Membrane associated collagen XIII promotes cancer metastasis and enhances anoikis resistance

MDA-MB-231 (ATCC) were maintained in DMEM/F12 with 10% FBS and 1% Pen/Strep. BT549 (ATCC) cells were maintained in RPMI-1640 with 10% FBS and 1% Pen/Strep. MCF-10A cells were kind gifts from Michael W. Kilgore, University of Kentucky, Lexington, KY. MCF10A cells were cultured as previously described [35]. Hs-578 T cells (ATCC) were kept as previously described [36]. S1 and T4–2 cells are kind gifts from Dr. Mina J Bissell, and they were cultured as previously described [37]. All the cells were tested for mycoplasma contamination every two months.

Anti-flag M2 (Sigma, F1804, 1:1000 for WB, 1:500 for IF). Anti-human β1 integrins,active (Millipore, MAB2079Z, 1:500 for IF). Anti-human Collagen XIII α1 (R&D systems, AF6346, 1:200 for WB). Anti-smad2/3 (BD Transduction Laboratories, 610842, 1:1000 for WB). Anti-p-smad2 (cell signaling, 130D4, 1:1000 for WB). Anti-tubulin (Cell Signaling, 2148, 1:5000). Anti-Integrin β1 subunit (AIIB2) (DSHB, 528306, 80 μg/ml for block assay). Anti-rat IgG1 (Santa Cruz Biotechology, sc-3882, 1:1000). Anti-PARP (Cell Signaling, 9542, 1:1000). Bovine Collagen Solution, Type I (Advanced BioMatrix, 5005).

Dual-luciferase reporter assay system (Promega, E1960). Click-It EdU Alexa Fluor 488 Imaging kit (Invitrogen, C10337). Growth Factor Reduced BD Matrigel™ (BD Biosciences, 354230). Annexin V (Thermo, A13201). LentiCRISPR v2 (Addgene, 52961), pCDH-EF1-MCS-T2A-Puro (System Biosciences, CD520A-1), p3TP-lux (Addgene, 306281), pGL4.10 (Addgene, 66128).

Mouse Collagen XIII (NM_007731.3) were amplified from constructs pCMV-SPORT6-COL13a1 (Transomic, BC034164), and cloned into pCDH-EF1-MCS-T2A-Puro (System Biosciences, CD520A-1) with the primers of

Forward: 5’ AATTGAATTCGCCACCATGGTGGCGGAGCGCACCCGC 3′;

Reverse: 5’ACTGGCGGCCGCCTTATCGTCGTCATCCTTGTAATCCTGCCCTCCAGGCCTGCTTCT3’.

Human Collagen XIII (NM_001130103.1) was amplified from an expression construct described by Dennis et al. [38], and cloned into FLAG-tagged pCDH-EF1-MCS-T2A-Puro with the primers of

Forward: 5’AATTGCTAGCGCCACCATGGTAGCGGAGCGCACCCAC 3′;

Reverse: 5’ACTGGAATTCCTTGTTCCAGCAGCCTTGGAC 3′.

Three-dimensional (3D) IrECM on-top culture was performed as previous described [39]. Briefly, Growth Factor Reduced BD Matrigel™ was plated on the bottom of the cell culture dish. MDA-MB-231 and MCF10A cells were seeded on the top of the matrigel layer, and additional medium containing 10% Matrigel was added on the top.

CRISPR-Cas9 plasmid for collagen XIII (NM_001130103.1) deletion was constructed with gDNA primers: 5’ CACCGCAGCTCGGCCGTCCGAAAGT 3′ (Forward) and 5’AAACACTTTCGGACGGCCGAGCTGC 3′ (Reverse). MDA-MB-231 cells were infected with the lentivirus containing the CRISPR-Cas9 construct, and monoclones were selected and verified by genomic DNA sequencing and western blot. The collagen XIII-knockout luciferase-expressing or GFP-expressing MDA-MB-231 cells were pooled together for the mouse experiments.

Protein samples were harvested by using 2% SDS in PBS with protease inhibitor cocktail and NaF (2.5 mM), NaVO4 (2 mM). SDS gel electrophoresis, immunoblot and a LI-COR Odyssey Infrared Imaging System were employed for detecting the target protein as previously described [39]. The Image Studio Lite software was used for quantification.

