Background
STAT1 has been described as a tumor suppressor because of its function as a mediator of IFN-γ - dependent immunosurveillance [
1]. This anti-tumor activity of STAT1 appears to be particularly important at the onset of tumor formation and is supposed to lead to the elimination of transformed cells by the innate and adaptive immune system. At a cellular level, STAT1 can exert this function via shaping the immune effector phenotype [
2,
3]: In dendritic cells, genes required for antigen processing and presentation are up-regulated, and in macrophages cytotoxic activity is increased, e.g. by induction of
iNOS expression. STAT1 also participates in the differentiation of B cells [
4,
5] and T cells [
6,
7]. In addition to its function in immune cells, expression of
STAT1 in the tumor epithelium has been shown to exert an inhibitory effect on the development of the tumor [
8,
9]. This has been attributed to its cell-autonomous role in mediating apoptosis and proliferation arrest in response to cellular stress such as oncogenic transformation [
10], as well as to the transcriptional induction of chemokine and MHC class I genes, which promote recruitment of immune effector cells and recognition of tumor antigens [
2].
The proof of principle for the importance of STAT1 in impeding the development of tumors came from experiments with MMTV-neu tumor STAT1 null mice, which develop mammary tumors with shorter latency as compared to STAT1-proficient controls [
8,
9,
11]. Furthermore, STAT1 deficiency predisposed multiparous wild-type mice to intraepithelial neoplasias [
12]. Expression of
STAT1 in the tumor epithelium as well as in the stroma cells was shown to contribute to these anti-tumor effects of STAT1 [
8,
9,
12,
13]. It has been postulated from these observations that tumors may adapt to the anti-tumor action of STAT1 by down-regulating its expression and/or by impairing its activation [
1]. This notion is supported by immunohistochemical (IHC) analysis of
STAT1 expression in estrogen receptor (ER) - positive primary mammary carcinomas, which revealed lower
STAT1 expression levels in the tumor epithelium as compared to the adjacent normal epithelium in a considerable number of cases [
14].
In addition to these effects of STAT1 in preventing development and progression of early lesions an influence of STAT1 on the progression of established tumors and their response to therapy has been described. Forced over-expression of
STAT1 in tumor cells was found to confer resistance to radiotherapy [
15] and tumor cells with an increased propensity to metastasize to the lung after serial transplantations were shown to acquire a phenotype characterized by high expression levels of
STAT1 [
16]. Furthermore, increased expression of genes belonging to the so-called interferon-related gene signature including
STAT1 was shown to correlate with elevated frequency of relapse in human breast cancer [
17]. Moreover, the presence of tumor STAT1 activity correlated with disease progression from ductal carcinoma in situ to invasive carcinoma [
18]. It has been proposed that high expression of
STAT1 in established tumors could be the result of a selection process and promote the escape of tumor cells from IFN-γ-mediated tumor surveillance [
19]. On the other hand, activation of STAT1 in established mammary tumors as determined by specific DNA binding activity and tyrosine phosphorylation was linked to good prognosis and decreased frequency of disease recurrence [
20], indicating that high expression levels and activation of STAT1 might represent distinct prognostic and /or predictive parameters.
Alternatively to its potential direct impact on tumor progression, STAT1 expression and activation might serve as markers for chronic or acute inflammatory processes in the tumor, which are known to potentially influence the progress of disease depending on the type of infiltrating cells and tumor subtypes [
21,
22]: this is because IFNs, the major triggers of STAT1 expression and activation in the tumor epithelium and stroma, are secreted during acute as well as chronic inflammatory responses [
23]. In order to better understand the interrelationship between STAT1 expression and activation, progression of disease and immune infiltration, the expression of
STAT1 and STAT1 target genes as well as marker genes for infiltrating immune cells was analyzed in primary mammary carcinoma tissue derived from two independent patient cohorts. The data were evaluated by correlation analysis for a link between STAT1 and immune infiltrates as well as for their significance in predicting progression of disease and patient’s survival. The study revealed a link of potential mechanistic significance between elevated expression of
STAT1 and its target genes with markers of infiltrating immune cells, in particular with tumor-associated macrophages.
