Introduction

Aggressive fibromatosis, also called desmoid tumor, is a locally invasive soft tissue lesion composed of a clonal proliferation of mesenchymal cells (Alman et al., 1997a). Most sporadic cases of aggressive fibromatosis contain a somatic mutation in either the adenomatous polyposis coli (APC) or -catenin genes, resulting in -catenin protein stabilization (Alman et al., 1997b; Li et al., 1998; Tejpar et al., 1999). Amino-terminal phosphorylation sites of -catenin are necessary for its degradation at the proteosome. Gsk3-, axin/conductin and APC form a complex that regulates the phosphorylation of these sites. Mutations in exon three of -catenin that remove the phosphorylation sites or mutations in APC, result in the stabilization of -catenin protein. Stabilized -catenin protein can bind to transcription factors in the Tcf-Lef family, resulting in transactivation of transcription (Miller et al., 1999; Roose and Clevers, 1999). Studies of genetically engineered mice show that -catenin stabilization is sufficient to cause aggressive fibromatosis, as demonstrated by the development of this tumor type with overexpression of a stabilized form of -catenin driven by a tetracycline-regulated CMV promoter. -catenin has also been shown to be transiently stabilized and transcriptionally active in normal mesenchymal (dermal fibroblast) cells during the proliferative phase of cutaneous wound healing. Using genetically engineered mice, stabilization of -catenin protein was found to cause a larger size of healing wounds, suggesting that cells in aggressive fibromatosis behave like unchecked proliferating wound fibroblasts (Cheon et al., 2002).

Familial adenomatous polyposis (FAP) and familial infiltrating fibromatosis (FIF) are two inherited conditions in which patients develop aggressive fibromatosis. Both result from germ-line mutations in APC. FIF is associated with a more 3' location of a mutation in APC than that responsible for FAP, and there is an almost universal occurrence of aggressive fibromatosis in FIF. An animal model of FIF, the Apc1638N mouse, carries a targeted mutation in the 3' end of the APC gene. By 6 months of age, male Apc/Apc1638N animals develop as many as 45 fibromatoses, and females form as many as 16 fibromatoses (Fodde et al., 1994; Smits et al., 1998; van der Houven van Oordt et al., 1999). Mice also develop gastrointestinal polyps, although in fewer numbers than in the Min mouse, which harbors an Apc mutation located 5' to that of the Apc1638N mouse.

The extracellular matrix environment plays an active role regulating cell proliferation and invasion during wound healing and tumor progression (Bissell and Radisky, 2001; Gutierrez-Ruiz et al., 2002). The accumulation and degradation of one extracellular matrix component, the polysaccharide hyaluronan (HA), regulates the extent of fibrosis in adult wounds (Gerdin and Hallgren, 1997), and enhanced accumulation of HA is part of a connective tissue desmoplastic reaction that is associated with a poor outcome in several human cancers (Toole, 2002). Overexpressing HA synthases, or adding purified HA in vitro, promotes tumor cell proliferation and invasive/metastatic properties, suggesting a role for this polysaccharide in neoplastic progression (Tammi et al., 2002; Turley et al., 2002). Cellular effects of HA on tumor cells are largely mediated by HA receptors or hyaladherins (Turley et al., 2002), the most characterized of which are CD44 and Rhamm. Both of these receptors, which regulate signaling through Ras and other small GTPases (Gares and Pilarski, 2000; Turley et al., 2002; Ponta et al., 2003), are linked to fibroproliferative (Lovvorn et al., 1998) and neoplastic processes (Naor et al., 1998). Rhamm (Hardwick et al., 1992), the first cellular HA-binding protein identified, was originally referred to as HABP (hyaluronan-binding protein, Turley, 1982), then later designated as HMMR (for human Rhamm, Spicer et al., 1995), IHABP/Rhamm (Assmann et al., 1999) for intracellular Rhamm, and CD168 for the cell surface form of Rhamm. There are now several HA-binding proteins that resemble Rhamm/HMMR/IHABP/CD168 by commonly occurring within the intracellular, extracellular and cell surface compartments (Turley et al., 2002). Here we use Rhamm to distinguish this protein from other, similar HA-binding proteins. Rhamm occurs on the cell surface, as a peripheral or GPI-linked protein, and such intracellular compartments as the cell nucleus, cytoskeleton, podosomes, lamellae and mitochondria (Turley et al., 1990; Pilarski et al., 1994; Zhang et al., 1998; Assmann et al., 1999; Lynn et al., 2001). Rhamm expression is low or undetectable in most normal tissues, but is upregulated following wounding in vivo (Savani et al., 1995), culture at low confluence in vitro in the presence of growth factors (Turley and Auersperg, 1989; Hardwick et al., 1992; Cheung et al., 1999; Savani et al., 2001) or upon neoplastic transformation in vitro (Hall et al., 1995) and in vivo (Tammi et al., 2002; Turley et al., 2002). For instance, high expression of human Rhamm is prognostic of a poor outcome in some tumors (Wang et al., 1998; Li et al., 2000; Assmann et al., 2001) and high Rhamm expression is characteristic of many aggressive human neoplasms (Crainie et al., 1999).

Rhamm is most extensively studied in mesenchymal cells. It is strongly upregulated following wounding of fibroblasts and smooth muscle cells in vitro (Savani et al., 1995) and during hypertrophic scarring of human skin grafts in vivo (Lovvorn et al., 1998). Furthermore, Rhamm antibodies, dominant-negative protein forms, soluble recombinant Rhamm protein and antisense Rhamm cDNA inhibit proliferation, motility and Ras-mediated transformation of immortalized fibroblasts in vitro (Hall et al., 1995; Mohapatra et al., 1996). Conversely, overexpression of Rhamm cDNA can be transforming and progressing fibroblast-like cells in vitro (Hall et al., 1995). Since Rhamm regulates mesenchymal cell functions during wound healing and neoplastic transformation, we examined the role of Rhamm in the mesenchymal tumor, aggressive fibromatosis.

