Background
Worldwide, of the 250,000 women who are diagnosed annually with ovarian cancer, about 140,000 die within 5 years. Ovarian cancer patients have the lowest survival rate among gynaecological cancers [
1]. Despite a high initial response to debulking surgery and first line chemotherapy consisting of taxane and platinum-based drugs, almost all patients relapse within a few months with a chemoresistance-associated disease, a trend that remains stagnant for the last three to four decades [
1]. Even though different mechanisms of chemoresistance have been described [
2,
3], none have had an individual impact in a clinical setting. Hence, a more detailed knowledge of the biological mechanisms leading to chemoresistance is essential to achieve a better treatment outcome for persistent recurring ovarian cancer patients.
Matrix metalloproteinases (MMPs) are members of the metzincin family that facilitate extracellular matrix (ECM) degradation and thereby promote tumour angiogenesis, invasion and metastasis [
4‐
6]. Conversely, tissue inhibitors of MMPs (TIMPs) are multifunctional proteins, which belong to a family of secreted and ECM bound proteins that naturally inhibit the proteolytic activity of MMPs [
7,
8]. The four TIMP family members, TIMP-1, − 2, − 3 and − 4 share a substantial homology in their sequences [
8]. Of all the TIMPs, TIMP-2 only interacts with a cell-membrane bound MMP, commonly known as MT1-MMP and can act as either an initiator or an inhibitor of MMP-2 activation. As an initiator of MMP-2 activation, the catalytic domain of MT1-MMP binds to the N-terminal region of TIMP-2. This leaves the C-terminal region of TIMP-2 free for binding to the hemopexin-like domain of pro-MMP-2 [
9]. This ternary MT1-MMP-TIMP2 complex facilitates assembly of the extracellular secreted pro-MMP-2 on the cell surface in close proximity to TIMP-free active MT1-MMP. TIMP-free MT1-MMP then cleaves the pro-peptide from cell membrane bound pro-MMP-2 to produce mature MMP-2. In cases where TIMP-2 acts as an inhibitor of MMP-2, the C-terminal end of TIMP-2 acts as a receptor for the C-terminal region of MMP-2, binding of which prevents the interaction of TIMP-2 with MT1-MMP, thereby preventing subsequent MMP-2 activation [
10].
The expression of TIMP-2 is universal in most cell types where it functions as an endogenous inhibitor of MMPs [
11]. In addition to its MMP-2 dependent functions, TIMP-2 can regulate signalling pathways by direct interaction with the cell surface receptors on normal and cancer cells [
12,
13]. TIMP-2 also mediates anti-angiogenic effects by inhibiting endothelial cell migration and invasion through α3β1 integrin [
14]. The tumour microenvironment may provide paracrine cues that regulate these TIMP-2-dependent roles in cancer cells [
15].
The concept that cancer stem cells (CSCs) are small populations of slowly proliferating, drug resistant cells that contribute to tumour initiation and progression, relapse and metastasis has gained substantial importance [
16,
17]. Standard cancer therapies usually target and kill actively proliferating cancer cells thereby reducing tumour burden and decreasing cancer-associated symptoms. This however, does not remove the small population of drug-resistant CSCs. As a result, CSCs can contribute to persistent residual disease even after effective anti-cancer treatments. Hence, it has been proposed that CSCs initiate cancer relapse, which ultimately results in mortality. We and others, have previously shown a correlation between activated STAT3 pathway and the existence of chemoresistance-associated CSCs in residual tumours [
18,
19,
20,
21]. Activated STAT3 was shown to be essential for the sustenance of glioblastoma stem cells [
22], fast proliferating intestinal [
23] and mammary stem cells [
24]. In addition, the activated STAT3 pathway has been associated with the progression and chemoresistance in ovarian and other cancers [
25,
26].
