2HS-Glycoprotein, an Antagonist of Transforming Growth Factor in Vivo

[CANCER RESEARCH 64, 6402–6409, September 15, 2004] ␣2HS-glycoprotein, an Antagonist of Transforming Growth Factor ␤ In vivo, Inhibits Intestinal Tumor Progression

Carol J. Swallow,1,2 Emily A. Partridge,1 Jennifer C. Macmillan,1 Tania Tajirian,1 Gianni M. DiGuglielmo,1 Kazy Hay,1 Melanie Szweras,1 Willi Jahnen-Dechent,3 Jeff L. Wrana,1,2 Mark Redston,1 Steven Gallinger,1,2 and James W. Dennis1,2 1Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada; 2Departments of Molecular and Medical Genetics and Surgery, University of Toronto, Toronto, Ontario, Canada; and 3Interdisziplina¨ren Zentrums fu¨r Klinische Forschung BIOMAT, Klinikum der Rheinisch-Westfa¨lischen Technischen Hoschschule, Aachen, Aachen, Germany

ABSTRACT ␤1-induced epithelial-mesenchymal transition is characterized by down-regulation of E-cadherin expression, loss of tight junctions, and ␤ Transforming growth factor (TGF)- 1 is associated with tumor pro- increased cell motility (1, 10) and is dependent on phosphatidylinosi- gression and resistance to chemotherapy in established cancers, as well as tol 3Ј-kinase/Akt activation (11), a key pathway promoting tumor cell host immune suppression. Here, we show that the serum glycoprotein ␣2-HS-glycoprotein (AHSG) blocks TGF-␤1 binding to cell surface re- invasiveness (12). Epithelial-mesenchymal transition appears to be a ceptors, suppresses TGF-␤ signal transduction, and inhibits TGF-␤- close in vitro correlate of metastatic capacity, and both require coop- induced epithelial-mesenchymal transition, suggesting that AHSG may erative signaling of TGF-␤ and Ras/mitogen-activated protein kinase play a role in tumor progression. In 66 consecutive sporadic human pathways (13, 14). Tumor-derived TGF-␤1 also has paracrine effects colorectal cancer specimens, we observed a 3-fold depletion of ASHG in on host cells that promote angiogenesis, alter matrix turnover, and tumor compared with normal tissue, whereas levels of other abundant suppress immune cell functions (1, 7, 15, 16). Furthermore, mice plasma proteins, albumin and transferrin, were equivalent. Using the treated with neutralizing anti-TGF-␤ antibody or with the competitive -Multiple intestinal neoplasia/؉ (Min/؉) mouse model of intestinal tumor inhibitor protein Fc:T␤RII and mice expressing dominant negative igenesis, we found twice as many intestinal polyps overall, twice as many TGF-␤ type II receptor (T␤RII) display suppression of tumor growth, large polyps (>3 mm diameter), and more progression to invasive adeno- .(carcinoma in Min/؉ AhsgϪ/Ϫ mice than in littermates expressing Ahsg. invasion, and/or metastasis (17–21 Phosphorylated Smad2 was more abundant in the intestinal mucosa and TGF-␤/bone morphogenetic protein cytokines are present in most tumors of Min/؉ mice lacking Ahsg, demonstrating increased TGF-␤ tissues but their availability to signaling receptors is regulated by signaling in vivo. Furthermore, TGF-␤-mediated suppression of immune soluble and matrix-associated binding proteins. During embryogene- Ϫ Ϫ cell function was exaggerated in Ahsg / animals, as shown by inhibition sis, chordin, noggin, twisted, and follistatin regulate local cytokine of macrophage activation and reduction in 12-O-tetradecanoylphorbol concentrations and thereby morphogenic activity (22, 23). In adult Ϫ/Ϫ 13-acetate–induced cutaneous inflammation. Reconstitution of Ahsg tissues, TGF-␤-binding proteins limit cytokine activity, thereby con- mice with bovine Ahsg suppressed endogenous TGF-␤-dependent signal- trolling tissue remodeling and inflammation. For example, TGF-␤ ing to wild-type levels, suggesting that therapeutic enhancement of AHSG ␣ levels may benefit patients whose tumors are driven by TGF-␤. latency-associated peptide and 2-macroglobulin regulate inflammation and response to infection (24, 25). ␣2-HS-glycoprotein (Ahsg)/ fetuin is a glycoprotein present in serum and extracellular matrix that INTRODUCTION binds to TGF-␤ cytokines (bone morphogenetic proteins 2, 4, and 6 and TGF-␤1 and TGF-␤2) via a 19 amino acid cytokine-binding Transforming growth factor (TGF)-␤ cytokines are conserved in domain homologous to an extracellular sequence in the T␤RII (26). metazoans and function as morphogens that regulate cell proliferation Ahsg antagonized binding of TGF-␤1toT␤RII in surface plasmon and differentiation during embryogenesis. In postnatal life, TGF-␤ resonance assays and inhibited the effect of TGF-␤1 on mink lung cytokines regulate tissue remodeling, wound repair, and immune cell epithelial cell proliferation, rat bone marrow cell osteogenic differen- functions. TGF-␤1 inhibits normal epithelial cell proliferation, and in tiation (27), and human monocyte matrix metalloproteinase 9 release a subset of epithelial malignancies, inactivating mutations in genes of (28). Ahsg and TGF-␤1 are also concentrated in mineralized bone, the signaling pathway occur early, presumably allowing expansion of Ϫ/Ϫ premalignant cell populations (1–4). However, in established human and Ahsg mice display a bone phenotype characterized by in- cancers, elevated TGF-␤1 expression correlates with progression and creased trabecular bone remodeling, osteoblast density, femur thick- adverse outcome (5–7). Transgenic mice that express TGF-␤1in ness, and bone mineral density (29). keratinocytes exhibit both reduced incidence of skin tumor initiation These observations suggest that Ahsg might inhibit cancer progres- ␤ and enhanced malignant progression, exemplifying the disparate ef- sion in vivo both by blocking TGF- 1 activity in tumor cells and by ␤ fects of TGF-␤1 at early and late stages in tumorigenesis (8). Simi- opposing TGF- -dependent immunosuppression. Here, we demon- ␤ ␤ larly, loss of TGF-␤ responsiveness promotes carcinogenesis in non- strate that AHSG inhibits TGF- 1 binding to cell surface TGF- malignant human breast-derived cell lines but inhibits metastatic receptors and blocks phosphorylation and nuclear translocation of ␤ behavior in genetically related high-grade malignant cells (9). Smad2/3. Ahsg inhibited signaling by endogenous TGF- in colon TGF-␤ signaling promotes the epithelial-mesenchymal transition, a cancer cells and blocked induction of epithelial-mesenchymal transi- morphogenic event governing cell fate during embryogenesis. TGF- tion by TGF-␤1, suggesting that its depletion in vivo could promote cancer cell autonomous invasiveness. Indeed, in compound mutant ϩ Received 3/29/04; revised 5/20/04; accepted 7/16/04. Min/ mice of three Ahsg genotypes, we found that the presence of Grant support: Canadian Institutes for Health Research (J. Dennis) and the Physi- Ahsg inhibited intestinal tumor progression, reducing polyp number, cians Services Incorporated (C. Swallow). size, and invasion. Compared with Ahsg wild-type Min/ϩ mice, The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with littermates lacking Ahsg displayed increased phosphorylated Smad2 18 U.S.C. Section 1734 solely to indicate this fact. in intestinal epithelium and tumors, indicating enhanced TGF-␤ sig- Requests for reprints: Carol J. Swallow, Mount Sinai Hospital, 600 University Ϫ/Ϫ Avenue, #1224, Toronto, Ontario, M5G 1X5 Canada. E-mail: [email protected]. naling in vivo. Treatment of Ahsg mice with exogenous Ahsg ©2004 American Association for Cancer Research. restored normal TGF-␤ signaling, as demonstrated in freshly har- 6402

