Introduction
The Signal transducers and activators of transcription (Stat) family of proteins are transcription factors known for their role as integrators of cytokine and growth factor receptor signaling and are required for cell growth, survival, differentiation, and motility [
1,
2]. Stat activation is dependent upon tyrosine phosphorylation, which induces dimerization via reciprocal phosphotyrosine-src homology domain 2 (phosphotyrosine-SH2) interaction between two Stat molecules. Activated Stat's translocate to the nucleus where they bind to consensus promoter sequences of target genes and activate their transcription [
3]. In normal cells, Stat tyrosine phosphorylation is transient. However, in numerous cancer-derived cell lines and in an ever growing number of primary tumors, Stat proteins (in particular Stat3) are persistently tyrosine phosphorylated [
4]. Stat3 is found to be constitutively phosphorylated to high levels in >50% of breast cancer derived cell lines and in >30% of breast adenocarcinomas and may be a poor prognostic indicator [
5,
6]. Constitutive activation of Stat3 in epithelial cancers and cancer derived cell lines is frequently due to aberrant autocrine or paracrine IL-6 signaling [
7]. Inhibition of Stat3 activity in tumor-derived cell lines both
in vitro and
in vivo, by the introduction of antisense, small interfering RNA, decoy molecules, dominant-negative Stat3 constructs, and/or blockade of tyrosine kinases has been associated with growth arrest, apoptosis, decreased angiogenesis and invasion [
2,
4,
8,
9]. More recently, non-canonical functions for Stat3 have been identified including non-tyrosine phosphorylated Stat3 mediating transcriptional activation, non-tyrosine phosphorylated Stat3 binding to stathmin a microtubule associated protein and regulating migration, non-tyrosine phosphorylated Stat3 regulating metabolic functions in the mitochondria leading to Ras-dependent transformation [
10‐
12].
The ras proto-oncogene encodes a guanine nucleotide binding protein that plays an essential role in diverse cellular responses, including cell proliferation and differentiation [
13]. Although ras mutations are infrequent in human breast cancers, elevated amounts of the ras protein have been found in 60 to 70% of human primary breast carcinomas [
14]. Ras expression has been suggested to be a marker of tumor aggressiveness in breast cancer, including the degree of invasion into fat tissue, infiltration into lymphatic vessels and tumor recurrence [
14‐
16]. Rodent fibroblasts and human mammary epithelial cell lines transformed by the H-Ras oncogene do not express tyrosine phosphorylated Stat3 [
17‐
19]. Moreover, non-tyrosine phosphorylated Stat3 was demonstrated to regulate metabolic functions in the mitochondria leading to Ras-dependent transformation [
20].
Here we further investigated the role of non-tyrosine phosphorylated Stat3 in Ras-mediated mammary tumorigenesis. Specifically, we examined the consequences of reducing Stat3 levels in Ras transformed mammary epithelial cells. We determined that Stat3 deficient Ras transformed MCF10A cells were less capable of mediating migration, invasion and tumorigenesis than the control MCF10A-Ras cells. Surprisingly, tumors derived from MCF10A-Ras cells expressed high levels of tyrosine phosphorylated Stat 3 (pStat3) as did mammary tumors from MMTV-expressing K-Ras mice. Furthermore, the interleukin-6 ligand (IL-6) which was recently shown to be a principal regulator of Stat3 activation in breast cancer [
6], was found to be elevated in both MCF10A-Ras and MMTV-K-Ras tumors. In addition, growth of MCF10A-Ras cells in the presence of basement membrane proteins (Matrigel) resulted in high levels of pStat3. Reduction of Stat3 levels or inhibition of its activity led to the up-regulation of E-cadherin in MCF10A-Ras cells. We demonstrated that culturing and passaging primary Ras-expressing tumors from 3-D to 2-D resulted in a diminution of pStat3 and IL-6 levels suggesting that depending on the context in which MCF10A-Ras expressing cells are grown can significantly alter the levels of pStat3 and the subsequent behavior of the cells.
