Syndecan-1 Promotes Angiogenesis in Triple-Negative Breast Cancer through the Prognostically Relevant Tissue Factor Pathway and Additional Angiogenic Routes

Simple Summary Triple-negative breast cancer is an aggressive subtype of breast cancer characterized by tumor angiogenesis and poor patient survival. Here, we analyzed the function of the cell surface molecule Syndecan-1 in tumor angiogenesis in a 3D cell culture system. As a novel finding, we demonstrate that downregulation of Syndecan-1 reduces angiogenesis by decreasing the amount of angiogenesis factors of the tissue factor pathway. Furthermore, we show that the components of this pathway are associated with the prognosis of breast cancer patients. Our study identifies Syndecan-1 and the tissue factor pathway as novel potential therapeutic targets in the aggressive triple-negative subtype of breast cancer, for which no targeted therapies are currently available. Abstract Triple-negative breast cancer (TNBC) is characterized by increased angiogenesis, metastasis, and poor survival. Dysregulation of the cell surface heparan sulfate proteoglycan and signaling co-receptor Syndecan-1 is linked to poor prognosis. To study its role in angiogenesis, we silenced Syndecan-1 in TNBC cell lines using a 3D human umbilical vein endothelial cell (HUVEC) co-culture system. Syndecan-1 siRNA depletion in SUM-149, MDA-MB-468, and MDA-MB-231 cells decreased HUVEC tubule network formation. Angiogenesis array revealed reduced VEGF-A and tissue factor (TF) in the Syndecan-1-silenced secretome. qPCR independently confirmed altered expression of F3, F7, F2R/PAR1, F2RL1/PAR2, VEGF-A, EDN1, IGFBP1, and IGFBP2 in SUM-149, MDA-MB-231, and MDA-MB-468 cells. ELISA revealed reduced secreted endothelin-1 (SUM-149, MDA-MB-468) and TF (all cell lines) upon Syndecan-1 depletion, while TF pathway inhibitor treatment impaired angiogenesis. Survival analysis of 3951 patients demonstrated that high expression of F3 and F7 are associated with better relapse-free survival, whereas poor survival was observed in TNBC and p53 mutant basal breast cancer (F3) and in ER-negative and HER2-positive breast cancer (F2R, F2RL1). STRING protein network analysis revealed associations of Syndecan-1 with VEGF-A and IGFBP1, further associated with the TF and ET-1 pathways. Our study suggests that TNBC Syndecan-1 regulates angiogenesis via the TF and additional angiogenic pathways and marks its constituents as novel prognostic markers and therapeutic targets.


Introduction
Breast cancer is the most frequently diagnosed malignant tumor in women worldwide, with over 1.6 million cases each year. It is the main cause of cancer-related mortality among female patients [1]. Based on the expression of the surrogate receptors, breast cancer can be stratified into different molecular subtypes [2] and this reflects the biological diversity of breast cancer [3]. The hormone receptor-positive molecular subtype represents 64% of the diagnosed breast cancer cases and is characterized by the expression of estrogen receptor (ER) or progesterone receptor (PR) (ER and/or PR) [4]. About 23% of cases are positive for human epidermal growth factor receptor 2 (HER2), of which 32% are hormone receptor-negative and 67% are hormone receptor-positive. The TNBC subtype is ER-, PR-, and HER2-negative and represents only 13% of breast cancer cases [4]; however, it is an aggressive subtype associated with poor outcome.
Angiogenesis-the formation of new blood vessels from existing vasculature-is a hallmark of cancer and is an essential process for cancer growth and progression [5]. Therefore, interference with the angiogenesis process is of particular relevance as a therapeutic option in cancer. Angiogenesis is a complex process that is highly regulated through a plethora of cues secreted by different tumor microenvironmental cells and multiple signaling pathways, such as vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), and tissue factor (TF) [3,6]. Notably, the activity of angiogenic factors depends on their interaction with proteoglycan coreceptors at the cell surface [7,8].
Syndecan-1 (Sdc-1/CD138), a transmembrane heparan sulfate proteoglycan, is ubiquitously expressed on epithelial cells and functions as a coreceptor for a myriad of chemokines, cytokines, and growth factors [9][10][11]. Sdc-1 is involved in various vital processes relevant to cancer progression, such as cell adhesion, proliferation, migration, invasion, and metastasis [10][11][12][13][14][15]. Of particular relevance to the process of angiogenesis, co-expression of Sdc-1, E-cadherin, and c-Met represents a signature associated with (lymph) angiogenesis-related factors in ductal breast carcinoma in situ [16]. Sdc-1 knockout mice show enhanced corneal angiogenesis mediated by increased leukocyte adhesion [17], whereas Sdc-1 overexpressing mice show abnormal angiogenesis during skin wound repair [18]. A further study showed that treatment with synstatin, a peptide inhibitor that disrupts the interaction of Sdc-1 with αvβ3 and αvβ5 integrins, impairs angiogenesis in vitro and in a mouse model of breast cancer [19]. Sdc-1 is physically associated with VEGFR-2, facilitating VEGF/VEGFR-2 signaling in glomerular endothelial cells [20]. Moreover, in breast cancer, Sdc-1 expression by stromal fibroblasts enhances tumor growth and angiogenesis [21]. We have reported that Sdc-1 is overexpressed and governs the cancer stem cell properties of triple-negative inflammatory breast cancer cells [14]. Interestingly, in the same study, we showed that Sdc-1 downregulation in SUM-149 cells was associated with decreased expression and secretion of the angiogenesis-related factors GRO, IL-6, and IL-8 in 2D models [14]. However, the role of tumor cell-autonomous Sdc-1 expression in the regulation of angiogenesis in the context of tumor-endothelial interactions has not yet been elucidated. Moreover, the role of Sdc-1 in breast cancer angiogenesis has previously not been investigated using unbiased screening approaches. Therefore, we aimed in this study to examine the impact of interference with Sdc-1 expression in TNBC cells on angiogenesis in a 3D co-culture model and to decipher the underlying molecular mechanisms.