vested peritoneal macrophages. We additionally show that ASHG immunohistochemical analysis of deparaffinized sections of proximal small levels are significantly and selectively reduced in tumor tissue com- intestine, according to the manufacturer’s suggested protocol. Photographs of pared with adjacent normal mucosa in human colorectal cancer. These three crypts, three villi, and two to four polyps per mouse were taken under oil results indicate that AHSG repletion holds promise as a treatment immersion using identical charge-coupled device settings by an observer strategy to combat epithelial cancer progression. blinded to the genotype, and luminosity of nuclei was determined using Adobe Photoshop. In total, three random groups of five nuclei were assessed for each crypt, each villus, and each polyp to yield a mean nuclear luminosity score for MATERIALS AND METHODS each. This was subtracted from the mean background luminosity for each slide, which was measured and calculated similarly, to yield a nuclear intensity Human Tissue Specimens. Sixty-six sequential sporadic colorectal cancer score. A mean crypt, villus, and polyp nuclear intensity score was then specimens from our tumor bank were analyzed (Table 1). Median patient age calculated for each mouse. was 70 years (range, 22 to 96 years). None of the patients had received To minimize platelet degranulation, whole blood obtained by cardiac punc- preoperative radiation or chemotherapy. Samples of colorectal tumor tissue ture was allowed to clot at room temperature for 1 hour, centrifuged at 3000 and paired adjacent normal colonic mucosa were obtained fresh at the time of rpm, and serum samples collected. Total serum TGF-␤1 was determined by resection and snap frozen in liquid nitrogen or fixed in 10% formalin. Normal ELISA (R&D Systems, Minneapolis, MN), using acid activation. For Western liver samples were handled similarly. blot analysis of serum Ahsg, 10 ␮L of serum (diluted 1:10 with PBS) were Frozen tissue samples were homogenized, and protein samples (10, 1, and loaded per well, separated by SDS-PAGE, and Ahsg identified using mouse ␮ 20 g for detection of AHSG, albumin, and transferrin, respectively) were antihuman AHSG monoclonal antibody (30), which reacts with both mouse separated by SDS-PAGE. Proteins were identified with mouse or goat antihu- and bovine Ahsg. For some experiments, mice received i.p. injections of man primary antibody followed by application of secondary sheep antimouse bovine Ahsg (Sigma), 3 mg in 1 mL of PBS, 72 and 24 hours before sacrifice. antibody or rabbit antigoat antibody. All antibodies were purchased from All animal protocols were reviewed and approved by the Samuel Lunenfeld Sigma (St. Louis, MO), except mouse antihuman AHSG, which was prepared Research Institute Institutional Review Board. as described previously (30). The enhanced chemiluminescence photodetection Peritoneal Macrophage Preparation and Analysis. Except where indi- system was used and densitometric analysis (Imagequant Iq) performed. For cated, mice were injected intraperitoneally with 25 ␮g lipopolysaccharide each protein, three separate blots of tumor and normal mucosa protein extracts (from Escherichia coli serotype 026:B6; Sigma) in 1 mL of PBS 5 days before were analyzed, and the final normal/tumor (N/T) ratio calculated was the mean sacrifice. The peritoneal cavity was lavaged with 10 mL of sterile ice-cold of these three. Use of human tissue specimens was approved by the Mount Ca2ϩ- and Mg2ϩ-free PBS, lavage fluid centrifuged at 4°C at 1100 rpm, and Sinai Hospital Institutional Review Board. pellets resuspended in serum-free macrophage specific medium (Invitrogen Mice. Ahsg-deficient mice were generated by targeted gene mutation in ES Life Technologies, Burlington, Ontario, Canada) with penicillin and strepto- cells that removed the entire coding sequence of the gene, as described mycin. The cell suspension was plated onto wells or coverslips, incubated for previously (30). AhsgϪ/Ϫ mice of 129/Sv ϫ C57BL6 background were crossed 2hat37°Cin5%CO, washed vigorously with warm PBS, and fresh with C57BL/6 mice; female Ahsgϩ/Ϫ offspring were bred with male C57BL/6 2 serum-free macrophage-specific medium added. For measurement of nitric Min/ϩ mice from a colony established at our institution. Min/ϩAhsgϩ/Ϫ oxide release and tumor necrosis factor ␣, macrophages in 12-well plates were offspring were then crossed to produce a F2 generation of Min/ϩ mice of three incubated for an additional 48 hours in serum-free macrophage-specific me- Ahsg genotypes, which were obtained at the expected Mendelian frequency. dium Ϯ human AHSG or Ϯ TGF-␤1 (R&D Systems). Supernatants were Littermates were maintained on standard chow, weighed weekly, aged to the collected, nitrites quantitated using the Griess method (Sigma), tumor necrosis indicated time points, and sacrificed by CO inhalation. The small and large 2 factor ␣ by ELISA assay (R&D Systems), and both normalized to protein intestines were harvested, washed with PBS, split longitudinally, preserved in content in each well (BCA assay). For immunofluorescence assays, macro- formalin, and examined for lesions as previously described (31). No mice phages on coverslips were fixed with 4% formaldehyde ϫ 15 minutes, washed required sacrifice before 180 days. One Min/ϩ Ahsgϩ/ϩ mouse, one Min/ϩ with PBS, permeabilized with 100% methanol ϫ 2 minutes, washed, and Ahsgϩ/Ϫ mouse, and two Min/ϩ AhsgϪ/Ϫ mice developed wasting symptoms incubated with mouse antihuman AHSG monoclonal antibody (30) at requiring sacrifice before 340 days; these mice were eliminated from subse- 1:100 ϫ 2 hours, followed by Cy3-labeled sheep antimouse antibody (1:50) quent analysis. An observer blind to the genotype of the mice examined the plus Hoechst (1:1000) ϫ 1 hour. intestinal segments. Tumor number and location and tumor diameter to a Skin Inflammation Assay. 12-O-Tetradecanoylphorbol 13-acetate (2 ␮g precision of 0.1 mm in 340-day-old mice were scored using a dissecting in 25 ␮L of ethanol) was applied to the right ear and ethanol alone to the left microscope as described previously (31). A second observer verified tumor ear of each mouse. Ear swelling was measured with a micrometer at 24-hour counts in a subset of samples. For histologic analysis, sections of the proximal intervals and reported as the difference in thickness between the right and left small intestine were fixed in 10% buffered formalin, embedded in paraffin, ϩ ϩ Ϫ Ϫ ears. Five mice of each genotype (Ahsg / and Ahsg / ) were tested. sectioned, and stained with H&E. Polyp histology was analyzed by a pathol- TGF-␤ Receptor Binding, Smad2/3 Nuclear Translocation, and Phos- ogist blinded to the genotype of the samples. Phospho-Smad2 (Ser465/467) pho/Total Smad2 Analysis. TGF-␤1 cross-linking to cell surface receptor antibody (1:1500; Cell Signaling Technology, Inc., Beverly, MA) was used for was performed as described previously (32). Briefly, MvLu1 cells were incubated for 30 minutes at 37°C in Krebs-Ringer-HEPES (KRH) plus 0.5% BSA and placed on ice. 125Iodine-labeled TGF-␤1 at 250 pmol/L was preincubated Table 1 Clinical characteristics and AHSG N/T levels in 66 colorectal cancer patients for 30 minutes at 20°C with Ahsg or transferrin, then added to cells and AHSG N/T AHSG N/T incubated for 2 hours at 4°C with agitation. The cells were then incubated with Category n ratio* ratio Ͼ 1 (%) DSS (60 ␮g/mL) for 15 minutes at 4°C. Cells lysates were centrifuged, and the Sex 125 ␤ Ϯ supernatants subjected to SDS-PAGE. I-TGF- 1 bound to proteins corre- Male 40 2.94 0.60 75 ␤ ␤ ␤ Female 26 2.43 Ϯ 0.56 85 sponding in size to T RIII, T RII, and T RI, and immunoprecipitation of 125 Age I-TGF-␤1 receptor complexes with anti-T␤R antibodies confirmed their Ͻ65 yrs 20 2.26 Ϯ 0.44 75 identity. Ն Ϯ 65 yrs 46 2.94 0.58 83 Smad2/3 localization was measured in MvLu1, SW620 cells and peritoneal Location Right/Transverse 28 2.42 Ϯ 0.54 79 macrophages plated in 96-well plates at a density of 5000 cells/well, and serum Left/Sigmoid 21 2.22 Ϯ 0.66 67 starved for 24 h. TGF-␤1 was preincubated with human AHSG (Calbiochem), Rectum 17 3.89 Ϯ 1.11 94 bovine Ahsg or human transferrin (Sigma) for 30 minutes at 37°C, then added Stage to cells for 30 min. For experiments assessing the effect of endogenous TGF-␤, I 7 3.65 Ϯ 1.62 71 II 25 2.17 Ϯ 0.59 64 Ahsg alone was added. The cells were washed with warm PBS, fixed for 10 III 20 3.70 Ϯ 1.00 95 minutes with 3.7% formaldehyde at 37°C, then washed with PBS plus 1% IV 14 2.73 Ϯ 0.37 86 FBS, and permeabilized with methanol for 2 minutes. PBS plus 10% FBS was * Mean Ϯ SE. added and left overnight at 4°C. Mouse anti-Smad2/3 antibody (BD Trans- 6403

