We began our S100A4 knock down studies using HCT116-LUC, HCT116-LUC-shNC and HCT116-LUC-shS100A4 cells. HCT116-LUC cells also express firefly luciferase. HCT116-LUC-shNC and HCT116-LUC-shS100A4 cells were additionally transfected with either an unrelated control shRNA-sequence (negative control, NC) or an S100A4 specific shRNA. S100A4 mRNA level showed a significant reduction to 13% in HCT116-LUC-shS100A4 cells, compared to HCT116-LUC and HCT116-LUC-shNC cells, respectively (P < 0.001). Western blotting of total cell lysates and immunostaining against S100A4 confirmed the decrease of endogenous S100A4 expression level in S100A4-shRNA transfected cells, compared to the respective control cells (Figure 1A). Immunocytochemistry of HCT116-LUC, HCT116-LUC-shNC and HCT116-LUC-shS100A4 cells demonstrated a high expression of S100A4 in HCT116-LUC and HCT116-LUC-shNC cells, but a strong S100A4 protein reduction in HCT116-LUC-shS100A4 cells (Figure 1B).

S100A4 has previously been linked to enhanced tumor migration and formation of metastasis of colorectal cancer [⁶],[²⁴],[²⁵]. Therefore, we analyzed the ability of these cells to migrate through porous membranes with the xCELLigence system, which allows real time data recording of cellular processes. In the xCELLigence-based assay, migrated cells attach on the bottom side of the membrane and increase the electrical impedance at the electrodes. HCT116-LUC-shS100A4 cells showed a delay of the signal increase of almost 3 hours and a lower signal increase than the control cells HCT116-LUC and HCT116-LUC-shNC (Figure 1C). We integrated the area under the signal curves of independent experiments and observed a significant reduction of migrating HCT116-LUC-shS100A4 cells to 49% (P = 0.031) compared to the control cell lines HCT116-LUC and HCT116-LUC-shNC (Figure 1D).

The directed cellular migration was evaluated by closing an applied scratch in a cell layer, documented daily until day 4. HCT116-LUC-shS100A4 cells showed a strong delay in wound closure compared to the control cell lines (Figure 1E). The closure of the wound was quantified by image analysis, resulting in a decrease of 41% in HCT116-LUC-shS100A4 cells (P < 0.001), compared to the control cell lines HCT116-LUC and HCT116-LUC-shNC (Figure 1F).

Beside increased migration, cancer cells have to pass through an intercellular matrix barrier to invade adjacent tissues and form distant metastases. We measured the ability of the cell lines to penetrate an extracellular matrix (ECM) like structure by adding a layer of Matrigel on top of the membranes. Using the xCELLigence system, S100A4-shRNA transfected cells showed a lower increase of the cell index after 24 hours (Figure 1G). The integration of the curves showed a decrease to 55% (P = 0.035), compared to the control cells (Figure 1H). The xCELLigence-based in vitro motility assays were confirmed by classical Boyden chamber assays for cell migration and invasion (Figure S1A,B).

We also analyzed the proliferative abilities of HCT116-LUC, HCT116-LUC-shNC, and HCT116-LUC-shS100A4 cells. However, neither the doubling time nor the ability to form colonies in soft agar differed significantly (Figure S1C,D).

We verified the reduction of cellular motility after S100A4 knock-down in the colorectal cancer cell lines SW620 and DLD-1. We generated stably shRNA transfected clones thereof, SW620-shNC and SW620-shS100A4, as well as DLD-1-shNC and DLD1-shS100A4. In SW620-shS100A4 and DLD-1-shS100A4 cells, S100A4 mRNA levels were reduced to 17% (P = 0.004) and 28% (P = 0.017), respectively, compared to the respective control cell lines containing either no or control shRNA (Figure 2A,B). By counting migrated cells in the Boyden chamber assay, we observed a reduction in cell migration in the cell lines SW620-shS100A4 to 53% (P = 0.030), and DLD-1-shS100A4 to 59% (P = 0.041), compared to the respective control cells (Figure 2C,D). No change in the doubling time of the cell lines derived from SW620 and DLD-1 was observed (Figure 2E,F).

