The unique N-terminal region of SRMS regulates enzymatic activity and phosphorylation of its novel substrate docking protein 1

Authors


Abstract

SRMS (Src-related tyrosine kinase lacking C-terminal regulatory tyrosine and N-terminal myristoylation sites) belongs to a family of nonreceptor tyrosine kinases, which also includes breast tumour kinase and Fyn-related kinase. SRMS, similar to breast tumour kinase and Fyn-related kinase, harbours a Src homology 3 and Src homology 2, as well as a protein kinase domain. However, unlike breast tumour kinase and Fyn-related kinase, SRMS lacks a C-terminal regulatory tail but distinctively possesses an extended N-terminal region. Both breast tumour kinase and Fyn-related kinase play opposing roles in cell proliferation and signalling. SRMS, however, is an understudied member of this family. Although cloned in 1994, information on the biochemical, cellular and physiological roles of SRMS remains unreported. The present study is the first to explore the expression pattern of SRMS in breast cancers, its enzymatic activity and autoregulatory elements, and the characterization of docking protein 1 as its first bonafide substrate. We found that, similar to breast tumour kinase, SRMS is highly expressed in most breast cancers compared to normal mammary cell lines and tissues. We generated a series of SRMS point and deletion mutants and assessed enzymatic activity, subcellular localization and substrate recognition. We report for the first time that ectopically-expressed SRMS is constitutively active and that its N-terminal region regulates the enzymatic activity of the protein. Finally, we present evidence indicating that docking protein 1 is a direct substrate of SRMS. Our data demonstrate that, unlike members of the Src family, the enzymatic activity of SRMS is regulated by the intramolecular interactions involving the N-terminus of the enzyme and that docking protein 1 is a bona fide substrate of SRMS.

Structured digital abstract

  1. SRMS physically interacts with Dok-1 by pull down (View Interaction: 1, 2)
  2. Dok-1 physically interacts with SRMS by anti bait coimmunoprecipitation (View Interaction: 1, 2, 3)
  3. SRMS phosphorylates Dok-1 by protein kinase assay (View interaction)
  4. Dok-1 physically interacts with SRMS by anti tag coimmunoprecipitation (View interaction)
Abbreviations
BRK

breast tumour kinase

DAB

3,3′-diaminobenzidine tetrahydrochloride

DAPI

4′,6-diamidino-2-phenylindole

Dok1

downstream of tyrosine kinases 1/docking protein 1

FRK

Fyn-related kinase

GFP

green fluorescent protein

GST

glutathione S-transferase

HEK 293

human embryogenic kidney 293

IHC

immunohistochemistry

PEI

polyethyleneimine

PTK

protein tyrosine kinase

PΔN-SRMS

partial N-terminus deletion SRMS

SH2

Src-homology 2

SH3

Src-homology 3

SRMS

Src-related tyrosine kinase lacking C-terminal regulatory tyrosine and N-terminal myristoylation sites

ΔN-SRMS

N-terminus deletion SRMS

Introduction

Protein tyrosine kinases (PTKs) comprise a distinct cohort of enzymes that function to phosphorylate the tyrosine residues on other proteins or, alternatively, those that lie within their own sequences, by autophosphorylation [1, 2]. Deregulated activities of tyrosine kinases have been associated with many diseases, notably cancer. The Src family kinases comprise a prominent class of nonreceptor tyrosine kinases, whose family members include Src, Yes, Fyn, Lck, Hck, Blk and Lyn [3, 4]. SRMS (Src-related kinase lacking C-terminal regulatory tyrosine and N-terminal myristoylation sites), also known as PTK70, is a 488 amino acid nonreceptor tyrosine-kinase whose cellular role is unknown. SRMS (pronounced ‘shrims’) is a member of a small family of intracellular Src-related tyrosine kinases, including breast tumour kinase (BRK) or PTK6 and Fyn-related kinase (FRK) or PTK5, called the FRK/PTK6 family [5]. The FRK/PTK6 family is distantly related to the Src-family, although members of this family share an exon–intron structure distinct from the well-characterized Src family of tyrosine kinases [5]. Nonetheless, similar to Src family kinases, the functional structure of the FRK/PTK6 family proteins comprises one Src homology 3 (SH3) domain, one Src homology 2 (SH2) domain and one protein kinase domain [5].

Unlike PTK6 and FRK and other Src family kinases, SRMS is an understudied tyrosine kinase and its biochemical function, as well as its cellular and physiological roles, remains poorly characterized or unknown. SRMS was first cloned in 1994 by Kohmura et al. [6] from mouse embryonic neuroepithelial cells in an attempt to identify the genes that regulate the growth and differentiation of neuroepithelial cells [6]. The study ascribed SRMS expression to most mouse tissues, with the strongest expression levels being reported in the lung, testes and liver [6]. In addition, although SRMS was only poorly detected in embryonic day 11 brain tissues, an augmented and uniform expression of the transcript was observed in tissues derived from embryonic day 15 brain [6]. In 1997, Kawachi et al.[7] found that the SRMS transcript was highly expressed in the murine keratinocytes derived from the nontransformed epidermis sheet compared to tissues derived from melanomas or fibroblasts, suggesting an involvement in proliferation or differentiation of keratinocytes in the skin [7].

Interestingly, the human SRMS maps to chromosome 20q13.33, which is only 1.5 kbp upstream of the PTK6 gene, and this tight linkage not only complicates analysis of the PTK6 regulatory region, but also indicates that both genes may interact genetically [8]. The murine SRMS gene maps to chromosome 2 [6], which is syntenically conserved with the human chromosome 20q13.3, and therefore adjacent to the PTK6 gene in the same [9]. Interestingly, SRMS knockdown mice displayed no obvious phenotypical deviations, suggesting that the function of SRMS may be compensated for by other tyrosine kinases [6]. However, PTK6-null mice displayed an increased intestinal villus length and delayed enterocyte differentiation with a concomitant increase in intestinal epithelial cell proliferation and Akt activity [10]. This suggests that, although SRMS and PTK6 are genetically linked, they may discharge distinct cellular and physiological functions.

Although PTK6 has been catalogued with over 20 characterized substrates and binding partners [8], no SRMS targets have been characterized to date. However, Takeda et al. [11] have described a proteomics study that revealed for the first time the adapter protein docking protein 1 (Dok1/p62) as a potential substrate of SRMS. Dok1 is a scaffolding protein that functions to mediate protein–protein interactions and has been characterized as a substrate of several tyrosine kinases, including the Src family kinases [12-16].

There are several uninvestigated facets regarding the physiological properties of SRMS, including its expression, activity, regulation and substrate specificity. PTK6 and FRK display opposing expression patterns and appear to exert contrasting functions in breast cancers [8, 17], yet the expression pattern and function of SRMS in breast cancers is unknown. Because SRMS lacks the C-terminal autoinhibitory domain, a concise mechanism delineating the regulation of its catalytic activity cannot be predicted. In addition, unlike PTK6, the involvement of the SH3 and SH2 domains of SRMS in intramolecular interactions (i.e. towards autoinhibition) remains unclear. Similarly, the structure–function correlation of the unique, extended N-terminal region of SRMS is unknown. The keys residues that contribute to the catalytic stability of SRMS, as well as the identification of the cellular substrates of SRMS, also remain to be probed.

In the present study, we analyzed the expression of SRMS in breast cancer cell lines and tissues. We also generated a series of SRMS mutants aiming to study the regulation of SRMS localization, activity and substrate specificity. We identify the N-terminus of SRMS as an unforeseen essential element aiding in the perpetuation of enzymatic activity and we validate Dok1 as a bona fide SRMS substrate.

Results

The novel nonreceptor tyrosine kinase SRMS is overexpressed in breast carcinomas

It was shown previously that the closest relative of SRMS, BRK, is overexpressed in the majority of breast cancer samples tested and has oncogenic properties [8, 18-20]. The expression pattern of SRMS in the normal mammary epithelium or breast carcinomas or their derived cell lines has not been reported. We therefore aimed to determine SRMS expression in eight disparate breast cancer cell lines, namely, BT-20, MCF7, MDA-MB-231, MDA-MB-435, MDA-MB-468, Au565, HBL-100 and SKBR3, and to compare its expression with that in a nontumourigenic mammary epithelial cell line, 184B5. We also compared the expression of SRMS with that of its highly characterized family member, BRK. The results obtained indicated a general overexpression of SRMS in six out of eight breast cancer cell lines, with HBL-100 exhibiting the highest levels of SRMS and MDA-MB-468 and AU565 displaying a lower expression (as demonstrated with a longer exposure of the X-ray film to the immunoblotted membrane) (Figs 1A and S1). Interestingly, we observed that SRMS was significantly downregulated in the nontumourigenic mammary epithelial cell line, 184B5. By contrast, BRK expression, although also observed in the majority, was found to clearly evade certain cell lines, namely MDA-MB-435, MDA-MB-468 and HBL-100. The data suggest that (a) SRMS, akin to BRK, is potentially overexpressed in the majority of breast cancer cell lines and (b) the expression of SRMS appears to evade the normal human breast cells at the same time as being significantly expressed in those derived from breast carcinomas.

