Identification of drivers of breast cancer invasion by secretome analysis: insight into CTGF signaling

An altered consistency of tumor microenvironment facilitates the progression of the tumor towards metastasis. Here we combine data from secretome and proteome analysis using mass spectrometry with microarray data from mesenchymal transformed breast cancer cells (MCF-7-EMT) to elucidate the drivers of epithelial-mesenchymal transition (EMT) and cell invasion. Suppression of connective tissue growth factor (CTGF) reduced invasion in 2D and 3D invasion assays and expression of transforming growth factor-beta-induced protein ig-h3 (TGFBI), Zinc finger E-box-binding homeobox 1 (ZEB1) and lysyl oxidase (LOX), while the adhesion of cell-extracellular matrix (ECM) in mesenchymal transformed breast cancer cells is increased. In contrast, an enhanced expression of CTGF leads to an increased 3D invasion, expression of fibronectin 1 (FN1), secreted protein acidic and cysteine rich (SPARC) and CD44 and a reduced cell ECM adhesion. Gonadotropin-releasing hormone (GnRH) agonist Triptorelin reduces CTGF expression in a Ras homolog family member A (RhoA)-dependent manner. Our results suggest that CTGF drives breast cancer cell invasion in vitro and therefore could be an attractive therapeutic target for drug development to prevent the spread of breast cancer.

. Identifying drivers of breast cancer cell invasion. (A) Transwell-invasion co-culture assay of MCF-7 breast cancer cells and MG-63 osteosarcoma cells without FBS addition and Matrigel or Collagen I coated insert. Invaded cells under the filter were stained and counted in four randomly selected regions after 96 h in co-culture. Data represent mean ± SEM. MCF-7 (Matrigel) n = 12, MCF-7 (gelatin) n = 6, unpaired, two-sided t-test to respective control. *P < 0.05; **P < 0.01 (B) Volcano plot demonstrating potential bone-directed breast cancer invasiveness related targets using secretome analysis. Detected target proteins were stated as discovery when adjusted p-value (adj. p-value) was below 0.0016 (dotted line) with a false-discovery rate (FDR) of 1% and a log twofold change (FC) higher 1.3 or lower -1.3. Every dot indicates one target, green dots indicate upregulated discoveries and red dot indicates downregulated discovery. n = 6, discovery determined using the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli, with Q = 1%. Each row was analyzed individually, without assuming a consistent SD. (C) Heat map visualizing all discoveries with a color gradient of log10 integrated area of mean values of three biological and two technical replicates corresponding to B. (D) Scheme of overlapping targets from microarray analysis of MCF-7 cells under dynamic EMT program and secretome analysis of co-cultured MCF-7 cells with a fold change of higher 1.3 or lower -1.3 and FDR 5% (microarray) and FDR 1% (secretome analysis). (E) Comparison of CTGF expression in the secretome of MCF-7 and MG63 cells. Data represent mean ± SEM. n = 6 using unpaired, two-tailed t-test analysis to MCF-7 (= 100%). ***P < 0.001 (F) Comparison of CTGF expression in the proteome of MCF-7 and MG63 cells. Data represent mean ± SEM. n = 6 using unpaired, two-tailed t-test analysis to MCF-7 (= 100%). ***P < 0.001b.

