Introduction
The Quiescin Sulfhydryl Oxidase 1 (
QSOX1) gene was identified by our group in primary culture of guinea pig endometrial glandular epithelial cells [
1]. The human gene is located on chromosome 1 (1q24) and encodes two major isoforms by alternative RNA splicing: QSOX1S (66 kDa) and QSOX1L (82 kDa) [
2]. The short transcript appears to be ubiquitous, whereas the expression of the longer form seems to be tissue specific [
3]. The longer form of the QSOX1 protein retains a potential transmembrane segment that could allow for anchorage to the membrane. The QSOX1 N-terminus contains a sequence targeting the nascent protein to the endoplasmic reticulum. Moreover, no signal for permanent retention in the endoplasmic reticulum (KDEL sequence) was identified, suggesting an extracellular destination [
4]. In addition, QSOX1 proteins have been detected in the endoplasmic reticulum, the Golgi apparatus and the secretion vesicles [
5]. These proteins can also be found in culture supernatant and in extracellular spaces, confirming that they are secreted [
1].
QSOX1 is the product of an ancient fusion between thioredoxin domains and Flavin Adenine Dinucleotide (FAD) -binding module, ERV/ALR. A first CXXC motif is located in N-terminus and can act as a reducer or an oxidant. The other CXXC motif is located in a FAD domain within C-terminus [
6]. The QSOX1 protein belongs to a family of FAD sulfhydryl oxidases and catalyzes the oxidation of thiols to disulfides.
In vitro, enzymatic studies on avian QSOX1 have demonstrated that this enzyme is able to both catalyze disulfide bridges of a large array of monothiol substrates (such as glutathione) and reduce proteins and peptides [
7,
8]. Moreover, it seems that QSOX1 is not a disulfide isomerase but instead assists the Protein Disulfide Isomerase (PDI) by establishing disulfide links in mature proteins [
9,
10].
Previous reports showed that serum depletion-induced quiescence, as well as cell contact inhibition, led to a QSOX1 mRNA accumulation in guinea pig endometrial glandular epithelial cells [
1] and in human lung fibroblasts [
3]. These experimental data suggest that QSOX1 could be involved in the negative control of the cell cycle. Furthermore, in our laboratory it was demonstrated that over-expression of guinea pig QSOX1-S in MCF-7 cells decreased the cellular proliferation and protected cells against oxidative stress [
11]. It is now known that cellular damage due to an accumulation of Reactive Oxygen Species (ROS) leads to tumorogenesis [
12,
13]. By the reducing activity of its first CXXC motif, QSOX1 could prevent tumorogenesis by down-regulating ROS levels in cells.
Another study suggested that QSOX1 could take part in the cell anchorage mechanism. Indeed, increased mRNA levels have been detected in human lung fibroblast when cell/plate or cell/cell adhesion was disturbed by a mechanical or chemical action [
14].
Several systemic studies have demonstrated an alteration of
QSOX1 expression in cancer cell models. In fact, one study demonstrated the presence of peptide fragments of QSOX1 at highly significant rates in plasma from patients suffering from pancreatic cancer [
15]. Moreover, very recently, it was reported that QSOX1 could promote invasion of pancreatic tumor cell lines by activating matrix metalloproteinase [
16]. In another, a correlation was observed between the overexpression of
QSOX1 and the initiation of prostate tumor growth [
17]. On the other hand,
QSOX1 expression is repressed by epigenetic regulation, especially by histone deacetylation in a cell model of endothelial tumors. Moreover, this down-regulation seems to be necessary for angiogenesis, an essential phenomenon for metastasis development [
18]. These data suggest an involvement of QSOX1 in the mechanisms of carcinogenesis.
In the present study, QSOX1 mRNA expression was investigated in a retrospective cohort of 217 invasive ductal carcinomas (IDC) of the breast. The impact of the QSOX1 expression on characteristic phenotypes of breast cancer cells and tumor growth was subsequently determined.
Discussion
In this study, we provide, for the first time, an insight into the QSOX1 expression level in breast tumors and into its role in breast cancer cells and tumor development.
Even if the enzymatic function of QSOX1 is well described in the literature, its biological function is not clearly established. In fact, QSOX1 expression is differentially modulated in some cancers. It turns out that QSOX1 is overexpressed during the early stages of prostate cancer, and in pancreatic tumor cells [
15‐
17]. However, it has also been shown that QSOX1 is down-regulated in an endothelial cancer cell model [
18]. As such, we decided to investigate the QSOX1 expression in normal and cancerous human breast tissues.