Materials and Methods
All supplies and chemicals were from Sigma-Aldrich Chemie GmbH (Taufkirchen, Germany) unless otherwise stated.

Capillary Tube Formation Assay
A 96-well plate was coated with 50 µL/well Matrigel Matrix (Corning, cat. no. 354230, New York, NY, USA) and allowed to solidify at 37 • C incubator for 30 min. HUVECs (2 × 10 4 /well) with a passage number less than 6 were plated on top of the Matrigel in EGM as a positive control or without EGM as a negative control. Cells were monitored using a confocal time-lapse LSM 880 microscope (Zeiss, Oberkochen, Germany). Tube formation analysis was performed using the "Angiogenesis Analyzer for ImageJ" software [24].

In Vitro 3D Co-Culture System
One hour prior to co-culture, 10,000 of HUVECs and either control and Sdc-1-silenced SUM-149 or MDA-MB-231 cells were maintained in 50 µL Endothelial Basal Medium extract (BEM) (Corning ® Matrigel ® Basement Membrane Matrix Growth Factor Reduced, cat. no. 354230) and then cells were grown in a 96-well plate as follows: HUVECs were maintained in BEM with and without FCS, control, and Sdc-1 siRNA transfected cells, HUVECs with control, and Sdc-1 siRNA transfected cells. The 96-well plate was maintained for 24 h in a humidified atmosphere of 5% CO 2 at 37 • C. Cells were stained with Cell Tracker Green CMFDA (Molecular Probes, Thermo Fisher Scientific, cat. no. C2925, Massachusetts, MA, USA) or Cell Tracker Orange CMTMR (Molecular Probes, Thermo Fisher Scientific, cat. no. C2927) according to the manufacturer's instructions and monitored using confocal time-lapse microscopy (LSM 880), followed by quantitative analysis of tube formation as described above. For TF pathway inhibition, the aforementioned steps were analogously repeated in absence and presence of 50 ng/mL recombinant human CellExpTM tissue factor pathway inhibitor (TFPI), (Sigma-Aldrich Chemie GmbH, cat. no. SRP6458-10 UG).

Proteome Profiler™ Human Angiogenesis Antibody Array
The relative expression levels of 55 angiogenesis-related proteins were detected in the cell culture supernatant using a Proteome Profiler™ Human Angiogenesis Antibody Array according to the manufacturer's instructions (R&D Systems, cat. no. ARY007, Minneapolis, MN, USA). Briefly, after the 1-h membrane blocking step, cell culture supernatant (200 µL) was pre-incubated for 1 h with biotinylated detection antibodies cocktail (15 µL) and then added to the membrane before incubating overnight at 4 • C on a rocking platform shaker. After a series of washing steps, the membrane was incubated with 2 mL of diluted streptavidin-horseradish peroxidase for 30 min before chemiluminescence detection on a Fusion SL system (PeqLab/VWR, Langenfeld, Germany).

Quantitative Real-Time PCR
After 48 h from siRNA transfection, total cellular RNA was isolated from SUM-149, MDA-MB 231, and MDA-MB-468 cells using innuPREP RNA mini-Kit (Analytik Jena AG, cat. no. 845-KS-2040050, Jena, Germany) and reverse transcribed into cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, cat. no. 4368814, Foster City, CA, USA). Quantitative real-time PCR was performed in duplicate for each target gene using RT 2 SYBR Green qPCR Primer Assay (Qiagen, cat. no. 330500, Hilden, Germany) and Takyon™ ROX probe qPCR Kit (Eurogentec GmbH, cat. no. UF-RPMT-B0100, Cologne, Germany) in a 7300 real-time PCR detection system (Applied Biosystems). The relative gene expression levels were assessed using the 2 −∆∆Ct method after normalization to the βactin gene as an internal control. Melting curve analysis was conducted to confirm specific product amplification. The following target genes were assessed: VEGF-A, F3, F7, IGFBP1, IGFBP2, EDN1, F2R, and F2RL1. Primer sequences are listed in Supplementary Table S1.

