Background
Heat shock protein 90 (Hsp90) targeting has emerged as a valuable strategy for cancer therapy [
1,
2], because these proteins are being up-regulated in malignant and non-malignant cells types upon exposure to a variety of stressors [
3]. At constitutive levels, heat-shock proteins regulate proper folding and stabilization of abundant intracellular proteins, and their stress-associated induction improves cell survival. Hsp90, one of the most studied molecular chaperons, is overexpressed in tumor cells and is essential for the stability and function of a wide range of oncogenic client proteins [
4]. These Hsp90 clients comprise kinases such as ERBB2, EGFR, CDK4, RAF, AKT, cMET and BCR-ABL, and transcription factors such as HIF-1α, STAT3, and STAT5 [
2,
5,
6]. Thus, Hsp90 is a promising target for cancer therapy, as demonstrated by the expanding armamentarium of Hsp90 inhibitors and by new clinical studies incorporating the use of these inhibitors [
7]. Nevertheless, due to the broad and complex inhibition of multiple signaling pathways affected by Hsp90, the biological effects remain poorly defined and incompletely understood.
We recently demonstrated that therapeutic inhibition of Hsp90 not only elicits antineoplastic efficacy through blocking oncogenic signaling, but also up-regulates certain signaling molecules in human colon carcinoma cell lines. One of these molecules is activating transcription factor-3 (ATF3), which is Hsp90-inhibitor inducible in HCT116, SW620 and HT29 colon cancer cells [
8]. Importantly, such protein up-regulation in response to Hsp90 inhibition has thus far only been reported for certain other heat-shock proteins such as HSF1 and Hsp70. This response may counteract the anti-neoplastic potential of Hsp90 inhibitors for the following reasons [
9,
10]. ATF3 belongs to the ATF/cyclic AMP response element binding (CREB) family of transcription factors and most cells have very weak or absent ATF3 expression under steady-state conditions. A significant increase in ATF3 can be observed when cell-stress is induced [
11], making ATF3 an universal „adaptive response gene" [
12,
13].
Importantly, different roles for ATF3 have been proposed. In normal tissues, ATF3 may promote both apoptosis and cell proliferation [
13], while in neoplasms it has been identified as either an oncogene or as tumor suppressor, depending on tumor entity and grade [
13‐
15]. For instance, ATF3 can mediate pro-apoptotic effects in human mammary epithelial cells, whereas in breast cancer cells (MCF10A) it may promote cell survival, motility and invasiveness [
15]. Transgenic mice that overexpress ATF3 in basal epithelial cells develop epidermal hyperplasia, dysplastic lesions and oral squamous cell carcinoma [
16]. Also in favor of oncogenicity, the tumor suppressor gene Drg-1 mediates its anti-metastatic properties through ATF3 down-regulation in prostate cancer [
17].
In colon cancer, the effects of ATF3 expression are particularly perplexing. In one respect, ATF3 was shown to be overexpressed in human colon cancer specimens and appears to promote tumor growth and migration in an experimental HT29 colon cancer model [
18,
19]. In another respect, ATF3 has been described to mediate anti-neoplastic and anti-invasive effects of non-steroidal anti-inflammatory drugs (i.e. COX-2 inhibitors) in colorectal cancer [
14]. In the present study, we sought to clarify ATF3 regulation and its role in human colon cancer using xenogenic mouse models. We hypothesized that Hsp90 inhibitor-mediated induction of ATF3 expression does not counteract the anti-neoplastic and anti-metastatic potential of Hsp90 targeting agents.
Discussion
Our recent observation that Hsp90 inhibition induces ATF3 in cancer cells and the lack of clarity regarding the biological effect of this transcription factor in oncology pressed our aim to define the role of ATF3 in colon cancer. We now have confirmed that blocking Hsp90 does indeed induce ATF3 in various cancer derived cell lines, including colon (HT29, HCT116), gastric (TMK1), and pancreatic (L3.6pl) cancer derived cells. Furthermore, this study is the first to demonstrate that loss of ATF3 via shRNA-mediated down-regulation increases the migration properties of HCT116 colon cancer cells in vitro and promotes tumor growth and metastasis in vivo. Hence, results from this study suggest that ATF3 functions as a tumor suppressor and anti-metastatic factor in HCT116 colon cancer, which is therapeutically inducible by blocking Hsp90.
