded to stimulation through the TCR. Differentiated Th2 cells and established Th2 clones had an accessible chromatin structure as evidenced by the presence of five clusters of HS sites over the IL-4 locus. Naive T helper cells, in contrast, like Th1 cells, possessed only one of the five HS sites. Within 48 hr of antigen activation, naive Thp rapidly acquired the chromatin phenotype of differentiated Th2 cells, implicating these regions in important regulatory functions (Agarwal and Rao 1998b). The acquisition of these sites was dependent on both IL-4 and Stat6 (Agarwal and Rao 1998b). The Th2-specific transcription factor GATA-3 induced the same changes when transduced into differentiated Th1 cells or Stat6-deficient cells, indicating that GATA-3 acts to remodel this locus downstream of Stat6 (Ouyang et al. 2000). Furthermore, naive Thp cells possessed a hypermethylated IL-4 locus, whereas differentiated Th2 cells display decreased CpG methylation at the IL-4 locus. This latter result is consistent with the observation that treatment of naive Thp cells with trichostatin A and 5-azacytidine (histone deacetylase and methylase inhibitors, respectively) accelerates the kinetics of IL-4 production (Bird et al. 1998). Factors like GATA-3 and Stat6 may directly remodel chromatin structure to allow TCR-induced factors like c-Maf and NFAT to access their specific binding sites in the IL-4 locus and promote rapid transcription of IL-4 (Agarwal and Rao 1998b); NFAT proteins may also directly alter chromatin configuration. Mice that lack both NFATc2 and NFATc3 vastly overproduce Th2 cytokines and display a highly allergic phenotype (Ranger et al. 1998b) suggesting that NFAT proteins may help regulate the balance between the active/inactive state of the IL-4/IL-5/IL-13 locus during the initiation of Th2 differentiation.

Summary

The genetic programs that specify lineage commitment in the T helper cell are beginning to be elucidated (Fig. 6). An attractive model to posit considers T-bet and GATA-3 as the master regulators of Th1 and Th2 differentiation, respectively, with c-Maf as the downstream factor that directly and selectively controls IL-4 gene transcription. The expression of T-bet and GATA-3 during Th differentiation are mirror images. T-bet and GATA-3 are strikingly similar in their ability to induce one lineage while simultaneously repressing the opposing lineage. T-bet may lie downstream of IL-12/Stat4 and IFN (Stat1), the major inducers of Th1 differentiation, whereas GATA-3 is activated by IL-4/Stat6, the critical Th2-inducing cytokine. It is tempting to postulate that T-bet and GATA-3 regulate each other with T-bet suppressing GATA-3 expression during Th1 differentiation and GATA-3 inhibiting T-bet expression during Th2 development. Cytokine gene expression may more broadly reflect a balance between repressors and activators. Transcription