Decreased expression of ADAMTS-1 in human breast tumors stimulates migration and invasion

Quantitative real-time RT-PCR (qPCR) was used to analyze mRNA expression level of ADAMTS-1 in 60 primary breast tumors. ADAMTS-1 transcripts were expressed at varying levels (ranging from 0.1 to 7.3 AU) in the primary breast tumors analyzed. We determined a gene expression cut-off value of 0.7 (median value) that differentiated between ADAMTS-1 low expression and high expression in breast tumors. Using this cut-off, 39/60 (65%) of the breast tumors were classified as ADAMTS-1 low expressors, and 21/60 (35%) were classified as ADAMTS-1 high expressors. The clinicopathological characteristics of the patients are summarized in Table 1.

Table 2 compares the mean ADAMTS-1 mRNA expression values in relation to the clinicopathological features of the 60 breast cancer patients.

Statistical analyses demonstrated that high ADAMTS-1 mRNA expression was significantly associated with advanced clinical stages (p = 0.006) (Table 2 and Figure 1). However, there were no significant differences between ADAMTS-1 mRNA expression and age, tumor size, lymph node metastasis or the levels of estrogen or progesterone receptors. Neither overall survival (log-rank test; p = 0.459), nor disease-free survival (log-rank test; p = 0.481) were significantly different between patients with low or high ADAMTS-1 mRNA expression levels (data not shown).

Immunoblot analyses using an antibody raised against the N-terminus of ADAMTS-1 produced at least four different bands in samples from human patients, with approximate molecular weights of 110, 80, 70 and 60 kDa (Figure 2A). Moreover, the levels of ADAMTS-1 expression in tumors was lower compared to the expression levels in adjacent, normal tissues from the same patients, and this comparison was obtained by calculating the ratio between ADAMTS-1 and tubulin expression. The 80 kDa ADAMTS-1 form produced the strongest band, and its expression was lower in tumors as compared to normal adjacent tissues (Figure 2B). No differences were found among the other three bands when comparing the tumor samples to normal adjacent tissues (not illustrated).

Immunohistochemistry (IHC) was carried out in samples from normal tissue; breast cancer tumors of grade IIA, IIB, IIIA and IIIC; and metastatic lymph nodes, and we evaluated ADAMTS-1 expression in the tumor cell cytoplasm and the surrounding stroma (Figure 3A). No significant differences were observed regarding the staining area fraction, even in the higher graded tumors (Figure 3B). However, high-grade tumors (IIIA and IIB) exhibited a trend towards decreased ADAMTS-1 expression in comparison to the lower tumor grades.

Next, we compared tumor and normal tissue subgroups, with tumor samples classified as either 1) estrogen (ER) and progesterone (PR) receptor positive tumors or 2) triple-negative tumors (ER-, PR-, Her-2). In triple-negative tumors, ADAMTS-1 protein expression was reduced in 40% of the cases as compared to both normal tissues and the ER or PR single-positive samples (Figure 3C).

We then assessed differences in staining intensity with regard to the ADAMTS-1 distribution among tumor cells or stroma. The labeling intensity profile (Figure 3D) showed that the expression of ADAMTS-1 in normal tissues was prominent in both epithelial cells and stroma. In contrast, ADAMTS-1 staining was discrete or absent in the tumor stroma, and this discrete or absent staining pattern was more evident in cases with higher grades of malignancy.

We also investigated ADAMTS-1 expression levels in breast tumor cell lines, such as MCF7 and MDA-MB-231. In these cells, siRNA treatment reduced the expression of ADAMTS-1. In addition, cell lysates and conditioned medium were analyzed by immunoblot, and an 80 kDa band was observed for the conditioned medium from both cell lines (Figure 4), whereas a 40 kDa band was present predominantly in the cell lysates. In addition, the immunoblot illustrated the efficiency of ADAMTS-1 siRNA knockdown (Figure 4).

Next, the effect of ADAMTS-1 on cell migration was analyzed using time-lapse imaging. For these experiments, we utilized shRNA-GFP. Cells were transfected with either ADAMTS-1 shRNA-GFP or control shRNA-GFP, and only fluorescent cells were analyzed. We measured the cell distance from the initial position during a set time interval, which yielded a mean velocity measurement (μm/hours). Functional ADAMTS-1 knockdown stimulated the migratory activity of MDA-MB-231 cells, and these cells presented a mean velocity of 11.74 ± 0.9603 μm/hour, which was at least 2 times higher than that of the controls (4.756 ± 0.3993 μm/hour) (Figure 5A-D).

