A Growth-suppressive Function for the c-Fes Protein-Tyrosine Kinase in Colorectal Cancer*

The human c-fes locus encodes a non-receptor protein-tyrosine kinase implicated in myeloid, vascular endothelial, and neuronal cell differentiation. A recent analysis of the tyrosine kinome in colorectal cancer identified c-fes as one of only seven genes with consistent kinase domain mutations. Although four mutations were identified (M704V, R706Q, V743M, S759F), the consequences of these mutations on Fes kinase activity were not explored. To address this issue, Fes mutants with these substitutions were co-expressed with STAT3 in human 293T cells. Surprisingly, the M704V, R706Q, and V743M mutations substantially reduced Fes autophosphorylation and STAT3 Tyr-705 phosphorylation compared with wild-type Fes, whereas S759F had little effect. These mutations had a similar impact on Fes kinase activity in a yeast expression system, suggesting that they inhibit Fes by affecting kinase domain structure. We have also demonstrated for the first time that endogenous Fes is strongly expressed at the base of colonic crypts where it co-localizes with epithelial cells positive for the progenitor cell marker Musashi-1. In contrast to normal colonic epithelium, Fes expression was reduced or absent in colon tumor sections from most individuals. Fes protein levels were also low or absent in a panel of human colorectal cancer cell lines, including HT-29 and HCT 116 cells. Introduction of Fes into these lines with a recombinant retrovirus suppressed their growth in soft agar. Together, our findings strongly implicate the c-Fes protein-tyrosine kinase as a tumor suppressor rather than a dominant oncogene in colorectal cancer.

tiation in physiological sites of expression including myeloid, vascular endothelial, and neuronal cells (3,5,6). Furthermore, a human chronic myelogenous leukemia-derived cell line (K-562) that lacks Fes expression can be induced to undergo growth arrest and terminal differentiation following the re-introduction of Fes, implicating this kinase as a potential suppressor of chronic myelogenous leukemia progression (7). More recently, a role for Fes in colorectal cancer progression was suggested when c-fes mutations were identified in colorectal tumors (8).
Colorectal cancer is the second leading cause of cancer-related deaths in the United States, with ϳ145,000 new cases diagnosed annually (9). The progression of colorectal cancer involves the successive accumulation of mutations that drive the transformation of normal colonic epithelium through a discrete series of stages that culminate in invasive adenocarcinoma (10). Many genetic changes associated with transformation have been identified and are the subject of recent reviews (10 -12). Recently, Bardelli et al. (8) performed a global screen of genomic kinase domain sequences from all known protein-tyrosine kinase, kinase-like, and receptor guanylate cyclase genes for colorectal cancerassociated mutations. Of the 138 kinase domains sequenced, c-fes was one of only seven genes with consistent mutations in the 182 tumors analyzed. However, the functional consequences of these mutations on Fes kinase activity and the broader biological role of Fes in colorectal cancer have not been explored.
In this study, the colorectal cancer-associated mutations found in the Fes kinase domain (M704V, R706Q, V743M, S759F) were assessed for their effects on Fes kinase activity. Surprisingly, all four mutations failed to activate Fes. Instead, three of the four mutations dramatically reduced Fes kinase activity in both human 293T cells and in a yeastbased expression system, whereas the fourth had little impact. Using immunohistochemistry, we observed strong Fes expression in a distinct subpopulation of crypt cells from normal human colonic epithelium. Moreover, Fes was expressed in Musashi-1-positive cells and localized near the base of colonic crypts, suggesting that Fes expression is restricted to early progenitor cells. In contrast, Fes protein was consistently reduced or absent in primary human colon tumor tissue and in colorectal cancer cell lines, including HT-29 and HCT 116. Introduction of Fes into these cells with a recombinant retrovirus suppressed their anchorage-independent growth in soft agar. Together, these studies suggest that c-Fes functions as a tumor suppressor in colonic epithelium, despite its protein-tyrosine kinase activity.