Figure 1.

(A) SRMS is overexpressed in the human breast carcinomas. Lysates prepared from breast carcinoma cell lines and a cell line derived from the normal mammary epithelial tissue (184B5) were evaluated for SRMS expression via immunoblotting with antibodies against SRMS. β-tubulin was used as the loading control. (B) A representative image of SRMS expression via IHC on biological specimens procured from the breast cancer and normal adjacent tissues of female breast cancer patients. Tissue sections corresponding to (a) and (b) represent Grade 1; (c) and (d) represent Grade 2; and (e) and (g) represent Grade 3 breast carcinoma samples. (i, j) Adjacent normal breast tissues. Tissue sections were stained with anti-SRMS serum and specific binding was detected with ImmPRESSTM reagent followed by colour development in peroxidase substrate DAB.

To corroborate these findings, we evaluated the expression of SRMS in human breast tissues and the normal mammary epithelium by immunohistochemistry (IHC). Tissue sections were derived from invasive ductal carcinomas of increasing pathological grade and the normal tissue sections were derived from adjacent sites. Representative images are shown in Fig. 1B. We detected a strong staining for SRMS in all carcinoma cases (Fig. 1B, a–g) compared to normal tissue controls (Fig. 1B, i,j). SRMS intensity in the normal tissues ranged from +1 to +2. However, we observed intensities of +2 in grade 1 (Fig. 1B, a,b) and grade 2 (Fig. 1B, c,d) and +3 in grade 3 (Fig. 1B, e,g) invasive carcinoma samples. The specificity of the SRMS antibody was confirmed via staining with normal rabbit serum alone as a control (Fig. S2). Taken together, these findings suggest that the expression of SRMS is induced more in breast cancers compared to the normal mammary gland. Furthermore, both the breast carcinoma and the normal tissue samples exhibited cytoplasmic and nuclear staining of SRMS (Table S1). Details of the SRMS staining pattern across different pathological breast tissue samples are compiled in Table S1.

SRMS localizes to punctate cytoplasmic structures

Figure 1 shows that SRMS is highly expressed in breast cancer cell lines and tumours. The subcellular localization of SRMS is unknown and, in the absence of a putative myristoylation signal (dictating plasma-membrane localization) as found in Src [4] or a characteristic FRK-like nuclear localization signal (directing nuclear expression) [19], it was imperative to determine its unpredictable cellular localization pattern. To determine the true native and characteristic intracellular habitation of the protein, we began by pursuing an immunocytochemical analysis on endogenous SRMS in two breast cancer cell lines, MDA-MB-231 and AU565, and one cervical cancer cell line, HeLa. We found that endogenous SRMS localized to distinct punctate cytoplasmic structures in all three cell lines (Fig. 2A). Interestingly, the same pattern of localization was also detected in one other breast cancer cell line, SKBR3 (data not shown). No obvious nuclear co-localization with 4′,6-diamidino-2-phenylindole (DAPI) (Fig. 2A, c, f, i) was observed, indicating that SRMS is predominantly a cytoplasmic protein. Next, we aimed to examine whether the endogenous localization pattern coincided with ectopically-expressed SRMS. A green fluorescent protein (GFP)-SRMS construct was used to transiently transfect the human embryogenic kidney 293 (HEK 293) cell line and localization was determined using fluorescence microscopy. The nuclei were stained with DAPI. Consistent with endogenous SRMS, overexpressed SRMS also produced a punctate localization pattern in the cytosol (Fig. 2B, a,c), whereas GFP alone was expressed diffusely throughout the cell (Fig. 2B, d,f). These findings suggest that SRMS is a cytoplasmic tyrosine kinase that localizes to specific punctate structures. Nevertheless, the localization of SRMS in sites of the cellular membrane was not abundantly clear from fluorescence microscopy analysis.

Figure 2.

SRMS localizes to punctuate cytoplasmic structures in vivo. (A) Intracellular localization of endogenous SRMS was detected via indirect immunofluorecence in the MDA-MB-231 and AU565 human breast cancer cell lines, as well as the human cervical cancer cell line, HeLa. Immunoreactivity was visualized using primary anti-SRMS antibodies and fluorescein isothiocyanate-conjugated secondary antibodies. Cells were counterstained with DAPI (blue). (B) HEK 293 cells were transiently transfected with plasmids encoding GFP-SRMS or GFP. The intracellular localization of exogenous GFP-tagged SRMS (a, c) and GFP alone (d, f) were detected via fluorescence microscopy. Cells were counterstained with DAPI (b, c, e, f). (C) Cells (from the indicated cell lines) were fractionated into the cytosolic, membrane and nuclear fractions and immunoblotted for the detection of SRMS. β-tubulin and Sam68 were used as controls for the cytosolic/membrane and nuclear compartments, respectively. SRMS is found in the cytosolic fraction of the indicated cell lines.

To further substantiate these results and to specifically determine whether SRMS localizes to the membrane, a subcellular fractionation was employed in MDA-MB-231 and HeLa cell lines, including one other breast cancer cell line, HBL-100. Indeed, out of the three segregated cellular fractions, SRMS was found to be exclusive to the cytosolic fraction in all three cell lines, with no detectable levels determined in either the membrane or the nuclear fractions (Fig. 2C). Immunoblotting with antibodies against Sam68 and β-tubulin served as controls for the enrichment of nuclear and cytosolic/membrane proteins respectively. Therefore, using two different approaches (immunofluorescence microscopy and a subcellular fraction), we have demonstrated that SRMS has a cytosolic localization in breast cancer cells, as well as HeLa cervical cancer cells.

The unique N-terminus region of SRMS regulates it tyrosine kinase activity

Although SRMS belongs to the same family of nonreceptor tyrosine kinases as BRK and exhibits significant structural homology with this PTK, there are subtle variations in their overall primary amino acid sequence. First, SRMS lacks the C-terminal regulatory tail found in BRK and Src family kinases. Second, SRMS has an extended and unique N-terminus region, which is absent in other Src-related kinases (Fig. 3A). To understand the functional significance of this N-terminal region, to assess the enzymatic activity of SRMS and to identify key regulatory elements within the structure, we generated a series of GFP-tagged SRMS mutants. We constructed a SRMS mutant lacking the first 51 amino acids (ΔN-SRMS). We also generated K258M, Y342F and W223A mutants. These residues are conserved in BRK (K219, Y342 and W184) and Src-family kinases including, but not limited to, c-Src (K298, Y419 and W263). Y342 and Y419 in BRK and c-Src, respectively, lock the catalytic loop in the fully-active conformation when autophosphorylated [21-23]. Previous studies have demonstrated that BRK is activated by autophosphorylation at Y342 because mutation of this site to alanine drastically reduced BRK activity [21]. K219 and K298 are critical for ATP binding and K219M and K298M mutants are kinase defective, dominant inhibitory forms of the enzymes [22-24]. The W184A mutant of BRK was shown to abolish kinase activity [22], whereas the analogous mutant in Src-family kinase, Hck, increased enzyme activity [25]. The enzymatic activity of wild-type SRMS and the contribution of the analogous residues in the regulation of SRMS activity have not been reported.

Figure 3.

The N-terminus region of SRMS and its SH2 domain sustain a constitutively active form of the kinase. (A) Wild-type SRMS containing the globular functional domains; SH3, SH2 and the kinase domain. SRMS mutants; ΔN-SRMS (N-terminally deleted), PΔN-SRMS (Partial N-terminus deletion), ΔSH3 (SH3 deletion), ΔSH2 (SH2 deletion) and its point mutated variants for the indicated amino acid residues are also shown. The critical regulatory residues W223A, K258 and Y380 are conserved with BRK and FRK. (B) Lysates prepared from HEK-293 cells transfected with the indicated SRMS and BRK constructs were evaluated to determine the relative kinase activity of the variants using antibodies against total phosphotyrosines (bottom). Expression of the ectopic proteins was probed via anti-GFP (top). (C) Wild-type BRK and wild-type SRMS, as well as its indicated mutants, were immunoprecipitated from HEK 293 cell lysates and immunoblotted with anti-phosphotyrosine (top) to reveal autophosphorylation of the respective proteins. The immunoprecipitated proteins were probed via anti-GFP (bottom). SRMS auto-phosphorylation is strongly diminished in its Y380F mutant. (D) Transiently transfected variants of SRMS were analyzed for relative catalytic activity compared to wild-type SRMS via immunoblotting with phosphotyrosine antibodies (top). Anti-GFP was used to probe the ectopically-expressed proteins (middle). Anti-SRMS sera, targeting an epitope in the N-terminus region of SRMS, depict abrogated immunoreactivity with ΔN-SRMS and PΔN-SRMS (bottom).