Prior to our study, it was not previously reported where QSOX1 was expressed in breast tissues. IHC results showed that the protein was expressed in the endomembranous system in epithelial cells, but not in stroma cells and in adipocytes. Results were consistent with the fact that QSOX1 is a protein specifically expressed in epithelial cells and secreted [
1,
30]. The differential localization of QSOX1 in normal and cancer cells could be due to post-translational modifications and/or a differential regulation of the expression of QSOX1 isoforms. The signal surrounding tumor cells could also be due to an epithelial cell depolarization, causing a loss of punctuate labeling and the appearance of a diffuse staining. Moreover, QSOX1 is expressed in pancreatic tumor cells and not in non-cancerous peripheral cells [
16]. Together, these results indicate that QSOX1 is not constitutively expressed in different tissues, suggesting that the cellular role for this protein may also be tissue-dependent.
As QSOX1 was expressed in breast cancer tissues, we assessed if its expression level could be correlated to the outcome of breast cancer patients. Thus, we studied the QSOX1 mRNA expression in a retrospective cohort of 217 invasive ductal carcinomas of the breast. It appears that breast tumors presenting good prognostic criteria, such as low histological grade and steroid receptor-positive status, express a high level of QSOX1 mRNA (Figure
1A and Table
1). Furthermore, Kaplan-Meier curves (Figure
1B) showed that patients with high QSOX1 levels presented a significantly better metastasis-free survival compared to patients with low QSOX1 mRNA levels. These data suggest that the QSOX1 expression level could negatively correlate with the aggressiveness of breast tumors.
To determine the extent of association between QSOX1 expression and breast cancer patient outcome, we investigated the roles of QSOX1 on characteristic cancer cell phenotypes in MCF-7 overexpressing QSOX1 and MDA-MB-231, in which QSOX1 expression is knocked down. It must be underlined that QSOX1 levels are higher in the invasive MDA-MB-231 cells compared to the more indolent MCF7 cells. Differences between these two cell lines are not limited to basal QSOX1 expression but also include the expression of several genes, such as steroid hormone receptors and the receptor 2 of the epidermal growth factor [
31]. Therefore, QSOX1 expression could not be responsible in itself for the difference in aggressiveness of these cell lines.
After validation of our cellular models, we investigated the role of QSOX1 on proliferation, clonogenic capacities and invasion. Above all, it is interesting to note that all observed phenotypes obtained with the overexpression of QSOX1 were opposite of those obtained with the QSOX1 knocked-down model. Moreover, in the latter cell model, the intensity of the effect was correlated to the extent of QSOX1 knock-down.
We report in this study that QSOX1 inhibits breast cancer cell proliferation. These data are in accordance with previous studies showing that QSOX1 decreased epithelial and endothelial cell proliferation [
11,
18]. The fact that QSOX1 was induced in quiescent WI38 fibroblasts, but not in cycling lung fibroblasts, suggests that it could act in the negative control of the cell cycle and reinforces the idea that it could downregulate cell proliferation [
3]. Conversely, Katchman and coworkers suggested that a QSOX1 knock-down in different pancreatic cancer cells decreased proliferation. But, cell cycle analyses did not give clear conclusions about QSOX1 role on cell cycle control [
16]. All these results suggested that QSOX1 role on proliferation and cell cycle could be cell type or tumor stage dependent.
The study of cancer cell proliferation is generally performed in parallel with the study of the cell clonogenic capacity, often associated with tumor development and establishment of metastases [
32]. Thus, the fact that QSOX1 reduced cell clonogenicity and anchorage-independent growth is in accordance with our observations on proliferation. In addition, we studied the effect of QSOX1 on cell-matrix adhesion, which is often disrupted in breast cancer. Our findings demonstrate that QSOX1 increased cell adhesion to the extracellular matrix but more specifically early in the adhesion process. Since QSOX1 is implicated in protein folding, we can hypothesize that QSOX1 could enhance the folding or the addressing of newly synthesized proteins implicated in cell adhesion to the matrix. In fact, it is already known that integrin activation can be mediated by thiol disulfide exchanges within the extracellular domain of the beta subunit [
33,
34], a process often performed by the Protein Disulfide Isomerase (PDI) [
35,
36]. QSOX1 could activate integrins directly by creating a thiol-disulfide bond or indirectly via an oxidation of PDI [
9].
We showed in our retrospective cohort of invasive ductal carcinomas of the breast that a high QSOX1 expression is associated with low expression of a well-established marker of migration and invasion phenotype in breast cancer: PAI1 (Table
1) [
37]. In fact, the tumor cell must first degrade basement membrane components before being able to migrate and establish itself in another organ [
38]. We demonstrated that QSOX1 decreased the ability of cells to invade the Matrigel. Furthermore, a low QSOX1 expression level confers to the invading cells this stellate morphology described as characteristic of a high metastatic potential [
39].