Scientific Reports
| (2020) 10:17889 | https://doi.org/10.1038/s41598-020-74838-8 www.nature.com/scientificreports/ In a second analysis, we tested 94 tissue sections of 47 patients (2 samples per patient) including non-cancerous tissues to analyze whether CTGF expression correlates with expression of androgen (AR), estrogen (ER), progesterone (PR) receptors or epidermal growth factor receptor 2 (HER2) (supplementary table 7b). Of these tissues, 3 were normal breast, 1 periductual mastitis, 3 hyperplasias, 2 fibrocystic changes, 3 fibroadenomas, 29 invasive ductal carcinomas, 1 phyllodes sarcoma, 2 intraductal carcinomas, 1 ductal carcinoma in situ, 1 invasive mucinous adenocarcinoma and 1 invasive lobular carcinoma. Two of the normal breast tissues showed no and 1 normal breast tissue a weak expression of CTGF (33.3%). It is notable that it is the normal breast tissue with increased expression of HER2 which is also slightly positive for CTGF. A positive correlation between HER2 and CTGF expression was recently found in gallbladder cancer and matched adjacent normal tissue 18 . There is also evidence in breast cancer that CTGF has an association with resistance to Lapatinib, a HER2-targeting therapeutic 19 . Twenty-eight (96.5%) of the 29 invasive ductal carcinomas showed CTGF expression. All other above mentioned tissues showed CTGF expression. Apart from the indication for a possible positive correlation between HER2 and CTGF in one normal breast tissue, no recognizable correlations between expression of CTGF and of AR, ER, PR or HER2 were found.
In a third series with 104 breast cancer tissue samples and their matched lymph node metastases tissues, we analyzed whether expression of CTGF in metastatic breast cancer tissues correlates with expression in matched lymph node metastases (supplementary table 7c). Of these cancer/metastatic tissue pairs, 86 were invasive ductal carcinomas, 8 invasive lobular carcinomas, 1 medullary carcinoma, 1 invasive micropapillary carcinoma and 8 mixed carcinomas (invasive ductal and invasive lobular carcinoma) including the matched metastatic tissue. Eighty-one (94.2%) of the 86 invasive ductal carcinomas including their matched metastatic tissue showed CTGF expression. In 3 of these cases (3.7%), CTGF expression in the metastatic tissue was clearly higher than in the primary tumor tissue. In 5 cases (5.8%), CTGF was not detectable either in the invasive ductal carcinoma or in the matched metastatic tissue. All of the 8 invasive lobular carcinomas, the medullary carcinoma, the invasive micropapillary carcinoma and the 8 mixed carcinomas including their matched metastatic tissues showed CTGF expression. In 1 case (12.5%) of the 8 invasive lobular carcinomas and in 2 cases (25%) of the 8 mixed carcinomas, CTGF expression in the matched metastatic tissue was clearly higher than in the primary tumor tissue. Noticeable correlations between CTGF expression and expression of ER, PR and HER2 were not found.
Detection of mesenchymal transformed and aggressive breast cancer cells is a major requirement to select specific treatment options. Previously, it was demonstrated that cells in transient transitional stages express specific cell receptor markers 20 . We found, that highly plastic breast cancer cells and TNBC do not only express more CTGF but co-express CD106 (Vascular cell adhesion molecule 1) and CD51 (Integrin subunit alpha V) in a higher probability than non-invasive MCF-7 cells (Fig. 2E, F and supplementary Fig. 4; MCF-7-EMT 72.67 ± 18.21 counts CD106 high CD51 high vs. MCF-7; P = 0.043; n = 3; MDA-MB-231 197 ± 49 counts CD106 high CD51 high vs. MCF-7; P = 0.0217, n = 3).

CTGF differentially regulates potential drivers of invasion and EMT-markers in mesenchymal transformed and triple negative breast cancer cells.
To further analyze underlying mechanisms of CTGF-induced invasion and suppressed adhesion we examined, if reduced CTGF expression alters expression of TGFBI, CD44, SPARC, FN1, LOX and FSTL1 which were all identified potential drivers for invasion by secretome analysis. We could detect, that reduced CTGF in mesenchymal transformed breast cancer cells suppressed expression of TGFBI ( We found that CTGF had in impact on TGFBI-expression, and further wanted to test, whether a reduced CTGF expression can regulate expression of EMT transcription factors. We examined expression of Cadherin 1 (CDH1), Vimentin (Vim), ZEB1 and SNAIl family transcriptional repressor 2 (SNAI2) after transient CTGF suppression in mesenchymal transformed and TNBC cells. We found that downregulation of CTGF led to reduced ZEB1 expression in mesenchymal transformed breast cancer cells ( Fig. 5C; 0.7767 ± 0.063 FC vs. control; P = 0.0138; n = 3). In contrast, suppressed CTGF resulted in downregulated Vimentin expression in TNBC cells ( Fig. 5D; 0.65 ± 0.0985; P = 0.0237; n = 3).
