The enigma of CD4 lineage specification

Introduction

CD4 T cells are essential players in the immune response, and work by recruiting and controling the functions of most cells involved in defenses against pathogens. Their importance is dramatically illustrated by the disseminated infections that occur in latestage HIV infection or after ablative cancer chemotherapy. By far the most abundant of these cells, referred to as ‘conventional’ CD4 cells, recognize peptide antigens bound to MHC-II molecules, although there are important subsets of CD4 cells restricted by other MHC or MHC-like molecules, including CD1d-restricted ‘invariant’ iNK T cells. Because the thymus involutes with age, T cell loss in adults (e.g. after chemotherapy) cannot be compensated by thymic production of new T cells; thus, much interest has focused on in vitro T cell generation systems. However, generating CD4 T cells in vitro has proven challenging [1], underscoring the need for a better understanding of the intracellular events that lead to CD4 cell development. In the present review, we will discuss recently identified transcriptional ‘nodes’ that appear involved in the development of CD4 cells regardless of their antigen specificity or function, and focus on how transcription factors acting in thymocytes promote the emergence of CD4-lineage specific gene expression patterns.

Making T cells from DP thymocytes

CD4 T cells carry αβ TCRs and differentiate in the thymus from ‘double-positive’ (DP) precursors that express both CD4 and CD8 coreceptors and form the most abundant subset in the thymus [2]. Although DP thymocytes have not yet acquired immune functions, they no longer harbor the multipotency that characterizes early thymic precursors. DP thymocytes not only have lost the ability to differentiate into non-T cells but they have rearranged their TCRβ and TCRα gene loci and no longer retain γδ lineage potential [2]. Their differentiation into mature T cells is triggered by the engagement of their TCR by MHC-peptide complexes expressed by the thymic epithelium, and involves three conceptually distinct, even if possibly overlapping, steps: positive selection, lineage differentiation and terminal maturation. In its strictest sense, positive selection is the rescue of DP cells from apoptotic death, their default fate in the absence of TCR signaling [3, 4]. It involves the up-regulation of anti-apoptotic factors of the Bcl-2 family, including Bcl-2 itself and Mcl1, and is accompanied by changes in expression of chemokine receptors and adhesion molecules that cause the migration of TCR-signaled thymocytes from their initial cortical to a thymic medullary location [5].

At the other end of the developmental pathway, positively selected thymocytes prepare their exit from the thymus and acquire functional and survival properties typical of mature T cells. This ‘maturation’ step involves several transcription factors, including Klf2, Foxo1 and Foxo3, and Foxp1 (structurally related to Foxp3, the master regulator of the regulatory T cell fate) [6–8]. An emerging feature of this maturation step in conventional thymocytes is that it restrains the expression of effector genes so that these cells remain quiescent and functionally inactive. In specific thymocyte subsets, other factors instead promote the acquisition of effector properties before thymic egress. The best characterized of these is PLZF, a zinc finger factor required for iNK T cell effector differentiation [9, 10].

Thpok, Runx and the separation of CD4 and CD8 lineages

The divergence of CD4 and CD8 lineages appears to take place between positive selection and terminal maturation. Genetic analyses over the last few years have identified several transcription factors required for the development of CD4 but not CD8 cells. An important notion that emerged from these findings is that of a CD4-commitment checkpoint, at which CD4-differentiating thymocytes lose their ability to adopt a CD8 fate. How this checkpoint operates has been reviewed in detail recently [11–13], and we will only summarize a few key points below.