MCF7 cells were used in migration experiments to address whether ADAMTS-1 knockdown stimulated migration in other cell lines, including this non-invasive breast adenocarcinoma cell line. The mean velocity of the control cells was 6.479 ± 0.4896 μm/hour, whereas the velocity of the treated cells was 4.676 ± 0.3026 μm/hour. Therefore, in MCF7 cells, migration was not stimulated and was instead reduced (Figure 5E-H).

As ADAMTS-1 knockdown had different effects in MCF7 and MDA-MB-231 cells, we next focused on the mechanisms of ADAMTS-1 that were responsible for these effects on cell migration. Previous reports have demonstrated that ADAMTS-1 sequesters VEGF [7]. Thus, we hypothesized that decreasing ADAMTS-1 could increase the availability of VEGF. ELISAs were performed to determine the relative VEGF concentrations in conditioned medium from either MDA-MB-231 or MCF7 cells that received ADAMTS-1 siRNA knockdown treatment. We found that MDA-MB-231 cells in conditioned medium exhibited a statistically significant increase in VEGF when ADAMTS-1 expression was reduced (Figure 6A). Conversely, MCF7 cells in conditioned medium showed a decrease in VEGF when ADAMTS-1 was silenced using siRNA (Figure 6A).

To address the functional role of decreased ADAMTS-1 levels in MDA-MB-231 cells and the consequent increase in VEGF, we carried out endothelial tubulogenesis assays.

In MDA-MB-231 cells, ADAMTS-1 expression was silenced by siRNA, and cells transfected with scrambled siRNA served as the controls. We obtained conditioned medium from treated and control cells, and we used this conditioned medium to induce tubulogenesis in human endothelial cells (HUVECs).

HUVECs were grown in conditioned medium derived from MDA-MB-231 cells transfected with ADAMTS-1 siRNA, which created a condition where endothelial cells could be cultured in media with reduced levels of ADAMTS-1. In this situation, tubular network formation was increased as compared to HUVECs grown with MDA-MB-231 control cells (Figure 6B). We next measured the length of the tubes formed in the tubulogenesis assays and determined that tube formation with conditioned medium from MDA-MB-231 cells and reduced ADAMTS-1 expression was increased by 4-fold as compared to the controls (Figure 6C).

Recent reports have also shown that VEGF is related to the migration of cancer cells [13‐18]. Therefore, we hypothesized that the differences between MCF7 cell migration and MDA-MB-231 cell migration may be associated with VEGF expression levels in these cell lines. Thus, we performed immunoblots to compare VEGFR2 protein expression levels between MCF7 and MDA-MB-231 cells, and we found that VEGFR2 expression levels in MDA-MB-231 cells were 1.5-fold higher than in MCF7 cells (Figure 6D).

To analyze the putative role of VEGF during the cell migration of ADAMTS-1 siRNA-treated cells, we carried out cell migration and invasion assays. In these assays, a VEGF blocking antibody was included. We found that decreasing ADAMTS-1 levels enhanced both the migration and invasion of MDA-MB-231 cells. Furthermore, treatment with a VEGF blocking antibody partially rescued both cell migration and invasion to the levels obtained with the control (Figure 7A-B).

Furthermore, ADAMTS-1 knockdown increased invadopodia activity in comparison to control siRNA-treated cells, as shown using a fluorescent gelatin assay (Figure 8A-F). The digested gelatin areas (black spots in the fluorescent background) were measured using Image J, and MDA-MB-231 cells treated with ADAMTS-1 siRNAs exhibited at least a 5-fold increase in digested areas as compared to the controls.

