Purification and initial characterization of the lymphocyte-specific protein-tyrosyl kinase p56lck from a baculovirus expression system.

A baculovirus expression system has been used to express large quantities of the lymphocyte-specific protein-tyrosyl kinase p56lck. A series of chromatographic steps, including the novel application of metalchelate affinity chromatography for protein kinase purification, were employed to obtain p56lck in a highly active form. Recombinant p56lck was purified to apparent homogeneity as determined by polyacrylamide gel electrophoretic analyses and was found to migrate in SDS gels as two related species, both with apparent molecular masses close to 56 kDa. p56lck phosphorylated all assayed substrates exclusively on tyrosyl residues, and underwent autophosphorylation at one principal site, also on a tyrosyl residue. p56lck displayed a high affinity for a synthetic peptide corresponding to the cytoplasmic domain (residues 52-164) of the T-cell receptor zeta-chain (TCR-zeta) (Km approximately 6.5 microM) but a low affinity for a peptide corresponding to its own autophosphorylation site (Km approximately 900 microM). p56lck was also found to be highly active for a purified protein-tyrosyl kinase (Vmax greater than 400 pmol.min-1.micrograms-1 using the TCR-zeta (52-164) as a substrate). A variety of agents were tested for their ability to inhibit p56lck, with zinc ions (I50 approximately 1.7 mM) and staurosporine (I50 approximately 500 nM) proving the most potent.

to be highly active for a purified protein-tyrosyl kinase ( Vmax >400 pmol*min"*~g" using the TCR-{(62-164) as a substrate). A variety of agents were tested for their ability to inhibit p66lCk, with zinc ions (IB0 -1.7 mM) and staurosporine (160 -500 nM) proving the most potent.
A number of studies have revealed that the enzymatic activity of the protein-tyrosyl kinase ~5 6 "~ is an integral component in the process of T-cell activation (reviewed in Ref. 1). ~5 6~~ is encoded by the kk gene, a member of the src family of protein-tyrosyl kinases (2). These tyrosyl kinases are associated with the cytoplasmic face of the plasma membrane by virtue of myristic acid covalently bound to their Nterminal glycine residues (3, 4). Additionally, all src family members share related sequences which include the catalytic, SH2, and SH3 domains as well as negative C-terminal regulatory sequences (reviewed in Ref. 5). Importantly, these molecules encode unique N-terminal sequences that specify, at least in kk, the association with specific cell surface receptor proteins.
CD4 and CD8 are T-cell-specific cell surface glycoproteins which are known to interact with the N-terminal region of p56 Lk via cysteine-containing sequences (6-8). CD4 and CD8 have been shown to bind nonpolymorphic regions of major * This work was supported in part by operating grants from the Medical Research Council of Canada (MRC) (to R. A. and C. R. A.).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby histocompatibility complex class I1 and class I proteins, respectively, during antigen-dependent activation of T-lymphocytes (9-12) and during T-cell maturation (13-15). Subsequent to thymocyte maturation and emigration to the periphery, differential surface expression of CD4 and CD8 defines distinct T-cell lineages as well as at least some of the immunological responses to foreign antigens observed in the mature cells (14,16-18). ~5 6 "~ may therefore be a critical downstream element in signal transduction from CD4 or CD8. Indeed, it has recently been shown that T-cell activation via an antigen presenting cell requires the correct interaction of ~5 6 "~ with CD4 (19). Thus to understand fully T-cell development and activation, it is of great importance to determine the role that ~5 6 "~ plays in this process and how the enzyme itself is regulated.
All src-related tyrosine kinases are capable of autophosphorylation on specific tyrosyl residues (at Tyr-394 of ~5 6 "~, reviewed in Ref. 5). Autophosphorylation occurs within a conserved motif in the catalytic domain and is closely associated with the in vitro and in vivo activation of ~5 6 "~ as a kinase (20). The C-terminal regulatory domain of src family kinases contains a tyrosyl residue (Tyr-505 in ~5 6 "~) that when phosphorylated results in down-regulation of enzymatic activity. This is supported by experiments showing that mutation of this tyrosine to phenylalanine results in a hyperactive enzyme and tumorigenic transformation of 3T3 cells (21,22).
