The cytoskeleton-associated TCR ζ chain is constitutively phosphorylated in the absence of an active p56^lck form

2.1 Phosphorylated ζ forms with different isoelectric points are localized to the cytoskeleton

Previous studies have demonstrated that the cska-ζ chain is phosphorylated in both non-activated 11 and activated T cells 12, 14, 15. Moreover, differences in the apparent molecular mass of phosphorylated cska-ζ (16 kDa) and the phosphorylated non-cska-ζ chains (21 kDa and 23 kDa) strongly suggest that the ζ forms may be phosphorylated at different sites and / or may differ in the number of their phosphorylated tyrosine residues. Indeed, Kersh et al. 16 have used sequence-specific anti-phosphotyrosine antibodies to delineate the phosphorylated tyrosine residues of the p21 and p23 non-cska-ζ chains. While the p21 form has only three phosphorylated tyrosines, each localized to a different ITAM, the p23 form has six phosphorylated tyrosine residues, with two in each of the three ζ-ITAM. A different mode of phosphorylation for these non-cska-ζ forms as well as for cska-ζ might have important functional ramifications. It has been previously shown that differences in phosphorylation affect interactions with various intracellular signaling molecules. For example, ZAP-70 binds to ζ via its two SH2 domains that recognize tandemly phosphorylated tyrosines in the same ITAM 17, while Fyn has the ability to bind to a ζ ITAM phosphorylated on a single tyrosine residue 18. Moreover, a recent study shows that partially phosphorylated TCR ζ molecules in either resting T cells or T cells activated with antagonist ligands can inhibit T cell activation 19.

Accordingly, we aimed to determine whether differently phosphorylated ζ forms exist within the cytoskeletal fraction of non-activated T cells by using the two-dimensional non-equilibrium pH gradient electrophoresis (NEpHGE) / SDS-PAGE analysis. Detergent-insoluble pellets were boiled in sample buffer and sonicated to extract detergent-insoluble (INS) proteins. The extracted phosphorylated and non-phosphorylated cska-ζ forms were immunoprecipitated with anti-ζ antibodies. While this was the most efficient procedure for immunoprecipitation, the poor yield of the precipitated proteins observed (Fig. 1 A) rendered this method of analysis not feasible for subsequent analysis. We therefore analyzed total INS lysates rather than immunoprecipitated proteins. To separate proteins from the cytoskeletal fraction of T cells, on the basis of their isoelectric points (PI) in two-dimensional NEpHGE / SDS-PAGE, we extracted the proteins from the INS fraction with urea and ampholines (A. Ben-Zeev, pesonal communication). Total lysates [INS and soluble (SOL)] were separated on NEpHGE / SDS-PAGE. As shown in Fig. 1 B, ζ chain from the SOL fraction of freshly isolated lymphocytes following separation by NEpHGE / SDS-PAGE revealed several ζ forms displaying both a lower PI and increasing apparent molecular mass (see lower panel). These forms result from differential phosphorylation as determined by anti-phosphotyrosine antibodies (data not shown). The different phosphorylation pattern changes the charge of the protein and its migration in the first dimension and induces a conformation that migrates more slowly in SDS-PAGE (second dimension). However, since immunoblotting with anti-phosphotyrosine antibodies was not sufficiently sensitive to observe phosphorylated cska-ζ by two-dimensional NEpHGE / SDS-PAGE analysis, it was necessary to first label splenocytes with [³²P] orthophosphate in vivo. Our results revealed (Fig. 1 B, upper panel) a pattern of cytoskeleton-localized, phosphorylated proteins of varying molecular mass. Upon immunoblotting with anti-ζ antibodies, only three closely migrating bands of 16 kDa could be discened, each with a slightly different PI. The ζ-band with the lowest PI overlaps precisely with the heavily phosphorylated band observed upon exposure. This band is likely phosphorylated on more tyrosine residues than the other two ζ bands which exhibited a higher PI. In accord with our previous results 12, unlike the non-cska-ζ chain, cska-ζ maintains its apparent molecular mass of 16 kDa even when phosphorylated on tyrosine residues. Although we 11, 12 and others 14, 15 have previously shown that cska-ζ is phosphorylated on tyrosine residues, from these experiments we cannot rule out the possibility that cska-ζ might also be phosphorylated on serine and / or threonine residues. These results demonstrate that cska-ζ chain is phosphorylated in non-activated normal murine lymphocyts, and that there are apparently at least two forms of differently phosphorylated cska-ζ, which maintain a molecular mass of 16 kDa.

