The Nuclear Tyrosine Kinase BRK/Sik Phosphorylates and Inhibits the RNA-binding Activities of the Sam68-like Mammalian Proteins SLM-1 and SLM-2*

Expression of the intracellular tyrosine kinase BRK/ Sik is epithelial-specific and regulated during differentiation. Only a few substrates have been identified for BRK/Sik, including the KH domain containing RNA-binding protein Sam68 and the novel adaptor protein BKS. Although the physiological role of Sam68 is un-known, it has been shown to regulate mRNA transport, pre-mRNA splicing, and polyadenylation. Here we demonstrate that the Sam68-like mammalian proteins SLM-1 and SLM-2 but not the related KH domain containing heterogeneous nuclear ribonucleoprotein K are novel substrates of BRK/Sik. The expression of active BRK/Sik results in increased SLM-1 and SLM-2 phosphorylation and increased retention of BRK/Sik within the nucleus. The phosphorylation of SLM-1 and SLM-2 has functional relevance and leads to inhibition of their RNA-binding abilities. We show that SLM-1, SLM-2, and BRK/Sik have restricted patterns of expression unlike the ubiquitously expressed Sam68. Moreover, BRK/Sik, SLM-1, and RNA-binding Assays— Myc-tagged Sam68, SLM-1, or SLM-2 and pcDNA3 or BRK/Sik YF were introduced into NMuMG cells, and cell lysates were prepared. These cell lysates were used for homopolymeric RNA-binding assays using poly(A), poly(G), and poly(U) (Sigma) co-valently coupled to beads in lysis buffer supplemented with 2 mg/ml heparin as described incubat-ing lysates with Sepharose 4B beads (Sigma) alone. Beads washed twice with lysis buffer and once with phosphate-buffered saline, and proteins sample buffer. analyzed sues HindIII Sam68 cDNA, HindIII/XhoI SLM-1 ApaI/HindIII cDNA pBluescript P-labeled

The intracellular tyrosine kinase BRK/Sik (also named PTK6) was identified in a screen for protein tyrosine kinases involved in breast cancer (1), from the mouse small intestine in a screen for factors that regulate epithelial cell differentiation (2), and from cultured human melanocytes (3). Highest levels are expressed in differentiating epithelial linings of the gastro-intestinal tract and skin (4,5) and in prostate epithelial cells (6). Activation of BRK/Sik has been correlated with the differentiation of cultured keratinocytes (7). Although BRK/Sik is expressed in many breast carcinoma cell lines and primary breast tumors, it has not been detected in normal human breast tissue (1,8) or at any stage of mammary gland differentiation in the mouse (5). Modest increases in BRK/Sik levels have been detected in colon tumors relative to normal colonic tissue (5). BRK/Sik expression has also been detected in normal human oral epithelium and oral squamous cell carcinomas (9).
Although related to Src, BRK/Sik belongs to a distinct family of intracellular tyrosine kinases (10). It lacks an amino-terminal myristoylation signal, and it is not specifically targeted to the membrane. Like the Src family kinases, mutation of the carboxyl-terminal tyrosine of BRK/Sik (Y447) results in increased enzyme activity, supporting a role for this residue in autoinhibition (11)(12)(13). However Csk, the kinase that phosphorylates the carboxyl-terminal tyrosine in Src kinases, does not appear to be the enzyme that regulates carboxyl-terminal phosphorylation of BRK/Sik (12). Mutation of the BRK/Sik regulatory tyrosine did not enhance its ability to induce anchorageindependent growth of NIH 3T3 cells (14).
Thus far two substrates have been identified for BRK/Sik, the novel adaptor-like protein BKS (15) and the nuclear RNAbinding protein Sam68 (11). Sam68 is a member of the STAR (signal transducers and activators of RNA) family of KH (heteronuclear ribonucleoprotein K homology) domain containing RNA binding proteins that regulate different aspects of RNA metabolism, including transport, stability, translation, and processing (16 -18). The Sam68 KH domain is required for RNA binding, and it is embedded in a larger conserved domain called the GSG (GRP33, Sam68, GLD1) domain that is found in several RNA-binding proteins that regulate developmental processes (19,20).
