ADAM28 Is Overexpressed in Human Breast

Research Article

ADAM28 Is Overexpressed in Human Breast Carcinomas: Implications for Carcinoma Cell Proliferation through Cleavage of Insulin-like Growth Factor Binding Protein-3 Yoko Mitsui,1,2 Satsuki Mochizuki,1 Takahide Kodama,1 Masayuki Shimoda,1 Takashi Ohtsuka,1 Takayuki Shiomi,1 Miyuki Chijiiwa,1 Tadashi Ikeda,2 Masaki Kitajima,2 and Yasunori Okada1

Departments of 1Pathology and 2Surgery, School of Medicine, Keio University, Shinjuku-ku, Tokyo, Japan

Abstract domain structure composed of a signal sequence and the following domains: propeptide, metalloproteinase, disintegrin, cysteine-rich, A disintegrin and metalloproteinases (ADAMs) are involved in epidermal growth factor-like, transmembrane, and cytoplasmic various biological events including cell adhesion, cell fusion, domains (2–5). Thirty-three members have been identified in membrane protein shedding, and proteolysis. In the present mammalian species, Caenorhabditis elegans, Xenopus, and Dro- study, our reverse transcription-PCRanalysis showed that sophila.3 However, only one third of the members possess the zinc- among the 12 different ADAM species with a putative binding consensus motif in their metalloproteinase domains (4, 5). metalloproteinase motif, prototype membrane-anchored These metalloproteinase-type ADAM proteins play a role in ADAM28m and secreted-type ADAM28s are selectively shedding of membrane proteins, cell adhesion, cell fusion, and expressed in human breast carcinoma tissues. By real-time cell signaling (2, 3, 6–9). ADAM28 is one of the metalloproteinase- quantitative PCR, their expression levels were significantly type ADAMs and is expressed in two alternative forms [i.e., a higher in carcinomas than in nonneoplastic breast tissues. prototype membrane-anchored form (ADAM28m) and a short In situ hybridization, immunohistochemistry, and immuno- secreted form (ADAM28s); ref. 10]. ADAM28 cleaves myelin basic blotting analyses indicated that ADAM28 is predominantly protein and the ectodomain of CD23 (11, 12). In addition, we have expressed in an active form by carcinoma cells within recently reported that ADAM28, prepared by activation of the carcinoma tissues. A direct correlation was observed between precursor with matrix metalloproteinase (MMP)-7, releases insulin- mRNA expression levels and proliferative activity of the like growth factor (IGF)-I from the IGF-I/IGF binding protein carcinoma cells. Treatment of ADAM28-expressing breast (IGFBP)-3 complex by selective digestion of IGFBP-3 into the two carcinoma cells (MDA-MB231) with insulin-like growth major fragments (13). factor-I (IGF-I) increased cell proliferation, cleavage of IGF A characteristic of malignant tumor cells is their ability to binding protein (IGFBP)-3, as well as IGF-I cell signaling; these autonomously proliferate, invade surrounding tissues, and metas- processes were all significantly inhibited by treatment with tasize to distant organs. Accumulated lines of evidence have shown ADAM inhibitor or anti-ADAM28 antibody. Down-regulation that MMPs play an important role in tumorigenesis and of ADAM28 expression in MDA-MB231 cells with small progression of various human cancers including breast carcinomas interfering RNA significantly reduced cell proliferation, through degradation of the extracellular matrix such as basement IGFBP-3 cleavage, and growth of xenografts in mice. In membrane components (14, 15). Because the catalytic site within addition, cleavage of IGFBP-3 in breast carcinoma tissues the metalloproteinase domain of ADAMs has high sequence was correlated with ADAM28 expression levels and inhibited homology to that of MMPs, it is expected that similar to MMPs, by treatment with ADAM inhibitor or anti-ADAM28 antibody. ADAMs may be involved in carcinogenesis as well as in invasion These results show that ADAM28 is overexpressed in an and metastases of human breast carcinomas, as they are expressed activated form in human breast carcinoma cells and suggest and activated within carcinoma tissues. The expression of ADAM9 that ADAM28 is involved in cell proliferation through (16) and ADAM12 (17) has been shown in human breast enhanced bioavailability of IGF-I released from the IGF-I/ carcinomas, and ADAM10, 15, 17, and 20 are expressed in gastric IGFBP-3 complex by selective IGFBP-3 cleavage in human carcinomas (18). Our previous study also showed that ADAM12 is breast carcinomas. (Cancer Res 2006; 66(20): 9913-20) overexpressed in an active form by glioblastoma cells in human astrocytic tumors and is involved in glioma cell proliferation (19). Introduction In addition, we have recently reported that ADAM28s and A disintegrin and metalloproteinases (ADAMs) are a recently ADAM28m are overexpressed by carcinoma cells in human non– discovered gene family of proteins with sequence similarity to the small-cell lung carcinomas (20). However, there have been no reprolysin family of snake venomases (1). Typical members of the systematic studies on the expression of ADAM family members in ADAM family are type I transmembrane proteins with a unique human breast carcinomas and little is known about the role of ADAM28 in these carcinomas. In the present study, we screened the expression of 12 different

Note: Y. Mitsui and S. Mochizuki contributed equally to this work. metalloproteinase-type ADAM species and found selective expres- Requests for reprints: Yasunori Okada, Department of Pathology, School of sion of ADAM28 in human invasive breast carcinoma tissues. Medicine, Keio University, 35 Shinanomachi, Shinjuku-ku Tokyo 160-0016, Japan. Phone: 81-3-5363-3763; Fax: 81-3-3353-3290; E-mail: [email protected]. I2006 American Association for Cancer Research. doi:10.1158/0008-5472.CAN-06-0377 3 See http://www.people.virginia.edu/%7Ejw7g/Table_of_the_ADAMs.html. www.aacrjournals.org 9913 Cancer Res 2006; 66: (20). October 15, 2006