2.2 Cska-ζ does not require p56^lck activity to undergo phosphorylation in vivo

The different mode of phosphorylation exhibited by cska-ζ raised the question as to whether p56^lck, which plays a crucial role in the generation of the phosphorylated p21 / 23 non-cska-ζ form 1, 2, is also involved in the phosphorylation of cska-ζ. To resolve this issue, we took advantage of the p56^lck-inactive JCaM1.6 cell line 20 and its parental (p56^lck-active) Jurkat human T cell leukemic line. The relative levels of ζ chain localized to the SOL and cytoskeletal fractions were compared using non-activated and activated cells from each line (Fig. 2, insets). The ratio of cska-ζ to non-cska-ζ was similar in both cell lines, indicating that an active p56^lck form is not obligatory for the localization of ζ chain to the cytoskeleton, at least in JCaM1.6 cells.

We next compared the phosphorylation levels of the two ζ forms (Fig. 2). The non-activated Jurkat cells maintained a basal level of constitutively phosphorylated cska-ζ similar to that previously observed in normal splenocytes 12. However, unlike normal splenocytes, the SOL fraction of Jurkat cells displayed high levels of constitutively phosphorylated p21 / 23 non-cska-ζ forms which dramatically increased upon activation via the TCR. However, the basal level of phosphorylated cska-ζ in these cells remained virtually unchanged even after TCR-mediated activation.

We then asked whether an active p56^lck form is required for the generation of phosphorylated 16-kDa cska-ζ, similar to its requirement for the generation of the phosphorylated p21 / 23 non-cska-ζ forms. As depicted in Fig. 2, non-activated JCaM1.6 cells displayed an undetectable level of p21 / 23 phosphorylated non-cska-ζ chain. Moreover, phosphorylation of non-cska-ζ was severely impaired upon TCR-mediated stimulation. This is in agreement with previously published observations 20 and residual levels of p21 / 23 non-cska-ζ chain were likely due to the compensatory effect of other kinases, such as p59^fyn. In contrast, the absence of an active p56^lck form in these cells did not prevent phosphorylation of the 16-kDa cska-ζ form.

To provide further evidence suggesting that p56^lck is not involved in the phosphorylation of cska-ζ, we compared normal splenocytes isolated from C57BL / 6 mice to those isolated from knockout mice lacking expression of an active p56^lck form. The normal C57BL / 6 splenocytes exhibited properties similar to those previously observed in other strains: ratios of cska-ζ / non-cska-ζ were similar, the constitutively phosphorylated 16-kDa cska-ζ form was unique to the INS fraction, and the 21 / 23-kDa ζ form was localized entirely to the SOL fraction (Fig. 3 A; top panel). p56^lck could not be detected at all by Western blotting analysis in splenocytes of the p56^lck-inactive mice (Fig. 3 B). However, splenocytes from these mice still contained basal levels of phosphorylated 16-kDa cska-ζ chain (Fig. 3 A; lower panel) The relatively low levels of ζ chain depicted in p56^lck-inactive splenocytes compared to normal mice (Fig. 3) are in accordance with the low T cell / B cell ratio previously observed 21. These results provide further evidence that p56^lck is not involved, directly or indirectly, in the in vivo phosphorylation of cska-ζ.

