The cell-surface anchored serine protease TMPRSS13 promotes breast cancer progression and resistance to chemotherapy

Abstract

Breast cancer progression is accompanied by increased expression of extracellular and cell-surface proteases capable of degrading the extracellular matrix as well as cleaving and activating downstream targets. The type II transmembrane serine proteases (TTSPs) are a family of cell-surface proteases that play critical roles in numerous types of cancers. Therefore, the aim of this study was to identify novel and uncharacterized TTSPs with differential expression in breast cancer and to determine their potential roles in progression. Systematic in silico data analysis followed by immunohistochemical validation identified increased expression of the TTSP family member, TMPRSS13 (transmembrane protease, serine 13), in invasive ductal carcinoma patient tissue samples compared to normal breast tissue. To test whether loss of TMPRSS13 impacts tumor progression, TMPRSS13 was genetically ablated in the oncogene-induced transgenic MMTV-PymT tumor model. TMPRSS13 deficiency resulted in a significant decrease in overall tumor burden and growth rate, as well as a delayed formation of detectable mammary tumors, thus suggesting a causal relationship between TMPRSS13 expression and the progression of breast cancer. Complementary studies using human breast cancer cell culture models revealed that siRNA-mediated silencing of TMPRSS13 expression decreases proliferation, induces apoptosis, and attenuates invasion. Importantly, targeting TMPRSS13 expression renders aggressive triple-negative breast cancer cell lines highly responsive to chemotherapy. At the molecular level, knockdown of TMPRSS13 in breast cancer cells led to increased protein levels of the tumor-suppressive protease prostasin. TMPRSS13/prostasin co-immunoprecipitation and prostasin zymogen activation experiments identified prostasin as a potential novel target for TMPRSS13. Regulation of prostasin levels may be a mechanism that contributes to the pro-oncogenic properties of TMPRSS13 in breast cancer. TMPRSS13 represents a novel candidate for targeted therapy in combination with standard of care chemotherapy agents in patients with hormone receptor-negative breast cancer or in patients with tumors refractory to endocrine therapy.

Introduction

The type II transmembrane serine protease (TTSP) family was discovered at the turn of the millennium and constitutes twenty structurally unique multi-domain serine proteases that are anchored directly to the plasma membrane [1, 2]. TTSPs have been shown to play critical and essential roles in normal development, tissue homeostasis, and diseases, including cancer [3–10]. To identify novel TTSP family members that are potentially involved in breast cancer progression, we performed a systematic in silico expression analysis to identify unexplored TTSPs that have increased expression in breast cancer. Our screening and expression analysis revealed that Transmembrane Protease Serine 13 (TMPRSS13, also known as MSPL [11, 12]) has increased expression in invasive breast cancer samples. To date, TMPRSS13 remains among the least characterized TTSPs, and currently, no published study has examined the expression or function of TMPRSS13 in breast cancer.

Previous studies identified that TMPRSS13 can proteolytically modify the viral protein hemagglutinin in vitro, and it was proposed that the protease may play a promotional role in influenza infection [12, 13]. TMPRSS13 has also been shown to cleave and activate the pro-form of hepatocyte growth factor (HGF) in a cell-free system [14]. A physiological role in epidermal development has been demonstrated for TMPRSS13 through characterization of TMPRSS13-deficient mice, which display a mild, transient epidermal defect with delayed barrier acquisition [15]. The phenotype observed in TMPRSS13-deficient mice is similar, albeit less severe, to the phenotypes observed in mice deficient in the TTSP matriptase and in mice lacking the glycophosphatidylinositol (GPI)-anchored cell-surface serine protease, prostasin. Thus, it has been suggested that TMPRSS13 may act either in parallel with or in the same pathway as these proteases [15]. In cancer, we have previously shown that matriptase acts as a tumor promoter in breast and squamous cell carcinoma in vivo [16, 17]. In contrast, prostasin has been described as a tumor suppressor in cellular and xenograft models of breast, prostate, bladder, esophageal, colorectal, oral, and ovarian cancer [18–25] and in a genetic in vivo model of colorectal cancer [20].

In the current study, in vivo loss-of-function strategies using a genetic mouse model along with complementary human cell culture models were performed to examine the role of TMPRSS13 in breast cancer. We demonstrate here that TMPRSS13 expression is increased in breast cancer, and that genetic ablation of TMPRSS13 in mice causes decreased proliferation and increased apoptosis in tumor cells leading to abated cancer progression. TMPRSS13 silencing in human breast cancer cell lines impairs cell survival and invasion, increases apoptosis, and reduces resistance to chemotherapy. Mechanistically, loss of TMPRSS13 leads to increased protein levels of prostasin, and we demonstrate that elevated prostasin impairs cell survival and invasion in human breast cancer cells.

Results

Elevated TMPRSS13 transcript and protein levels in breast cancer

As part of ongoing efforts to determine the expression and function of the TTSP family members in breast cancer, a systematic expression analysis using the Oncomine™ database was performed. The in silico analysis revealed two uncharacterized TTSPs in breast cancer, corin and TMPRSS13, whose transcript levels are significantly increased in invasive ductal carcinoma (IDC) patient samples compared to normal breast tissue (Fig. 1a, Supplementary Fig. S1a, c, and supplementary Table 1). We next assessed whether expression levels of corin or TMPRSS13 are associated with patient survival using the PROGgeneV2 database [26]. Gene expression data from a cohort of breast cancer patients indicated that high expression of TMPRSS13 is a significant prognostic factor for poor overall survival (Fig. 1b) [27]. No differences in overall survival were observed based on low vs. high corin gene expression using the same dataset (Supplementary Fig. S1b) [27]. Based on the in silico findings that transcript levels of TMPRSS13 are increased in IDC patient samples and high expression is associated with decreased overall survival, we set out to investigate TMPRSS13 in breast cancer.