To assess the overall activity of SRMS and its mutants, including ΔN-SRMS, K258M and W223A (Fig. 3A), we transiently transfected the GFP-tagged variants along with the wild-type protein in HEK 293 cells. For comparison, we also transfected the GFP-tagged BRK wild-type and two other mutants (kinase-dead BRK-K219M and constitutively active BRK-Y447F) and evaluated the total levels of tyrosine phosphorylation in cell lysates by immunoblotting using PY20, an antibody against phosphorylated tyrosines. First, we noted that wild-type SRMS exhibited strong intrinsic tyrosine kinase activity that is almost comparable to that of the wild-type BRK but not to that of its hyperactive mutant BRK-Y447F (Fig. 3B). Second, and more significantly, we noted that deleting the 51 amino acid chain in the N-terminus region of SRMS (ΔN-SRMS) totally abolished kinase activity. Third, we observed an anticipated absence of kinase activity with the K258M mutant because abrogating the ATP-contacting site is expected to eliminate the intrinsic ability of the kinase to acquire and utilize the phosphate group derived from ATP for tyrosine phosphorylation. Because autophosphorylation of Src family kinases including BRK is a measure of enzyme activation [21, 25], we aimed to examine whether the SRMS activity levels observed in Fig. 3B corresponded with the autophosphorylation of SRMS. We therefore analyzed the degree of autophosphorylation of wild-type SRMS and its three mutants, ΔN, W223A and Y380F. Y380 is the predicted autophosphorylation site within the kinase domain of SRMS. BRK wild-type was used as a positive control. The GFP-tagged constructs were transiently transfected into HEK 293 cells and the expressed proteins immunoprecipitated from the cell lysates using anti-GFP sera. The immunoprecipitates were then subjected to immunoblotting with anti-phosphotyrosine (PY20). As shown in Fig. 3C, although wild-type SRMS and BRK displayed prominent autophosphorylation, the SRMS-Y380F exhibited markedly reduced autophosphorylation, thereby conforming that this tyrosine residue is indeed the site for autophosphorylation in SRMS. Furthermore, with a longer exposure of the X-ray film, a subtle degree of autophosphorylation was noted in the W223A mutant, whereas ΔN-SRMS displayed null autophosphorylation (data not shown).

We extended these studies to include additional SRMS mutants (Fig. 3D). The importance of the SH2 and SH3 domains in regulating the catalytic activity of BRK via intramolecular interactions has been demonstrated previously [20, 21]. To determine the involvement of these globular domains in the regulation of the catalytic activity of SRMS, we generated GFP-tagged SH3- and SH2-deleted mutants of SRMS (ΔSH3-SRMS and ΔSH2-SRMS, respectively). We also mutated a conserved arginine residue (R147) to alanine (R147A). This arginine residue resides in the SH2 domain and is part of the FLVRS motif that constitutes the phosphotyrosine recognition site in Src family kinases [26]. The analogous R175 in v-Src makes contact with the phosphate group on tyrosine residues [26], and mutating this residue was shown to abrogate phosphotyrosine binding [27]. Furthermore, mutation of an analogous arginine (R105) in BRK resulted in reduced phosphorylation and association with its substrate STAP2 (signal transducing adaptor protein 2) [28]. Nonetheless, although ΔSH2 BRK mutant exhibits elevated kinase activity [25], the effect of the spatially disruptive R105 mutation on catalytic activity has not been reported. Therefore, we aimed to compare an analogous mutation in SRMS (R147) with its ΔSH2 mutant to evaluate catalytic repercussions. Figure 3B,C shows that the N-terminal region plays a critical role in stabilizing the intrinsic tyrosine kinase activity of SRMS. This N-terminus is replete with proline residues, where six such prolines, located proximal to the SH3 domain, were found to be closely stacked in a manner reflective of a discrete proline-rich motif. To assess its requirement wiith respect to SRMS kinase activity, we deleted this proline-rich segment to generate a partial N-terminal deletion mutant (PΔN-SRMS). The SH2-kinase linker region of BRK contains proline residues that exhibit the PXXP SH3-binding motif and were also shown previously to form intramolecular interactions with the SH3 domain of BRK [22]. This linker region in SRMS contains three proline residues, albeit not conforming to a PXXP type motif (Fig. 3A). Therefore, we constructed two different mutants with knowledge of the spatial arrangement of the three prolines. One construct contained a single mutation of the proline (P214A and designated SRMS-PA) residing closest to the SH2 domain. The other two prolines, situated distally from the SH2 domain, were mutated together (P218A/P226A and referred to as SRMS-2PA) in the second construct. All of these mutants, together with wild-type SRMS, ΔN-SRMS, SRMS-K258M and SRMS-W223A, were transiently transfected in HEK 293 cells to test the overall effects on kinase activity. Cell lysates were analyzed for immunoreactivity with anti-phosphotyrosine sera to measure the relative kinase activity (Fig. 3D, top). The variants were generally expressed at comparable levels (Fig. 3D, middle and bottom). Of all the SRMS mutants examined, ΔSH2-SRMS (lane 7) displayed the lowest kinase activity compared to the wild-type SRMS, suggesting that the SH2 domain is essential for SRMS kinase activation. However, deletion of the SH3 domain did not significantly affect kinase activity (lane 8). P214A (lane 4) and 2PA (lane 5) mutations in the linker region also had little effect on kinase activity. Furthermore, the R147A mutation within the SH2 domain displayed kinase activity that was consistent with that of wild-type SRMS (compare lanes 1 and 6). Also, the PΔN-SRMS failed to alter the catalytic activity of SRMS, suggesting that the short segment rich in proline residues does not display an exclusive role with respect to regulating SRMS tyrosine kinase activity. Taken together, our data demonstrate that the kinase activity of SRMS is regulated by its unique N-terminus sequence. Furthermore, by contrast to BRK [22], deletion of the SH2 domain of SRMS resulted in significantly reduced kinase activity.

SRMS mutants display diverging subcellular localization patterns

Figure 2A,B shows that both endogenous SRMS and ectopically-expressed GFP-SRMS localized to punctate cytoplasmic structures. The mechanisms regulating this punctate cytoplasmic localization are unknown. It is also not known whether this localization is dictated by intermolecular interactions with SRMS targets. It is conceivable, however, that the interaction of SRMS with certain endogenous cellular targets via SH3 and/or SH2 domains might result in the retention of the protein kinase in specific cellular compartments. It is also possible that the unique, extended N-terminus region of SRMS regulates not only the activity of the enzyme, but also its subcellular localization via an unknown intermolecular interaction. We therefore tested the abilities of the N-terminal region, as well as the SH3 and SH2 domains, of SRMS to influence subcellular localization. Accordingly, full-length GFP-SRMS, as well as its GFP-tagged variants, ΔSH3-SRMS, ΔSH2-SRMS, ΔN-SRMS and SRMS-K258M, were expressed in HEK 293 cells and their localization was assessed by fluorescence microscopy (Fig. 4). As shown in Fig. 4 and consistent with Fig. 2B, wild-type SRMS localizes to punctate cytoplasmic structures in over 90% of the transfected cells. However, approximately 70% of the cells transfected with ΔN-SRMS exhibited a diffused localization, whereas the remainder displayed a punctate pattern of localization. Although almost 80% of the cells transfected with ΔSH3-SRMS localized to punctate cytoplasmic structures. ΔSH2-SRMS exhibited a diffused localization pattern in approximately 90% of transfected cells. Unexpectedly, SRMS-K258M displayed a cytoplasmic or nucleo-cytoplasmic diffused localization pattern in approximately 80% of transfected cells. Taken together, our data suggest that the N-terminal region of SRMS not only regulates the activity of the enzyme, but also can modulate the subcellular localization. Our findings also indicate that the SH2 domain of SRMS may contribute to sequester SRMS in specific cytoplasmic structures.

Figure 4.

SRMS mutants exhibit divergent subcellular localization. GFP-tagged wild-type SRMS, as well as its mutants, ΔN-SRMS, K258M, ΔSH3 and ΔSH2, were transiently transfected in HEK 293 cells and localization was determined via fluorescence microscopy. ΔN-SRMS localization was found to be diffused in a major proportion of the transfected cell population. K258M exhibited a diffused cytoplasmic/nucleo-cytoplasmic pattern. ΔSH2 displayed a predominantly diffused localization, whereas ΔSH3 revealed predominant punctate localization pattern.