ADAM28 expression was compared with clinicopathologic variables Canada), or anti-h-actin antibody (0.2 Ag/mL; Sigma-Aldrich, St. Louis, MO); and its tissue localization in carcinomas was determined by immunoreactive protein bands were detected with enhanced chemilumi- in situ hybridization and immunohistochemistry. We also examined nescence Western blotting reagents (Amersham Pharmacia Biotech, the implications of ADAM28 in breast carcinoma cell proliferation, Piscataway, NJ; ref. 20). To study the effects of a synthetic ADAM inhibitor (KB-R7785; a kind gift from Dr. Koichiro Yoshino, Carnabioscience, Kobe, IGFBP-3 degradation, and IGF-I-induced cell signaling. The results Japan; refs. 8, 13) or anti-ADAM28 antibody (297-2F3; ref. 20) on IGFBP-3 provide the first evidence that ADAM28 is overexpressed in breast cleavage in breast carcinomas, carcinoma tissue fragments (f3 Â 3 Â 2 carcinoma cells and promotes carcinoma cell proliferation through mm) were cultured in serum-free DMEM for 24 to 48 hours in the presence degradation of IGFBP-3. or absence of 1 Amol/L KB-R7785 or 5 Ag/mL anti-ADAM28 antibody, and the media (800 AL/lane) were subjected to immunoblotting with anti- IGFBP-3 antibody specific to the COOH-terminal domain of IGFBP-3 (0.2 Materials and Methods Ag/mL; Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Inhibitor activity of Clinical samples and histology. Fresh tissue samples were resected KB-R7785 against ADAM species including ADAM28 was previously from patients with primary breast carcinoma (34 cases) who underwent reported (8, 13). Inhibition of ADAM28 activity with anti-ADAM28 antibody surgery at Keio University Hospital, Tokyo, Japan. Control nonneoplastic (297-2F3) was shown in an assay of IGFBP-3 degradation by ADAM28s (data breast tissues (16 cases) were obtained from sites remote from the not shown). carcinomas. All tumor tissues were obtained at primary resection, and none Immunoblotting analyses were also done using the human breast car- of the patients had been subjected to chemotherapy or radiation therapy cinoma cell lines MDA-MB231 and MCF-7. After synchronization and growth before surgery. The samples were snap-frozen immediately after surgical arrest, cells were treated with 1 Amol/L KB-R7785, 5 Ag/mL anti-ADAM28 removal and stored at À80jC. Human breast carcinomas were classified antibody (297-2F3), or 100 Amol/L aprotinin (Sigma-Aldrich) 30 minutes based on the standard criteria of WHO International Classification of Breast before stimulation with 100 ng/mL IGF-I (Sigma-Aldrich) in the presence of Tumors (21). The 34 cases of carcinomas included invasive ductal each agent for 24 hours, and culture media were subjected to immunoblotting carcinomas (32 cases) and invasive lobular carcinomas (2 cases). Informed with anti-ADAM28 antibody or anti-IGFBP-3 antibody as described above. consent for experimental use of the surgical samples was obtained from Inhibition of IGFBP-3 degradation by the treatments was analyzed by each patient according to institutional ethical guidelines. densitometry using NIH Image 1.63. For analysis of IGF-I cell signaling, Reverse transcription-PCR and real-time quantitative PCR. Total supernatants of the cell homogenates (10 Ag/lane) were subjected to RNA was isolated from frozen sections of breast carcinoma and nonneo- immunoblotting with anti–phospho-IGF type I receptor (p-IGF-IR) antibody plastic control breast tissues and evaluated with Agilent 2100 Bioanalyzer (0.2 Ag/mL; Cell Signaling Co., Beverly, MA), anti-IGF-IR antibody (0.2 Ag/mL; (Agilent Technologies, Palo Alto, CA). Reverse transcription-PCR (RT-PCR) Cell Signaling), anti–phospho-extracellular signal-regulated protein kinase for ADAM8, ADAM9 (ADAM9m and ADAM9s), ADAM10, ADAM12 (p-ERK)-1/2 antibody (0.2 Ag/mL; Cell Signaling), or anti-ERK1/2 antibody (ADAM12m and ADAM12s), ADAM15, ADAM17, ADAM19, ADAM20, (0.2 Ag/mL; Cell Signaling). The effect of mitogen-activated protein kinase ADAM21, ADAM28m, ADAM28s, ADAM30, and housekeeping gene h-actin kinase (MEK) inhibitor on ERK1/2 phosphorylation in MDA-MB231 cells was carried out, and specific PCR amplification from the target mRNAs was was examined by treatment of cells with 50 Amol/L PD98059 (Calbiochem, confirmed by sequencing of PCR products as previously described (19, 20). San Diego, CA) 30 minutes before the IGF-I stimulation for 24 hours. For quantitative analysis of ADAM28s and ADAM28m expression, cDNA Culture media of MDA-MB231 cells, treated with or without IGF-I under was used as template in a TaqMan real-time PCR assay (ABI Prism 7000 serum-free DMEM containing 0.2% lactalbumin hydrolysate, were subjected Sequence Detection System, Applied Biosystems, Foster City, CA), and to gelatin and carboxymethylated transferrin zymography for detection of sample data were normalized to glyceraldehyde-3-phosphate dehydroge- MMPs (23, 24). nase (GAPDH) according to our previous methods (20). Cell proliferation assay. The mitogenic effects of IGF-I on MDA-MB231 In situ hybridization. To identify cells expressing ADAM28 mRNA, the and MCF-7 cells were measured with 5-bromo-2¶-deoxy-uridine (BrdUrd) carcinoma samples (five cases) were subjected to in situ hybridization Labeling and Detection Kit III (Roche Molecular Biochemicals, Basel, according to our previous methods (20). BLASTN searches were done to Switzerland) according to the instructions of the manufacturer. After confirm the specificity of the probe encoding ADAM28 nucleotides 952 to synchronization and growth arrest, cells were treated with 100 ng/mL IGF-I 1,132. Paraffin sections of the tissues were hybridized with digoxigenin- in DMEM containing 1% fetal bovine serum (FBS). After 6 hours, BrdUrd labeled RNA antisense or sense probes, followed by immunohistochemistry (10 Amol/L) was added to the media and cultured for the next 42 hours. To using the peroxidase-labeled avidin-biotin-complex method. After the determine the involvement of ADAM28 activity in the mitogenic effect of reaction, the sections were counterstained with hematoxylin. IGF-I, cells were incubated with 1 Amol/L KB-R7785 or 5 Ag/mL anti- Immunohistochemistry. Paraffin sections were subjected to immuno- ADAM28 antibody 30 minutes before the IGF-I treatment, and then reacted histochemistry for ADAM28 using mouse monoclonal antibody against with BrdUrd for 42 hours in the presence of IGF-I. Similarly, the effects of ADAM28 (10 Ag/mL; 297-2F3) and biotinylated secondary antibody to MEK inhibition on cell proliferation were examined by treating the cells mouse immunoglobulin G (IgG) according to our previous methods (20). As with 50 Amol/L PD98059. For determination of MIB1-positive cell index, control, sections were reacted by replacing the first antibody with breast carcinoma cells cultured on Lab-Tek II Chamber Slides (Nalge Nunc, nonimmune mouse IgG. For immunostaining of Ki-67, the sections were International, Rochester, NY) were treated with 100 ng/mL IGF-I after reacted with mouse monoclonal antibody against Ki-67 (MIB1; DAKO A/S, reaction with 1 Amol/L KB-R7785 or 5 Ag/mL anti-ADAM28 antibody. Cells Glostrup, Denmark) following antigen retrieval (19, 20). The percentage of were immunostained with antibody against Ki-67 as described above. ADAM28-immunostained carcinoma cells to total carcinoma cells was Transient transfection of ADAM28 small interfering RNA to cell measured by observing five different fields at Â200 magnification without lines. A 21-oligonucleotide small interfering RNA (siRNA) with the sequence knowledge of the clinical and real-time PCR data as previously described of ADAM28 (AAGACTTCATCCACTGCATAA) and control nonsilencing (22). MIB1-positive cell index was determined in a similar way (19, 20). oligonucleotide (AATTCTCGGAACGTGTCACGT) were custom synthesized Immunoblotting and zymographical analyses. For immunoblotting of at Qiagen, Inc. (Valencia, CA). Transfection was done with the Nucleofector ADAM28, homogenate supernatants of breast carcinoma tissues (10 Ag/ Kit (Amaxa, Inc., Gaithersburg, MD); siRNA (2 Ag) was added to 1 Â 106 lane) were subjected to SDS-PAGE under reducing conditions and the cells suspended in 100 AL of solution Tfor MDA-MB231 cells and solution V resolved proteins on gels were transferred onto polyvinylidene difluoride for MCF-7 cells. The program A-23 (for MDA-MB231) or P-20 (for MCF-7) membranes. The membranes were blotted with anti-ADAM28 antibody was selected for high density of transfection. The cells were then plated in specific to the metalloproteinase domain of ADAM28 (5 Ag/mL; 297-2F3; 12-well plates and incubated in DMEM containing 5% FBS. After 48 hours, ref. 20), anti-ADAM28m antibody specific to the cytoplasmic domain of they were subjected to immunoblotting analysis or proliferation assay as ADAM28m (1 Ag/mL, CL1ADAM28; Cedarlane Labs, Hornby, Ontario, described above.