Several studies support roles for Sam68 in the regulation of RNA metabolism. It may act as a functional homologue of the human immunodeficiency virus (HIV) 1 type 1 Rev protein, which transports RNA from the nucleus to the cytoplasm (21)(22)(23)(24). Sam68 may also regulate polyadenylation and has been found to enhance the 3Ј-end processing of HIV-1 RNA (25). In addition, Sam68 colocalizes with and associates with RNA splicing factors (26,27) and was shown to be a regulator of alternative splicing (28). Because of its interactions with a number of signaling proteins, including Src, Fyn, Grb2, and phospholipase C-␥ it has been proposed that Sam68 may also act as an adaptor protein (29).
We showed previously that BRK/Sik phosphorylation of Sam68 inhibits its ability to bind RNA and function as a cellular HIV-1 Rev homologue (11). Although Sam68 can be phosphorylated by other intracellular tyrosine kinases, only BRK/ Sik has been shown to colocalize with Sam68 in the nucleus (11). BRK/Sik also regulates the ability of Sam68 to regulate utilization of specific RNAs in the cytoplasm (30).
Two Sam68-like mammalian proteins, SLM-1 and SLM-2, were identified based on their high degree of sequence similarity with Sam68 within the GSG domains (31,32). SLM-1 shares many similarities with Sam68; it interacts with many of the same proteins and is also tyrosine-phosphorylated by Src during mitosis (31). SLM-2 was also identified by its ability to interact with RNA-binding motif in spermatogenesis and was also named T-STAR or ETOILE (33). Because of the similarities that SLM-1 and SLM-2 share with Sam68, we asked if SLM-1 and SLM-2 are BRK/Sik substrates. We then examined the possible biological significance of BRK/Sik tyrosine phosphorylation of SLM-1 and SLM-2 by examining their RNAbinding activities after tyrosine phosphorylation and their coexpression with BRK/Sik in mouse tissues.
Cell Lines and Protein Expression-Cell lines were obtained from the American Type Culture Collection. NMuMG cells were transfected using the Lipofectamine reagent (Invitrogen Corp.). HeLa cells were maintained in Dulbecco's modified Eagle's medium with 1.0 mM sodium pyruvate and 10% bovine calf serum and transfected with the vaccinia virus T7 expression system (29).
Immunoprecipitations and in Vitro Kinase Assays-pBS-KS Myc-Sam68, SLM-1, or SLM-2 was expressed in HeLa cells using the vaccinia-T7 expression system as described previously (29). Transfected cells were lysed and immunoprecipitations were performed with anti-Myc antibodies (31). Immunoprecipitated proteins were incubated with GST-BRK/Sik and [␥-32 P]ATP for 15 min at 30°C. Proteins were separated by SDS-PAGE and exposed to x-ray film.
RNA-binding Assays-Myc-tagged Sam68, SLM-1, or SLM-2 and pcDNA3 or BRK/Sik YF were introduced into NMuMG cells, and cell lysates were prepared. These cell lysates were used for homopolymeric RNA-binding assays using poly(A), poly(G), and poly(U) (Sigma) covalently coupled to beads in lysis buffer supplemented with 2 mg/ml heparin as described (31,36,37). Controls were performed by incubating lysates with Sepharose 4B beads (Sigma) alone. Beads were washed twice with lysis buffer and once with phosphate-buffered saline, and proteins were eluted in Laemmli sample buffer. Samples were analyzed on 9% SDS-PAGE.
RNase Protection Assays-Total RNA was isolated from various tissues of CD1 mice using TRIzol reagent (Invitrogen). A 135-bp BglII/ HindIII fragment of the Sam68 cDNA, a 300-bp HindIII/XhoI fragment of the SLM-1 cDNA, and a 230-bp ApaI/HindIII fragment of SLM-2 cDNA were subcloned into the vector pBluescript KS, and 32 P-labeled antisense cRNA probes were generated. RNase protection assays were performed as described previously (2,38).