Recent publications have demonstrated a dichotomous role of ATF3. Depending on the cell type and malignancy, ATF3 can mediate either proliferative and pro-migration properties, or anti-proliferative and pro-apoptotic effects [
26‐
29]. For instance, Yin and co-workers have demonstrated in
in vitro experiments that ATF3 induces apoptosis in non-malignant mammary epithelial cells, but reduces apoptosis and enhances motility in breast cancer cells, suggesting an oncogenic role of ATF3 in breast cancer [
15]. In colon cancer, down-regulating ATF3 in HT29 colon cancer cells with antisense oligonucleotides apparently diminished entopic tumor growth and metastasis in mice [
19]. In contrast, we could show that in HCT116 colon cancer, loss of ATF3 function does result in a higher pro-migration capacity
in vitro and an accelerated tumor growth with increased metastasis
in vivo. One explanation of this discrepancy might be the different genetic background of HT29 and HCT116 colon cancer cells. While HCT116 harbors mutant KRAS, HT29 colon cancer cells are wildtype for KRAS but harbor mutant BRAF. Recent publications have shown that the KRAS and BRAF mutation status of colon cancer cells influence the expression rates of multiple proliferative as well as apoptotic signaling intermediates (Kikuchi et al, Cancer Res 2009, Oliveira et al, Oncogene 2003, Seruca et al, 2009), including HIF1α signaling and the MAPK/Erk and PI3K/Akt pathways which we identified as interacting with ATF3 (Figure
2). Furthermore, EGFR-targeting agents are clinically effective in the treatment of KRAS and BRAF wildtype tumors, whereas no clinical benefit could be proven for KRAS or BRAF mutant tumors (Lievre et al, Oncogene 2010). Thus, drug-induced overexpression of ATF3 may have beneficial effects in only a subset of colon cancer cells. This important result will be further addressed in future experiments, where loss of ATF3 function as well as ATF3-overexpression will be investigated in colon cancer cells with different genetic background.
In line with our findings in HCT116 colon cancer, tumor suppressive properties of ATF3 were suggested in a study by Oh
et al., describing that ATF3 acts as tumor-inhibiting factor in HeLa cervical cancer cells
in vitro [
30]. Moreover, Lu and co-workers elegantly demonstrated that ATF3 is capable of suppressing a Ras-mediated tumorigenicity of murine fibroblasts (ATF3
-/- versus ATF3
+/+ fibroblasts) in an
in vitro, as well as in an
in vivo model, hence supporting our hypothesis of a tumor suppressive role. In conclusion, these discrepancies mirror the complex role of ATF3 which may not solely depend on the investigated cell line. The biological function of ATF3
in vivo may rather highly rely on the microenvironment of a defined tumor entity.
One clinical significance of our findings is that treatment-induced up-regulation of ATF3, as for example via Hsp90-inhibition or COX-2 inhibition, may be beneficial in some tumors for reducing growth and metastasis [
8,
14]. With respect to COX-2 inhibitors, experimental studies have nicely demonstrated that ATF3 may mediate anti-neoplastic and anti-invasive effects of such non-steroidal anti-inflammatory drugs [
14]. In this study, overexpression of ATF3 inhibited invasion to a similar degree as sulindac sulfide treatment and antisense ATF3 increased invasion
in vitro. This tumor suppressive effect of ATF3 is also supported by their findings, where transfection of cancer cells with a full-length ATF3 vector suppressed tumorigenicity and invasiveness
in vitro and tumor growth
in vivo [
14]. However, this group was not able to validate in an
in vivo setting that loss of ATF3 function is conversely associated with increased growth rates and metastasis, hence our study further expands the knowledge on ATF3 function beyond these aspects. We observed an enhanced migration behavior after ATF3 inhibition
in vitro and hypothesized that loss of ATF3 function may also lead to an increased tumor metastasis
in vivo, an aspect that has not been comprehensively investigated to date. In subsequent hepatic and peritoneal tumor models, we were able to demonstrate a significant increase in tumor burden, cancer dissemination, and tumorigenicity upon further down-regulating ATF3. Thus, we propose that ATF3 functions as a tumor suppressor and anti-metastatic factor in HCT116 colon cancer. Moreover, in a recent publication, Ameri and colleagues could show that induction of ATF3 in hypoxic conditions, a common feature detectable in solid tumors, is independent of the transcription factor HIF-1α [
11]. The factors HIF-1α and ATF3 are both induced by hypoxia and other cellular stressors, and both transcription factors regulate the expression of multiple genes during tumor progression and metastasis [
11].
Importantly, and of high clinical relevance, we could show in the current and in one preliminary previous study that ATF3 expression can be induced in cancer cells by Hsp90 inhibition
in vitro and
in vivo. Inhibitors to Hsp90 are currently being investigated in a growing number of clinical trials
http://www.clinicaltrials.gov. Thus, the present study not only adds an interesting new aspect to the multiple mechanisms of Hsp90-inhibition, but also provides reasonable evidence that an ATF3-induction by Hsp90 inhibition could be favorable for therapy of advanced colon cancer.
Acknowledgements
The authors thank Christine Wagner and Kathrin Stengel for excellent technical assistance.
These studies were supported in part by the AACR-Littlefield Grant for Metastatic Colon Cancer Research (American Association for Cancer Research, Philadelphia, PA) (O.S.), the German Cancer Aid (Deutsche Krebshilfe, Max-Eder Nachwuchsgruppen-Programm, Bonn, Germany) (O.S.), and a grant from the University of Regensburg, Medical Faculty (ReForM) (S.A.L.; C.M.)
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CH - performed majority of experiments, adjusted study design and contributed to manuscript preparation; SAL - performed experiments and involved in study design; CM - performed experiments; AM - performed experiments and aided in animal study; SF - performed statistical analyses and helped with manuscript preparation; CH - collaborator for cell culture experiments; WD - provided human tissues from tissue bank and performed analyses of human specimens; HJS - manuscript editing and study refining; EKG - manuscript preparation and animal study supervision; OS - principal investigator, manuscript preparation, study design, animal study supervision. All authors read and approved the final manuscript