Dok1 is differentially expressed in breast cancer

SRMS is a tyrosine kinase, although no SRMS substrate has been characterized to date. Recently, Takeda et al. [11] described a proteomics study that revealed potential substrates of the human SRMS protein for the first time. One of these potential substrates was the adapter protein, Dok1. Dok1 has not been characterized as a SRMS substrate and/or binding partner. Dok1 was first identified as an abundant tyrosine-hyperphosphorylated protein in chronic myelogenous leukaemia cells [29]. The scaffolding protein has also been validated as a substrate of various tyrosine kinases, including Src family kinase members [12, 30]. Dok1 is a reported tumour suppressor and has been shown to inhibit cell proliferation and leukaemogenesis, and to promote apoptosis [31, 32]. The gene encoding this adapter protein is located in the human chromosome 2p13, a locus that is unstable in various human tumours [33]. Gene and protein expression studies have shown that Dok1 is repressed in several human cancers, including head and neck cancer, lung, liver and gastric cancers, and Burkitt's lymphoma [34]. To our knowledge, the expression pattern of Dok1 and interaction with its regulators in breast cancer cells has not been reported to date. Therefore, we analyzed Dok1 expression in various breast cancer cell lines, as well as in the nontumourigenic breast epithelial-derived 184B5, and correlated this with the expression of SRMS (Fig. 5A). We show that Dok1 expression is low or absent in four cell lines, namely 184B5, MDA-MB-468, AU565 and SK-BR3 (Fig. 5A, lanes 6, 7 and 9). Three of these cell lines (184B5, MDA-MB-468 and AU565) also show decreased levels of SRMS. Interestingly, the elevated expression of Dok1 in the other cell lines, especially in HBL-100, corresponded with the expression of SRMS (Fig. 5A, lane 8). Taken together, these data reveal a positive correlation between Dok1 and SRMS expression in the breast cancer cells studied.

Figure 5.

Dok1 is differentially expressed in breast cancer cells. (A) Dok1 expression was surveyed alongside SRMS in eight breast cancer cell lines, as well as the normal breast epithelial, 184B5, using antibodies against Dok1 and SRMS, respectively. β-tubulin was used as the loading control. (B) Endogenous SRMS and Dok1 co-localization was determined via indirect immunofluorescence in the HBL100 cell line using antibodies against SRMS (a) and Dok1 (b). Cells were counterstained with DAPI (c). The composite images are shown in (d), (e), (f) and (g).

Figure 2A,B shows that SRMS localizes to punctate cytoplasmic structures. Previous studies have shown that Dok1 has a predominantly cytoplasmic/membrane localization [35], although the protein can also shuttle between the nucleus and the cytoplasm [15]. We next investigated whether SRMS and Dok1 co-localized in vivo. Using antibodies specific to SRMS and Dok1, we found that SRMS and Dok1 display different localization patterns (Fig. 5B). As with AU565 and MDA-MB-231, SRMS displays punctate cytoplasmic localization in HBL100 cells (Fig. 5B, a,d). Although Dok1 is predominantly nuclear, minimal cytoplasmic staining was also observed (Fig. 5B, b,e). Overall, the merged images did not reveal significant co-localization (Fig. 5B, f,g). A similar observation was made with HEK 293 cells (Fig. S3). Furthermore, via immunoprecipitation analysis using anti-Dok1 sera, endogenous SRMS failed to co-precipitate with endogenous Dok1 from the MDA-MB-231 and HBL-100 cell lysates, indicating the absence of binding interactions between the endogenous proteins (data not shown).

SRMS directly interacts with Dok1 via its SH3 and SH2 domains

SRMS has SH3 and SH2 domains, which, in Src-family kinases, are known to interact with proline-rich motifs and phosphorylated tyrosine residues, respectively. The C-terminal axis of Dok1 harbours several proline residues and numerous tyrosine residues that are potential targets for phosphorylation by Src-family kinases (Fig. 6A). Because Dok1 was identified as a potential target of SRMS [11], we first investigated whether both proteins interact. Accordingly, we transiently transfected GFP-Dok1 and GFP-SRMS alone or together in HEK 293 cells and subjected the cell lysates to immunoprecipitation with antibodies against Dok1 and SRMS, followed by immunoblotting with the same antibodies. As shown in Fig. 6B (lane 7), the co-presence of GFP-SRMS and GFP-Dok1 in the same immunocomplex was detected reciprocally in the lysates immunoprecipitated with anti-Dok1 and anti-SRMS, respectively. To determine whether Dok1 interacts with ΔN-SRMS and to confirm that such an interaction between the exogenous GFP-tagged proteins is not a result of GFP dimerizaton, HEK 293 cells, transiently transfected with either pEGFP control vector or co-transfected with pEGFP and GFP-Dok1 or GFP-Dok1 and ΔN-SRMS, were subjected to immunoprecipitation with anti-Dok1 sera and immunoblotted with anti-GFP antibodies. It was found that (a) the interaction between the proteins is not via GFP dimerization (Fig. 6C, lane 4) and (b) Dok1 interacts with ΔN-SRMS (Fig. 6C, lane 7). These results suggest that SRMS physically interacts with Dok1 and that the N-terminal region of SRMS, although being indispensible for sustaining its kinase activity, is expendable for binding interactions of the enzyme with its targets.

Figure 6.

SRMS interacts with Dok1 in vivo and in vitro. (A) Schematic diagram of Dok1 depicting its two functional domains, namely the pleckstrin homology (PH) domain and the phosphotyrosine-binding (PTB) domain, as well as a proline and tyrosine-rich C-terminal axis. (B) HEK 293 cell lysates from GFP-SRMS, GFP-Dok1 or GFP-SRMS/GFP-Dok1 co-transfected cohorts were subjected to immunoprecipitation with anti-Dok1 and immunoblotted with Dok1 and SRMS (top). Conversely, SRMS was immunoprecipitated from such lysates using anti-SRMS and the immunoprecipitates probed for SRMS and Dok1 (bottom). Anti-IgG (rabbit) was used as the control. Total cell lysates indicate the relative expression of the proteins. (C) Cell lysates were subjected to immunoprecipitation with anti-Dok1 in the GFP-Dok1/pEGFP or GFP-Dok1/GFP-ΔN SRMS co-transfected cohorts and immunoblotted with antibodies against GFP to probe for GFP dimerization-mediated interactions between the ectopic proteins. (D) Endogenous Dok1 from HEK 293 lysates binds to the purified recombinant GST-SH3 domain of SRMS in a GST pull-down assay, as demonstrated upon immunoblotting with anti Dok1 (left). Expression of the bacterially expressed proteins is shown via Coomassie Blue staining (right). (E) Tyrosine phosphorylated endogenous Dok1 from GFP-SRMS-transfected HEK 293 cell lysates binds to the GST-SH2 domain of SRMS, as shown via immunoblotting with anti-Dok1. Expression of the GST-fused proteins is shown in the Coomassie Blue stained image of the gel (right). (F) Tyrosine phosphorylation of endogenous Dok1 upon exogenous expression of GFP-SRMS was probed with phosphotyrosine antibodies. (G, H) Endogenous Dok1 binding to GST-SH2 domain of SRMS was examined upon exogenous expression of the pEGFP vector control or GFP-SRMS-K258M (kinase inactive mutant). (I) The expression of the GST-fused proteins is shown via blotting with anti-GST sera.

To determine whether SRMS interacts with Dok1 via its SH3 and/or SH2 domains, we performed in vitro binding assays using glutathione S-transferase (GST)-fused SH3 and SH2 domains of SRMS. First, we evaluated the association between endogenous Dok1 and GST-SH3 using lysates from untransfected HEK 293 cells. Using antibodies against Dok1, we observed interactions between GST-SH3 and Dok1 (Fig. 6D, left). However, we did not observe such binding interactions with the GST-SH2 protein (data not shown). Because the SH2 domain interacts with phosphorylated tyrosine residues, we transiently transfected HEK 293 cells with GFP-SRMS to determine whether ectopically-expressed SRMS promote the binding of endogenous Dok1 to the SH2 domain of SRMS. GST-SH2 was found to form a complex with tyrosine phosphorylated endogenous Dok1 in the presence of GFP-SRMS (Fig. 6E,H, left). However, a SRMS GST-SH2 association with endogenous Dok1 was not observed in the presence of exogenously expressed pEGFP vector or GFP-SRMS K-M (kinase dead) (Fig. 6F,G) demonstrating that tyrosine phosphorylation is necessary for Dok1 interactions with SRMS GST-SH2. The expression and molecular sizes of the GST and GST-SH3 and GST-SH2 proteins are shown by Coomassie Blue staining and immunoblotting with anti-GST sera (Fig. 6D,E,I). GST alone bound to beads was used as a control and total cell HEK 293 lysates were used to determine the relative expression of endogenous Dok1. Taken together, these data show that both the SH3 and SH2 domains of SRMS mediate interactions with Dok1.