The Dual-luciferase reporter assay was performed as previously described [40]. Briefly, MDA-MB-231cells or MCF10A cells were seeded into a 24-well plate at the density of 0.1 × 10⁶/well. 24 h later the cells will reach 80% confluence. Then 0.5 μg p3TP-lux and 0.025 μg renilla plasmids were transfected into the cells using Fugene HD Transfection Reagent. The cells were starved before 5 ng/ml TGF-β was added. The relative luciferase activity was defined as firefly luciferase activity normalized by renilla luciferase activity. The final results were normalized by the relative luciferase activity of the control vector pGL4.10.

The Transwell invasion assay was performed as previously described [41]. As for the single cell migration assay, MCF-10A and MDA-MB-231 cells were seeded into a 4 chamber glass bottom dish (Invitro Scientific, D35C4–20-1-N) at the density of 2000 cell/cm². About 4 h after seeding, the single cell migration was monitored by Nikon BioStation (Nikon, IMQ) every 10 min for 10 h [41]. In some experiments, the cells were pre-incubated with β1 integrin blocking antibody AIIB2 with the final concentration of 80 μg/ml for 30 min, and then the Transwell invasion and single cell migration assay were performed in the presence of the blocking antibody.

Cells were seeded in poly-HEMA (12 mg/ml in 95% ethanol) pre-coat regular plastic culture dish [42] and cultured in tumorsphere medium [DMEM/F12 medium supplemented with B27 (1:50), EGF (20 ng/ml), bFGF (20 ng/ml), insulin (5 μg/ml), hydrocortisone (0.5 ng/ml), Gentamicin (10 μg/ml)] for 5 days without moving or disturbing the plates. The phase images of mammosphere/tumorsphere were taken by Nikon eclipse 80i microscope. Mammosphere or tumorspheres forming efficiency (%) was calculated as follows: (Number of mammosphere or tumorspheres per well / number of cells seeded per well) × 100.

Poly-HEMA (12 mg/ml in 95% ethanol) pre-coated dishes were used for detachment cell culture [42]. MCF-10A and MDA-MB-231 cells were cultured in regular media with 0.5% methyl cellulose in suspension at a density of 30,000 cells per cm² [42]. Cells were cultured in suspended condition for 24 h. And then were collected for annexin V analysis as manufacturer’s instructions.

For immunofluorescence staining, cells were cultured in a chamber slide (Nalge Nunc International, 154526). The cells were fixed with Methanol/Acetone (1:1) or formalin and permeabilized with 0.5% Triton X-100. The slides were blocked by 10% goat serum at room temperature for 60 min, and incubated with the primary antibodies (anti-flag/anti-active-β1 integrin/anti-caspase 3) at 4 °C overnight. The slides were incubated with secondary antibodies at room temperature for 60 min. Images were taken with Nikon Eclipse 80i fluorescence microscope and Nikon eclipse Ti2 confocal microscope.

Cell proliferation was performed per the instructions of the Click-It EdU Alexa Fluor 488 Imaging kit and assessed by quantification of the proportion of cells with EdU-positive staining. The number of nuclei positive for EdU was counted and divided by the total number of nuclei (DAPI).

For the xenograft experiment, 6-week old female SCID mice were randomly grouped and injected with 2 × 10⁶ control or collagen XIII-silenced MDA-MB-231-luc-D3H2LN cells at 4th mammary fat pad. Tumors were measured with a caliper every other day. Tumor volume (mm³) was estimated using the formula [volume = π × (width)² × (length) / 6]. Twenty five days after tumor cell implantation, the primary tumors are removed by surgery. To detect lung metastasis, bioluminescent images were taken at 3 weeks after primary tumor removal with in vivo imaging system (IVIS).

For lung colonization experiment, 6-week old female SCID mice were randomly grouped and injected with 0.5 × 10⁶ (in 200 μl PBS) control or collagen XIII-silenced MDA-MB-231-luc-D3H2LN cells via tail vein. To detect lung metastasis, bioluminescent images were taken once a week begining 4 weeks after the injection of cancer cells with IVIS. At the experimental endpoint, lung tissues were harvested and fixed with 4% PFA for paraffin-embedded section. H&E staining was performed in lung tissue sections, and images were taken by a Nikon microscope. Metastasized tumors in the lung were quantified by counting three sections per lung sample. For the intracardiac inoculation experiment, 6-week old female nude mice were randomly grouped. 0.2 × 10⁶ (in 100 μl PBS) control, collagen XIII-silenced MDA-MB-231-luc-D3H2LN cells or collagen XIII-silenced MDA-MB-231-GFP cells were injected into left cardiac ventricle. Bioluminescent images were taken once a week to detect lung and bone metastasis. Illumatool was used to detect GFP labeled cells metastasis.