In a next step we compared protein expression of ZEB1, Cadherin 1 (CDH1), SNAIl family transcriptional repressor 2 (SNAI2) and Vimentin (Vim) in mesenchymal transformed MCF-7-EMT and MDA-MB-231 cells with their protein expression in MCF-7 cells ( Fig.6A and supplementary Fig. 3). We found that protein expression of ZEB1 and Vimentin (Vim) in MCF-7-EMT and MDA-MB-231 cells was increased in comparison to MCF-7 cells, whereas protein expression of Cadherin 1 (CDH1) and SNAIl family transcriptional repressor 2 (SNAI2) was reduced (  . CTGF regulates invasiveness in breast cancer cells. (A) Scheme illustrating 2D invasion experiment using a co-culture transwell invasion assay. BCR = breast cancer cell, ECM = extracellular matrix. (B) Following CTGF siRNA transfection invaded cells under filter were counted in four randomly selected regions, using a co-culture Matrigel invasion assay for 96 h. Data represent mean ± SEM. MCF-7-EMT n = 15, MDA-MB-231 n = 9 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05. (C) Scheme illustrating 3D spheroid invasion assay. Cells were seeded in ultra-low attachment wells and after initial spheroid formation (48 h), spheroids were surrounded by Matrigel matrix and further cultivated. (D) 3 D spheroid assay was performed after transient CTGF siRNA transfection. Invaded area was assessed using ImageJ software and relative area growth was calculated corresponding to respective control. Data represent mean ± SEM.MCF-7-EMT n = 15, MDA-MB-231 n = 9 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05; **P < 0.01. (E) Representative experiment illustrating area measurement of 3D spheroids. Green shape corresponding to initial spheroid size right after adding Matrigel and red shape corresponding to time point 48 h after Matrigel adding. (F) 3 D spheroid assay of MCF-7 cells treated with different combinations of 1 µg/ ml rhCTGF, 1 µg/ml hFN1, 1 µg/ml rhMMP2, and/or 4 nM BB-94 (Batimastat) for 48 h every 24 h. Area growth of spheroids was assessed using ImageJ software and relative area growth was calculated corresponding to untreated control. Data represent mean ± SEM. n = 4-6 using one-way ANOVA and a Dunnett 's multiple comparison test with no matching or pairing between groups was calculated to assess significant differences compared to untreated control. *P < 0.05; **P < 0.01; ***P < 0.005. The graphics A, C, E were created using Microsoft PowerPoint 2016; www.micro soft.com.

GnRH agonist regulates CTGF expression through altered RhoA activity in mesenchymal transformed breast cancer cells.
Most luminal breast cancer will metastasize to bone 27 . Suppression of ovarian function is part of therapy of endocrine-sensitive premenopausal early and advanced hormone receptorpositive breast cancer. Triptorelin, a GnRH agonist, revealed clinical benefit in high-risk patients by suppressing ovarian steroids and it has been investigated in attempt to preserve ovarian function during chemotherapy in young female patients 28 . GnRH receptor is expressed in 50-64% of all human breast cancers [29][30][31][32][33] . Around 15% of all human breast cancers are stated as TNBC, which is associated with high risk recurrence and metastasis 34,35 . Approximately 74% of all TNBC express GnRH receptor 13,36,37 . It was observed that GnRH agonist Triptorelin has impact on breast cancer invasiveness 13,14,38 . Accordingly, we wanted to assess, whether Triptorelin treatment suppresses CTGF expression. Mesenchymal transformed breast cancer cells were treated for 48 h with 10 -9 M or 10 -7 M Triptorelin every 24 h. We found that treatment with 10 -7 M Triptorelin reduced CTGF expression ( It was suggested earlier that RhoA determines mesenchymal cell fate and regulates CTGF cleavage 39 . We wanted to test, if GnRH agonist Triptorelin facilitates reduced invasiveness and increased adhesion by regulating RhoA activity. We found that Triptorelin