Dok1 is a direct substrate of SRMS

To characterize and validate Dok1 as a SRMS substrate, we first transiently transfected HEK 293 cells with GFP-Dok1, in the presence or absence GFP-SRMS. Dok1 was immunoprecipitated from cell lysates using antibodies against Dok1 and the immunoprecipitates were immunoblotted with antibodies against phosphotyrosine, Dok1 and SRMS. The results obtained showed an implicit tyrosine phosphorylation of the ectopically-expressed Dok1 in cells overexpressing SRMS (Fig. 7A, top, lane 4). Autophosphorylation of GFP-SRMS was also detected in the total cell lysates (Fig. 7A, top, lane 6). In addition, overexpressed SRMS co-immunoprecipitated with Dok1 as detected by anti-SRMS serum (Fig. 7A, middle, lane 4), further validating the interaction between Dok1 and SRMS. The expression of GFP-Dok1 in the immunoprecipitates is also shown (Fig. 7A, bottom). These data indicate that ectopically-expressed GFP-SRMS interacts with and mediates the phosphorylation of overexpressed GFP-Dok1.

Figure 7.

Dok1 is a bonafide substrate of SRMS. (A) Dok1 was immunoprecipitated from GFP-Dok1 alone or GFP-Dok1/GFP-SRMS co-transfected cohorts of HEK 293 cell lysates and probed for tyrosine phosphorylation against phosphotyrosine antibodies (top), SRMS (middle) and Dok1 (bottom). Anti-IgG (rabbit) was used as a control. Total cell lysates were used to indicate the relative expression of both proteins. (B) Dok1 was immunoprecipitated from GFP-SRMS or GFP-ΔN-SRMS transfected cohorts of HEK 293 cell lysates and immunoblotted for Dok1 (top), SRMS (middle) and phosphotyrosine (bottom). Total cell lysates were used to indicate the relative expression of the respective proteins. Anti-IgG (rabbit) was used as a control. (C) Schematic representation of the 5 Dok1 mutants that were constructed from wild-type GFP-Dok1(Dok-WT). Each C-terminally truncated mutant contains an increasing number of tyrosine residues. (D) Dok1 mutants were transfected either alone or with mcherry-SRMS in HEK 293 cells and total cell lysates were used for immunoblotting with antibodies against phosphotyrosine (top), GFP (middle) and SRMS (bottom). (E) HEK 293 cells were co-transfected with the Dok1 mutants and mcherry-SRMS and subjected to immunoprecipitation with anti-GFP (rabbit) antibodies. Immunoprecipitates were probed for tyrosine phosphorylation using antibodies against phosphotyrosine (top), Dok1 using anti-GFP (mouse) (middle) and SRMS using anti-SRMS (bottom). (F) An in vitro kinase assay was performed using the active kinase, GST-SRMS, and the substrate, GST-Dok1, in the presence or absence of ATP. Tyrosine phosphorylation was probed via anti-phosphotyrosine sera (top) and expression of the GST-fused proteins via anti-GST (bottom).

Upon exogenous expression of GFP-SRMS in HEK 293 cells, endogenous Dok1 bound to the purified GST-SH2 domain of SRMS in a GST pull-down assay (Fig. 6E). To validate that such interaction was a result of tyrosine phosphorylation of endogenous Dok1 by GFP-SRMS, we transiently transfected GFP-SRMS or ΔN-SRMS in HEK 293 cells and immunoprecipitated Dok1 via anti-Dok1 sera. The immunoprecipitates were then immunoblotted with antibodies against phosphotyrosine, SRMS and Dok1. We demonstrate that GFP-SRMS binds to and phosphorylates endogenous Dok1 (Fig. 7B, middle, lane 4; bottom, lane 4). Furthermore, the finding that ΔN-SRMS co-immunoprecipitated with endogenous Dok1 corroborates the results shown in Fig. 6C, attesting to the dispensability of the 51 amino acid N-terminus sequence with respect to regulating SRMS substrate recognition and interaction.

As shown in Fig. 6A, Dok1 possesses a dense array of prolines and tyrosine residues along its C-terminal segment. In an attempt to map the cluster of tyrosine residues phosphorylated in the presence of SRMS, we generated five deletion mutants of Dok1, each containing a progressively increasing number of tyrosine residues (Fig. 7C). Wild-type Dok1 and its mutants were transiently transfected in HEK 293 cells, either alone or with mcherry-SRMS. The Dok1 mutants were immunoprecipitated with anti-GFP sera and the immunoprecipitates and total cell lysates were immunoblotted with antibodies against phosphotyrosine, GFP and SRMS. Immunoblotting the total cell lysates (Fig. 7D, top) revealed that, besides wild-type Dok1 (lane 12), Dok-Δ3 (1-345), Dok-Δ4 (1-380) and Dok-Δ5 (1-415) were all phosphorylated in the presence of mCherry-SRMS. Barring Dok-Δ1(1-259) as the only exception, Dok-Δ2 (1-317) was found to be phosphorylated upon longer exposure of the X-ray film to the immunoblotted membrane (data not shown). The expression levels of transfected GFP-Dok1 mutants (Fig. 7D, middle) and mcherry-SRMS proteins (Fig. 7D, bottom) are shown. Immunoblotting performed on the immunoprecipitates also demonstrated identical results (Fig. 7E), which confirmed that all Dok1 mutants, except Dok-Δ1, were phosphorylated in the presence of ectopically-expressed mCherry-SRMS. Furthermore, the finding that co-immunoprecipitation of ectopic SRMS progressively diminished with the smaller Dok1 mutants (Fig. 7E, bottom, lanes 1–6) highlights the underlying significance of the C-terminal proline residues in mediating binding interactions with SRMS. Taken together, these data provide strong evidence that Dok1 is a potential substrate of SRMS.

To determine whether Dok1 was a direct substrate of SRMS, we performed an in vitro kinase assay. We incubated GST-SRMS and GST-Dok1 in the presence or absence of ATP in a kinase reaction. The proteins were resolved by SDS/PAGE and analyzed by immunoblotting using anti-phosphotyrosine sera. It was found that GST-Dok1 was strongly phosphorylated in the presence of GST-SRMS and ATP (Fig. 7F, lane 3, top). The activity of SRMS was confirmed by the autophosphorylation of GST-SRMS in the presence and absence of GST-Dok1 (Fig. 7F, lanes 3 and 1). Immunoblotting with anti-GST (Fig. 7F, bottom) was used as a control for the presence of the GST-tagged proteins. The data reported in the present study are the first to indicate that Dok1 is a direct and bona fide substrate of SRMS.

Discussion

The nonreceptor tyrosine kinase, SRMS, was cloned in 1994 but, subsequently, has remained understudied. Nothing is known or has been reported about its enzymatic activity, substrate identification and recognition, nor its expression profile in human tissues, cell lines or carcinomas. The urgency to study this mysterious protein appeared to be even more compelling because SRMS shares a conservative structural commonality with its other well-characterized family members, BRK and FRK. The present study is the first of its kind to investigate and unravel (a) the expression profile of SRMS in breast cancer cell lines and tissues and (b) the mechanisms for SRMS autocatalytic regulation and substrate specificity. Along these lines, we have established that: (a) SRMS is overexpressed in the breast cancer cell lines studied and breast carcinomas but is found to be low in human mammary epithelial cells obtained from normal tissues and normal mammary tissues; (b) the endogenous and ectopically-expressed SRMS localizes to punctate cytoplasmic structures; (c) the N-terminal region of SRMS is an essential element regulating its enzymatic activity; and (d) Dok1 is a cellular target and a direct substrate of SRMS.

SRMS was originally cloned from the murine embryonic neuroepithelial cells and the mouse skin, and was found to be strongly expressed in mouse lung, testes and liver, in addition to the epithelial keratinocytes [6, 7]. Our data show that SRMS is expressed or overexpressed in all human breast carcinoma samples tested compared to BRK whose expression is observed in most (but not all) cell lines (Fig. 1). However, we detected only diminutive SRMS expression in the nontumourigenic/normal breast epithelial cell line, 184B5, and in normal tissues. The SRMS gene, similar to that of BRK, comprises eight exons and, because the former localizes to chromosome 20q13.33, mapping within 1.5 kbp upstream of BRK, both PTK genes remain tightly linked [8]. This locus, notably, resides in a region of the genome that is frequently amplified in breast cancers [36]. Studies on BRK have shown that the protein is expressed in more than 60% of breast tumours and breast cancer-derived cell lines, whereas it is absent in normal mammary tissues and benign lesions [8]. BRK is also overexpressed in various other types of cancers, including those of the ovary, colon, head and neck, and prostate [9, 37-41]. The dramatic induction of BRK in a significant percentage of human breast tumours suggests a role for BRK in the aetiology of breast cancers. BRK has oncogenic properties and can promote cell proliferation, migration and tumour formation [8, 20]. Consistent with its potential role in tumourigenesis, BRK has also been shown to associate with epidermal growth factor receptor, as well as enhance the mitogenic signals of EGF and promote the recruitment of phosphatidylinositol 3-kinase, in addition to activating Akt and mitogen-activated protein kinase [42, 43]. By contrast, the other BRK family protein, FRK, is a putative tumour suppressor in breast cancer and has been shown to inhibit cell growth and suppress tumourigenesis [44]. One mechanism of action of FRK is via the phosphorylation and stabilization of the tumour suppressor phosphatase and tensin homologue, which results in the inhibition of Akt signalling [45]. The human FRK gene maps to the chromosomal locus 6q21-23, a region that is destabilized by a loss of heterozygosity in 30% of breast tumours [46] and 40% of melanomas [47]. Our data from IHC analysis indicate that SRMS expression is associated with the mechanistic onset of human breast cancer and is apparently gradually augmented in vivo as the disease aggravates with time (Fig. 1B). This is also indicative of an intricate involvement of SRMS in the subcellular molecular aberrations occurring during the progressive transformation from a presumably early lesion-type condition to a full-blown cancer phenotype. Although the present study is the first to examine SRMS expression in breast cancer samples, given the significance of the investigation's outcome, it is highly likely that, similar to BRK and FRK, induction of SRMS in breast cancer correlates with the aetiology of the disease.

