Interferons and cytokines: where we stand after the first joint meeting of the international cytokine society and the international society for interferon and cytokine research (Geneva, October 6-10, 1996)

ARTICLE

INTERFERONS, INTERFERON RECEPTORS: GENE REGULATION AND SIGNAL TRANSDUCTION

Dr. I. Gresser gave an historical and personal philosophical view of interferon discovery, properties, of its role in host defense and the immune system and its involvement in pathological situations. He presented strong arguments supporting the notion that the major physiological role is "after all antiviral defense".

Dr. Georges Stark presented an overview and new data on the genetic analysis of mammalian signaling pathways. The use of mutant cell lines lacking proteins required for IFN-alpha and IFN-gamma signaling provided a great amount of information. Analysis of U6A cells lacking STAT2 allowed to show that IFNAR2-2 subunit of the IFN-alpha receptor is pre-associated with STATs 1 and 2 and that, following activation with ligand, STAT2 binds through its SH2 domain to phosphotyrosine 466 of IFNAR-1 where it becomes phosphorylated itself. STAT1 then uses its SH2 domain to bind to the STAT2 phosphotyrosine, followed by phosphorylation of STAT1 and eventual dissociation of the STAT1-STAT2 heterodimer. NH2 terminal of STAT2 is involved in specificity but STAT1-NH2 terminal is required for the interaction with the receptor.

Dr. A. Kimchi gave a lecture on cytokine mediated cell cycle arrest and programmed cell death. The identification of the genes that initiate or execute programmed cell death is a major challenge. A functional approach of gene cloning was applied to HeLa cells in an attempt to isolate positive mediators of programmed cell death induced by IFN-gamma. The approach was based on random inactivation of genes by transfection with antisense cDNA expression libraries prepared from cytokine-treated cells, followed by the selection of cells that survived in the presence of IFN-gamma. This approach led to the identification of three novel genes (named DAPs for Death Associated Proteins) that function as positive mediators of IFN-gamma-induced cell death. In addition, an antisense cDNA fragment identical to human cathepsin D aspartic protease was rescued by this positive selection. One of the DAP genes coded for a novel Ca²⁺/calmodulin-dependent serine/threonine kinase that carries eight ankyrin repeats and the death domain and is localized at the cytoskeleton. The two other genes directed the synthesis of a basic proline-rich 15 kDa and a 46 kDa protein (that carries a p loop motif) is a nucleotide binding protein. The DAP kinase is widely expressed in normal tissues but the mRNA and the protein were below detection limits in 70% of human B cell lymphoma. 5-azadesoxycytidine experiments suggest that loss of expression was due to gene silencing by specific methylation. Interestingly, Lewis lung carcinoma is DAP kinase negative. Transfection with the DAP kinase wild gene reduced the growth of the primary tumor and the metastatic capacity was suppressed.

Silverman presented data showing that the (2'-5') oligoadenylate-dependent RNAse L activity contributes to apoptosis by facilitating the degradation of RNA and suggests that this mechanism contributes to antiviral protection induced by IFN via the induction of the (2'-5') oligoadenylate synthase. Sangfelt showed that induction of apoptosis and inhibition of cell growth induced by IFN-alpha are two independent responses and that susceptibility to these IFN effects varies between cell types.

R. Schreiber gave a lecture on the specificity and in vivo relevance of IFN signaling through the JAK/STAT pathway. He presented an overview of the structure-function relationships that are operative within the two subunits of the IFN-gamma receptor. Results from his laboratory and others (Aguet, Pestka) allowed to identify the JAK1 and JAK2 binding sites and the STAT1 docking site. Interestingly, JAK1 and JAK2 kinases are associated with the receptor prior to activation. Upon activation by the ligand, they became phosphorylated and phosphorylate the receptor triggering STAT1 association, its phosphorylation and its dimerization. Generation of JAK1 and STAT1 deficient mice allowed to define the physiological significance of the in vitro observation. STAT1 deficient mice show no developmental defects but display a global and specific inability to respond to IFN-gamma. Surprisingly, no defects in any other cytokine signaling pathway (such as GH, EGF, IL-1) was observed, indicating that in vivo STAT1 specifically mediates the biological response to IFNs. STAT1 appears to be an obligatory requirement for IRF-1, GBP, CIITA induction. STAT1-deficient knock-out mice are extremely susceptible to viral (VSV) and microbial infection (Listeria). Thus, in vivo STAT1-mediated signaling is specific for the IFN system. Target disruption of the murine JAK1 gene led to perinatal lethality and abnormal thymic development, decreased birth-weight and failure to nurse neonates. Embryonic fibroblasts from JAK1 -/- mice display defective responsiveness to IFN-alpha and IFN-gamma, responsiveness for MHC I induction. JAK1 is required for all IFN-induced biological properties but it is also important for signaling by receptors that use the gp130. STAT1 appears to be an obligatory requirement for IL-7 response in lymphocyte development. The STAT1 -/- mice show normal response for GM-CSF/MCSF myeloid development but drastic defect for IFN-gamma and LIF signaling was observed in macrophages.