Fig. 1. TMPRSS13 expression in breast cancer.

a Box and whisker plot representing TMPRSS13 mRNA expression data for normal breast (N = 61, median = −0.8) and IDC (N = 389, median = 0.3) from the Oncomine^™ microarray database (TCGA Breast dataset). Boxes show interquartile range, and medians are indicated by horizontal lines, P = 1.73E-22. b Overall survival in patients with low (N = 79, green line) versus high (N = 80, red line) TMPRSS13 transcripts levels. Hazard ratio = 5.16 (1.9–14.9). Log-rank P = 0.002 [27]. The analysis was conducted using the GSE1456_U133B breast cancer patient cohort [27] in the PROGgeneV2 prognostic database and bifurcated into “high” and “low” based on the median gene expression value [26]. c Elevated TMPRSS13 expression in seven breast cancer lines compared to two non-malignant breast epithelial lines assessed by western blotting. Black arrowhead = full-length, glycosylated TMPRSS13 [28], white arrowhead = glycosylated, phosphorylated form of TMPRSS13 [28], gray arrowhead = potential alternate glycosylated isoform of TMPRSS13 [28].

To validate the observed increased TMPRSS13 expression at the protein level, we analyzed human breast cancer cell lines and patient tumors by western blot analysis and immunohistochemistry (IHC), respectively. TMPRSS13 could readily be detected in all seven human breast cancer cell lines analyzed (SK-BR-3, T47D, MCF-7, HCC1954, BT-20, MDA-MB-468, and HCC1937) (Fig. 1c). In contrast, the expression of TMPRSS13 was near or below detection levels in two non-malignant immortalized human breast epithelial cell lines (MCF-10A and hTERT-HME1) (Fig. 1c).

IHC analyses of patient samples were also performed (Intraductal carcinoma; N = 3, Grade I; N = 3, Grade II; N = 13, Grade III; N = 5) (Fig. 2). The majority (12/14) of the non-malignant patient samples displayed little to no detectable TMPRSS13 protein levels (representative example in Fig. 2a). Strong (N = 7), moderate (N = 7), or weak (N = 3) expression of TMPRSS13 was detected in 17 of 24 cancer patient samples and expression was strictly confined to malignant epithelial cells (Fig. 2b–g). Cell-surface staining of TMPRSS13 was observed as expected (Fig. 2b, c), along with diffuse cytoplasmic staining (Fig. 2e–f). The phosphorylation status of TMPRSS13 could potentially contribute to its subcellular distribution, since previous studies have indicated that the main TMPRSS13 species localized on the cell surface is highly phosphorylated [28]. Staining intensity was scored on a scale from 0 to 3 (see “Materials and Methods”), and statistical analyses showed a significant increase in staining intensity in IDC compared to normal tissue or benign lesions, respectively (Fig. 2h and Supplementary Table 2). Taken together, these data show that TMPRSS13 expression is increased in breast cancer at both transcript and protein levels.

Fig. 2. TMPRSS13 protein expression is elevated in IDC.

Staining intensity was assessed using defined scoring criteria (see “Materials and Methods”). a IHC analysis of human patient samples reveals very low to no detectable TMPRSS13 in normal mammary glands (duct indicated with arrowhead and stroma with asterisk). b Strong TMPRSS13 staining in carcinoma cells (Ca, brown, white arrowheads, staining intensity score = 3) with no detectable expression in the stroma (asterisk). c High magnification to illustrate TMPRSS13 cell membrane staining (red arrowheads). d TMPRSS13 moderate staining in carcinoma cells (staining intensity Score = 2). e Example of weak TMPRSS13 staining (staining intensity score = 1). Strong staining (staining intensity Score = 3) (f) with Rabbit-anti-TMPRSS13 but no visible staining (g) in serial sections when the ant-TPMRSS13 antibody was substituted with non-immune rabbit IgG. h Dots represent staining scores from individual patient samples (N = 7 normal/caner adjacent normal, N = 7 benign, N = 24 IDC). Scale bars, 40 μm. Significant differences are observed between normal versus IDC *(p < 0.05) and benign (adenosis and fibroadenoma) versus IDC *(p < 0.05, Kruskal–Wallis ANOVA with Dunn’s post-hoc analysis). Blue line = median.

Increased TMPRSS13 expression in the MMTV-PymT model of breast cancer

Transgenic mouse mammary tumor virus (MMTV) Polyoma middle T (PymT) mice develop multifocal mammary carcinomas with tumor progression that is similar to that seen in human breast carcinomas [29, 30]. To ensure that TMPRSS13 expression in the MMTV-PymT cancer model closely mimics the observations in human breast cancer, the expression of TMPRSS13 in normal mammary glands, mammary tumors, and lung metastases was characterized (Fig. 3). Therefore, we took advantage of the β-galactosidase reporter gene cassette under the transcriptional control of the endogenous Tmprss13 promoter in TMPRSS13^+/− mice [15]. X-gal staining of tissue followed by whole mount and histological examination allows for the precise detection of TMPRSS13 protein expression in mouse tissue (Fig. 3a–d). Whole-mount mammary glands and lungs from heterozygous TMPRSS13^+/− mice with or without the MMTV-PymT transgene were examined and TMPRSS13 was detected in mammary carcinomas (Fig. 3a) and metastatic lung lesions in resected lungs from 129-day old mice (Fig. 3b). In addition, histological analysis revealed that TMPRSS13 is exclusively expressed in the malignant epithelial cells and not the surrounding tumor stroma (Fig. 3d). Importantly, expression was not detected in the normal mammary gland (Fig. 3c). These TMPRSS13 expression patterns align with those observed in human patient samples as described above.