ET-1 and Coagulation Factor III/TF ELISA
A total of 2.5 × 10 5 SUM-149, MDA-MB-231, and MDA-MB-468 cells were seeded in a 6-well plate, then transfected with Sdc-1 siRNA and negative control siRNA. After 24 h, normal medium with FCS was added for only one hour, and the cells were then detached and counted. A total of 1.5 × 10 5 cells were re-suspended and seeded in a 12-well plate with medium without FCS. After 48 h, the supernatant was collected, centrifuged at 4 • C for 15 min, and used for quantification of ET-1 or coagulation factor III/TF by ELISA (Quantikine ELISA endothelin-1 and coagulation factor III/TF kits, R&D Systems, cat. no. DET-100, and DCF300, respectively).

Survival and Expression Analysis
Using the online tool kmplot.com [25], prognostic analysis for angiogenesis-related genes was performed in correlation with the relapse-free survival of 3951 breast cancer patients (see [25] for details on patient data). The median was used as a cut-off value to differentiate between high and low expression and to produce a strong correlation between different categories by reducing the influence of the outliers. The Jet-Set probe set was selected as the recommended ideal probe set for each target gene. Breast cancer patients were molecularly classified based on the expression of the surrogate markers and clinical features, namely ER, PR, and HER2 status, lymph node status, the grade of the tumor, intrinsic subtype, and p53 status. The Affymetrix IDs are 204363_at (F3) 204363_at (F7), 203989_x_at (F2R); 213506_at (F2RL1), and 210512_s_at (VEGF-A).

Protein-Protein Interaction Analysis
STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) (version 11) (http://string-db.org/) (accessed on 29 January 2021) [26] was used to create an in silico protein interaction network for Sdc-1 in relation to the TF pathway signaling-related products; F3, F7, F2R, and F2RL1 in addition to VEGF-A, EDN1, and insulin-like growth factor (IGF) signaling pathways. For the enrichment analysis, STRING implements wellknown classification systems such as Gene Ontology (GO) and KEGG (Kyoto Encyclopedia of Genes and Genomes).

Statistical Analyses
GraphPad Prism 4.02 (GraphPad Software, La Jolla, CA, USA) was used for the statistical analysis of experimental assays, and Student's t-test was used to assess the significant differences between the two groups. For survival analysis, in the R statistical environment, we utilized the Kaplan-Meier Plotter database via the statistical package "survival" to calculate Kaplan-Meier survival curves and the number-at-risk. Furthermore, the hazard ratio (and 95% confidence intervals) and log-rank p were calculated for each gene [27]. The statistical analysis was considered significant when p < 0.05.

Sdc-1 Silencing in TNBC Cells Impairs Capillary-Like Tube Formation in a HUVEC Co-Culture Model
Sdc-1 has an influence on a myriad of biological processes, relevant to tumor progression [28]; however, the role of cell-autonomous Sdc-1 expression in TNBC cells on in vitro angiogenesis has not been explored. Therefore, we employed an in vitro 3D co-culture system comprising HUVECs and TNBC cells. After the successful confirmation of siRNAmediated Sdc-1 silencing by~80% using qPCR ( Figure 1A, Supplementary Figure S1A), control and Sdc-1-silenced TNBC cells were co-cultured with HUVECs and the tube network formation was monitored under 3D co-culture conditions. Our data indicate decreased angiogenesis evident by the reduced HUVEC tube formation when HUVECs were co-cultured with MDA-MB-468 ( Figure 1B Figure 1C) subjected to Sdc-1 depletion, in comparison with those co-cultured with control cells assessed by phase-contrast and confocal immunofluorescence microscopy ( Figure 1B-D). The length of tubes formed by HUVECs was significantly suppressed upon 3D co-culturing with both Sdc-1 knockdown SUM-149 (p < 0.05), MDA-MB-468 (p < 0.05), and MDA-MB-231 (p < 0.01) cells relative to control transfectants ( Figure 1D). Furthermore, a significant reduction of the number of meshes of HUVECs was observed when co-cultured with Sdc-1-  Figure 1C and Supplementary Figure S1B, confocal immunofluorescence microscopy demonstrated that angiogenesis tubes in co-culture were exclusively formed by HUVECs. However, we observed that MDA-MB-231 cells were able to form tubules when cultured alone in 3D on Matrigel, and that tube formation was reduced upon Sdc-1 silencing and abolished when the cells were co-cultured with HUVECs ( Figure 1C,D).