CD45 is a membrane-associated protein-tyrosyl phosphatase present in T-cells (23,24). Recent studies have implicated CD45 as the in vivo regulator of ~5 6 "~ activity via dephosphorylation of Tyr-505. Mutant cell lines lacking CD45 exhibit increased ~5 6 "~ phosphorylation at Tyr-505 (25) and antibody clustering of CD45 with either CD4-~56"~ or CD8-~56"~ complexes leads to a decrease in phosphotyrosine levels associated with ~5 6 "~ (26). In all studies of ~5 6 "~ properties in the milieu of an activated T-cell, increased levels of serine and threonine phosphate have been observed and found to be associated with alterations in the mobility of ~5 6 "~ by SDS-PAGE' analysis (27, 28). While these studies clearly show that these sites of in vivo phosphorylation induced on ~5 6 "~ are important regulatory events during T-cell activation, their functional signifi-

902
Purification and Characterization of ~5 6 "~ cance and the protein-seryl/threonyl kinase(s) responsible for modifications have yet to be determined. While distinct progress has been made in understanding the mediators of T-cell signalling, the lack of purified enzymes or knowledge of target substrates has hampered characterization of the activity, specificity, regulation, and structure of these important mediators of cellular signalling.
In spite of recent advances, very little is known regarding the substrate identities of protein-tyrosyl kinases both in vitro and in uiuo (reviewed in Ref. 29). One potential substrate of ~5 6 "~ is a component of the T-cell receptor (TCR) CD3 complex. When ligated with CD3 via antibody conjugation, C D 4 .~5 6 "~ complexes are able to evoke the tyrosine phosphorylation of the TCR {-chain (30). Phosphorylation of the {-chain on tyrosine is a strong correlate of an activated state of the T-cell (31, 32) and can be observed in response to antigen or in lymphoproliferative states such as revealed in the lprllpr mouse (33).
To determine the molecules and events that govern the regulation of ~5 6 "~ via posttranslational modification and to define substrate targets of ~5 6~~ following T-cell activation, a useful tool would be an abundant source of purified and active enzyme. We have therefore overproduced ~5 6 "~ using a baculovirus expression system. Subsequent purification strategies including the novel application of metal-chelate affinity chromatography for the purification of a protein kinase, resulted in active ~5 6 "~ preparations of very high purity. Initial characterization of the purified recombinant ~5 6 '~ has been accomplished and its enzymatic properties are described herein.

RESULTS
Purification of p5@k from Sf9 Cells-Recombinant murine ~5 6 "~ was overexpressed in and subsequently purified from the Spodoptera frugiperda cell line Sf9, as described under "Experimental Procedures." Table I summarizes the yield and efficiency observed at each purification stage. ~5 6 '~ activity profiles for each chromatographic step were determined via an in uitro kinase assay and are given in Fig. 1 in the miniprint section. The resultant protein content and composition following each purification step was determined by SDS-PAGE, visualizing either by Coomassie Blue staining, autoradiography following autophosphorylation, or by Western blot. These results are shown in Fig. 2. Fig. 3 demonstrates that the recombinant ~5 6 "~ exhibits exclusively tyrosyl kinase activity and shows that the enzyme autophosphorylates at only one major site. Fig. 4 shows a two-dimensional Western blot illustrating the presence of several size/charge variants of ~56"' expressed in the Sf9 cells. For a more complete description of the ~5 6 "~ purification and its electrophoretic characterization, see the miniprint section.