Our results showed a clear decrease in the metastatic potential of colorectal cancer cells, when stably transfected with S100A4-shRNA expression plasmids. To evaluate the anti-metastatic effect of S100A4 knock-down in vivo, the cell lines HCT116-LUC-shNC and HCT116-LUC-shS100A4 were transplanted into the spleens of female NOD/SCID mice (n = 8, each group). The ability of the cells to form metastases in the liver was examined by in vivo bioluminescence imaging of animals and ex vivo scoring of visible metastases after organ resection (Figure 3). Both groups showed luminescence signals near the transplantation site, which increased over time during the experiment (Figure 3A, rows 1 and 3). However, additional luminescence signals distant from the transplantation site were only observed in the control mice (Figure 3A, rows 2 and 4). After the sacrifice of the animals, spleens and livers were removed. The tumors in the spleens of representative mice and the metastatic burden of the corresponding livers are shown as luminescence overlays (Figure 3B). Scoring of the liver metastases showed a significant decrease of number and size of metastases in the animals with transplanted HCT116-LUC-shS100A4 cells (P = 0.004). Liver metastases of control mice had an average score of 27.3 (± 35.1), whereas liver metastases of the mice with transplanted HCT116-LUC-shS100A4 cells had an average score of 0.3 (± 0.8) (Figure 3C). Quantification of bioluminescence signal intensities of organs confirmed the reduction of liver metastases with transplanted HCT116-LUC-shS100A4 cells (P = 0.035) compared to transplanted HCT116-LUC-shNC cells (Figure 3D). No significant differences were found in primary tumor size, average body weight, as well as the weight of spleens and livers (Table 1). The reduction of S100A4 expression in tumor tissue of transplanted HCT116-LUC-shS100A4 mice was verified by qRT-PCR (Figure 3E). When we analyzed changes in the expression of genes associated with S100A4-mediated metastasis formation, we observed a significant decrease of matrix metalloproteinase (MMP) 9 in the tumor tissue derived from HCT116-LUC-shS100A4 cells (Figure 3F).

In order to evaluate therapeutic approaches to interfere with S100A4 driven metastasis formation, we improved the experimental setup by systemic application of S100A4 shRNA plasmids. First, we determined the plasmid copy number of systemically administered shRNA expression plasmids (10 μg plasmid DNA) in the blood of NOD/SCID mice (n = 5) over time before and 5 minutes, 30 minutes, as well as 1, 2, 4, and 24 hours after single tail vein injection. Although the plasmid is cleared rapidly in the blood, 10⁵ copies per μg of total DNA were detectable after 24 hours (Figure 4A).

Based on this, the cell line HCT116-LUC was transplanted intrasplenically into female NOD/SCID mice, and the animals were randomly divided into two groups (n = 8, each). We then applied shRNA expression plasmids systemically by repeated tail vein injection every second day until day 26. One group was treated with shRNA plasmids specific for S100A4, the other with a plasmid containing an unrelated control sequence. As we have observed before, the in vivo luminescence signals at the site of transplantation increased over time in both groups (Figure 4B, rows 1 and 3). However, an increase of distant, additional signals was mainly observed in control mice (Figure 4B, rows 2 and 4). The tumors in the spleens of representative mice and the metastatic burden of the corresponding livers are shown as ex vivo luminescence overlays (Figure 4C). The average score of liver metastases in shS100A4 treated animals was also statistically lower, compared to control mice, 16.0 (± 14.8) and 59.9 (± 42.6), respectively (P = 0.005, Figure 4D). Consistent with our previous data, the quantification of bioluminescence signal intensities of livers and spleens confirmed the reduction in liver metastases in animals treated with S100A4 shRNA expression plasmids (P = 0.016), compared to a treatment with plasmids for control shRNA expression (Figure 4E). No significant difference in primary tumor size, average body weight or the weight of spleens and livers were detectable (Table 1). By plasmid specific qPCR, we quantified the copy number of each plasmid and determined a similar plasmid concentration of about 1 × 10⁶ copies per μg of total DNA in primary tumors and 4 × 10⁶ copies per μg of total DNA in liver metastases, respectively, in both animal groups (Figure 4F,G).