Downloaded from cancerres.aacrjournals.org on September 24, 2021. © 2004 American Association for Cancer Research. Ahsg AND INTESTINAL NEOPLASIA duction Laboratories, Mississauga, Ontario, Canada) was added at 1:200 in tirabbit IgG conjugated to horseradish peroxidase (Amersham Biosciences, PBS plus 10% FBS for1hat20°C. After washing three times with PBS plus Inc.), and blots developed using the Renaissance chemiluminescence kit 1% FBS, Alexa Fluor 488 goat antimouse immunoglobulin (Molecular Probes) (NEN-PerkinElmer, Boston, MA). was added at 1:200 with Hoechst ϫ 1 hour at 20°C. The plates were scanned E-Cadherin Staining. NMuMG cells were plated at 30% confluence and using the Scan Array automated fluorescence microscope and cytoplasmic- cultured in DMEM for 48 hours with 10 ␮g/mL insulin Ϯ TGF-␤1 (200 nuclear software (Cellomics, Inc., Pittsburgh, PA). The difference in nuclear- pmol/L) and/or Ahsg (20 ␮mol/L). E-Cadherin was detected with mouse cytoplasmic staining intensity was determined for 100 cells per well, generat- monoclonal antibody (500 ng/mL ϫ 2 hours; Transduction Laboratories), ing a mean Ϯ SE for each condition. Experiments were performed in triplicate. followed by Alexa Fluor 488 goat antimouse immunoglobulin (2 ␮g/mL ϫ 1 For analysis of total and phosphorylated Smad2, SW620 cells at 75% hour; Molecular Probes) and imaged using a deconvolution microscope. confluence were washed three times with PBS, preincubated without or with Statistical Analysis. All values are reported as mean Ϯ SE. Statistical Ahsg (20 ␮mol/L) in serum-free RPMI medium at 37°C for 30 minutes, after analysis to compare means was by ANOVA or Mann-Whitney when data were which, 200 pmol/L TGF-␤1 was added for the indicated time, cells washed or were not distributed normally, respectively, and proportions were compared three times with PBS, and lysed with Tris-NaCL-Triton X 100-EDTA (TNTE) by ␹2 test. Statistical significance was established at P ϭ 0.05 (two-sided). buffer with protease inhibitor mixture tablets (Roche Diagnostics, Mannheim, Germany) at 4°C ϫ 20 minutes. Protein lysates were diluted with TNT buffer, RESULTS immunoprecipitated with anti-Smad2/3 antibody (N-19; Santa Cruz Biotech- nology, Santa Cruz, CA) on ice ϫ 3 hours, shaken with a 1:5 slurry of protein Ahsg Competes for TGF-␤1 Binding to Cell Surface Receptors G-Sepharose beads (Pharmacia, Uppsala, Sweden) in Tris-NaCL-Tween and Blocks Signaling. The disulfide-looped sequence in Ahsg from (TNT) buffer at 4°C ϫ 1 hour, centrifuged, washed, resuspended in loading Cys114 to Cys130 (bovine sequence) shares homology with the extra- buffer with ␤-mercaptoethanol, separated by SDS-PAGE, and transferred to ␤ 84 101 polyvinylidene difluoride membranes. Blots were incubated at 4°C overnight cellular domain of T RII (Cys to Cys ). In surface plasmon ␤ with anti-Smad2/3 antibody at a 1:2000 dilution (BD Transduction Laborato- resonance assays, these peptides bind to TGF- and bone morphoge- ries) or anti-phospho-Smad2 antibody at a 1:1000 dilution (Upstate Biotech- netic protein cytokines with specificity characteristic of native Ahsg nology, Inc., Lake Placid, NY), followed by antimouse IgG conjugated to and T␤RII, suggesting these are the major cytokine-binding domains horseradish peroxidase (Amersham Biosciences, Inc., Piscataway, NJ) or an- in both glycoproteins (26). Because Ahsg binds to TGF-␤1 with a

Fig. 1. Ahsg inhibits TGF-␤1 binding to cell surface receptors and blocks Smad2/3 nuclear translocation. A, inhibition of TGF-␤1 binding to cell surface receptors by Ahsg. 125I-labeled TGF-␤1 was preincubated with either Ahsg or transferrin, added to MvLu1 cells, chemically cross-linked, and protein extracts separated by SDS-PAGE. The position of 125I-TGF-␤1 in complex with T␤RIII, T␤RII, and T␤RI is indicated by the arrows. The four lanes on the right represent transferrin controls. B, competition by Ahsg (F, f, Œ), or transferrin (E, Ⅺ, ‚), for 125I-TGF-␤1 binding to T␤RIII (Œ, ‚), T␤RII (f, Ⅺ), and T␤RI (F, E) quantified by densitometry. C, localization of Smad2/3 in MvLu1 cells treated with AHSG (20 ␮mol/L), TGF-␤1 (200 pmol/L), or both for 30 minutes, imaged by Scan Array automated fluorescence microscopy. D, nuclear-cytoplasmic Smad2/3 intensity difference in MvLu1 cells treated with 200 pmol/L TGF-␤1 plus varying concentrations of AHSG (f) or transferrin (Œ) or treated with AHSG alone (E), measured by Scan Array. Data are means Ϯ SE for 100 cells. E, nuclear-cytoplasmic Smad2/3 intensity difference in MvLu1 cells treated with varying doses of TGF-␤1 in the presence of increasing concentrations of AHSG arrayed in two dimensions in a 96-well plate. Data are means Ϯ SE for 100 cells per well.

