In this study, we investigated the function of EphB6, a member of the Eph family of receptor tyrosine kinases, in the immune system. EphB6 shares about 30% amino acid identity with other EphB family members (18
), but human EphB6 and mouse EphB6 have more than 90% amino acid similarity (18
). This suggests that EphB6 must have important conserved functions, even though it has no kinase activities because of a mutation in its kinase domain (18
). Indeed, we showed that EphB6 was essential in T cell function.
We have demonstrated here that EphB6–/–
T cells were defective in their response to TCR stimulation in vitro and in vivo. One might argue that the compromised T cell response seen in EphB6–/–
mice is due to defective T cell development, not to lack of EphB6 expression on the mature T cells. Our data show that this is unlikely: when T cells from normal individuals were sorted by flow cytometry according to EphB6 expression, EphB6–/–
T cells responded poorly to anti-CD3 and anti-CD28 stimulation, compared with EphB6+/+
T cells (19
), which suggests that EphB6 is essential for the function of normally developed T cells. We previously assessed the expression of EphB6 and its ligands in both CD45RA+
T cells. Although the expression on CD45RA+
cells was higher than that on CD45RO+
cells (58% versus 28%), they had similar proliferative response to anti-EphB6 costimulation (19
). This suggests that both memory and naive T cells require EphB6 for optimal function.
Although the T cell proliferation and lymphokine production in vitro, and the T cell–mediated cellular immune response in vivo (such as DTH and EAE induction), were compromised in the EphB6–/–
mice, the T cell–dependent humoral response, such as anti-TT IgG production and serum IgG, IgA, and IgE levels, in these mice was not defective. A likely explanation to this seemingly contradictory observation is that the T cell functions are reduced but not totally ablated in the EphB6–/–
mice, and the remaining capacity of the T cell function is sufficient to support the T cell–dependent humoral response. The T cell–independent Ab production, as shown by the serum IgM level, in the EphB6–/–
mice was also normal, further indicating that EphB6 is not essential for B cell function. This is consistent with the low expression of EphB6 and EFNBs on B cells (19
). We attempted to investigate the Th1 versus Th2 differentiation of EphB6–/–
T cells in vitro. However, as these T cells could not proliferate well, they could not be driven into either Th1 or Th2 status in vitro using conventional protocols. As an alternative, we tested their serum Ig isotypes, including IgE. These parameters in the EphB6–/–
mice were comparable to those in the WT mice, which suggests that the lack of EphB6 does not skew the Th1/Th2 balance.
Recent findings based on anti–IFN-γ Ab administration or IFN-γ– or IFN-γ receptor–null mutation indicate that IFN-γ has a protective role in EAE induction (39
). In our study, although IFN-γ secretion by EphB6–/–
T cells was reduced, EAE induction in the EphB6–/–
mice was diminished. How do we reconcile these 2 phenomena? It is possible that the degree of IFN-γ reduction in EphB6–/–
mice is not as severe as in those published EAE models, in which IFN-γ or IFN-γ receptor levels were increased or suppressed on more drastic scales; moreover, EphB6–/–
mice have other defects, such as decreased IL-2 production and T cell proliferation. As a result, the observed decrease of EAE severity in EphB6–/–
mice is the sum of all these defects and is not only related to the reduced IFN-γ level.
We demonstrated, using EphB6–/–
T cells, that all 3 EFNBs could trigger T cell costimulation via EphB6; such an effect of the EFNBs on EphB6 was only revealed in the presence of soluble EphB4, which could block the interaction between the plate-bound EFNBs and EphB4 on the cells. It is to be noted that even in the absence of EphB6 and the blocking of EphB4, the inhibition of the costimulation by EFNBs was still not complete, which suggests the involvement of other Ephs in EFNB-triggered costimulation. These results demonstrate that, as with other Eph kinases, EphB6 binds multiple ligands for its function in T cells; they also suggest that the EFNBs can exert their costimulation through EphB6 as well as other Ephs. As EFNB family members are cell surface molecules and can function as receptors to reversely transduce signals into cells using their normal receptors, Eph kinases, as ligands (3
), the phenotype of the EphB6–/–
mice could, in theory, be caused by either or both of the following 2 mechanisms: (a) lack of EphB6 as a receptor to receive stimulation from its ligands (e.g., EFNB1, EFNB2, and/or EFNB3); or (b) lack of stimulation from EphB6 to EFNB1, EFNB2, and/or EFNB3. The immunological phenotype of the EphB6–/–
mice is likely due to the former mechanism, since we have proven that mAb or the EphB6 ligand EFNB2 on wells can costimulate T cells (19
) but EphB6 on wells failed to have any effect on T cells (data not shown).