Expression of Coagulation Factor III/TF Is Associated with Worse Relapse-Free Survival in Breast Cancer Patients with Different Classifications
Sdc-1 knockdown affected the expression of different angiogenic factors including the TF pathway constituents' coagulation factor III/TF, Factor VII/F7, F2R/PAR1 and F2RL1/PAR2, and the key angiogenic factor VEGF-A. To assess the prognostic value of the Sdc-1-dependent factors, we correlated their expression with the relapse-free survival (RFS) for 3951 breast cancer patients using the online tool Kaplan-Meier Plotter as we described before [29]. RFS was determined in the whole collective, and in subgroups of patients stratified according to the expression of hormonal receptors (ER and PR) and HER2, intrinsic subtypes, lymph node status, grade status (I-III), and p53 status (wildtype vs. mutant). In agreement with several previous studies establishing a prognostic value for VEGF-A in breast cancer [30,31], a high VEGF-A expression was associated with poor RFS in all patients (n = 3951), ER-patients (n = 801), PR-patients (n = 549), patients with HER2-status (n = 800), triple-negative patients (n = 255), and patients with grade II tumors (n = 901) (Supplementary Table S3). Regarding the TF pathway, in the collective of all 3951 patients, high F3 expression was associated with longer RFS (HR = 0.78, p = 8.4 × 10 −6 ) ( Figure 4A). In contrast, high F3 expression was correlated with shorter RFS, when patients were stratified according to p53 mutation status (HR = 1.85, p = 0.012, Figure 4B) and to p53 mutation with a basal subtype (HR = 2.56, p = 0.032, Figure 4C). For all other stratifications, we did not observe any prognostic value for F3 expression (results not shown). In addition, a trend for a poor survival with high F3 expression was seen in the subgroup of triple-negative tumors of the basal subtype (HR = 1.63, p = 0.06, Figure 4D) ( Table 1).

Expression of Coagulation Factor III/TF is Associated with Worse Relapse-Free Survival in Breast Cancer Patients with Different Classifications
Sdc-1 knockdown affected the expression of different angiogenic factors including the TF pathway constituents' coagulation factor III/TF, Factor VII/F7, F2R/PAR1 and F2RL1/PAR2, and the key angiogenic factor VEGF-A. To assess the prognostic value of the Sdc-1-dependent factors, we correlated their expression with the relapse-free survival (RFS) for 3951 breast cancer patients using the online tool Kaplan-Meier Plotter as we described before [29]. RFS was determined in the whole collective, and in subgroups of patients stratified according to the expression of hormonal receptors (ER and PR) and HER2, intrinsic subtypes, lymph node status, grade status (I-III), and p53 status (wildtype vs. mutant). In agreement with several previous studies establishing a prognostic value for VEGF-A in breast cancer [30,31], a high VEGF-A expression was associated with poor RFS in all patients (n = 3951), ER-patients (n = 801), PR-patients (n = 549), patients with HER2-status (n = 800), triple-negative patients (n = 255), and patients with grade II tumors (n = 901) (Supplementary Table S3). Regarding the TF pathway, in the collective of all 3951 patients, high F3 expression was associated with longer RFS (HR = 0.78, p = 8.4e-06) ( Figure 4A). In contrast, high F3 expression was correlated with shorter RFS, when patients were stratified according to p53 mutation status (HR = 1.85, p = 0.012, Figure  4B) and to p53 mutation with a basal subtype (HR = 2.56, p = 0.032, Figure 4C). For all other stratifications, we did not observe any prognostic value for F3 expression (results not shown). In addition, a trend for a poor survival with high F3 expression was seen in the subgroup of triple-negative tumors of the basal subtype (HR = 1.63, p = 0.06, Figure 4D) ( Table 1).