E.F. Petricoin et al. presented results implying that the protein tyrosine phosphatase CD45 is required for the antiproliferative actions of IFN-alpha in Jurkatt cells. Whereas IFN-alpha stimulation of JAK/STAT pathway and antiviral activity were the same in wild type or CD45 negative cells, the antiproliferative effect of IFN-alpha was completely abrogated in CD45 negative cells. In T-cells, IFN-alpha-induced association of ZAP70 tyrosine kinase with the IFN receptor was found to be dependent of CD45 expression. Several intracellular tyrosine phosphatases are implicated in signalling by cytokine receptors, and this first report of an association with a transmembrane tyrosine phosphatase is opening new views in the understanding of cytokine receptor activation.

J.K. Riley transferred the IL-10 receptor docking sites to Ba/F3 cells expressing a truncated IFN-gamma receptor alphachain, which retained the JAK1 binding sites but lacked the STAT1 docking site. Ba/F3 cells expressing the chimeric receptor stimulated with IFN-gamma resulted in STAT3 activation and cell proliferation, indicating that STAT3 activation is directly linked to the induction of at least some IL-10-dependent biological activities.

B. Willman reported results on the IFN-induced double-stranded RNA activated Ser/thr protein kinase (PKR). PKR regulates gene transcription by phosphorylating IkB. In cells lacking PKR, dsRNA also fails to induce NFkB binding to its target sequence. Binding can be restored by transfection of wild type PKR. NFkB induction in response to TNF or LPS is normal in PKR -/- cells. Embryo fibroblasts deleted in PKR also show signaling deficiency in response to IFN-gamma since IRF-1 is not induced, although STAT1 binding to its target sequence is unaltered.

Results from several laboratories (Uzé, Rubinstein, Colamonici, Petska) have shown that type I IFN receptor is a multi-subunit receptor of the cytokine superfamily. Type I receptor consists of three membrane spaning subunits IFNAR-1 (alpha chain, Uzé), IFN-R2 (ß chains) also called IFNAR-2alpha (ß short, Rubinstein) and IFNAR-2ß (ß long, Colamonici). The last two chains are splice variants, having identical extracellular domains, while the cytoplasmic domain of IFNAR-2ß is longer. Co-expression of IFNAR-1 and IFNAR-2ß reconstitute a functional type I receptor. Antibodies developed against IFNAR-2 block antiviral activity without affecting the induction of ISRE-dependent genes, suggesting that an additional pathway is required for induction of antiviral activity.

O. Colamonici. The type I IFN receptor (IFNAR) is formed by at least two chains termed alpha and ß subunits. The alphasubunit has a Mr of 110 kDa while the ß subunit a short and a long forms with Mrs of 55 and 95-100 kDa, that are generated by alternative splicing of the same gene. Co-expression of the alphachain and ß chain 95-100 kDa reconstitutes the antiviral response. Mapping of the cytoplasmic domains of IFNAR-2 long chain demonstrates that JAK1 binds to amino acids 300-346. This chain also directly interacts with STAT2 (amino acids 376-462) and indirectly with STAT1. Colamonici showed that only the first 82 (amino acids 265-376) of the cytoplasmic domain are required to elicit an antiviral response to type I IFN.