Although SRMS has three functional domains (an SH3 domain, an SH2 domain and a tyrosine kinase catalytic domain) that are conserved in Src family kinases, the primary amino acid sequence of SRMS shows the presence of a 51 amino acid extended N-terminal region. Another striking characteristic of SRMS is the absence of a C-terminal regulatory tail present in most Src-family kinases, including BRK and FRK. In addition, unlike the Src-family members, SRMS lacks myristoylation and palmitoylation membrane-anchoring signals, which potentially renders the protein a reasonable degree of flexibility in subcellular localization, and therefore also bestows upon it a broader access to potential cellular targets. We found that both endogenous and ectopically-expressed SRMS localizes to punctate cytoplasmic structures (Fig. 2A,B), which was demonstrated even with fractionation studies (Fig. 2C). However, because the localization of wild-type SRMS is characteristically punctate as opposed to diffused, we suggest that such localization is tightly regulated and that the protein anchors to its location via interactions with specific endogenous protein targets. Our results show that the localization of SRMS may be dictated by interactions involving the extended N-terminus or the SH2 domain (Fig. 4). The deletion of any of these regions at least partially altered the cytoplasmic localization of SRMS from punctate to diffuse. It is possible that the N-terminal region and the SH2 domain mediate intermolecular interactions with other target proteins that sequesters SRMS to distinct punctuate structures within the cell. However, how these critical regions mediate such associations with other proteins is as yet unknown. Intermolecular interactions involving the SH2 and SH3 domains of Src family kinases are also known to regulate the subcellular localization of these kinases [22, 48]. Interestingly, the localization of the inactivating K258M SRMS mutant was cytoplasmic and diffused in all cells. It is not obvious why mutating the ATP-binding site would alter the intracellular localization of SRMS. This observation therefore requires further investigation. The cellular localization of BRK, for example, may not be as tightly regulated. The protein has been shown to associate with its nuclear substrate, Sam68, in certain breast carcinomas besides the normal human prostate epithelial cells and well-differentiated prostate carcinomas, although it localizes to the cytoplasm in poorly differentiated prostate tumours [8]. In addition, strategically targeting BRK specifically to the membrane has been shown to enhance its oncogenic abilities, a feature that functionally contrasts with its nuclear-targeted counterpart [49]. Nonetheless, it remains to be determined how the localization of SRMS may affect its function.

The catalytic activity of Src family kinases is regulated by intramolecular interactions. Studies on BRK have also shown that, similar to Src kinases, BRK is regulated negatively by phosphorylation of C-terminal tyrosine 447 (which is analogous to the regulatory Y530 of human Src) and positively by phosphorylation of tyrosine 342 in the catalytic domain (as with Y419 of human Src) [21, 50]. C-Src tyrosine kinase (Csk) regulates Src family kinase activity by phosphorylating their C-terminal tyrosines, thereby promoting the intramolecular interactions that inactivate these enzymes [51]. Mutating the C-terminal tyrosine to phenylalanine in BRK (BRK-Y447F) and Src family kinases results in the constitutive activation of the enzymes. SRMS lacks a C-terminal regulatory tail and our data are the first to show that full-length SRMS is active, as revealed by autophosphorylation and phosphotyrosine content as a measure of in vivo protein activity (Fig. 3B). However, the full-length SRMS induced a relatively lower level of phosphorylation of cellular targets compared to BRK-Y447F (Fig. 3B). It was observed that SRMS is sequestered in specific cytoplasmic compartments within the cell and has lower access to cellular substrates compared to BRK, which may freely shuttle between the nucleus and cytoplasm. We generated a series of SRMS mutants aiming to understand the autoregulatory mechanism of the kinase. Deletion of the entire 51 amino acid spanning N-terminus region completely abrogated catalytic activity. BRK has a much shorter N-terminus and, unlike SRMS, the N-terminal amino acid sequence preceding the SH3 domain in BRK is known to be dispensable for the regulation of kinase activity [22]. In addition, and in contrast to SRMS, the deletion of the 81 amino acid N-terminus region of c-Abelson murine leukaemia viral oncogene homologue 1 relieved the enzyme of its autoinhibitory intramolecular interactions, thereby activating the enzyme [52, 53]. Interestingly, deletion of the SH2 domain of SRMS also abolished kinase activity. By contrast, Qiu and Miller [21] found that the Y447F and ΔSH2 variants of BRK induced comparable and higher levels of substrate phosphorylation, which implied that wild-type BRK is potentially maintained in an inactive (autoinhibited) conformation by the putative SH2-pY447 interaction. Although the BRK-SH2 and SH3 domains play an inhibitory role, our data show that the SH2 domain in SRMS is essential for the enzyme to sustain its catalytic activity in a yet unidentified mechanism. Thus, the autoregulatory mechanism of SRMS is distinct from that of BRK and Src kinases, where the modes of catalytic autoregulation, unlike the N-terminus, depend on the inhibitory C-terminal tail possessing a regulatory tyrosine residue [8, 17].

W260 and W184 situated within the SH2-kinase linker in Src family kinases and BRK respectively, have been shown to play opposing roles in maintaining the interactions of the SH2 and SH3 domains with their respective intramolecular ligands [22, 25]. The crystal structure of inactive Hck revealed that W260 interacts with its kinase domain, maintaining the kinase in an inactive conformation [54]. Thus, the W260A mutation in Hck activated the enzyme [25]. However, the analogous W184A mutation in BRK was found to abrogate the kinase activity of the enzyme [22]. The effect of W184A is reminiscent of the analogous W342A mutation in c-rapidly accelerated fibrosarcoma, which also resulted in a loss of kinase activity [55]. We found that the analogous W223A mutation in SRMS also caused a loss of kinase activity (Fig. 3D). Thus, similar to BRK, W223 in SRMS is predicted to interact with the kinase domain and promote the active conformation of the enzyme.

Finally, we also show that Dok1 is a binding partner and substrate of SRMS (Figs 5, 6 and 7). To our knowledge, Dok1 expression has not been characterized in breast tumour samples previously, even as the tumour suppressive adaptor protein has gained vital interest over the years and been studied in several other types of cancer [34, 56]. We have identified a lucid differential pattern of Dok1 expression in the eight breast cancer cell lines, amongst which, akin to SRMS, an elevated expression was witnessed in the HBL-100 cell line. Notably though, in the 184B5, AU565, MDA-MB-468 and SK-BR3 cell lines, Dok1 was either too low or absent. This, we conjecture, appears to correlate with the expression pattern of SRMS, albeit in a small cohort.

Dok1 is functionally characterized by a pleckstrin homology domain that allows anchorage to the membrane, a phosphotyrosine-binding domain that is involved in protein–protein interaction and a C-terminal region rich in tyrosine, proline and serine residues [57]. Even though the candidate tumour suppressor protein has been shown to inhibit cell proliferation and transformation, it also promotes cell spreading and cell migration. Dok1 is a substrate of a variety of receptor and nonreceptor tyrosine kinases, including the Src tyrosine kinase family members Tec and breakpoint cluster region-Abelson murine leukaemia viral oncogene homologue 1. Such tyrosine phosphorylation of Dok1 has been implicated in the regulation of its cellular functions [12, 14, 35, 58]. A more recent study characterized the nuclear export signal embedded within the C-terminal segment of Dok1 and demonstrated that phosphorylation via Src prevents the entry of Dok1 into the nucleus, thereby regulating its subcellular localization and functions [15]. Takeda et al. [11] identified Dok1 as a substrate of several tyrosine kinases, including SRMS [11]. Our data confirm and validate Dok1 as the first bona fide SRMS substrate. Src phosphorylates Dok1 on Y296, Y362 and Y449 [15]. Our studies show that SRMS indiscriminately phosphorylates the C-terminal tyrosines on Dok1 (Fig. 7D). Although we observed that Dok1 is predominantly nuclear in the HBL100 breast cancer cell line, it is possible that, similar to Src, phosphorylation of Dok1 by SRMS may also promote the cytoplasmic sequestration of Dok1. This hypothesis is currently being investigated. Accordingly, we also investigated whether there exists a potential interaction between endogenous SRMS and Dok1 in the MDA-MB-231 and HBL-100 cell lines via immunoprecipitation analysis, although we could not detect any association between the proteins (data not shown). However, it is possible that endogenous Dok1 may not be constitutively associated with endogenous SRMS and dissociates with the latter upon phosphorylation.