CD4-lineage commitment requires the zinc finger transcription factor Thpok (encoded by the Zbtb7b gene, that we will refer to as Thpok for simplicity). Thpok, then called cKrox, was initially identified as a collagen promoter binding protein [14]. It entered the immunology field in 2005 when it was shown to be mutated in a spontaneously occurring mouse line (HD) lacking helper T cells, and in another study through microarray comparisons of gene expression in differentiating thymocytes [15, 16]. Studies of HD mice, subsequently corroborated by knock-out analyses, showed that Thpok-deficient MHC II-restricted cells differentiate into mature CD8 T cells [15, 17–20]. Such ‘lineage redirection’ demonstrates that Thpok is required for CD4- differentiation, but not for the development of MHC II-restricted thymocytes. Several key features of Thpok fit with this commitment role. Although Thpok is expressed in ‘nonconventional’ T cells (including CD1d-restricted iNK T cells and some γδ T cells [21–23]), its expression in conventional αβ lineage thymocytes appears limited to the MHC II-restricted subset [15, 16]. Furthermore, enforcing its expression in MHC I-restricted thymocytes prevents their differentiation into CD8 cells [15, 16], and Thpok represses the expression of CD8-lineage genes, including Cd8 itself and the transcription factor Runx3 that promotes the development of CD8 lineage cells [18, 19, 24, 25].

Reciprocating the repression of Runx3 by Thpok, which mediates at least in part the ‘committing’ function of Thpok [19], Runx3 and the related factor Runx1 inhibit Thpok expression in CD8-lineage and DP thymocytes [26]. These findings have led to the concept that a dual negative regulatory loop based on Thpok and Runx3, and alimented by other factors, including the zinc finger protein Mazr [27], is the keystone of CD4-CD8 lineage commitment in the thymus.

CD4 lineage specification

It was initially envisioned that Thpok could act as a ‘master switch’ of CD4 differentiation, that is, that it would not only turn off CD8-lineage genes but also turn on CD4-lineage genes, which we will refer to as ‘CD4 lineage specification’. Indeed, mirroring the ‘CD8-redirection’ of Thpok-deficient MHC II-restricted thymocytes, enforced expression of Thpok in MHC I-restricted thymocytes not only prevents their CD8 differentiation, but redirects them into the CD4 lineage [15, 16]. However this only results in the generation of small numbers of MHC I-restricted CD4 T cells [16, and S.R. Jenkinson, K. F. Wildt and R.B., unpublished data], raising the possibility that Thpok is not sufficient for CD4 lineage specification. In addition, small numbers of CD4 T cells develop in mice that lack both Thpok and Runx activity, suggesting that Thpok, at least in such circumstances, is not even necessary for CD4-lineage specification [19].

Do these findings establish that Thpok is not involved in CD4-lineage specification? Not necessarily. The relative inefficiency with which Thpok transgenic MHC I-restricted thymocytes give rise to CD4 cells could simply be due to their impaired survival when they shut-down CD8 expression, because CD8 is critical for MHC-I induced TCR signaling in the thymus [28]. Analyses of T cell differentiation in Thpok-Runx deficient mice are complicated by the fact they require early disruption of Runx activity, prior to the DP stage, resulting in inefficient CD4 T cell generation [19]. Furthermore, Thpok counteracts the Runx-mediated repression of Cd4 and Thpok [18, 29], and there is evidence that this effect is at least in part direct [18]. Thus, whether Thpok contributes to promote CD4-lineage specification in physiological circumstances remains an open question.

Analyzing this CD4-specification step is complicated by the paucity of markers of the CD4 lineage. Distinguishing differentiating CD4 from CD8 and DP thymocytes may seem a trivial issue, because CD4 and CD8 T cells differ in their coreceptor expression and functional helper vs. cytotoxic properties. However, these features are largely inoperative to define CD4 lineage precursors. Most remarkably, Cd8 gene expression is transiently repressed during positive selection in MHC I-restricted thymocytes, so that the down-regulation of surface CD8 expression is not indicative of CD4-lineage differentiation [30, 31]. Conversely, Cd4 expression in CD4-differentiating cells is not as dependent on a cis-regulatory element (‘proximal’ Cd4 enhancer) as is Cd4 expression in DP thymocytes [32]. While this underscores that the differences in gene regulation between DP and CD4 T cells are perhaps greater than they appear, this property cannot at the present time be used to follow CD4-lineage differentiation because the additional element that would direct Cd4 expression in mature CD4 cells is not yet known. Lastly, while mature CD4 cells differ from their CD8 counterparts by their broader potential for cytokines expression (notably Th2 cytokines IL-4, IL-5 and IL-13 that CD8 T cells typically express at much lower levels), these effector cytokines are not expressed in CD4-differentiating thymocytes.