Complementary analysis was performed using T cells isolated from the spleen of lpr mice. These T cells are known to display enhanced tyrosine phosphorylation levels of the non-cska-ζ chain 22, most likely due to enhanced p56^lck and p59^fyn kinase activities 23. As shown in Fig. 4, in the SOL fraction of non-activated lpr splenocytes enhanced tyrosine phosphorylation was observed when compared to normal non-activated splenocytes. This was reflected by the elevated levels of the p21 / 23 ζ chain form. These results indicate that basal levels of PTK activity are indeed elevated in the lpr cells. In contrast, levels of phosphorylated and non-phosphorylated 16-kDa cska-ζ (INS) were unchanged in the lpr splenocytes and resembled phosphorylation levels of cska-ζ in normal splenocytes. Again, these results indicate that phosphorylation levels of cska-ζ are independent of p56^lck kinase activity and most likely also of the p59^fyn activity.

2.3 The cska-ζ chain serves as a potential in vitro substrate for p56^lck

Our results show that while p56^lck is required for the phosphorylation of non-cska-ζ chain, it is not essential for the in vivo phosphorylation of the cska-ζ chain. One compelling question that arose is why p56^lck does not phosphorylate cska-ζ in vivo. We suggested two possible general mechanisms: (1) cska-ζ is in a conformation that prohibits binding and / or in vivo phosphorylation by p56^lck; (2) a "compartmentalization effect", where an active form of p56^lck is not accessible to cska-ζ in vivo, perhaps as a result of differential subcellular localization to separate microdomains. To provide evidence that might support either possibility, we sought to determine whether cska-ζ has the potential to serve as a substrate for p56^lck in vitro.

To this aim, we designed a reconstituted in vitro kinase assay where ζ chain is first immunoprecipitated from either the SOL or the DNasel-treated cytoskeletal fraction (INS; see Sect. 4.5). Some of the ζ immunoprecipitates were then "reconstituted" with immunoprecipitates containing p56^lck (Fig. 5). The samples were subsequently subjected to in vitro kinase assays, and phosphorylated proteins were analyzed. Under these experimental conditions, the anti-ζ mAb precipitated non-cska-ζ along with a kinase(s), as reflected by a level of kinase activity observed in vitro even without "reconstitution" with exogenous p56^lck (Fig. 4 A; upper left panel). We then endeavoured to combine and "reconstitute" the immunoprecipitates of the SOL non-cska-ζ chain with immunoprecipitates of Lck. The results show an increase in the phosphorylation of ζ (Fig. 5 A; upper right panel), demonstrating that the "reconstitution system" of combining immunoprecipitates that the "reconstitution system" of combining immunoprecipitates of both kinase (p56^lck) and substrate (ζ) could be utilized. We then determined whether the cska-ζ chain could also undergo phosphorylation upon in vitro "reconstitution" with p56^lck. Under the conditions used, little or no co-associated kinase activity was found in the cska-ζ immunoprecipitates, and therefore basal phosphorylation levels of the cska-ζ chain in vitro were not observed (Fig. 5 A; lower left panel). However, upon in vitro "reconstitution" with p56^lck, cska-ζ underwent substantial phosphorylation and in some of the experiments performed, several phosphorylated forms of cska-ζ could be obtained (Fig. 5 A; lower right panel). In several cases, the CD3 subunits that co-immunoprecipitated with cska-ζ also served as substrates for p56^lck in the in vitro "reconstitution" assays (Fig. 5 A; lower right panel). Control experiments were performed for non-cska-ζ (data not shown) and for cska-ζ (Fig. 5 B). In the latter, cska-ζ was "reconstituted" with immunoprecipitates using protein A-Sepharose beads precoated with normal rabbit serum, and was devoid of any kinase activity which phosphorylates cska-ζ (Fig. 5 B; upper right panel). Immunoprecipitations of p56^lck under the conditions specified did not "drag along" ζ chain (data not shown), indicating that the phosphorylation of ζ was indeed due to the "reconstitution" with the p56^lck immunoprecipitate. These results show that while p56^lck is not responsible for the phosphorylation of cska-ζ in vivo, it retains the capacity to phosphorylate cska-ζ in in vitro kinase assays. This hints that cska-ζ is not phosphorylated by p56^lck in vivo due to the physical separation between kinase and substrate. In support of such a possibility is the study by Rodgers and Rose 24 demonstrating that the localization of Lck to glycolipid-enriched membrane domains induces a physical separation between phosphatase (CD45) and substrate (Lck). In this case, Lck is hyperphosphorylated and displays weakened kinase activity, an effect also exhibited for detergent-soluble Lck in T cells lacking CD45 24.