MATERIALS AND METHODS
Plasmids and Retroviruses-Coding sequences for wild-type human Fes, a kinase-active coiled-coil domain mutant (L145P), and a kinaseinactive mutant (KE) were fused to the enhanced yellow-shifted variant of green fluorescent protein (YFP) in the plasmid vector pEYFP-C1 (BD Bioscience-Clontech) as described elsewhere (4). In this report, YFP is referred to as green fluorescent protein (GFP) for simplicity. Codons for the colon cancer-associated mutations M704V, R706Q, V743M, and S759F were incorporated into wild-type or L145P Fes cDNA clones via standard PCR-based methods. Yeast expression vectors for the wildtype and L145P forms of Fes have been described elsewhere (13) . The  M704V, R706Q, V743M, and S759F single mutants as well as the  L145P,M704V, L145P,R706Q, L145P,V743M, and L145P,S759F double mutants were subcloned from the corresponding YFP-Fes fusion plasmids into the yeast expression vector pESC-Ura (Stratagene). Retroviruses for the expression of GFP alone or wild-type Fes fused to GFP have been described previously (14). To generate the active Fes-Fps chimera, the coding region for residues 610 -822 of c-Fes were replaced with the homologous region of v-Fps (15). The Fes-Fps coding region was then amplified by PCR, subcloned into pSR␣GFP (14), and used for retrovirus production as described (6,14). STAT3 was expressed in 293T cells using the mammalian expression vector pCDNA3.1(Ϫ) (Invitrogen) as described elsewhere (16).
Immunoprecipitation, Immunoblotting, and Antibodies-Transiently transfected 293T cells as well as stably transduced HT-29 and HCT 116 cells were lysed in radioimmune precipitation assay buffer, whereas all other cell lines were lysed in Fes lysis buffer as described previously (6). Immunoprecipitation and immunoblotting assays were performed as described elsewhere (4). Fes immunoprecipitation utilized a rabbit polyclonal antiserum raised against the Fes amino-terminal and SH2 domains (18). Fes was immunoblotted with goat polyclonal antiserum raised against the amino-terminal region of Fes (Fes N-19; Santa Cruz Biotechnology), and YFP-Fes was detected with a mouse monoclonal antibody raised against GFP (GFP B-2; Santa Cruz). The active phosphorylated form of Fes was detected with either a mouse monoclonal anti-phosphotyrosine antibody (PY99; Santa Cruz) or a rabbit polyclonal antiserum produced against a peptide corresponding to the phosphorylated activation loop of Fes (pY713) (13). STAT3 was immunoblotted with a rabbit polyclonal antibody against the carboxyl terminus of STAT3 (STAT3 C-20; Santa Cruz), and pY705 STAT3 was detected with a mouse monoclonal antibody (Upstate). Immunoblot analysis of actin was performed with a mouse monoclonal antibody (Chemicon International).
Immunohistochemistry-Formalin-fixed and paraffin-embedded adult human colon tissue microarrays (Clinomics Biosciences, Cytomyx, and Asterand) were deparaffinized in xylenes, rehydrated through a graded alcohol series, and permeabilized for 20 min in 0.5% Tween 20. Endogenous peroxidase activity was quenched in 3% H 2 O 2 for 20 min, and antigen retrieval was performed by microwave oven incubation (high power, 12 min) in IHC select citrate buffer (Chemicon). Slides were cooled in citrate buffer for 20 min and blocked for 1 h with normal serum. Tissue sections were then incubated with rabbit polyclonal antibodies raised against Musashi-1 (Chemicon) or with goat polyclonal antibodies generated against either Fes amino-terminal (Fes N-19) or carboxyl-terminal (Fes C-19; Santa Cruz) peptides. Antibody blocking experiments were performed by preincubation of the Fes carboxyl-ter-minal antibodies with an excess of the antigenic peptide (Fes C-19 blocking peptide; Santa Cruz) for 1 h prior to tissue staining. Fes antibodies were detected with the Vectastain Elite ABC kit and the DAB substrate kit for peroxidase detection (Vector Laboratories) according to the manufacturer's instructions. Sections were counterstained with Harris hematoxylin solution, modified (Sigma-Aldrich), dehydrated, and mounted with Vectamount (Vector Laboratories). HT-29 cells were fixed in 4% paraformaldehyde, permeabilized in 0.5% Tween 20, and processed as described for tissue sections.
Yeast Studies-Saccharomyces cerevisiae strain YPH499 (Stratagene) was transformed with uracil-selectable pESC-Fes expression vectors as described previously (13). For growth suppression assays, equivalent A 600 units of fresh liquid cultures were spotted onto agar plates containing either glucose (to repress Fes expression) or galactose (to induce Fes expression), and the plates were incubated for 2-4 days at 30°C. To assess Fes protein levels and phosphotyrosine content, liquid cultures were grown in the presence of galactose for 6 h, lysed, and analyzed by immunoblotting as described previously (13).
