Independent Regulation of Cytokine Genes in T Cells

The dominant paradigm in the field of T-cell immunoregulation today is the notion that cytokine-producing T cells can be classified into two major types. Type 1 cells of CD4+ or CD8+ phenotype synthesize interleukin(IL*)-2, interferon (IFN)-γ, and lymphotoxin-α, whereas type 2 cells synthesize IL-4, IL-5, IL-6, IL-10, and IL-13. Both synthesize IL-3, granulocyte-macrophage colony-stimulating factor (GM-CSF), and tumor necrosis factor (TNF)-α. The type 1 and type 2 cytokines then differentially regulate a wide variety of antigen-specific and nonspecific immune effector mechanisms. A simple, widely accepted working model for T-cell regulation of the immune response has developed around this division of labor and guides the interpretation of many experimental and clinical observations in transplantation, infection, autoimmunity, and other immunological conditions. This model suggests that signals received from the antigen-presenting cell and bystander cells during T-cell activation direct the development of the T-cell cytokine response toward the production of either type 1 or 2 cytokines. The longer the immunogen persists, the more polarized the T-cell cytokine response becomes. The preferential production of type 1 or type 2 cytokines in turn regulates the development and activation of effector cells of the inflammatory and immune systems and therefore determines the outcome of the response(1, 2). This model is supported at the T-cell population level by the demonstration that cytokine profiles are polarized in immune responses to certain infectious agents, allergens, and other persistent immunogens in mouse and man, and that these polarized profiles play a key role in antigen clearance and pathogenesis (1, 2). A central premise of this model is that these two sets of cytokine genes are reciprocally expressed in T cells, while those within each set are coordinately expressed. There is, however, a large body of evidence that synthesis of each cytokine can be regulated independently of the others(3). As discussed below, differential expression can be demonstrated at multiple levels in response to the same and different signals, within and between T cells, and within and between the type 1 and 2 cytokine sets. Here we review the evidence for linked and for independent regulation of cytokine genes in T cells, in the latter case with particular emphasis on work done in our laboratory. We also discuss how the reconciliation of these two, apparently paradoxical, aspects of cytokine gene regulation may allow a more sophisticated understanding of the complexity of many immune reactions. LINKED REGULATION OF CYTOKINE GENE EXPRESSION IN T CELLS The regulation of T-cell cytokine gene expression varies according to the stage of differentiation of the cells. In naive T cells after primary activation and early differentiation, the most important influences driving type 1 and type 2 polarization seem to be cytokines such as IL-12 and IL-4, the signals they transmit through their cognate receptors, and the pathways involving the Janus kinases and the signal transducers and activators of transcription (STAT) (1, 4). As evidenced by transgenic knock-out mice that lack specific members of the STAT family, these signals can have significant effects on the ability of T cells to develop toward coordinated expression of type 1 or type 2 cytokines(5, 6). Understanding of this level of regulation will require additional information on the relative roles of other transcription factors, such as GATA-3 (7), and the network of molecular interactions that potentially can be used during T-cell priming and early differentiation. In differentiated T cells, especially mature cells in the postactivation effector phase of cytokine expression, other regulatory influences take effect and three major molecular levels of regulation have the potential to link and coordinate the expression of multiple cytokines: convergent receptor-triggered signaling pathways, common and cooperating transcription factors, and coordinated posttranscriptional mRNA degradation (Fig. 1). T-cell cytokine expression is largely inducible and initiated by cross-linking of the T-cell antigen receptor (TCR) and/or other accessory cell-surface molecules. This is followed by activation of the following major signaling pathways and transcription factors: a calcium-dependent and cyclosporinsensitive pathway that regulates the nuclear factor of activated T cells (NFAT); a protein kinase C-dependent pathway that regulates activator protein-1 (AP-1) and nuclear factor kappa B (NF-κB); and a p21ras-activated mitogen-activated protein kinase cascade that affects AP-1 and other transcription factors (8, 9). Thus, these signaling pathways converge, coordinately activating multiple transcription factors that are present in all normal lymphocytes and are involved in the regulation of expression of multiple cytokine genes(10-14). Activated transcription factors localized in the nucleus recognize specific DNA sequences that may be shared by different cytokine genes, and cytokine enhancer/promoter elements usually contain several different transcription factor binding sites (Table 1). In addition, cooperative binding interactions between transcription factors in complexes occur in a number of cytokine genes and can integrate transcription factor activities(14-17). The general importance of this level of linked regulation is illustrated by transgenic knockout mice lacking members of the NFAT or NF-κB transcription factor families. The T cells from these animals exhibit dysregulated or defective expression of multiple cytokines and gross immunological perturbations (18, 19). Decay of