A Physical Interaction between the Cell Death Protein Fas and the Tyrosine Kinase p59 fynT *

The Fas antigen (Apo1/CD95) is a transmembrane protein belonging to the nerve growth factor receptor fam-ily. It is expressed on a variety of cells, including activated T lymphocytes. Ligation of Fas with its natural ligand or with anti-Fas antibodies often results in the apoptotic death of the cell, making Fas an important mediator of down-regulating immune responses. The signal transduction pathways utilized by Fas are currently unknown, although tyrosine kinase activity has recently been strongly implicated. Here, we report that the tyrosine kinase p59 fyn physically associates with Fas in Fas-sensitive cells. In addition, we show that activated T lymphocytes from fyn knockout mice exhibit elevated lifespans and reduced apoptosis in vitro com-pared to their normal counterparts. Furthermore, activated T lymphocytes from the fyn - deficient mice are less sensitive to killing by both anti-Fas antibody and Fas-ligand cytotoxic T cells. These results suggest that p59 fyn plays an important role in Fas signal transduction. Deletion of activated peripheral T cells is important in the down-regulation of immune responses and controlling T cell numbers. The mechanisms by which this is accomplished are unclear, but the expression and action of a transmembrane death protein known as the Fas antigen has recently been implicated

Deletion of activated peripheral T cells is important in the down-regulation of immune responses and controlling T cell numbers. The mechanisms by which this is accomplished are unclear, but the expression and action of a transmembrane death protein known as the Fas antigen has recently been implicated (1)(2)(3)(4)(5). It is now believed that activated T cells eventually co-express Fas and Fas ligand, leading to the apoptotic death of the cells upon interaction with each other (see Refs. 6 -8 for recent reviews). The nature of the Fas death signal is currently unknown and has become the focus of intense research.
A cytoplasmic region in Fas that is essential for killing has been identified and termed the "death domain" (9). A similar motif exists in the cytotoxic TNF 1 receptor (TNFR1), a member of the nerve growth factor receptor family, to which Fas also belongs (10). This death domain also shares some homology with part of the Drosophila protein reaper, which is an important mediator in apoptosis (11). Both Fas and TNFR appear to be coupled to a sphingomyelinase, ceramide-producing pathway, which could explain, at least in part, the apoptotic outcome of ligating these receptors (12)(13)(14)(15). Whether this coupling is direct or not is unknown.
Fas contains no previously known signaling motifs or catalytic activities. In an effort to identify molecules that might interact with Fas and link it to signal transduction pathways, a number of groups have employed the yeast 2-hybrid system, with the Fas cytoplasmic region as "bait." This approach has identified three different proteins that can interact with Fas: FADD (MORT1) (16,17), RIP (18), and FAP-1 (PTP-BAS) (19), although evidence that these proteins interact in a physiologically relevant system is lacking. FADD and RIP both contain death domain-like motifs, and interaction with Fas may occur via dimerization of this region. The functions of FADD and RIP are not known, although RIP may be a protein kinase. FAP-1 is a tyrosine phosphatase and, interestingly, interacts with the Fas carboxyl terminus, which has been shown to be involved in negatively regulating Fas killing (9). The level of endogenous FAP-1 seems to correlate with the Fas sensitivity of a cell, and transfection of cells with FAP-1 encoding DNA also leads to a reduction in Fas sensitivity. This suggests that one or more tyrosine kinases may play a role in Fas killing. Indeed, it has previously been demonstrated that Fas activity is inhibited by tyrosine kinase inhibitors, and ligation of Fas leads to the tyrosine phosphorylation of a number of cellular proteins (20). Therefore, there is now strong evidence from two different sources that tyrosine kinase activity is involved in Fas signal transduction.