Enzymatic Characterization of Buculovirus-expressed p56"k-For the purposes of determining some characteristics of the recombinant ~5 6 "~, it is not necessary to use a homogeneous preparation of enzyme providing that there are no contaminating kinases or high-affinity substrates. As can be seen from Fig. 2B, tyrosine-agarose peak A contains only one phosphorylated band with a M , of -56,000. Following prolonged autoradiography of the thin-layer chromatography * Portions of this paper ("Experimental Procedures," part of "Results'' including p56"' purification and electrophoretic characterization, Tables I and 111,  plate in Fig. 3A (not shown) it was also clear that both ~5 6 "~ and LCK-AP (a peptide corresponding to the p56" autophosphorylation site) are phosphorylated solely on tyrosyl residues by this fraction. Additionally, myelin basic protein (MBP, an efficient substrate for many protein-tyrosyl, -seryl, and -threonyl kinases) phosphorylated in uitro by an aliquot of the DEAE-Sepharose peak was found (by two-dimensional peptide mapping and direct protein sequencing of phosphopeptides) to be phosphorylated only at a single, specific tyrosyl residue (43). Thus the more highly purified Zn2+-chelating Sepharose peak was considered to be of more than sufficient enzymatic homogeneity (see lane 4, Fig. 2, A and B ) for further studies of ~5 6 "~ activity. ~5 6 "~ was incubated with a variety of potential substrates to determine how efficiently each was phosphorylated. A timecourse of phosphorylation had revealed that incorporation of phosphate into the LCK-AP by ~56"' was linear for the first 8-10 min (data not shown). V , , and K,,, values were therefore determined graphically for the various substrates following 3min incubations. Table I1 summarizes these findings and   preparations of p60"" (44-46) and a partially purified p6Wes (47) all three kinases being members of the src gene family. ~5 6 '~ demonstrated the highest affinity ( K , -6.5 PM) and Vmax (-420 pmol min" pg") for a synthetic peptide corresponding to the entire cytoplasmic domain of the TCR {chain (residues 52-164) (35) and demonstrated a V,,,,/K, for this molecule equivalent to those previously attained for any substrate of p60"" and p6Wes (see Table 111). ~5 6 "~ also phosphorylated MBP relatively well and a poly-glutamic acid/ tyrosine copolymer (4:l) less efficiently ( Vma,/K, -3.4 and 0.4, respectively). However, LCK-AP was found to be a poor substrate (K, -900 PM) though we were able to obtain a Vmex of -80 pmol. min" .pg-' ( Vma,/K, -0.1). There was no detectable phosphorylation of bovine serum albumin by ~5 6 '~ (data not shown).
A number of agents were also tested for their ability to inhibit ~5 6 "~. These results are summarized in Table IV. DTT has been found to activate certain protein-tyrosyl kinases (48) but was found to have little effect on ~5 6 "~, though it was inhibitory at high concentrations (I6o -100 mM). Sodium fluoride, a known protein kinase inhibitor (49), molybdate (a phosphatase inhibitor (50),) additional divalent cations (calcium), and Triton X-100 also did not significantly affect ~5 6 "~ activity. Sodium chloride, however, was found to have a much lower 150 (-300 mM) than is observed with some protein-seryl/ threonyl kinases (~4 4 ' "~~, a member of the MAP kinase family for example, gives an 150 with NaCl of >1 M (49)). Interestingly, the addition of DTT (1 mM) to the NaCl buffer increased ~5 6 "~ sensitivity to salt (Iso -150 mM). Predictably, manganese ions activated the enzyme with a maximum at 5-10 mM (data not shown) and were inhibitory at high concentration (I50 -100 mM). The pH optimum for ~5 6 "~ was determined using MES, Tris, and CAPS buffers at 0.5 pH-unit intervals between pH 5 and 11, inclusive. ~5 6 "~ was completely inactivated at extremes of pH (activity detected only above pH 5.5 and below pH 10) with 50% inhibition occurring at pH 5.8-6.3 and pH 8.8-9.3 compared with the maximum at pH 7.5-8.0. The results for all three buffers also indicated that at the relatively high concentrations used (100 mM final) Tris was mildly inhibitory compared with MES and CAPS (data not shown).
Of the agents thus far tested, the most potent inhibitory effects were observed with zinc ions and staurosporine, a molecule previously shown to inhibit protein kinase C as well as a series of protein-tyrosyl kinase activities (51). The ISo value for zinc ions was found to be -1.7 mM and -500 nM for staurosporine. It could be that zinc ions compete with manganese ions as cofactors (the strong interaction of ~5 6 '~ with the Zn'+-chelating column could support this hypothesis), or it is possible that N-terminal cysteine-containing motifs found in ~5 6 "~ can bring about inactivation via a zinc-mediated aggregation of the enzyme.