Results

Rhamm is expressed in aggressive fibromatosis

We compared six aggressive fibromatosis tumors with patient-matched normal fascial tissues for the expression of Rhamm using RT–PCR and immunohistochemistry. In all the six cases, there was increased Rhamm protein as detected by immunohistochemistry in the fibromatosis tumors compared to normal mesenchymal tissues on the same slide (Figure 1). Rhamm staining in the tumor cells was quite intense, and all cells exhibited positive staining (Figure 1b). For controls, we examined the mesenchymal cells surrounding blood vessels on the same histological sections and there was minimal detectable Rhamm staining in these cells (Figure 1c). The specificity of the antibody for Rhamm was established by absence of staining in Rhamm-/- cells as in Figure 2 and by lack of staining of fibromatosis tumor sections using nonimmune serum (Figure 1d). In five of the six cases, RT–PCR showed upregulation of RHAMM at the mRNA level, compared to normal fascia from the same patient and as detected using semiquantitative RT–PCR (Figure 1a). When all cases were averaged, the expression of RHAMM mRNA was 3.4 times higher in the tumors compared to normal tissue. Interestingly, in the one case where RHAMM mRNA was not elevated but, rather, was similar to matched normal tissue, Rhamm protein expression was nevertheless increased in the tumor relative to normal mesenchymal cells, as measured using immunohistochemistry. The lack of detection of a difference in mRNA expression in this one sample may have been due to tumor infiltrating into the specific region from which the normal fascia sample was obtained.

Figure 1.

Rhamm is expressed in human aggressive fibromatosis (RT–PCR and immunostaining). Panel a shows the result of semiquantitative RT–PCR data, showing a higher level of expression of RHAMM in aggressive fibromatosis than in normal fibrous tissue from the same patient, compared to GAPDH as a control. Data from two tumors and normal tissue are shown, with PCR products from the tumors loaded in lanes labeled 'T' and PCR products from normal tissues loaded in lanes labeled 'N'. Panel b shows immunohistochemistry for Rhamm in aggressive fibromatosis tumor cells, showing intense staining in all of the tumor cells, compared to sparce staining in normal cells on the same histological slide as shown in Panel c. Panel d shows a histological section from an aggressive fibromatosis using nonimmune serum as an additional negative control. Histological sections are shown magnified 120

Full figure and legend (250K)

Figure 2.

Strategy for homologous recombination of the Rhamm gene. Panel a shows the homologous recombination strategy. Exons 8–16 of the 18 exons containing Rhamm gene were replaced by the Hprt selection marker which was flanked by 9 kb homologous genomic 129 DNA. Probes used in Southern blot analysis of ES cell or tail DNA are indicated. Panel b shows Southern blot analysis of ES cell and tail DNA. Sequences of exons 6 and 7 (5' probe) as well as exons 17 and 18 (3' probe) were used as hybridization probes in Southern blot analysis of ES cell or tail DNA. The presence of a BglI site in the selection marker results in the detection of a 16 kb (5' probe) or 5 kb (3' probe) BglI fragment in addition to a 20 kb BglI fragment of the wt allele. Lanes are labeled wt for wt, het for heterozygous wt and knockout, and hom for homozygous knockout alleles

Full figure and legend (81K)

Characterization of Rhamm-/- mice

To investigate the role of Rhamm in vivo, Rhamm-deficient mice were utilized. Exons 8–16 of the murine Rhamm gene were deleted by homologous recombination in embryonic stem (ES) cells using the strategy outlined in Figure 2. Heterozygous ES cell clones were identified by detecting both a 20 and 16 kb BglI fragment using Southern Blot analysis (Figure 2b). Based on these analyses, the efficiency of homologous recombination was estimated to be approximately 10%. The Rhamm knockout allele was transmitted to germ line in one out of five chimeric males.

The offspring of brother–sister matings of heterozygous mice were genotyped by Southern Blot analysis (Figure 2) and results show that the 20 kb Rhamm BglI fragment is reduced in heterozygotes and absent in homozygotes. The expected 1 : 2 : 1 ratio of wild-type (wt), heterozygous and homozygous mice was obtained, indicating that embryonic lethality did not occur. The appearance of Rhamm-/- mice was normal at birth both in the size and gross anatomy of neonates. Further, an overview of tissue histology revealed no obvious defects. Rhamm-/- animals were viable for a normal lifespan of 2 years. However, homozygous matings resulted in significantly fewer viable litters compared to either heterozygote or wt matings, suggesting a fertility defect.

Rhamm-/- animals retain exons 1–7 and 17–18 of the Rhamm gene (Figure 3). The splice donor site of exon 7 is intact and the next splice acceptor site downstream of exon 7 is located on exon 17. Fusion of exons 7 and 17 by splicing should result in a frame shift between exon 7 and the HA-binding region located in exon 18 (Yang et al., 1994). Thus, our strategy for deleting Rhamm could permit the expression of an N-terminal 920 bp mRNA, but not the expression of its carboxyl terminus encoding the HA-binding region. To both confirm that full-length (fl) Rhamm mRNA is not expressed and to assess whether or not a 920 bp fragment is expressed, Rhamm mRNA and protein expression in sample tissues isolated from adult wt and Rhamm-/- mice, as well as primary wt and Rhamm-/- mouse embryonic fibroblasts (MEF), were analysed (Figure 3). Since previous reports have noted that the highest expression of Rhamm mRNA in normal tissues occurs in spleen, thymus, testes and placenta (Line et al., 2002), we isolated mRNA from Rhamm-/- and wt spleen for RT–PCR analysis. We also used mRNA from C3 cells as a positive control, since these have previously been reported to express high levels of Rhamm (Hall et al., 1995). RT–PCR analysis of RNA from wt spleen tissue and C3 cell lysates, using primers specific for exons 3 and 12, amplified the expected 1.1 kb mRNA product of Rhamm (Figure 3). In contrast, this 1.1 kb PCR product was absent in Rhamm-/- spleen and Rhamm-/- MEF (Figure 3). Using primers specific for exons 3 and 17, low levels of the expected 650 bp RT–PCR product (see methods) were expressed in Rhamm-/- MEF, but this product was not detected in wt samples (Figure 3). Sequencing the 650 bp product confirmed that it encodes N-terminal exons of Rhamm but not the carboxyl-terminal 2 exons coding for the HA-binding region. Rhamm mRNA is subject to alternative splicing (Hall et al., 1995, Assmann et al., 1999; Crainie et al., 1999), and although the fl form of Rhamm mRNA is most commonly expressed in normal tissues, protein analysis has not previously been reported. Therefore, the Rhamm proteins expressed by normal spleen tissue were determined using a Western blot assay (Figure 3). Comparison of wt and Rhamm-/- spleen tissue also permitted assessment of the specificity of our antibodies for Rhamm protein. A Rhamm-positive protein corresponding to fl Rhamm protein (95 kDa in mouse) was observed using an affinity-purified polyclonal anti-Rhamm antibody (Figure 3), but was absent in the Rhamm-/- tissue. These results confirm the specificity of the Rhamm antibody and indicate that only fl Rhamm protein is detectably expressed in spleen. We did not detect the expression of a protein corresponding to the predicted 950 bp mRNA in Rhamm-/- tissue using Western blot assays. Since it is possible that antibodies to sequences encoded in exons 17 and 18 are not represented in the polyclonal population of antibodies used for Western analysis, we utilized polyclonal antibodies prepared against the peptide sequence encoded in exon 7, which would be expressed from the 920 bp mRNA fragment of Rhamm-/- tissue (Lynn et al., 2001). Positive immunofluorescence was detected in wt but not Rhamm-/- MEF with this antibody. As shown in Figure 3, the Rhamm antibody stained MEF rescued with fl Rhamm and staining decorated the cytoskeleton as described previously for the intracellular Rhamm (Assmann et al., 1999). Rhamm-/- MEF (Figure 3) did not stain above that of IgG controls. These results show that this anti-Rhamm antibody is specific for Rhamm protein.