Although it has not been shown that Fas itself is tyrosinephosphorylated, we examined the context of the tyrosine residues in the cytoplasmic domains of both human and mouse Fas for any similarities to those in other signaling molecules that are known to interact with tyrosine kinases. Interestingly, one of the tyrosines is found in the death domain within a conserved YXXL motif. This sequence is reminiscent of half of an immunoreceptor tyrosine-based activation motif (ITAM) found in the T cell receptor CD3 complex chains and elsewhere (see Ref. 21 for a review) and is also suggestive of a binding site for proteins with SH2 domains (22). ZAP-70 and p59 fyn are two tyrosine kinases known to interact with similar sequences, although p59 fyn is much more flexible in its binding sequence requirements than ZAP-70 (23). Also, since disruption of the fyn gene has been correlated with lack of programmed cell death in certain regions of the mouse brain (24) (which, along with T lymphocytes, expresses the highest level of p59 fyn ), we decided to investigate the possibility that p59 fyn might interact with Fas and contribute to Fas signal transduction. Here we report that there is indeed a specific physical interaction be-tween p59 fyn and Fas in Fas-sensitive cell lines. The biological relevance of this interaction was confirmed by the observation of an increased survival and decreased rate of apoptosis in alloantigen-stimulated splenocytes from fyn knockout mice. Furthermore, fyn-deficient activated T cells are resistant to killing when their surface Fas is ligated by either anti-Fas antibody or Fas ligand presented by Fas-dependent cytotoxic T cells. This provides the first report of a protein with known catalytic activity capable of linking Fas to a signal transduction pathway that has been shown to occur in a physiologically relevant system.
Immunoprecipitates were washed extensively in lysis buffer without EDTA and either dissociated in SDS sample buffer, separated by SDSpolyacrylamide gel electrophoresis (8%), and electroblotted onto nitrocellulose, or subjected to in vitro kinase assays. Western blotting was performed using either rabbit-anti-fyn antibody (FYN3, Santa Cruz Biotechnology) or rabbit anti-Fas peptide Fab 2 fragment (kindly provided by Dr. K. Elkon, Ref. 28) and horseradish peroxidase-conjugated donkey anti-rabbit (Amersham) and visualized using the ECL method (Amersham).
Analysis of DNA Fragmentation-DNA fragmentation gels were done as described previously (29). Genomic DNA was isolated from 10 6 Ficoll-purified splenocytes from MLCs on the days indicated and electrophoresed in 1% agarose gels in Tris-borate-EDTA buffer. DNA was visualized by ethidium bromide staining and UV fluorescence.
Determination of Sensitivity of Cells to Fas-mediated Killing-C57BL/6 and Fyn Ϫ/Ϫ splenocytes were examined for cell viability and sensitivity to PMM-1 induced cell death after stimulation with irradiated Balb/c splenocytes (1:1 ratio). On day 5 poststimulation, the splenocytes were exposed to either cross-linked anti-Fas antibody or PMM-1 cells pretreated with PMA and ionomycin. The cell viability assays were carried out in 96-well microtiter plates in the following manner. The antibody was diluted to a final concentration of 40 g/ml in PBS, then 50 l of this suspension was added per well. Plates were allowed to sit for 2 h at room temperature or at 4°C overnight before being washed twice with 100 l of PBS. Triplicate wells with and without anti-Fas antibody containing 2 ϫ 10 5 cells each were incubated for 24 h at 37°C. Cell viability was determined by staining with eosin or trypan blue.
Standard cytotoxicity assays were carried out with PMM-1 effector cells stimulated with PMA (10 ng/ml) and ionomycin (3 g/ml). Effector cells were stimulated for 3 h at 37°C and then washed in RPMI media. Target cells, C57BL/6 or Fyn Ϫ/Ϫ splenocytes were labeled with chromium sulfate for 60 min at 37°C before being washed in RPMI and PBS. Assays were set up in 200-l volumes containing a maximum of 5 ϫ 10 4 effector cells and 1 ϫ 10 4 target cells. To examine the impact of effector to target cell ratios, the concentration of the PMM-1 cells was serially diluted from 5 ϫ 10 4 cells per well to 0.625 ϫ 10 4 cells per well.
The assays were incubated for 4 h at 37°C after which the cells were pelleted and the amount of 51 Cr in the supernatants was measured in a LKB Rack Gamma Counter. Cell-specific lysis was calculated to determine the relative amount of cell death.