DISCUSSION
Since one of our goals was to obtain an abundant supply of active ~5 6 '~ for in vitro phosphorylation studies, a baculovirus 9.0 f 0.2 expression system was chosen to overexpress the molecule. The method presented in this paper describes the subsequent purification of ~5 6 "~ in an active form. This enzyme preparation has been shown to autophosphorylate solely on tyrosyl residues (see Fig. 3A) as well as phosphorylate several model substrates, including a synthetic peptide corresponding to the cytoplasmic domain of the TCR {-chain (see Table 11). The availability of this purified enzyme should greatly facilitate the study of both the kinase(s) and phosphatase(s) involved in the regulation of ~5 6 "~, as well as the determination of its in vivo substrates and hence lead to a greater understanding of the role of ~5 6 "~ in the process of T-cell activation. The highly abundant sources of intact enzymes afforded by the baculovirus expression system could also lead to the ultimate determination and comparison of x-ray crystal structures of different members of the src gene family as well as functionally significant mutants.
Previously among the src family of kinases, only p60"" has been obtained at such a high degree of purity. This has been accomplished from both physiological sources (44-46, 52) as well as from a baculovirus system (53). The proto-oncogene c-yes product, p 6 P , has also been partially purified (-43%) from rat liver (47). Table I11 demonstrates that our purified recombinant ~5 6 "~ is as least as active as any other src family kinase currently purified from more natural sources. It is clear that both the autophosphorylation peptides SRC-AP and LCK-AP are poor substrates for their respective kinases (K, -670 and -900 p~, respectively). Also, the difference in observed V,.,/K, values obtained with a poly glutamic acid/ tyrosine copolymer for ~5 6 "~, p60"", and p6oy'" ( V,.,/K, -0.4, 1.0, and 60, respectively) suggests that at least p 6 P may have a different substrate specificity compared with the other two enzymes.
More recently, a baculovirus-produced ~5 6 "~ has also been reported partially purified (54). However, this preparation was only -30% pure, and only short peptides were assayed as model enzyme substrates. This partially purified enzyme gave a K, of -1.8 mM for a peptide corresponding to residues 386-399 of ~5 6 "~ with a Vmax of -0.5 nmol-min" .pg" ( V,,,/K, -0.3) while our preparation gave a K, of -900 JIM and a V,,, of -8 0 pmol.min" .pg" for the closely related LCK-AP It is known that the TCR {-chain is phosphorylated on tyrosyl residues in vivo (31,32) and that the src-related kinase p 5 9 " is associated with certain TCR molecules (55). We therefore synthesized the cytoplasmic domain of the TCR {chain (residues 52-164) in its entirety in order to determine its effectiveness as a substrate for ~5 6 "~. We observed a low K, for this molecule (-6.5 pM) with a V,,, of -420 pmol. It now seems clear that the SH2 domains of many receptorassociated molecules such as PLC-7 and GAP interact directly with phosphotyrosine residues present in the cytoplasmic domains of various receptor molecules (56, 57). Since ~5 6 ' "~ also contains an SH2 domain, it is easy to envisage that the affinity of ~5 6 "~ for the synthetic TCR {-chain (52-164) would increase as the molecule becomes phosphorylated (it contains 6 tyrosyl residues). This effect would not be observed with the shorter TCR {peptide (107-117). Additionally, since the nature of tyrosyl phosphorylation of the TCR {-chain and the kinase(s) responsible for these modifications still remain unclear, the finding that ~5 6 '~ efficiently phosphorylates a syn- in vitro model for the determination of its in vivo phosphorylation sites. ~5 6 "~ was found to be fairly resistant to inhibition by a selected group of salts, detergent, and divalent cations known to affect the activity of some protein-seryl/threonyl kinases and phosphatases (see Table 11). However, ~5 6 "~ was strongly inhibited by zinc ions at concentrations in the low millimolar range, possibly due to competition for manganese binding sites. p60"" is also strongly affected by zinc ions, but is far less sensitive to salt than ~5 6 "~ (45). It is noteworthy that while zinc ions inhibit ~5 6 "~, they have been found to activate CD45 (24) the protein-tyrosyl phosphatase thought to play a role in the in vivo regulation of ~5 6 "~ phosphotransferase activity. Staurosporine, previously shown to inhibit a number of protein kinases including protein kinase C and some receptor protein-tyrosyl kinases (51), was found to inhibit ~5 6 "~ at concentrations in the nanomolar range. The availability of purified ~5 6 "~ should allow screening of other molecules known to inhibit T-cell activation in vivo and thus allow the question of the mode of action of such inhibitors to be addressed in greater detail.