Thus, unlike for differentiating CD8 thymocytes that express genes characteristic of the cytotoxic program, it has been difficult to identify functional markers of CD4-helper differentiation in the thymus. In fact, only Thpok seems so far unambiguously associated with CD4 cell differentiation, whereas Cd40lg and to a lesser extent Gata3 are preferentially but not exclusively expressed in CD4-differentiating cells [33–35]. This paucity of developmental markers has undoubtedly complicated analyses of lineage differentiation in the thymus, and explains in large part the preponderance of genetic studies in our understanding of this process.

Early birds: Tox, E-proteins, Gata3 and Myb

Indeed, other genes are required for CD4 T cell development at an earlier stage than Thpok. Such an early block raises the possibility that these genes promote an early ‘lineage specification’ step that would precede commitment, similar to what has been characterized in other biological systems [36]. Notwithstanding the experimental limitations highlighted in the previous section, we will adopt this hypothesis as the leading thread of our discussion of these factors, including Tox, E-proteins, Gata3 and Myb.

Tox is an HMG (High Mobility Group) domain DNA binding protein preferentially expressed in the thymus [37]. There is little if any Tox expression in preselection DP thymocytes, but it is strongly up-regulated by TCR signaling, through a calcium9 calcineurin dependent pathway [38]. Unlike for Thpok, there does not seem to be preferential Tox expression in MHC-II vs. MHC-I signaled cells. Nonetheless, Tox disruption essentially precludes the generation of CD4 SP thymocytes and T cells, whereas its impact on the development of CD8 cells, although detectable, is less pronounced [39]. Arrested Tox-deficient thymocytes accumulate as CD4^{CD8 cells that do not express Thpok, consistent with a role of Tox in lineage specification. In contrast, these cells normally up-regulate Gata3; since Gata3 up-regulation is induced by TCR signaling [}35], this results fit with experimental evidence that Tox disruption does not impair TCR signal transduction [39].

Given these observations, it is striking that transgenic Tox expression promotes the development of CD8 lineage cells, even in circumstances where there is no intrathymic TCR engagement [37, 38]. Although why this is so remains to be determined, an appealing possibility is that Tox allows thymocytes to clear an early positive selection checkpoint but does not on its own promote Thpok expression. Enforced Tox expression could then allow thymocytes that have not received TCR signals to reach the CD4 commitment checkpoint, at which they would fail CD4 commitment because they lack Thpok (since Thpok expression requires TCR signaling), but would be nonetheless be able to adopt a CD8 fate as they express Runx3 [38]. While a detailed discussion of this possibility, and of how CD8 cells could differentiate in the absence of TCR signals [40] is beyond the scope of this review, it is notable that E-proteins E2A and HEB enforce an early positive selection checkpoint and that, similar to transgenic Tox expression, disruption of genes encoding these factors results in the generation of CD8-lineage cells in the absence of TCR signaling [41]. A tantalizing hypothesis would be that Tox and E10 proteins have opposite effects in early positive selection, possibly because Tox antagonizes E-proteins by upregulating E-protein inhibitors encoded by the Id gene family [42], as recently observed during NK cell differentiation [43]. The similar development of TCR-independent CD8 cells from E-protein-deficient and Tox-overexpressing thymocytes is consistent with this perspective.

However, it is difficult to infer from these observations whether E-protein and Tox serve similar functions during CD4-lineage specification. Because of the pleiotropic activity of E-proteins in thymocytes, we cannot at present distinguish their potential effects on CD4-lineage specification from those on positive selection and Cd4 gene expression (and thereby MHC II-induced TCR signaling) [41, 44]. As an illustration of this complexity, genetic inactivation of E proteins or of their inhibitor Id3 paradoxically both impair CD4 T cell development [41, 45]. Further analyses will be needed to gain a better understanding of the respective roles of these factors in lineage specification.