mRNA is the third major level of potential linked regulation of cytokine expression. Rapid mRNA deadenylation and degradation are mediated by sequence motifs within the 3′-untranslated regions of mRNA that are common to many cytokine genes (20, 21). Events triggered by the stimulation of T-cell accessory molecules such as CD28, including changes in the activities of a number of RNA-binding proteins, can coordinately modulate the expression of multiple cytokines through regulation of cytokine mRNA half-lives (10, 22). Therefore, the pathways that convert TCR and accessory molecule cross-linking into an induced burst of cytokine secretion include both transcriptional and posttranscriptional mechanisms with the potential to integrate or link expression of different cytokine genes. Although this potential is clearly realized in some T cells for some combinations of cytokines, the arguments outlined below strongly support the existence of independent and flexible cytokine gene expression in T cells activated in vitro and in vivo. INDEPENDENT REGULATION OF CYTOKINE GENE EXPRESSION IN T CELLS Differences between signals. There is unequivocal evidence that established T-cell lines, T-cell clones, and individual T cells can differentially express cytokines in response to different stimuli. Both TCR-dependent and -independent stimuli can elicit dissociated expression of many cytokines, including those generally thought to be coordinately expressed such as the pairs GM-CSF and IL-3, or IL-4 and IL-5, or TNF-α and lymphotoxin-α(23-28). After TCR stimulation, cytokine expression can vary according to the strength of the stimulus, i.e., ligand dose and the number of TCR molecules triggered (23, 29-32). Qualitative effects add further diversity because altered peptide ligands can elicit different patterns of cytokine production as a consequence of activation of only a subset of signaling molecules in the TCR complex(33, 34). Different cytokine genes are also differentially responsive to the activated signaling pathways and transcription factors downstream of the TCR (Table 1). For example, inhibition of activation of the extracellular-regulated kinases within the mitogen-activated protein kinase signaling cascade(Fig. 1) can result in quantitatively distinct effects on different T-cell cytokines, with IFN-γ, GM-CSF, IL-5, and IL-10 production being substantially reduced, whereas IL-4 is least affected(9).3 Thus, rather than a simple on-off switch for predetermined patterns of gene expression, the population of TCR complexes on a T cell has the capacity to assess certain properties of the antigenic stimulus, transmit signals in a number of ways, and elicit differential responses in cytokine genes. Modes of in vitro T-cell stimulation that bypass the TCR, such as phorbol myristate acetate and ionomycin treatment, can elicit dissociated expression of many cytokines within and between the type 1 and 2 groupings(9, 24, 26, 35). Similar potential is also manifest when defined costimulatory molecules are triggered. For example, the effects of CD28 ligation on TCR-initiated cytokine production are characterized by a hierarchy of cytokine responsiveness, with IL-2 production the most augmented and IFN-γ and IL-4 production the least affected(10, 29, 36). However, it is not yet clear whether this is entirely due to TCR-independent signaling pathways triggered by CD28 ligation or whether a strength of stimulus effect via the TCR is also involved (37, 38). Therefore, the potential for integrated signaling, transcription factor activation, and mRNA turnover is counter-balanced by the flexibility inherent in the TCR complex and the diversity of the accessory T-cell surface molecules that can modulate TCR-derived signals. Differences between cells. It has been recognized for a long time that individual CD4+ and CD8+ T-cell clones can vary widely in the combinations and amounts of cytokines they secrete, even under apparently identical stimulation conditions(27, 39-44). Although clones cultured long-term often display distinct type 1 or type 2 profiles(1, 2), short-term clones derived both from polyclonal and from antigen-specific responses can show almost unlimited diversity in the combinations of cytokines they express. A feature of this diversity among the panels of clones we have analyzed is that the frequency of coexpression of any pair of cytokines was generally close to that predicted by multiplying their individual frequencies. Thus, producers of one cytokine occurred at a similar frequency among both producers and nonproducers of other cytokines, suggesting that expression of each cytokine was independently regulated (3, 41, 45). This was true for the signature type 1 and type 2 cytokines, IFN-γ and IL-4, as well as most other cytokine pairs and was observed even when the population of clones as a whole was polarized toward type 1 or type 2 cytokine synthesis. Individual cells within populations and within clones also display diverse cytokine profiles in response to the same nominal stimulus, whether assayed at the mRNA or protein level(23, 29, 43, 45-50). Again, expression of each cytokine seems to be independently regulated because each can be found with or without any other cytokine and, in at least some instances, coexpression frequencies are close to those expected for random association despite type 1 or type 2 polarization at the population level. It is important to note that random association does not imply random synthesis. Thus, as for short-term clones, the frequency at which a cytokine is synthesized within a group of cells reflects the cytokine profile of the population from which the cells were derived and can change as populations become more