Cancer Res 2006; 66: (20). October 15, 2006 9914 www.aacrjournals.org

In vivo effects of ADAM28 siRNA on growth of xenografts in mice. MDA-MB231 cell suspension (2 Â 106/0.1 mL HBSS) was injected s.c. to male and female BALB/c-nu/nu mice (Charles River, Yokohama, Japan). Because there was no difference in tumor growth in male and female mice, male mice were used for further experiments with ADAM28 siRNA. To enhance the in vivo transfection efficiency of siRNA with less toxicity, cationic cardiolipin analogue (CCLA)–based liposome (NeoPharm, Inc., Waukegan, IL) was used as a transfection reagent (25). At 3 days after tumor cell injection, the mice were randomized into three groups. Two groups were treated with ADAM28 siRNA or nonsilencing siRNA in CCLA-based liposome. Based on data of a previous study (25) and our preliminary experiments, we determined optimal siRNA doses, charge ratios of siRNA to CCLA-based liposome, and injection schedule. Then, the siRNA solutions (100 AL) were injected into the s.c. tissues near the xenografts on days 3, 5, and 7 at 3.5 mg/kg/d of siRNA in CCLA-based liposome with a 1:2 charge ratio of siRNA to CCLA-based liposome. One group of mice were injected with an equal volume of 10% sucrose (w/v) as a vehicle control. The three dimensions, height (h), length (l), and width (w), of each tumor were measured at the indicated times, and volumes were calculated according to the following formula: volume = [ph(h2 +3a2)] / 6, where a =(w + l) / 4 (ref. 26). Statistical analysis. Results between the two independent groups were determined with the Mann-Whitney U test. Comparisons among more than three groups were determined by the Kruskal-Wallis test. Statistical analyses were carried out using StatView statistical software on a personal computer. P < 0.05 was considered significant.

Results mRNA expression of ADAM species in human breast carcinomas. The mRNA expression of ADAM8, ADAM9, ADAM10, ADAM12, ADAM15, ADAM17, ADAM19, ADAM20, ADAM21, ADAM28s, ADAM28m and ADAM30 was screened by RT-PCR in eight matched carcinoma and control nonneoplastic breast tissues, in the latter of which normal histology was microscopically verified. As shown in Fig. 1A, ADAM8, ADAM20, and ADAM30 Figure 1. The mRNA expression of ADAM28s and ADAM28m in the human were expressed in neither the carcinoma nor the nonneoplastic breast carcinomas and its correlation with carcinoma cell proliferation. A, tissues, and the expression of ADAM19 was detected in only 12.5% RT-PCR analysis of 12 different ADAM species with a putative metalloproteinase motif in the invasive breast carcinoma and nonneoplastic breast tissues. of the carcinoma samples. In contrast, ADAM21 was constitutively Total RNA was extracted from the carcinoma tissues (lanes 1-8, T) and the expressed in all the carcinoma and control breast tissues. ADAM10 nonneoplastic control breast tissues (lanes 1-8, N), and reverse transcribed into and ADAM17 were expressed in all the carcinoma samples, but cDNA followed bya PCR reaction as described in Materials and Methods. C, positive control for RT-PCR (see ref. 20). B, real-time quantitative PCR they were also expressed in 37.5% and 12.5% of the control tissues, analysis of the relative mRNA expression levels of ADAM28s and ADAM28m in respectively. The expression of ADAM9, ADAM12, ADAM15, the breast carcinomas and nonneoplastic breast tissues. Bars, indicate mean ADAM28s, and ADAM28m seemed to be selective in the carcinoma value. ***, P < 0.0001. n.d., not detected. C, correlation between MIB1-positive cell index and ADAM28 mRNA expression levels in the breast carcinomas. Note tissues because none of the eight control breast tissues showed that MIB1-positive cell index directlycorrelates with ADAM28s/GAPDH ( r = expression of these ADAM species (Fig. 1A). Although the 0.759, P < 0.001; n = 34) or ADAM28m/GAPDH (r = 0.577, P < 0.01; n = 34). expression rate of ADAM9, ADAM12, and ADAM15 was 87.5% in the carcinoma tissues, ADAM28s and ADAM28m were expressed in and ADAM28m, but others showed an expression level similar to all the carcinoma samples examined. Thus, we focused on that of nonneoplastic breast tissues (Fig. 1B), suggesting that breast ADAM28s and ADAM28m and analyzed their expression levels in carcinomas can be classified to carcinomas with high expression of a larger number of carcinoma and control breast tissues by real- ADAM28 and carcinomas with low expression of ADAM28. time quantitative PCR. MIB1-positive cell index determined by immunohistochemistry mRNA expression levels of ADAM28 species and their was 27.9 F 26.0% in the breast carcinomas, and no staining was correlation with MIB1-positive cell index. RNA samples of the observed in the nonneoplastic breast tissues (data not shown). breast carcinomas (n = 34) and control nonneoplastic breast When the index was plotted against mRNA expression levels of tissues, some of which showed partial fibrocystic changes (n = 16), ADAM28s and ADAM28m in each case, there were direct were subjected to real-time quantitative PCR for ADAM28s and correlations between the index and the expression levels (r = ADAM28m. As shown in Fig. 1B, the relative expression level of 0.759, P < 0.001 for ADAM28s and r = 0.577, P < 0.01 for ADAM28m; ADAM28s mRNA in the carcinomas (3.26 F 8.80, mean F SD) was n = 34; Fig. 1C), suggesting that the ADAM28 species play a role in significantly 36-fold higher than that in the control tissues (0.09 F cell proliferation. On the other hand, no significant correlations 0.17; P < 0.0001). Similarly, the expression level of ADAM28m in were obtained between the mRNA expression levels and the carcinomas (2.22 F 3.67) was significantly 20-fold higher than that clinicopathologic variables of breast carcinomas including age, in the control tissues (0.11 F 0.31; P < 0.0001; Fig. 1B). Most breast menopausal status, tumor size, histologic type, lymph node carcinoma samples (f65%) showed high expression of ADAM28s metastasis, and tumor stage (data not shown). www.aacrjournals.org 9915 Cancer Res 2006; 66: (20). October 15, 2006