Fig. 3. Delayed tumor development, smaller tumors, and decreased lung metastasis in TMPRSS13^−/− mice.

a–d Endogenous TMPRSS13 visualized by X-gal staining in TMPRSS13^+/−/MMTV-PymT (blue). a TMPRSS13 is expressed in MMTV-PymT mammary carcinomas (arrows) and b in lung lesions (arrows) in whole mounts. LN = lymph node. Histological analysis of X-gal stained whole mounts with no detectable TMPRSS13 in normal mammary ducts (arrow) (c) whereas TMPRSS13 is readily detected (blue, arrows) in carcinoma cells in tumors from MMTV-PymT transgenic mice (d). No detectable expression of TMPRSS13 in the stroma (asterisk). e Kaplan–Meier plot of prospective cohort of littermate TMPRSS13+/MMTV-PymT (N 15) and TMPRSS13^−/−/MMTV-PymT mice (N = 21). Mice were palpated weekly to detect time of tumor formation. All control mice had palpable tumors at 84 days and TMPRSS13^−/−/MMTV-PymT mice at 127 days (P = 0.0013, Log-rank test (Mantel-Cox)). f Tumor burden at 129 days. Mice were euthanized, mammary tumors were resected, and the total tumor weight was recorded. Median values; 3.03 g (TMPRSS13⁺/MMTV-PymT, N = 14) and 1.46 g (TMPRSS13^−/−/MMTV-PymT, N = 18) (p < 0.01, Mann–Whitney U-test). g Growth rate of mammary tumors in TMPRSS13⁺/MMTV-PymT (N = 16) and TMPRSS13^−/−/MMTV-PymT^−/− mice (N = 19) (p < 0.05, Mann–Whitney U-test). h Presence of lung metastasis was determined at the termination of the cohort experiment (day 129). Lung lesions were detected using a dissection microscope and in H&E slides; Control mice 8/12 (67%) and TMPRSS^−/−/MMTV-PymT⁻ mice 4/14 (29%) (p = 0.052, chi-squared test.).

TMPRSS13 deficiency impairs mammary carcinogenesis

To test the effects of genetic ablation of TMPRSS13 on breast cancer progression in vivo, comparative analyses were performed in TMPRSS13-deficient and TMPRSS13-sufficient mice harboring the MMTV-PymT transgene (referred to as TMPRSS13^−/− and TMPRSS13⁺ mice, respectively). Prospective littermate-controlled cohorts were established and monitored biweekly for detection of the first palpable mammary mass (tumor latency). At the 129-day end-point, total mammary tumor burden was assessed and calculated by the total weight of the postmortem excised mammary glands of each individual mouse, and total tumor area was calculated by caliper measurements. In addition, the presence of lung metastases was determined by macroscopic and microscopic examination of lung surfaces and histological examination of lung sections (Supplementary Fig. S2). TMPRSS13 deficiency affected all tumorigenic parameters measured (Fig. 3e–h). Genetic ablation of TMPRSS13 resulted in a significant delay in the formation of palpable mammary tumors in TMPRSS13-deficient mice (Fig. 3e). Moreover, a significant decrease in the overall final tumor burden was observed in TMPRSS13-deficient mice compared to control, with a 52% overall reduction in the final tumor burden in TMPRSS13^−/−/MMTV-PymT mice (median tumor burden of TMPRSS13⁺/MMTV-PymT, 3.03 g, TMPRSS13^−/−/MMTV-PymT, 1.46 g) (Fig. 3f). The growth rate was significantly reduced in TMPRSS13-deficient mice with a median growth rate of 1.53 mm²/day in TMPRSS13-deficient mice, compared to TMPRSS13 control mice that displayed a growth rate of 2.67 mm²/day (Fig. 3g). Furthermore, TMPRSS13-deficient mice displayed a decreased incidence of lung metastasis (Fig. 3h). Taken together these data indicate that TMPRSS13 deficiency impairs mammary cancer progression.

TMPRSS13 deficiency decreases tumor cell proliferation and increases apoptotic cell death in vivo

TMPRSS13 deficiency results in a decreased final tumor burden and tumor growth rate in vivo. To determine whether the decreased tumor growth in TMPRSS13^−/− mice was due to decreased proliferative abilities and/or increased programmed cell death rates, tumor tissues were stained for BrdU incorporation as a measure of cells in S-phase (Fig. 4a), and for presence of cleaved caspase-3 as a marker for cells undergoing apoptosis (Fig. 4c). TMPRSS13^−/− tumors displayed a 44% reduction in the number of proliferating cells compared to control tumors (Fig. 4b). Tumors in TMPRSS13^−/− mice also displayed a marked increase in apoptotic cell death compared to control littermate mice, with a six-fold increase in the number of apoptotic cells in TMPRSS13^−/− mice (Fig. 4d).

Fig. 4. Decreased proliferation and increased apoptosis in TMPRSS13-deficient mammary tumors.

a Representative BrdU staining in TMPRSS13⁺/PymT (left) and TMPRSS13^−/−/PymT (right) mice. b Quantification of the number of BrdU positive, proliferating carcinoma cells relative to total carcinoma cell number in mammary tumors from 129-day old TMPRSS13⁺/PymT (N = 4) and TMPRSS13^−/−/PymT (N = 3) mice. Means: TMPRSS13+ = 16.5% TMPRSS13^−/− = 9.2% (p < 0.05, unpaired Student’s t-test). c Representative staining for cleaved caspase-3 in TMPRSS13⁺/PymT (left) and TMPRSS13^−/−/PymT (right) mice. d Quantification of the number of apoptotic carcinoma cells relative to total carcinoma cell number in 129-day old TMPRSS13⁺/PymT (N = 4) and TMPRSS13^−/−/PymT (N = 4) mouse tumors. Means: TMPRSS13⁺ = 0.59% TMPRSS13^−/− = 3.55% (p < 0.05, unpaired Student’s t-test). Positive and negative cells were counted in 20× microscopy fields by an investigator who was unaware of genotypes. All panels: Error bars represent SD, Scale bars, 50 μm.

TMPRSS13 silencing in human breast cancer cells decreases proliferation and induces apoptosis

To test whether loss of TMPRSS13 affects cancer cell proliferation/survival in human breast cancer cells, TMPRSS13 expression was silenced in three separate breast cancer cell lines. The MCF-7, BT-20, and HCC1937 cell lines were chosen for these analyses because they, in addition to expressing endogenous TMPRSS13 (Fig. 1c), represent different breast cancer subtypes. MCF-7 is a poorly invasive, ER+, PR+ cell line of the luminal subtype. BT-20 and HCC1937 are both highly invasive, triple-negative (ER−, PR−, HER2−) cell lines (TNBC) of the basal subtype, and HCC1937 cells carry mutated BRCA1 [31]. TMPRSS13 was silenced by using two non-overlapping synthetic siRNA duplexes and scrambled GC-matched duplexes were used as controls (Fig. 5a). MCF-7 cells displayed a decrease in cell numbers at day 3 after TMPRSS13 silencing, indicative of cell death. In HCC1937 and BT-20 cells, significantly lower cell numbers were counted in TMPRSS13-silenced cells compared to control cells starting at days 5 and 7, respectively.