Soft Agar Assays-HT-29 or HCT 116 cell lines were infected with retroviruses expressing the indicated Fes constructs and selected in G418 for 7-14 days as described above. Cell aliquots were mixed with culture medium containing 0.33% agarose and plated on pre-solidified bottom layers of the same medium containing 0.5% agarose in 35-mm Petri dishes. An additional 1 ml of culture medium was added upon solidification of the top layer, and cells were incubated for 2 weeks at 37°C. Plates were then stained with 3-[4,5-dimethylthiazol-2-yl]-2,5diphenyl tetrazolium bromide (Sigma), and colony numbers were determined from scanned plates using Quantity One colony-counting software (Bio-Rad). Fig.  1A shows a model of the Fes kinase domain in its down-regulated form and illustrates the positions of the four residues mutated in colon cancer (Met-704, Arg-706, Val-743, and Ser-759). Both Met-704 and Arg-706 localize to the activation loop (shown in red) and are in close proximity to the major site of tyrosine autophosphorylation (Tyr-713) (17). Tyrosine 713 plays a critical role in the regulation of Fes kinase and biological activities (17,19), and Met-704 and Arg-706 are positioned to directly impact autophosphorylation and kinase domain function. The other two sites of colon cancer-associated mutations, Ser-759 and Val-743, localize to the carboxyl-terminal lobe of the kinase domain away from the active site but may affect kinase activity through an allosteric mechanism.

Colon Cancer-associated Mutations Block Fes Kinase Activity-
To determine the impact of the colon cancer-associated mutations on Fes kinase activity, individual mutants were created and expressed in human 293T cells. Kinase activity was assessed via immunoblot analysis with a general anti-phosphotyrosine antibody (pTyr) as well as a phospho-specific antibody against the Fes activation loop (pY713). As shown in Fig. 1B, wild-type (WT) Fes reacted strongly with both antibodies, indicative of kinase activation resulting from the high level of Fes expression in this system. Surprisingly, the M704V, R706Q, and V743M mutations all substantially reduced Fes autophosphorylation, whereas S759F caused a partial reduction. To address the effects of these mutations on substrate phosphorylation in vivo, wild-type and mutant forms of Fes were co-expressed with STAT3, a previously identified Fes substrate (16). As demonstrated in Fig. 1C, co-expression of wild-type Fes (WT) resulted in robust STAT3 Tyr-705 phosphorylation, consistent with previous results (16). The M704V, R706Q, and V743M mutations substantially reduced STAT3 phosphorylation by Fes, whereas the S759F mutant had little impact, consistent with the effects of these mutations on Fes autophosphorylation. The observation that three of four colon cancer-associated mutations inhibit Fes kinase activity argues against a contributory role for Fes in colon cancer.
Colon Cancer-associated Mutations Inhibit Fes Kinase Activity Independently of Mammalian Cell Factors-To determine whether the suppressive effects of the colon cancer-associated mutations are intrinsic to the structure of Fes or require interaction with other proteins in mammalian cells, the WT and mutant Fes proteins were expressed in S. cerevisiae. Yeast expression provides an excellent model system to explore the functional consequences of these mutations on Fes structure and activity. Prior studies have shown that wild-type Fes naturally adopts an inactive conformation in yeast that can be activated with a point mutation (L145P) in the amino-terminal coiled-coil regulatory region (13). The yeast system therefore allowed us to assess the effects of the colon cancer-associated mutations in both the down-regulated and active forms of Fes without interference from endogenous tyrosine kinases. Moreover, ectopic expression of active Fes (as well as other non-receptor tyrosine kinases) inhibits yeast cell growth, providing a convenient in vivo biological assay for Fes activity (13,20,21).
S. cerevisiae cultures transformed with Fes expression plasmids were started in liquid culture, and aliquots were spotted on agar plates containing either galactose to induce Fes expression or glucose to repress it (Fig. 2). As reported previously, wild-type Fes expression had little effect on yeast growth, whereas the active variant (L145P) significantly impaired growth on galactose medium (13). To correlate growth suppression with kinase activity, protein extracts from yeast cultures grown in the presence of galactose were analyzed by immunoblotting. Wildtype Fes showed no evidence of autophosphorylation, whereas the active L145P mutant reacted strongly with both the pTyr and pY713 antibodies, in agreement with the growth suppression data. Moreover, expression of wild-type Fes did not result in detectable tyrosine phosphorylation of yeast cell proteins, whereas L145P expression significantly increased yeast protein phosphotyrosine content.