Purified ~5 6 "~ may now be used in in vitro kinase assays with potential substrates, purified protein-seryl/threonyl kinases, and partially purified activated and inactivated Tcell extracts. Recent advances in protein sequencing technology allowing the determination of phosphoserine, phosphothreonine (58), and phosphotyrosine (43) residues at the low picomole level of sensitivity, developed in part in this laboratory, will permit the direct sequencing of phosphopeptides derived from such assays. Subsequent comparison of twodimensional peptide maps of in vitro and in vivo phosphorylated molecules will thus allow the determination of the apparent regulatory sites of seryl phosphorylation induced on ~5 6 "~ following T-cell activation as well as sites of tyrosyl phosphorylation induced on potential substrates by p56". In a similar manner, utilizing the TCR-{ 52-164 peptide, we should also be able to determine the in vivo sites of tyrosyl phosphorylation found on the TCR {"chain. Knowledge of such phosphorylation sites is vital for our understanding of the role played by both ~5 6~~ and the TCR-Jc chain in the process of T-cell activation as well as the regulation of ~5 6 '~ activity and should greatly facilitate studies into their functional significance by the expression of mutated genes both in cell culture and transgenic expression systems.
The tyrosme-agarose fractmn resulting from peak A was loaded onto a CS column (10 x 1 5 cm) bncluding p5Sc*, were removed from the coiumn wlth a 100 mi gradient from 0 to 50 mM Imldazoie ~n CS buffer, while tightly bound species were removed wtlh 50 mM EDTA In CS buller. Fractlons conlaming kinase activity were again pooled [ Figure IC) and dialysed Into 25 mM Tns-HCI pH 7.5, Tyrosine-agarose peak B was also chromatographed under these condn!ons (not shown).
10% glycerol. 1 mM EDTA, 0.1% Trlton X-100: MonoQ column buffer) w n g a YMlO membrane The CS fraction resulting f r m tyrosine-agarose peak A was diluted three-fold wcth MonoQ column buller and loaded Onto a 1 ml Mono0 column. The column was washed wnh buller. and bound material eluted with a 0 lo 1 M NaCl gradient In runnlng buffer. p56rC*-conta~nmg fractions were #dentdied by the In virro kinase assay (see Flgure 1D) Purlfled enzyme was dlaiysed into Mono0 column buller. made up to 30% glycerol and stored at -60°C untli requred.

139)
However. lnlial anempts to purify p%k*lrom LSTRA cells were unsuccessful. and the quantdies Separate preparatlons of Sf9 cell microsomes and cylosol showed Substamial enzymatic activity lo be Present In both Iractlons (data not shown) thus total detergent-solubb cell emracts were prepared.
