Background
SOCS proteins comprise a family of eight members (SOCS1-7 and CIS), which have initially been described as STAT induced STAT inhibitors or more generally as negative regulators of cytokine signaling via the Jak/Stat pathway. Their ability to modulate signal transduction is based on two functional domains, an SH2 domain, which binds to phosphorylated tyrosine residues and a SOCS box, which serves as a recruiting site for ubiquitin ligases [
1]. Recent reports, showing the interaction of SOCS proteins with various other signaling molecules like p65, FAK, c-kit, IRS1/2 and vav, indicate a more general role of SOCSs proteins in the maintenance of cellular homeostasis [
2‐
6]. Tumors often exhibit alterations in SOCS protein expression: CpG island methylation of SOCS gene loci, and consequently, inhibition of SOCS expression was observed in a variety of solid tumors and hematological malignancies [
7‐
13]. On the other hand, forced overexpression of SOCS1 and SOCS2 results in diminished transformation and decreased metastatic potential [
8,
14]. Taken together, these in vivo and in vitro data support the role of SOCS proteins as tumor suppressors. To evaluate the potential significance of SOCS expression in mammary carcinoma for prognosis and its association with clinicopathological characteristics we have investigated the mRNA levels of SOCS1, SOCS2, SOCS3 and CIS in a representative collection of primary breast cancers specimens. Since our study revealed SOCS2 expression as a predictor for prognosis, and growth hormone is a major inducer of SOCS2, we became interested in the expression of the growth hormone regulated gene IGF-I. IGF-I is a growth and differentiation factor, which acts in an endocrine – via the classical hypothalamus/pituitary/liver axis – and in a paracrine/autocrine manner [
15]. In the normal mammary gland, IGF-I is required for structural development, promoting ductal growth and differentiation [
16]. Whereas epidemiological data indicate that high levels of circulating serum IGF-I are associated with an increased risk for the development of solid tumors including breast cancer, little is known on the role of autocrine produced IGF-I [
15,
17]. We therefore evaluated the expression levels of locally produced IGF-I in the tumor. Interestingly, high IGF-I mRNA levels were associated with good prognosis, suggesting that secretion of IGF-I by the tumor is a property of a less malignant and more differentiated tumor.
Discussion
Several recent reports have investigated the role of SOCS proteins in the oncogenesis of various solid tumors and hematological malignancies [
7‐
13] indicating a central role of SOCS proteins in the regulation of cellular growth and differentiation. SOCS promoter methylation and consequent gene silencing has been observed in hepatocellular carcinoma, lung cancer, head and neck cancer, multiple myeloma and AML [
9‐
13]. Conversely, restoration of SOCS protein expression resulted in induction of apoptosis and suppression of growth. These findings suggest an antioncogenic role of SOCS proteins.
However, little is known on the prognostic significance and the association of SOCS expression with clinicopathological features. In this study, we therefore sought to elucidate the relationship between pathological and clinical parameters and the expression of the SOCS family members SOCS1, SOCS2, SOCS3, and CIS in patients with breast cancer by RT-PCR.
Our results showed that high SOCS2 expression is associated with increased overall survival and after adjusting for age, lymph-node status, estrogen receptor status and tumor grade, high SOCS2 expression remained an independent predictor for good prognosis (Table
3). Farabegoli et al. recently provided first evidence that high SOCS2 protein expression is associated with lower pathological grade and lower cell proliferation indices [
23]. Together with our findings of highly significant negative correlation between SOCS2 expression and histopathological grade, positive correlation with steroid receptors, which are markers of differentiation, and the higher expression in normal tumor adjacent tissue as compared to carcinoma tissue, these results indicate that SOCS2 is a marker for differentiated tumors. Similar to our observations in breast cancer low SOCS2 expression in prostate cancer is associated with an increased incidence of metastasis and SOCS2 mRNA levels decrease during prostate cancer progression [
27]. Furthermore, SOCS2 expression was found to be down-regulated in pulmonary adenocarcinoma [
28]. The important role of SOCS2 as a regulator of growth hormone signaling raises the intriguing question, whether in mammary carcinoma, SOCS2 serves as an essential regulator of cellular growth and tissue homeostasis and its expression is required for the maintenance of a more differentiated and less malignant phenotype. However, Raccurt et al. described an increased expression in cancerous ducts and reactive stroma as compared to normal breast tissues by using in situ hybridization [
29]. This result is not in agreement with our findings obtained by RT-PCR, immuno-blotting and immunohistochemistry and might be due to the differences in the detection methods.
The fact that SOCS2 has been described to be induced by and to modulate the GH receptor/Jak/STAT pathway prompted us to further investigate the expression of another downstream target of the GH receptor, IGF-I [
30,
31]. Similarly to our observations with SOCS2, high IGF-I mRNA expression proved to be a predictor for good prognosis (Figure
1 &
2), and normal adjacent tissue was characterized by increased IGF-I mRNA levels (Figure
3). Our study confirms the results from a previous report from a large sample collection in which patients with high IGF-I expression tended to have a favorable prognosis [
25].