Subsequent analysis of relative S100A4 mRNA levels in micro-dissected spleen and liver tissue showed a significant decrease (P < 0.001) in those animals, treated with shS100A4 expression plasmids, to 62% in tumors and 50% in metastases of the control mice (Figure 5A,B). Again, we found the expression level of MMP9 significantly reduced in mice with specific S100A4-shRNA treatment. MMP9 mRNA expression dropped to 20% (P = 0.037) in spleen tumor tissue and to 37% (P = 0.022) in liver metastases, compared to control mice (Figure 5C,D). Reduced S100A4 and MMP9 expression was confirmed on protein level by immunohistochemistry of tumor tissue from both groups (Figure 5E,F).

Investigation has been conducted in accordance with the ethical standards and according to the Declaration of Helsinki and according to national and international guidelines and has been approved by the authors’ institutional review board.

The human colorectal cancer cell line HCT116 was stably transfected with a pcDNA3.1 plasmid, carrying firefly luciferase under the CMV promoter, and the puromycin resistance cassette [¹⁹]. The resulting cell line HCT116-LUC was further transfected with SureSilencing™ control or S100A4 shRNA expression plasmids, containing the neomycin resistance cassette (SABiosciences, Frederick, MD). HCT116-LUC-shS100A4 cells express a shRNA, specifically targeting S100A4-mRNA, while HCT116-LUC-shNC cells (negative control) express an unrelated control sequence. Transfection of the cell lines SW620 and DLD-1 with the shRNA-expression plasmids resulted in the cell lines SW620-shS100A4, SW620-shNC, DLD-1-shS100A4, and DLD-1-shNC, respectively. All transfections were performed with Fugene HD (Roche Diagnostics, Germany), according to the manufacturer's instructions. Transfected cells were selected with 1 mg/ml neomycin (PAA Laboratories, Austria) or 1 μg/ml puromycin (Life Technologies Corporation, Carlsbad, CA). All cell lines derived from HCT116 and SW620 were grown in RPMI-1640 medium, cell lines derived from DLD-1 were grown in DMEM (PAA Laboratories), each medium supplemented with 10% FCS (Life Technologies Corporation), in a humidified incubator at 37°C and 5% CO₂. The cell lines were tested with the MycoAlert® Mycoplasma Detection Kit (Lonza, Rockland, ME), and found to be free of mycoplasma. The genotype of the parental cell lines HCT116, SW620, and DLD-1 were confirmed by short tandem repeat (STR) genotyping at the DSMZ (German Collection of Microorganisms and Cell Cultures; Braunschweig, Germany). The STR genotypes were consistent with published genotypes for each cell line.