In this study, we report significantly elevated expression of TIMP-2 in high-grade serous compared to benign ovarian tumours. By using siRNA, we examined the effect of TIMP-2 knock down (T2-KD) on MT1-MMP and MMP-2 expression levels, proliferation, invasion and chemosensitivity in Fallopian tube secretory epithelial cells and two ovarian cancer cell lines. We demonstrate that T2-KD cells are associated with increased expression of MT1-MMP and decreased expression and activity of MMP-2 in ovarian cancer and Fallopian tube epithelial cells. The knock down of TIMP-2 resulted in enhanced cell proliferation and increased sensitivity to chemotherapy treatments in ovarian cancer and Fallopian tube epithelial cells. However, increased invasion was only observed in ovarian cancer T2-KD cells but not in T2-KD Fallopian tube cells. Increased sensitivity to chemotherapy in T2-KD cancer cells was associated with an inability to activate the STAT3 pathway and a consequent deficit in chemotherapy resistance and CSC traits. These data reveal a potential role of TIMP-2 in the regulation of proliferation, invasion and chemoresistance in ovarian cancer cells and identifies TIMP-2 as a potential target to circumvent chemoresistance in ovarian cancer.
Discussion
There has been a substantial progress in the last few years in understanding the biological role of MMPs and their endogenous tissue inhibitors (TIMPs) in tumour progression [
41,
6,
42]. It is now evident that TIMP-2, defined as a tissue inhibitor of MMPs, plays an important regulatory role in tumour biology through both MMP-inhibitory and MMP-independent mechanisms [
7,
43,
44]. In this study, we demonstrate that TIMP-2 expression in high-grade serous ovarian tumours is significantly higher than in benign tumours of the same origin. We also demonstrate for the first time that suppression of TIMP-2 expression in Fallopian tube secretory epithelial cells and two ovarian cancer cell lines by an in vitro transient knock down (siRNA) modulates ovarian cell proliferation, invasion and sensitivity to chemotherapy.
The expression of TIMP-2 coincided mostly with the expression of CA125 in high-grade serous tumours, indicating that TIMP-2 is mainly expressed by epithelial tumour cells. However, some diffuse stromal staining of TIMP-2 was also evident in some malignant tumours. In benign tumours, weak expression of TIMP-2 was confined to the ovarian surface epithelium. Previous reports on the expression of TIMP-2 in ovarian tumours and its prognostic impact on the clinical outcome in patients have been ambiguous [
45‐
48]. These indefinite findings regarding the expression of TIMP-2 in ovarian tumours may be due to different affinities of the antibodies used and differences in the experimental methods used for analysis of the studied protein. Alternatively, it could be due to differences in the pathology of high-grade serous tumours as distinct histopathological and genetically different sub-types of high-grade serous tumours have been described [
49,
50].
Since our study demonstrated higher expression of TIMP-2 in high-grade serous ovarian tumours, we used normal secretory Fallopian tube and cancer cell lines that expressed relatively more TIMP-2 than TIMP-1 or TIMP-3, to investigate in vitro the cellular functions of TIMP-2 by transient knockdown. We transiently suppressed the TIMP-2 expression by siRNA in a Fallopian tube secretory epithelial cell line, FT282, and two ovarian cancer cell lines, JHOS2 and OVCAR4. Suppression of TIMP-2 in vitro in the three cell lines was associated with an increase in MT1-MMP expression but inhibition of MMP-2 expression and enzyme activation. Enhanced expression of MT1-MMP and the loss of expression and activation of MMP-2 in response to suppression of TIMP-2 may result due to unavailability of TIMP-2 to bind to MT1-MMP to initiate the activation MMP-2.