Expression of F7/Factor VII Is Associated with Better Relapse-Free Survival and Expression of F2R/PAR1 and F2RL1/PAR2 with Worse Relapse-Free Survival in Breast Cancer Patients with Different Classifications
We next aimed at establishing a prognostic value for the additional TF pathway constituents that were expressed in a Sdc-1-dependent manner in our study employing KM Plotter survival analysis. A high expression of factor VII/F7 was associated with improved RFS in the whole collective of all 3951 patients (HR = 0.69, p = 1.9 × 10 −6 ) ( Figure 5A) and patients with luminal A tumors (HR = 0.59, p = 3.4 × 10 −5 ) ( Figure 5B). In contrast, a high expression of F2R/PAR1 and F2RL1/PAR2 was associated with a decreased RFS in breast cancer patients of different subtypes. A high F2R expression was associated with poor prognosis in the ER-(HR 1.38, p = 0.0054) ( Figure 5C) and HER2 + (HR1.85, p = 0.031) ( Figure 5D) classifications, whereas high expression of F2RL1/PAR2 was associated with a worse outcome independent of the subtype (HR1.18, p = 0.0023) ( Figure 5E), in the ER-(HR1.41, p = 0.0031) ( Figure 5F) and HER2 + (HR1.6, p = 0.035) ( Figure 5G) classification, as well as the basal subtype (HR1.34, p = 0.022) ( Figure 5H). Overall, these data highlight the prognostic relevance of the Sdc-1-dependent angiogenic pathways identified in this study. To assess if the differential prognostic value of the Sdc-1-dependent angiogenic factors was associated with altered expression patterns in well-characterized human breast cancer model cell lines, we compared the expression of the factors analyzed in this study in a panel of 33 cell lines using RNASeq data of the EMBL-EBI expression atlas (release 37 March 2021) [32] (Supplementary Figure S3). Apart from the observation of a low expression of IGFBP1 and F7 in many cell lines, no clear subtype-assignment of angiogenic factor expression could be made based on RNA Seq data. Importantly, F3 and VEGFA were strongly expressed across the cell line panel.

Protein Network Analysis of Sdc-1 and Angiogenesis-Related Proteins
STRING network analysis was used to reveal the biological interactions between Sdc-1 and the angiogenesis-related factors, namely F3, F7, VEGF-A, EDN1, F2R, F2RL1, IGFBP1, and IGFBP2 [26]. The analysis showed that there are close interactions between Sdc-1 and VEGF-A and that Sdc-1 interacts with IGF1. VEGF-A, in turn, has close interactions with EDN1, F3, F2R, IGF1, IGFBP1, and IGFBP2. As shown in Figure 6A, F3, F2R, F7, and F2RL1 are highly interconnected and there is a strong interaction between EDN1, F2R, and F2RL1. We analyzed GO enrichment to evaluate the cellular component (orange), molecular functions (green), and biological processes (blue) related to the proteins of interest. The most significant GO terms (p < 0.05) in each classification are presented in Figure 6B and Supplementary Table S4. KEGG pathway analysis (red) is shown in Figure 6C and Supplementary Table S5. Remarkably, Rap1, PI3K-Akt, HIF-1, focal adhesion, Ras, and VEGF signaling pathways, regulation of actin cytoskeleton, and proteoglycans in cancer were found to be linked to the proteins of interest ( Figure 6C and Supplementary Table S5). Overall, the STRING analysis confirms the interconnection of the Sdc-1-dependent factors identified in this study. Curves were analyzed using the log-rank test. Log-rank p-values and hazard ratios (HRs; 95% confidence interval in parentheses) are shown. The corresponding Affymetrix IDs are: F7: 204363_at; F2R: 203989_x_at; F2RL1: 213506_at.