Localization and protein expression of ADAM28 in breast lanes 5-7). After a longer exposure of the films, carcinoma samples carcinoma tissues. By in situ hybridization, the signal for showed, besides the 42-kDa band, the 55/57-kDa bands, which ADAM28 was observed predominantly in the carcinoma cells with were also immunoreactive with antibody specific to the cytoplas- the antisense probe, although the sense probe gave only a mic domain of ADAM28m (Fig. 2D). No immunoreactive bands background signal in the carcinoma tissues (Fig. 2A). Immunohis- were detected in the tissue samples when the immunoblotting was tochemistry showed that ADAM28 is localized predominantly to done with nonimmune IgG (data not shown). cytoplasm of the carcinoma cells in f65% of the carcinoma Involvement of ADAM28 in IGF-I-induced cell proliferation samples (22/34 cases; Fig. 2B). A few samples (2 of 22 positive of human breast carcinoma cells. When the mRNA expression cases) showed membrane staining as well as cytoplasmic staining of ADAM28s and ADAM28m in human breast carcinoma cell (data not shown). The proportion of ADAM28-immunostained lines, MDA-MB231 and MCF-7, was examined by RT-PCR, MDA- carcinoma cells to total carcinoma cells was 35.4 F 38.7%, and a MB231 cells showed definite expression of ADAM28s and significant correlation was observed between the immunoreactivity ADAM28m, but negligible expression of these ADAM28 species and the mRNA expression levels (data not shown). Negligible or no was observed in MCF-7 cells (Fig. 3A). By immunoblotting with staining was observed with nonimmune mouse IgG (Fig. 2B). By anti-ADAM28 antibody (297-2F3), a 42-kDa immunoreactive band immunoblotting with anti-ADAM28 antibody (297-2F3), an immu- was detected in culture media of MDA-MB231 cells, but not noreactive band of 42 kDa was identified in the carcinoma tissues MCF-7 cells (Fig. 3A). When the cell lines were treated with (Fig. 2C, lanes 1-4), but no such species was recognized in the IGF-I, BrdUrd incorporation was significantly increased 2.5-fold control nonneoplastic breast tissues with normal histology (Fig. 2C, and 1.4-fold with MDA-MB231 and MCF-7 cells, respectively (P < 0.0001; Fig. 3B). On the other hand, enhanced BrdUrd incorporation in IGF-I-treated MDA-MB231 cells, but not in MCF-7 cells, was significantly inhibited after treatment with KB- R7785 (P < 0.0001) or anti-ADAM28 antibody (P < 0.001; Fig. 3B). KB-R7785 did not affect the proliferation of MDA-MB231 cells without IGF-I treatment (data not shown). Stimulatory effects of IGF-I on cell proliferation was confirmed by MIB1-positive cell index: 24.1 F 3.1% versus 68.2 F 6.0% for the control and IGF-I-stimulated MDA-MB231 cells and 29.4 F 2.3% versus 48.7 F 10.7% for the control and IGF-I-stimulated MCF-7 cells, respectively (Fig. 3C). MIB1-positive cell index in IGF-I-stimulated MDA-MB231 cells was significantly reduced after treatment with KB-R7785 (34.2 F 7.4%; P < 0.001) or anti-ADAM28 antibody (38.0 F 9.2%; P < 0.001; Fig. 3C). In contrast, no such inhibition was observed in MCF-7 cells (Fig. 3C). To further show a direct involvement of ADAM28 in the cell proliferation, we transfected the cells with ADAM28 siRNA. As shown in Fig. 3D, the mRNA and protein expression of ADAM28s and ADAM28m in IGF-I- stimulated MDA-MB231 cells was almost completely abolished with ADAM28 siRNA but not with nonsilencing siRNA. Impor- tantly, transfection of siRNA to the cells significantly decreased IGF-I-mediated proliferation of MDA-MB231 cells (P < 0.05; Fig. 3D), whereas ADAM28 siRNA did not influence the proliferation of MCF-7 cells (Fig. 3D). Growth inhibition of MDA-MB231 cell xenografts with ADAM28 siRNA. ADAM28 siRNA or nonsilencing siRNA in CCLA-based liposome was s.c. administered around the MDA- MB231 cell tumors on days 3, 5, and 7 after inoculation of tumor cells, and the average tumor volume was measured at 0, 3, 5, 7, 14, 21, and 31 days. As shown in Fig. 4A and B, the tumor volume on F 3 Figure 2. Tissue localization of ADAM28 mRNA and protein in the breast day 31 in the ADAM28 siRNA group (132 21 mm ) was carcinomas and immunoblotting of ADAM28. A, in situ hybridization of ADAM28 significantly reduced to <40% of the control groups treated with in the breast carcinoma tissues. In situ hybridization was done as described in nonsilencing siRNA (356 F 31 mm3) or 10% sucrose (347 F Materials and Methods. Note the positive and negative signals for ADAM28 in 3 the carcinoma cells with the antisense and sense probes, respectively. Bar, 34 mm ; P < 0.01). No toxic effects of ADAM28 or nonsilencing 50 Am. B, immunohistochemistryof ADAM28 in the breast carcinoma tissues. siRNA in CCLA-based liposome and 10% sucrose were observed, Paraffin sections of the carcinoma tissues were subjected to immunostaining and none of the mice died during the treatments. Immunoblotting with anti-ADAM28 antibodyor nonimmune mouse IgG as described in Materials and Methods. Bar,50Am. C, immunoblotting with anti-ADAM28 antibody analysis indicated that the 42-kDa immunoreactive band of (297-2F3) in the breast carcinoma and nonneoplastic breast tissues. Note the ADAM28 in the xenografts was reduced on day 7 (at the point positive 42-kDa band of ADAM28 in the breast carcinoma samples (lanes 1-4) but not in the nonneoplastic breast samples (lanes 5-7). h-Actin is used as a after the second treatment with ADAM28 siRNA), almost negligible loading control. D, immunoblotting of the two carcinoma samples with on day 14 (7 days after the third treatment), and recovered on day anti-ADAM28 antibody(297-2F3) and anti-ADAM28 antibodyspecific to the 31 (Fig. 4C). cytoplasmic domain of ADAM28m (CL1ADAM28). A longer exposure of the membrane blotted with the 297-2F3 antibodygives 55/57-kDa bands besides the IGFBP-3 degradation and its inhibition by suppression of 42-kDa band, and 55/57-kDa bands are reactive with the CL1ADAM28 antibody. ADAM28 activity. Immunoblotting analysis of the conditioned

Cancer Res 2006; 66: (20). October 15, 2006 9916 www.aacrjournals.org

Figure 3. Involvement of ADAM28 in the IGF-I-induced cell proliferation in MDA-MB231 cells. A, determination of ADAM28 expression in MDA-MB231 cells. The expression of ADAM28 mRNA and protein was examined byRT-PCR and immunoblotting as described in Materials and Methods. Note that ADAM28 is expressed in MDA-MB231 cells but not in MCF-7 cells. h-Actin is used as a loading control. B, inhibition of the IGF-I-induced cell proliferation of MDA-MB231 cells bya synthetic ADAM inhibitor (KB-R7785) and anti-ADAM28 antibody. Breast carcinoma cells were treated with 1 Amol/L KB-R7785 or with 5 Ag/mL anti-ADAM28 antibody, and then cultured in the presence or absence of 100 ng/mL IGF-I. BrdUrd incorporation was measured bya microplate reader as described in Materials and Methods. Columns, mean; bars, SD. C, culture in the absence of IGF-I. ***, P < 0.0001; **, P < 0.001. C, MIB1-positive cell index in MDA-MB231 and MCF-7 cells. Carcinoma cells were treated in a similar wayas described in ( B), and then MIB1-positive cell index (%) was determined byimmunostaining with anti-Ki 67 antibodyas described in Materials and Methods. D, inhibition of MDA-MB231 cell proliferation bytransfection with ADAM28 siRNA. Breast carcinoma cells were transfected with nonsilencing siRNA (NS siRNA) or ADAM28-specific siRNA (ADAM28 siRNA), and then cell proliferation was assessed byBrdUrd incorporation. Effects of ADAM28 siRNA were assessed byRT-PCR and immunoblotting. Note that ADAM28 siRNA abolishes ADAM28 expression and decreases cell proliferation in MDA-MB231 cells. *, P < 0.05.