Mammary epithelial cell specific overexpression of IGF-I which mimics autocrine IGF-I, increased the incidence of mammary carcinomas in mice [
32], whereas liver-specific deletion of IGF-I, eliminating serum IGF-I, results in reduced tumor development [
33]. Several large epidemiological studies highlighted this association also in humans, revealing an increased risk for breast cancer especially in women with elevated serum levels of IGF-I before the age of 50 [
34]. Compiled evidence suggests that high IGF-I renders cells susceptible to transformation by secondary events and thereby contributes to tumor progression [
15]. We observed an increased expression of IGF-I in normal tumor adjacent tissue as compared to cancerous tissue. This is in line with a previous reports comparing IGF-I expression in large collections of normal mammary glands and breast cancer specimens [
25,
35]. Differences in IGF-I expression levels between normal and cancerous tissue could partly be explained by differences in the prevailing cell types. Several reports showed that IGF-I expression is restricted to stroma cells in the mammary gland [
36]. High stroma content, frequently found in highly differentiated tumors would therefore result in higher IGF-I levels, which could explain the positive prognostic value of high IGF-I expression. However, MCF7 and T47D mammary carcinoma cells showed expression of IGF-I (Haffner MC, Doppler W, unpublished observation) indicating that also epithelia cells are capable of producing IGF-I.
Interestingly, we observed that high IGF-I expression is a very potent predictor for good prognosis in patients with lymph-node negative disease (Figure
2). Histological status of the axillary lymph-nodes represents an important prognostic factor. Patients with even a single lymph-node metastasis have a poorer outcome than those with negative nodes and decision on systemic chemotherapy is largely based on lymph-node status [
26]. Risk stratification in patients without lymph-node involvement is difficult. Hence markers predicting the outcome of patients in this subgroup are needed. Gene-expression profiling based on microarray analysis of 70 genes is a powerful predictor for disease outcomes in patients with lymph-node negative disease. [
37] We demonstrate here in a limited number of patients that expression levels of a single gene (IGF-I) can be used to independently predict the risk of disease recurrence and death in lymph-negative patients. Further studies evaluating this marker in a larger cohort of patients are needed to determine its prognostic and predictive potential.
Our results point to the hypothesis that both IGF-I and SOCS2 expression in breast cancer is controlled by similar mechanisms. A possible common signaling intermediate could be STAT5, since both IGF-I and SOCS2 have been described as STAT5 target genes [
38,
39]. STAT5 is essential for mammary gland development and lactogenesis [
40,
41], and alterations and repression in the signaling cascade could result in less differentiated and more malignant tumors. This is in line with recent findings showing that breast lesions with aberrantly proliferating cells like atypical ductal hyperplasia, ductal carcinoma in situ and invasive carcinomas are characterized by a reduction or absence of STAT5a expression [
42]. Activation of STAT5, as determined by tyrosine phosphorylation, decreases during metastatic progression of breast cancer, and in a large patient cohort, activated STAT5 emerged as an independent prognostic marker for good prognosis in women with breast cancer [
43]. Furthermore, STAT5 expression levels in the primary tumor as determined by immunohistochemisty was associated with better response to endocrine therapy suggesting a possible crosstalk between ER and STAT5, which could be mediated by SOCS2 [
44]. Interestingly, CIS, another known STAT5 target gene, was strongly correlated with SOCS2 and IGF-I, but did not prove to be a prognostic marker.
Whereas SOCS2 was initially described as a negative regulator of GH signaling [
30], recent reports suggest that depending on the expression level, SOCS2 can act as either an enhancer or suppressor of this signaling pathway [
45,
46]. SOCS2 knock out mice show only modest increase in STAT5 phosphorylation [
39] and overexpression of SOCS2 even results in an enhancement of GH signaling in vivo and in vitro [
46]. Therefore, high STAT5 activation in the tumor is compatible with higher SOCS2 levels. Interestingly, GH-transgenic mice lacking one copy of SOCS2 show a high incidence in aberrant lesions in the colon indicating that small variations of SOCS2 expression levels can have profound implications on cell proliferation and eventually tumor growth [
47].
In a comprehensive study Kate Sutherland et al. evaluated promoter CpG island methylation and loss of heterozygosity (LOH) of the SOCS1, SOCS2 and SOCS3 genes in ovarian and breast cancer [
8]. In 48 primary breast cancer samples no methylation was observed and only 13% of tumors showed LOH in the SOCS2 locus. Similar, unpublished results from our group show no significant CpG island methylation in SOCS2 locus in a limited number of breast cancer samples (Haffner MC and Auer D, unpublished data). In conclusion, the current data do not support methylation as a major factor contributing to differences in SOCS2 expression levels in primary breast cancers.
Our observation of higher SOCS2 expression levels in ER positive tumors, raises the hypothesis that functional ER signaling contributes to enhanced SOCS2 transcription. This is in accordance with a recent report showing selective upregulation of SOCS2 expression by estrogen in vitro [
48]. In addition, mice carrying a targeted deletion of the estrogen receptor failed to show an increased SOCS2 expression upon estrogen stimulation [
49]. It still remains to be shown whether high SOCS2 expression per se is causative for the differences in tumor differentiation and prolonged overall survival. Preliminary results from MCF7 mammary carcinoma cells show that overexpression of SOCS2 results in a significant reduction of anchorage independent growth, which supports the hypothesis of SOCS2 as an antioncogene [
8]. Whether the favorable prognostic characteristics of ER positive tumors can be attributed to some extent to higher SOCS2 expression needs to be further clarified.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MCH conceived the study, performed experiments and wrote the paper. BP participated in RT-PCR analysis and study design. JPP and FR contributed patients' samples and clinical data. EM-H carried out the pathological studies. GD and CM contributed to study design and provided samples. WD conceived and coordinated the study and wrote the manuscript. All authors read and approved the final manuscript.