Total RNA from cell culture or tissues was isolated using the Universal RNA Purification Kit (Roboklon, Germany), according to the manufacturer's instructions. Quantification of RNA concentration was performed with Nanodrop (Peqlab, Germany), and 50 ng total RNA was reverse transcribed with random hexamers in a reaction mix (10 mM MgCl₂, 1 × PCR-buffer II, 250 μM pooled dNTPs, 1 U/μL RNAse inhibitor, 2.5 U/μL MuLV reverse transcriptase; all from Life Technologies Corporation). Reaction occurred at 42°C for 15 minutes, 95°C for 5 minutes, and subsequent cooling at 4°C for 5 minutes. The cDNA product was amplified in a total volume of 10 μl in 96-well plates using the LightCycler 480 (Roche Diagnostics, Mannheim, Germany) and the following PCR conditions: 95°C for 10 minutes, followed by 45 cycles of 95°C for 10 seconds, 60°C for 30 seconds, and 72°C for 4 seconds. For S100A4 cDNA quantification, the following primer and probes were used for a 124 bp amplicon: forward primer, 5′-gagctgcccagcttcttg-3′; reverse primer, 5´-tgcaggacaggaagacacag-3´; fluorescein isothiocyanate (FITC) probe, 5′-tgatgagcaacttggacagcaaca-3′; and LCRed640-probe, 5′-gacaacgaggtggacttccaagagt-3′. For cDNA quantification of the housekeeping gene glucose-6-phosphate dehydrogenase (G6PDH) the LightCycler-h-G6PDH Housekeeping Gene Set (Roche Diagnostics) was used, according to manufacturer's instructions, leading to a 113 bp amplicon. For MMP9 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA quantification with a SYBR® Green based reaction mix (GoTaq®; Promega, Germany), the following primers were used: MMP9, forward-primer 5´-tggggggcaactcggc-3´, reverse-primer 5´-ggaatgatctaagcccag-3´, 224 bp amplicon; GAPDH, forward-primer 5´-gaaggtgaaggtcggagtc-3´, reverse-primer 5´-ggtggaatcatattggaacatgtaa-3´, 151 bp amplicon. Data analysis was performed with LightCycler 480 Software release 1.5.0 SP3 (Roche Diagnostics). Mean values were calculated from duplicate qRT-PCR reactions. Each mean value of the expressed gene was normalized to the respective mean value of the housekeeping gene cDNA. The cDNA for the calibrator and the in-run standard derived from HCT116 cells was employed in serial dilutions simultaneously in each run.

For total protein extraction, cells were lysed with RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Nonidet P-40, supplemented with complete protease inhibitor tablets; Roche Diagnostics) for 30 minutes on ice. Protein concentration was quantified with Coomassie Plus (Bradford) Protein Assay Reagent (Pierce; Rockford, IL), according to manufacturer's instructions, and lysates of equal protein concentration were separated with SDS-PAGE and transferred to Hybond-C Extra nitrocellulose membrane (GE Healthcare; Germany). Membranes were incubated in blocking solution containing 5% nonfat dry milk for 1 hour at room temperature. Membranes were incubated overnight at 4°C with rabbit anti-human S100A4 antibody (Dako, Denmark; dilution 1:1000) or goat anti-human GAPDH antibody (Santa Cruz Biotechnology; dilution 1:500) followed by incubation for 1 hour at room temperature with HRP-conjugated anti-rabbit IgG (dilution 1:10000) or anti-goat IgG (dilution 1:10000), respectively. Antibody-protein-complexes were visualized with electrochemical-luminescence reagent (100 mM Tris/HCl, 0.025% w/v luminol, 0.011% w/v para-hydroxycoumaric acid, 10% v/v dimethylsulfoxide, 0.004% v/v H₂O₂, pH 8.6) and subsequent exposure to CL-XPosure™ Films (Pierce). Immunoblotting for GAPDH served as protein loading control. All experiments were performed at least three independent times.

For immunocytochemistry, cells were cultured on double-chamber slides (Nunc, Rochester, NY). For immunohistochemistry, cryosections (5 μm) of tumor tissue were transferred onto glass slides. Samples were fixated for 15 minutes with 0.04% glutaraldehyde, endogenous peroxidase was inactivated for 20 minutes with 0.9% H₂O₂, and cells were permeabilized for 10 minutes with 0.5% Triton X100 in 2.5% BSA. After blocking for 1 hour with 5% BSA, cells were incubated with the S100A4 antibody (dilution 1:400, 2 hours) or rabbit h-MMP9 antibody (Cell Signaling, dilution 1:400, 2 hours). Detection was performed using the biotin-based ABC kit (Dako, Denmark) and diaminobenzidine as substrate, according to manufacturer's instructions. Positive signals resulted in brown staining. The slides were counterstained with haematoxylin and analyzed using a light microscope Axioplan2 (Leica, Germany) at a magnification of 40x. Negative control experiments were carried out by omitting the primary antibody.