Discussion
In this study, we employed an in vivo-inspired 3D co-culture model of Sdc-1-depleted TNBC cells and HUVECs to study the impact of tumor cell Sdc-1 on angiogenesis in a defined setting. Sdc-1 depletion in the TNBC cells reduced capillary tubule-like formation by HUVECs and suppressed expression and secretion of angiogenesis-related factors. An important and novel finding of this study is that TF pathway components governing angiogenesis of HUVECs are prominently regulated by Sdc-1 expression on TNBC cells. It has been shown that MDA-MB-231 and MDA-MB-468 TNBC cells, as well as clinical TNBC tissue samples, express TF, and RNASeq data support this finding (Supplementary Figure S3) [33]. Given the lack of targeted therapies for TNBC, TF emerges as a promising target. Notably, our KM Plotter survival analysis establishes a prognostic value for several constituents of the TF pathway in breast cancer, emphasizing the clinical relevance of our findings.
Sdc-1 acts as a co-receptor for several angiogenic and growth factors [9,11,13,34]. We and others have shown that Sdc-1 is a modulator of angiogenesis in preclinical and clinical settings. For example, Sdc-1-deficient mice featured increased corneal angiogenesis [18] and delayed dermal wound repair caused by the soluble Sdc-1 ectodomains was observed in mice overexpressing Sdc-1 [18]. Mechanistically, altered recruitment of leukocytes secreting proangiogenic cytokines and an altered proteolytic milieu contributed to the Sdc-1-dependent angiogenesis phenotype in vivo. In breast cancer, Sdc-1 along with Ecadherin and c-met constitutes an expression signature associated with angiogenic and lymphangiogenic factors in ductal carcinoma in situ [16]. Moreover, Sdc-1 of stromal fibroblasts induced breast carcinoma growth and angiogenesis in vivo [21]. Clinically, the same study confirmed this finding in primary human breast cancer specimen, where stromal Sdc-1 expression was significantly associated with tumor vascularity (i.e., both vessel density and total vessel area). The regulatory role of Sdc-1 in angiogenesis network formation was also assigned in another study, where Sdc-1 was shown to be a prerequisite for the activation of αvβ3 and αvβ5 integrins during angiogenesis and breast tumorigenesis in vivo [19,35]. These data suggest that Sdc-1 is able to modulate angiogenesis via several different pathways, which is in accordance with its pleiotropic role in physiology and tumor progression [36]. Apart from the central angiogenesis regulator VEGF-A, our study extends these findings to the tissue factor pathway, which apparently plays an important Sdc-1dependent role in the communication of TNBC and endothelial cells during angiogenesis.
Angiogenesis is a complex process regulated by multiple molecular factors. VEGF-A is a key player regulating angiogenesis of different human malignancies, including breast cancer [37]. Our qPCR data show downregulation of VEGF-A mRNA levels in SUM-149 and MDA-MB-231 cells. Furthermore, relative to control cells, the angiogenesis array analysis revealed that the secretome of Sdc-1-silenced TNBC cells contained lower VEGF-A, which may explain the observed inhibition of in vitro 3D capillary tube formation. This is in agreement with the notion that Sdc-1 knockdown myeloma cells formed fewer and smaller tumors exhibiting diminished levels of VEGF and impaired development of blood vessels in mice [38]. Moreover, Sdc-1 facilitates angiogenesis of multiple myeloma endothelial cells by modulating VEGF/VEGFR-2 signaling [39], providing an additional mechanistic link.
Other important factors regulating angiogenesis are TF, F7, and the protease-activated receptors (F2R/PAR-1 and F2RL1/PAR-2) [40,41]. The local and systemic activation of the coagulation cascade evoked by tumor TF represents the main cause of cancer-associated thrombosis [42], a feature associated with breast cancer [43]. ELISA data revealed significantly decreased secretion of TF in both SUM-149 and MDA-MB-231 cells following Sdc-1 siRNA knockdown. Moreover, significantly downregulated F3 mRNA levels were detected in Sdc-1-depleted SUM-149 and MDA-MB-468 cells. Additionally, our pharmacological inhibition experiments of TF pathway indicate that SUM-149 cells promote capillary tube formation and angiogenesis via the TF pathway. The non-enzymatic activity of heparanase-known to be modulated by Sdc-1 expression-induces F3 expression in cancer cells and impacts cancer progression [44,45]. We have recently discovered an overexpression of heparanase and Sdc-1 in triple-negative inflammatory breast cancer, an aggressive form of breast cancer represented by the SUM-149 cell line employed in this study [14,46]. F3 promotes angiogenesis of the microvascular endothelial cells [47] and the overexpression of tumor F3-induced increased angiogenesis enables human melanoma xenografts to form lung metastasis [48]. Similar to Sdc-1 mode of action, spliced F3 binds to the endothelial cell integrin αvβ3 and α6β1 to evoke tumor angiogenesis via protein tyrosine kinase-2 signaling [49]. Notably, full-length F3 induces tumor angiogenesis via inducing the expression of angiogenesis-related factors IL-8, CXCL-1, and VEGF-A [50]. Consistently, we have previously shown that IL-8 was downregulated upon Sdc-1 silencing in SUM-149 cells [14]. TF expression is essentially regulated by different signaling pathways, including EGF/EGFR, HGF/c-Met, TGF-β/TGF-βR, and VEGF/VEGFR [51]. Expression of these ligands and/or their receptors besides their downstream effectors have been demonstrated to be regulated by Sdc-1 expression. For example, we have shown that EGFR expression was repressed in SUM-149 cells when subjected to Sdc-1 depletion [14]. Of note, co-expression of Sdc-1 and c-Met represents a signature associated with angiogenic and lymphangiogenic-related factors in ductal breast carcinoma in situ [16]. Taken together, it would be conceivable that a Sdc-1/F3/VEGF-A axis may exist, which regulates angiogenesis in a coordinated fashion.
KM survival analysis unraveled that higher F3 and VEGF-A expression correlates with shorter RFS in breast cancer of the TNBC subtype and p53 mutant patients of the basal subtype, respectively. A subtype-specific impact on breast cancer survival was also noted for other constituents of the TF pathway, with F7 being associated with a better prognosis in the whole patient collective and the luminal A subtype, and F2R (PAR1)/F2RL1 (PAR2) acting as negative prognostic indicators particularly in the aggressive ER-and HER2 + subtypes. In agreement, breast cancer TF/F3 is an independent prognostic factor for overall survival and has a predictive value for distant metastasis [52]. Moreover, previous pilot studies have suggested that PAR1 is upregulated in aggressive breast cancer [53,54], and mechanistic studies have demonstrated that PAR1 induction promotes breast cancer progression and metastasis [55]. PAR2 is highly expressed in infiltrative ductal breast cancer tissue [56] and has been shown to be associated with distant disease-free survival in a previous study on 221 breast cancer patients [57]. It is well-established that VEGF is overexpressed in breast cancer tissue specimens, and high VEGF expression is coupled with poor prognostic factors, namely tumor size, TNM stage II and III, and lymph node metastasis, which affect the overall survival [58]. Connecting these findings, another study reported a significant link of F2RL1, F3, and VEGF-A expression in the clinical setting of primary breast cancer, where patients with phosphorylated F3 and F2RL1-expressed tumor tissue specimen are prone to recurrence. Co-expression of phosphorylated F3 and F2RL1 significantly correlates with shorter recurrence-free survival, while co-expression of F2RL1 and VEGF-A is linked with an aggressive phenotype [59]. Our study on over 3000 patients firmly establishes a prognostic role for the TF pathway in breast cancer and provides a novel link to Sdc-1.