BRK/Sik Phosphorylates SLM-1 and SLM-2 in Vitro and in
Vivo-To determine whether SLM-1 and/or SLM-2 are substrates of BRK/Sik, the normal murine mammary gland cell line NMuMG was transiently cotransfected with either wildtype BRK/Sik WT or activated BRK/Sik YF, and GFP-tagged GFP-SLM-1 and GFP-SLM-2. The NMuMG cell line does not express endogenous BRK/Sik and is an excellent epithelial cell line model for evaluating BRK/Sik activities (5). Tyrosinephosphorylated GFP-SLM-1 and GFP-SLM-2 were detected by immunoblotting with anti-phosphotyrosine in total cell lysates from NMuMG cells cotransfected with either wild-type BRK/ Sik or activated BRK/Sik YF but not in cells cotransfected with vector alone (Fig. 1). BRK/Sik YF, which contains a substitution of the carboxyl-terminal negative regulatory tyrosine, was most effective in phosphorylating the SLM proteins because this mutation results in increased BRK/Sik activation (11). Immunoblotting with BRK/Sik, ␤-actin, and GFP antibodies confirmed the equivalent expression of the transfected proteins (Fig. 1). These data demonstrate that SLM-1 and SLM-2 are substrates for BRK/Sik. BRK/Sik phosphorylation of SLM-1 and SLM-2 was also demonstrated in vitro using recombinant GST-BRK/Sik. Myctagged Sam68 (Myc-Sam68), Myc-SLM-1, and Myc-SLM-2 were immunoprecipitated from HeLa cells and incubated with GST-BRK/Sik and [␥-32 P]ATP. The GST-BRK/Sik phosphorylated Myc-tagged Sam68, SLM-1, and SLM-2 as visualized in duplicate Myc immunoprecipitations ( Fig. 2A), providing further evidence that Sam68, SLM-1, and SLM-2 are substrates of BRK/Sik.
Tyrosine Phosphorylation of Nuclear SLM Proteins and Enhanced Nuclear Retention of BRK/Sik-The consequences of tyrosine phosphorylation on BRK/Sik and SLM protein localization were examined using confocal microscopy. NMuMG cells were cotransfected with BRK/Sik expression vectors and GFP-SLM-1 and GFP-SLM-2. GFP-tagged SLM-1 and SLM-2 were preferentially localized to nuclei of transfected cells (Figs.  3 and 4). Although GFP-SLM-1 expression was localized to the nuclei of transfected cells (Fig. 3A, panels A, C, E, G, I, and K), wild-type BRK/Sik protein, visualized by rhodamine-avidin (red), was distributed throughout the cells (Fig. 3, panels F and G). Colocalization of GFP-SLM-1 and wild-type BRK/Sik in the nucleus is shown in Fig. 3G (yellow). In contrast to the wildtype protein, constitutively active BRK/Sik YF was present primarily in nuclei of transfected cells where it colocalized with GFP-SLM-1 (Fig. 3A, panels J and K). A rare instance in which both expression constructs were not taken up by the same cell is shown in Fig. 3B. BRK/Sik YF is localized throughout the cell in the cell lacking expression of GFP-SLM-1, but it is nuclear in the cell expressing GFP-SLM-1 (Fig. 3B, panels B and C). These data suggest that activated BRK/Sik YF is more efficiently retained in the nuclei of cells expressing high levels of substrate. Similar patterns of colocalization of GFP-SLM-2 with wild-type BRK/Sik and BRK/Sik YF were detected (data not shown).
Following transfection of wild-type BRK/Sik and BRK/Sik YF differences in the intracellular localization of phosphotyrosine specific immunoreactivity were also observed. Differences in the localization of phosphotyrosine in cells transfected with wild-type BRK/Sik and BRK/Sik YF and GFP-tagged SLM-2 are shown in Fig. 4. Expression of wild-type BRK/Sik resulted in the tyrosine phosphorylation of cytoplasmic and nuclear proteins (Fig. 4, B and C), whereas expression of activated BRK/Sik YF led to enhanced phosphotyrosine immunoreactivity within the nucleus (Fig. 4, F and G). Phosphotyrosine activity was dependent on BRK/Sik activity and not detected in cells expressing kinase-defective BRK/Sik KM with a mutation in its ATP binding site (Fig. 4, J). Most of the BRK/Sik tyrosine phosphorylated proteins colocalized with SLM-2 (Fig. 4, C and  G). Cotransfection of the GFP expression vector pEGFP-C1 and the empty BRK/Sik expression vector pcDNA3 resulted in diffuse GFP fluorescence throughout the cell and no detectable anti-phosphotyrosine staining (Fig. 4, M and N). In additional control experiments with IgG, no specific fluorescent signal was detected (data not shown). Increased nuclear localization of activated BRK/Sik YF (Fig. 3) and increased phosphotyrosine immunoreactivity in cells expressing BRK/Sik YF suggest that the carboxyl-terminal tyrosine of BRK/Sik functions to regulate both activity and subsequent localization of the kinase.