S. Pestka, with the use of chimeric receptors involving the IFN-gamma receptor complex as a model system, demonstrated that JAK2 activation is not an absolute requirement for IFN-gamma signaling since other kinases can substitute JAK2.

M. Aguet reported about the orphan receptor subunit CRF4, whose gene is located near the two type I IFN receptor genes within a gene cluster which also contains the ß subunit of the type II IFN receptor. However, analysis of ES cells with a homozygous disruption of the CRF-4 gene showed that IFN-alpha and -gamma responses are not altered, suggesting that CRF2-4 is not an essential component of IFN receptors but may be part of another cytokine receptor system.

J.L. Casanova reported the first evidence for a human inherited IFN-gamma receptor deficiency in a child who died after BCG infection, following vaccination. The genetic defect, inherited as an homozygous trait consists of a single base deletion in the IFN-gamma receptor alpha chain, leading to a stop codon and a lack of detectable cell surface protein and IFN-gamma binding. These results point for the unreplaceable role of IFN-gamma receptor-mediated activation pathway in the control of mycobacteria in man.

Croze showed that Daudi cells stimulated with IFN-ß but not IFN-alpha induced in a few minutes the tyrosine phosphorylation of IFNAR-1 and a protein of 100 kDa. This phosphorylated protein co-immunoprecipitate with IFNAR-1 and was recognized by anti-IFNAR-2 antibodies. Evidence was presented that the 100 kDa phosphorylated protein was IFNAR-2. Since IFNAR-2 did not co-immunoprecipitate with IFNAR-1 upon IFN-alpha binding, it can be suggested that IFN-alpha and IFN-ß induce different conformation changes after interaction with the receptor, leading to variation in the IFN-alpha and IFN-ß signaling pathway.

STAT proteins comprise a family of related proteins containing 7 members. STATs 1 and 2 are activated by tyrosine phosphorylation in IFN-alpha and IFN-gamma signalling, while other STAT proteins mediate signalling by a wide variety of other cytokines. Structure/function analyses of STAT proteins have identified several discrete functional domains including SH2, SH3-like, DNA-binding and transcriptional activation domains but has failed to reveal the function of the N-terminal one-half of theses proteins. Previously published results revealed that an alternatively spliced product of STAT1 lacking the carboxy-terminal 38 amino acid also lacks transactivation ability. Based on this observation, Pisharody and Levy have further characterized the STAT1 carboxy-terminus. Using chimeric proteins based on Gal4 fusion and a UAS reporter assay, they identified the minimal transactivation domain of STAT1. Similar studies with other STAT proteins have demonstrated transactivation function in the carboxy-termini of these proteins as well. Interestingly, the carboxy-termini of the STAT proteins are the least conserved region and the sequence divergence is reflected in widely varying degrees of transactivation, with the STAT2 transactivation domain being two hundred-fold more potent than that of STAT1. The minimal transactivation domain of STAT1 and STAT2 have been mapped. These domains do not resemble any previously described transactivation motifs.

INTERLEUKINS, INTERLEUKIN RECEPTORS AND SIGNAL TRANSDUCTION

Many results reporting the characterization of signalling proteins that associate to the IL-1 receptor or to the TNF receptors/FAS family members were presented at the ICS/ISICR joint meeting, in Geneva last October. Immunex scientists also described the resulting phenotypes of mouse gene inactivation for a number of these receptors and ligands (TNF-Rec.I and II, CD30, TNF-ß, CD40 lig.). A new member of this cytokine family, TRAIL (TNF Related Apoptosis Ligand) displays some interesting antitumoral and antiviral protective effects in vitro, that are mediated through a specific receptor with a restricted tissue distribution. In contrast to its relative cytokines, TRAIL did not seem to display toxic shock-like syndrom when injected to mice, which might represent an important feature for this molecule.

Many presentations were also devoted to the signalling pathways activated by the cytokines belonging to the IL-6 family. Tyrosine residues located in the intracellular part of the common gp130 receptor signalling protein have preferential interactions with the subsequent relaying components of this pathway. Activation of gp130 by its ligands leads to the recruitment of STAT1 and STAT3 signalling proteins, and P. Heinrich (Germany) has shown a preferential interaction of STAT1 with the two more distal tyrosine residues of the gp130 receptor (Tyr 905 and 915), whereas STAT3 can indifferently contact the last 4 tyrosine residues of gp130. T. Kishimoto (Japan) also reported the possibility for Tyr 759 to recognize the Grb2 adaptor and participate to the activation of the MAP kinase pathway. Activation of the MAP kinase pathway, but also STAT3 activation, were found to be sensitive to the H7 serine/threonine phosphorylation inhibitor, suggesting a tight intrication between these 2 signalling cascades.