Discussion
We sought to determine the role of non-tyrosine phosphorylated Stat3 in tumorigenesis by examining the breast epithelial cell line MCF10A cells transformed with the H-RasV12 oncogene. Non-tyrosine phosphorylated Stat3 can function as a transcription factor in association with NF-kB driving expression of a number of genes involved in tumorigenesis including BCL2A1, Rho GAP6, MRAS, MET, RANTES and Cyclin B1 [
11]. We examined and compared levels of these transcripts in MCF10A-Ras cells either expressing or lacking Stat3 and found no significant differences (data not shown). Furthermore, we performed gene expression profiling on these RNA populations and found only 10 transcripts that were potentially differentially regulated as a function of non-tyrosine phosphorylated Stat3 (data not shown). Thus in this cell line, it does not appear that non-tyrosine phosphorylated Stat3 plays a significant role in regulating transcription.
We examined cell proliferation and observed no differences as a function of Stat3 (Figure
1). Furthermore, stimulation of cells with exogenous IL-6 led to robust Stat3 phosphorylation but did not affect cell proliferation (Figures
3 and
5c). Thus, Stat3 (either phosphorylated or non-phosphorylated) has no significant impact on 2-D growth. These observations have previously been made demonstrating a marginal role for gp130, Stat3 or constitutively activated Stat3 (Stat3-C) in 2-D cell proliferation but a dominant one for
in vivo growth [
40,
41]. In contrast to cell proliferation, we determined that Stat3 was required for migration and invasion (Figure
1). It was recently shown that Rac1 activation leads to enhanced IL-6 expression and gp130/Jak/Stat3 activation leading to gp130 dependent cell migration [
42]. Activated Stat3 has been shown to mediate migration of cancer cells by regulating genes such as
integrin β6,
tenascinC, twist and
liv1 [
39,
43‐
46]. In addition to its transcriptional activating function, phosphorylated Stat3 was shown to interact with focal adhesion kinase (FAK) and was shown to play a role in cell migration [
10,
47]. We hypothesize that migrating or invading Ras transformed MCF10A cells activate Rac1 which leads to increased IL-6 expression, Stat3 tyrosine phosphorylation and enhanced cell migration and invasion. This process can be enhanced by paracrine IL-6 and partially inhibited by reducing IL-6 levels (Figure
5).
IL-6 was shown to be expressed to high levels in numerous Ras-expressing cell lines including kidney, fibroblasts, human mammary epithelial cells and pancreatic cancer derived cell lines when grown in 2-D [
23]. In contrast, we do not see any appreciable IL-6 mRNA or protein expression in Ras transformed MCF10A cells grown in 2-D (Figures
3b and
5a). Perhaps, expression levels of Ras influence IL-6 production which may have been lower in our cells than in those described by the Counter laboratory. In contrast to cells grown on plastic, we observed that MCF10A-Ras cells grown in 3-D either in basement membrane (Matrigel) cultures or as xenografts expressed high levels of IL-6 and pStat3 (Figures
2,
4 and Supplemental figure S1 in Additional file
1). In addition, MMTV-Ras transgenic mice also developed tumors expressing IL-6 and pStat3 (Figure
2). Thus, our data suggest that the environment in which Ras transformed cells are grown can regulate the expression levels of IL-6.
MCF10A cells are immortalized human mammary epithelial cells that undergo a program of apical-basolateral polarization, proliferation, growth arrest and apoptosis leading to acinar formation when grown in matrigel [
48‐
50]. These 3-D cultures are felt to be a more relevant system to examine cell growth, cell adhesion and cell-ECM interactions recapitulating some of architectural changes observed
in vivo [
49]. Expression of activated H-Ras in MCF10A cells led to changes in the morphology of these structures from organized hollow acini to solid irregularly shaped structures lacking E-Cadherin expression (Figure
4) [
51]. We previously demonstrated that prostate epithelial cells expressing a constitutively activated Stat3 (Stat3-C) had decreased E-Cadherin levels [
39]. Furthermore, IL-6 stimulated mammary epithelial cells downregulate E-Cadherin expression [
38]. Here we provide further evidence that pStat3 negatively regulates E-Cadherin expression in Ras transformed MCF-10A cells as inhibiting its activity led to an increase in E-Cadherin expression (Figures
4 and
6). Although Jak inhibition restored E-Cadherin expression in MCF10A-Ras cells we did not observe any hollowing out of the acini suggesting that E-Cadherin expression is insufficient to induce a reorganization of the acini or induce apoptosis of the centrally located cells (Figure
4 and Supplemental figure S3 in Additional file
3). Although Jak inhibition had no affect on the 2-D growth of MCF10A-Ras cells we did see a loss of viability in the acinar structures grown in 3-D over a seven-day period (data not shown). These observations further support the hypothesis that inhibition of IL-6/Jak/Stat3 signaling inhibits 3-D growth and tumorigenesis but not 2-D growth.