Fig. 5. TMPRSS13 silencing in human breast cancer cells decreases proliferation, increases apoptosis, and impairs invasion.

a Relative mean cell numbers by hemocytometer counting following siRNA-mediated knockdown (siRNA 1 and siRNA 2) of TMPRSS13 were assessed over time in serum-containing media supplemented with phenol red in three breast cancer cell lines (MCF-7, BT-20, and HCC1937). Each experimental condition was done in triplicate. Error bars represent SD. Graphs are representative of at least three independent experiments. Asterisks indicate significant difference from scrambled GC-matched control, *p < 0.01. **p < 0.001, ***p < 0.0001 (two-way ANOVA with Tukey multiple comparison test). Black lines = control, red lines = TMPRSS13 knockdown. b–e Western blot analysis of TMPRSS13, PARP and β-actin upon silencing with two non-overlapping RNA duplexes (siRNA 1 and siRNA 2) targeting TMPRSS13 in MCF-7 cells b, BT-20 cells c, HCC1937 cells d, and HCC1954 cells e. f Western blot analysis of TMPRSS13, Bcl-xL, and β-actin upon silencing with two different RNA duplexes (siRNA 1 and siRNA 2) targeting TMPRSS13 in HCC1937 cells. g Quantification of Bcl-xL protein normalized to β-actin. h Invasion assays were performed in siRNA-treated BT-20 cells using two non-overlapping synthetic RNA duplexes (siRNA 1 and siRNA 2) targeting TMPRSS13 and a %GC-matched RNA duplex control (triplicate conditions for each siRNA construct and matched %GC control). 48 h following siRNA treatment, cells were seeded in serum-free media onto transwell inserts coated with 1 mg/mL Matrigel basement membrane gel, inserted into 24-well plates with serum-containing media. Cells were incubated for 16 h and invading cells were fixed and stained. Representative images of Matrigel-coated transwell membranes containing invading cells are shown. i Invading cells were counted, and numbers analyzed by one-way ANOVA, with Tukey’s post-hoc test for multiple comparisons (upper panel). Western blot analysis to confirm TMPRSS13 silencing (lower panel). Error bars = SD, (N = 3, **= p < 0.01).

To determine whether increased apoptosis contributed to the observed decreases in cell numbers, lysates were collected at the indicated days for all three cell lines. The levels of cleaved Poly (ADP-ribose) polymerase (PARP), a marker of apoptosis, were detected by western blot analysis. Loss of TMPRSS13 expression led to increased PARP cleavage in both the luminal-like MCF-7 and the TNBC cancer cell lines BT-20 and HCC1937 (Fig. 5b–d). Inclusion of the HCC1954 (ER−, PR−, HER2+) cell line showed similar increases in cleaved PARP (Fig. 5e) suggesting that increases in apoptotic events upon TMPRSS13 silencing occur in different breast cancer subtypes. In addition, western blot analysis for the anti-apoptotic protein Bcl-xL in HCC1937 cells revealed a decrease in expression (Fig. 5f–g and Supplementary Fig. S3), suggesting that TMPRSS13 may play a role in Bcl-xL regulation. These data, together with the mouse tumor data above, demonstrate that loss of TMPRSS13 expression reduces proliferation and increases apoptotic cell death in breast cancer.

TMPRSS13 promotes breast cancer cell invasive potential

TMPRSS13^−/− mice display decreased incidence of spontaneous breast cancer metastatic lesions compared to control mice (Fig. 3h). For progression from localized to advanced/metastatic breast cancer to occur, cancer cells acquire properties consistent with a propensity to invade into surrounding tissues and distal organs. The role of TMPRSS13 in the invasion potential of TNBC cells was assessed using a transwell assay in which breast cancer cells were seeded on top of a reconstituted matrix in serum-free media and allowed to invade overnight towards serum-containing media in the bottom chamber (Fig. 5h–i). Upon silencing of TMPRSS13 using two non-overlapping synthetic RNA duplexes, a significant decrease in invasive potential was observed in TNBC cells. Importantly, the experiment was carried out 2 days after siRNA transfection when no difference in cell number is observed between control and TMPRSS13-silenced cells (Fig. 5a) minimizing interference from differences in cell proliferation/survival. These data suggest that loss of TMPRSS13 impairs breast cancer cell invasion.

Loss of TMPRSS13 leads to increased levels of the tumor suppressor, prostasin

To uncover candidate TMPRSS13 substrates and mechanisms participating in its pro-oncogenic properties, we examined the effect of TMPRSS13 silencing on the TTSP, matriptase, and the cell surface GPI-anchored serine protease, prostasin. A possible functional connection between these three proteases has previously been proposed in normal epidermal development [15]. Prostasin protein levels were consistently increased upon TMPRSS13 silencing in breast cancer cells (Fig. 6a, b, Supplementary Figs. S4, S5b). In contrast, no consistent changes in matriptase protein levels were observed upon TMPRSS13 silencing (Fig. 6a and Supplementary Fig. S4a).