Figure 3.

Characterization of Rhamm-/- mice. Panels a and b show RT–PCR and Western blot analysis of spleen, confirming an absence of Rhamm expression in tissue from Rhamm-/- animals, and the presence of both Rhamm mRNA and protein in wt tissue. In panel a, 10 g of total RNA isolated from the spleen was reverse transcribed using oligo dT as primer. Using oligo nucleotide primer specific for Rhamm exons 3 and 12, a 1.1 kb product was amplified from Ras-transformed C3 cells and wt spleen cDNA, but not from Rhamm-/- spleen cDNA. Amplification of -actin was used as loading control. In panel b, proteins from spleen lysates (30 g protein) was separated on SDS–PAGE and probed for Rhamm using Western blots. Antibodies against a recombinant Rhamm protein detects fl (95 kDa) Rhamm protein in wt spleen, but not in Rhamm-/- spleen lysates. -actin protein was used as a loading control. Panel c shows RT–PCR detection of a truncated Rhamm transcript expressed by Rhamm-/- fibroblasts. Total RNA was isolated from subconfluent cultures of wt and Rhamm-/- embryonic fibroblasts plated on fibronectin-coated cell culture plates. RNA (10 g) was reverse transcribed using oligo dT as primer. Using primer specific for Rhamm exons 3 and 17, a 2.2 kb product corresponding to fl Rhamm was amplified from wt MEF cDNA. This 2.2 kb product is absent in Rhamm-/- MEF, but a minor 650 bp product could be detected, although this was not detected in spleen tissue. Amplification of -actin was used as loading control. Panels d–g show immunofluorescence of Rhamm-/- MEF with Rhamm exon 8 specific antibodies 200. Fl Rhamm was re-expressed in immortalized Rhamm-/- fibroblasts by retroviral infection. Subconfluent cultures plated on fibronectin were immunostained with antibodies raised against a peptide sequence encoded in Rhamm exon 8 (d, e). Most cells infected with fl Rhamm express Rhamm protein and staining is observed on the cytoskeleton. Cells infected with empty vector (f) do not stain with the Rhamm-specific antibodies. Incubation with rabbit IgG was used as negative control. The bar represents 50 m (g)

Full figure and legend (218K)

Rhamm deficiency results in a decreased number of murine aggressive fibromatosis tumors, but does not alter the number of gastrointestinal tumors formed in the Apc1638n mouse

The Apc1638N mouse develops a large number of aggressive fibromatosis tumors by 6 months of age, with male mice developing a larger number of lesions than female mice. To determine the role of Rhamm in this mesenchymal tumor, we mated Apc1638N mice with the Rhamm-/- mice. Both male and female Apc/Apc1638N-Rhamm-/- mice developed significantly fewer aggressive fibromatosis lesions, than did their Apc/Apc1638N-Rhamm+/+ littermates at 6 months of age (Figure 4). In female mice, there was a significant difference in the average volume of the tumors that formed in Apc/Apc1638N-Rhamm+/+ vs Apc/Apc1638N-Rhamm-/- mice (17 vs 7 mm³, P<0.05). However, there was no difference in the average volume of the tumors that formed in male mice (22 vs 19 mm³). Apc1638N mice also develop gastrointestinal polyps and subdermal cysts. However, in contrast to aggressive fibromatosis, there was no difference in the number or size of polyps or cysts that developed between Rhamm-deficient animals and their wt littermates.

Figure 4.

Incidence of aggressive fibromatosis tumors in the Apc1638N mice deficient in Rhamm. Male mice develop more fibromatosis tumors than do female mice. Mice deficient in Rhamm develop significantly fewer tumors than do littermates that express Rhamm (P<0.01 for both male and female mice). There were 10 male and 10 female mice in each genotype. Error bars are 95% confidence intervals

Full figure and legend (43K)

Rhamm regulates proliferation at culture subconfluence

Primary wt MEF, immortalized MEF rescued by transfection with fl Rhamm cDNA, and wt primary transformed fibroblasts incorporate more Bromo-deoxyuridine (BrDu) when plated at sparse culture density (<50%) than their Rhamm-/- counterparts do (Figure 5). All cell types incorporate more BrDu in the presence of serum supplements than in defined medium (Figure 5A), but the response of cells lacking Rhamm does not reach the level of BrDu incorporation observed in wt cells. These results indicate that primary Rhamm-/- cells have not lost the ability to respond to growth factors, but this response is blunted. This effect of Rhamm on cell replication in response to serum is dependent upon culture confluence. At higher culture confluence (70%), there is no difference in BrDu incorporation in the presence or absence of Rhamm, although cells are still incorporating BrDu in response to serum supplements (Figure 5b).

Figure 5.

Effect of Rhamm on cell proliferation in fibroblasts. Panel a: Rhamm plays a role regulating proliferation in response to serum. Primary MEF were plated onto sterile glass coverslips in growth medium at sparse density for several hours and the growth medium was then either replaced with another aliquot of growth medium or defined medium (serum free), both containing BrDu as described in Materials and methods. wt Primary MEF significantly (P<0.0001) increase BrDu incorporation in response to serum, compared to serum-starved controls. Rhamm-/- primary MEF also respond to serum by a significant enhancement of BrDu incorporation (P<0.00001), but levels of incorporation in the presence of serum supplements remain significantly lower than for the wt cells (P<0.0001). Values represent the mean and 95% confidence intervals of triplicates sampling (approximately 100 cells/replicate). Panel b: Culture density determines whether or not loss of Rhamm affects cell proliferation. Immortalized Rhamm-/- MEF (-/- line) were studied after transfection with fl Rhamm (FL-R, labeled '+'), or with an empty construct (labeled '-'). Cells were plated onto sterile glass coverslips and exposed to BrDu in growth medium or serum-free culture medium as described in Materials and methods and in Panel a above. Cells expressing fl Rhamm incorporated significantly more BrDu in the presence of serum supplements (P<0.0001) than cells not expressing fl Rhamm under conditions of low confluence. When cells were plated at higher density so that proliferation was reduced relative to sparse culture conditions, a difference in incorporation between cells expressing fl Rhamm and those not expressing Rhamm was not observed. Cells still exhibit significantly greater proliferation in the presence of serum supplements compared to their absence (P<0.01 for both fl Rhamm and Rhamm-/-, data not shown). Values represent the mean and 95% confidence intervals of triplicate samples sampling approximately 100 cells/replicate