The intracellular localization of wild-type and activated BRK/Sik YF was also examined following fractionation of transfected NMuMG cells (Fig. 5A). Higher levels of BRK/Sik YF were present in the nuclear fractions than wild-type BRK/ Sik. Immunoblotting with antibodies against Sp1 and ␣-tubulin served as a control to demonstrate the enrichment of nuclear and cytoplasmic/membrane proteins in the respective fractions. Confocal microscopy also indicated that higher levels of BRK/Sik YF were present in nuclei of transfected NMuMG Immunoblotting was performed with antibodies against BRK/Sik, Myc, and phosphotyrosine. BRK/Sik YF expression led to tyrosine phosphorylation of Myc-Sam68, Myc-SLM-1, and Myc-SLM-2 but not the related protein hnRNP K or Sam68 (68 -347) lacking the tyrosine-rich carboxyl terminus. cells (Fig. 5B, panels C and F) than wild-type BRK/Sik (Fig. 5B,  panels B and E). These two different experimental approaches demonstrated that a greater amount of BRK/Sik YF is concentrated in nuclei than wild-type BRK/Sik even in the absence of ectopically expressed substrate. However, NMuMG cells do express Sam68, which may promote nuclear localization of BRK/Sik in these cells. Activated BRK/Sik YF appears to have greater access to its nuclear substrates, which could contribute to increased nuclear tyrosine phosphorylation and increased retention of BRK/Sik in the nucleus through SH2-domain mediated interactions with tyrosine-phosphorylated proteins.
BRK/Sik Inhibits the RNA-binding Abilities of SLM-1 and SLM-2-To determine whether BRK/Sik regulates RNA-binding functions of SLM-1 and SLM-2, we performed RNA-binding studies in the presence and absence of activated BRK/Sik. NMuMG cells were cotransfected with Myc-Sam68, Myc-SLM-1, or Myc-SLM-2 and the empty expression vector pcDNA3 or constitutively activated BRK/Sik YF. Total cell lysates were probed with anti-phosphotyrosine, anti-Myc, and anti-BRK/Sik antibodies. The Myc-tagged constructs as well as BRK/Sik YF were expressed following transfection and expression of BRK/Sik YF resulted in increased tyrosine phosphorylation of Myc-SLM1, Myc-SLM-2, and Myc-Sam68 (Fig. 6). Sam68 served as a positive control (Fig. 6C), because we have shown previously that BRK/Sik can negatively regulate Sam68 homopolymeric RNA binding (11). Cell lysates were divided equally and incubated with either poly(A), -(G), or -(U) immobilized to agarose or Sepharose alone. SLM-1 bound poly(A) and SLM-2 bound poly(G) homopolymeric RNA when cotransfections were performed with the empty expression vector. However, little or no RNA binding was detected when SLM-1 or SLM-2 were coexpressed with BRK/Sik YF (Fig. 6). These data indicate that BRK/Sik negatively regulates the RNA binding abilities of both SLM-1 and SLM-2.