media of IGF-I-stimulated MDA-MB231 and MCF-7 cells showed treatment resulted in only 31% inhibition. In the culture media of 45-kDa and 40-kDa bands of intact IGFBP-3 protein (27), but MDA- human breast carcinoma tissues, intact IGFBP-3 and 22-kDa MB231 and MCF-7 cells showed different digestion fragments of fragments were also detected, and the intensity of the 22-kDa 22 kDa and 26 kDa, respectively (Fig. 5A and B). Without IGF-I fragments was decreased when the carcinoma tissues were treatment, MDA-MB 231 cells showed low-level production of incubated with KB-R7785 or anti-ADAM28 antibody, but not with intact IGFBP-3 and 22-kDa fragments (data not shown). nonimmune mouse IgG (Fig. 5C). By densitometric analysis, Importantly, the fragment formation in IGF-I-stimulated MDA- treatment with KB-R7785 and anti-ADAM28 antibody resulted in MB231 cells was attenuated after treatment of the cells with KB- 97% and 88% inhibition, but nonimmune IgG showed no R7785, anti-ADAM28 antibody, or ADAM28 siRNA, but not inhibition. In addition, formation of the 22-kDa fragments in the aprotinin (Fig. 5A), although the 26-kDa fragments in MCF-7 cells breast carcinoma tissues seemed to correlate with the expression were not influenced by these treatments (Fig. 5B). Densitometric levels of ADAM28 (Fig. 5D). analysis of the ratios of the 22-kDa fragments to the total amount Zymographical analysis of the culture media of MDA-MB231 of IGFBP-3 protein showed that treatment of MDA-MB231 cells cells stimulated with IGF-I showed proteolytic bands corres- with KB-R7785, anti-ADAM28 antibody, and ADAM28 siRNA gave ponding to pro-MMP-1, pro-MMP-2, pro-MMP-3, pro-MMP-7, and 93%, 87%, and 88% inhibition, respectively, whereas aprotinin pro-MMP-9, but not to their active forms (data not shown). www.aacrjournals.org 9917 Cancer Res 2006; 66: (20). October 15, 2006

Discussion In the present study, we provide the first evidence that among the 12 different metalloproteinase-type ADAMs, ADAM28 is selectively overexpressed in activated form by carcinoma cells in human invasive breast carcinomas with a direct correlation to carcinoma cell proliferation. In addition, our in vitro and in vivo experimental studies have shown that inhibition or down- regulation of ADAM28 activity attenuates cell proliferation and IGF-I-induced cell signaling, together with decreased IGFBP-3 degradation. Based on these findings, we propose the hypothesis that ADAM28 plays a key role in cell proliferation in human breast carcinomas through enhanced bioavailability of IGF-I released from the IGF-I/IGFBP-3 complex by selective cleavage of IGFBP-3. Previous studies showed that ADAM9 or ADAM12 is predominantly expressed in the breast carcinomas by immunostaining and RT-PCR (16, 17) and that ADAM17 is overexpressed in breast carcinomas, although nonneoplastic breast tissues showed low- level expression (30). The present RT-PCR data confirmed the previous results on ADAM9 and ADAM12, but further indicated that ADAM15, ADAM28s, and ADAM28m are overexpressed in human breast carcinomas. Our real-time PCR and immunohisto- chemical analyses showed that f65% of the breast carcinoma cases have strong ADAM28 expression at mRNA and protein levels. Importantly, in situ hybridization and immunohistochemistry indicated expression to be predominantly by the carcinoma cells,

Figure 4. Growth inhibition of MDA-MB231 cell xenografts implanted in mice by injection of ADAM28 siRNA. A, time-course changes in the tumor volume. Mice bearing tumors in the s.c. tissue received injections of ADAM28 siRNA in CCLA-based liposome (n; n = 9), nonsilencing siRNA in CCLA-based liposome (E; n = 9), or 10% sucrose (d; n = 6), and tumor volumes were calculated as described in Materials and Methods. **, P < 0.001. B, macroscopic observation of representative mice bearing tumors (arrows) treated with nonsilencing siRNA or ADAM28 siRNA. C, immunoblotting analysis of ADAM28 and h-actin in the implanted tumors on days 3, 7, 14, and 31. Lanes 1, 3, 5, and 7, nonsilencing siRNA treatment; lanes 2, 4, 6, and 8, ADAM28 siRNA treatment.

Involvement of ADAM28 in IGF-I-triggered cell signaling in MDA-MB231 cells. Because IGF-IR is tyrosine phosphorylated after binding with IGF-I (28), we examined phosphorylation of IGF- IR in IGF-I-treated MDA-MB231 and MCF-7 cells. As shown in Fig. 6A, a definite increase in the phosphorylation of IGF-IR was observed in MDA-MB231 cells after treatment with IGF-I (lanes 1 and 2), and their treatment with KB-R7785 or anti-ADAM28 antibody significantly inhibited the IGF-IR phosphorylation to Figure 5. Inhibition of IGFBP-3 cleavage byADAM inhibitor, anti-ADAM28 basal level (P < 0.001; lanes 3 and 4). On the other hand, similar antibody, or ADAM28 siRNA in breast carcinoma cell lines and breast carcinoma effects were not observed in MCF-7 cells (Fig. 6A, lanes 5-8). To tissues. A and B, MDA-MB231 cells (A) and MCF-7 cells (B) were treated study the IGF-IR-mediated post-receptor signal transduction, the without (lane 1) or with 1 Amol/L KB-R7785 (lane 2), 5 Ag/mL anti-ADAM28 antibody( lane 3), ADAM28 siRNA (lane 4), or 100 Amol/L aprotinin (lane 5), phosphorylation of ERK1/2, one of the downstream molecules in and then treated with IGF-I. Conditioned media were analyzed for IGFBP-3 this pathway (28, 29), was evaluated. Figure 6B shows that ERK1/2 proteolysis by immunoblotting as described in Materials and Methods. Arrows, is phosphorylated through IGF-IR activation in MDA-MB231 cells position of intact IGFBP-3; arrowheads, IGFBP-3 proteolytic fragments. Note the IGFBP-3 fragments with different molecular weights in MDA-MB231 cells (lanes 1 and 2) and the phosphorylation is prevented by their (closed arrowhead) and MCF-7 cells (open arrowhead). C, breast carcinoma incubation with KB-R7785 or anti-ADAM28 antibody (P < 0.001; tissues were cultured in the absence (lane 1) or presence of 1 Amol/L KB-R7785 (lane 2), 5 Ag/mL anti-ADAM28 antibody( lane 3), or 5 Ag/mL nonimmune mouse lanes 3 and 4). However, these reactions of ERK1/2 phosphoryla- IgG (lane 4), and then IGFBP-3 proteolysis was analyzed by immunoblotting. tion were not observed in MCF-7 cells (Fig. 6B, lanes 5-8). In Note that formation of IGFBP-3 fragments (closed arrowhead) is inhibited by addition, the IGF-I-induced ERK1/2 phosphorylation and cell KB-R7785 and anti-ADAM28 antibody. D, immunoblotting of ADAM28 (top), IGFBP-3 (middle), and h-actin (bottom) in homogenate samples of the control proliferation in MDA-MB231 cells were prevented by treatment nonneoplastic breast (lanes 1 and 2) and breast carcinoma tissues (lanes 3-5). with PD98059, a specific inhibitor of MEK1/2 (data not shown). Arrow and arrowhead, intact IGFBP-3 and 22-kDa IGFBP-3 fragments.