Full figure and legend (73K)

Primary cell cultures derived from aggressive fibromatosis tumors from Apc/Apc1638N-Rhamm+/+ mice also exhibited a higher proliferation rate than did cells from tumors derived from Apc/Apc1638N-Rhamm-/- animals at low cell confluence (Figure 6). Cells derived from both Rhamm+/+ and Rhamm-/- animals exhibited higher proliferation rates at higher serum levels only at low culture confluence, similar to our findings using nontumor fibroblasts. Taken together, these results suggest that Rhamm plays a role regulating cell proliferation when cells are grown under sparse conditions where cell contact is limited, both in primary fibroblast cultures and in cultures derived from aggressive fibromatosis tumors.

Figure 6.

Effect of Rhamm on proliferation of aggressive fibromatosis cell cultures. Proliferation is lower in aggressive fibromatosis tumor cell cultures from Rhamm-/- animals than given for aggressive fibromatosis tumor cells from Rhamm+/+ animals when grown under conditions of low cell density. The percentage of cells incorporating BrDu is given from primary cell cultures grown under conditions of low confluence and high confluence, as in Figure 5. Values represent the mean and 95% confidence intervals of triplicate samples sampling approximately 100 cells/replicate. There is a significant difference in proliferation between cells grown from tumors that developed in Rhamm-/- animals and those that developed in Rhamm+/+ animals at low confluence (P<0.001). Proliferation was higher when cells were grown in high serum concentration (10%) media. However, at high confluence, there was no difference in proliferation detected between tumor-derived cells and normal fibroblast-derived cells

Full figure and legend (59K)

Discussion

Aggressive fibromatosis shares histological and molecular characteristics with mesenchymal cells that accumulate during the proliferative phase of dermal wound healing. Here, we show that Rhamm, a protein important in wound healing, also plays an important role in aggressive fibromatosis. Although increased Rhamm expression is found in a variety of human neoplasias (Assmann et al., 2001; Zhou et al., 2002; Kong et al., 2003), this is the first study to directly assess a role for Rhamm in tumor biology in vivo. We found that the genetic deletion of Rhamm significantly reduces the number and size of fibromatosis, arising as a result of an APC mutation, providing genetic evidence for a role of this HA-binding protein in tumor pathology. Our data showing an effect of Rhamm loss on tumor cell proliferation suggest that aberrant cell growth regulation or clonal expansion may contribute to this effect. Interestingly, although our murine studies showed a decrease in the number of aggressive fibromatosis tumors that developed in Apc1638N mice deficient in Rhamm, we found no difference in the number of gastrointestinal tumors that formed.

Rhamm alters cell proliferation at low confluence, but not at high confluence in fibroblasts, a finding consistent with a role for this protein in coupling cell signaling with the extracellular environment. Rhamm was originally isolated from the supernatant medium of subconfluent migratory embryonic heart fibroblasts in vitro (Turley et al., 1985). Initial studies showed a role for this protein in the motile behavior of subconfluent fibroblasts (Turley et al., 1991) and subsequent studies demonstrated its role in both tissue fibrosis (Lovvorn et al., 1998) and transformation of fibroblasts (Hall et al., 1995). Previous studies using immortalized 10T1/2 cells suggest that cell surface Rhamm forms (CD168) are required for Ras-mediated transformation of MEF cell lines, since truncated soluble forms of Rhamm blocked transformation (Mohapatra et al., 1996). However, roles of cell surface vs intracellular forms of RHAMM in motility or proliferation have not yet been clearly separated. Our present study does not assess the specific effects of these Rhamm forms in the transformation or invasive process of fibromatosis tumors.

Intriguingly, unlike the effect of genetic deletion of Rhamm on the formation of aggressive fibromatosis tumors, we did not see a significant effect on the formation of gastrointestinal tumors. Thus, in the Apc1638N mouse, Rhamm does not play a role in the transformation of gut epithelial cells. Despite this finding, a number of studies show high Rhamm expression in many human carcinomas that arise from tissue epithelial cells, including human colorectal tumors (Line et al., 2002). In addition, RHAMM is mutated in subsets of human colorectal tumors and linked with a group of genes that contribute to the genomic instability and progression of these tumors (Duval et al., 2001). Our results showing an effect of Rhamm loss on mesenchymal, but not epithelial, transformation may be due to several possibilities. Rhamm may act as a progression factor in neoplasms of epithelial cell origin rather than as an oncogene that plays a role in tumor initiation. For instance, aberrant expression of this hyaladherin in transformed epithelial cells could result from acquisition of fibroblast-specific gene expression due to epithelial-to-mesenchymal transformations (EMT) that are common in aggressive carcinomas and that are frequently associated with progression and poor prognosis. Alternatively, Rhamm expressed by fibroblasts surrounding tumor cells could play a paracrine role in carcinoma progression as paracrine interactions of carcinomas with the surrounding stroma and ECM can contribute to tumor initiation and progression. Another possibility is based on our observation that the effect of Rhamm on cell proliferation is limited to conditions where cell–cell contact is limited such as in very sparse culture conditions. Rhamm may not play a role in tumors arising from epithelial cells because they characteristically exhibit extensive cell–cell contact that might mimic dense culture conditions. Therefore, the effect of Rhamm on fibromatosis formation may occur because this tumor is made up of cells with few cell–cell contacts and surrounded by extracellular matrix, conditions which mimic the sparse culture conditions in vitro where Rhamm plays a role regulating cell proliferation. The role of Rhamm in wound responses, which would also involve proliferation and migration during low cell–cell contact (Lovvorn et al., 1998), are consistent with this possibility.