It is conceivable that "compartmentalization" or physical separation of TCR subunits from specific intracellular molecules may be utilized by T cells to regulate signal transduction pathways. Several reports provide strong evidence that ζ and the TCR localize to detergent-resistant "rafts" 25. Most of the studies suggest that TCR localization to the "rafts" appears following TCR stimulation 26 – 29. Results are conflicting regarding the appearance of the TCR within "rafts" in resting T cells. However, unlike the non-cska- and cska-ζ forms, the raft-associated ζ chain appears to localize to these domains alone, without the other TCR subunits. It is noteworthy that while detergent insolubility is common to both raft- and cytoskeleton-associated proteins, the insolubility of the former is considered to be unaffected by inhibitors of the cytoskeleton 30. Moreover, while the cholesterol-sequestering agent saponin dissociates ζ from its "rafts" localization 27, it is not known to promote dissociation of cytoskeleton-linked proteins. Since we 11 and others 14 have demonstrated that the detegent insolubility of ζ chain depends on an intact actin microfilament system, it is likely that there are two distinct detergent-insoluble pools, only one of which localizes to the cytoskeleton and lacks an active form of Lck. However, these two detergent-insoluble ζ forms may not be mutually exclusive. A recent study by Harder and Simons 28 shows that T cell rafts accumulate filamentous actin and tyrosine-phosphorylated proteins. Another study by Moran and Miceli 29 demonstrates that cholesterol depletion in T cells not only impairs raft formation, but also interferes with the association between ζ and the cytoskeleton.

2.4 The cska-ζ-associated kinase(s) requires different conditions for optimal in vitro activity than the kinases associated with non cska-ζ chain

The results presented herein show that p56^lck is not required for the phosphorylation of cska-ζ. Moreover, treatment of cells with the Src-family selective tyrosine kinase inhibitor PP1 31 did not affect levels of phosphorylated cska-ζ (data not shown).

The apparent involvement of a different kinase(s) in the phosphorylation of cska-ζ suggests a difference in function of the two TCR populations. We attempted to precipitate kinase activity with cska-ζ by immunoprecipitating non-cska- and cska-ζ from normal non-activated or activated T lymphocytes, and subjecting them to in vitro kinase assays. These immunoprecipitates were washed according to condition A (see Sect. 4.7). As indicated in Fig. 6 A (upper panels), the non-cska-ζ chain was weakly phosphorylated in non-activated cells, but showed greatly enhanced phosphorylation upon activation. Another kinase-dependent measure of TCR-mediated activation, the appearance of the ubiquitinated ζ chain 32, could also be observed. The CD3 γ, δ and ϵ chains also served as substrates for the TCR-associated kinase(s). However, under these conditions, precipitations of cska-ζ failed to display kinase activity (as observed after long exposures) in either non-activated or activated T cells (Fig. 6 A; lower panels).