The mechanisms of IL-6 transcriptional regulation in transformed cells involves the activation and recruitment to the IL-6 promoter of a number of transcription factors including AP-1, NF-kB, (NF-IL6) or C/EBPβ and CREB [
52,
53]. Furthermore, a myriad of other transcription factors in association with the above mentioned proteins can modulate expression of the IL-6 gene for example nuclear hormone receptors (GR and ER), PPARγ and Stat3 [
12,
54‐
60]. IL-6 mRNA stability is also tightly regulated through the association of RNA-binding proteins (AUF1) with the 3'UTR and activation of p38 [
61,
62]. We examined levels of activated NF-kB, AP-1, CREB and C/EBPβ in MCF10A-Ras cells (grown in 2-D culture) by EMSA and did not observe any binding of these factors suggesting that Ras expression in MCF10A cells is insufficient to mediate activation of the IL-6 gene (data not shown). Furthermore, our data suggest that in order for Ras transformed cells to produce high levels of IL-6, cells need to be exposed to extracellular matrix proteins such as those found in matrigel (principally laminin) or to an
in vivo environment which exposes epithelial cells to extracellular matrix proteins but also fibroblasts, endothelial cells and macrophages which produce growth factors capable of mediating IL-6 expression. When MCF10A-Ras cells were cultured on plates coated with matrigel, collagen I, collagen IV, fibronectin and laminin we found that matrigel and laminin could induce modest expression of pStat3 (Supplemental figure S1 in Additional file
1). We suggest that integrin engagement of ECM proteins can enhance pStat3 through upregulation of IL-6. Indeed, there are examples whereby integrin engagement with extracellular matrix proteins such as collagen and laminin leads to increased IL-6 production [
63‐
65]. The work by the Bissell laboratory has demonstrated that the nature of the ECM matrix (solid or gel-like) can profoundly influence cell morphology and gene expression [
66]. Furthermore, stromal cells surrounding epithelial cells secrete IL-6 and in a paracrine manner can induce epithelial cells to produce IL-6 in an autocrine manner [
37,
67]. Since murine IL-6 does not engage the human IL-6R, other factors are therefore implicated in promoting human IL-6 expression and Stat3 phosphorylation in these tumors [
68]. For example, aberrant EGFR signaling in glioblastoma, lung cancer and MCF10A cells led to enhanced IL-6 production and signaling [
32,
69].
In order to determine whether the 3-D environment is required for sustained IL-6 expression by epithelial cells we cultured tumor cells (either from MCF10A-Ras xenografts or MMTV-Ras tumors) and passaged them on plastic dishes (2-D environment). After a few passages in culture, we have an enriched epithelial cell population (devoid of fibroblasts, endothelial cells and immune cells) which no longer expresses IL-6 nor pStat3 (Figure
6). These data suggest that the micro-environment (including stromal cells, endothelial cells and immune cells) is critical for the expression of the IL-6 ligand which results in activation of Stat3. A requirement for IL-6 signaling in tumor formation has been demonstrated by using IL-6 knock-down approaches as well as blocking antibodies to IL-6 in a variety of cell types [
23,
32,
70,
71]. Here we also demonstrate a requirement for IL-6 in MCF10A-Ras mediated tumor formation with no apparent effect on 2-D growth.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KL performed and designed most of the experiments and helped write the manuscript. SG and MB performed several experiments. HH worked in the laboratory of LI generating the lentiviral Stat3shRNA. KP provided the MMTV-K-Ras mice and tumors and technical assistance. JB conceived and designed the experiments and critically revised the manuscript. All authors read and approved the final manuscript.