The molecular mechanisms regulated by Rhamm that affect neoplastic processes including proliferation and migration have only begun to be dissected. For example, Rhamm has been shown to regulate protein tyrosine phosphorylation of actin-binding proteins and activation of Erk through Ras in response to serum, HA, and growth factors such as PDGF (Zhang et al., 1998; Lokeshwar and Selzer, 2000). These effects of Rhamm on signaling pathways are consistent with its reported ability to affect remodeling of the actin cytoskeleton in response to mutant active Ras and v-Src (Hall et al., 1995,1996). Since actin remodeling and Erk hyperactivity and expression have been linked to progression of many human tumors (Sivaraman et al., 1997), the ability of Rhamm to regulate Erk activity is likely one important mechanism related to its oncogenic potential. We showed that high levels of -catenin protein can itself result in the formation of fibromatosis tumors (Cheon et al., 2002). Our results here raise the additional possibility that Rhamm is acting on a -catenin-regulated pathway that controls proliferation. A role for Rhamm on a Wnt-Apc--catenin pathway has not previously been demonstrated and, conversely, a direct link between the -catenin signaling pathway and Ras-regulated pathways such as Erk has also not been established. However, both Ras- and Wnt-regulated pathways control actin cytoskeleton remodeling necessary for cell proliferation/invasion and these pathways may synergize to coordinate actin remodeling (Savanger, 2001). Ras-regulated pathways have been reported to promote release of -catenin from adherens complexes and/or translocation of -catenin to the cell nucleus, where its association with the LEF family of transcription factors is required for the expression of a number of genes, including another HA-binding protein that is also involved in tumor metastasis, CD44 (Wielenga et al., 1999; Wong and Pignatelli, 2002). Given that both Rhamm and -catenin play a role in fibroproliferative processes associated with wound repair or cell transformation, and that Ras and -catenin-regulated pathways interface to regulate cell adhesion and actin remodeling, it is quite likely that functional interactions occur between these two proteins although the precise nature of such associations requires further investigation.

Unlike Cox-2, whose deficiency was found to result in a smaller size to aggressive fibromatosis tumors, and a smaller number of gastrointestinal lesions in the Apc1638N mouse (Poon et al., 2001), Rhamm deficiency resulted in a smaller number of aggressive fibromatosis tumors, but no difference in the number of gastrointestinal lesions. In addition, there was no difference in the size of aggressive fibromatosis tumors that formed in male mice, although the tumors that formed in female mice were of smaller size. The smaller size in the female mice may be partially related to the relatively small number of tumors that formed, or may be related to another factor linked to the X-chromosome. Thus, Rhamm has a very different effect on the phenotype of the Apc1638N mice than Cox-2, suggesting that Rhamm and Cox-2 regulate tumor cell behavior through different mechanisms. This raises the possibility that their effects could be additive, and as such, deficiency of Cox-2 and Rhamm together might result in a larger effect on the tumor phenotype than deficiency of either of these proteins alone.

Current treatments for aggressive fibromatosis are less than satisfactory with radical surgical procedures often being required. Our data suggest that agents that block Rhamm function could be used to treat aggressive fibromatosis. In addition, it is possible that such agents could be used in combination with other drugs blocking -catenin targets, such as cyclooxygenase inhibitors, to produce an effective multidrug regimen to treat patients with aggressive fibromatosis. Such pharmacological therapies are urgently required to treat this troublesome-to-manage tumor.

Materials and methods

Human tumor samples and RHAMM expression studies

Six cases of sporadic aggressive fibromatosis and normal fascia from the same patients were investigated for expression of RHAMM. All the tumors contained a mutation resulting in -catenin protein stabilization, as previously reported (Tejpar et al., 2001). The tumors were treated with surgical excision and samples were cryopreserved as soon as possible after resection and stored in liquid nitrogen vapor for later RNA analysis. Expression at the RNA level was determined utilizing RT–PCR as previously described (Wang et al., 1998), using the primer pair GCAAACACTGGATGAGCTTGA and TGGTCTGCTGATCTAGAAGCA, which produces a 416 bp product. PCR was performed in a semiquantitative manner, with amplification in the linear range at 28, 31 and 34 cycles with GAPDH as a control as previously reported (Hopyan et al., 1999). Both primer pairs were amplified in the same tube. The densities of the resultant PCR bands were measured. Relative differences in expressions between the tumor and normal tissues were determined by averaging the relative density of the RHAMM product compared to the GAPDH product over the three cycle numbers used to amplify each sample.

Formalin-fixed, paraffin-embedded material was utilized for immunohistochemical analysis using a polyclonal antibody raised against peptides 269–288 of Rhamm, a region that is conserved between mouse and human. Sections (4 m) of formalin-fixed paraffin-embedded material were prepared on slides. The antibody was diluted 1 : 600, and incubated overnight at -4°C. Nonimmune serum was also utilized as a control. Slides were incubated with biotinylated goat ant-rabbit IgG (1 h at room temperature) and detected using an avidin–biotin–peroxidase complex (Vector Laboratories, Burlingame, CA, USA).

Generation of Rhamm-/- mice

Rhamm -/- mice used in this study were re-derived from breeding pairs of heterozygous Rhamm+/- mice obtained from the laboratory of Dr. A. Berns (Netherlands Cancer Center, Amsterdam Ned). Rhamm-/- mice were originally prepared as a collaboration between the lab of EAT and AB. Rhamm-/- mice were originally prepared as follows. To delete the exons 8–15 of the murine Rhamm gene (using a 129SVP1 genomic clone provided by Dr AP Spicer, University of California at Davis), a 5.5 kb genomic SacI fragment which contains sequences upstream of exon 8 was cloned into Bluescript SK. Using primer specific for exons 16 and 18, a 3.5 kb genomic fragment was amplified from the 129SVP1 DNA and fused to the SacI fragment. An Hprt selection marker consisting of the Hprt minigene comprised of the promoter and poly A sequence of the Pgk gene (van der Lugt et al., 1991) was inserted between the genomic fragments in antisense orientation. Plasmid DNA was isolated by CsCl gradient centrifugation. Prior to electroporation into ES cells, the construct was digested with NotI/XhoI and the insert was isolated by agarose gel electrophoreses followed by phenol extraction. 4 10⁷ Hprt negative HM-1 cells were electroporated with 100 g DNA using a Biorad gene pulser at 0.8 kV, 3 F. Cells were plated on gelatin coated dishes and selected in HAT medium containing 60% BRL conditioned growth medium, 10³ U/ml LIF, 0.1 Hypoxanthine, 0.8 M Aminopterin, 20 M Thymidine. Construct integration by homologous recombination was identified by Southern blot analysis, positive clones were expanded and then these were injected into C57BL/6 blastocysts. Heterozygous mice were obtained by breeding chimeric males with C57BL/6 females, which were then bred to obtain homozygous mice and wt littermates.