In summary, we have provided the first characterization of the catalytic activity of the novel nonreceptor tyrosine kinase SRMS. In addition to the SH2 domain, we found an unexpected critical and indispensable role for the first 51 residues of the SRMS protein in maintaining the activated state of the protein. We have also characterized Dok1 as the first and bona fide substrate of SRMS. This work is significant because it paves the way for further research aiming to identify additional factors involved in the regulation of SRMS expression and activity, and also aiming to determine how SRMS regulates the cellular role of Dok1. BRK and FRK display opposing expression patterns and functions in breast cancer. It will be interesting to correlate the expression of SRMS in a larger cohort of breast cancer samples. In the long run, it will be important to determine the functional, diagnostic, prognostic and therapeutic significance of SRMS expression in breast cancer.

Materials and methods

SRMS and Dok1 expression vectors and mutagenesis

pANT7-cGST-SRMS plasmid, purchased from DNASU Plasmid Repository (Tempe, AZ, USA), was used as a template to PCR-amplify full-length SRMS. The full-length SRMS cDNA was amplified from the above-mentioned construct via PCR and cloned C-terminal to the GFP sequence into the HindIII and BamHI sites of the pEGFP-C1 plasmid using the oligonucleotide primers: 5′-CTC AAG CTT CGG AGC CGT TCC TCA GGA GGC G-3′ and 5′-CCG GGA TCC TCA GGG GTG GCA TCT GGT GGA T-3′. The same primers were also used to clone the SRMS cDNA C-terminal to the mcherry sequence, into the same sites of the pmcherry-C1 plasmid (a kind gift from Scot Stone, Department of Biochemistry, University of Saskatchewan, Saskatoon, Saskatchewan, Canada). All other variants of SRMS were generated using the GFP-SRMS construct as the template employing single or double pairs of degenerate primers: 5′-CTC AAG CTT CGG AGC CGT TCC TCA GGA GGC G3'/5′-CTC AAG CTT CGC CCT TCC CTC AGC TCT TCC TT-3′ and 5′-CCG GGA TCC TCA GGG GTG GCA TCT GGT GGA T-3′ (GFP-ΔN-SRMS); 5′-AAA CTC GAG GCC ATG GTC CGG CTC GCC-3′ and 5′-CCG GGA TCC TCA GGG GTG GCA TCT GGT GGA T-3′/5′-CCG GGA TCC TCA GGG GTG GCA TCT GGT GGA T-3′ (GFP-PΔN-SRMS, degenerate primers); 5′-CTC AAG CTT CGG AGC CGT TCC TCA GGA GGC G-3′/5′-AAA CTC GAG TTG GTC TGA GAG CGT CTC-3′ and 5′-AAA CTC GAG CCC CAG AAG GCC CCG AGG-3′/5′-CCG GGA TCC TCA GGG GTG GCA TCT GGT GGA T-3′ (GFP-ΔSH2-SRMS, degenerate primers); 5′-CTC AAG CTT CGG AGC CGT TCC TCA GGA GGC G-3′/5′-AAA CTC GAG GCA AGG CTC GGC GGG GAG-3′ and 5′-AAA CTC GAG ACG CTC TCA GAC CAA CCC-3′ and 5′-CCG GGA TCC TCA GGG GTG GCA TCT GGT GGA T-3′ (GFP-ΔSH3-SRMS, degenerate primers). GST-SRMS constructs were generated by cloning the SRMS-SH3 and SH2 cDNAs into the BamHI and XhoI sites located within the multiple cloning site, C-terminal to the GST sequence, in the pGEX-6-p-3 vector backbone. The pairs of primers used included: 5′-CTG GGA TCC GAG ACG CTC TCA GAC CAA CCC-3′ and 5′-CGG CTC GAG TCA CCT CGG GGC CTT CTG GGG CAT-3′ (GST-SRMS-SH2); and 5′-CTG GGA TCC CCC GCC GAG CCT TGC AGC CCC-3′ and 5′-CGG CTC GAG TCA GTC TGA GAG CGT CTC AGG-3′ (GST-SRMS-SH3).

The GFP-Dok1 construct was a kind gift from Bakary S. Scylla (Lyon, France). Five Dok1 deletion mutants were generated using the GFP-Dok1 construct as the template. Five pairs of primers were used to amplify five Dok1 cDNA variants differing progressively in length and cloned C-terminal to the GFP sequence in the EcoRI and SmaI sites of the pEGFP-C1 vector backbone: DokΔ1: 5′-AGT GAA TTC GGA CGG AGC AGT GAT GGA A-3′ and 3′-ATT CCC GGG TCA AGT CTC AAC TGC CTG-5′; DokΔ2: 5′-AGT GAA TTC GGA CGG AGC AGT GAT GGA A-3′ and 3′-ATT CCC GGG TCA CTT CCG TTG TAC TCC-5′; DokΔ3: 5′-AGT GAA TTC GGA CGG AGC AGT GAT GGA A-3′ and 3′-ATT CCC GGG TCA CTT GGC CTT CAG CAA-5′; DokΔ4: 5′-AGT GAA TTC GGA CGG AGC AGT GAT GGA A-3′ and 3′-ATT CCC GGG TCA CTT CAC CCG AGC TTG-5′; DokΔ5: 5′-AGT GAA TTC GGA CGG AGC AGT GAT GGA A-3′ and 3′-ATT CCC GGG TCA CTT GGG AGC AAG GAG-5′.

The GFP-Dok1 template was also utilized, via PCR, to clone Dok1 C-terminal to the GST cDNA sequence into the EcoRI and NotI sites of the pGEX-5-x-3 vector, using the oligonucleotide primers: 5′-ATA GAA TTC CGA CGG AGC AGT GAT GGA A-3′ and 5′-ATA GCG GCC GCT CAG GTA GAG CC-3′. For the Dok1 deletion mutants, the amplified cDNA products were cloned into the EcoRI and SmaI sites of the pEGFP-C1 vector.

All point mutations of SRMS were generated using the QuikChange site-directed mutagenesis kit (Stratagene, Santa Clara, CA, USA) in accordance with the manufacturer's instructions.

Cell culture

HEK 293, HeLa, BT20, MCF7, MDA-MB-231, MDA-MB-435, MDA-MB-468, AU565, SKBR3 and HBL100 were purchased from the American tissue type culture collection (ATCC, Manassas, VA, USA). All cell lines were maintained in high glucose (4.5g·l−1) DMEM supplemented with 10% fetal bovine serum (Thermo Scientific, Logan, UT, USA), 4 mm l-glutamine, 100 units·ml−1 penicillin and 100 μg·ml−1 streptomycin (Sigma-Aldrich, St Louis, MO, USA).

Mammalian cell expression and immunoprecipitation

All transfections were carried out in the HEK 293 cell line, cultured as described above. Cells, cultured in six-well plates, were transiently transfected with a total of 2.5 μg of DNA using 1% polyethyleneimine ‘Max’ (PEI) (Polysciences Inc., Warrington, PA, USA) at a DNA to transfection reagent ratio of 1 : 3. For each well in the six-well plate, 2.5 μg of the appropriate DNA was mixed with 107.5 μl of 0.15 m sterile NaCl via gentle vortexing for 10 s. In total, 15 μl of the transfection reagent, PEI, was then added to this mixture followed by another 10 s of gentle vortexing. DNA–PEI complex formation was allowed to take place by incubating the mixture at room temperature for 10 min followed by dispensing it dropwise into the wells. The cells were incubated for 24 h post transfection and harvested the next day.

Whole cell lysates were directly prepared in 2 x Laemmli buffer (Sigma-Aldrich). For immunoprecipitation, cells washed in cold 1 × NaCl/Pi, were lysed in freshly prepared lysis buffer constituting 20 mm Tris (pH 7.5), 1% Triton, 150 mm NaCl, protease inhibitors: aprotinin 5 mg·l−1 and 0.1 mm phenylmethanesulfonyl fluoride, as well as 0.3 mm sodium orthovanadate (Enzo Life Sciences, Farmingdale, NY, USA). Lysates were prepared by incubating the harvested cells in ice-cold lysis buffer for 30 min followed by centrifugation for 10 min at 13 000 g. Supernatants were collected and transferred into fresh tubes and incubated with 1 μg of the appropriate antibody and maintained on a gyrorotator for 1 h at 4 °C. In total, 20 μl of protein A beads were then added to the samples and incubated for another 40 min on the gyrotator at 4 °C. The beads were washed twice with ice-cold lysis buffer and 1 × NaCl/Pi and the immunoprecipitated proteins were resolved via SDS/PAGE.