We also report for the first time that the suppression of TIMP-2 in ovarian cancer cell lines is associated with a loss of E-Cad mRNA expression with a corresponding increase in the mRNA expression of N-Cad, VIM and a decrease in the mRNA expression of COL12A1. These results are consistent with the study in a non-small cell lung cancer model which showed that overexpression of TIMP-2 upregulated E-Cad expression in in vitro and in vivo models, contributing to the maintenance of cell-cell adhesion and inhibition of tumour growth [
44,
51]. Another study using gastric carcinoma cells has shown an inverse relationship between MT1-MMP expression and E-Cad expression, where knock down of MT1-MMP resulted in increased E-Cad expression which was related to inhibition of proliferation and invasion of gastric cancer cells [
37]. The loss of E-Cad expression in response to suppression of TIMP-2 expression in ovarian cancer cells in our study did not lead to a change in the expression of classical EMT transcription factors downstream of E-Cad such as
SLUG,
SNAIL and
TWIST. However, a decrease in
COL12A1 (alpha chain of type XII collagen, associated with type I collagen) mRNA expression in response to TIMP-2 suppression may be correlated with increased expression of MT1-MMP, a potent protease involved with the degradation of ECM-related fibrillar collagen implicated in tissue remodelling [
39]. In this context, it should be mentioned that the mesothelial layer of the peritoneum is rich in interstitial collagen that provides structural support for optimum tissue assembly and hinders foreign implantation [
52]. MT1-MMP is also essential for the release of ovarian cancer cells (either as sheets of cells or single cells) from primary tumours, which later accumulate as multicellular aggregates in the peritoneum prior to attachment on the mesothelial lining of the peritoneum [
53]. MT1-MMP also promotes migration, cell-matrix detachment, ECM invasion, angiogenesis, formation of multicellular aggregates and growth in three-dimensional collagen matrices in ovarian cancer cells [
54]. In addition, active MT1-MMP facilitates the shedding of ectodomain of MUC16/CA125 in ovarian cancer which restrains adhesion and invasion of cancer cells to the peritoneum [
55]. These MT1-MMP-mediated functions may explain the enhanced invasion observed in the ovarian cancer cell lines with decreased TIMP-2 expression after siRNA treatment. Enhanced proliferation in response to TIMP-2 suppression may result due to enhancement in the expression of cell cycle analogues CDC25B and CDC25C, which may drive cells through G2 and M phases [
56]. However, in the FT282 cell line, knockdown of TIMP-2 may only enhance the M2 phase of the cell cycle through CDC25B. This indicates that TIMP-2 may mediate proliferation in FT282 and ovarian cancer cell lines through different cell cycle mediated mechanisms.
Enhanced proliferation and invasion by knockdown of TIMP-2 in ovarian cancer cell lines is consistent with previous studies that have shown that overexpression of TIMP-2 reduced invasion and proliferation in cells. In melanoma B16F10 cell line TIMP-2 overexpression reduced invasion and angiogenic abilities of these cells [
57]. Overexpression of TIMP-2 in rat smooth muscle cells produced a dose-dependent reduction in proliferation [
58]. Consistent with that study, silencing miR939 produced an overexpression of endogenous TIMP-2 with consequent significant loss of proliferation of non-small cell lung cancer cell line (NSCLC) [
59].
We have previously reported that both paclitaxel and cisplatin, standard chemotherapies used for the treatment of ovarian cancer patients, promotes an increase in the expression of the chemoresistant markers ERCC1 and TUBB3 and the CSC markers CD44, CD133, OCT4A and EpCAM in ovarian cancer cells, [
16,
18,
19,
60]. In this study, we report similar findings in OVCAR4 cells. The increase in chemoresistance and CSC marker expression coincided with the upregulation of TIMP-2 expression and overlapped with the activation of STAT3 pathway in Cont OVCAR4 cells. However, T2-KD OVCAR4 cells, which had reduced TIMP-2 expression, did not exhibit activation of STAT3 or an increase in the expression of chemoresistance and CSC markers in response to chemotherapy treatments. These results may suggest that TIMP-2 and STAT3 activation are intrinsically associated with chemotherapy resistance in ovarian cancer. It can be speculated that the extracellular microenvironment initiated by the cytotoxic damage of the cancer cells may be associated with the activation of STAT3. In this context, the synthesis and secretion of cytokines like interleukin-6 (IL-6), a potent STAT3 activator, have been reported in response to cytotoxic damage in cancer cells [
61]. In addition, the concentration of IL-6 increases after platinum treatment of ovarian tumours, and IL-6 secreted by stromal fibroblasts activates STAT3 and enriches the numbers of ALDH
+ CSCs in residual tumours [
38]. To date, no study has provided a drect link between TIMP-2, STAT3 and chemoresistance. However, a recent paper has demonstrated TIMP-1 mediated chemoresistance in a non-small cell lung carcinoma model via induction of IL-6 secretion [
62]. In addition, in vitro TIMP-1 production in primary mouse hepatocytes was enhanced by IL-6 treatment, but it was less in STAT3-deficient hepatocytes [
63].