RESULTS AND DISCUSSION
To determine if p59 fyn associates with the Fas antigen, coimmunoprecipitation experiments were performed. We chose to study YAC-1 cells which express relatively high levels of both Fas and p59 fyn2 and are very sensitive to killing by PMM-1, a killer cell that has been shown to kill via a Fas-dependent pathway (25). Either Fas or p59 fyn were immunoprecipitated from extracts of YAC-1 target cells. Western blot analysis revealed that p59 fyn was co-immunoprecipitated with anti-Fas antibody, but not with an isotype-matched control or antibodies to another cell surface molecule CD45 (Fig. 1). Interestingly, ligation of Fas on YAC-1 cells does not appear to be required for the association of p59 fyn , as we obtained the same results from YAC-1 cells preincubated in the presence or absence of PMM-1 Fas-dependent killer cells (data not shown).
In order to establish the reciprocity of this interaction, extracts were first immunoprecipitated with either anti-Fas or anti-fyn antibodies. The resulting precipitates were then probed with both antibodies. As can be seen in Fig. 2, fyn is present in the Fas precipitates, and Fas can be detected in the anti-fyn precipitated sample. Together, these experiments establish an association between Fas and fyn.
Elimination of activated T cells is believed to occur via Fas/ Fas ligand interactions (1)(2)(3)(4)(5)(6)(7)(8). Therefore, in order to determine if p59 fyn plays a role in Fas function, we studied mixed lymphocyte cultures (MLCs) of splenocytes obtained from normal and fyn Ϫ/Ϫ mice (26). Throughout the early stages of MLC, the T cells from normal and fyn Ϫ/Ϫ mice exhibited nearly identical growth characteristics (Fig. 3A). [ 3 H]thymidine uptake experiments also indicated comparable levels of proliferation within the early stages of culture (data not shown). However, by day 17 after stimulation, the cultures of fyn-deficient T cells had substantially more viable cells than control cultures, indicating an impairment in the normal (presumably Fas-based) mechanisms of activated T cell deletion (Fig. 3A). Additionally, analysis of several MLCs indicated an increase in fyn Ϫ/Ϫ lymphocyte lifespan manifested as a greatly elevated cloning efficiency 2 E. A. Atkinson and R. C. Bleackley, unpublished observations.

FIG. 1. Antibodies to the Fas-antigen immunoprecipitate fyn.
Western blot showing that p59 fyn specifically associates with Fas. Various antibodies were used to precipitate material from YAC cell lysates. Material bound to protein A/G-agarose was fractionated on an 8% polyacrylamide gel, and blotted with anti-fyn antibody. Anti-Fas and anti-fyn antibodies immunoprecipitate detectable levels of p59 fyn from cell lysates, but an isotype-matched control and anti-CD45 antibodies do not. Even extreme overexposure of the filter did not reveal any fyn in the CD45 immunoprecipitates.
Interaction of p59 fyn with Fas 5969 (Fig. 3B). A representative cloning experiment is shown. In 3 other cloning experiments, the actual number of clones obtained varied, but an enhancement in cloning efficiency of the fyn Ϫ/Ϫ T cells of approximately 5:1 was consistently observed. This apparent increase in fyn Ϫ/Ϫ T cell lifespan was closely paralleled by a reduction in the level of apoptosis as assessed by characteristic DNA fragmentation (Fig. 4). No discernable differences in the relative amounts of Fas surface expression on fyn Ϫ/Ϫ and control splenocytes was observed upon flow cytometric analysis (data not shown). These data are all consistent with a role for p59 fyn in Fas signal transduction.
Finally, we sought to directly determine if fyn played a role in Fas-mediated cell death. To do this, we utilized alloantigenstimulated splenocytes from fyn Ϫ/Ϫ and fyn ϩ/ϩ mice and determined their susceptibility to anti-Fas antibody. Wells of 96-well plates were coated with either anti-Fas or bovine serum albumin as a control, and day 5 alloantigen-activated splenocytes were incubated in them overnight. The next day, cell viability was assessed by vital dye exclusion. As seen in Fig. 5A, the fyn Ϫ/Ϫ splenocytes were resistant to the effects of anti-Fas antibody while the control cells were susceptible. We also determined the relative sensitivities of fyn Ϫ/Ϫ and fyn ϩ/ϩ splenocytes to killing by PMM-1. Normal and fyn-deficient splenocytes were used as targets in a 51 Cr-release lytic assay. The fyn Ϫ/Ϫ cells were found to be far less sensitive to PMM-1 than their normal counterparts (Fig. 5B). These results provide strong and direct evidence that fyn activity is involved in the Fas death pathway in these cells.