We have shown here that after TCR activation, EphB6 translocated into aggregated rafts, to which TCRs also migrate. This provides a morphological basis for the signaling pathways of EphB6 and TCR to interact. Moreover, rafts function as a scaffold for many signaling molecules; the migration of EphB6 to rafts may allow it to interact with these signaling molecules for its signaling. We examined the raft aggregation after TCR activation in EphB6+/+ and EphB6–/– T cells, but no significant difference was found (data not shown); this indicates that EphB6 is not essential for raft aggregation. The movement of rafts during TCR activation is a cytoskeleton-dependent process. We did not find any abnormality in actin polymerization in EphB6–/– T cells during TCR ligation (data not shown); further, EphB6 cross-linking alone did not result in apparent actin polymerization, which suggests that the essential function of EphB6 is not related to cytoskeleton reorganization.
Through our study, a putative model is emerging to explain the mechanism of EphB6 costimulation of T cells. It seems that EphB6 is an essential component of the TCR signaling complex, which is now often referred to as the TCR signalosome. During T cell activation, EphB6 is recruited to the raft in which the signalosome resides; ZAP-70, LAT, PLCγ1, SLP-76, and TCR are all interconnected in the signalosome. As Grb2 is physically associated with EphB6 (25
) and LAT (36
), it might function as a bridge connecting EphB6 and the signalosome. We showed that EphB6 was essential for activation of ZAP-70, which is the kinase responsible for LAT phosphorylation; without EphB6, ZAP-70 activation and, subsequently, LAT phosphorylation were compromised, even under strong TCR stimulation. (In our experiments, strong TCR stimulation was mimicked by anti-CD3 plus anti-CD4; anti-CD3 alone was effective, but less so than anti-CD3 plus anti-CD4 [data not shown].) EphB6 itself has a mutated kinase domain and thus has no intrinsic kinase activity (18
); how EphB6 activates ZAP-70 remains to be elucidated. There are 2 obvious possibilities. (a) As we have demonstrated (25
), EphB6 is associated with a number of adaptor molecules, such as CrkL, CrkII, Grb2, and Cbl. Any of these adaptors could associate with certain kinases that initiate the cascade of EphB6 signaling. (b) EphB6 can form dimers with EphB1 (33
), which has competent kinase activity; with such dimerization, EphB6 is no longer kinase-incompetent. The phosphorylation of LAT is pivotal for recruitment of other signaling molecules such as SLP-76 via PLCγ1, and for downstream MAPK activation (36
); indeed, we showed that in the absence of EphB6, these further signaling events could not develop to a full scale. We attempted to assess whether EphB6 cross-linking alone could increase LAT phosphorylation and augment SLP-76 binding to PLCγ1 in WT thymocytes; however, in our liquid cross-linking system (i.e., the cross-linking of TCR and EphB6 was conducted in solution by secondary Ab or streptavidin), the increase was not obvious. The likely reason is that it was difficult to adjust the TCR cross-linking to a suboptimal level to reveal the effect of EphB6 in this system. However, it is conceivable that under a physiological condition, EphB6 might be cross-linked by EFNBs on the neighboring cells. Such cross-linking results in activation of ZAP-70 followed by augmented LAT phosphorylation, and in association between PLCγ1 and SLP-76.
The significance of our study with respect to the immune system is as follows: That all 3 ligands of EphB6 are prominently expressed on T cells (refs. 28
, and data not shown) suggests that interaction between EphB6 and EFNB is important for T cell–T cell cooperation during T cell activation. The necessity of such cooperation is often neglected but can well explain the fact that T cells need to reach a certain density in in vitro activation, and the fact that T cells are best activated in lymphoid organs where they are tightly packed and have ample opportunity to interact with fraternal EphB6 ligand–expressing T cells. In our in vitro activation model (Figure ), highly purified CD4 and CD8 cells (more than 98.5% pure) were used, and the lack of EphB6 in these cells led to reduced proliferation upon stimulation; this proves that T cell–T cell collaboration via EphB6 and its ligands is essential in optimizing T cell activation and proliferation. In vivo, it is possible that EphB6 on T cells will receive signals not only from other T cells, but also from APCs, since APCs express some of the EphB6 ligands as well (28
). The overall effect of EphB6 seems to reduce the threshold of T cell response to antigen stimulation.