Sixty tumor samples and 20 normal tissues adjacent to the tumors were obtained from 60 breast cancer patients at the Hospital do Câncer, A. C. Camargo, São Paulo, Brazil. Patient age ranged from 23 to 85 years (median 54 years). Tumor samples were dissected to remove any residual normal tissue before freezing and storing the samples in liquid nitrogen. The largest diameter of the tumors was recorded. Microscopic examinations were performed to determine the average number of lymph node metastases in 24 patients. Tumor metastasis into the lymph nodes was detected in 40 patients. Histopathological review of all the tumor slides was performed to confirm the diagnosis. All tumors were classified according to the WHO Histological Typing of Breast Tumors. Infiltrating ductal carcinomas were studied, and the clinical stage was assigned based on the UICC TNM (tumor, nodes, metastases) staging system.

The Institutional Ethics Committee approved this study, and all subjects provided informed consent.

Tissue specimens were pulverized under liquid nitrogen using a Frozen Tissue Pulverizer (Thermovac Industries Corporation, Copiague, NY, USA), and total RNA was extracted using the acid guanidinium thiocyanate-phenol-chloroform method [38]. Ten micrograms of total RNA, previously treated with DNaseI, was reverse transcribed using a High Capacity cDNA Archive Kit (Applied Biosystems, Carlsbad, CA, USA). qPCR was performed using an Applied Biosystems 7500 Real-Time PCR System, and each cDNA sample was analyzed in duplicate. The PCR reactions were carried out in a total volume of 25 μl according to the manufacturer’s instructions for the SYBR Green PCR Core reagent (PE Applied Biosystems). The following PCR primers were used: ADAMTS-1 forward, 5’- TGTGGTGTTTGCGGGGGAAATG-3’ and reverse, 5’- TCGATGTTGGTGGCTCCAGTT -3’; and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) forward, 5’-CCTCCAAAATCAAGTGGGGCG-3’ and reverse, 5’-GGGGCAGAGATGATGACCCTT-3’. The algorithm geNorm was used to define the calibrator gene among the RLP13, HPRT, ACTB and GAPDH transcripts. As a result, the relative gene expression was normalized, with GAPDH expression serving as the internal control. The average value from two groups of 10 normal tissue samples served as a reference sample. The results were expressed as the n-fold difference in target gene expression relative to the expression of the GAPDH gene and the reference sample. The relative expression was calculated using the 2^-ΔΔCT method (CT = fluorescence threshold value; ΔCT = CT of the target gene - CT of the reference gene (GADPH); ΔΔCT = ΔCT of the tumor sample - ΔCT of the reference sample).

ImageJ public domain software (http://​rsb.​info.​nih.​gov/​ij/​) was used for image analysis. The DAB channel was separated by the color deconvolution plugin. The stained areas were segmented and measured. Quantification involved either DAB labeling divided by hematoxylin staining or scoring the labeled intensity of tumor cells or stroma. Immunohistochemical staining was quantitatively assessed by three independent observers (VMF, JBA, RGJ) with minimal interobserver variability (<5%).

MCF7 and MDA-MB-231 cells were cultured in Dulbecco’s Modified Eagle’s Medium-F12 (DMEM-F12, Sigma) supplemented with 10% fetal bovine serum (FBS; Cultilab, Campinas, SP, Brazil). Human umbilical vascular endothelial cells (HUVECs) were cultured in 199 Medium supplemented with LSGS (Gibco). The cells were maintained in 25 cm² flasks in a humidified atmosphere of 5% CO₂ at 37°C.

MCF7 and MDA-MB-231 cells were transfected with commercially available siRNA targeting ADAMTS-1 (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA), according to the manufacturer’s instructions. One day prior to transfection, subconfluent MCF7 and MDA-MB-231 cells were cultured in DMEM supplemented with 10% FBS without antibiotic-antimycotic solution. The cells were incubated with a complex formed by the siRNA (50 nM), transfection reagent (Lipofectamine 2000, Invitrogen) and transfection medium (Opti-MEM I, Invitrogen) for 30 h at 37°C.

SureSilencing shRNA plasmids for human ADAMTS-1 with GFP (#KH01149G, SABiosciences, Frederick, MD, USA) were used for the time-lapse experiments.

Cells transfected with scrambled siRNA or shRNA served as controls. These assays were performed in triplicate.