was overcome by the incorporation 01 protease lnhlbnors (PMSF to 1 mM and aprotinm to 10 pglml) in It was found. however, that ~5 6~k was hlghly susceptible to proteolysis under these conditions. This the runnlng bufler. Table I  Important to fun the column. screen the fractlons lor ~5 6 '~ activny. and dlalyze and concentrate the Two major peaks of ~5 6~~ actlvlty were observed from the tyrosme-agarose column. together peak A , see Figure 18) One potentlal dilticulty wnh the use of the bacubwrus system IS the (unknown) containing -30% of the total actlvny loaded onto the Column (-70% of this bang in peak Band -30% m extent of post-translattonal modltlcat1ons. Slnce pSkk was over-expressed. It 18 lhkely that the total resubng enzyme would exhlblt a dlstnbmon of phosphorylatlon and myrlstoylation states. Thus because the tyroslns-agarose column was operated m a hydrophobic lnteractlon mode, n IS posslble that the two peaks observed represent dtlterent myrlstoylatlon slates of p56bk However. rechromatography of peak A on the tyroslne-agarose column yielded the same two peaks. thm time recoverlng -50% 01 the actwity loaded onto the column (data not shown). The reason for these two peaks IS thus unclear. but IS unlhely to be due solely to the presence of two subpopulations. Analysis autophosphorylations pertotmed uslng both tyrosine-agarose peaks revealed no dlllerences in by Western blonmg gave the same pattern for both peaks A and B and all in vmo klnase assays and enzymatlc actlvity (data not shown). We have therefom not investigated thts phenomenon further.
Am-Cys-at posnlons 20-23. These cystetnes are known to be Involved in banding to CD4lCD8 (8) thus The N-termmal region of ~5 6~~ contalns the molds -Cys-Val-Cys-at posdlons 3.5 and -Cyr-Glulndlcatlng that these two sequences are freely avallable for lnteranlon with other molecules The found In many sterold receptors. transcrlptlonal activators and transcriptton factors (41). The pooled SPatlal arrangement of these cysteine residues is slmdar lo that of the swalled zmc-tmger domavls column. and bound materlal eluted wlth a gradlent of mdarole as a competitor lor zmc bmding (42) fractions from both tyrosine-agarose peaks A and B were therefore loaded onto a Zn*+-cheIatlng 907 This step proved hlghly succesful. renderlng samples of p56Mat very hiah levels of purily (see Figure   2A. lane 4) and retaining high enzyme activity. Since enzyme adrvily eluted from the column as a very sharp peak during the gradient (see Figure IC) the interaction between p56"and the column must be to the column (eg. Nit+. Cat+, Cu+*. Fe+**, etc.). However. these possibilities have not been very wai l defined. It Is also possible that similar Mecls could be observed by chelating other metal ions investigated.
An a l q w l of the p56M resuninp originally from the tyrosinsagarore peak A was subjected lo a single M o n a Nn (Figure 1D) which removed the last malor contaminant remainlng following h i +chelating chromatography. yielding essentially pure p56" (see Figure 2A. lane 5). The final yield 01 p5BMtr0m the Mono0 column usnQ tyrosmsagarose peak A alone was calculated lo be -1 .l% d the total aaivily present m the original cel l lysate. However, since tyrosinsagarose peak 8 contamed -70% of the told a d w recovered from this column and peak A proved a bener preparation tor further purification. thia number Is slightly misleading. me fracfii resuning from tyrosinsagarose peak 6 was also subjecled to Zn+*-chelating chromatography. and w h o l e the resuning enzyme sample was of a very high Wree d purily. severai higher mobcuiar weight minor contaminants were nlll present (data not shown). Nevertheless, thm sample was still very uselul for in vifro phosphorylatlon experiments since l contained. as tar as could be determined. p56kk as the only proteln kinase adwity (data not shown). We are currently anemptmg lo optimize the tyrosineagarose column condnions lo give the best wsslble samples for application to the Zn+"chelattng column.

EkIrwhorefIc characterization ot bacubvirus expressed p58w
electrophoresis and Western bloning. Autophosphorylated aliquots were added to the SDS gel Fractions resuning from each stage of the purlflcation process were ln#t!aily analysed by gel samples to i) ~n l l r m lwation of the p56kkand it) assess the levels of contammatlon wlth other protein purlluatlon step outloned above. and Figure 26 has the in vilm phosphorylatlon Panerns observed kinases and their substrates. Figure 2A shows the Cwmassie blue-stamed panern obtained alter each following autoradiography Figure 2C  & m e n ; mo&lar &ss. Analysis 01 the size and charge varlants of p56" present In the Westem blmlng (see Figure 4). The panern of spots Observed demonstrates that whlle some of the MonoQ fraction was therefore carrled out by two-dimenslonai eledrophorests and vIsuaiIzed by p56kk specks seen in Figure 2C

Table 111
Comparison