regulates RhoA activity in a time-dependent manner. After 4 h Triptorelin treatment (10 -7 M) no increased RhoA activity could be detected by active RhoA pulldown. After 24 h a clear increased RhoA activity appeared (Fig. 7F). Furthermore, we found that mesenchymal transformed breast cancer cells treated with a Rho activator exhibit a decreased invasive capacity ( Fig. 7G; RhoA activator II: 22.99 ± 9.922% vs. untreated; P = 0.0401; n = 7), which could be verified for TNBC cells as well (supplementary Fig. 6E). Besides, non-invasive MCF-7 breast cancer cells with transiently suppressed RhoA expression exhibit an increased invasiveness ( Fig. 7H; RhoA -: 183.1 ± 23.45% vs. control; P < 0.05; n = 9). This increased invasiveness was partially reduced by treatment with Triptorelin ( Fig. 7H; RhoA -+ Trip 10 -7 M: 138.48 ± 23.23 vs. control; not significant; n = 9). Furthermore, we tested if this increased invasiveness is due to an increased CTGF expression. We could observe that through reduction of RhoA expression (verification; supplementary Discussion. Tumor metastasis is highly regulated by micro environmental changes. Drugs are needed to modify breast micro environment were tumor cells gain ability to disseminate and bone micro environment, which is the niche where breast cancer cells preferentially colonize and remain in a state of survival and dormancy. Micro environmental modifications may be lethal for isolated, dormant cancer cells, reducing risk of reactivating dormant cells and growth of distant metastases over time is a high priority in preventing metastasis. Here we suggest potential drivers of initial dissemination of tumor cells with regards to bone-directed metastasis. An increased CTGF expression in human breast cancer correlates with poor patient outcome and drug resistance 40 . It was suggested previously that downregulation of CTGF inhibits bone metastasis in a where counter-stained with crystal violet and absorption was measured at 570 nm. Data represent mean ± SEM. MCF-7-EMT n = 3, MDA-MB-231 n = 3 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05; **P < 0.01 (B) Representative images corresponding to A. (C) Extracellular CTGF was reduced using a blockingantibody against CTGF and cell-ECM adhesion was assessed. Data represent mean ± SEM. MCF-7-EMT n = 6, MDA-MB-231 n = 3 using unpaired, two-tailed t-test analysis to respective control (IgG control). *P < 0.05; **P < 0.01 (D) Representative images corresponding to C. (E) MCF-7 cells where treated with recombinant human CTGF (rhCTGF) in different concentrations prior to assessing of cell-ECM adhesion. Data represent mean ± SEM. n = 3 using one-way ANOVA with F = 6.244 and a Dunnett 's multiple comparison test with no matching or pairing between groups. *P < 0.05 (F) Representative images corresponding to E. (G) Following transient transfection mesenchymal transformed and triple negative breast cancer cells were seeded on FITCconjugated gelatin (0.2%). Degradation of gelatin /proteolytic activity results in an increase of fluorescence. Data represent mean ± SEM. MCF-7-EMT n = 3, MDA-MB-231 n = 3 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05 (H) Assessment of proteolytic activity of MCF-7 breast cancer cells after treatment with rhCTGF. Data represent mean ± SEM. M n = 3 using unpaired, two-tailed t-test analysis to respective control (untreated).*P < 0.05. Scale bar gauges 200 µm. www.nature.com/scientificreports/ BMP9-dependent manner 41 . Two major questions have remained: will targeting CTGF help to prevent breast cancer cell dissemination into surrounding tissue, which underlying molecular mechanisms are involved in breast cancer directed bone metastasis. We found that CTGF is highly upregulated in invasive ductal carcinoma and during co-culture of breast cancer cells with osteosarcoma cells. Furthermore, CTGF expression is comparable in bone and mammary gland tissue.