Gata3 is expressed throughout T cell development and required at several key checkpoints of thymocyte differentiation, notably for T lineage commitment and β-selection [46]. During positive selection, Gata3 expression is higher in MHC-II than in MHC I-signaled cells [35]. The most dramatic effect of Gata3 disruption in DP thymocytes (obtained by Cre-mediated excision of a ‘floxed’ Gata3 allele) is an almost complete loss of CD4 T cells contrasting with better preserved CD8 cell populations [47]. Gata3 is required for Thpok expression, both in conventional and iNK T precursors [20, 23], an effect linked at least in part to its binding to the Thpok locus; in that sense, Gata3 contributes to specification.

However, Gata3 contribution to CD4 T cell differentiation is not limited to promoting Thpok expression, as enforced expression of Thpok fails to restore the CD4 differentiation of Gata3-deficient thymocytes [20]. One possibility is that Gata3 promotes the expression of other genes required for CD4 lineage differentiation, whose identity remains elusive. Enforced expression of Gata3 is not sufficient to promote expression of Thpok or Cd40lg, indicating that such a ‘CD4-specifying’ function would require cooperation with other factors [48, and Y.X. and R.B., unpublished results]. An additional possibility is that Gata3 is needed for CD4 cell differentiation because it promotes TCR signaling. This seems paradoxical at first because TCR signaling is necessary for the differentiation of both CD4 and CD8 cells, whereas Gata3 is only required for the former. However, CD4 cell differentiation depends on sustained TCR signaling, unlike CD8 cell differentiation [for a recent review, see Ref. 49]. If Gata3 is necessary for the expression of TCR signaling intermediates, the amount of such intermediates remaining in DP thymocytes after Cre-mediated Gata3 disruption could be sufficient for ‘transient’ but not for ‘persistent’ TCR signals, therefore allowing CD8 but not CD4-lineage differentiation. Although TCR signaling in DP thymocytes was reported to be unaffected by Gata3 disruption [47], the reduced expression of TCR and of CD69 (a surface molecule up-regulated by TCR signaling) in Gata3-deficient thymocytes undergoing positive selection is consistent with this view [20, 50]. Thus, the current data suggests that Gata3 promotes early CD4 T cell differentiation through at least two mechanisms: directly by contributing to expression of Thpok and possibly other genes specific of the CD4 lineage, and indirectly by promoting TCR signal transduction.

In addition to Gata3 and Tox, the transcription factor Myb is important for CD4 T cell differentiation [51], and has been reported to promote Gata3 expression, by binding the Gata3 promoter [52]. Because TCR expression or signaling appear normal in Mybdeficient thymocytes, unlike in their Gata3-deficient counterparts, it is unlikely that Myb is necessary for Gata3 expression, although it could be partly redundant with the related genes Amyb and Bmyb. Conversely, overexpression of a constitutively active form of Myb prevents the development of CD8 T cells, whereas forced expression of Gata3 does not have such an effect, suggesting that the targets of Myb and Gata3 are only partly overlapping during lineage differentiation. Of note, while Myb has pleitropic effects on DP thymocytes, including promoting their survival, it promotes CD4 cell differentiation independently from its survival effect [53, 54].

Themis

Although there is no evidence suggesting that Themis acts as a transcription factor, its discussion in this review is justified by its nuclear localization and by its apparent involvement in both positive selection and CD4 lineage differentiation. Themis, which was identified last year in five independent studies [55–59], is required for the generation of mature T cells, both CD4 and CD8, from DP thymocytes, in part because it promotes rescue from programmed cell death upon TCR engagement. The connection with our current focus comes from experiments assessing the fate of Themis-deficient thymocytes forced to express Bcl-2, which effectively prevent their apoptotic death: Bcl-2 restores the number of CD8 SP thymocytes, whereas it has little or no effect on the number of CD4 SP cells [56]. This suggests that, in addition to its effect on positive selection, Themis could be specifically required for CD4-lineage differentiation. Indeed, Themis13 deficient thymocytes fail to normally up-regulate Gata3 and Thpok. Of note, enforced expression of either factor fails to restore CD4 differentiation, suggesting that Themis affects multiple nodes of the CD4 differentiation circuitry [56].

