Kallikrein-related peptidases in triple negative breast cancer: prognostic value of quantitatively assessed KLK8, KLK10 and KLK11 mRNA expression

Background. Triple-negative breast cancer (TNBC) is a breast cancer subtype with poor prognosis and limited targeted therapy options. Multiple KLKs have been described to play key roles in carcinogenesis and metastasis of breast cancer. Purpose. In the present study, the clinical signicance of KLK8, KLK10, and KLK11 mRNA expression in tumor tissue of TNBC patients was investigated. Methods. The mRNA expression levels of KLK8, 10, and 11 were quantied by quantitative PCR and their prognostic values were analyzed in a large, well-characterized TNBC cohort (n = 123). Results. Signicantly positive correlations were observed between all three KLK mRNA levels indicating coordinate expression of these proteases in TNBC. In univariate analyses, both elevated KLK8, KLK10 as well as all combinations of the three factors (KLK8 + KLK10, KLK8 + KLK11, KLK10 + KLK11, KLK8 + KLK10 + KLK11) were signicantly associated with shortened disease-free survival (DFS), while high mRNA levels of KLK11, as well as KLK10 + KLK11 were signicantly associated with shortened overall survival (OS). In multivariate Cox regression analyses, KLK10 and all combined factors remained unfavorable independent predictive markers for DFS, while high KLK11 mRNA expression represented an unfavorable independent predictor for OS.


Background
Triple-negative breast cancer (TNBC) is characterized by the lack of estrogen receptor (ER) and progesterone receptor (PR) expression as well as the absence of human epidermal growth factor receptor 2 (HER2) ampli cation. TNBC accounts for about 17% of breast cancers and is associated with a more aggressive course and a worse prognosis of the disease, compared to hormone-related breast cancer [1].
TNBC occurs more commonly in younger women (< 50 years), and more likely shows distant recurrence.
TNBC tends to disseminate to vital organs like the brain and lungs, leading to a median survival time of fewer than 13 months in metastatic TNBC, compared to other subtypes of breast cancer which preferentially metastasize to the bones [2][3][4]. Furthermore, the molecular features of TNBC exclude these patients from available targeted, anti-hormonal or HER2-targeting therapies.
As approximately 20% of TNBC patients show a BRCA1/2 mutation, genetic testing for all TNBC patients diagnosed at the age of 60 years or younger is recommended. For BRCA1/2 mutation carriers, treatment with PARP inhibitors seems to be a promising therapeutic tool [5]. Apart from this, few additional systemic chemotherapies (like platinum-or taxane-containing chemotherapy regimens or post adjuvant capecitabine therapy) have been introduced into the management of TNBC, with only minor changes in outcome for the price of higher burden in toxicity [1,5,6]. Therefore, valid biomarkers for the development of individualized treatment and improvment of TNBC management are urgently needed.
The kallikrein-related peptidase (KLK) family encompasses 15 closely related serine proteases, which are encoded by the genes KLK1-15 located on the long arm of chromosome 19q13.4 [7]. Accumulating evidence has revealed that KLKs are dysregulated and implicated in tumor-relevant processes in various hormone-dependent malignancies like ovarian, breast and prostate cancer [8,9]. In breast cancer, KLK3-10 and 12-15 but not KLK11 have been reported to serve as diagnostic and prognostic markers [10][11][12][13].
KLK8, a serine protease with trypsin-like activity, has been described to be elevated in several malignancies like lung, colon, cervical, and ovarian cancer. It was reported to e ciently cleave ECMrelated proteins like casein, bronectin, and gelatin, thus promoting proteolysis in the pericellular space of cancer cells and potentially facilitating invasion of tumors [14][15][16][17][18]. Still, in ovarian cancer, elevated KLK8 expression was suggested to be a favorable prognostic marker for prolonged disease-free and overall survival. Additionally, tumors with increased KLK8 concentrations were diagnosed with lower grading and reduced residual tumor mass after surgery [19]. Moreover, overexpression of KLK8 in non small lung cancer cell lines suppresses cell invasion and tends to be a positive predictor for a prolonged tumor-free survival as well [20].
In breast cancer, KLK8 is downregulated on the mRNA level, compared to non-tumorous tissue. Still, high KLK8 mRNA expression in breast tumors was associated with advanced TNM stages, positive nodal status and poor prognosis [21]. Particularly, Michaelidou et al. [13] also showed KLK8 mRNA expression levels to be signi cantly higher in TNBC than in other subtypes of breast cancer.
KLK10 is widely distributed in various human organs and biological uids and is broadly involved in pathophysiological processes [21]. Cumulative evidence indicates that KLK10 represents an unfavorable prognostic marker in colon, renal clear cell, colorectal and gastric cancer [22][23][24], but contrariwise KLK10 has also been associated with tumor-suppression in prostate and testicular cancer by modulation of proliferation and apoptosis. Moreover, KLK10 has been suggested to be a favourable biomarker, e.g. in ovarian cancer [25][26][27].
Similar to KLK8, also KLK10 mRNA as well as antigen expression levels were found to be downregulated in breast cancerous tissue, compared to non-cancerous breast tissue [28,29]. KLK10 has been identi ed to be an important factor in tamoxifen and trastuzumab resistance, thus representing a possible therapeutic target [30,31].
KLK11 seems to be an important enzyme for several physiological processes in humans and is expressed in the tissue of the esophagus, salivary gland, skin, vagina, and cervix [32]. Besides that, it has been described as cancer biomarker: Geng et al. [25,33] reported that both high KLK11 mRNA and protein levels are signi cantly associated with a favorable prognosis, revealing that KLK11 may exert an anti-tumorigenic role in ovarian cancer. KLK11 expression levels are higher in low grade breast cancer tissues and a loss of KLK11 was described as the malignancy increases [34]. However, the role of KLK11 as a biomarker in breast cancer has not yet been explored.
Besides the individual effects of the KLKs, it is known that these proteases are often parts of proteolytic networks and can be coordinately expressed and regulated. It was shown before that KLK8, KLK10 and KLK11 are expressed in breast tissue [35]. Together with the close proximity of these genes within the kallikrein locus [36] and the absence of individual classical promoter regions for each kallikrein [37], this raises the possibility for coordinate expression of these genes.
Altogether, KLK8, KLK10, and KLK11 may have the potential to serve as prognostic biomarkers and represent possible targets for therapy in breast cancer. The present study was designed to further explore the hypothesis, that KLK8, KLK10, and KLK11 are co-regulated in TNBC and that mRNA expression levels of these KLKs or their combinations are prognostic biomarkers.