Expression of Fes proteins carrying the individual colorectal cancerassociated mutations had little impact on yeast growth, Fes autophosphorylation, or yeast protein phosphotyrosine content, clearly demonstrating that these mutations do not activate wild-type Fes (Fig. 2). In contrast, the M704V, R706Q, and V743M mutations all strongly suppressed Fes-L145P autophosphorylation, caused a substantial reduction in tyrosine phosphorylation of yeast cell proteins, and reversed the growth inhibitory effect. This reduction in tyrosine phosphorylation of Fes substrate proteins in yeast agrees with the STAT3 result in Fig. 1C and provides further support for the idea that these mutations directly affect Fes kinase function. As in 293T cells, the S759F mutation did not markedly alter Fes-L145P kinase or biological activity in yeast. Demonstration that colon cancer-associated mutations affect Fes kinase activity in the same way in yeast and human cells suggests that these mutations affect kinase structure independently of cellular regulatory factors.
Fes Expression in Human Colonic Epithelium-Although high Fes mRNA levels have been observed in the developing gut (22), Fes protein expression in the adult has not been reported for this tissue type. To determine whether Fes is expressed in the adult colon, sections of normal colon tissue from 58 colon cancer patients were probed with antibodies raised against the carboxyl-terminal region of human c-Fes. Nearly 86% of these patients exhibited a staining pattern similar to that represented in Fig. 3A, in which Fes expression was restricted to the cytoplasm of a distinct subpopulation of intestinal crypt epithelial cells. The remaining sections exhibited either diffuse Fes staining in all cells or no detectable expression (data not shown). Interestingly, longitudinal  sections of colonic crypts (Fig. 3, B and C) exhibited cells with strong Fes staining that localized to the lower portion of the crypts.
Normal colon tissue stained with the Fes carboxyl-terminal antibody is shown at higher magnification in Fig. 3D and more clearly shows robust Fes expression in the cytoplasm of individual crypt cells. The basal nuclei of these cells are evident from their blue counterstaining with hematoxylin. A serial section from this same sample was stained with an antibody raised against the amino-terminal region of Fes and shows a very similar pattern of Fes immunoreactivity (Fig. 3E). Preincubation of the Fes carboxyl-terminal antibody with the corresponding peptide antigen completely eliminated immunostaining to this tissue section, providing evidence for antibody specificity (Fig. 3F). As an additional control for antibody specificity, the Fes-negative colorectal cancer cell line HT-29 was transduced with retroviruses carrying either GFP alone (Fig. 3G) or GFP fused to Fes (Fig. 3, H and I). Immunostaining with the Fes carboxyl-terminal antibody demonstrated no immunoreactivity in GFP-expressing cells consistent with the lack of Fes protein in HT-29 cells observed via immunoblotting (see Fig. 5B). In contrast, robust Fes staining was observed in HT-29 cells expressing the GFP-Fes fusion protein (Fig. 3H), further validating Fes antibody specificity. Together, these data provide the first evidence for Fes protein kinase expression in the crypt cells of normal human colonic epithelium.
Fes Localizes to Musashi-1-expressing Cells-Intestinal epithelial cells are believed to arise from a small population of stem cells located near the base of colonic crypts (23)(24)(25). Localization of Fes to distinct epithelial cells near the base of the colonic crypts (Fig. 3, B and C) suggested that Fes expression might be restricted to these colonic epithelial progenitors. Unfortunately, lack of clear morphological criteria and definitive molecular markers has hindered identification of intestinal stem cells. However, Musashi-1, an RNA-binding protein required for asymmetric cell division during Drosophila neural development, was recently characterized as a putative stem cell marker in human colonic epithelium where it exhibited an expression pattern similar to that of Fes described here (26 -28). To investigate whether Fes is expressed in Musashi-1-positive cells, serial human colon tissue sections were stained with Fes or Musashi-1 antibodies. As demonstrated in Fig. 4, serial sections from three separate patients exhibited Musashi-1 and Fes staining in the same cells, suggesting that Fes expression may be restricted to the stem cells of the gut epithelium.