To address the clinical relevance of increased collagen XIII expression, we assessed the association between mRNA levels of Col13A1 and recurrence or distant recurrence free survival using the published microarray dataset generated from 3554 human breast cancer tissue samples [43] (2014 version). Patients were equally grouped into low and high Col13A1 expression based on the mRNA levels. Significant differences in recurrence or distant recurrence survival time were assessed with the Cox proportional hazard (log-rank) test.

All experiments were conducted by three independent experiments. Data were reported as mean ± s.e.m.. Student’s t-test (two groups) or one-way ANOVA (three or more groups) were used to determine the significant differences between means. Statistical analysis was performed with Graph Pad Prism 5 and IBM SPSS Statistics 22. p < 0.05 represents statistical significance and p < 0.01 represents sufficiently statistical significance. All reported p values were from two-sided tests.

To determine whether collagen XIII expression is induced during breast cancer development, we analyzed collagen XIII protein levels in a panel of non-malignant and malignant mammary epithelial cells. Triple negative breast cancer cell line MDA-MB-231, Hs578T, BT549, and T4–2 expressed higher levels of collagen XIII protein than luminal type breast cancer cell lines and non-malignant mammary epithelial cell lines (Fig. 1a). By analyzing TCGA and Finak datasets (www.​oncomine.​com), we found that collagen XIII mRNA levels were significantly increased in human breast cancer tissue compared to normal mammary gland tissue (Fig. 1b, c). Collagen XIII expression in ER negative breast cancer was much higher than the expression in ER positive breast cancer (Fig. 1d). Consistence with cancer cell line data, we also found that triple negative breast cancer tissue had higher level of collagen XIII expression compared with other subtypes (Fig. 1e).

Next, we asked whether collagen XIII expression is associated with clinical outcome in human breast cancer patients. Breast cancer patients were divided into two groups based on collagen XIII mRNA levels (low and high). Kaplan-Meier log rank analysis showed that patients whose tumors had high collagen XIII expression levels had a significantly shorter overall survival period (Fig. 1f). Moreover, the association of collagen XIII expression with poor clinical outcome is stronger in ER negative breast cancer (Fig. 1g). Although collagen XIII mRNA levels are lower in ER positive breast cancer compared to ER negative cancer, increased collagen XIII expression in ER positive breast cancer is still associated with poor clinical outcome (Additional file 1: Figure S1). These results indicate that breast cancer development and progression are accompanied by increased expression of collagen XIII.

Increased expression of collagen XIII has been detected in several types of cancer [24, 25]. However, function of collagen XIII in cancer progression is not clear. Since collagen XIII protein is highly expressed in MDA-MB-231 cells but non-detectable in MCF-10A cells, these cell lines were used for loss- and gain-of function experiments, respectively, to define roles of collagen XIII in breast cancer progression. Collagen XIII expression was silenced in MDA-MB-231 cell line using CRISPR technology (Fig. 2a, b and Additional file 2: Figure S2). 3D culture models have been widely used to examine the malignant mammary tissue morphogenesis [44, 45]. Invasive branching structure in 3D culture is associated with cancer invasion and aggressive cancer phenotypes [36, 41]. We showed that silencing collagen XIII in MDA-MB-231 significantly reduced the number of invasive branches and inhibited invasive growth in 3D culture (Fig. 2c). For the gain-of function experiments, MCF10A cells were infected with lentivirus containing the collagen XIII expression construct. The majority of exogenous collagen XIII was detected on cell surface (Fig. 2d, 2e, and Additional file 3: Figure S3a). We found that collagen XIII expression increased 3D colony size in MCF10A cells (Fig. 2f). The ratio of EdU positive cells was significantly higher in the collagen XIII expression group compared with the control group (Fig. 2g, Additional file 3: Figure S3b). A reduction of activated caspase 3 staining was also detected in collagen XIII-expressing cells (Additional file 3: Figure S3c).