Cancer Res 2006; 66: (20). October 15, 2006 9918 www.aacrjournals.org

(31, 37, 38). In the present study, we have provided in vitro and in vivo data that the IGF-I-induced proliferation of MDA-MB231 cells is significantly inhibited by treatment with KB-R7785, neutralizing anti-ADAM28 antibody, or transfection with ADAM28 siRNA, and that injection of ADAM28 siRNA to the MDA-MB231 cell xenografts significantly suppresses tumor growth. Thus, these data show that activity of ADAM28 is involved in IGF-I-induced proliferation of MDA-MB231 cells. The cellular action of IGFs is strictly regulated by IGFBPs because the affinities of IGFs to IGFBPs are higher than those to IGF receptors (27). Therefore, proteolysis of IGFBPs directly controls the bioavailability of IGFs to the IGF receptors, and thereby indirectly modulates cell proliferation (27). Although the IGFBP family is composed of six proteins with high affinity to IGFs, the major IGF transport function is attributed to IGFBP-3, which is the most abundant circulating IGFBP species synthesized by liver and locally produced by cancer tissues (39). Proteolytic cleavage has been shown for IGFBP-3 and has gained wide acceptance as the predominant mechanism for IGF release from the IGF/IGFBP-3 complex (40). MMP-1, MMP-2, MMP-3, MMP-7, MMP-9, MMP-19, and ADAM12 are capable of digesting IGFBP-3 (41–45). We have also reported that ADAM28 can release IGF-I from the IGF-I/ IGFBP-3 complex through selective cleavage of IGFBP-3 (13). MDA- MB231 cells have a potential to express IGFBP-3-degrading proteinases, which include MMP-1, MMP-2, MMP-3, MMP-7, MMP-9 (46–48), and ADAM12 (49), as well as ADAM28. However, the COOH-terminal fragments of IGFBP-3 generated by the action

Figure 6. Inhibition of IGF-I-triggered cell signaling byADAM inhibitor or of MMP-1, MMP-2, MMP-3, or MMP-7 (41, 42) are much smaller anti-ADAM28 antibody. A, serum-starved breast carcinoma cells were treated than the 22-kDa fragments observed in our present and previous with 1 Amol/L KB-R7785 or 5 Ag/mL anti-ADAM28 antibodyin the presence studies (13), whereas MMP-9 degrades IGFBP-3 into the 30-kDa or absence of IGF-I. Whole-cell lysates separated by SDS-PAGE were immunoblotted with anti-phospho-IGF-IR antibody( top) and membranes were and 19-kDa fragments (43), which are also different from the reblotted with anti-IGF-IR antibodyas a control for protein loading ( bottom). ADAM28-digested fragments. In addition, MMPs are produced in Columns, mean densitometric values of phopho-IGF-IR/IGF-IR ratios from inactive pro-MMPs under monolayer cultures used in the present three independent experiments. **, P < 0.001. B, breast carcinoma cells were treated in a similar wayas in ( A), and ERK1/2 activation was evaluated by study, and our zymographical analyses actually showed no active immunoblotting with an anti–phospho-ERK1/2 antibody( top) and anti-ERK1/2 forms of MMP-1, MMP-2, MMP-3, MMP-7, and MMP-9. ADAM12 antibody( bottom). Columns, mean densitometric values of phopho-ERK1/2/ ERK1/2 ratios from three independent experiments. **, P < 0.001. digests IGFBP-3 into several small fragments of 10 to 20 kDa (45), which differ from those seen in the present study. In addition, down-regulation of ADAM12 by siRNA did not prevent the 22-kDa although our methods did not differentiate the isoforms of fragment formation in MDA-MB231 cells.4 In contrast, fragment ADAM28s and ADAM28m. In a parallel study, we recently reported formation was inhibited by treatment with anti-ADAM28 antibody that human non–small-cell lung carcinomas selectively overexpress and ADAM28 siRNA as well as KB-R7785. All these data strongly ADAM28s and ADAM28m (20). Our immunoblotting analyses with suggest that in MDA-MB231 cells, ADAM28 is involved in the two different antibodies showed that ADAM28 is present in protein degradation of IGFBP-3. Importantly, IGFBP-3 digestion in breast bands of 42 kDa and 55/57 kDa in breast and lung carcinoma carcinoma tissues was also inhibited by treatment of the tissues tissues (data not shown for lung carcinomas). Based on the with anti-ADAM28 antibody and KB-R7785, suggesting the molecular weight and reactivity with the antibodies, the 42-kDa possibility that a similar mechanism is present in the human species is considered to be an active form of ADAM28s and/or breast carcinoma tissues. ADAM28m shed from the cell membranes (20), and the 55/57-kDa The physiologic effects of IGFs are mainly mediated through species are propeptide-deleted ADAM28m. Altogether, these data IGF-IR (29). Our data that IGF-I-induced phosphorylation of IGF-IR suggest that ADAM28s and ADAM28m are selectively overex- and ERK1/2 in MDA-MB231 cells is prevented by treatment with pressed in active form by carcinoma cells of these human solid KB-R7785 or anti-ADAM28 antibody suggest the possibility that tumors. proteolytic cleavage of IGFBP-3 by ADAM28 causes IGF-I-induced One of the interesting findings in the present study is that the signaling in the cells. Because IGF-I-induced MDA-MB231 cell mRNA expression of ADAM28s and ADAM28m directly correlates proliferation was also suppressed by treatment with a specific with MIB1-positive cell index of the breast carcinomas. Our recent inhibitor of MEK (50), signaling of the IGF-I-induced cell study on human non–small-cell lung carcinomas (20) has shown a proliferation may predominantly involve the ERK pathway in similar finding. Several growth factors such as IGFs (31–33), MDA-MB231 cells. Further work is necessary to show that in epidermal growth factor (34, 35), and transforming growth factor h (36) have been reported to modulate breast and lung carcinoma cell proliferation. Among them, however, IGF-I and IGF-II are considered to be the most potent mitogens for carcinoma cells 4 S. Mochizuki, Y. Okada, unpublished data. www.aacrjournals.org 9919 Cancer Res 2006; 66: (20). October 15, 2006

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. Cancer Research human breast carcinomas, ADAM28 overexpressed by carcinoma Grant support: Grant-in-Aid from the Ministry of Education, Science, and Culture of Japan, no. 17014079 (Y. Okada). cells promotes cell proliferation through the ERK pathway of The costs of publication of this article were defrayed in part by the payment of page IGF-IR signaling that is triggered with IGF-I released from the charges. This article must therefore be hereby marked advertisement in accordance IGF-I/IGFBP-3 complex by the action of ADAM28. with 18 U.S.C. Section 1734 solely to indicate this fact. We thank Dr. Kiran Chada (Department of Biochemistry, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey) and Dr. Atsushi Acknowledgments Suzuki (Department of Obstetrics and Gynecology, School of Medicine, Keio University) for reviewing the manuscript; and Mayumi Kishi and Michiko Uchiyama Received 1/31/2006; revised 8/12/2006; accepted 8/23/2006. for their technical assistance.