Patients
One-hundred and twenty-three patients a icted with triple-negative breast cancer (TNBC, n = 123) were enrolled in the study conducted between the years 1988 and 2012 at the Department of Obstetrics and Gynecology, Klinikum rechts der Isar, Technical University of Munich (TUM). Tumor tissues were collected after inspection by pathologists and immediately stored in liquid nitrogen. Tumors were routinely tested for ER, PR, and HER2 expression and categorized as triple-negative breast cancer, if there was a lack of estrogen receptor (ER) and progesterone receptor (PR) protein expression and absence of or low HER2 expression (immunohistochemically determined score 0, 1+; or score 2+ with negative uorescence in situ hybridizations [FISH] test ) [38].
All patients primarily underwent standard surgical procedures including mastectomy or breast conserving surgery and none of them had distant metastasis at the time of surgery. Invasive ductal carcinoma subtype accounted for 87% (109/123) and the rest were more rare subtypes like medullary, lobular and others. Regarding treatment, adjuvant therapy was administered based on consensus recommendations at that time. 72% (89/123) of the patients were treated with anthracycline-or cyclophosphamide-based chemotherapy. The biomarker study in breast tumor tissues was approved by the local Ethics Committee (TU Munich, No. 491/17 S), and performed in accordance with the Declaration of Helsinki. Written informed consent of the study was available for all patients.

Real-time polymerase chain reaction
Universal ProbeLibray probes (Roche) were used in the quantitative polymerase chain reaction (qPCR) (for details see [14].. Gene-speci c primers were designed with the Universal Probe Library Assay Design approximated that of HPRT. As described previously, the 2ΔΔCt method was applied for the calculation of relative KLK8 and KLK11 mRNA expression [14]. As to KLK10, relative KLK10 mRNA expression was quantitatively accessed using the e ciency correction method [33].

Statistics
The correlation of mRNA expression levels of KLKs with clinicopathological parameters of the patients was assessed by the Chi-square test. Survival analyses were performed by constructing Kaplan-Meier curves. The log-rank test is proposed to evaluate group differences in the survival functions. Associations of KLKs and clinical parameters with patients' survival were additionally determined by univariate and multivariable Cox regression analysis and expressed as hazard ratio (HR) as well as its 95% con dence interval (95% CI). The correlations between continuous variables of KLKs were examined with the Mann-Whitney U test and the Spearman rank correlation (r s ). Box plots were drawn to indicate differences. All calculations were performed with the SPSS statistical analysis software (version 20.0; SPSS Inc., Chicago, IL, USA). P values of less than 0.05 were considered statistically signi cant. For further details, see [14;33].