Fes Expression Is Reduced or Absent in Colorectal Cancer-To determine whether Fes expression is altered in colorectal cancer, human colon tumor tissue and matched normal control sections from 49 patients were immunostained with the Fes carboxyl-terminal antibody. As shown in Table 1, Fes expression was observed in all 49 normal tissues examined but was reduced in tumor tissue from 67% (33 of 49) of these patients. Moreover, Fes expression was undetectable in 16 tumor sections. Fig. 5A illustrates Fes localization to select crypt cells in the normal mucosa from two representative patients (panels a and b). In contrast to normal colonic epithelium, Fes expression was not observed in tumor sections from these same individuals (panels c and d). The reduction or loss of Fes expression in nearly 70% of the tumor samples is consistent with a tumor suppressor function for Fes in colonic epithelium.
Fes expression was also examined in six independent human colorectal cancer cell lines (DLD-1, SNU 1040, HCT 116, COLO 320, Caco-2, and HT-29) and from two human myeloid leukemia cell lines (TF-1 and  K-562) for comparison (Fig. 5B). Fes was strongly expressed in TF-1 cells but was not detected in K-562 cells, in agreement with prior studies (7,29,30). Consistent with the loss of Fes expression in human colon tumor samples, Fes protein was undetectable in all colorectal cancer cell lines with the exception of Caco-2 cells. Here, Fes expression was observed although at a level significantly lower than that seen in the TF-1 myeloid leukemia cell line. These data further suggest that the loss of Fes protein expression, like the inactivating mutations described above, may contribute to colorectal cell transformation.
Fes Suppresses Anchorage-independent Growth of Colorectal Cancer Cell Lines-To test directly whether Fes modulates colorectal cancer cell growth, the Fes-negative HT-29 and HCT 116 cell lines were transduced with recombinant retroviruses carrying either GFP alone or fused to wild-type or kinase-active forms of Fes. The active form of Fes used in these experiments (Fes-Fps) is a chimera in which residues 610 -822 of the c-Fes kinase domain have been replaced with the corresponding region of an active viral homologue (v-Fps) (15). Each of the cell lines was then assayed for anchorage-independent growth in a soft agar colony-forming assay. As shown in Fig. 6, wild-type Fes expression reduced soft agar colony formation to 60% of the levels observed with the GFP control in HT-29 cells and to 54% of control levels in HCT 116 cells. Expression of the active Fes-Fps mutant further reduced colony numbers to 31% of control levels in HT-29 cells but did not further suppress the growth of the HCT 116 cell line. This difference in Fes-Fps biological activity may reflect genetic differences between these cell lines. Assessment of Fes phosphotyrosine content via immunoblot analysis demonstrated robust activity for Fes-Fps in both cell lines, whereas wild-type Fes activity was low (HCT 116) or fully repressed (HT-29). In contrast to Fes, expression of an active form of the Src family kinase Hck had very little impact on HT-29 colony formation (data not shown), supporting the idea that the Fes tyrosine kinase has a unique role in growth suppression. Together, these data show that Fes suppresses anchorage-

TABLE 1 Fes expression in normal colon versus tumor
Colon tissue sections from 49 colorectal cancer patients were immunostained with an antibody raised against a c-Fes carboxyl-terminal peptide as described under "Materials and Methods" and the legend to Figure 3. Relative Fes staining was scored using the following scale: Strong, robust Fes expression in five or more cells; Moderate, robust Fes expression in 1-4 cells; Weak, low levels of diffuse staining with an absence of robust Fes-positive cells; Undetectable, no expression detected. Patient age, sex, and TNM tumor staging are also included. A detailed explanation of the TNM tumor staging system (T, extent of primary tumor; N, metastasis to lymph nodes; M, distal metastasis) can be found at the American Cancer Society web site (www.cancer.org). NA, not available.

DISCUSSION
Data presented here define an unexpected tumor suppressor function for the c-Fes protein-tyrosine kinase in colorectal cancer. We have shown that c-fes mutations previously associated with colon cancer (8) dramatically inhibit rather than stimulate Fes kinase activity. Fes protein levels were consistently reduced or absent in primary human colon tumors compared with matched normal tissue. Moreover, in colon cancer cell lines where Fes expression was undetectable (HT-29 and HCT 116), re-establishing Fes expression by gene transfer was sufficient to suppress anchorage-independent growth, providing direct evidence that Fes acts as a growth suppressor in this cell type.