Deletion mutants of intracellular parts of gp130 revealed that the first 65 residues starting from the membrane were sufficient to get a transient proliferative response to IL-6, but that a minimal portion of 141 residues was required to have a long term response to the cytokine. These results are in sharp contrast with the recruitment of the distal tyrosine residues of gp130 for signalling, and suggest the implication of additional adaptator proteins in the STAT and MAP kinase pathway. In line with these observations, an interesting presentation was done by R. Collum, from Columbia University. By using a yeast two-hybrid system to screen for proteins that interact with the N-terminal half of STAT3, he has isolated a novel interacting protein named StIP-1 (for STAT Interacting Protein) with an apparent M.W. of 92kD and which is widely distributed. StIP-1 can be co-precipitated in the cells with the non-phosphorylated forms of STAT1, STAT3 and STAT5, but also with the JAK kinases activated in the IL-6 signalling pathway. In addition, over-expression of a truncated form of StIP-1 inhibits the functional responses and behaves as a dominant negative regulator. Thus, although the model of STAT docking to receptor P-Tyr cannot be discarded, it may not represent the whole story and StIP-1 may serve as the intermediary between the JAK kinases and STAT proteins.

The gp130 signalling cascade and its in vivo relevance was further studied with the help of gene inactivation experiments in the mouse. STAT3 K.O. mice were presented by T. Kishimoto. The resulting phenotype in homozygous animals is very severe and all the embryos die between day 5 and 8 of gestation, rendering the subsequent studies very difficult to develop.

Cao (San Francisco) described a kinase associated with IL-1R, called IRAK (Interleukin-1 Receptor Associated Kinase). In association with TRAF6, it is implicated in the activation of NFkappaB. Two other molecules associated with this signal were also reported by Sims (Seattle, W). One of these molecules, ITAK, displays GDP-Exchange Factor sequence homology, and may be involved in the regulation of G proteins (yet to be defined). The same group used the two-hybrid technique to isolate a molecule associated with the IL-1 receptor. This molecule (IIP1) contains a GAP region and overexpression of this protein blocks the activation of NFkappaB by IL-1. The ITAK and IIP1 molecules may regulate one or more G proteins (although they have not been identified) implicated in IL-1 signalling, and they may regulate in opposite directions. Along the same lines, Matthews (Dublin, Ireland) reported a direct relationship between the activation of Rac1 and p38 MAP Kinase by MEK3 during induction of the IL-2 gene by IL-1 in T lymphocytes. Wallach (Rehovot, Israel) used the two-hybrid technique to characterize the NIK molecule, which carries a kinase motif and is associated with TRAF2 during TNF signalling. This kinase can autophosphorylate and its suppression activates NFkappaB.

New substrates for the JAK kinases were reported. Using JAK1-deficient cells, Kerr (London, GB) showed that the activation of IRS1 and P13K by IL-4 involves the kinase JAK1. Only the activation of IRS1 can be partially re-established by the overexpression of TYK2. Kinases other than JAKs can phosphorylate and thus activate STATs. Pfeffer (Memphis, TN) showed that IFN activation induces an association between STAT3 and PI3K, and suggested that STAT3 Ser phosphorylation may result from the activity of PI3K as it is inhibited by wortmannin. The list of transcription factors able to dimerise with STAT molecules is growing. Kerr (London, GB) reported that IFN activation can cause heterodimerisation between STAT3 and c-Jun, and between STAT1 and SP1 and ISGF3g. Furthermore, as recently published in Nature, STATs interact with activated steroid receptors leading to mutual inhibition of transcriptional activity.