Genotyping of mice and murine Rhamm expression analysis

Although Southern analysis was initially utilized to genotype the mice, PCR primers were developed for this use. The primer sequences for amplification of the wt allele were GGC GTC TCC TAT ATG AAG AAC and GAT TTG AGT TGG CTA TTT TCA TC. The primer sequences for amplification of the Rhamm knockout allele were GCC GAG GAT TTG GAA AAA GTG and TCC AAC AAC AAA CTT GTC TGG. Both primer pairs were amplified for 40 cycles for 3 min at 94°C, 45 s at 55°C and 1 min 30 s at 72°C. Total RNA was isolated using Trizol (Gibco/BRL) for expression studies. Reverse transcriptase (Gibco/BRL) was utilized to produce cDNA, which was then amplified using PCR with specific primers for -actin, and with primers for Rhamm located in exons 3, 12 and 17. The Rhamm primer sequences were RHex3 (Rhamm exon 3): CTC AAC AAA ATC TTA ACA TTG AC; RHex12 (Rhamm exon 12): GGG CGT GAG CAG CAA TAT G; and RHex17 (Rhamm exon 17): GGA GTA AAA TTC TCC TTA GAT G. Amplification profiles were as follows: RHex3 and RHex17: 3 min 94°C, 45 s 94°C, 45 s 55°C, 1 min 72°C for 35 cycles followed by 10 min 72°C; RHex3 and RHex12: 3 min 94°C, 45 s 94°C, 45 s 55°C, 1 min 72°C for 35 cycles followed by 10 min 72°C.

Isolation, culture and immortalization of MEF

Pregnant females were killed and the wt and Rhamm-/- embryos were removed under aseptic conditions. The heads and livers of the embryos were discarded and the remaining tissue was minced to small pieces in 0.25% Trypsin, 1 mM EDTA. After 10 min incubation at 37°C, the trypsin digest was stopped by addition of growth medium (Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS)). The digested tissue was passed several times through a 1 ml sterile plastic pipette to obtain a homogenous mixture. The tissue of each embryo was plated on a 10 cm culture dish and cultured in growth medium containing an antibiotic–antimycotic (Gibco/BRL) and cells were cultured under the standard culture conditions of 37°C in a humidified 5% CO₂ atmosphere. Attached cells generally reached culture confluence at 24–48 h after isolation. Confluent cultures were trypsinized (0.25% trypsin/EDTA, Gibco/BRL), spun out of trypsin, frozen in FBS containing 10% dimethylsulfoxide (DMSO) and stored at -150°C. All experiments using cell cultures used litter-matched embryos. Spontaneously immortalized Rhamm-/- MEF cell lines were obtained from primary MEF by forcing cells into crisis as a result of subculture every 3 days (1 : 3 dilution) for 15–20 subcultures. Cells were confirmed to be Rhamm-/- by Southern blot analyses and by RT–PCR analyses of total RNA. Cells were also assessed for tumorigenic properties using growth in soft agar and foci formation in monolayer cultures as transformation indices. Neither immortalized cell line was tumorigenic by these criteria.

Retroviral rescue of immortalized Rhamm-/- MEF with fl Rhamm cDNA

Murine Rhamm cDNA encoding the fl form of this protein was cloned into the pBSF vector (provided by Dr Nolan, Stanford). Culture and transfection of packaging cells, as well as infection of immortalized MEF by ecotropic virus particles, were performed as described elsewhere (Pear et al., 1993). In brief, 2 10⁶ Phoenix packaging cells were plated on 6 cm cell culture plates. Just before transfection, chloroquine was added to the medium (final concentration 25 M). DNA (10 g) was dissolved in 439 l H₂O, 61 l 2 M CaCl₂ and mixed with 500 l 2 Hank's buffered salt solution (HBS) (280 mM NaCl, 10 mM KCl, 1.5 mM Na₂HPO₄, 50 mM HEPES, pH 7). The DNA/HBS mixture was added dropwise to packaging cell monolayers. At 24 h after transfection, the culture medium was changed to growth medium, and at 48 h after transfection, virus was harvested by filtration of supernatant medium through a 0.45 m filter. To infect immortalized Rhamm-/- MEF, the harvested virus-containing medium was diluted 1 : 2 with DMEM+10% FBS and to this was added 5 g/ml polybrene. This culture medium mixture was then added to 50% subconfluent cultures of immortalized Rhamm-/- MEF.

Analysis of Rhamm protein expression

Western analysis was performed on protein lysates from cells or tissues isolated in modified detergent buffer (RIPA buffer) composed of 1% Triton X-100, 1% Na-deoxycholate, 0.1% SDS, 50 mM HEPES pH 7.4, 150 mM NaCl, 10% Glycerol, 1.5 mM MgCl₂, 1 mM EGTA containing proteinase and phosphatase inhibitors (Protease Inhibitor Cocktail, Sigma; Phosphatase Inhibitor Cocktail II, Sigma). Mutant active Ras-transformed 10T1/2 (C3) cells which overexpress Rhamm protein (Hall et al., 1995), primary MEF or immortalized MEF isolated from wt or Rhamm-/- 12-day-old embryos were plated at 50% confluence on fibronectin-coated dishes (25 g/ml) to promote adhesion. At 24 h after plating, when cultures were still 50–60% subconfluent, monolayers were washed with cold phosphate-buffered saline (PBS) and were then lysed in modified RIPA buffer and lysates were combined with detergent insoluble material that was removed by scraping. Tissue samples were homogenized in cold lysis buffer and centrifuged at 14 000 r.p.m. for 15 min at 4°C. The supernatant protein concentration was quantified using a Bio-Rad Dc Protein Assay (Bio-Rad Laboratories). Protein lysate (10–20 g) was electrophoresed on SDS–PAGE using an 8% gel. Proteins were transferred on nitrocellulose membrane overnight at 4°C using a transfer buffer containing 39 mM glycine, 48 mM Tris base pH 8.3, 0.037% SDS and 20% methanol. After transfer, protein-binding sites on the membrane were blocked using a buffer (TBST) composed of 50 mM Tris, 150 mM NaCl pH 7.5 and 0.25% Tween 20 containing 5% fat-free milk for 2 h at room temperature (RT). Blots were then incubated with different dilutions of Rhamm antisera or antiactin antibodies in TBST containing 1% fat-free milk for 2 h at RT. Following incubation with antibody, the blots were washed 2 15 min in TBST, 1% skim milk. Blots were incubated with secondary antibody in TBST, 1% skim milk for 30 min at RT. The blots were then washed 4 20 min with TBST at room temperature and signal was detected using an ECL kit (Amersham Pharmacia Biotech).

Cells infected with pBSF-fl Rhamm or pBSF empty vector were plated onto fibronectin-coated (25 g/ml) coverslips. In all, 50% subconfluent monolayers of either primary wt and Rhamm-/- MEF (passage 5–6) or immortalized Rhamm-/- and Rhamm-rescued MEF were fixed with cold 4% paraformaldehyde/PBS (pH 7.4) for 15 min. Fixed monolayers were washed 2 5 min with cold PBS, 1 15 min with 1 M glycine (pH 7.4) and then 2 5 min with PBS. Nonspecific binding sites were blocked by incubation with 3% bovine serum albumin (BSA) in PBS for 2 h at RT. Polyclonal rabbit Rhamm antibodies prepared against sequence encoded in exons 7 and 8 (Lynn et al., 2001) were used at 1 : 2000 dilution in 1% BSA/PBS containing 0.3% Triton X-100. After 24 h incubation at 4°C, cells were washed 4 15 min with PBS followed by incubation with FITC-labeled goat anti-rabbit antibodies,1 : 200 dilution in 1% BSA/PBS and 30 min at room temperature. After 5 5 min washes in PBS, coverglasses were mounted onto standard glass microscope slides using DAKO Fluorescent mounting medium.