What we know from Themis offers fragmented yet interesting clues to its potential function. Cell fractionation experiments found the proteins in both nuclear and cytosolic fractions, raising the possibility that the protein may shuttle specific signals to or from the nucleus [56]. Given that the requirement for Themis becomes manifest when TCR signaling is required, it could be conceived that this molecule promotes TCR signaling. Consistent with this possibility, T cell development in Themis-deficient animals resembles that in mice in which TCR signaling is specifically compromised after the initiation of positive selection [60]. However, whether Themis is directly involved in TCR signal transduction is controversial, and further work will be needed to ascertain its function in T cell differentiation.

Conclusions and Perspectives

The recent years have seen much progress in elucidating the mechanisms of CD4 lineage ‘commitment’, i.e. the loss of CD8 developmental fate, notably through the identification of the Thpok-Runx interplay [11–13]. In contrast, much less is known about the genes involved in CD4-lineage specification, i.e. the initiation of gene expression programs specific of the CD4 lineage. Part of the remaining uncertainty is due to the absence of a clear definition of what is ‘CD4-lineage specification’, as so far only Thpok expression has been shown to be specific of CD4-lineage cells. The difficulty to distinguish factors involved in MHC II- induced positive selection from those promoting a putative CD4 specification step has been a leitmotiv in this review: we have 14 encountered this problem with Tox, E-proteins, Gata3 and Themis. Although these difficulties may be due to experimental limitations, they could indicate that these two events, positive selection of CD4 cells and CD4-lineage specification are intricately directed by the same transcriptional circuitry.

Pushing this logic to the extreme, it is conceivable that, with the exception of inducing Thpok expression, there is no ‘CD4-lineage specification’ prior to CD4 commitment. In that perspective, all it would take to make a CD4 T cell would be to inhibit CD8 and cytotoxic differentiation in positively selected cells, through Thpok expression or, as recently proposed, by inhibiting cytokine signaling [40]. In that sense, with the critical exception of their Thpok expression, CD4 T cells would remain ‘unspecified’ until they undergo effector differentiation. Alternatively, it is conceivable that lineage-specific gene expression is established after commitment, and that the transcriptional control of thymocyte maturation would be distinct in CD4- and CD8- lineage cells, even though there is so far little evidence to support this idea. An important objective of current research is to determine whether transcription factors required for CD4 T cell generation and Thpok expression act directly by promoting expression of CD4-lineage genes, or indirectly by enabling signaling conditions conducive to CD4 lineage choice. These alternative are of course not mutually exclusive, as we have seen for Gata3.

Another theme emerging from these studies is the conservation of key nodes in the circuitry that promotes lineage differentiation. Several subsets of CD4 T cells, some of them with pre-programmed effector properties (e.g. iNK T, Treg), differentiate from DP thymocytes [2]. In CD1d-restricted iNK T cells, similar to MHC II-restricted cells, Thpok 15 is required for CD4 expression and for repression of CD8 [23, 61], and Thpok expression is also controled by Gata3 [23]. These findings suggest a degree of commonality in the development of all CD4 cells, regardless of their MHC-antigen specificity. However, they should not be overemphasized. In contrast to Thpok and Gata3, which are needed for the development of CD4 ^{but not CD4 iNK T cells [}23, 61, 62], Tox appears required for the development of all iNK T cells [39]. More strikingly, while Thpok in iNK T cells inhibits Cd8 gene expression, it may promote, rather than repress, cytotoxic gene expression [23, 61]. Thus, multiple variations seem to occur on the common theme of Thpok inhibiting CD8-lineage differentiation.

Important challenges remain. The first is to connect the transcriptional circuits with extra-cellular signals that promote lineage choice [49, 63]. At the other end of the spectrum, it will be important to build the circuitry and connections between factors, that with a few exceptions so far can only be speculated from genetic studies; large scale sequencing analyses will undoubtedly shed new light on this important question. Finally, despite emerging clues, it remains largely unclear how factors like Thpok or Runx actually affect gene expression. Thus, it will be important to understand how DNA-binding transcription factors are connected, through co-activator and co-repressor complexes, to the general transcription machinery and RNA polymerase, and how they install epigenetic marks to achieve persistent gene silencing [64–66].

Acknowledgments