Consistent with recent findings we could assess that an elevated expression of CTGF led to increased cell invasiveness and correlated with bone-directed metastasis. Reducing CTGF expression resulted in a decreased invasion in 2D and 3D invasion assays. It was suggested earlier, that FN1 has a protective function against Data represent mean ± SEM. MCF-7-EMT n = 3 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05; **P < 0.01 (B) Relative quantification of TGFBI, CD44, SPARC, FN1, LOX and FSTL1 mRNA expression in triple negative breast cancer cells treated transiently with CTGF siRNA for 48 h. Data represent mean ± SEM. MDA-MB-231 n = 3 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05 (C) Relative quantification of EMT markers VIM, CDH1, SNAI2 and ZEB1 mRNA expression in mesenchymal transformed breast cancer cells treated transiently with CTGF siRNA for 48 h. Data represent mean ± SEM. MCF-7-EMT n = 3 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05 (D) Relative quantification of EMT markers VIM, CDH1, SNAI2 and ZEB1 mRNA expression in triple negative breast cancer cells treated transiently with CTGF siRNA for 48 h. Data represent mean ± SEM. MDA-MB-231 n = 3 using unpaired, two-tailed t-test analysis to respective control. *P < 0.05.   www.nature.com/scientificreports/ (BB-94). But surprisingly this inhibitor was not effective enough to reverse effect of CTGF treatment, which could be an indicator for a MMP2-independent mechanism.
Loss of intercellular and cell-ECM adhesion allows malignant cells to escape from their site of origin 43 . To further analyze, why cancer cells treated with extracellular CTGF are highly invasive, we analyzed their cell-ECM adhesive and proteolytic abilities. We suggest that reduced CTGF increases cell-ECM adhesion, while ECM degradation was decreased. Increased extracellular CTGF expression led to decreased cell-ECM adhesion and increased ECM degradation. This is supported by previous findings that CTGF induces expression of ECM degradations genes and fibronectin 44 .
MCF-7-EMT cells exhibited increased expression of TGFBI, Twist, Vimentin and N-cadherin, while E-cadherin expression was reduced. Also MCF-7-EMT cells are more invasive 14 . We could furthermore identify, that these mesenchymal transformed breast cancer cells revealed a high ITGαV (CD51) and VCAM-1 (CD106) co-expression compared to non-invasive MCF-7 breast cancer cells. Interestingly, it was suggested, that CTGF stimulates osteosarcoma metastasis by upregulating VCAM-1 expression. Additionally, VCAM-1 may have a role in activation of dormant micro metastasis 9,45,46 . CTGF enhances cell motility in breast cancer through integrinαVβ3-ERK1/2 dependent S100A4 upregulation 47 . We analyzed impact of CTGF on other secretome analysis detected targets and could detect that reducing CTGF expression represses TGFBI, LOX and ZEB1 expression in mesenchymal transformed breast cancer cells. LOX was demonstrated to be involved in collagen I stabilization leading to chemo resistance 48 . It was proposed previously that EMT-TFs SNAI1 and SNAI2 activate TGBFI signaling in breast cancer and that CTGF and SPARC are upregulated as well 49 . Reduced CTGF expression led to increased CD44, SPARC, and FN1 expression in mesenchymal transformed breast cancer cells. CD44 is a stem cell marker and appears to have a dual nature regarding tumor progression and metastasis 50 . SPARC has anticancer effects 51 , inhibits bone metastasis 52 , and was suggested to be involved in same biological pathways than CTGF 53 . We could assess earlier in that study that an increased FN1 expression prevents 3D invasion, even when CTGF is added as well. This could indicate that downregulation of CTGF leads to an increased FN1 expression. We found that suppressed CTGF upregulated FN1 in TNBC cells, and downregulated Vimentin. Except for similar CTGF-dependent FN1 regulation, regulated targets are cell-type specific and could be related to expression of hormone-receptors or to MDA-MB-231 cell line specific mutations. These interesting observations need further evaluation by analyzing CTGF driven mechanism in another TNBC cell line and a hormone receptor positive mesenchymal transformed cell line.