Fig. 6. TMPRSS13 silencing leads to increased protein levels of the tumor suppressor protease prostasin.

a Western blot analysis of TMPRSS13, matriptase, prostasin, and β-actin upon silencing with two non-overlapping RNA duplexes (siRNA 1 and siRNA 2) b Quantification of prostasin protein normalized to β-actin. c Invasion assays were performed in siRNA-treated HCC1937 cells 96 h post transfection with two non-overlapping synthetic RNA duplexes (siRNA-P1 and siRNA-P2) targeting prostasin and a %GC-matched RNA (in triplicates). Quantification of prostasin knockdown (KD) cells (red) and control cells (blue) invading through Matrigel (16 h) (upper panel) *p < 0.05, **p < 0.01 (one-way ANOVA, with Tukey’s post-hoc test for multiple comparisons). Western blot analysis to confirm prostasin silencing (96 h) (lower panel). Data are representative of three independent experiments. d Whole-cell protein lysates from HEK293T cells expressing empty expression vector (EV), expression vector with V5-tagged full-length human TMPRSS13 (TMPRSS13-V5) and EV, human full-length prostasin and EV, or TMPRSS13-V5 and prostasin were separated by SDS-PAGE under reducing conditions. Prostasin and β-actin were detected by western blotting. A lower molecular weight form of prostasin (cleaved prostasin?) is detected when co-transfected with TMPRSS13. e HEK293T cells expressing EV, prostasin/EV, or TMPRSS13-V5/prostasin, were immunoprecipitated with a mouse anti-V5 antibody and analyzed by western blotting (right panels) using the mouse anti-V5 antibody or a mouse anti-prostasin antibody. Whole-cell lysates were analyzed by western blotting to verity expression before precipitation (input, left panels). IgG H = IgG heavy chain detected by the secondary anti-mouse antibody. f HEK293T cells were transfected with full-length human wildtype prostasin (Pros-WT) and full-length human wildtype TMPRSS13 (T13-WT), T13-WT and zymogen locked prostasin (Pros-ZL), or Pros-WT and catalytically dead TMPRSS13 (T13-CD). Cells were treated with PI-PLC to release GPI-anchored prostasin 48 h post transfection. Released soluble prostasin was incubated with PN-1 (+) or buffer (−) for 1 h at 37 °C and samples were analyzed by SDS-PAGE under reducing conditions and western blotting using a mouse anti-prostasin antibody. The position of prostasin and prostasin-PN-1 complexes are indicated. A recombinant, soluble, truncated active form of purified prostasin protein (rProstasin) with or without PN-1 was included as positive control for complex formation. Lane 1: rProstasin; lane 2: rProstasin with addition of PN-1; lane 3: HEK293T cells expressing Pros-WT and T13-WT; lane 4: HEK293T cells expressing Pros-WT and T13-WT with addition of PN-1; lane 5: HEK293T cells expressing Pros-ZL and T13-WT; lane 6: HEK293T cells expressing Pros-ZL and T13-WT with addition of PN-1; lane 7: HEK293T cells expressing Pros-WT and T13-CD; lane 8: HEK293T cells expressing Pros-WT and T13-CD with addition of PN-1.

One of the contributing tumor-suppressing functions of prostasin has been reported to be maintenance of the epithelial, differentiated state by suppressing epithelial-mesenchymal transition (EMT) and invasive potential [20, 23]. Interestingly, transient knockdown of TMPRSS13 led to a simultaneous increase in both prostasin and zonula occludens-1 (ZO-1), a tight junction protein lost during EMT [32] (Supplementary Fig. S5a, b). We next tested whether prostasin levels alone affect the invasiveness of breast cancer cells by performing an invasion assay upon silencing prostasin expression using two non-overlapping synthetic RNA duplexes. Loss of prostasin expression resulted in a significant increase in invasion (Fig. 6c), confirming that prostasin expression suppresses the aggressive/invasive phenotype in breast cancer cells. Taken together these data indicate that loss of TMPRSS13 expression results in increased expression of the tumor suppressor prostasin and the epithelial marker ZO-1 and a corresponding decrease in invasive potential.

TMPRSS13 interacts with and activates prostasin

To further interrogate a potential molecular connection between TMPRSS13 and prostasin, a series of single and co-expression experiments were performed using a V5-tagged version of human full-length TMPRSS13 (TMPRSS13-V5) and human full-length prostasin (untagged). Prostasin is a ~40-kDa inactive trypsin-like serine protease zymogen that is activated by a single proteolytic cleavage after Arg¹² [33–36]. The prostasin zymogen has been reported to be devoid of enzymatic activity and does not, unlike TMPRSS13 [28], autoactivate in cell-based or in purified biochemical systems [37].

First, HEK293T cells were co-transfected with TMPRSS13-V5 and prostasin, and whole-cell lysates were analyzed by western blotting (Fig. 6d). In the presence of TMPRSS13, prostasin was detected as a form with increased SDS-PAGE mobility, indicative of a potentially cleaved form of prostasin, compared to the prostasin form detected without TMPRSS13 expression. This mobility change required catalytically competent TMPRSS13, since no change was observed when prostasin was co-expressed with a catalytically dead (CD) form of TMPRSS13 (S506A mutation in catalytic triad, CD-TMPRSS13-V5 [28]; Supplementary Fig. S6). Furthermore, a functional activation cleavage site in prostasin was required since a zymogen-locked (ZL) prostasin mutant (R44Q, mutation of zymogen activation cleavage site, ZL-Prostasin [38]) co-transfected with wildtype (WT) TMPRSS13 did not result in mobility shift (Supplementary Fig. S6).

In a second set of experiments to assess potential protein-protein associations between TMPRSS13 and prostasin, lysates from cells co-transfected with TMPRSS13-V5 and prostasin, or prostasin alone, were immunoprecipitated using an anti-V5 antibody followed by western blot analysis. Prostasin was clearly detected in immunoprecipitants of double-transfected cells with no detection of prostasin in the control in which TMPRSS13-V5 was not expressed (Fig. 6e). This finding further indicates an interaction between TMPRSS13 and prostasin.

To confirm the ability of TMPRSS13 to cleave and activate prostasin, a functional prostasin/inhibitor complex formation analysis was performed. Activation of the prostasin zymogen leads to the formation of active two-chain prostasin, which can be distinguished from the zymogen by a small increase in electrophoretic mobility (due to loss of 12 amino acids upon activation cleavage) in SDS-PAGE gels after reduction of the single disulfide bridge that links the two chains [33, 39, 40]. However, the sized-based separation of the two forms is frequently not clearly distinguishable. Therefore, the presence of active prostasin was confirmed by its ability to form a covalent complex with exogenously added protease nexin 1 (PN-1), a cognate inhibitor of prostasin that only binds to the active form of prostasin [33, 39, 40].