Since it is likely that a different kinase(s) is needed to phosphorylate the cska-ζ chain, we hypothesized that the conditions required for its optimal activity may be distinct from those needed by kinases associated with the non-cska-ζ form. We therefore performed kinase assays under a wide variety of conditions, in an attempt to observe in vitro phosphorylation of cska-ζ by its associated kinase(s) (Fig. 6 B and data not shown). Levels of phosphorylated cska-ζ could be observed when cska-ζ immunoprecipitates were washed under condition B (see Sect. 4.7) (Fig. 6 B; left panel). These results were also obtained after preclearing with a non-relevant antibody, suggesting that the cska-ζ-associated kinase activity is specific. Under these conditions, no ζ-associated kinase activity was detected in the SOL fraction (Fig. 6 B; central panel). However, when a wash with water was added (condition C, see Sect. 4.7), non-cska-ζ chain was readily phosphorylated by an associated kinase (Fig. 6 B; right panel), but cska-ζ did not undergo in vitro phosphorylation (data not shown). Thus, while cska-ζ is phosphorylated in vitro by a kinase that co-immunoprecipitates with it and is active in the presence of trace levels of Triton X-100, the non-cska-ζ-associated kinase is most active in vitro in the absence of detergent. Although the activity obtained in vitro is generally weak, these results further support our data indicating that different kinases are responsible for the phosphorylation of these two ζ forms. One possible explanation for the weak level of kinase activity in the cytoskeletal fraction is that the interaction between cska-ζ and its phosphorylating kinase may be a transient one.

A recent study has reported the involvement of lck in phosphorylating a form of ζ that localizes to the INS fraction 33. However, this lck-dependent phosphorylated, dimeric ζ form is induced upon TCR stimulation, similar to the non-cska ζ chain and a ζ form that associates with cholesterol and sphingomyelin-enriched "rafts". While we have been able to detect only the 16-kDa uniquely phosphorylated ζ form in the INS cytoskeletal fraction, it is clear that the reported INS ζ forms may have distinct functions and will need to undergo further investigation.

One way to elucidate a functional role for cska-ζ may be to identify molecules that might link the TCR to the actin microfilaments. Indeed, two recent studies 34, 35 provide strong evidence that the proto-oncogene Vav may well be such a linker protein. Determining both how zeta associates with the cytoskeleton, and identifying the kinase responsible for its phosphorylation should help unravel the role of cska-ζ.

4.1 Animals

BALB / c and C57BL / 6 female mice were bred in our SPF facility. lpr mice were purchased from The Jackson Laboratory (Bar Harbor, MA). Mice lacking expression of an active Lck form were generated and generously provided by Dr. T.W. Mak 21.

4.2 Cells

Splenocytes were isolated from mice aged 12 – 16 weeks and "rested" overnight, unless otherwise indicated, at 37 °C in RPMI 1640. The human leukemic Jurkat C cell J.E6-1 and JCaM1.6 Lck-inactive lines 20 were grown in the same medium.

4.3 Antibodies

The 145-2C11 (2C11) and OKT3 mAb have been described 36, 37. Polyclonal rabbit anti-ζ antibodies were generated as described 38. The anti-ζ mAb H-146 39 was a kind gift from Dr. R. Kubo and Dr. D. Wiest. Anti-phosphotyrosine mAb (4G10) were obtained from Upstate Biotechnology.

4.4 Cell activation

Normal splenocytes were harvested after being incubated overnight at 37 °C ("rested"), washed and resuspended in full media at a concentration of 1 × 10⁷ cells per ml. Cells were stimulated by anti-CD3ϵ antibodies (2C11 ascites diluted 1 : 250) for the time indicated. Jurkat and JCaM1.6 cell lines were activated by incubation with OKT3 ascites (1 : 100) for 2 min.

4.5 Cell lysis and separation of SOL and INS cytoskeletal fractions

Cell pellets were lysed with MES lysis buffer, as previously described 12. Following lysis, the detergent-soluble fraction was isolated and the supernatant was designated as "SOL". The detergent-insoluble pellet was washed and designated as "INS". For analysis of the total insoluble fraction, proteins were extracted by incubation at 95 °C in sample buffer containing 4 % SDS. When in vitro actin microfilament depolymerization of the cytoskeletal fraction was required, the INS pellet was treated with DNasel as previously described 12. Unless otherwise indicated, proteins from each fraction were resolved by two-dimensional non-reducing / reducing SDS-PAGE on 12 % and 13 %, gels, respectively.