Discovering the prominent role of CTGF during breast cancer invasion by modifying cell adhesion, ECM degradation and FN1 expression, we wanted to test if CTGF can be targeted and elucidated molecular mechanism by which CTGF can be repressed to suppress cell dissemination and colonization at distant sites. We found that GnRH agonist Triptorelin, which is in clinical use for ovarian function suppression of premenopausal breast cancer with high clinical risk of recurrence 28 , and was demonstrated to reduce breast cancer invasion 37 , reduced CTGF expression in mesenchymal transformed breast cancer in a dose-dependent manner. Furthermore, we found that CTGF was downregulated by Triptorelin treatment in TNBC cells. GnRH receptor is expressed in 50-60% of all human breast cancer and to a further extent in approximately 74% of all TNBC 13,36,37 . We could demonstrate that treatment with 10 -7 M Triptorelin led to an increased cell-ECM adhesion in mesenchymal transformed breast cancer cells and TNBC cells as it was detected by CTGF suppression as well. Data represent mean ± SEM. MCF-7-EMT n = 3 using one-way ANOVA with F = 8.366 and a Dunnett 's multiple comparison test with no matching or pairing between groups. **P < 0.01 (B) Quantification and representative experiment of CTGF protein expression after Triptorelin treatment for 48 h (10 -7 M). CTGF band intensity was quantified by densitometry and normalized to GAPDH. Lower panel shows loading control GAPDH that was detected in the same sample and were run in the same gel lane and detected in the same Western blot membrane. Data represent mean ± SEM. MCF-7-EMT n = 3 using unpaired, two-tailed t-test analysis to respective control (untreated). **P < 0.01 (C) Adhesion analysis of mesenchymal transformed breast cancer cells treated with 10 -7 M Triptorelin. Adhesive cells where counter-stained with crystal violet and absorption was measured at 570 nm. Data represent mean ± SEM. MCF-7-EMT n = 5 using unpaired, two-tailed t-test analysis to respective control (untreated). www.nature.com/scientificreports/ It was suggested that RhoA determines lineage fate of mesenchymal stem cells in ECM and that RhoA activity controls CTGF cleavage 39 . Beside, Arguilar-Rojas and colleagues found out that Busrelin, a GnRH agonist, regulates RhoA activity in MDA-MB-231 breast cancer cells thereby decreasing invasiveness 54 . We wanted to examine, if Triptorelin regulates RhoA activity and also if RhoA expression has an impact on CTGF expression. We could observe that Triptorelin induces RhoA activity in a time-dependent manner in mesenchymal transformed breast cancer cells. As expected, invasiveness of mesenchymal transformed breast cancer cells was reduced when RhoA was activated. Later we wanted to assess if reducing RhoA expression has an impact on invasiveness of non-invasive MCF-7 breast cancer. We found that transient RhoA suppression led to increased invasion, which is facilitated through upregulation of CTGF. The increased invasiveness was partially reduced by treatment with Triptorelin, indicating a specific regulation of RhoA by Triptorelin. In addition, CTGF treatment reduced the effect of Triptorelin, indicating a specific regulation of CTGF by Triptorelin. This led to the conclusion that CTGF expression is dynamically regulated through RhoA activation and thereby regulates cell-ECM adhesion.
On molecular level it would be interesting to evaluate, if Triptorelin treatment has an impact on cell plasticity by regulating EMT-TF expression. CTGF activates ERK1/2 signaling through ITGαV cascade 47 and plastic breast cancer cell co-express higher ITGαV and VCAM-1 receptors and exhibit an increased CTGF expression. ERK1/2 appears to be a new treatment option with promising preclinical phase I trials 55,56 . Targeting CTGF when cancer cells gained drug resistance, could help to identify new treatment options. In addition, a new phase III trial study (HOrmonal BOne Effects-2, HOBOE-2) revealed interesting results using zoledronic acid which is approved to treat osteoporosis 57 . In this context it may be worthwhile to examine if zoledronic acid reduces extracellular CTGF, which may open up possibilities for preventing bone metastasis.