HEK293T cells were transfected with WT human full-length prostasin, in which the recombinant prostasin is primarily present in its single-chain pro-form [40], or in combination with WT-TMPRSS13 or mutant forms of either prostasin or TMPRSS13 as controls (Fig. 6f). The recombinant pro-prostasin/prostasin was released from the cell surface with Phosphatidylinositol-Specific Phospholipase C (PI-PLC) to generate a soluble form of the protease, which was then incubated with PN-1. In cells co-expressing prostasin and WT-TMPRSS13, complex formation between WT-prostasin and PN-1 is readily detected with an anti-prostasin antibody (Fig. 6f, lane 4). In contrast, no prostasin/PN-1 complex formation was detected when prostasin was co-expressed with CD-TMPRSS13 indicating that prostasin activation is mediated by active TMPRSS13 (Fig. 6f, lane 8). In addition, expression of the uncleavable, ZL-prostasin mutant control co-transfected with WT-TMPRSS13 showed no prostasin/PN-1 complex formation (Fig. 6f, lane 6).

Targeting TMPRSS13 increases chemosensitivity in TNBC cells

TNBC is typically treated with a combination of radiation therapy and chemotherapy; yet, TNBC patients have a significantly lower 5-year relative survival rate than patients with other breast cancer subtypes. The major reason for low response rates in TNBC patients is chemoresistance and lack of targeted therapies. Therefore, there is an urgent need to identify new targets that can be used for monotherapy or in combination therapy with established drugs. To explore expression of TMPRSS13 in TNBC, we analyzed IDC in silico data sets stratified into TNBC versus non-TNBC. Two representative data sets revealed significantly higher levels of TMPRSS13 in TNBC compared to non-TNBC (Fig. 7a).

Fig. 7. TMPRSS13 levels are increased in TNBC patients and increased chemosensitivity is observed upon TMPRSS13 silencing in TNBC cell lines.

a Increased TMPRSS13 levels in TNBC. Box and whisker plots representing TMPRSS13 mRNA expression data for Patient samples 1; Non-TNBC (N = 250, median = 0.01) and TNBC (N = 46, median = 0.9) (TCGA) and Patient samples 2; Non-TNBC (N = 129, median = −0.9) and TNBC (N = 39, median = 0.4) (Bittner) from the Oncomine™ microarray database. Boxes show interquartile range and medians are indicated by horizontal lines. p < 0.0001 for both plots. b HCC1937 cells were transfected with two non-overlapping RNA duplexes (siRNA 1 and siRNA 2) targeting TMPRSS13 or scrambled GC-matched siRNA (Control) and treated with vehicle or 100 μM paclitaxel for 48 h. Cell survival was measured using Calcein AM read-out values with N = 4 experimental replicates per condition. The combination treatment (purple bar) reduced cell survival by 91% compared to 31% with paclitaxel alone (red bar). Negative control = no cells. Error bars = SD. *p < 0.05 **p < 0.001, ***p < 0.0001 compared to control/vehicle, red asterisks ***p < 0.0001 combination treatment compared to each mono-treatment (one-way ANOVA with Tukey multiple comparison test). Graph representative of three independent experiments. c TMPRSS13-silenced cells or control BT-20 cells were treated with increasing concentrations of paclitaxel or d carboplatin for 48 h (N = 4 experimental replicates per condition). Relative decrease in viability upon drug treatment normalized to vehicle-treated cell viability for each transfection (control, siRNA1, siRNA2) is shown. Error bars = SD. *p < 0.05, ***p < 0.001 (Two-way ANOVA with Tukey multiple comparison test). Graphs representative of three independent experiments per drug treatment. e Corresponding lysates of BT-20 treated with 100 nM paclitaxel or vehicle and f HCC1937 cells treated with 100 μM paclitaxel or vehicle after 48 h of treatment. Detection of TMPRSS13, PARP, and β-actin by western blotting. g HCC1937 cells were transfected with full-length human prostasin cDNA (Pros OE) or an empty expression vector (EV). Twenty-four hours after transfection, cells were treated with 100 μM paclitaxel or vehicle. Cell lysates were collected 24 h after treatment. Detection of prostasin and β-actin by western blotting. h HCC1937 cells were transfected with either human prostasin (Pros OE) or EV, as described above. Twenty-four hours after transfection, cells were treated with increasing concentrations of paclitaxel for 24 h. Cell survival was measured using Calcein AM read-out values. Relative decrease in viability upon drug treatment normalized to vehicle treated EV cell viability for each transfection (EV, Pros OE) is shown. Error bars = SD. *p < 0.05, **p < 0.01 prostasin overexpression (OE) compared to empty vector (EV) (Student’s t-test). Graph representative of three independent experiments.

The clinical therapeutic strategies for management of TNBC include treatment with taxanes (e.g., paclitaxel) and platinum compounds (e.g., carboplatin) [41]. To test whether TMPRSS13 may represent a new target to increase TNBC cell chemosensitivity, experiments were performed to determine the TMPRSS13-dependent response to the chemotherapeutic drugs paclitaxel and carboplatin. The effects of silencing TMPRSS13 alone, drug treatment alone, or combination in HCC1937 and BT-20 TNBC cell lines were assessed by measuring cell viability (Fig. 7b). The HCC1937 cells were highly resistant to paclitaxel treatment (Fig. 7b, red bar) which was expected based on previous reports [42, 43]. HCC1937 cells showed significantly increased sensitivity to paclitaxel upon TMPRSS13 silencing as compared to either treatment alone (Fig. 7b, purple bar). BT-20 cells were also significantly more responsive to paclitaxel (Fig. 7c) and carboplatin (Fig. 7d) upon TMPRSS13 silencing. On the molecular level, analysis of lysates from treated cells by western blotting revealed increased PARP cleavage in combination-treated cells compared to either TMPRSS13 silencing or paclitaxel treatment alone in both cell lines, indicative of increased apoptosis (Fig. 7e and f).