Invasive growth of breast cancer in 3D culture depends on cancer cell invasion and migration [46, 47]. Thus, we asked whether collagen XIII regulates invasion and migration of mammary epithelial cells. Silencing collagen XIII significantly reduced invasion of MDA-MB-231 cells in Transwell experiments (Fig. 3a). Using single cell tracking analysis, we showed that knockout of collagen XIII reduced the velocity and the distance of cell migration in MDA-MB-231 cells (Fig. 3b). In contrast, MCF-10A cells with collagen XIII overexpression were more invasive than control cells (Fig. 3c). Collagen XIII expression also significantly increased the cell migration velocity and distance in MCF-10A (Fig. 3d, and Additional file 4: Figure S4a, b). These results indicate that increased expression of collagen XIII in MDA-MB 231 cells promotes malignant phenotypes in 3D culture by enhancing cancer cell migration and invasion. Collagen XIII is expressed at the intermediate level in malignant T4–2 cells (Fig.1a), and we performed both loss- and gain-of-function experiments using this cell line. Silencing collagen XIII reduced T4–2 invasion, while overexpression of collagen XIII did not significantly enhance cell invasion in the Transwell experiments (Additional file 5: Figure S5). Therefore, the moderate expression of collagen XIII may be sufficient to promote cancer invasion. To determine whether collagen XIII regulates cell invasion in a cell-autonomous fashion, we performed co-culture experiments by mixing GFP-labeled collagen XIII-silenced MDA-MB-231 cells with non-labeled wild type MDA-MB-231 cells. We found that wild type MDA-MB-231 cells could not rescue cell invasion in the collagen XIII-silenced cells (Additional file 6: Figure S6). These results suggest that collagen XIII promotes cancer cell invasion through the receptor on the same cells.

Tumor-initiating cells are the driver of cancer relapses and metastasis [48, 49]. The tumorsphere assay has been used to enrich tumor-initiating cells and to study their colony formation activity [50, 51]. By analyzing the published microarray dataset [52], we found that collagen XIII expression was upregulated in tumor spheroids compared to the corresponding primary tumors (Fig. 4a). To determine whether collagen XIII contributes to tumorsphere formation, we cultured control, collagen XIII-silenced MDA-MB-231 and T4–2 cells in non-adhesive plates. We found that silencing collagen XIII significantly reduced tumorsphere formation in both MDA-MB-231 and T4–2 cells (Fig. 4b and Additional file 7: Figure S7a). In contrast, expression of collagen XIII enhanced mammosphere formation in MCF10A cells (Fig. 4c). Interestingly, overexpression of collagen XIII in T4–2 had little effect on tumorsphere formation, suggesting that the moderate expression of collagen XIII is sufficient to enhance cancer cell stemness (Additional file 7: Figure S7b). Next, we performed another tumorsphere formation experiments by mixing GFP-labeled collagen XIII-silenced MDA-MB-231 cells with non-labeled wild type MDA-MB-231 cells. Wild type MDA-MB-231 cells did not increase the tumorsphere formation efficiency in GFP-labeled collagen XIII-silenced cells (Additional file 8: Figure S8). These results suggest that collagen XIII enhances cancer cell stemness in a cell-autonomous manner.

Tumor initiating cells are more resistant to detachment-induced anoikis, which is crucial for the cancer cell survival during cancer metastasis [53]. To determine whether collagen XIII promotes anoikis resistance, control, collagen XIII-silenced MDA-MB-231, and collagen-expressing MCF10A cells were cultured in methyl cellulose. Cell anoikis was assessed by analyzing PARP cleavage and cell surface-expressed phosphatidylserine with FITC-labeled annexin-V. We found that collagen XIII expression reduced detachment-induced cell surface-expressed phosphatidylserine and PARP cleavage in MCF-10A cells (Fig. 4d, e), while silencing collagen XIII increased annexin-V staining in MDA-MB-231 cells (Fig. 4f). These results suggest that collagen XIII promotes breast cancer progression by enhancing cancer cell stemness and its associated anoikis resistance.