The clinicopathological findings are further supported by work in vitro and in model systems: LX-2 hepatic stellate cells-expressing F2RL1 enhance tumor growth via angiogenesis in vivo using a xenograft of hepatocellular carcinoma model [60] and in an intestinal tumor model [61]. In a mouse model of mammary tumor virus-polyoma middle T (PyMT), F2RL1 (PAR2), but not F2R (PAR1), signaling accelerates angiogenesis and tumor growth [62]. These findings further underscore the clinical significance of Sdc-1 expression in regulating multiple interconnected factors (F3, F2RL1, and VEGF-A) involved in various hallmarks of cancers, angiogenesis, and tumor recurrence. This lines up well with our previous study, indicating the pivotal role of Sdc-1 in regulating cancer stem cells in IBC tumors [14]. Angiogenesis of prostate cancer was inhibited possibly via suppressor of cytokine signaling 6 (SOCS6) overexpression-mediated F7 downregulation [63]. Concordantly, our qPCR data analysis uncovered diminished F7 mRNA expression levels in both Sdc-1-silenced SUM-149 and MDA-MB-231 cell relative to control cells. In addition, F2RL1 mRNA levels were decreased in MDA-MB-231 cells upon Sdc-1 depletion. This may serve as a further explanation for alleviating angiogenesis in our experimental model. Interestingly, in retinal ganglion cells, F2RL1 translocates from the plasma membrane to the nucleus facilitating recruitment of the transcription factor Sp1, which in turn leads to induction of VEGF-A expression and subsequent neovascularization [64]. We and others have shown that Sp1 expression is influenced and regulated by Sdc-1 expression [23,65]. Accordingly, we propose a similar mechanism may function to regulate VEGF-A expression by Sdc-1 in our models. Taken together, it can be inferred that the identified panel of factors may represent a further plausible explanation for the suppressive effect of Sdc-1 expression interference on angiogenesis. F2R expression has been found to promote angiogenesis via thrombin-induced VEGF synthesis and secretion in prostate cancer [66]. Surprisingly, we found overexpression of F2R mRNA upon Sdc-1 suppression by siRNA treatment in MDA-MB-231 and MDA-MB-468 cells, but no change of F2R mRNA expression was observed in SUM-149 cells. However, the effect of F2R overexpression on VEGF-A was not observed in our experimental models, as VEGF-A secretion was diminished post-Sdc-1 depletion in SUM-149 cells, suggesting a context-dependent effect. Moreover, as we previously mentioned above, F2R expression is dispensable for mammary tumor growth and angiogenesis in the mammary tumor virus-polyoma middle T (PyMT) mouse model [62].
Apart from TF, ET-1 is one of the major angiogenic factors and correlates with VEGF expression in breast cancer [67,68]. Our qPCR and ELISA data analysis demonstrated that EDN1/ET-1 expression on mRNA levels and in secretome was reduced in Sdc-1-deficient SUM-149 and MDA-MB-468 cells. In accordance, downregulation of EDN1 expression dampened endothelial cell tube formation in human gastric cancer cells [69], implying that Sdc-1 may modulate angiogenesis via modulation of EDN1 expression.
Further clues for angiogenesis regulation are IGFBP1 and IGFBP2. Our qPCR data show downregulation of IGFBP1 mRNA expression levels upon Sdc-1 knockdown in MDA-MB-231 and MDA-MB-468 cells, but IGFBP2 mRNA expression was only repressed in MDA-MB-468 cells. In addition, the secretome analysis revealed downregulation of IGFBP1 and IGFBP2 in MDA-MB-231 cocultures, and of IGFBP1 in MDA-MB-468 cocultures. IGFBP1 acts as a proangiogenic factor contributing to HCV-associated with hepatocellular carcinoma pathogenesis [70] and that elevated IGFBP1 expression is associated with hematogenous metastasis and poor outcome of gastric cancer [71]. Moreover, the nuclear IGFBP2 augments angiogenesis via increasing the promoter transcriptional activity of VEGF in cancer cells [72]. Together, this provides further explanation for impaired angiogenesis upon Sdc-1 expression ablation in cancer cells, which apparently acts in a cell context-dependent manner.
A few caveats are associated with this study. Regarding independent confirmation of the angiogenesis assay with Sdc-1 siRNA sequence #2, a reduction of angiogenesis was seen in SUM-149 and MDA-MB-231 cells; however, these data were only statistically significant in the case of SUM-149. In contrast, downregulation of TF secretion was significant in both models for siRNA #1 and siRNA #2. Regarding the impact of Sdc-1 on angiogenic factor expression, we have highlighted the TF pathway as a novel finding of this study. Nevertheless, additional factors of this work, such as VEGF-A, ET-1 and the IGF-pathway, can be expected to make an additional contribution to the Sdc-1-dependent angiogenesis phenotype, albeit in a cell type-specific manner. Cell type-and context-dependent effects for Sdc-1 are not surprising, as it acts as an important co-receptor for a multitude of signaling pathways, including receptor tyrosine kinases, G-protein-coupled chemokine receptors, morphogen signaling, and possible notch signaling [7,8,11]. Independent of the nature of the signaling pathways, it is remarkable that Sdc-1 depletion in all three TNBC cell lines resulted in a general decrease in the secretion of multiple proangiogenic factors (Figure 2). While we focused our in vitro investigation on the clinically highly relevant subtype of TNBC, our survival analysis demonstrated a prognostic impact of the TF pathway on different breast cancer subtypes, with F3 being of prognostic value only in p53 mutated TNBCs, barely missing significance (p = 0.06) in TNBC in general. This finding suggests that a further mechanistic evaluation of the role of Sdc-1 in connection to the TF pathway in other subtypes of breast cancer may be worthwhile. Our previous observation of an association of Sdc-1 with angiogenic factors in ductal carcinoma in situ of the breast is in line with this idea [16], as are our previous studies demonstrating altered signaling through the MAPK, Notch, Wnt, and IL-6-STAT3-pathways in Sdc-1-depleted cells as a mechanistic element of aberrant cytokine and angiogenic factor expression [9,14,23,73,74].