We are currently working on determining how the fyn tyrosine kinase interacts with Fas. As mentioned previously, the conserved YXXL motif in the death domain of Fas is intriguing in that it is very similar to ITAM sequences involved in signal transduction via the T cell receptor-associated CD3 chains. We have so far not been able to demonstrate tyrosine phosphorylation of Fas. Furthermore, to our knowledge, no death domaincontaining protein has ever been shown to be phosphorylated on tyrosine, even though many of these proteins, including TNFR1 and RIP, possess conserved YXXL motifs in their death domains. 3 Clearly, the importance of these sequences in binding fyn and/or relaying an apoptotic signal needs further study.
It should be noted that lack of phosphorylation of the tyrosine residue within the YXXL motif of Fas would not preclude the possibility that p59 fyn interacts with Fas at this sequence. These sequences can bind tyrosine kinases in at least two different ways: via SH2 domain interactions with the phosphorylated tyrosine residues or via a mechanism independent of both SH2 domains and tyrosine phosphorylation. An intact tyrosine activation motif consists of 2 YXXL sequences separated by approximately 10 amino acids. Binding of ZAP-70 to the CD3-⑀ ITAM has recently been shown to be dependent on very specific sequence criteria (23). These include the requirement for an intact ITAM, accurately spaced tyrosine residues, and prior phosphorylation of the tyrosine residues. Binding of ZAP-70 to the ITAM is likely achieved via its 2 SH2 domains (23,30,31). Binding of fyn to an ITAM is much more flexible, 3  Interaction of p59 fyn with Fas 5970 however. Even mutating the tyrosine residues in the ITAM or altering their spacing does not affect the interaction of p59 fyn , which is thought to be initially mediated by its unique aminoterminal region (23). Given the plasticity of the sequence requirements for fyn binding to YXXL-containing motifs, it is tempting to speculate that this region of Fas could represent an "ITAM-equivalent" for fyn interaction.
It is interesting to note that in some circumstances, Fas ligation can cooperate with T cell receptor/CD3 stimulation in signaling T cell mitogenesis (32), and that T cells from Fasdefective lpr mice exhibit reduced responsiveness to antigen (2). Our observation that p59 fyn binds to Fas provides an explanation for both of these phenomena, by identifying a specific tyrosine kinase that is common to the two systems. Fas-bound p59 fyn may, under some conditions, be brought into close proximity with the CD3 chains and may be able to contribute additional tyrosine phosphorylation of CD3 or other complexed proteins, leading to enhanced T cell stimulation. It will be interesting to determine if the lpr mutation disrupts the association between Fas and p59 fyn .
Although we believe that p59 fyn plays an important role in Fas signal transduction, it is clearly not absolutely essential. Apoptosis still occurs, albeit at a reduced rate, in activated fyn Ϫ/Ϫ splenocytes. It is likely that other tyrosine kinases may be able to partially compensate for p59 fyn in its absence, as may be the case in CD3 signal transduction (26). To our knowledge, fyn-deficient mice have not been shown to exhibit an lpr-like phenotype, with large numbers of T lymphocytes accumulating with age. This has not yet been specifically studied, however, and our data would predict that in older fyn Ϫ/Ϫ mice, there may be some degree of lymphoproliferative disease apparent. Our observations are also important in that they predict that some types of human lymphoproliferative and autoimmune diseases could result in mutations that affect fyn-Fas interactions or p59 fyn activity. Clearly, the identification of a physiologically relevant kinase associating with the apoptosis-inducing Fas antigen opens up numerous new avenues of investigation for elucidating the entire nature of the death signal.