For the time-lapse fluorescence microscopy, shRNA-GFP was used to silence ADAMTS-1, as previously described. Cells transfected with scrambled interference RNA served as control. The migration was determined using a time-lapse fluorescence microscope with an Olympus IX 81 inverted microscope equipped with an Orca R2 CCD camera. Video recordings were conducted using Cell Observer image software (Olympus). Treated and control cells were maintained at 37°C in a temperature-controlled chamber, and images were collected every 5 minutes (over a total of 4 hours and 30 minutes). MTrack J plugin (Image J software) was used to measure cell velocity.

Transwell inserts (with 8 μm pores) in 12-well plates (BD Biosciences) were used for the migration and invasion assays, where ADAMTS-1 was silenced by siRNA (Santa Cruz).

In the migration assays, treated and control cells (10⁵) were plated into the upper chamber containing 1 mL of DMEM-F12 without serum. The lower chamber was filled with 1.5 mL of DMEM-F12. After 24 hours in culture, the cells were fixed with 4% paraformaldehyde and post-fixed with 0.2% crystal violet in 20% methanol. Cells on the upper side of the filter were removed with a cotton swab. The migrating cells on the lower side of the filter were photographed and counted. These experiments were performed in triplicate and were repeated at least three times.

For the invasion assays, the filters were coated with 10 μl of Matrigel (10-13 mg/ml). Treated and control cells (10⁵) were plated into the upper chamber containing 1 mL of DMEM-F12 without serum. The lower chamber was filled with 1.5 mL of DMEM-F12. After 48 hours in culture, the cells were fixed and stained. The cells on the upper side of filter were removed, as described above, whereas the invading cells on the lower side of the filter were photographed and counted. These experiments were performed in triplicate and were repeated at least three times.

To determine the role of VEGF in migration and invasion following ADAMTS-1 depletion, cells with silenced ADAMTS-1 (10⁵) and controls were plated into the upper chamber containing 1 mL of DMEM-F12 without serum. The cells were incubated with anti-VEGF blocking antibody (1 μg/ml R&D Systems) or non-specific mouse IgG (1 μg/ml Millipore). The lower chamber was filled with 1.5 mL of DMEM-F12 supplemented with 10% of serum. The migration and invasion assays were conducted as previously described.

The level of VEGF in conditioned medium was quantified using MCF7 and MDA-MB-231 cells and the Human VEGF ELISA kit (Invitrogen). Cells were transfected with either ADAMTS-1 siRNA or control siRNA, followed by serum starvation for 24 h. The conditioned media were collected and centrifuged at 1,000 rpm for 5 min at 4°C to remove cellular debris. The conditioned medium was then analyzed by ELISA according to the manufacturer’s instructions, and the results were expressed in pg/mL.

ADAMTS-1 expression was silenced by siRNA in MDA-MB-231 cells, and cells that were transfected with scrambled siRNAs served as controls. We obtained conditioned medium from treated and control cells, as previously described, and this conditioned medium was used to induce tubulogenesis in human endothelial cells (HUVECs).

Three-dimensional HUVEC cultures were seeded on a solidified layer of reduced growth factor (RGF) Matrigel, approximately 1-2 mm in thickness. A round 13-mm coverslip coated with 20 μl of RGF-Matrigel (Trevigen, Gaithersburg, MD, kindly provided by Dr. Matthew Hoffman, NIDCR, NIH) was placed in each well of a 24-well plate. Then, a 100-μl drop of the cell suspension (3.5 × 10⁵ cells/ml) was placed on top of the Matrigel and incubated in HUVEC complete medium. After two hours, the non-adherent cells were washed, and the complete medium was replaced with serum-free medium and conditioned for 24 hours with confluent two-dimensional cultures of either control or siRNA-treated MDA-MB-231 cells. The HUVEC cells remained in these conditions for 4 hours, followed by fixation in 4% paraformaldehyde in PBS, permeabilization with 0.5% Triton X-100, actin staining with rhodamine-phalloidin (Invitrogen-Molecular Probes, Eugene, OR) and mounting with ProLong-DAPI (Invitrogen). The images were acquired using a Zeiss LSM 510 laser scanning confocal microscope (Carl Zeiss, Oberkochen, Germany) and analyzed using Imaris 7.1 (Bitplane Inc, South Windsor, CT, USA) and Wimasis (Wimasis GmbH, Munich, Germany).