To understand how collagen XIII regulates cancer progression, we determined whether collagen XIII induces β1 integrin activation. Monoclonal antibody HUTS-4 is specific against active human β1 integrin [54]. Activated β1 integrin was analyzed by immune fluorescence staining with HUTS-4 in control, collagen XIII-silenced or collagen XIII-expressing cells. Quantified data showed that silencing collagen XIII reduced the activated β1 integrin foci in MDA-MB-231 cells (Fig. 5a). In contrast, expression of collagen XIII induced β1 integrin activation in MCF-10A cells (Fig. 5b). AIIB2 is an antibody that blocks β1 integrin activity [33]. To determine whether β1 integrin activation is crucial for collagen XIII-induced cell function, the collagen XIII-expressing MCF10A cells were treated with the control IgG or AIIB2 antibody in the single cell migration, invasion, and mammosphere formation experiments. We showed that the AIIB2 treatment blocked collagen XIII-induced cell migration (Fig. 5c), invasion (Fig. 5d), and mammosphere formation in MCF10A cells (Fig. 5e), while IgG treatment had little effect. In addition, inhibition of β1 integrin activation enhanced PARP cleavage in collagen-expressing cell (Fig. 5f). These results indicate that collagen XIII enhances cell invasion and mammosphere formation at least partially through the β1 integrin pathway.

Aberrant activation of the TGF-β pathway promotes cancer progression by enhancing cancer cell stemness and invasion [55‐57]. It has been shown that β1 integrin contributes to the activation of the TGF-β pathway [58, 59]. We asked whether collagen XIII enhances the TGF-β signaling through β1 integrin. Control and collagen XIII knock-out MDA-MB-231 clones were treated with TGF-β1. The cell lysates at various time points were analyzed by western blot to determine the ratio of phosphorylated SMAD2/3 compared to total SMAD. The TGF-β-induced SMAD2/3 phosphorylation was decreased in the collagen XIII knockout clones compared to control MDA-MB-231 cells (Fig. 6a). Consistent with western blot data, the TGF-β-induced TPA response elements (TREs)-driven luciferase activities were significantly reduced in collagen XIII-silenced cells (Fig. 6b). These experiments were repeated in the MCF-10A cells. Results showed that the TGF-β-induced SMAD2/3 phosphorylation and reporter activities were enhanced by exogenous collagen XIII expression (Fig. 6c, d). Importantly, AIIB2 treatment significantly reduced the collagen XIII-enhanced SMAD2/3 phosphorylation (Fig. 6e). Therefore, collagen XIII expression may enhance the TGF-β signaling through β1 integrin which subsequently promotes cancer progression.

We showed that expression of collagen XIII was associated with short distant recurrence free survival in patients with ER negative (Fig. 7a) and ER positive breast cancer (Additional file 9: Figure S9), suggesting that collagen XIII contributes to cancer metastasis. To determine whether collagen XIII expression promotes cancer cell colonization at distant organs, we silenced collagen XIII expression in MDA-MB-231-luc-D3H2LN cells and pooled multiple clones together (Additional file 10: Figure S10). Control and collagen XIII-silenced cells were injected into the tail veins of SCID mice. Lung colonization of the cancer cells was monitored by IVIS imaging. We showed that the mice injected with control cells developed lung metastasis within 5 weeks, while silencing collagen XIII significantly reduced the lung metastasis (Fig. 7b). Haemotoxylin and Eosin (H&E) staining further confirmed that silencing collagen XIII inhibited the lung colonization of cancer cells in SCID mice (Fig. 7c). Intracardiac inoculation of MDA-MB-231 cells has been used as a model to investigate breast cancer bone metastasis. Using this model, we also found that silencing collagen XIII reduced colonization of MDA-MB-231-luc-D3H2LN cells in nude mice (Fig. 7d). We further analyzed cancer cell colonization in bone using the GFP-labeled MDA-MB-231 cells. Interestingly, bone metastasis was detected in all four mice in the collagen-silenced group, while only three mice had bone metastasis in the control group (Additional file 11: Figure S11). Thus, function of collagen XIII in breast cancer bone metastasis remains for further clarification.

Next we defined roles of collagen XIII in primary tumor growth and cancer metastasis using the MDA-MB-231-luc-D3H2LN orthotopic mammary tumor model [60]. The orthotopic mammary tumor model is a physiologically relevant model to study cancer metastasis. It reproduces the entire metastatic process, including tumor cell dissemination from the mammary fat pad, followed by colonization and outgrowth at distant organs. The same amount of control and collagen XIII-silenced cancer cells were injected into the mammary fat pads of 6-week-old female SCID mice, and tumor growth was monitored twice per week. Then the primary tumors were removed, and mice were maintained for another month before analyzing the metastases in the lung. We found that silencing collagen XIII slightly reduced tumor growth (Fig. 7e). We also observed a reduction of lung metastases in collagen XIII-silenced group (Fig. 7f). Therefore, increased collagen XIII expression may promote cancer colonization and metastasis by enhancing cancer cell invasion and stemness.