Silencing TMPRSS13 results in increased expression of the tumor suppressor prostasin (Fig. 6a, b), and increased prostasin levels have previously been shown to decrease viability in several cancer types including colon [18], gliomas [44], and ovarian cancer [25]. To test whether increased prostasin expression influences breast cancer cell survival and chemosensitivity, HCC1937 cells were transfected with either WT-prostasin (overexpressing, OE), or empty vector (EV) control, and treated with increasing concentrations of paclitaxel. Western blot analysis was performed to confirm TMPRSS13 expression compared to endogenous TMPRSS13 in EV transfected cells (Fig. 7g). HCC1937 OE cells showed a decrease in cell viability compared to EV transfected cells both under normal culture conditions (no paclitaxel) and upon paclitaxel treatment (Fig. 7h), indicating that prostasin overexpression in itself decreases HCC1937 viability leading to overall increased chemosensitivity. To further analyze whether TMPRSS13 and prostasin are mechanistically linked in breast cancer, we performed double-silencing of TMPRSS13 and prostasin compared to single silencing of either TMPRSS13 or prostasin alone, and control cells were transfected with scrambled siRNA. Silencing TMPRSS13 led to an increase in cleaved PARP protein as expected when assessed by western blot analysis. Silencing of prostasin led to reduced PARP cleavage both when cells received no chemotherapy and when treated with paclitaxel (Supplementary Fig. S7), further substantiating the proposed role of prostasin as a tumor suppressor in TNBC cells congruent with the prostasin overexpression experiments described above. Interestingly, the apoptosis-inducing effect of TMPRSS13 silencing was reversed when prostasin expression was silenced simultaneously, suggesting that TMPRSS13-mediated apoptosis is mechanistically linked to prostasin expression (Supplementary Fig. S7). Taken together, these experiments indicate that loss of TMPRSS13 expression decreases TNBC viability and increases chemosensitivity and that TMPRSS13-mediated regulation of prostasin protein levels may contribute to these effects.

Discussion

We report that TMPRSS13 expression is increased in human and murine breast cancer, and that TMPRSS13 deficiency impairs breast cancer progression in vivo. At the cellular level, loss of TMPRSS13 decreased proliferation and increased apoptotic cell death both in mouse mammary tumors and in human breast cancer cell culture models, and silencing of TMPRSS13 in TNBC cells led to increased sensitivity to chemotherapy. Furthermore, a reduced incidence of lung metastasis was observed in TMPRSS13^−/− mice compared to control mice, and decreased invasion potential was observed upon TMPRSS13 silencing in human TNBC cells. Most extracellular proteases have multiple targets in vitro and it is predicted that many proteases, including TMPRSS13, mediate pleiotropic effects during cancer progression. Two potential contributing mediators of TMPRSS13’s pro-tumorigenic properties, the anti-apoptotic Bcl-xL protein and the tumor suppressor protease prostasin, were identified. Among the anti-apoptotic proteins of the Bcl-2 family, Bcl-xL has the most potent anti-apoptotic activity as it binds to the widest spectrum of pro-apoptotic counterparts [45]. Importantly, its overexpression correlates with chemoresistance in cancer cell lines [46] and in TNBC patient samples [47]. The mechanisms by which TMPRSS13 silencing in TNBC cells decreases protein levels of Bcl-xL awaits further study.

Prostasin has been proposed to be a tumor-suppressing protease based on expression data in human cancer, studies of prostasin loss-of-function, and gain-of-function in cancer cell lines and xenograft models, and genetic mouse studies. Clinical studies have demonstrated that prostasin expression is reduced in multiple cancer types including prostate [21, 48], breast [22], bladder [23], colorectal [20, 49], glioblastoma [44], and gastric cancers [50]. In esophageal and colorectal cancers, loss of expression correlates with poor prognosis [18, 19]. In glioma, breast, and colorectal cancer models, increased levels of prostasin inhibited cancer cell proliferation and invasion, and xenograft experiments showed that prostasin overexpression suppressed colorectal and glioma cell growth in vivo [18, 20, 22, 44]. Conversely, mice with conditional deletion of prostasin in the colon displayed enhanced colorectal carcinogenesis and metastasis accompanied by increased intestinal cell proliferation and invasion and decreased cell differentiation [20]. In TNBC cells, we observed that forced overexpression of prostasin alone led to decreased cell viability both with and without chemotherapy treatment and silencing of prostasin enhanced their invasive potential, further supporting a tumor-suppressive role for prostasin in breast cancer. The precise mechanisms by which prostasin exerts its tumor suppressor functions in vivo are currently unknown. Candidate pathways and mechanisms include suppression of the Sphk1/S1P/Stat3/Akt and Wnt/β-catenin pathways, suppression of EMT, and regulation of epidermal growth factor receptor (EGFR) signaling [18, 20, 23, 44, 51, 52]. Increased prostasin levels observed upon TMPRSS13 silencing may lead to enhanced suppression of these pathways causing a decreased EMT phenotype and increased chemosensitivity in breast cancer. In support of this, we observed an increase in the epithelial marker ZO-1 and decreased invasive potential upon TMPRSS13 knockdown in TNBC. Furthermore, the apoptotic effect observed upon TMPRSS13 silencing was counteracted by simultaneous silencing of prostasin, indicating that induction of apoptosis observed upon TMPRSS13 silencing is mediated through a prostasin-dependent tumor suppressor pathway.

It is currently unclear whether the tumor-suppressive effects of prostasin are mediated by its catalytic activity and thus whether activation of prostasin by other serine proteases is required for prostasin-meditated tumor suppression. Recent studies investigating the molecular mechanisms of prostasin have revealed that prostasin is capable of eliciting physiological functions independent of its proteolytic activity. It has been demonstrated in normal development that while complete deletion of prostasin in mice is perinatally lethal due to epidermal defects [53], knock-in mice with expression of either the catalytically inactive prostasin or the activation site cleavage-resistant (zymogen-locked) endogenous prostasin fully support postnatal survival and homeostasis [54, 55]. Interestingly, the zymogen-locked prostasin accumulates in mouse tissues due to lack of activation site cleavage [54]. Prostasin, like TMPRSS13, is expressed in multiple epithelia and different cancer types [7], and it is conceivable that zymogen prostasin may carry out non-enzymatic functions in different physiological processes and cancer progression.