Crossing Rhamm-/- mice with Apc1638N mice

To determine a role for Rhamm in aggressive fibromatosis in vivo, we mated the Apc1638N mice with the mice we generated that were deficient in Rhamm. C57B1/6JIco-Apc1638N mice were generated as previously described (Fodde et al., 1994). These mice develop about 5–6 gastrointestinal lesions by 6 months of age, and male mice develop up to 45 fibromatoses by 6 months of age. The females develop a lower number of fibromatoses than male mice (Smits et al., 1998). Apc+/Apc1638N male mice were crossed with Rhamm-/- female mice. The resulting compound heterozygote Apc+/Apc1638N-Rhamm+/- mice were bred, producing Apc+/Apc1638N-Rhamm-/- and Apc+/Apc1638N-Rhamm+/+ littermate mice. Of each genotype, 20 mice (ten male and ten female) were studied. The Rhamm-deficient mice were thus compared with wt littermates. The mice were killed at 6 months of age, and the number and size of fibromatoses and gastrointestinal lesions scored as previously reported (Smits et al., 1998). Primary cell cultures from fibromatosis tumors that formed in the Rhamm-/- and wt littermates were prepared using mechanical dissociation. They were grown as a monolayer, initially in DMEM with 50% FBS added, but changed to DMEM with 10% FBS added after 5 days in culture as previously reported (Li et al., 1998). The cultures were divided when confluent, and all studies were performed within the first three passages.

Cell proliferation

Cells were seeded onto sterilized glass coverslips for 2–4 h in 10% serum containing DMEM until cells had attached and spread. This medium was then replaced with either fresh 10% medium supplemented with 10% fetal bovine serum, medium supplemented with 0.5% FBS, or defined DMEM (DMEM containing insulin (4 mg/ml) and transferrin (4 mg/ml)). Cells were left overnight and BrDu was added to the culture medium. The cells were then processed for immunofluorescence using a monoclonal antibody to BrDu as per the instructions of the manufacturer (Roche, Immunofluorescence kit for BrDu incorporation). For sparse cultures, 1 10⁴ cells were added per well of a 24-well plate and 4 10⁴ per well for denser cultures.