For monitoring of cell migration and invasion in real-time the xCELLigence Real Time Cell Analyzer Dual Plate (RTCA-DP) instrument was used according to the manufacturer's recommendations (Roche Applied Science, Germany). The impedance is expressed as a dimensionless parameter, termed cell index, and is directly proportional to the area covered by cells. For detection of cellular migration and invasion, electrical impedance changes are measured at a gold microelectrode plated on the bottom of a membrane separating the upper and lower chambers. The cell line (HCT116-LUC) and the clones expressing control- or S100A4-shRNA (HCT116-LUC-shNC and HCT116-LUC-shS100A4, respectively) were subjected to serum starvation, 12 hours before the start of measurement. For cell migration assays, 5 × 10⁴ cells in RPMI-1640 were seeded per well of a 16-well CIM plate, and the lower chamber was loaded with RPMI-1640, supplemented with 10% FCS. For cell invasion assays, the membranes were coated with Matrigel (dilution 1:40 with medium) 12 hours before seeding 1 × 10⁵ cells. Cell proliferation was determined by seeding 5 × 10³ cells per well into E-Plates and recording the cell index every 15 minutes for 75 hours. Quantitative measurements of changes of the electrical impedance are expressed as cell index. For cell migration, cell index values were monitored every 5 minutes for 24 hours. For invasion, the cell index was measured every 15 minutes for 48 hours. At least three independent experiments were performed for monitoring cell migration or invasion, respectively, each carried out in triplicates. Cell index values were calculated and plotted using the RTCA software 1.2.1 of the RTCA xCELLigence system. For cell migration, the cell index curves were normalized 2 hours after seeding and signals between 2 hours and 8 hours were used to calculate the area under the curve. For cell invasion, the cell index curves were normalized 12 hours after seeding and signals between 12 hours and 36 hours were used to calculate the area under the curves. Average values were normalized to HCT116-LUC.

Wound healing assay was performed as described previously [¹⁹]. HCT116-LUC, HCT116-LUC-shNC and HCT116-LUC-shS100A4 cells were grown to form 60% confluent monolayers, in which a wound of about 0.3 mm width was set with a sterile pipette tip. The medium was exchanged to remove non-adherent cells. The progress of wound closure was monitored daily with microphotographs of 10 × magnification taken with the Leica DM IL light microscope and a Kappa CF 15/4 MCC-MLUII Modul camera (Leica Microsystems) up to day 4. The wound healing experiment was performed three independent times. The spreading of cells across the area of the wound at day 4 was quantified with ImageJ [⁴⁹]. The background value (day 1) was substracted and the average values were normalized to HCT116-LUC.

Two × 10³ cells of the cell lines SW620 and DLD-1, including the stable shRNA expression clones SW620-shNC, SW620-shS100A4, DLD-1-shNC and DLD-1-shS100A4, respectively, were seeded in 96-well-plates (for each day one plate) and incubated for 24 hours to allow the cells to attach to the bottom of the wells. Determination of viable cells was performed by adding 3-(4,5-dimethyl-2-thiazol)-2,5-diphenyl-2H-tetrazolium bromide (MTT; Sigma-Aldrich) to a final concentration of 0.5 mg/ml. After 3 hours incubation at 37°C and 5% CO₂ in a humidified incubator, the formazan crystals were resolved by 10% SDS in 10 mM HCl and the absorption was measured at 560 nm. MTT measurements were performed daily for 5 consecutive days, as triplicates in three independent experiments. The absorption data were used to calculate the doubling time of each cell line (V. Roth 2006; wwww.doubling-time.com).

Anchorage-independent cell proliferation was analyzed by soft agar colony formation assay. HCT116-LUC, HCT116-LUC-shNC and HCT116-LUC-shS100A4 cells were resuspended in RPMI-1640, supplemented with 10% FCS and 0.33% (w/v) agarose (Life Sciences Corporation). One × 10³ cells/ml were seeded into soft agar and colony growth was analyzed after 7 days by light microscopy. Only colonies with more than three cells were counted. Independent colony formation experiments were repeated twice, each in triplicate.