6404

Downloaded from cancerres.aacrjournals.org on September 24, 2021. © 2004 American Association for Cancer Research. Ahsg AND INTESTINAL NEOPLASIA ϳ ␮ dissociation constant (KD)of 2.0 mol/L (26) and serum AHSG concentration is ϳ12 ␮mol/L, the laws of mass action indicate that Ahsg should influence cytokine availability in vivo. To determine whether Ahsg competes for TGF-␤1 binding to cell surface signaling receptors, MvLu1 cells were incubated with 125I-labeled TGF-␤1 and a chemical cross-linking agent to reveal the ligand-receptor complexes (Fig. 1A). Preincubation of 125I-TGF-␤1 with increasing concentrations of Ahsg resulted in dose-dependent inhibition of cytokine binding to T␤RIII, T␤RII, and T␤RI, whereas the control glycoprotein transferrin had no effect (Fig. 1, A and B). Ahsg reduced 125I- ␤ TGF- 1 cross-linking to receptors with an estimated IC50 for the individual receptors of 20 to 30 ␮mol/L (Fig. 1B). TGF-␤1 binding to T␤RII stimulates transphosphorylation of T␤RI, which in turn phosphorylates the RSmads Smad2/Smad3. The latter heterodimerize with Smad4 and translocate into the nucleus where in complexes with other proteins, they regulate gene transcrip- tion. Anti-Smad2/3 antibody was used to probe the translocation of protein from the cytoplasm to the nucleus after TGF-␤1 stimulation in MvLu1 cells (Fig. 1C). Preincubation of TGF-␤1 with either bovine Ahsg or human AHSG blocked the TGF-␤1-induced nuclear translocation of Smad2/3 in MvLu1 cells, whereas transferrin protein had no ␮ effect (Fig. 1, C and D). The IC50 for AHSG was 10 mol/L when TGF-␤1 was added at 200 pmol/L, a concentration 10-fold greater

than the D50 for maximal signaling (Fig. 1, D and E). Furthermore, the competitive interaction between TGF-␤1 and AHSG was observed over a range of concentrations for both proteins (Fig. 1E). To determine whether Ahsg can regulate TGF-␤ signaling in malignant cells, we examined the SW620 human colorectal cancer line, which releases ϳ4 ng of TGF-␤1 per 106 cells in 24 hours when cultured in serum-free medium. Ahsg suppressed Smad2 phosphorylation induced by both endogenous and exogenous TGF-␤ (Fig. 2A). At concentrations in the physiologic range, Ahsg suppressed the basal level of Smad2/3 nuclear localization in SW620 cancer cells (Fig. 2B). Ahsg also blocked TGF-␤1-induced loss of E-cadherin at cell tight junctions, a hallmark of the epithelial-mesenchymal transition phenotype associated with progression of epithelial tumors to an invasive Fig. 2. AHSG inhibits signaling by endogenous TGF-␤ and blocks TGF-␤1-induced phenotype (Fig. 2C). These results suggest that Ahsg has the potential epithelial-mesenchymal transition. A, total Smad2/3 and phosphorylated Smad2, assessed ␤ by Western blot, in protein lysates prepared from SW620 colorectal cancer cells incubated to block TGF- 1-driven tumor progression in vivo. without (control) or with Ahsg (20 ␮mol/L) plus 200 pmol/L TGF-␤1 for the indicated Depletion of AHSG in Human Colorectal Cancer. Tumor pro- time. B, nuclear-cytoplasmic Smad2/3 intensity difference in SW620 colorectal cancer motion by autocrine TGF-␤ might be enhanced by selective depletion cells incubated with Ahsg at the indicated concentration for 48 hours. C, E-cadherin localization in NMuMG cells incubated for 48 hours Ϯ TGF-␤1 (200 pmol/L) Ϯ Ahsg (20 of TGF-␤ antagonist proteins in the tumor microenvironment. To ␮mol/L), examined by confocal fluorescence microscopy. explore this possibility, we measured the relative abundance of AHSG in tumor versus paired normal mucosa in 66 consecutive sporadic colorectal cancer specimens from our tumor bank (Fig. 3, A and B). Ͼ1 in 65.6% of stage I/II versus 91.1% of stage III/IV cancers (n ϭ 32 The AHSG transcript was below the level of detection in tumor and and 34, respectively, P ϭ 0.03 by ␹2), suggesting an association of normal mucosa but readily detected in liver by Northern blotting and AHSG depletion with tumor progression. Immunohistochemical anal- by in situ RNA hybridization, confirming that tissue AHSG was ysis of AHSG in tissue sections revealed staining in the extracellular derived largely from serum. As a control for serum-derived proteins in matrix of the mucosal and submucosal layers of the bowel wall in normal and tumor tissues, albumin and transferrin content were as- normal colon, but tumor tissue stained only weakly (Fig. 3C). sessed by the same method (Fig. 3, A and B). The mean normal Ahsg Inhibits Intestinal Tumor Progression. The progressive mucosa:tumor ratio (N/T) for AHSG was significantly greater than loss of AHSG in late-stage tumors suggests that this TGF-␤ antagonist that for transferrin or albumin (2.73 Ϯ 0.42 versus 1.11 Ϯ 0.05 and may oppose intestinal tumor progression in vivo. To test this hypoth- 1.13 Ϯ 0.05, respectively, P ϭ 0.001; Fig. 3B). In addition, the mean esis, adenomatous intestinal tumors were studied in Min/ϩAhsgϩ/ϩ, compound N/T ratios for AHSG:transferrin and AHSG:albumin were Min/ϩAhsgϩ/Ϫ, and Min/ϩAhsgϪ/Ϫ mice. Min/ϩ mice have a loss- both significantly Ͼ1 (2.71 and 2.43, respectively, P ϭ 0.001). These of-function point mutation in one allele of the Apc gene, and with results indicate that AHSG is selectively depleted in tumor tissue stochastic loss of the wild-type allele, they develop multiple small when compared with adjacent normal colonic mucosa. Furthermore, bowel adenomas (33). This is an apt model because sporadic human the AHSG:transferrin N/T and the AHSG:albumin N/T ratios were colorectal cancers commonly have loss-of-function mutations in APC Ͼ1 in 69 and 68% of specimens, respectively, with congruence and loss of the wild-type allele (34). At 180 days of age, there was no between the two ratios in 85% of patients, demonstrating selective statistically significant difference between groups in the number of depletion of AHSG in the majority of colorectal tumors. The AHSG small bowel adenomatous polyps (Fig. 4A), although there was a trend N/T ratio did not correlate with patient gender, age, or tumor location toward increased polyp multiplicity in Min/ϩAhsgϪ/Ϫ mice compared (Table 1) but did correlate with tumor stage: the AHSG N/T ratio was with Min/ϩAhsgϩ/ϩ and Min/ϩAhsgϩ/Ϫ littermates. By 340 days, 6405