Results
KLK8, 10, and 11 mRNA expression in tumor tissues of triple-negative breast cancer (TNBC) patients and their correlation with clinical parameters KLK8, 10, and 11 mRNA expression levels were quantified by established qPCR assays [25,33] in 123 cases of triple-negative breast cancer tissues. KLK8 mRNA levels ranged The associations between low and high expression groups of the three KLK pairs were also validated by box plot analysis (Mann-Whitney test; p-value in all cases < 0.001, Fig. 1) indicating co-regulation of the three KLK genes in TNBC. Due to this observation, we further dichotomized the mRNA levels into either a KLK8+KLK10, KLK8+KLK11, KLK10+KLK11, or KLK8+KLK10+KLK11 low-expressing group (mRNA levels below the respective cutoff value) versus high-expressing groups (KLK8 and/or KLK10 and/or KLK11 mRNA levels above the respective cutoff value) for statistical analyses.
The associations between relative KLK mRNA expression levels and the established clinical variables, including age, lymph node status, and tumor size, are summarized in Table 1. Neither KLK8 nor KLK10 nor KLK11 mRNA expression levels were associated with clinicopathological parameters of TNBC. Similarly, no statistically significant associations were observed between KLK combinations and clinical parameters (see Suppl. Table 1). Association of KLK8, 10, and 11 mRNA expression levels with disease-free survival (DFS) and overall survival (OS) in univariate Cox regression analysis Univariate Cox regression analysis was performed to evaluate the prognostic relevance of KLKs, the combination thereof, and the clinicopathological parameters regarding the patient outcome in the TNBC cohort (Table 2). Among the clinical variables, advanced age and a positive lymph node status indicated a significantly shorter DFS and OS. The impact of these factors on patient survival was also confirmed by Kaplan-Meier estimation. As shown by the respective survival curves, high KLK8 ( Fig. 2A) and high KLK10 levels (Fig. 2B) were significantly associated with shortened DFS (p = 0.033, p = 0.042, respectively), whereas high KLK11 expression (Fig. 2C) was notably associated with both worse DFS (p = 0.049) and OS (p = 0.013). In case of the combined factors, all combinations represented prognostic factors for DFS (Fig. 3), whereas only the combination KLK10+KLK11 displayed a significant association with OS (see Table 2).
Association of KLK8, 10 and 11 mRNA expression levels with disease-free survival (DFS) and overall survival (OS) in multivariable Cox regression analysis The independence of KLKs as prognostic factors in TNBC was studied in multivariable Cox regression analysis (  c Dichotomization into a low versus high expressing group was performed as described in