CYTOKINES AND HEMATOPOIESIS

Although presentations dealing with the effect of cytokines on hematopoiesis were relatively rare, the three studies examining the effect of chemokines on lymphoid and myeloid maturation provided some interesting perspectives to this field. Thus, T. Kishimoto reported on the ligand of mouse FUSIN (CXCR4) termed SDF-1 (Stromal-Derived Factor-1) which acts in synergy with IL-7 on the generation of B cell progenitors. In accordance with this notion, he showed that SDF-1 knock-out mice have a defect in B lymphopoiesis, while myelopoiesis is normal in fetal liver but not in adult bone marrow.

B.L. Kreider described a new ß-family chemokine (C-C), most abundantly expressed in hematopoietic tissues, which is identical to the recently characterized chemokine HCC1. It specifically inhibits M-CSF-mediated colony formation and the development of the monocytic lineage in response to IL-3, SCF, GM-CSF, TPO and EPO from both mouse and human bone marrow progenitors. This chemokine does apparently not affect the function of normal mature macrophages (oxidative burst, cytokine production...). Two other novel ß chemokines have been investigated by V.P. Patel for their inhibitory function on myeloid progenitors. They are both strongly homologous to MIP1alpha and have been termed MPIF-1 and MPIF-2 (Myeloid Progenitor Inhibitory Factor). The two molecules elicit chemotaxis of peripheral T-cells and have no effect on monocytes and granulocytes. MPIF-1 is a potent suppressor of murine LPP-CFC (Low Proliferative Potential Colony Forming cells) with granulocyte monocyte differentiation potential, while MPIF-2 strongly inhibits the development of more primitive murine HPP-CFC (High Proliferative Potential-CFC).

In addition to these studies on chemokines, some other presentations can be briefly summarized by the following points: 1) direct effect of IL-11 on megakaryocytopoiesis, independent from TPO; 2) characterization of the NFS-60 cell line as a potent producer of cytokines (IL-6, IL-4) whose cytokine mRNA expression pattern is similar to that of normal bone marrow progenitors purified on the basis of high rhodamine retention and 3) mimicry of the effect of EPO by small peptides whose primary sequence is not related to that of the growth factor.

CYTOKINES AND APOPTOSIS

This session comprised several important presentations on apoptosis-inducing ligands of the TNF/FasL family. A new member of this family, TRAIL (TNF-Related Apoptosis-Inducing Ligand) has been recently cloned by S.R. Wiley at Immunex. This molecule is widely expressed in human tissues, including spleen, thymus, prostate and lung. It promotes apoptosis in a variety of cell lines and virus-infected primary cells, acting in synergy with IFN-gamma. The expression of its receptor is restricted to a few cell types (not discussed during this meeting), mainly after induction. TRAIL possesses 30% homology with FasL, but conversely to the latter, its injection does apparently not induce toxicity.

Generally speaking, it is becoming increasingly clear that the outcome of the signalling through the TNF-R/Fas family depends on the activation state of the cells, resulting either in growth and/or differentiation or growth arrest and/or apoptosis.

It is now well established that IFN-gamma is also a potent inducer of apoptosis. In this meeting, A. Kimchi reported on four new molecules that act as positive mediators of IFN-gamma-induced cell death. These are 1) DAP 1 (Death Associated Protein 1) which is a small proline-rich cytoplasmic protein; 2) DAP kinase, a novel Ca²⁺/calmodulin-dependent serine/threonine kinase; 3) DAP 3, a 46 kDa nucleotide-binding protein and 4) the well-known aspartic protease, cathepsin D. DAP kinase is widely expressed in normal tissues and in non-tumorigenic cell lines, while it is below detection in a large percentage of human B lymphoma cells and bladder carcinoma cell lines. Restoration of DAP kinase to highly metastatic lung tumor cell clones strikingly reduced their metastatic activity, suggesting that this molecules may function as a tumor suppressor gene.

Some other data belonging to this field of investigation should also be pointed out, such as 1) requirement of STAT-1 for IFN-gamma-induced ICE expression; 2) Fas expression on glioma cells and evidence for their sensitivity to FasL; 3) resistance to apoptosis resulting from the deficiency or blockade of (2'-5') oligoadenylate-dependent RNase L and 4) the demonstration that apoptosis and cell cycle arrest represent two distinct responses to IFN, depending on the nature of the target cell.

CONCLUSION