In our study we observe that TMPRSS13 associates with and is capable of cleaving prostasin at its activation site, which is dependent on TMPRSS13 proteolytic activity. The interacting domains/sites between prostasin and TMPRSS13 are currently unknown. Previous studies of the complex formations between matriptase and prostasin suggested that prostasin forms exosite interactions with the serine protease domain of matriptase [38]. It is possible that TMPRSS13 and prostasin exhibit similar interactions. Future studies are needed to identify specific prostasin-TMPRSS13 interaction sites.

Our data indicate that silencing TMPRSS13 expression results in increased expression of prostasin in TNBC cells. In addition, we observed increased prostasin levels in the colorectal cancer line DLD1 with simultaneous impairment of proliferation and invasion upon TMPRSS13 silencing (F. A. Varela and K. List, unpublished data), indicating that TMPRSS13-mediated prostasin regulation may be relevant in multiple cancer types. One possible molecular mechanism underlying TMPRSS13-mediated modulation of prostasin protein levels is by direct proteolytic modification of prostasin to potentially regulate prostasin protein turnover. Since prostasin is capable of mediating molecular functions in cultured cancer cell models and in vivo independent of its catalytic activity [54, 55], it is possible that cleavage-activation of prostasin, under some conditions, represents a mechanism to regulate prostasin functions via regulation of its total protein levels. Examples of prostasin having catalytic activity-independent functions in cancer include the findings that the expression of the tumor promoters urokinase-type plasminogen activator (uPA), the uPA receptor (uPAR), cyclooxygenase-2 (COX-2), and the inducible nitric oxide synthase (iNOS) were decreased by both the WT and the active-site mutant prostasin in prostate cancer cells [56]. Interestingly, the epithelial differentiation protein, E-cadherin, was increased selectively by the activesite mutant prostasin [56]. Therefore, we hypothesize that one of the pro-oncogenic roles of TMPRSS13 is mediated through the two-chain conversion of prostasin leading to decreased total prostasin expression, which in turn renders the potential tumor-suppressive function of zymogen prostasin diminished. Further studies are needed to elucidate the potential zymogen-specific functions of prostasin in cancer progression, which may provide insights into novel downstream pathways mediated by prostasin that are independent of its catalytic activity.

While TMPRSS13 has been reported to activate pro-HGF in a cell-free system, we did not observe any changes in activation of the HGF-receptor c-MET or the downstream targets AKT and Gab1 upon TMPRSS13 silencing in human breast cancer cell lines (C. Martin and K. List, unpublished data). However, it cannot be ruled out that TMPRSS13 contributes to pro-HGF activation in whole organisms under certain physiological or pathological conditions. Functional studies of TMPRSS13 in other cancers have not yet been described. However, a recent whole-exome sequencing study identified somatic mutations in TMPRSS13 in renal cancer patients and found that these mutations might be predictive of programmed death-ligand 1 (PD-L1) positive expression in cells of this tumor type [57]. It is plausible that TMPRSS13 plays critical roles in additional malignancies.

Based on the data described in this study, TMPRSS13 is critically involved in breast cancer progression in vivo and breast cancer cell survival and invasion in cultured cells. From a translational perspective, the data presented in this study suggest that TMPRSS13 may be a promising therapeutic target for treatment of breast cancer due to its convenient pericellular location and its capacity to regulate fundamental pro-tumorigenic processes. Thus, the findings presented in this paper encourage ongoing efforts towards the development and use of inhibitors of TMPRSS13 for cancer treatment.

Materials and methods

Animals

All procedures involving live animals were performed following institutional guidelines and standard operating procedures. For details on generation of experimental mice, genotyping, and cohort establishment please see “Supplemental Materials and Methods.”

In silico analysis

The gene expression of TTSP family members was analyzed using the TCGA Breast dataset in Oncomine^™. Detailed description in “Supplemental Materials and Methods,” Supplementary Table 1, and Supplementary Fig. 1.

Cell culture of human breast cancer cell lines

Malignant MCF-7 (ER+/PR+, HER2−), HCC1954 (ER−/PR−, HER2+), HCC1937/BT-20 (ER−/PR−,HER2−) cells [58], and non-malignant HEK293T cells (ER−/PR−, HER2−) were cultured as described in “Supplemental Materials and Methods.”

TMPRSS13 and prostasin silencing

Transient knockdown of TMPRSS13 or prostasin expression described in “Supplemental Materials and Methods.”

X-gal staining of mouse tissue

Resected mouse tissue was stained as described [59]. Detailed description in “Supplemental Materials and Methods.”

Immunohistochemistry analyses of mouse tissue

Immunohistochemistry was performed as described [17]. Detailed description in “Supplemental Materials and Methods.”

Breast cancer tissue array

In the BR723 and T088A arrays were processed as described [17] (Detailed description in “Supplemental Materials and Methods”). In the BR723 array, normal/cancer adjacent normal (N = 5), benign (N = 7) (adenosis and fibroadenoma cases), intraductal carcinoma (N = 3), and invasive ductal carcinoma (N = 17) were scored. In the T088A array, normal (N = 2) and invasive ductal carcinoma (N = 4) were scored (age, diagnosis, type, grade, stage, and TNM information in Supplementary Table 2).

Proliferation assays in human breast cancer cell lines

MCF-7, BT-20, and HCC1937 cells were analyzed as described in “Supplemental Materials and Methods.”

Invasion assays

BT-20 cells were used as described in “Supplemental Materials and Methods.”

Immunoprecipitation

HEK293T were used as described. Detailed protocol in “Supplemental Materials and Methods.”

Prostasin transfection in HEK293T cells

Full-length human prostasin cDNA was used for the generation of the S238A and R44Q mutants as previously described [38, 40]. The V5-tagged forms of human full-length TMPRSS13 were generated as described [28]. Transfection and PI-PLC release of prostasin was performed as described [40] (details in “Supplemental Materials and Methods”).

Chemotherapy treatment of breast cancer cells and cell viability assay

HCC1937 and BT-20 cells were analyzed as described “Supplemental Materials and Methods.” Chemotherapy drugs were paclitaxel (T7402 - Sigma-Aldrich, St. Louis, MO) and carboplatin (C2538, Sigma-Aldrich St. Louis, MO).

Statistical analyses

Detailed description of statistical tests used in “Supplemental Materials and Methods.”