References

1. Alman BA, Li C, Pajerski ME, Diaz-Cano S and Wolfe HJ. (1997b). Am. J. Pathol., 151, 329–334. | PubMed | ISI | ChemPort |
2. Alman BA, Pajerski ME, Diaz-Cano S, Corboy K and Wolfe HJ. (1997a). Diag. Mol. Pathol., 6, 98–101. | Article | ChemPort |
3. Assmann V, Jenkinson D, Marshall JF, Hart IR. (1999). J. Cell. Sci., 112, 3943–3954. | PubMed | ISI | ChemPort |
4. Assmann V, Gillett CE, Poulsom R, Ryder K, Hart IR and Hanby AM. (2001). J. Pathol., 195, 191–196. | Article | PubMed | ChemPort |
5. Bissell MJ and Radisky D. (2001). Nat. Rev. Cancer, 1, 46–54. | Article | PubMed | ChemPort |
6. Cheon SS, Cheah AY, Turley S, Nadesan P, Poon R, Clevers H and Alman BA. (2002). Proc. Natl. Acad. Sci. USA, 99, 6973–6978. | Article | PubMed | ChemPort |
7. Cheung WF, Cruz TF and Turley EA. (1999). Biochem. Soc. Trans., 27, 135–142. | PubMed | ISI | ChemPort |
8. Crainie M, Belch AR, Mant MJ and Pilarski LM. (1999). Blood, 93, 1684–1696. | PubMed | ISI | ChemPort |
9. Duval A, Rolland S, Compoint A, Tubacher E, Iacopetta B, Thomas G and Hamelin R. (2001). Hum. Mol. Genet., 10, 513–518. | Article | PubMed | ISI | ChemPort |
10. Fodde R, Edelmann W, Yang K, van Leeuwen C, Carlson C, Renault B, Breukel C, Alt E, Lipkin M and Khan PM. (1994). Proc. Natl. Acad. Sci. USA, 91, 8969–8973. | Article | PubMed | ChemPort |
11. Gares SL and Pilarski LM. (2000). Dev. Immunol., 7, 209–225. | PubMed |
12. Gerdin B and Hallgren R. (1997). J. Intern. Med., 242, 49–55. | Article | PubMed | ISI | ChemPort |
13. Gutierrez-Ruiz MC, Robles-Diaz G and Kershenobich D. (2002). Arch. Med. Res., 33, 595–599. | PubMed |
14. Hall CL, Lange LA, Prober DA, Zhang S and Turley EA. (1996). Oncogene, 13, 2213–2224. | PubMed | ISI | ChemPort |
15. Hall CL, Yang B, Yang X, Zhang S, Turley M, Samuel S, Lange LA, Wang C, Curpen GD, Savani RC, Greenberg AH and Turley EA. (1995). Cell, 82, 19–26. | Article | PubMed | ISI | ChemPort |
16. Hardwick C, Hoare K, Owens R, Hohn HP, Hook M, Moore D, Cripps V, Austen L, Nance DM and Turley EA. (1992). J. Cell. Biol., 117, 1343–1350. | Article | PubMed | ISI | ChemPort |
17. Hopyan S, Gokgoz N, Bell RS, Andrulis IL, Alman BA and Wunder JS. (1999). J. Orthop. Res., 17, 633–638. | Article | PubMed | ISI | ChemPort |
18. Kong QY, Liu J, Chen XY, Wang XW, Sun Y and Li H. (2003). Oncol. Rep., 10, 51–55. | PubMed | ISI | ChemPort |
19. Li C, Bapat B and Alman BA. (1998). Am. J. Pathol., 153, 709–714. | PubMed | ISI | ChemPort |
20. Li H, Guo L, Li JW, Liu N, Qi R and Liu J. (2000). Int. J. Oncol., 17, 927–932. | PubMed | ISI | ChemPort |
21. Line A, Slucka Z, Stengrevics A, Silina K, Li G and Rees RC. (2002). Cancer Immunol. Immunother., 51, 574–582. | Article | PubMed | ChemPort |
22. Lokeshwar VB and Selzer MG. (2000). J. Biol. Chem., 275, 27641–27649. | PubMed | ISI | ChemPort |
23. Lovvorn III HN, Cass DL, Sylvester KG, Yang EY, Crombleholme TM, Adzick NS and Savani RC. (1998). J. Pediatr. Surg., 33, 1062–1069. | Article | PubMed | ISI |
24. Lynn BD, Li X, Cattini PA, Turley EA and Nagy JI. (2001). J. Comp. Neurol., 439, 315–330. | Article | PubMed | ChemPort |
25. Miller JR, Hocking AM, Brown JD and Moon RT. (1999). Oncogene, 18, 7860–7872. | Article | PubMed | ISI | ChemPort |
26. Mohapatra S, Yang X, Wright JA, Turley EA and Greenberg AH. (1996). J. Exp. Med., 183, 1663–1668. | Article | PubMed | ISI | ChemPort |
27. Naor D, Sionov RV, Zahalka M, Rochman M, Holzmann B and Ish-Shalom D. (1998). Curr. Top. Microbiol. Immunol., 231, 143–166. | PubMed |
28. Pear W, Nolan GP, Scott M and Baltimore D. (1993). Proc. Natl. Acad. Sci. USA, 90, 8392–8396. | Article | PubMed | ChemPort |
29. Pilarski LM, Masellis-Smith A, Belch AR, Yang B, Savani RC and Turley EA. (1994). Leuk. Lymphoma, 14, 363–374. | PubMed |
30. Ponta H, Sherman L and Herrlich PA. (2003). Nat. Rev. Mol. Cell. Biol., 4, 33–45. | Article | PubMed | ISI | ChemPort |
31. Poon R, Smits R, Li C, Jagmohan-Changur S, Kong M, Cheon S, Yu C, Fodde R and Alman BA. (2001). Oncogene, 20, 451–460. | Article | PubMed | ChemPort |
32. Roose J and Clevers H. (1999). Biochim. Biophys. Acta, 1424, M23–M37. | Article | PubMed | ISI | ChemPort |
33. Savanger P. (2001). BioEssays, 23, 912–923. | Article | PubMed | ISI | ChemPort |
34. Savani RC, Cao G, Pooler PM, Zaman A, Zhou Z and DeLisser HM. (2001). J. Biol. Chem., 276, 36770–36778. | Article | PubMed | ISI | ChemPort |
35. Savani RC, Wang C, Yang B, Zhang S, Kinsella MG, Wight TN, Stern R, Nance DM and Turley EA. (1995). J. Clin. Invest., 95, 58–68.
36. Sivaraman VS, Wang H, Nuovo GJ and Malbon CC. (1997). J. Clin. Invest., 99, 1478–1483. | PubMed | ISI | ChemPort |
37. Smits R, van der Houven van Oordt W, Luz A, Zurcher C, Jagmohan-Changur S, Bruekel C, Khan PM and Fodde R. (1998). Gastroenterology, 114, 275–283. | Article | PubMed | ISI | ChemPort |
38. Spicer AP, Roller ML, Camper SA, McPherson JD, Wasmuth JJ, Hakim S, Wang C, Turley EA and McDonald JA. (1995). Genomics, 30, 115–117. | PubMed |
39. Tammi MI, Day AJ and Turley EA. (2002). J. Biol. Chem., 277, 4581–4584. | Article | PubMed | ISI | ChemPort |
40. Tejpar S, Li C, Yu C, Poon R, Denys H, Sciot R, Van Cutsem E, Cassiman JJ and Alman BA. (2001). Br. J. Cancer, 85, 98–101. | Article | PubMed | ISI | ChemPort |
41. Tejpar S, Nollet F, Li C, Wunder JS, Michils G, dal Cin P, Van Cutsem E, Bapat B, van Roy F, Cassiman JJ and Alman BA. (1999). Oncogene, 18, 6615–6620. | Article | PubMed | ISI | ChemPort |
42. Toole BP. (2002). Glycobiology, 12, 37R–42R. | Article | PubMed | ISI | ChemPort |
43. Turley EA. (1982). Biochem. Biophys. Res. Commun., 188, 1016–1024.
44. Turley E and Auersperg N. (1989). Exp. Cell. Res., 182, 340–348. | PubMed |
45. Turley EA, Austen L, Vandeligt K and Clary C. (1991). J. Cell. Biol., 112, 1041–1047. | Article | PubMed | ISI | ChemPort |
46. Turley EA, Bowman P and Kytryk MA. (1985). J. Cell. Sci., 78, 133–145. | PubMed |
47. Turley EA, Brassel P and Moore D. (1990). Exp. Cell Res., 187, 243–249. | PubMed |
48. Turley EA, Noble PW and Bourguignon LY. (2002). J. Biol. Chem., 277, 4589–4592. | Article | PubMed | ISI | ChemPort |
49. van der Houven van Oordt CW, Smits R, Schouten TG, Houwing-Duistermaat JJ, Williamson SL, Luz A, Meera Khan P, Van der Eb AJ, Breuer ML and Fodde R. (1999). Genes Chromosom. Cancer, 24, 191–198. | PubMed |
50. van der Lugt N, Maandag ER, te Riele H, Laird PW and Berns A. (1991). Gene, 105, 263–267. | PubMed | ChemPort |
51. Wang C, Thor AD, Moore II DH, Zhao Y, Kerschmann R, Stern R, Watson PH and Turley EA. (1998). Clin. Cancer Res., 4, 567–576. | PubMed | ChemPort |
52. Wielenga VJ, Smits R, Korinek V, Smit L, Kielman M, Fodde R, Clevers H and Pals ST. (1999). Am. J. Pathol., 154, 515–523. | PubMed | ISI | ChemPort |
53. Wong NA and Pignatelli M. (2002). Am. J. Pathol., 160, 389–401. | PubMed | ISI | ChemPort |
54. Yang B, Yang BL, Savani RC and Turley EA. (1994). EMBO J., 13, 286–296. | PubMed | ISI | ChemPort |
55. Zhang S, Chang MC, Zylka D, Turley S, Harison R and Turley EA. (1998). J. Biol. Chem., 273, 11342–11348. | Article | PubMed | ISI | ChemPort |
56. Zhou R, Wu X and Skalli O. (2002). J. Neurooncol., 59, 15–26. | PubMed |

Acknowledgements

This work is funded by Grants from the National Cancer Institute of Canada (BAA) and Canadian Institutes of Health Research (EAT). ET is supported by the Pamela GreenawayKohlmeier Translation Breast Cancer Unit Salary Award and BAA is supported by the Canadian Research Chairs Program.