Considering that QSOX1 decreased cellular invasion, we studied the activity of MMP-2 and MMP-9, enzymes widely described to be involved in the mechanism of invasion [
29]. QSOX1 decreased both pro and active MMP-2 level in extracellular medium and thus the rate of MMP-2 activity. QSOX1 has also a lesser effect on MMP-9 activity. Nevertheless, QSOX1 had no effect on other MMPs implicated in the basal membrane degradation process, such as MMP-3 and MMP-7 [
29] (data not shown). It seems that QSOX1 reduced invasion by down-regulating MMP-2 activity, since the effect of QSOX1 on invasion was no longer observable in the presence of a specific MMP-2 inhibitor. The involvement of MMP-2 in cell proliferation [
40] could account for our results on this cellular process.
The MMP-2 regulation is very complex and the role of QSOX1 on this MMP remains to be determined. Our results suggest that QSOX1 could attenuate MMP-2 extracellular activity by probably acting on MMP-2 secretion and/or stability without modifying the mRNA level.
Thus, QSOX1 could inhibit MMP-2 activation by promoting cell matrix adhesion. In fact, when cells do not adhere to the matrix, integrin αvβ3 interacts with MMP-2 and so participates to its activation [
29]. The TIMPs (Tissue Inhibitors of MetalloProteinases), present six conserved disulfide bonds involved in their interaction domains with MMPs [
41]. QSOX1 could promote disulfide bond formation in TIMPs, and thus their interaction with MMP-2, leading to its inhibition.
Interestingly, the activation and secretion of MMP-2 could also be positively regulated by oxidative stress through the oxidation of its pro-peptide [
41]. Since QSOX1 protected cell against oxidative stress [
11], it could eventually prevent oxidation of the pro-peptide and so MMP-2 secretion and activation.
Conversely, it has been shown in pancreatic cancer cells that QSOX1 favored the activities but not the secretion of MMP-2 and MMP-9; whereas, QSOX1, in our breast cancer cell models, decreases both the activity and secretion of these MMPs. The discrepancies between our results on breast cancer cells and Katchman's results on pancreatic cells could be explained by differential expression of QSOX1 isoforms and its substrates availability. Furthermore, QSOX1 participates to the redox state control together with a network of proteins like PDI, Ero1, Gluthatione reductase or thioredoxins. It would be interesting to know the expression level of all these actors to understand how QSOX1 could generate opposite effects.
Given these conflicting results between the role of QSOX1 in the breast and pancreas cancer cell models, we performed an
in vivo study. We demonstrated that QSOX1 drastically inhibited tumor growth. Moreover, it should be noted that during the tumor resection, we observed that QSOX1 was unfavorable for tumor invasion in subjacent muscle tissue. By inhibiting MMP-2, an enzyme involved in tumor development, angiogenesis and metastasis [
42], QSOX1 could disfavor breast tumor development and aggressiveness. Indeed, it has been demonstrated in endothelial cancer cell models that QSOX1 could inhibit angiogenesis [
18]. It would consequently also be interesting to study angiogenesis to better understand the effect of QSOX1 on tumor growth through MMP-2 activity regulation.
Acknowledgements
We are grateful to Jean Y Bobin (Centre Hospitalier Lyon Sud), J Datchary (Centre Hospitalier Régional, Annecy) and G De Laroche (Clinique Mutualiste, St Etienne) for providing breast cancer tumors and clinical data. We thank F. Bonnefoy for CIEA NOG Mice breeding. We thank Pierre Y Rysold for his valuable help during immunohistochemistry experiments. We thank J Radom for the gift of coding sequence of QSOX1-S splice variant 2, I Idirene for the gift of HT-1080 cells, F Poncet (Plateforme séquençage, IFR133, Besançon), L Pagnot, V Perez and E Bedel for technical help and Nicole J Le Grand for critical reading of the paper.
This work was supported by a grant from the "Ligue Nationale Contre le Cancer" (Conférence de Coordination InterRégionale-Grand Est). N Pernodet is supported by a fellowship from "Ville de Besançon".
The clinical research work was supported by "Ligue Nationale Contre le Cancer" (Drôme, Haute Savoie, Loire, Saône & Loire and Rhône committees).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
NP performed research, analyzed data and wrote the paper. FH performed research and analyzed data. PA participated in data interpretation and in the writing of the paper. AV participated in experiment design. FD performed clinical data analysis and interpretation. MA performed clinical RT-qPCR analysis. CB provided his expertise in in vivo cancer experiments. JRP performed in vivo experiments execution. GV performed immunohistochemistry data analysis and interpretation. FE provided critical reagents. MJ participated in writing of the manuscript. GD designed research, analyzed data and wrote the paper. All authors have read and approved the manuscript for publication.