Expression of BRK/Sik and STAR Proteins in Mouse Epithelial Tissues-To determine the physiological relevance of SLM-1 and SLM-2 as substrates of BRK/Sik we examined the expression   antibodies (B, F, J, and N). 4Ј,6-Diamidino-2-phenylindole (DAPI) was used to stain the nuclei (D, H, L, and P). In NMuMG cells, SLM-2 displays diffuse, nuclear localization visible by green fluorescence (A, E, and I). Cells cotransfected with GFP-SLM-2 and wild-type BRK/Sik or BRK/Sik YF also stain strongly with the anti-phosphotyrosine antibody visualized using rhodamine (B and F), whereas no phosphotyrosine is detected in cells expressing kinasedefective BRK/Sik KM (J). Interestingly, phosphotyrosine immunoreactivity is seen in the cytoplasm and at the membrane when wild-type BRK/Sik is transfected into the cells but only in the nucleus following transfection of BRK/Sik YF. C, G, K, and O are composites demonstrating colocalization that appears yellow. GFP alone is expressed throughout the cell (M) and is negative for anti-phosphotyrosine staining (N). Bar represents 5 m.

FIG. 5. Intracellular localization of wild-type and activated BRK/Sik YF in transfected NMuMG cells.
A, NMuMG cells were transfected with empty vector (V), wild-type (WT), or activated BRK/Sik YF expression constructs. Cells were fractionated into nuclear and membrane/cytoplasmic fractions, and localization of transfected BRK/ Sik protein was examined by immunoblotting. A greater proportion of activated BRK/Sik YF is found in the nucleus. B, localization of transfected BRK/Sik protein was examined using immunohistochemistry and confocal microscopy. BRK/Sik YF exhibits increased nuclear localization in NMuMG cells (panel C). Bar represents 10 m.
of BRK/Sik, SLM-1, SLM-2, and Sam68 RNAs by performing RNase protection assays using total RNA from multiple different mouse tissues of male and female mice. Because BRK/Sik plays a role in breast cancer, we examined expression of its substrates in mammary glands isolated from nulliparous virgin female mice and multiparous breeder females that had raised litters (Fig. 7). As expected, BRK/Sik was expressed at highest levels in the gastrointestinal tract and skin. No BRK/Sik expression was detected in normal mammary gland at any stage. SLM-1 and SLM-2 exhibited restricted patterns of expression, whereas Sam68 was ubiquitously expressed. Notable levels of SLM-1 were coexpressed with BRK/Sik in skin, colon, and cervix, whereas SLM-2 exhibited testes-specific expression. No expression of BRK/Sik protein has been detected in the mouse testes (data not shown). These findings indicate that SLM-1 and Sam68 may be relevant substrates of BRK/Sik in vivo.

FIG. 8. Functions of BRK/Sik in normal tissues and tumors may be dependent on its intracellular localization and access to different substrates.
In breast cancer cells, BRK/Sik appears to promote growth by enhancing epidermal growth factor receptor and ErbB3 receptor signaling (14,53). However in normal tissues, BRK/Sik expression is associated with cell cycle exit and differentiation of epithelial linings (4 -7, 9). In normal prostate epithelial cells BRK/Sik is enriched in the nucleus where it associates with Sam68, whereas BRK/Sik is relocalized to the cytoplasm in prostate tumors (6). Differential localization of BRK/Sik in normal tissues and cancer cells may lead to activation of divergent signaling pathways. RNase protection assays were performed with total RNAs from a variety of tissues from the mouse (male mice and multiparous and virgin females as indicated) and 32 P-labeled antisense probes specific for BRK/Sik, Sam68, SLM-1, and SLM-2. Cyclophilin expression was examined as a control. Sam68 is ubiquitously expressed, whereas SLM-1 and SLM-2 have a much more restricted pattern of expression. Although BRK/Sik is induced in a high percentage of breast tumors, mouse BRK/Sik and its SLM substrates are not expressed in adult female mammary gland.

DISCUSSION
Previously we showed that BRK/Sik phosphorylates the nuclear protein Sam68 and inhibits its RNA-binding functions (11). Here we demonstrated that SLM-1 and SLM-2 are substrates of BRK/Sik and that their RNA-binding abilities are inhibited by tyrosine phosphorylation. The RNA-binding functions of these STAR proteins have been implicated in the posttranscriptional regulation of gene expression. Both SLM-1 and SLM-2 have been shown to stimulate HIV-1 Rev activity (22,39) and to regulate the selection of alternative splice sites in transcripts encoded by a CD44 minigene (40,41). However, the cellular functions of these two RNA-binding proteins and their RNA targets are still poorly understood.