Conclusions
Our study establishes a novel role for TNBC cell-derived Sdc-1 in regulating components of the TF pathway and additional angiogenic factors, which promote angiogenesis in a 3D in vitro model. Notably, the Sdc-1-dependent factors have a prognostic value in breast cancer beyond TNBC, highlighting their clinicopathological relevance. Pharmacological targeting of Sdc-1, therefore, emerges as a potential therapeutic approach in TNBC and possibly additional subtypes of breast cancer, which does not only affect tumorigenic properties such as cell proliferation, invasive growth, and the cancer stem cell phenotype [14,73,74], but also tumor angiogenesis. The novel connection between Sdc-1 and the TF pathway may even trigger studies in non-oncological areas, including inflammation and thrombosis. Several preclinical approaches targeting Sdc-1 have already yielded positive results [75][76][77][78][79]. For example, application of the Sdc-1-targeting monoclonal antibody OC-46F2 curtails vascular maturation and tumor growth in mouse models of human melanoma and ovarian cancer [80]. In light of our present study findings, these data emphasize the potential therapeutic benefit of targeting cell-autonomous and tumor microenvironmental Sdc-1 against tumor angiogenesis of TNBC, which is worth exploring in future translational studies.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cancers13102318/s1, Figure S1: Sdc-1 depletion in TNBC cells restrains angiogenesis network formation of HUVECs and reduces the secretion of coagulation factor III/TF. Figure S2: Quantitative analysis of angiogenic factor secretion as determined by angiogenesis array. Figure S3. RNASeq expression data of angiogenic factors in 33 breast cancer cell lines as retrieved from the EMBO EBI expression atlas. Table S1: Primers sequences used in the study. Table S2. Summary of angiogenesis array proteome profiling results for conditioned media collected from control and Sdc-1 siRNA-treated SUM-149 cells cultured in 2D or 3D co-culture with HUVEC cells and analyzed by Proteome Profiler™ Human Angiogenesis Antibody Array. Table S3. KM Plotter analysis reveals the prognostic impact of VEGF-A expression in breast cancer patients stratified according to different clinicopathological categories. Table S4. Gene ontology (GO) analysis of Sdc-1, F3, F7, VEGF-A, EDN1, F2R, F2RL1, IGFBP1, and IGFBP2. Table S5. The biological processes associated with Sdc-1, F3, F7, VEGF-A, EDN1, F2R, F2RL1, IGFBP1, and IGFBP2 according to KEGG enrichment analysis.  Institutional Review Board Statement: All patients' survival data were retrieved using the online tool kmplot.com, which combined publically available GEO datasets. The datasets are described in reference [25].

Informed Consent Statement: Not applicable.
Data Availability Statement: All data are available within the article.