Immunoblotting

The proteins, either from whole cell lysates or derived from immunoprecipitations, were resolved via SDS/PAGE on 10% polyacrylamide gels. The resolved proteins were then transferred onto nitrocellulose membranes (Bio-Rad, Hercules, CA, USA) and immunoblotted with the appropriate antibodies via incubation overnight at 4 °C. Polyclonal goat horseradish peroxidase-conjugated secondary antibodies (Bio-Rad) against rabbit or mouse were used at a working dilution of 1 : 10 000 for membrane incubation at 4 °C. Enhanced chemiluminiscence (Perkin Elmer, Groningen, Netherlands) was finally used to detect the immunoreactive proteins on the membranes.

Subcellular fractionation

Subcellular fractionation was carried out using the ProteoExtract Subcellular Proteosome Extraction Kit (Calbiochem, San Diego, CA, USA) in accordance with the manufacturer's instructions. Proteins from total cell lysates were fractionated in accordance with the manufacturer's instructions to obtain three different cellular fractions (cytosolic, membrane and nucleus), which were eventually resolved via SDS/PAGE and detected using the appropriate antibodies.

Primary antibodies

Primary antibodies were purchased from Santa-Cruz Biotechnology Inc. (Santa Cruz, CA, USA). These included: anti-SRMS (Sc-68341), anti-β-tubulin (Sc-9104), anti-GFP (Sc-8334) and anti-phosphotyrosine (Sc-508). Anti-SRMS (T2686) was obtained from Epitomics (Burlingame, CA, USA). Anti-Dok1 was a kind gift from Ryuji Kobayashi (University of Texas, Austin, TX, USA). The anti-Sam68 (AD1) polyclonal antibody has been described previously [59].

Immunofluorescence microscopy

Cells, seeded on coverslips, were cultured in six-well plates, fixed with 1% paraformaldehyde in 1 x NaCl/Pi (pH 7.4) for 5 min and permeabilized with 0.5% Triton X-100 in NaCl/Pi for 5 min at room temperature, and incubated with anti-SRMS or anti-Dok1 sera (dilution 1 : 200) for 1 h in NaCl/Pi at room temperature. The cells were washed with 0.1% Triton X-100 in NaCl/Pi and incubated with the appropriate secondary antibodies (dilution 1 : 200) in NaCl/Pi for 30 min. Goat anti-mouse coupled to Texas Red (Santa Cruz Biotechnology Inc.) and goat anti-rabbit coupled to fluorescein isothiocynate (Santa Cruz Biotechnology Inc.) were used as secondary antibodies. The coverslips were then mounted onto glass slides with glycerol containing 3 mg·ml−1 DAPI to stain the nuclei. The cells were observed under an Olympus 1X-71 inverted microscope (Nikon, Tokyo, Japan) and the images were captured.

Recombinant GST-fused protein expression and GST-pull-down assay

GST pull-down assays were performed as described previously [60]. With the exception of GST-SRMS, which was procured from SignalChem (Richmond, BC, Canada), all other GST-tagged constructs were expressed in Escherichia coli (BL21 strain), cultured in 2 x YT media. Protein induction was initiated with the addition of 1 mm isopropyl thio-β-d-galactoside to the bacterial cultures at D0.6. Bacterial cells were then lysed, via sonication using short intermittent pulses, in ice-cold 1 x NaCl/Pi buffer containing protease inhibitors, 1 μg·ml−1 aprotinin and 0.01% phenylmethanesulfonyl fluoride (Calbiochem, Gibbstown, NJ, USA), supplemented with a protease inhibitor cocktail comprising 23 mm 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, 2 mm bestatin, 100 mm EDTA, 0.3 mm trans-epoxysuccinyl-l-leucylamido-(4-guanidino)butane and 0.3 mm pepstatin A, in dimethylsulfoxide (P8465; Sigma-Aldrich). Lysates were then incubated with the glutathione sepharose beads (Novagen, San Diego, CA, USA).

In brief, the pull-down experiments were carried out using GST, GST-SRMS-SH3 and GST-SRMS-SH2 proteins immobilized on glutathione sepharose beads, which were incubated with cell lysates and the bound proteins were resolved via SDS/PAGE as described above.

In vitro kinase assay

In vitro kinase assays were performed using 50 ng of GST-SRMS (SignalChem) and a 10-μl bed volume of substrate (GST-Dok1) in a reaction volume of 50 μl, comprising 20 μl of kinase buffer (25 mm Mops, pH 7.2, 2.5 mm dithiothreitol, 12.5 mm and 5 mm EGTA) and 20 μl of H2O with or without 200 μm ATP. The reaction was allowed to proceed by incubating the mixture at 30 °C for 30 min and was ultimately terminated by the addition of 2 x Laemmli buffer. The samples were then boiled at 100 °C and resolved via SDS/PAGE (as described above).

IHC

IHC staining for SRMS was performed using tissue microarray BR243d in accordance with the manufacturer's instructions (US Biomax Inc., Rockville, MD, USA) on 4-μm paraffin-embedded sections. BR243d contained six cases of breast invasive ductal carcinoma and self-matched cancer adjacent normal breast tissue in quadruple cores per case format. Briefly, after deparaffinization and hydration, the slides were treated with 3% H2O2 for 5 min to block endogenous peroxidase, and washed twice in water for 5 min each. Antigen retrieval was performed by using 1 x antigen retrieval solution and incubated for 30 min in a microwave oven with simmering, and allowed to cool down at room temperature for 15 min. The sections were then washed for 3 x 5 min in NaCl/Pi-Tween and the sections were blocked for 30 min at room temperature with 2.5% blocking serum (Santa Cruz Biotechnology Inc.) in NaCl/Pi before reacting with anti-SRMS (Epitomics) at a dilution of 1 : 500. Sections were then incubated with primary antibodies for 1 h at room temperature in a moist chamber. After incubation, the sections were washed three times in NaCl/Pi and treated with ImmPRESS™ reagent (Vector Laboratories, Burlingame, CA, USA). The sections were then washed for 3 x 5 min in NaCl/Pi followed by colour development in peroxidase substrate 3,3′-diaminobenzidine tetrahydrochloride (DAB) (DAKO Cytomation, Brea, CA, USA). Finally, the slides were rinsed in tap water and lightly counterstained with haematoxylin QS (Vector Laboratories), cleared and mounted with permanent mounting medium (C0487; Sigma-Aldrich). Duplicate sections were immunostained without exposure to primary antibodies and served as a negative control. To quantitate SRMS protein expression, a scoring method was used in which the mean percentage of positive tumour cells was determined in at least five random fields at × 20 magnification in each section. The intensity of the SRMS immunoreaction was scored as: 1+, weak; 2+, moderate; and 3+, intense. The percentage of positive tumour cells and the staining intensity were multiplied to produce a SRMS-IHC staining score for each case. The DAB substrate-chromogen yielded a dark brown reaction end-product at the site of the target antigen. Haematoxylin was used for counterstaining cell nuclei, which yields a blue colour stain. To control for antibody specificity, the samples were incubated with secondary antibody alone and analyzed as described above. The data for the control experiment are presented in Fig. S2.

Acknowledgements

We express our indebted gratitude to Dr Ryuji Kobayashi (MD Anderson Cancer Centre, University of Texas, Austin, TX, USA) for providing us with anti-Dok1 serum and Dr Bakary S. Scylla (Lyon France) for providing us with the GFP-Dok1 plasmid. We would also like to thank Lexie Martin for her critical review of the manuscript. We also sincerely thank Dr Scott Stone (Department of Biochemistry, University of Saskatchewan, Saskatoon, Saskatchewan, Canada) for providing us with the mcherry-C1 plasmid. I especially thank Dr Gerald (Jerry) Davies for all of his help. In a breath of personal acknowledgement, I deem it invaluable to thank my mentor Dr Kiven Erique Lukong. I also laud the priceless support and faith bestowed upon me by my late father as my eternal mentor, as well as my mother and elder brother. R.K.G. contributed to the experimental design, set-up and interpretation of the results, amounting to 90% of the study's yield. The present study was also supported in part (10%) by S.M., K.B., N.K. and C.D. The manuscript was written and edited by R.K.G. and K.E.L. This work was funded by start-up funds awarded to K.E.L. by the Department of Biochemistry at the University of Saskatchewan. R.K.G. is a recipient of a meritorious scholarship from the Department of Biochemistry at the University of Saskatchewan.

Ancillary