Discussion
The members of the kallikrein-related peptidase family (KLKs) have been described to be coordinately expressed and to play a major role in different cancer entities including ovarian and breast cancer [14,25,28,33]. In the present study, we analyzed mRNA expression levels of KLK8, KLK10 and KLK11 in a wellde ned cohort of triple-negative breast cancer (TNBC), and further estimated their prognostic value.
When exploring the prognostic impact of KLK8, 10 and 11 in TNBC, both elevated KLK8 and KLK10 mRNA expression levels were shown to be signi cantly associated with shortened DFS while high expression levels of KLK11 were notably associated with worse OS. In multivariate Cox regression analysis, KLK8 mRNA expression lost its prognostic signi cance for DFS, while KLK10 mRNA expression signi cantly contributed to the base model for DFS.
In line with our ndings, Michaelidou and co-workers [13] found that high KLK8 mRNA expression was notably correlated with poor prognosis and more aggressive subtypes of breast cancer, such as advanced TNM stages and positive nodal status. Particularly, they also showed that KLK8 mRNA expression levels were signi cantly higher in TNBC than other subtypes of this disease.
Likewise, the positive association between KLK10 levels and worse prognosis has been previously observed by Wang and co-workers [31] by analyzing mRNA data form more than four hundred HER2positive breast cancer patients from the TCGA database. Similarly, Luo and co-woekers [40] also investigated the correlation of elevated KLK10 antigen levels with higher risk of disease relapse and cancer-related death in tumor tissues of breast cancer patients. Altogether, more and more evidence is evolving regarding KLK8 and KLK10 to represent unfavorable predictive biomarkers in breast malignancy.
For KLK11, there are only few data on it's prognostic impact in cancer, but it has been suggested to be a favorable prognostic marker e.g. in gastric, prostate, and ovarian cancer [41,42]. Furthermore, high serum levels of KLK11 in ovarian cancer and prostate cancer patients may also be related to a favorable prognosis in these entities [44]. On the other hand, higher KLK11 levels seem to be markers of early disease in breast cancer and vanish during malignancy progression [34]. However, the prognostic value of KLK11 for breast cancer progression needs to be further validated.
When analyzing the prognostic impact of KLK combinations (KLK8 + KLK10, KLK8 + KLK11, KLK10 + KLK11, KLK8 + KLK10 + KLK11), these biomarkers signi cantly contributed to the base model for DFS and thus represented independent predictive markers. This increase in predictive power for survival has already been described in ovarian cancer for the combination of KLK6 and 8 as well as KLK10 and KLK11 [14,26]. KLK8, KLK10, and KLK11 might be involved in synergistic mechanisms affecting tumor pathogenesis in TNBC. Hence, compared to KLKs alone, the combined paired factors allowed better identi cation of patients with unfavorable prognosis in this TNBC cohort.
There have been several studies to explore possible pathophysiological mechanisms including tumorigenicity of KLK8, KLK10 and KLK11. KLK8, abundantly expressed by highly reactive oligodendrocytes in the central nervous system, contributes to the degeneration of the spinal cord after injury and the modulation of neural plasticity development as well as demyelination [44,45]. Patients with Alzheimer's disease expressed a 11.5 higher concentration of KLK8 compared to healtyh patients, which implicates KLK8 being part of the disease development of Alzheimer [46]. Moreover, KLK8 has been identi ed to be part of a proteolytic cascade targeting the skin barrier, thus being involved in dermal wound healing but also in pathological skin diseases, with high KLK8 levels correlating with disease severity [47]. Studies with patients suffering from bowel cancer show an effect of KLK8 in proliferation, invasion and migration in vitro. High KLK8 expression is signi cantly associated to an advanced, more aggressive disease and therefore to a shorter disease free and overall survival [48]. Regarding the ability of tumors to metastasize, KLK8 holds an important role in tumor invasion. The serine protease activity of KLK8 has an impact on bronectin and collagen IV, which are both part of the extracellular matrix. Higher levels of bronectin lead to tumor growth and metastases [49]. The loss of collagen IV results in a different tumor environment and increased tumor invasion in bowel cancer [50].
KLK10 is commonly supposed to promot tumor invation and metastasis via degrrading extracellular matrix proteins [51]. Furthermore, Li et al. [52] observed that KLK10 might lower the cisplatin-induced apoptosis, indicating that KLK10 might be coorelated with cisplatin resistance and serve as a potential therapeutic target in esophageal cancer cells. Moreover, KLK10 expression has been found to be signi cantly enhanced in established herceptin-resistant gastric cancer cell lines [53], in where the lower expression of KLK10 or suppressing PI3K/AKT pathway was able to decrease herceptin resistance and the overexpression showed the reverse effects, indicating that KLK10 displayed a key role in herceptin resistance by impacting the PI3K/AKT signaling pathway.
Possible mechanisms of the unfavorable biological role of KLK11 in malignancies have been explored as well. Xu et al. [54] found that a knockdown of KLK11 expression signi cantly inhibits growth and induces apoptosis of tumor cells in colorectal cancer cells in vitro. There, KLK11 inhibition activates the apoptosis signaling pathway by upregulating Bax expression and diminishing Bcl-2 expression. Moreover caspase-3 activity was induced, resulting in cancer cell death [54]. Furthermore, Zhang et al. [55] showed that the downregulation of KLK11 could inhibit the phosphatidylinositol-3-kinase (PI3K)/AKT pathway, which plays a critical role in tumor cell proliferation, invasion, and metastasis, thus affecting oxaliplatin resistance, suppressing cell growth and activating apoptosis in colorectal cancer cells [56]. Moreover, KLK11 cleaves insulin-like growth factor (IGF) binding protein 3 (IGFBP-3) and modulates functions of IGFs [34], which play important roles in the tumor growth and cell survival [57].

Competing interests
The authors declare that they have no competing interests.

Funding
The present study was supported in part by a grant from the Deutsche Forschungsgemeinschaft, DO 1772/1-1 (JD), and from the Wilhelm Sander-Stiftung, contract number 2016.024.1 (TD). The funding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data, in the writing of the manuscript, and in the decision to publish the results.
Authors' contributions YL and SP were responsible for acquisition of data, formal analysis, interpretation of data and drafting the manuscript. VM was responsible for conception and design, project administration, formal analysis, supervision and writing -review and editing. JD was responsible for conception and design, data curation, project administration, supervision, drafting the manuscript, review & editing. TD was responsible for acquisition of data, formal analysis, interpretation of data, and review & editing. XG, WG and MK were responsible for acquisition of data and writing -review & editing. All authors approved the nal manuscript version.

Figure 1
Correlation of KLKs mRNA expression in tumor specimens of triple-negative breast cancer patients. By applying the Mann-Whitney test, KLK8 mRNA expression levels were found to be signi cantly correlated with KLK10 mRNA expression levels (A, rs = 0.612, p < 0.001). Similarly, a signi cantly positive correlation was also observed between KLK8 and KLK11 (B, rs = 0.460, p < 0.001) and between KLK10 and KLK11 (C, rs = 0.722, p < 0.001).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplFig1YLiu.pdf SupplFig2YLiu.pdf SupplTab1YLiu.pdf