To assess the role of ADAMTS-1 in MDA-MB-231 cell invadopodia formation, we carried out a fluorescent gelatin substrate degradation assay. The substrate was prepared using gelatin conjugated to Alexa 568 (Invitrogen, Eugene, OR, USA), and the conjugation followed the manufacturer’s instructions. MDA-MB-231 cells (2 × 10⁴/mL) that were transfected with ADAMTS-1 shRNA were plated onto the fluorescent gelatin substrate and incubated in DMEM with 10% FBS at 37°C for 6 hours. The control cells were treated with scrambled non-silencing shRNA. Treated and control cells were fixed in 4% paraformaldehyde in PBS and mounted using ProLong containing DAPI (Invitrogen).

The images were obtained with an Axiophot widefield fluorescence microscope using a 63x PlanApo 1.4 NA objective (Carl Zeiss) and acquired using a digital CCD monochromatic camera (CoolSnap HQ2, Photometrics Inc, Tucson, AZ, USA). To assess matrix digestion spots in the fluorescent substrate, at least ten Z sections per sample field were obtained using a piezoelectric device (PIFOC, Physik Instrumente, Germany) coupled to the objective. The microscope and devices were controlled using the Metamorph Premier 7.6 software (Molecular Devices, Sunnyvale, CA, USA).

We also examined shRNA-GFP-transfected cells (green channel) while screening for superimposed digested areas (dark spots) in the fluorescent gelatin matrix (red channel). ImageJ was used to measure the degraded areas. The red channel (fluorescent gelatin) was separately processed using a threshold tool to calculate the digested area (μm²) per cell. Volocity software (PerkinElmer, Waltham, MA, USA) was used to determine the orthogonal projections and image restorations using deconvolution algorithms.

Western blots were carried out to compare ADAMTS-1 levels in MCF7 and MDA-MB-231 cell lysates and conditioned medium and to verify siRNA transfection efficiency. The cells were lysed in RIPA buffer (150 mM NaCl, 1.0% NP-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris pH 8.0) containing a protease inhibitor cocktail (Sigma). After centrifugation (10,000 g) for 10 min at 4°C, the supernatants were recovered and quantified (BCA kit, Pierce). The samples were resuspended in Laemmli buffer containing 62.5 mM Tris–HCl pH 6.8, 2% sodium dodecyl sulphate (SDS), 10% glycerol, 5% mercaptoethanol and 0.001% bromophenol blue. The conditioned medium (1 mL) was then ethanol-precipitated, and equal amounts (30 μg) of the cell lysates were electrophoresed on 10% polyacrylamide gels. The proteins were transferred to a Hybond ECL nitrocellulose membrane (Amersham) and blocked in TBS with 5% non-fat milk overnight at 4°C. Following one wash in TBS with 0.05% Tween 20 (TBST), the membranes were probed with antibodies against ADAMTS-1 (1:1,000, Abcam 28284), VEGFR2 (1:500, Santa Cruz) and β-actin (1:2,000, Sigma). The ECL protocol was used to detect proteins on the membrane.

We also used OncoPair INSTA-Blot Breast Tissue membranes (IMB-130a, b, c and e; IMGENEX, San Diego, CA, USA). These ready-to-use PVDF membranes contain denatured protein lysates from ductal carcinomas and were matched with adjacent tissues obtained from seven donors. Each membrane was probed with ADAMTS-1 (Abcam 28284) and alpha-tubulin (Abcam 4074) and revealed using the ECL protocol.

The median values of ADAMTS-1 mRNA expression were in accordance with the clinical and pathological variables and were compared using the Kruskal-Wallis and Mann–Whitney tests. For survival analyses, ADAMTS-1 expression was classified as low (<0.7) or high (≥0.7) according to its median expression value. The overall survival and disease-free survival rates were calculated using the Kaplan-Meier method, and the curves were compared using the log-rank test. The overall survival and disease-free survival were calculated from the day of diagnosis to the date of death and to the date in which recurrence was detected, respectively. The significance level was set at 5% for all tests. The statistical analyses were performed using SPSS software 15.0 (SPSS Inc., Chicago, IL). The protein data analyses and statistical significance were obtained using the Wilcoxon matched test. For in vitro experiments, the data were analyzed using Graph Pad Prism 5 software (Graph Pad Software, Inc., San Diego, CA, USA). A Student’s t test or ANOVA was used to assess the differences.