Using proteome analysis it was detected, that heat shock proteins (HSP) are dysregulated when breast cancer cells are co-cultured with osteosarcoma cells (supplementary Fig. 2C, D and supplementary table 4). Nonetheless, further evaluation is necessary due to different basal expression of detected potential drivers within different cell lines. It was suggested previously, that cancer cells are more dependent on heat shock protein chaperonage due to an elevated level of misfolded onco-proteins 58,59 . Additionally, inhibiting HSP90 suppresses versatile pro-invasive and proangiogenic pathways 60 . Inhibiting HSP90 led to LATS1 and LATS2 depletion, which led to reduced YAP phosphorylation and decreased CTGF expression 61 . Targeting HSP90 could be of great interest to regulate CTGF expression and HSP90 inhibitors are currently under investigation for metastatic breast cancer [62][63][64] .
It should be noted that breast cancer tissue is a heterogeneous tissue with different cell types beside tumor cells, i.e. fibroblasts and stromal cells 65 . Increased expression of CTGF was found in the fibrous stroma of breast cancers 66,67 . And it has proposed that CTGF plays pivotal role in pathophysiology of many fibrotic disorders and is associated with TGFβ signaling 68 . We found that CTGF protein expression and secretion of MG-63 cells is highly increased as compared to MCF-7 breast cancer cells. It is therefore necessary to further investigate the influence that CTGF expression of fibroblasts and stromal cells has on initiation and progression of tumors in the breast. www.nature.com/scientificreports/ Figure 8 shows our proposed model of CTGF driven invasion in breast cancer cells in vitro. Mesenchymal transformed breast cancer cells with GnRH agonist Triptorelin treatment, CTGF blocking antibody or transiently suppressed CTGF expression reduced invasiveness, increased cell-ECM adhesion and reduced ECM degradation (Fig. 8A). On the other hand, co-cultured non-invasive MCF-7 breast cancer cells or mesenchymal transformed breast cancer cells exhibit an increased CTGF expression, higher invasion, decreased cell-ECM adhesion and increased ECM degradation (Fig. 8B).
In summary, we identified a novel mechanism by which extracellular CTGF drives cell dissemination by regulating cell adhesion, ECM degradation, and regulation of EMT inducing factor TGFBI in vitro. Furthermore, we propose that CTGF is a versatile regulator in breast cancer and facilitates SPARC, LOX, ZEB1, VIM and FN1 expression changes. Moreover, it was assessed that CTGF expression is regulated by RhoA activity. Performed experiments support value of CTGF as therapeutic target for invasive breast cancer, and GnRH agonist Triptorelin could be of value in clinical applications. However, due to the high complexity of metastasis and diverse interactions between different cell types, it is necessary to confirm our findings in an animal model. Transwell co-culture invasion assay. Using co-culture transwell assay as describes earlier 13 , 1 × 10 4 breast cancer cells were seeded in DMEM w/o phenol red, supplemented with 10% cs-FCS into a cell cultural insert (upper well) with a polycarbonate membrane (8 µm pore diameter, Merck Millipore, Cork, Ireland) coated with 30µL of a Matrigel (BD Bioscience, Bedford, MA, USA) solution (1:2 in serum-free DMEM) or gelatin (1 mg/ml in PBS, Sigma). Osteosarcoma cells were seeded (2.5 × 10 4 ) in DMEM supplemented with or without 10% cs-FCS into the lower well (24-well-plate). After 24 h cells were co-cultured for 96 h or 48 h when treated with RhoA activator II. Invaded cells on lower side of inserts were stained with hematoxylin and number of cells in four randomly selected fields of each insert was counted.

Methods
3D spheroid assay. Assessment of 3D cell invasion was pursued as describes earlier with minor changes 69 .
Adherence assay. Cell-ECM adherence was examined by coating 96-well plates with bovine collagen I (30 µL; 0.04 mg/ml; BD Bioscience) for 12 h at 4 °C. Solution was aspirated and plate was left to dry under bench.