To begin to understand the significance of SLM-1 and SLM-2 phosphorylation in vivo, we examined the expression of these two proteins in mouse tissues. SLM-1 is coexpressed with BRK/ Sik in some epithelial tissues, including colon and skin. However our data and that of others (33,42) indicate that SLM-2 is predominantly expressed in testis, a tissue lacking significant BRK/Sik expression. Although SLM-1 and Sam68 may be phosphorylated by other kinases such as Src and Fyn (31), BRK/Sik colocalizes with SLM-1 and Sam68 in the nucleus and may regulate nuclear functions.
In previous studies, SLM-2 was not demonstrated to be a Src or Fyn substrate (31), so it is interesting that this RNA-binding protein is a BRK/Sik substrate. However lack of coexpression of SLM-2 and BRK/Sik in epithelial tissues makes it unlikely that this STAR substrate is physiologically relevant. Further studies will be required to determine whether other nonmyristoylated members of the BRK/Sik kinase family such as Srms or FRK/Rak (reviewed in Ref. 10) are coexpressed with SLM-2 in the testis and able to regulate its nuclear functions.
Although BRK/Sik phosphorylates SLM-1, SLM-2, and Sam68 in the nucleus, it did not phosphorylate the KH domaincontaining protein hnRNP K (Fig. 2). In addition BRK/Sik does not phosphorylate the ubiquitous nuclear protein YT521-B that regulates alternative splicing and associates with Sam68 (43). YT521-B is a substrate of Src and Fyn in the cytoplasm and c-Abl in the nucleus (43). In the nucleus BRK/Sik has substrate specificity and appears to regulate specific STAR family signaling pathways.
Sam68 is one of the most well characterized members of the STAR family of RNA-binding proteins (reviewed in Ref. 16), but its physiological functions are still not well understood. For example, both growth inhibiting and growth promoting functions have been reported for Sam68. A retroviral-based antisense strategy revealed that Sam68 deficiency led to transformation of murine NIH 3T3 fibroblasts (44). A recent study indicated that ectopically expressed Sam68 could compromise cell proliferation independent of its RNA-binding abilities and induce apoptosis in an RNA-binding dependent manner (45). In contrast, a variant of Sam68 that lacks a functional KH domain and appears to act as a dominant negative form of the protein inhibits cell cycle progression (46). Disruption of the Sam68 gene in the chicken DT40 cell line also resulted in reduced growth rate because of elongation of the G 2 -M phase of the cell cycle (47). Sam68 has been shown to be a direct target of Cdc2 during mitosis (48) and an extracellular signal-regulated kinase target (28) providing further evidence that it functions in growth regulatory pathways.
Tyrosine phosphorylation of nuclear proteins regulates many cellular processes, including growth, differentiation, and apoptosis (reviewed in Refs. 49 and 50). Although BRK/Sik lacks an apparent nuclear localization signal, absence of an aminoterminal myristoylation signal gives the kinase flexibility in its intracellular localization, and it is one of a handful of tyrosine kinases found in the nucleus. Like c-Abl (51), the intracellular localization of BRK/Sik may influence its protein-protein interactions and the signaling pathways that it regulates.
Protein mislocalization has been shown to play an important role in the development of cancer and has been observed for a number of signaling proteins (52). BRK/Sik is localized to nuclei of normal prostate epithelial cells where it associates with Sam68 (6). However BRK/Sik nuclear localization is lost in high grade prostatic intraepithelial neoplasia and in prostate tumors, whereas its substrate Sam68 remains nuclear (6). Although it is not expressed in normal mammary gland epithelial cells, BRK/Sik is expressed in a significant proportion of primary breast tumors and breast tumor cell lines. In breast cancer cells, BRK/Sik promotes epidermal growth factor receptor (14) and ErbB3 (53) signaling and associates with these membrane receptors. The ability of BRK/Sik to associate with distinct sets of substrates in different cellular compartments in normal tissues and in cancer cells may lead to activation of divergent signaling pathways (Fig. 8). In some epithelial tissues, BRK/Sik appears to be positioned to inhibit RNA-binding activities of its nuclear STAR protein family substrates during differentiation. Further studies will be required to determine the significance of BRK/Sik and STAR family proteins in epithelial cell signal transduction.