References metalloproteinase) messenger RNA expression in Heli- Siegfried JM. Combined targeting of the estrogen cobacter pylori-infected, normal, and neoplastic gastric receptor and the epidermal growth factor receptor in 1. McLane MA, Marcinkiewicz C, Vijay-Kumar S, mucosa. J Infect Dis 2002;185:332–40. non-small cell lung cancer shows enhanced antiprolifer- Wierzbicka-Patynowski I, Niewiarowski S. Viper venom 19. Kodama T, Ikeda E, Okada A, et al. ADAM12 is ative effects. Cancer Res 2005;65:1459–70. disintegrins and related molecules. Proc Soc Exp Biol selectively overexpressed in human glioblastomas and is 36. Wakefield LM, Piek E, Bottinger EP. TGF-h signaling Med 1998;219:109–19. associated with glioblastoma cell proliferation and in mammary gland development and tumorigenesis. 2. Seals DF, Courtneidge SA. The ADAMs family of shedding of heparin-binding epidermal growth factor. J Mammary Gland Biol Neoplasia 2001;6:67–82. metalloproteases: multidomain proteins with multiple Am J Pathol 2004;165:1743–53. 37. Singer CF, Mogg M, Koestler W, et al. Insulin-like functions. Genes Dev 2003;17:7–30. 20. Ohtsuka T, Shiomi T, Shimoda M, et al. ADAM28 is growth factor (IGF)-I and IGF-II serum concentrations 3. Blobel CP. ADAMs: key components in EGFR overexpressed in human non-small cell lung carcinomas in patients with benign and malignant breast lesions: signalling and development. Nat Rev Mol Cell Biol and correlates with cell proliferation and lymph node free IGF-II is correlated with breast cancer size. Clin 2005;6:32–43. metastasis. Int J Cancer 2006;118:263–73. Cancer Res 2004;10:4003–9. 4. Okada Y. Proteinases and matrix degradation. In: J. 21. Ellis I, Schnitt SJ, Sastre-Garau X, et al. Invasive 38. Allen NE, Roddam AW, Allen DS, et al. A prospective Harris ED, Budd RC, Genovese MC, Firestein GS, Sargent breast carcinoma. In: Tavassoli FA, Devilee P, editors. study of serum insulin-like growth factor-I (IGF-I), IGF- JS, editors. Kelley’s textbook of rheumatology. 7th ed. Pathology and genetics of tumours of the breast and II, IGF-binding protein-3 and breast cancer risk. Br J Philadelphia: Elsevier Saunders; 2005. p. 63–81. female genital organs. Lyon: IARC Press; 2003. p. 9–112. Cancer 2005;92:1283–7. 5. White JM. ADAMs: modulators of cell-cell and cell- 22. Kayano K, Shimada T, Shinomiya T, et al. Activation 39. Hwa V, Oh Y, Rosenfeld RG. The insulin-like growth matrix interactions. Curr Opin Cell Biol 2003;15:598–606. of pro-MMP-2 mediated by MT1-MMP in human factor-binding protein (IGFBP) superfamily. Endocr Rev 6. Black RA, Rauch CT, Kozlosky CJ, et al. A metal- salivary gland carcinomas: possible regulation of pro- 1999;20:761–87. loproteinase disintegrin that releases tumour-necrosis MMP-2 activation by TIMP-2. J Pathol 2004;202:403–11. 40. Bunn RC, Fowlkes JL. Insulin-like growth factor factor-a from cells. Nature 1997;385:729–33. 23. Nomura H, Sato H, Seiki M, Mai M, Okada Y. binding protein proteolysis. Trends Endocrinol Metab 7. Allport JR, Ding HT, Ager A, Steeber DA, Tedder TF, Expression of membrane-type matrix metalloprotei- 2003;14:176–81. Luscinskas FW. L-Selectin shedding does not regulate nase in human gastric carcinomas. Cancer Res 1995;55: 41. Fowlkes JL, Enghild JJ, Suzuki K, Nagase H. Matrix human neutrophil attachment, rolling, or transmigra- 3263–6. metalloproteinases degrade insulin-like growth factor- tion across human vascular endothelium in vitro. 24. Nemori R, Yamamoto M, Kataoka F, et al. Develop- binding protein-3 in dermal fibroblast cultures. J Biol J Immunol 1997;158:4365–72. ment of in situ zymography to localize active matrix Chem 1994;269:25742–6. 8. Asakura M, Kitakaze M, Takashima S, et al. Cardiac metalloproteinase-7 (matrilysin-1). J Histochem Cyto- 42. Miyamoto S, Yano K, Sugimoto S, et al. Matrix hypertrophy is inhibited by antagonism of ADAM12 chem 2005;53:1227–34. metalloproteinase-7 facilitates insulin-like growth factor processing of HB-EGF: metalloproteinase inhibitors as a 25. Chien PY, Wang J, Carbonaro D, et al. Novel cationic bioavailability through its proteinase activity on insulin- new therapy. Nat Med 2002;8:35–40. cardiolipin analogue-based liposome for efficient DNA like growth factor binding protein 3. Cancer Res 2004;64: 9. Nagano O, Murakami D, Hartmann D, et al. Cell- and small interfering RNA delivery in vitro and in vivo. 665–71. matrix interaction via CD44 is independently regulated Cancer Gene Therapy 2005;12:321–8. 43. Manes S, Llorente M, Lacalle RA, et al. The matrix by different metalloproteinases activated in response to 26. Taguchi A, Blood DC, del Toro G, et al. Blockade of metalloproteinase-9 regulates the insulin-like growth extracellular Ca(2+) influx and PKC activation. J Cell RAGE-amphoterin signalling suppresses tumour growth factor-triggered autocrine response in DU-145 carcino- Biol 2004;165:893–902. and metastases. Nature 2000;405:354–60. ma cells. J Biol Chem 1999;274:6935–45. 10. Roberts CM, Tani PH, Bridges LC, Laszik Z, Bowditch 27. Firth SM, Baxter RC. Cellular actions of the insulin- 44. Sadowski T, Dietrich S, Koschinsky F, Sedlacek R. RD. MDC-L, a novel metalloprotease disintegrin cyste- like growth factor binding proteins. Endocr Rev 2002;23: Matrix metalloproteinase 19 regulates insulin-like ine-rich protein family member expressed by human 824–54. growth factor-mediated proliferation, migration, and lymphocytes. J Biol Chem 1999;274:29251–9. 28. Hamelers IH, Steenbergh PH. Interactions between adhesion in human keratinocytes through proteolysis of 11. Howard L, Zheng Y, Horrocks M, Maciewicz RA, estrogen and insulin-like growth factor signaling path- insulin-like growth factor binding protein-3. Mol Biol Blobel C. Catalytic activity of ADAM28. FEBS Lett 2001; ways in human breast tumor cells. Endocr Relat Cancer Cell 2003;14:4569–80. 498:82–6. 2003;10:331–45. 45. Loechel F, Fox JW, Murphy G, Albrechtsen R, Wewer 12. Fourie AM, Coles F, Moreno V, Karlsson L. Catalytic 29. LeRoith D, Werner H, Beitner-Johnson D, Roberts CT. UM. ADAM 12-S cleaves IGFBP-3 and IGFBP-5 and is activity of ADAM8, ADAM15, and MDC-L (ADAM28) on Molecular and cellular aspects of the insulin-like growth inhibited by TIMP-3. Biochem Biophys Res Commun synthetic peptide substrates and in ectodomain cleav- factor I receptor. Endocr Rev 1995;16:143–63. 2000;278:511–5. age of CD23. J Biol Chem 2003;278:30469–77. 30. Fabre-Lafay S, Garrido-Urbani S, Reymond N, Gon- 46. Wyatt CA, Geoghegan JC, Brinckerhoff CE. Short 13. Mochizuki S, Shimoda M, Shiomi T, Fujii Y, Okada Y. calves A, Dubreuil P, Lopez M. Nectin-4, a new hairpin RNA-mediated inhibition of matrix metallopro- ADAM28 is activated by MMP-7 (matrilysin-1) and serological breast cancer marker, is a substrate for teinase-1 in MDA-231 cells: effects on matrix destruc- cleaves insulin-like growth factor binding protein-3. tumor necrosis factor-a-converting enzyme (TACE)/ tion and tumor growth. Cancer Res 2005;65:11101–8. Biochem Biophys Res Commun 2004;315:79–84. ADAM-17. J Biol Chem 2005;280:19543–50. 47. Bachmeier BE, Albini A, Vene R, et al. Cell density- 14. Egeblad M, Werb Z. New functions for the matrix 31. Sachdev D, Yee D. The IGF system and breast cancer. dependent regulation of matrix metalloproteinase and metalloproteinases in cancer progression. Nat Rev Endocr Relat Cancer 2001;8:197–209. TIMPexpressionindifferentlytumorigenicbreast Cancer 2002;2:161–74. 32. Macaulay VM, Everard MJ, Teale JD, et al. Autocrine cancer cell lines. Exp Cell Res 2005;305:83–98. 15. Shiomi T, Okada Y. MT1-MMP and MMP-7 in function for insulin-like growth factor I in human small 48. Gordon LA, Mulligan KT, Maxwell-Jones H, Adams invasion and metastasis of human cancers. Cancer cell lung cancer cell lines and fresh tumor cells. Cancer M, Walker RA, Jones JL. Breast cell invasive potential Metastasis Rev 2003;22:145–52. Res 1990;50:2511–7. relates to the myoepithelial phenotype. Int J Cancer 16. O’Shea C, McKie N, Buggy Y, et al. Expression of 33. Lee HY, Chun KH, Liu B, et al. Insulin-like growth 2003;106:8–16. ADAM-9 mRNA and protein in human breast cancer. Int factor binding protein-3 inhibits the growth of non- 49. Iba K, Albrechtsen R, Gilpin BJ, Loechel F, Wewer J Cancer 2003;105:754–61. small cell lung cancer. Cancer Res 2002;62:3530–7. UM. Cysteine-rich domain of human ADAM 12 (meltrin 17. Kveiborg M, Frohlich C, Albrechtsen R, et al. A role 34. Biscardi JS, Ishizawar RC, Silva CM, Parsons SJ. a) supports tumor cell adhesion. Am J Pathol 1999;154: for ADAM12 in breast tumor progression and stromal Tyrosine kinase signalling in breast cancer: epidermal 1489–501. cell apoptosis. Cancer Res 2005;65:4754–61. growth factor receptor and c-Src interactions in breast 50. Dudley DT, Pang L, Decker SJ, Bridges AJ, Saltiel AR. 18. Yoshimura T, Tomita T, Dixon MF, Axon AT, cancer. Breast Cancer Res 2000;2:203–10. A synthetic inhibitor of the mitogen-activated protein Robinson PA, Crabtree JE. ADAMs (a disintegrin and 35. Stabile LP, Lyker JS, Gubish CT, Zhang W, Grandis JR, kinase cascade. Proc Natl Acad Sci U S A 1995;92:7686–9.