Our study is the first to report Fes protein expression in epithelial cells of the adult human colon, suggesting an important physiological role in normal colon function. Fes has previously been shown to promote differentiation of hematopoietic, endothelial, and neuronal cells (3,5,14,31), suggesting that it may affect the differentiation program of colonic epithelium as well. Surprisingly, Fes was expressed in a distinct subpopulation of colonic epithelial cells localized near the base of some crypts (Fig. 3). Colonic epithelial stem cells have also been proposed to reside in limited numbers at the base of colonic crypts (24,25), suggesting that Fes might be expressed in these cells. In support of a role for Fes in colon progenitor cells, Fes localized to cells expressing the putative colon stem cell marker Musashi-1 (Fig. 4). However, further studies are necessary to identify the precise physiological role of Fes in normal colon.
Data presented here suggest that both the presence of the c-Fes protein as well as its tyrosine kinase activity may contribute to its putative tumor suppressor function. Three of four mutations in the Fes kinase domain associated with colon cancer substantially inhibited Fes kinase activity, whereas a fourth had little effect (Figs. 1 and 2). Although Fes protein was expressed in normal colonic epithelium, it was reduced or absent in tumor tissue from 67% of these same individuals ( Fig. 5 and Table 1). It will be of great interest to determine whether colon tumors staining positive for Fes exhibit inactivating mutations in the c-fes locus. In HT-29 cells, which fail to express endogenous Fes, re-introduction of wild-type Fes induced partial suppression of anchorage-independent growth, whereas introduction of a kinase-active mutant suppressed growth further (Fig. 6). Taken together, these data suggest that although kinase-inactivating mutations may promote colorectal cancer progression, a further selective advantage may be conferred following loss of Fes expression.
The mechanism by which Fes inhibits colorectal cell growth is currently unclear. Mutations in components of the Wnt signaling network that activate ␤-catenin are very common in colorectal cancer (10,12,23,32), suggesting that Fes may exert growth suppression via inhibition of ␤-catenin function. In support of this idea, co-immunoprecipitation experiments showed strong association of active Fes with endogenous ␤-catenin in HCT 116 cells (data not shown), one of the colon cancer cell lines that responded to Fes expression with reduced colony-forming activity (Fig. 6). Moreover, the Fes homologue, Fer, has been shown to modulate ␤-catenin function at adherens junctions, resulting in enhanced cell adhesion (33). A similar role for Fes would fit with the suppression of anchorage-independent growth of colon tumor cells reported here.
Protein-tyrosine kinases are generally regarded as oncogenic, and activating mutations, enhanced expression, and constitutive kinase activity have been reported for a wide variety of cancers, including colon tumors (34 -37). For example, protein-tyrosine kinases of the Src family are constitutively active in many colon cancer cell lines where they disrupt cadherin-catenin complexes and promote anchorage-independent growth (37)(38)(39)(40). Along these lines, a tumor suppressor function has been suggested for Csk, the tyrosine kinase responsible for negative regulation of c-Src and other members of the Src kinase family (41).  ) and tumor (c and d) tissue were probed with antibodies raised against the Fes carboxyl-terminal region as described in Fig. 3. B, Fes was immunoprecipitated from the indicated cell lines with antiserum generated against the Fes amino-terminal and SH2 regions followed by immunoblotting with an antibody against the Fes amino-terminal domain (upper panel). Clarified lysates were also immunoblotted with actin-specific antibodies as a control for input protein levels (lower panel). Data are representative of two independent experiments. Although a tumor suppressor role for a non-receptor protein tyrosine kinase such as Fes is very unusual, a similar loss of Syk expression has been associated with development of breast cancer (42). However, Fes has previously been shown to inhibit growth and induce terminal differentiation of chronic myelogenous leukemia cells, suggesting a more general role as a tumor suppressor (6,7). Indeed, mice with either null or kinase-inactivating mutations have very recently been shown to develop breast cancer with a shorter latency, and this effect can be rescued with a c-fes transgene (43). The Fes kinase and/or the downstream mediators of its tumor suppressor function may therefore represent unexplored targets for drug discovery efforts aimed at reducing cancer progression. Conversely, an important consideration related to kinase-directed drug discovery (44) is that candidate inhibitors remain free from nonspecific effects on Fes tyrosine kinase activity, as inhibition of this kinase pathway may favor tumor development.