Accumulating data suggests that the chemotherapy treated tumours facilitate the selection of therapy-resistant CSCs through elimination of sensitive cells, which makes the residual recurrent tumour aggressive and resistant to therapy [
16,
64]. In addition, chemotherapy labile apoptotic/necrotic cells release intracellular metabolites and soluble cytokines/chemokines and growth factors which therapy-resistant cancer cells may require for the re-growth and establishment of recurrent tumours [
4,
65,
66]. Furthermore, CSCs may escape host immune surveillance by upregulating checkpoint regulators [
67]. The embryonic stem cell marker Nanog interacts with the cancer stem cell marker CD44 to activate the STAT3 pathway in ovarian cancer cells [
68]. An enhanced expression of STAT3 has been reported in recurrent ovarian tumours extracted from metastatic ovarian lesions and ascites-derived tumours, compared to primary tumours and chemonaive ascites-derived tumours [
69,
70]. The genomic and proteomic signatures of recurrent ovarian tumours have been associated with CSCs [
21,
69]. Analyses of clinical samples have shown that the expression of CSC markers (CD44, CD133 and ALDH1A) is low in primary tumours but is enhanced in tumours immediately after chemotherapy treatment but reduces back to their original levels at recurrence, suggesting that the initial expression of CSCs markers identifies chemoresistant cells [
71]. Our recent study has shown that daily oral treatment with Momelotinib (a potent JAK2/STAT3 inhibitor) as a maintenance treatment in conjunction with chemotherapy suppresses STAT3 activation, the CSC traits, and extends the disease-free period by deterring peritoneal spread in a mouse model of ovarian cancer [
31]. Our current study also highlights the important role of the activated STAT3 pathway in CSC-mediated chemoresistance whereby we show a lack of an active STAT3-associated CSC pathway in T2-KD cells.
We report that a STAT3 inhibitor, Momelotinib, inhibits P-STAT3 activation in parental OVCAR4 cells resulting in loss of paclitaxel-induced TIMP-2 upregulation and concomitant loss in the enhancement of the expression of chemoresistance and CSC markers. Although paclitaxel enhanced TIMP-1 expression on parental OVCAR4 cells, Momelotinib had no effect on its expression, strongly suggesting that TIMP-2 mediated chemotherapy-induced STAT3 activation is essential for chemoresistant and CSC phenotypes in ovarian cancer.
TIMPs are secreted by normal and tumour cells and most likely have paracrine effects on the cells in the cellular microenvironment. It is possible that high expression of TIMP-2 in tumours is a pre-requisite for constitutive activation of STAT3 in many cancers or vice versa
, constitutive activation of STAT3 may sustain enhanced TIMP-2 expression in certain cancers. In that context, we have demonstrated persistent activation of STAT3 in advanced-stage ovarian cancer [
72]. Nuclear existence of activated (phosphorylated) STAT3 has been observed in 70% of advanced-stage ovarian cancer and that has been associated with decreased survival [
25]. Whether this persistent STAT3 activation is responsible for enhanced expression of TIMP-2 in ovarian tumours or enhanced expression of TIMP-2 drives constitutive STAT3 activation remains to be determined. In the chemotherapy treatment scenario, it is possible that the chemotherapy-induced acute secretory process, which causes the release of soluble factors (including TIMP-2), may control the activation of STAT3 in ovarian cancer cells. In T2-KD OVCAR4 cells, this aspect of the secretory process may have been compromised resulting in the inability of the cells to activate the STAT3 pathway resulting in the loss of a chemoresistant population. This suggested role of TIMP-2 requires further investigation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.