Discussion
The most striking result of our study is the opposing impact of STAT1 pY701 levels versus STAT1 expression and transcription of STAT1 target genes on prognosis in mammary cancer. A similar distinct association of STAT1 and pY-STAT1 levels with patient’s survival has been recently reported for soft tissue sarcomas [
38]. Mechanistically, our findings can be potentially attributed to different outcomes of short-term and prolonged STAT1 signaling in the tumor. STAT1 Y701 phosphorylation is typically maximal within the first hour of extracellular stimulation of JAK/STAT signaling, e.g. by IFN-γ, and then decreases as a result of the action of counter-regulatory phosphatases [
39,
40] and the negative feedback elicited by SOCS1 [
41,
42]. On the other hand, STAT1 transcriptional activity usually remains elevated even after cessation of triggering signals and leads to a sustained upregulation of STAT1 target genes. Long-term effects of STAT1 activation are further enhanced by upregulation of STAT1, which stays under its own transcriptional control, and is described to act as a transcription factor even in unphosphorylated form [
43,
44]. Thus, the increased levels of pY-STAT1 and its association with good prognosis in breast cancer tissue may reflect the short-term mode of STAT1 signaling (Figure
6A), whereas elevated STAT1 and STAT1 target gene mRNA and its link with bad prognosis may be indicative of persistent stimulation of Jak/STAT1 signaling in malignant cells (Figure
6A and B). Furthermore, the postulated disparity in kinetics of STAT1 signaling in breast cancer tumors may underlie the lack of linkage between pY-STAT1 levels and STAT1 mRNA expression (Figure
2).
Another important posttranslational modification of STAT1 is the phosphorylation at S727 [
45]. pS727-STAT1 was reported to stimulate or inhibit the IFN-γ transcriptional response depending on the target gene [
46]. S727 phosphorylation usually follows after Y701 phosphorylation [
45]. However, there are number of reports describing a separate regulation of these two sites. For example, in NK cells the phosphorylation of S727 can occur without concomitant tyrosine phosphorylation [
47] and in macrophages adenosine A(3) receptor signaling selectively modulates S727 phosphorylation [
48]. Our findings on the lack of correlation of pS727-STAT1 and pY701-STAT1 levels in breast cancer tissue samples (Figure
2) provide a further example for the non-coordinate regulation of these two sites. In our study we could observe an apparent discrepancy, whereby the total STAT1 protein levels as determined by IHC or the pS-727 STAT1 levels as quantified by immunoblotting were not significantly correlated with bad prognosis (Figure
6A), despite being linked to STAT1 mRNA expression levels (Figure
2). The STAT1 protein, however, could be regulated at the levels of transcription and translation. Since only the
STAT1 mRNA levels were found to be predictive of unfavorable outcome, we postulate that only the transcriptional but not the posttranscriptional regulation is relevant for the prognosis. This implicates that the protein may serve as a rather unspecific readout of STAT1 transcriptional activity.
How STAT1 is activated in mammary tumors remains unclear. Its activation might be promoted by tumor-intrinsic mechanisms mediated by receptor tyrosine kinases, such as HER2/erbB2 [
9] or induced by its principle activator IFN-γ produced by immune cells. The latter possibility is supported by the association of IFN-γ and marker genes for infiltrating immune cells in tumors with high STAT1 levels (Figure
4). Elevated levels of IFN-γ in ER-positive tumors were predictive of bad prognosis, as were high STAT1 levels (Figure
6A). However in the multivariate Cox regression, association of STAT1 with bad prognosis did not depend on IFN-γ (Figure
6C). Thus, despite the significant association between the expression of IFN-γ and STAT1 transcripts in mammary tumors, the impact on tumor prognosis of these two parameters appears to be non-redundant.