Ahsg Regulates TGF-␤-Dependent Signaling and Immunosup- pression In vivo. TGF-␤1-deficient mice display an overwhelming inflammatory response shortly after weaning (35), which is mediated by hyperactivation of lymphocytes and monocytes. To determine whether TGF-␤1 signaling in immune cells was sensitive to changes in Ahsg levels in vivo, we compared freshly harvested peritoneal macrophages from Ahsgϩ/ϩ and AhsgϪ/Ϫ mice. Monocytes, including Kupffer cells (Fig. 5A) and peritoneal macrophages from wild-type mice (Fig. 5B), accumulate Ahsg protein in vivo, presumably via serum, because the cells do not express Ahsg mRNA (Fig. 5C). In the basal state, macrophages from AhsgϪ/Ϫ mice displayed Ϸ3-fold more Smad2/3 protein localized in the nucleus compared with Ahsgϩ/ϩ mice (Fig. 5, D and E). Smad2/3 nuclear localization in response to exogenous TGF-␤1 added for 30 or 60 minutes was also significantly higher in mutant than wild-type macrophages (Fig. 5, D and E). These results demonstrate hyper-TGF-␤ signaling in AhsgϪ/Ϫ cells. TGF-␤1 acts on monocytes and macrophages to suppress the release of cyto-

Fig. 3. AHSG is depleted in human colorectal cancer. A, three representative cases of colorectal cancer tumor (T) and paired normal colonic mucosa (N) protein extracts probed for AHSG, transferrin, and albumin by Western blotting. B, AHSG, transferrin, and albumin content expressed as a ratio of normal mucosa to tumor tissue, mean Ϯ SE, P ϭ 0.001, AHSG versus transferrin and albumin. C, serial sections of invasive ,ء ,n ϭ 66 colon cancer (T, bottom panels) and adjacent normal colon (N, top panels) stained with H&E or mouse monoclonal antihuman AHSG antibody, showing staining of blood vessels in normal and tumor, and of extracellular stroma in normal only. however, Ahsg-deficient Min/ϩ mice harbored an average of 21.2 Ϯ 2.4 small bowel tumors per mouse, significantly more than either wild-type (10.4 Ϯ 1.6, P ϭ 0.005) or heterozygous (13.8 Ϯ 2.7, P ϭ 0.047) littermates (Fig. 4A). The polyp counts were corrected for intestinal surface area at 340 days of age. Although the mean small bowel and colonic surface areas displayed a trend to increased dimensions in mice lacking Ahsg compared with wild-type and heterozygous littermates, the polyp incidence per small bowel surface area was still significantly elevated (ϳ2-fold) in Ahsg-deficient Min/ϩ mice (Table 2). For each segment of the small bowel (proximal, middle, and distal), the number of Fig. 4. Ahsg inhibits intestinal tumor progression in Min/ϩ mice. A, mean num- polyps per centimeter of bowel length, measured at 340 days, was ber Ϯ SE of small bowel polyps per mouse at 180 and 340 days of age. At 340 days, polyp greater in mice lacking Ahsg than in wild-type or heterozygous number was significantly greater in AhsgϪ/Ϫ compared with Ahsgϩ/ϩ and Ahsgϩ/Ϫ ϭ ء ϩ littermates, and the difference reached statistical significance for the Min/ mice; , P 0.005 and 0.047, respectively. Number of mice per group is indicated. B, mean number of polyps per segment of three equal small bowel segments at P ϭ 0.048 for AhsgϪ/Ϫ versus Ahsgϩ/ϩ (proximal segment) and ,ء .proximal and distal segments. The greatest difference between geno- 340 days of age types was observed in the distal small bowel (Fig. 4B), where Ahsg- P ϭ 0.018 for AhsgϪ/Ϫ versus Ahsgϩ/ϩ and Ahsgϩ/Ϫ (distal segment). C, incidence of P ϭ 0.008 for ,ء ;small bowel polyps Ͻ 1,1to3,orϾ3 mm diameter at 340 days of age deficient Min/ϩ mice harbored almost four times the number of Ϫ/Ϫ ϩ/ϩ ϩ/Ϫ ϩ ϩ Ahsg versus Ahsg and Ahsg . D, histologic sections of small bowel tumors polyps as Min/ϩAhsg / mice and twice that found in Min/ from Min/ϩAhsgϩ/ϩ (adenoma, top panel) and Min/ϩAhsgϪ/Ϫ (invasive carcinoma, ϩAhsgϩ/Ϫ mice. bottom panel; arrow shows penetration through the muscle layer of the bowel wall toward the adjacent fat) mice. Scale bar represents 1 mm. The differences between genotypes for polyp incidence increased in the late stages of disease (340 days), suggesting that Ahsg opposes tumor progression rather than initiation. In support of this interpreta- Table 2 Intestinal surface area and polyp multiplicity at 340 days of age tion, an increased frequency of large tumors (Ͼ3 mm in diameter) was Surface area (cm2)* No. of polyps/cm2* ϩ observed in the small bowel of Ahsg-deficient Min/ mice compared Genotype n Small bowel Colon Small bowel Colon ϩ ϩ with their wild-type and heterozygous littermates at 340 days (Fig. Min/ϩ Ahsg / 8 32.02 Ϯ 0.86 8.02 Ϯ 0.30 0.327 Ϯ 0.054 0.274 Ϯ 0.103 ϩ Ϫ 4C). Invasion into the muscularis propria layer of the bowel wall (Fig. Min/ϩ Ahsg / 17 32.12 Ϯ 0.94 7.88 Ϯ 0.19 0.419 Ϯ 0.083 0.175 Ϯ 0.051 Ϫ Ϫ 4D, bottom panel) was observed in 6 of 73 tumors Ͼ 1 mm diameter Min/ϩ Ahsg / 14 34.24 Ϯ 0.64† 8.46 Ϯ 0.23‡ 0.627 Ϯ 0.074§ 0.214 Ϯ 0.056 * Mean Ϯ SE. isolated from the proximal small bowel segment in Ahsg-deficient ϩ/ϩ ϩ/Ϫ ϩ ϩ † P ϭ 0.080 versus Ahsg , P ϭ 0.099 versus Ahsg , ANOVA. ϩ / ϩ ϩ ϩ Ϫ mice and in none of 18 similarly sized tumors from Min/ Ahsg ‡ P ϭ 0.246 versus Ahsg / , P ϭ 0.055 versus Ahsg / , ANOVA. ϩ ϩ ϩ Ϫ mice (P ϭ 0.01, ␹2). § P ϭ 0.006 versus Ahsg / , P ϭ 0.108 versus Ahsg / , Mann-Whitney. 6406