Cancer Res 2006; 66: (20). October 15, 2006 9920 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. Cancer Correction Research Correction: ADAM28 Is Overexpressed in Human Breast Carcinomas: Implications for Carcinoma Cell Proliferation through Cleavage of Insulin-like Growth Factor Binding Protein-3

In this article (Cancer Res 2006;66:9913–20), which appeared in the October 15, 2006, issue of Cancer Research (1), there were errors in Figs. 3C, 3D, 4C, and 5C, which the authors have corrected as follows:

Figure 3C. The authors repeated the experiment and obtained conﬁrmatory results. By utilizing the data, they prepared a corrected version of Fig. 3C.

Figure 3D. The authors found that the loading control b-actin of Fig. 3A was erroneously duplicated for Fig. 3D. Figure 3D has now been corrected by using the appropriate b-actin controls.

Figure 4C. The authors could not find the original figure for Fig. 4C, so they repeated the experiments and obtained confirmatory data to those in the original Fig. 4C; the corrected Fig. 4C below consists of these new data.

Figure 5C. The authors were able to find the original figure for Fig. 5C and realized that lane 4 was inadvertently duplicated from the original figure. The corrected Fig. 5C uses the appropriate image for the control from the original data.

The corrected ﬁgures appear below. These errors were caused in large part by inadequate oversight of data management. These errors, which are now corrected, do not change either the results or the conclusions of the article. The authors regret these errors.

Figure 3C.

Figure 3D.

www.aacrjournals.org 4437 Correction

Figure 4C.

Figure 5C.

Reference 1. Mitsui Y, Mochizuki S, Kodama T, Shimoda M, Ohtsuka T, Shiomi T, et al. ADAM28 is overexpressed in human breast carcinomas: implications for carcinoma cell proliferation through cleavage of insulin-like growth factor binding protein-3. Cancer Res 2006;66:9913–20.

Published online October 15, 2015. doi: 10.1158/0008-5472.CAN-15-2369 2015 American Association for Cancer Research.

4438 Cancer Res; 75(20) October 15, 2015 Cancer Research ADAM28 Is Overexpressed in Human Breast Carcinomas: Implications for Carcinoma Cell Proliferation through Cleavage of Insulin-like Growth Factor Binding Protein-3

Yoko Mitsui, Satsuki Mochizuki, Takahide Kodama, et al.

Cancer Res 2006;66:9913-9920.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/66/20/9913

Cited articles This article cites 48 articles, 20 of which you can access for free at: http://cancerres.aacrjournals.org/content/66/20/9913.full#ref-list-1

Citing articles This article has been cited by 10 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/66/20/9913.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/66/20/9913. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.