polarized. Apparently, expression of each cytokine gene can be independently regulated despite being subject to the influence of polarizing stimuli. This can be accommodated by a model in which polarizing stimuli act independently on each cytokine gene to influence its probability of expression(3). It also indicates that the “type 1” and“type 2” labels describe the behavior of some T-cell populations but not necessarily the cells within them. Differences between anatomical sites. Because different cytokine profiles are associated with different immune effector functions, the cytokine profiles of activated T cells in vivo might be expected to vary between tissues. Differences have been reported in the propensity of T cells from lymph nodes draining mucosal versus nonmucosal tissues to secrete IL-2 and IL-4 when activated polyclonally in vitro (51). These differences have been attributed, at least in part, to influences of local steroid hormones and to the proposed preferential migration of type 1 and 2 memory cells into nonmucosal and mucosal lymph nodes, respectively(51, 52). T-cell cytokine profiles can also vary between different compartments within the mucosal system (53, 54). In a murine model of influenza virus infection in the respiratory tract, for example, we found substantial differences in the cytokine profiles and other effector functions of CD4+ and CD8+ T cells isolated from the draining mediastinal lymph nodes, lung parenchyma, and airways(54-56). These differences were both qualitative and quantitative: T cells from the lung and airways displayed a more restricted, IFN-γ-biased profile and contained markedly higher frequencies of cytokine producers per activated (CD44high LFA-1high) T cell than those from lymph nodes. Many of the activated T cells found in the virus-infected lung are presumably derived from precursors primed in the draining lymph nodes. It is not yet clear in this influenza model whether the lung preferentially recruits T cells with certain cytokine profiles or whether local signals alter the cytokine profiles of indiscriminately recruited cells after they have reached the site. There is evidence from other studies, however, that T cells with a type 1 profile or a type 2 profile are differentially recruited into inflamed tissues: T cells with a type 1 profile selectively bound P selectin and E selectin and used these adhesion molecules to accumulate preferentially in inflamed tissues in type 1-biased, but not type 2-biased, responses (57). CONCLUDING REMARKS The interpretation of many immunological phenomena in the laboratory and the clinic is currently guided by the type 1/type 2 paradigm, which suggests that activated T cells preferentially express one of two groups of cytokines. Signals such as IL-12 and IL-4 can clearly promote the development of activated T-cell populations with polarized type 1 or type 2 cytokine profiles, and molecular mechanisms have been identified that enable such previously activated cells to express groups of cytokine genes coordinately in response to reactivation. It is therefore not surprising that some T-cell populations primed under strongly polarizing conditions in vivo and in vitro, and then reactivated by optimal TCR ligation in vitro, synthesize type 1 or type 2 cytokines as a set. During physiological responses in vivo, on the other hand, T cells are exposed to many different influences as they undergo primary activation, clonal expansion, effector cell differentiation, and reactivation by antigen. The outcome of interaction with antigen-presenting cells during priming will be affected by TCR affinity and avidity, peptide/MHC density, costimulatory signals, the duration and frequency of interaction, the activities of bystander cells, cytokines, endocrine hormones, and other soluble mediators, all of which can influence cytokine expression patterns. Therefore, even within a single tissue such as the priming lymph node, individual responding T cells may differ in the cytokine genes they express. Reactivation of primed T cells in effector sites is likewise subject to the diverse influences of antigen-presenting cells, bystander cells, and soluble factors, modified by the activation and differentiative state of the T cell itself. As discussed in this article, dissociated expression of cytokines within and between the type 1 and type 2 groups can be demonstrated in many ways, between similar cells responding to different signals and between different cells responding to the same signals. It is therefore not surprising that T-cell populations isolated from different tissue sites, and individual cells within populations, can display substantial diversity in their cytokine profiles. Such diversity in the physiological cytokine responses of T cells probably serves several purposes, ensuring the activation of multiple effector mechanisms early in an immune reaction and providing a choice of T-cell phenotypes for further differentiation, selective expansion, and preferential tissue migration if the immunogen persists. Importantly, the capacity of activated T cells to respond differentially to various signals also allows modulation of the cytokine secretion patterns of fully differentiated cytokine-producing T cells by their environment. There are many circumstances in which the selective enhancement or suppression of certain immune effector mechanisms would be clinically useful. A deeper understanding of the signals and pathways that enable dissociated as well as coordinated cytokine gene expression can be expected to suggest new ways to achieve these aims.Figure 1: Multiple levels of linked and independent regulation of cytokine gene expression in T lymphocytes. Selected major levels and pathways are shown.