Fig. 5. Ahsg localizes to macrophages in vivo and regulates TGF-␤- dependent signaling and function. A, serial sections of normal human liver stained with H&E, the monocyte-specific monoclonal antibody KP-1, and mouse monoclonal antihuman AHSG antibody, showing colocalization of staining in Kupffer cells (arrows). B, peritoneal macrophages from 10-week- old Ahsgϩ/ϩ or AhsgϪ/Ϫ mice stained with anti-AHSG antibody. Some AhsgϪ/Ϫ mice were injected intraperitoneally with bovine Ahsg (3 mg in 1 mL of PBS) 72 and 24 hours before lavage (bottom panel). Nuclei are stained with Hoechst. C, Northern analysis of Ahsg mRNA performed on total RNA isolated from peritoneal macrophages or liver homogenates from Ahsgϩ/ϩ or AhsgϪ/Ϫ mice that had (ϩ) or had not (Ϫ) been treated with 25 ␮gof lipopolysaccharide (LPS) i.p. 5 days previously. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression was measured as a loading control. D, localization of Smad2/3 in peritoneal macrophages from Ahsgϩ/ϩ or AhsgϪ/Ϫ mice, at baseline (control) or treated with 40 pmol/L TGF-␤1 for 60 minutes, imaged by Scan Array automated fluorescence microscopy. E, nuclear-cytoplasmic Smad2/3 intensity difference in peritoneal macrophages from Ahsgϩ/ϩ or AhsgϪ/Ϫ mice treated with 40 pmol/L TGF-␤1 for the indicated times; n ϭ 6 mice per genotype. Data are means Ϯ SE for 100 cells. E and F, nitric oxide release from peritoneal macrophages incubated with the indicated concentration of TGF-␤1(E) or Ahsg (F) for 48 hours.

kines and nitric oxide, thereby reducing tissue inflammation (1). peritoneal macrophages (Fig. 5B). Nuclear localization of Smad2/3 in Macrophages from AhsgϪ/Ϫ mice released 53% less nitric oxide in the macrophages from the AhsgϪ/Ϫ mice that received injections of Ahsg basal state and were ϳ10-fold more sensitive to suppression by protein was normalized, i.e., rendered comparable with that in cells exogenous TGF-␤1 than wild-type cells (Fig. 5F). Addition of Ahsg from Ahsgϩ/ϩ mice (Fig. 6E). This proves that Ahsg protein can be protein to the serum-free culture medium enhanced nitric oxide pro- administered to mice to suppress the TGF-␤ response in macrophages duction by both AhsgϪ/Ϫ and Ahsgϩ/ϩ macrophages (Fig. 5G). Tumor and quite possibly other tissues, including tumors. necrosis factor ␣ release by macrophages was affected by endogenous and exogenous Ahsg in an analogous manner (data not shown). Serum DISCUSSION TGF-␤1 levels were not significantly different in Min/ϩAhsgϩ/ϩ and Min/ϩAhsgϪ/Ϫ animals. In this article, we demonstrate that Ahsg competes with cell surface To assess the effect of Ahsg on TGF-␤ signaling in normal and receptors for TGF-␤1 binding, blocks TGF-␤1-induced Smad2/3 ac- neoplastic intestinal epithelium, we compared phospho-Smad2 stain- tivation in vivo, and suppresses tumor progression in the Min/ϩ model ing in Min/ϩAhsgϩ/ϩ and Min/ϩAhsgϪ/Ϫ littermates (Fig. 6A). For of intestinal neoplasia. Furthermore, AHSG levels were reduced in the epithelial cells located in crypts, villi, or polyps, the intensity of majority of human colorectal cancer tumors compared with adjacent phospho-Smad2 nuclear staining was significantly greater in mice normal mucosa, a reduction that was not dependent on vascularity or lacking Ahsg (Fig. 6B), demonstrating enhanced endogenous TGF-␤ plasma content. Postnatally, Ahsg is expressed almost exclusively by signaling. Consistent with this, we have also found that Ahsg-defi- hepatocytes, and Ahsg protein localized to other tissues is plasma cient mice have a greater capacity to suppress the inflammatory derived. In normal colonic mucosa, AHSG appeared to be localized to response. Cutaneous inflammation induced by topical application of the extracellular matrix, and it is likely that altered matrix structure 12-O-tetradecanoylphorbol 13-acetate was reduced at its peak and and composition in tumor tissue underlies the loss of AHSG. resolved more rapidly in AhsgϪ/Ϫ mice (Fig. 6C). To determine Polyp number was unchanged between 180 and 340 days in Min/ϩ whether extrinsic manipulation of Ahsg levels in vivo could modulate Ahsgϩ/ϩ mice, whereas Min/ϩ AhsgϪ/Ϫ mice showed a significant TGF-␤1 responses, AhsgϪ/Ϫ mice received i.p. injections of bovine increase in multiplicity, size, and invasion at the later time point. Gene Ahsg protein. This partially restored Ahsg in serum (Fig. 6D) and in mutations that increase polyp multiplicity in Min/ϩ mice at an early 6407