The cell lines HCT116-LUC, SW620, and DLD-1 and the respective clones expressing control- or S100A4-shRNA were used in cell migration and invasion analyses performed with Boyden chamber assay. 2.5 × 10⁵ cells were seeded into each transwell chamber with filter membranes of 12 μm pore size (Millipore, Germany). For invasion, filter membranes were coated with Matrigel (BD Biosciences; diluted 1:3 in growth medium) 4 hours before cell seeding, and fresh medium was added to the bottom chamber. After 24 hours (migration) or 72 hours (invasion), respectively, insets were removed and cells, which had migrated through the membrane to the lower chamber, were trypsinized and counted in a Neubauer chamber (LO-Laboroptik, Bad Homburg, Germany). Each well was counted ten times. Each migration or invasion experiment was performed in duplicate. The average number of migrated or invaded cells was determined from at least three independent experiments.

Plasmid concentration in the blood of NOD/SCID mice after single tail vein injection of 10 μg SureSilencing™ shRNA expression plasmid (QIAGEN, Germany), supplemented with 2 μg aurintricarbolic acid (Sigma, Germany) in 200 μl sterile PBS, were determined by total DNA extraction of 200 μl blood samples (Quick Blood DNA Extraction Kit; Roboklon, Germany), taken before single tail vein injection and after 5 minutes, 30 minutes, 1, 2, 4, and 24 hours. Plasmid copy numbers were determined by SYBR Green based quantitative real time PCR (forward-primer 5´-agacaatcggctgctctgat-3´, reverse-primer 5´-caatagcagccagtcccttc-3´, 203 bp amplicon).

Animal experiments were performed in accordance with the UKCCCR guidelines and approved by the responsible local authorities (State Office of Health and Social Affairs, Berlin, Germany). For analysis of stably transfected shRNA-expressing cell lines, 3 × 10⁶ cells HCT116-LUC-shNC or HCT116-LUC-shS100A4 were transplanted at day zero into the spleens of 6 to 8 weeks old female NOD/SCID mice (randomly assigned to 8 mice per group). The mice were imaged twice a week in a NightOWL LB 981 system (Berthold Technologies, Germany) by intraperitoneal injection of 150 mg/kg D-luciferin (Biosynth, Switzerland) under temporal anesthesia with 35 mg/kg hypnomidate (Jassen-Cilag, Germany). Mice were sacrificed at day 28. Spleens (site of tumor injection) and livers (metastasis target organ) were removed and shock-frozen in liquid nitrogen. To evaluate systemic application of shRNA-expressing plasmids, 3 × 10⁶ HCT116-LUC cells were transplanted at day zero into the spleens of 6 to 8 weeks old female NOD/SCID mice (randomly assigned to 8 mice per group). Ten μg SureSilencing™ shRNA expression plasmid (SABiosciences), containing either S100A4-specific (shS100A4) or negative control (shNC) shRNA sequences, respectively, supplemented with 2 μg aurintricarbolic acid in 200 μl sterile PBS, were injected within 1 second into the tail vein at days 1, 5, 7, 9, 12, 14, 16, 19, 21, 23, and 26. The mice were imaged twice a week, as described above. Mice were sacrificed at day 27. Spleens and livers were removed and shock-frozen in liquid nitrogen. The level of liver metastasis was evaluated by scoring. The score for each liver was calculated as the sum of the volumes V of the individual metastases, calculated by V=L² × W (with L as the length of the smaller axis, and W as the length of the larger axis). Luminescence signals of isolated organs were quantified with ImageJ. Serial consecutive cryosections were made from spleen tumors and liver metastases for immunohistochemistry or subsequent DNA and RNA isolations.

Statistical analyses were performed with Sigma Stat Version 3.5. Depending on the normal distribution of the dataset, comparison of two groups was done either by Student's t-test or Mann-Whitney-Rank-Sum-Test. Comparison of more than two groups was performed by one-way analysis of variance (ANOVA) and Bonferroni post hoc multiple comparison. P values less than 0.05 were defined as statistically significant.