Background
The hedgehog (Hh) signalling pathway is known to be essential for multiple aspects of embryonic development. It is implicated in processes of organ patterning, cell differentiation and cell proliferation but it also plays a crucial role in the development of the limb, lung, brain and foregut [
1]. An active involvement of Hh signalling has been shown for a variety of human tumour entities including tumours of skin, cerebellum, muscle, lung, digestive tract, pancreas and prostate [
2‐
7]. Furthermore, the malignancy of tumours and their progression to metastatic stages have been associated with the activity of the Hh signalling pathway [
8,
9]. GLI1 is a member of this pathway and belongs to the family of GLI transcription factors including GLI1, GLI2 and GLI3. These transcription factors act as downstream mediators of Hh signalling and they share in common a DNA binding zinc finger domain [
10]. The human
GLI1 gene is located on chromosome 12q13.2-q13.3 [
11] and functions as an activator of transcription [
12]. Downstream targets of GLI1 signalling include molecules with regulatory effects on cell cycle and apoptosis such as Cyclin D2 or FOXM1 in basal cell carcinomas [
13,
14]. Several studies on GLI1 analysed potential involvement of this transcription factor in tumour developmental processes. Dahmane et al. (1997) [
15] showed that ectopic expression of GLI1 in the basal cells of the embryonic frog epidermis is able to induce basal cell carcinomas. Additionally in human oesophageal squamous cell carcinomas an association was found between the expression of GLI1 and depth of tumour invasion, status of lymph node metastasis, as well as unfavourable overall survival [
16].
SHH is another important member of the Hh signalling cascade which acts upstream of GLI1 in the activation process of the Hh pathway [
17]. It binds to the receptor Patched (PTCH) 1 or 2, which then relieves repression of the membrane protein Smoothened (SMO), a G protein-coupled receptor related protein [
9,
17]. The relief of SMO inhibition leads to an activation of GLI1 [
18]. As a result of this GLI1 accumulates in the nucleus [
19] where it controls the expression of typical Hh target genes [
20]. Thus SHH expression is assumed to have a direct impact on GLI1 activity. A significant correlation between increased expression of both SHH and GLI1 in human breast cancer would support the hypothesis that aberrant Hh signalling contributes to breast cancer development or progression. Therefore we performed a systematic expression analysis of GLI1 in human breast cancer on the mRNA and protein level and correlated GLI1 expression with clinicopathological patient characteristics. Subsequently we compared GLI1 expression data to previous data on the expression of SHH in the same collective of breast tumours.
Discussion
GLI1, a mediator of the Hh signalling pathway has previously been implicated in the development of different human tumour entities such as oesophageal squamous cell carcinomas [
16], basal cell carcinomas [
29] and brain tumours [
30]. Consistent with a major role in carcinogenesis highly increased levels of GLI1 expression were found in basal cell carcinomas but not in the surrounding normal breast tissues [
15,
29,
31].
To analyse the role of GLI1 in human breast cancer we carried out an initial realtime PCR analysis and found increased levels of GLI1 mRNA expression comparing the malignant mammary cell lines SKBR3, MDA-MB468, BT20, MDA-MB231 and MDA-MB435s to a set of benign mammary cell lines (HMEC and MCF12A).
GLI1 mRNA was especially abundant in the cell line MDA-MB435s, which is a common
in vitro and
in vivo model for metastatic breast cancer behaviour [
32] stating a potential involvement of GLI1 signalling in human breast tumour metastasis. Interestingly we also detected high levels of GLI1 mRNA expression in the benign breast cell line MCF10A.
It is known that GLI proteins themselves are mediators of more than just the Hh signalling pathway [
12] and overexpression of GLI1 is reported from several human tumour entities [
16,
29,
30] as well as benign tissues of organs like prostate or colon [
33,
34]. Therefore a functional involvement of GLI1 activity in human tumour development [
35] but also within cellular processes and maintenance of normal adult tissues might be concluded [
33]. mRNA expression analysis was further applied to a set of five matched normal and tumourous breast tissues. The results from these analyses underlined a potential involvement of GLI1 as a transcriptional regulator important for breast cancer development as increased levels of GLI1 expression were shown in four out of five analysed samples. However data derived from these studies were statistically not significant maybe because of the small sample numbers. Therefore accuracy of this first explorative approach should be tested in a larger validation set.
GLI1 expression was also studied on protein level using a larger cohort of human breast cancers (
n = 204) as well as normal breast tissues (
n = 46). We were able to show significantly increased levels of nuclear GLI1 protein expression (median IRS = 8) in human breast carcinomas in comparison to normal breast tissues (median IRS = 6). Furthermore intensities of GLI1 expression varied strongly between the different analysed tissues. These data are in accordance to earlier findings from Kubo et al. (2004) where nuclear GLI1 overexpression was found in a set of human breast cancers (
n = 52) and ratios of GLI1 expressing carcinoma cells varied strongly between the different breast tumour samples [
35]. Thus overexpression of GLI1 may not be a general characteristic of all human breast cancers although it could be a useful marker for a subset of human breast tumours, what we will analyse in more detail in our upcoming studies.
Analysing the data from our TMA a positive significant association was found between overexpression of GLI1 and unfavourable overall survival outcome. This association has not been reported anywhere else so far, but similar tendencies were recently shown in human oesophageal cancer [
36]. Patients with abundant GLI1 expression in the tumour (IRS > 6) had an estimated mean OS of 117 months (95% confidence interval [CI]: 105 to 125 months) compared to an estimated mean OS of 102 months (95% CI: 91 to 112 months) which was found in patients with lower levels of GLI1 expression (IRS = 6). Our data indicate that nuclear GLI1 overexpression being associated with unfavourable overall survival could be a phenomena found in a variety of human solid tumours that have been shown to express members of the Hh signalling pathway [
2,
5,
6,
8,
37].