Fig. 6. Enhanced endogenous TGF-␤ signaling in mice lacking Ahsg. A and B, immunostaining for phosphorylated Smad2 on sections of intestinal epithelium from 340-day-old P Ͻ 0.01. C, mean difference ,ءء ,P Ͻ 0.05 ,ء ;mice. Intensity of nuclear staining in crypts, villi (A), and polyps (B) was greater in Min/ϩAhsgϪ/Ϫ than Min/ϩAhsgϩ/ϩ littermates in ear thickness between the right (treated) and left (control) ears for each individual mouse at the indicated time after topical application of 12-O-tetradecanoylphorbol-13-acetate to the right ear. P Ͻ 0.05 in AhsgϪ/Ϫ versus Ahsgϩ/ϩ mice; n ϭ 5 per group. D, serum Ahsg in Ahsgϩ/ϩ, Ahsgϩ/Ϫ, and AhsgϪ/Ϫ mice or AhsgϪ/Ϫ mice that received i.p. injections of bovine Ahsg 72 and 24 hours before lavage, detected by Western blot. E, Smad2/3 nuclear localization in peritoneal macrophages from Ahsgϩ/ϩ, AhsgϪ/Ϫ or AhsgϪ/Ϫ mice that received i.p. injections of bovine Ahsg; n ϭ 6 mice per group. Data are means Ϯ SE for 100 cells. age probably enhance tumor initiation because the histologic grade induced skin inflammation was reduced in Ahsg-deficient mice com- remains preinvasive. This has been observed for Min/ϩ mice lacking pared with wild-type littermates, consistent with an increased avail- the mismatch repair gene Msh2 (31). Similarly, Min/ϩ mice lacking ability of TGF-␤ in AhsgϪ/Ϫ mice. Mirroring this effect, reduced the growth factor insulin-like growth factor II have fewer polyps, serum AHSG has been correlated with impaired delayed-type hyper- whereas an insulin-like growth factor II transgene increases polyp sensitivity response in malnourished patients with solid tumors (44). multiplicity in young mice (36). Spontaneous and radiation-induced Cultured splenic T cells from AhsgϪ/Ϫ mice are less responsive to tumors in Min/ϩ mice commonly have loss of heterozygosity at the T-cell receptor agonists than cells from Ahsgϩ/ϩ mice.4 T-Cell– Ϫ/Ϫ Smad4 locus (37), and Smad4 Min/ϩ mice display increased specific blockade of TGF-␤ signaling is sufficient to restore an polyp multiplicity (38). Smad4 loss and insulin-like growth factor II immune response capable of eradicating tumors in mice (21). This overexpression may enhance intestinal epithelial cell proliferation at indicates the potential for immune cell regulation by Ahsg in the an early stage in tumorigenesis. By contrast, a delayed increase in control of tumor progression. ϩ Ϫ/Ϫ polyp number in the Min/ Ahsg mice was associated with The efficacy of TGF-␤-neutralizing antibodies in short-term pre- increased tumor size and invasive histology. This suggests that growth clinical tumor models, and suppression of cancer progression in mice of individual polyps is enhanced with tumor progression and allows by long-term exposure to the TGF-␤ antagonist Fc:T␤RII highlight more foci to reach the size of macroscopic detection (39). Indeed, the potential therapeutic value of cytokine antagonists in cancer TGF-␤-responsive tumor cells have a selective advantage in late-stage treatment (17–19, 42). In contrast to the exaggerated immune re- neoplasms, both in humans and in mouse models (5–9, 18–20, 40). Ϫ Ϫ sponse observed in TGF-␤1 / mice, mice transgenic for Fc:T␤RII TGF-␤1 down-regulates the tumor suppressor E-cadherin and thus exhibited no chronic adverse effects and a normal life span (20). promotes epithelial-mesenchymal transition, signaling via phosphati- Wakefield et al. (20) speculate that the incomplete blockade of en- dylinositol 3Ј-kinase/protein kinase B/glycogen synthase kinase-3 dogenous TGF-␤ activity effected by Fc:T␤RII was sufficient to (12, 13), presumably unopposed by the tumor suppressor APC/␤- catenin in late-stage Min/ϩ AhsgϪ/Ϫ tumors. inhibit tumor progression in their model without uncontrolled immune In addition to tumor cell autonomous effects, the selective depletion cell activation. Our results indicate that Ahsg similarly effects an ␤ of AHSG in tumor tissue may enhance TGF-␤-dependent immune incomplete blockade of endogenous TGF- and could offer an attrac- ␤ suppression and thereby promote tumor growth (15, 41, 42). In this tive safety profile in vivo. Elevated tumor TGF- 1 expression is regard, cyclosporine promotes tumor progression by suppressing host associated with resistance to cytotoxic chemotherapy (45), and colon ␤ immune function and by directly enhancing tumor cell invasiveness, cancer patients with microsatellite instability and T RII mutations and notably, both mechanisms are TGF-␤ dependent (43). Kupffer show a more favorable response to adjuvant chemotherapy (46). This cells and peritoneal macrophages do not express the Ahsg gene, but suggests that TGF-␤ antagonists could complement the antineoplastic the protein is taken up from extracellular sources, and we observed activity of conventional cytotoxic agents. On the basis of our results, strong Ahsg staining in an intracellular vesicular compartment. Ahsg could also have therapeutic applications in other pathologies. Macrophages produce and activate TGF-␤1, which suppresses release Notably, excess TGF-␤1 contributes to the pathogenesis of asthma, of inflammation-associated nitric oxide and tumor necrosis factor ␣. The extent and duration of 12-O-tetradecanoylphorbol 13-acetate- 4 M. Szweras, personal communication. 6408