4.6 In vivo labeling of cells with [³²P] orthophosphate and two-dimensionsal NEpHGE / SDS-PAGE analysis

Freshly isolated splenocytes were "phosphate-starved" and then suspended at 2 × 10⁷ cells / ml in phosphate-free RPMI containing 7 % dialyzed FCS and 0.3 mCi [³²P] orthophosphate / ml. Cells were labeled at 37 °C for 4 h, washed, lysed and separated into SOL and cytoskeleton-enriched INS fractions, as described above. Proteins were eluted from the INS pellet by the addition of NEpHGE sample buffer (containing 57 % urea w / v, 4 % ampholines pH 3.5 – 10, 0.2 % NP40 and 3 % 2-ME) followed by mechanical agitation. The solubilized proteins were then loaded onto NEpHGE capillary tubes and separated at 15 mA and 0.3 W / tube until the voltage reached 500 V. Samples were then separated for another 5 h at 500 V (15 mA and 0.3 W / tube). Tube gels were then equilibrated for 30 min in 0.5 M Tris pH 6.8 containing 0.1 % SDS and 0.5 % DTT (w / v), and resolved on 10 – 20 % gradient SDS-PAGE. All gels were transferred to nitrocellulose, exposed to PhosphorImager or X-ray films, and in some cases immunoblotting analysis was performed.

4.7 Immunoprecipitations and kinase assays

For immunoprecipitations, the INS pellets were either treated in vitro with DNasel (Boehringer Mannheim), as previously described 12 or sonicated (3 min, 4 °C) and eluted (30 min, 4 °C) in a buffer containing 20 mM Tris pH 8, 150 mM NaCl, 0.1 % SDS, 0.1 % Triton X-100, 0.5 % deoxycholic acid, 40 mM DTT, protease and phosphatase inhibitors. The extracted INS proteins and the SOL lysates were equilibrated to 150 mM NaCl and incubated with antibody-coated protein A-Sepharaose beads (Pharmacia) for 3 h at 4 °C. Immunoprecipitates were either washed and separated on one-dimensional, reducing SDS-PAGE or treated according to condition A (i. e. washed twice a buffer containing 0.1 % Triton X-100, and then once in kinase buffer containing 100 mM NaCl, 25 mM Hepes pH 7.5, 5 mM MgCl₂ and 5 mM MnCl₂), condition B (i. e. washed twice with a buffer containing 0.1 % Triton X-100, and then once in wash buffer lacking EDTA in the phosphatase inhibitors), or condition C (i. e. washed twice with a buffer containing 0.1 % Triton X-100, and then once on deionized water) and subjected to a kinase reactin (in kinase buffer) with the addition of 1 μCi [γ-³²P] -ATP / reaction. Reconstitution studies were performed using two sets of immunoprecipitations performed in parallel: (1) non-cska-ζ and cska-ζ were immunoprecipitated as described above; (2) Lck was immunoprecipitated from non-activated splenocytes using protein A-Sepharose beads precoated with rabbit serum containing anti-Lck antibodies. A control immunoprecipitation was done using protein A-Sepharose beads alone or protein A-Sepharose beads coated with preimmune normal rabbit serum. After the immunoprecipitation, the beads were washed and immunoprecipitates from each group were mixed together. The mixtures were then subjected to in vitro kinase assays as described above. The reaction was terminated by the addition of sample buffer containing SDS and boiling. The samples were then subjected to two-dimensional non-reducing / reducing SDS-PAGE. Separated proteins were transferred to nitrocellulose filters and either exposed to films or PhosphorImager. The filters were subsequently immunoblotted with the designated antibodies.