In vivo phosphorylation with TKB1 competent cells
David L. Dankort William J. Muller
Institute for Molecular Biology and Biotechnology, McMaster University, Ontario, Canada
GST-fusion proteins expressed in TKB1 bacteria are used to affinity purify cellular proteins, which interact with a specific autophosphorylation site on the ErbB2/Neu receptor tyrosine kinase (RTK). This approach is aptly suited to rapidly identify phosphotyrosine-protein interactions.
When receptor tyrosine kinases (RTKs) are activated, a number of proteins within the cell are rapidly phosphorylated, with the predominate tyrosine-phosphorylated proteins being the RTKs themselves. RTK tyrosine phosphorylation primarily functions to generate binding sites for cytoplasmic- or plasma membrane-associated proteins involved in transducing proliferative or differentiating signals to the nucleus. These signaling proteins each contain modular Src homology 2 (SH2) or protein tyrosine binding/interacting domains (PTB/PID), which enable them to directly interact with specific receptors in a phosphotyrosine-dependent and sequence-specific manner and, therefore, bind to short linear peptide sequences.1,2,3
Extensive mutagenesis of known autophosphorylation sites in the RTK Neu/ErbB-2 has revealed that four (Y1144, Y1201, Y1227, Y1253) of the five autophosphorylation sites are capable of mediating transformation from the receptor.4 Subsequent coimmunoprecipitation analyses from stable cell lines disclose that SHC and GRB2 proteins bind directly to the receptor at tyrosines 1227 and 1144, respectively. 4
To more quickly identify candidate signaling molecules, we developed an in vitro association assay based on the work of Pawson, et al.,5 using tyrosine-phosphorylated GST-fusion proteins produced in bacteria. Because E. coli do not contain detectable tyrosine kinases, we used the recombinant bacterial strain TKB1, which harbors an inducible Elk6 RTK domain. Plasmids encoding GST-fusion proteins were transformed into TKB1 cells or the parental RTK-deficient BL21 cells, and the fusion proteins were induced with IPTG, which is also capable of inducing expression of the Elk RTK domain in TKB1 bacteria but not in BL21 bacteria (Figure 1). The in vitro association assays carried out with bacteria produced unphosphorylated and phosphorylated fusions that recapitulate associations observed in vivo.
Neu contains five sites of tyrosine phosphorylation (Y1028, Y1144, Y1201, Y1226/7, and Y1253), which we renamed (to simplify) sites A through E in a membrane proximal to distal fashion. Three GST fusion proteins were constructed to individually contain phosphorylation sites Y1201, Y1226/7, and Y1253 (gT-YC, YD, and YE, respectively), as well as a fusion mutated at each of the three sites (gT-F3) (Figure 2). Plasmids encoding each GST-fusion protein were transformed into competent BL21 cells or TKB1 cells (following manufacturers instructions), and the fusion proteins from 3 ml of bacterial cultures were purified on glutathione sepharose as described.7 While Coomassie Blue staining revealed roughly equivalent protein levels from each strain, immunoblot analyses showed that gT-YC, gT-YD, and gT-YE were tyrosine phosphorylated when purified from TKB1 but were not phosphorylated from the parental BL21 strain (Figure 3). The phosphoproteins found in the TKB1 cells expressing gT-F3 likely represented contaminating bacterial proteins, as these have not been repeatedly observed in other experiments; this suggests that TKB1 efficiently induces the tyrosine phosphorylation of GST fusion proteins but does not phosphorylate GST itself.
As we previously determined that SHC proteins bind directly to tyrosine D on Neu (Y1227),4,8 we sought to determine whether these fusion proteins could be used to affinity purify cellular signaling molecules in vitro. Lysates from exponentially growing NIH-3T3 fibroblasts were incubated with gT-YC, gT-YD, gT-YE, and gT-F3 purified from BL21 or TKB1; specifically associated proteins were analyzed by immunoblotting with SHC specific antisera (Figure 3). SHC did not interact with fusion proteins produced in the BL21 strain, yet efficiently bound gT-YD when expressed in TKB1. The lack of binding can neither be attributed to the amounts nor the level of tyrosine phosphorylation of the fusion proteins (Figure 3, upper and middle panels). Thus, because this approach is specific, it permits rapid screenings for interactions with known proteins.
GST-fusion proteins that harbor tyrosine phosphorylation sites can be used to affinity purify cellular signaling proteins in vitro when purified from TKB1 bacteria. These in vitro interactions are highly specific, as demonstrated by the binding of SHC protein to the same site as observed in vivo. We failed to detect proteins that do not interact with these phosphorylation sites (e.g., Ras-GAP) and successfully used this approach to map interaction sites with another SH2-containing protein.8
This approach is superior to other methods: A number of known candidate molecules can be quickly and simultaneously screened by immunoblot analyses and be readily modified to identify novel proteins through the use of labeled cellular extracts. Fusion proteins that contain a single tyrosine residue can be used to map sites of interaction; and interaction sites can be identified which, when mutated in vivo, lead to a loss of protein activity or stability. For example, when two tyrosine residues mutate within the juxtamembrane region of the platelet-derived growth factor receptor, the kinase is rendered catalytically inactive;9 hence, interacting proteins cannot be identified.
Use the tyrosine-phosphorylated fusions for a variety of applications: Carry out direct interactions with soluble tyrosine-phosphorylated proteins and immobilized interaction SH2/PTB domains, both produced in E. coli; screen bacteriophage expression libraries (for a nonisotopic variation of the CORT assay,10 in which novel SH2-containing proteins are cloned using a carboxy-terminal region of receptor tyrosine kinases); and generate and purify tyrosine-phosphospecific antibodies.
Alternatively, phosphotyrosine-containing peptides that are chemically synthesized can also be used to yield similar results. However, synthetic peptide reagents are expensive, nonrenewable, and require prior immobilization for such an assay (e.g., biotinylation or chemical crosslinking to a support matrix). On the other hand, tyrosine-phosphorylated fusion proteins produced in bacteria are a renewable and inexpensive resource.
Tyrosine phosphorylated and unphosphorylated polypeptides are efficiently produced in TKB1 and BL21 bacteria, respectively. These proteins can be used to rapidly define a nd map the association of known SH2- and PTB-containing proteins, accelerating the process of dissecting the role individual phosphorylation sites play in signal transduction from receptor tyrosine kinases.
Pawson, T. (1995) Nature 373: 573-580.
Sadowski, I., et al. (1986) Mol. Cell. Biol. 6: 4396-4408.
van der Geer, P., et al. (1995) Current Biol. 5: 404-412.
Dankort, D.L., et al. (1997) Mol. Cell. Biol. 17: 5410-5425.
Larose, L., et al. (1993) Oncogene 8: 2493-2499.
Lhotak, V., et al. (1991) Mol. Cell. Biol. 11: 2496-2502.
Smith, D.B., et al. (1991) Gene 67: 31-40.
Dankort, D.L. Unpublished observations.
Mori, S., et al. (1993) EMBO J. 12: 2257-2264.
Lowenstein, E.J., et al. (1992) Cell 70: 431-442.
McGlade, C.J., et al. (1992) Mol. Cell. Biol. 12: 991-997.