Downloaded from cancerres.aacrjournals.org on September 24, 2021. © 2004 American Association for Cancer Research. Ahsg AND INTESTINAL NEOPLASIA persistent infections, chronic inflammation, and tissue fibrosis, and 22. Chang C, Holtzman DA, Chau S, et al. Twisted gastrulation can function as a BMP neutralization of TGF-␤ in animal models has therapeutic benefit (25, antagonist. Nature (Lond.) 2001;410:483–7. 23. Matzuk MM, Lu N, Vogel H, Sellheyer K, Roop DR, Bradley A. Multiple defects and 47, 48). perinatal death in mice deficient in follistatin. Nature (Lond.) 1995;374:360–3. 24. Webb DJ, Wen J, Lysiak JJ, Umans L, Van Leuven F, Gonias SL. Murine alpha- macroglobulins demonstrate divergent activities as neutralizers of transforming ACKNOWLEDGMENTS growth factor beta and as inducers of nitric oxide synthesis. A possible mechanism for the endotoxin insensitivity of the alpha2-macroglobulin gene knock-out mouse. J Biol We thank Aaron Pollett for helpful discussion, Patricia Wegrynowki, Naomi Chem 1996;271:24982–8. Dore, Pamela Cheung, Laura Daly, Carina Klemens, and Cindy Law for 25. Wilkinson KA, Martin TD, Reba SM, et al. Latency-associated peptide of transforming growth factor beta enhances mycobacteriocidal immunity in the lung during technical assistance, and Nancy Baxter, Michael A. Ko, and Erin Kennedy for Mycobacterium bovis BCG infection in C57BL/6 mice. Infect Immun 2000;68: assistance with statistical analyses. 6505–8. 26. Demetriou M, Binkert C, Sukhu B, Tenenbaum HC, Dennis JW. Fetuin/alpha2-HS glycoprotein is a transforming growth factor beta type II receptor mimic and cytokine REFERENCES antagonist. J Biol Chem 1996;271:12755–61. 27. Binkert C, Demetriou M, Sukhu B, Szweras M, Tenenbaum HC, Dennis JW. 1. Moustakas A, Pardali K, Gaal A, Heldin CH. Mechanisms of TGF-beta signaling in Regulation of osteogenesis by fetuin. J Biol Chem 1999;274:28514–20. regulation of cell growth and differentiation. Immunol Lett 2002;82:85–91. 28. Tajirian T, Dennis JW, Swallow CJ. Regulation of human monocyte proMMP-9 2. Derynck R, Akhurst RJ, Balmain A. TGF-beta signaling in tumor suppression and production by fetuin, an endogenous TGF-beta antagonist. J Cell Physiol 2000;185: cancer progression. Nat Genet 2001;29:117–29. 174–83. 3. Wakefield LM, Roberts AB. TGF-beta signaling: positive and negative effects on 29. Szweras M, Liu D, Partridge EA, et al. alpha2-HS Glycoprotein/Fetuin, a transform- tumorigenesis. Curr Opin Genet Dev 2002;12:22–9. ing growth factor beta/bone morphogenetic protein antagonist, regulates postnatal 4. Siegel PM, Massague J. Cytostatic and apoptotic actions of TGF-beta in homeostasis bone growth and remodeling. J Biol Chem 2002;277:19991–7. and cancer. Nat Rev Cancer 2003;3:807–21. 30. Jahnen-Dechent W, Schinke T, Tridl A, et al. Cloning and targeted deletion of the 5. Shim KS, Kim KH, Han WS, Park EB. Elevated serum levels of transforming growth mouse fetuin gene. J Biol Chem 1997;272:31496–503. factor-beta1 in patients with colorectal carcinoma: its association with tumor pro- 31. Reitmair AH, Cai J-C, Bjerknes M, et al. MSH2 deficiency contributes to accelerated gression and its significant decrease after curative surgical resection. Cancer (Phila.) APC-mediated intestinal tumorigenesis. Cancer Res 1996;56:2922–6. 1999;85:554–61. 6. Dumont N, Arteaga CL. Transforming growth factor beta and breast cancer: tumor 32. Wrana JL, Attisano L, Carcamo J, et al. TGF-beta signals through a heteromeric protein kinase receptor complex. Cell 1992;71:1003–14. promoting effects of transforming growth factor beta. Breast Cancer Res 2000;2: ϩ 125–32. 33. Luongo C, Moser AR, Gledhill S, Dove WF. Loss of Apc in intestinal adenomas 7. Wikstrom P, Bergh A, Damber JE. Transforming growth factor beta1 and prostate from Min mice. Cancer Res 1994;54:5947–52. cancer. Scand J Urol Nephrol 2000;34:85–94. 34. Powell SM, Zilz N, Beazer-Barclay Y, et al. APC mutations occur early during 8. Cui W, Fowlis DJ, Bryson S, et al. TGF-beta1 inhibits the formation of benign skin colorectal tumorigenesis. Nature (Lond.) 1992;359:235–7. tumors, but enhances progression to invasive spindle carcinomas in transgenic mice. 35. Shull MM, Ormsby I, Kier AB, et al. Targeted disruption of the mouse transforming Cell 1996;86:531–42. growth factor beta1 gene results in multifocal inflammatory disease. Nature (Lond.) 9. Tang B, Vu M, Booker T, et al. TGF-beta switches from tumor suppressor to 1992;359:693–9. prometastatic factor in a model of breast cancer progression. J Clin Investig 2003; 36. Hassan AB, Howell JA. Insulin-like growth factor II supply modifies growth of ϩ 112:1116–24. intestinal adenoma in Apc(Min/ ) mice. Cancer Res 2000;60:1070–6. 10. Comijn J, Berx G, Vermassen P, et al. The two-handed E box binding zinc finger 37. Haines J, Dunford R, Moody J, Ellender M, Cox R, Silver A. Loss of heterozygosity protein SIP1 down-regulates E-cadherin and induces invasion. Mol Cell 2001;7: in spontaneous and X-ray–induced intestinal tumors arising in F1 hybrid min mice: ϩ 1267–78. evidence for sequential loss of APC( ) and Dpc4 in tumor development. Genes 11. Bakin AV, Tomlinson AK, Bhowmick NA, Moses HL, Arteaga CL. Phosphatidyli- Chromosomes Cancer 2000;28:387–94. nositol 3-kinase function is required for transforming growth factor beta-mediated 38. Takaku K, Oshima M, Miyoshi H, Matsui M, Seldin MF, Taketo MM. Intestinal epithelial to mesenchymal transition and cell migration. J Biol Chem 2000;275: tumorigenesis in compound mutant mice of both Dpc4 (Smad4) and Apc genes. Cell 36803–10. 1998;92:645–56. 12. Keely PJ, Westwick JK, Whitehead IP, Der CJ, Parise LV. Cdc42 and Rac1 induce 39. Roberts RB, Min L, Washington MK, et al. Importance of epidermal growth factor integrin-mediated cell motility and invasiveness through PI(3)K. Nature (Lond.) receptor signaling in establishment of adenomas and maintenance of carcinomas 1997;390:632–6. during intestinal tumorigenesis. Proc Natl Acad Sci USA 2002;99:1521–6. 13. Janda E, Lehmann K, Killisch I, et al. Ras and TGF[beta] cooperatively regulate 40. Bennett WP, el Deiry WS, Rush WL, et al. p21waf1/cip1 and transforming growth epithelial cell plasticity and metastasis: dissection of Ras signaling pathways. J Cell factor beta 1 protein expression correlate with survival in non-small cell lung cancer. Biol 2002;156:299–313. Clin Cancer Res 1998;4:1499–506. 14. Oft M, Heider KH, Beug H. TGF-beta signaling is necessary for carcinoma cell 41. Hsieh CL, Chen DS, Hwang LH. Tumor-induced immunosuppression: a barrier to invasiveness and metastasis. Curr Biol 1998;8:1243–52. immunotherapy of large tumors by cytokine-secreting tumor vaccine. Hum Gene Ther 15. Kirkbride KC, Blobe GC. Inhibiting the TGF-beta signalling pathway as a means of 2000;11:681–92. cancer immunotherapy. Expert Opin. Biol Ther 2003;3:251–61. 42. de Visser KE, Kast WM. Effects of TGF-beta on the immune system: implications for 16. Hagedorn HG, Bachmeier BE, Nerlich AG. Synthesis and degradation of basement cancer immunotherapy. Leukemia (Baltimore) 1999;13:1188–99. membranes and extracellular matrix and their regulation by TGF-beta in invasive 43. Hojo M, Morimoto T, Maluccio M, et al. Cyclosporine induces cancer progression by carcinomas. Int J Oncol 2001;18:669–81. a cell-autonomous mechanism. Nature (Lond.) 1999;397:530–4. 17. Arteaga CL, Carty-Dugger T, Moses HL, Hurd SD, Pietenpol JA. Transforming 44. Baskies AM, Chretien PB, Weiss JF, et al. Serum glycoproteins in cancer patients: growth factor beta 1 can induce estrogen-independent tumorigenicity of human breast first report of correlation with in vitro and in vivo parameters of cellular immunity. cancer cells in athymic mice. Cell Growth Differ 1993;4:193–201. Cancer (Phila.) 1980;45:3050–8. 18. Wojtowicz-Praga S, Verma UN, Wakefield L, et al. Modulation of B16 melanoma 45. Liu P, Menon K, Alvarez E, Lu K, Teicher BA. Transforming growth factor beta and growth and metastasis by anti-transforming growth factor beta antibody and inter- response to anticancer therapies in human liver and gastric tumors in vitro and in vivo. leukin-2. J Immunother Emphasis Tumor Immunol 1996;19:169–75. Int J Oncol 2000;16:599–610. 19. Muraoka RS, Dumont N, Ritter CA, et al. Blockade of TGF-beta inhibits mammary 46. Watanabe T, Wu TT, Catalano PJ, et al. Molecular predictors of survival after tumor cell viability, migration, and metastases. J Clin Investig 2002;109:1551–9. adjuvant chemotherapy for colon cancer. N Engl J Med 2001;344:1196–206. 20. Yang YA, Dukhanina O, Tang B, et al. Lifetime exposure to a soluble TGF-beta 47. Sime PJ, O’Reilly KM. Fibrosis of the lung and other tissues: new concepts in antagonist protects mice against metastasis without adverse side effects. J Clin pathogenesis and treatment Clin Immunol 2001;99:308–18. Investig 2002;109:1607–15. 48. Border WA, Noble NA, Yamamoto T, et al. Natural inhibitor of transforming growth 21. Gorelik L, Flavell RA. Immune-mediated eradication of tumors through the blockade factor beta protects against scarring in experimental kidney disease. Nature (Lond.) of transforming growth factor beta signaling in T cells. Nat Med 2001;7:1118–22. 1992;360:361–4.

6409

Downloaded from cancerres.aacrjournals.org on September 24, 2021. © 2004 American Association for Cancer Research. α2HS-glycoprotein, an Antagonist of Transforming Growth Factor β In vivo, Inhibits Intestinal Tumor Progression

Carol J. Swallow, Emily A. Partridge, Jennifer C. Macmillan, et al.

Cancer Res 2004;64:6402-6409.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/64/18/6402

Cited articles This article cites 46 articles, 14 of which you can access for free at: http://cancerres.aacrjournals.org/content/64/18/6402.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/64/18/6402.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/64/18/6402. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.