Another intriguing question is the mechanistic link between high
STAT1 and STAT1 target gene expression and bad prognosis. We consider two possibilities, which are not mutually exclusive: First, the transcriptional activation of
STAT1 could lead to expression of one or more critical STAT1 target genes that directly influence tumor progression and metastasis; Second, high STAT1 levels might simply serve as a marker for a chronic inflammatory process which was described to drive the progression and dissemination of the tumor [
21,
49]. Among the STAT1 targets investigated in our study,
MX1 and
CXCL10 can be considered as genes influencing tumor progression, since their expression was significantly associated with bad patient‘s prognosis (Figure
6). MX1 is described to exert an antiviral activity by binding to cellular RNA helicases required for viral replication but was not ascribed any obvious function in tumor biology [
50]. By contrast, two described properties of CXCL10 may underlie its potential tumor-promoting effects: One is its direct action on the proliferation of breast cancer cells [
51]. The other is its N-terminal processing by proteases under conditions of chronic infections to a truncated antagonistic CXCL10 form, which impedes chemoattraction of activated lymphocytes and by this means acts as an immunosuppressor [
37]. The later mechanism could also be exploited by cells of established, highly inflammatory neoplasms to avoid recognition and killing by tumor-specific T lymphocytes.
In our study, immunohistochemical analysis was able to distinguish the expression of STAT1 in the tumor epithelium and stroma, yet it was not possible to discern whether expression in one of these two compartments was prognostically more relevant (Figure
6A,B). We could identify a correlation of STAT1 mRNA amounts and epithelial STAT1 with markers of infiltrating leukocytes, in particular macrophages (Figure
4A,B and Figure
5). This indicates that either tumor cells with high STAT1 expression are more likely to provide a favorable environment for the recruitment/expansion of macrophages in the tumor or that macrophages promote an environment leading to high
STAT1 expression in the tumor. Whereas the association of STAT1 with infiltrating leukocytes and its impact on bad prognosis in breast cancer is a novel finding of this study, several reports have already documented the impact of infiltrating immune cells, in particular tumor-associated macrophages, on progression of disease and bad outcome [
52,
53]. By contrast, increased infiltration of the tumor with lymphocytes, in particular T cells, has been associated with better outcome in breast cancer patients subjected to neoadjuvant therapy [
54‐
56]. It remains to be investigated whether the lymphocyte infiltration in the tumor correlates with a better survival in our studied patient collective.
Conclusions
Our study reveals a complex association between STAT1 activation and progression of breast cancer. STAT1 tyrosine phosphorylation, typically increased after short-term activation of STAT1, is linked to good prognosis for the patient, whereas high levels of STAT1 mRNA, characteristic for sustained activation, predict bad outcome of disease. Furthermore, there was a positive correlation between mRNA levels of STAT1, STAT1 target genes, and marker genes indicative for infiltration with macrophages, pointing to an interrelationship between these parameters. The results of the Cox regression analysis further support a relevant link between STAT1 and macrophage infiltration for explaining bad prognosis.
Acknowledgements
This work was supported by grants from Integrated Center for Research and Therapy (IFTZ) of Innsbruck Medical University (W.D., E.M.-H.) as well as the doctorate program MCBO (P.T., Z.T.) and the project SFB21 Cell proliferation and cell death in tumors (Z.T.) funded by the Austrian Science Fund FWF. We thank Anto Nogalo, Sonja Phillip, Claudia Soratroi, Martina Chamson, Stefanie Faserl, Inge Gaugg, Martina Fleischer for excellent technical support, Chiara Berlato and Michael Haffner for helpful discussions, Jonathan Vosper for critical reading the manuscript, and Petra Massoner and Ernst Werner for kindly providing TaqMan primers.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
PT designed and performed experiments and contributed to data analysis, interpretation and writing of the manuscript. PC, HH performed statistical data analysis. RS, EM-H and PO carried out the immunohistochemical evaluation of data. FR, JPP and HF contributed patient’s samples, clinical data and participated in the design of the study. ZT participated in the study design. WD conceived and coordinated the study, participated in the immunohistochemical analysis and wrote the manuscript. All authors read and approved the final manuscript.