RhoA pull-down. RhoA pulldown assay was examined using Rho activation assay biochem kit as described by the manufacturer (BK036-S; Cytoskeleton Inc.). Briefly, 300 µg proteins was loaded with 50 µg Rhotekin rho binding domain (RBD) glutathione agarose bound beads which binds/precipitates specifically active GTP-bond Rho proteins. To quantify active RhoA total RhoA protein was determined. A positive cellular control loaded with non-hydrolysable GTP analog (GTPγS) and a negative control loaded with GDP were determined from each examined sample. To assess functionality of assay one sample was treated with RhoA activator II (CN03, 1 µg/ml, Cytoskeleton). As quantitation estimate for endogenous Rho, His-RhoA protein was run on gel together with examined samples.
Mass spectrometric secretome and proteome analysis. Sample  MS sample processing. For generation of a peptide library, equal amount aliquots from comparable samples were pooled to a total amount of 100 µg, and separated into eight fractions using a reversed phase spin column (Pierce High pH Reversed-Phase Peptide Fractionation Kit, ThermoFisher Scientific). All samples were spiked with a synthetic peptide standard used for retention time alignment (iRT Standard, Schlieren, Schweiz).
Qualitative LC/MS/MS analysis was performed using a Top25 data-dependent acquisition method with an MS survey scan of m/z 350-1250 accumulated for 350 ms at a resolution of 30,000 full width at half maximum (FWHM). MS/MS scans of m/z 180-1600 were accumulated for 100 ms at a resolution of 17,500 FWHM and a precursor isolation width of 0.7 FWHM, resulting in a total cycle time of 2.9 s. Precursors above a threshold MS intensity of 125 cps with charge states 2 + , 3 + , and 4 + were selected for MS/MS, the dynamic exclusion time was set to 30 s. MS/MS activation was achieved by CID using nitrogen as a collision gas and manufacturer's default rolling collision energy settings. Three technical replicates per reversed phase fraction were analyzed to construct a spectral library.
For quantitative SWATH analysis, MS/MS data were acquired using 65 variable size windows 70 across the 400-1,050 m/z range. Fragments were produced using rolling collision energy settings for charge state 2 + , and fragments acquired over an m/z range of 350-1400 for 40 ms per segment. Including a 100 ms survey scan this resulted in an overall cycle time of 2.75 s. Two replicate injections were acquired for each biological sample.
Protein identification was achieved using ProteinPilot Software version 5.0 build 4769 (AB Sciex) at "thorough" settings. MS/MS spectra from combined qualitative analyses were searched against UniProtKB human reference proteome (revision 04-2018, 93.661 entries) augmented with a set of 52 known common laboratory contaminants to identify 217 proteins at a False Discovery Rate (FDR) of 5% in the secretome, and 2,033 proteins at an FDR of 1% for whole proteome analysis. We consciously allowed for a larger FDR in the secretome analysis since identified candidate proteins were further validated during SWATH data extraction and by biochemical experimentation.
Spectral library generation and SWATH peak extraction were achieved in PeakView Software version 2.1 build 11,041 (AB Sciex) using SWATH quantitation microApp version 2.0 build 2003. Following retention time correction using iRT standard, peak areas were extracted using information from MS/MS library at an FDR of 1% 71 . Resulting peak areas were then summed to peptide and finally protein area values, which were used for further statistical analysis.
MS data availability. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 72 partner repository with the dataset identifier PXD021539.

Statistical analysis.
All experiments were performed at least in three biological and technical replicates.
Data were analyzed by GraphPad Prism Software version 8.41 (GraphPad Software Inc., La Jolla, CA/USA) using unpaired, two-tailed, parametric t-test comparing two groups (treatment to respective control) by assuming both populations have same standard derivation or ANOVA one-way analysis when more than two groups were compared. F-values were recorded and a Dunnett 's or a Tukey 's multiple comparison test with no matching or pairing between groups was calculated. P < 0.05 was considered statistically significant.

Data availability
Secretome and proteome data are available via ProteomeXchange with identifier PXD021539. The datasets used and/or analyzed during current study are available from the corresponding author on reasonable request. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.