To assess the role of GLI1 for the progression of human breast cancer in more detail we next correlated GLI1 expression data to clinicopathological characteristics of the tumours. Interestingly we found a positive significant association between increased nuclear GLI1 expression and tumour stage, lymph node status of the analysed breast tumours supporting a role of GLI1 as a new prognostic marker. We also carried out multivariate analysis and found a significant association between advanced tumour stage (pT2 vs pT1) and increased likelihood of GLI1 overexpression in the analysed tumours (4.9 times; 95% CI 2.3 – 10.5). However associations between GLI1 overexpression and lymph node status as well as histological grade of the tumours as stated by univariate analysis were multivariate insignificant.
Similar to our study Kubo et al. (2004) correlated expression of GLI1 to clinicopathological characteristics of human breast tumours and found a significant association of GLI1 overexpression to oestrogen receptor status and histological subtype of the analysed tumours (
P = 0.0216 and
P = 0.0036) [
35]. These correlations could not be stated through the results from our study and therefore we would not support the hypothesis of an involvement of the Hh pathway in the hormone-induced development of human breast carcinomas [
35,
38].
With lower abundance (median IRS = 6) GLI1 was also expressed in the nucleus of normal breast cells, and therefore we suggest a general involvement of GLI1 in transcriptional regulatory processes that are not only important within tumour development but also in normal breast cells. However Kubo et al. analysed normal breast epithelia (n = 52) and did not detect GLI1 expression in the nucleus of normal breast cells adjoining the analysed tumourous tissues.
So far the discrepancies between these two studies remain unsolved and should therefore be topic of further analyses characterizing the potential of GLI1 as a new potential prognostic marker of human breast cancer.
To find out more about the complexity of GLI1 signalling and its functional implications on human breast cancer a systematic characterisation of potential upstream and downstream targets of GLI1 in tumourous breast samples that are characterised by GLI1 overexpression would be an important approach. Interestingly the results from our statistical analyses showed a significant association between GLI1 overexpression and the presence of Hh member SHH in human breast tumours. Our observations indicate that not only GLI1 expression but more important its implication in the Hh signalling pathway might be a more precise characteristic of human breast cancers.
SHH and GLI are both members of the Hh signalling cascade which was first described in the fruit fly Drosophila [
39] and it might be reasonable to assume that this signalling pathway is conserved and active in a proportion of human breast cancers. In mammals, a direct influence of SHH expression on increased GLI1 activity has been described in a variety of biological processes, e.g. during mouse limb bud development [
40]. Furthermore increased expression of
SHH mRNA in human colonic adenocarcinomas is known to correlate with downstream increased expression of GLI1 leading to promotion of cell proliferation [
41]. In human cancer, sustained activity of Hh-GLI signalling is indicated to be essential in growth and survival of human prostate cancer cells [
42]. The results from our correlation analyses between SHH and GLI1 are in accordance to these findings and may lead to the suggestion that GLI1 activity is mediated through the Hh initiating molecule SHH and that abundant levels of nuclear GLI1 expression observed in breast cancer could be a result of increased SHH activity. More interestingly Hh signalling is also active in human breast cancer stem cells and studies on
in vitro culture systems showed that an activation of this pathway with Hh ligands promotes the self-renewal of mammary stem cells, and also increases proliferation of mammary progenitor cells as reflected by increased numbers of mammosphere-initiating cells and increased mammosphere size [
43]. On regard of this an early oncogenic activation of Hh signalling as an initial step in breast cancer development starting at the level of tumour progenitor stem cells could be proposed.
To achieve therapeutic benefits of new oncogenic markers that are involved into tumour development, strategies have to be developed to inhibit their cancer initiating potential. Interestingly a number of molecules inhibiting the Hh signalling pathway have already been described. Cyclopamine for example is an inhibitory molecule for Hh signalling and has been shown to inhibit the proliferation of brain tumour cells which contain an active Hh signalling pathway [
44]. GANT61 a hexahydropyrimidine derivative and GANT58 possessing a thiophene core with four pyridine rings, are two other small-molecule antagonists, which act in the nucleus to block GLI1 function and inhibit GLI1 mediated transcription [
45]. Assuming that GLI1 might be involved into human breast cancer development and further proposing that GLI1 is activated through upstream members of the Hh signalling cascade, the mentioned Hh inhibitors could offer new drugs in the treatment of breast cancer as well.
In conclusion, we showed nuclear GLI1 overexpression in human breast carcinomas relative to normal breast tissue on the RNA and protein level. Interestingly, we also found a significant positive correlation between increased nuclear GLI1 expression and aggressive tumour characteristics (i.e. tumour stage and lymph node status). The aggressive behaviour of GLI1 positive breast cancers is supported by the finding that GLI1 overexpression is associated to unfavourable overall survival.
Our data raise the hypothesis that GLI1 represents a candidate oncogene in human breast cancer. Additional studies validating these data and further analysing possible interactions of GLI1 with other Hh signalling members are underway to clarify the role of this important developmental pathway for the pathogenesis of human breast cancer.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AtH: designed the study, carried out the experiments, interpreted the data and wrote the manuscript; NB: analysed the immunohistochemical data and critically revised the manuscript; SvS and IL: supported with expertise in molecular biology techniques and in data interpretation; ECA: supported in the statistical data analysis; RK: participated in design and coordination of the study; ED conceived the study, participated in study design and coordination, molecular and data analysis, data interpretation and drafting of the manuscript.