Background
DDX3 (encoded by
DDX3X) is a DEAD box RNA helicase with ATPase dependent helicase activity, which is involved in several steps of endogenous RNA metabolism and translation initiation [
1‐
4]. DDX3 has been implicated in neoplastic transformation due to its role in cell cycle progression [
5,
6] and its anti-apoptotic properties [
7‐
9]. In addition, DDX3 has been shown to promote several steps of tumor metastasis. Overexpression of DDX3 resulted in increased motility and migration by induction of an epithelial-to-mesenchymal (EMT) phenotype with loss of E-cadherin [
10,
11] and upregulation of Snail expression [
12]. Furthermore, DDX3 was found to promote anchorage independent growth and invasive capacities of cancer cells through regulation of mRNA translation [
13,
14]. DDX3 knockdown has also been shown to result in reduced breast cancer metastases in mice [
15]. These findings have led to the development of DDX3 inhibitors for the treatment of breast cancer [
15] among other malignancies [
5,
6,
16,
17]. The tumor-enhancing role of DDX3 was corroborated by studies on DDX3 expression in patient samples of primary tumors [
5,
6], but DDX3 expression was never studied in metastatic cancer samples.
Although therapeutic options for patients with metastatic breast cancer have increased, the vast majority of patients still develops resistance to treatment and eventually succumbs to the disease [
18]. With 5-year survival rates of 25% [
19] and approximately 40,000 deaths on a yearly basis in the United States, metastatic breast cancer still ranks second on the list of causes of cancer deaths in women, accounting for 15% of all cancer deaths [
20]. Therefore the identification of novel therapeutic targets that inhibit the development and outgrowth of breast cancer metastases remains urgently wanted.
Upregulation of DDX3 in metastases would confirm the role of DDX3 in metastatic tumor progression that has been suggested in functional studies. In addition, high DDX3 expression levels in metastatic lesions could indicate that breast cancer metastases are reliant on high DDX3 expression, and that patients with advanced disease could benefit from treatment with DDX3 inhibitors under development. Therefore, this study aimed to evaluate DDX3 expression in distant breast cancer metastases as compared to their primary tumor.
Discussion
DDX3 is an RNA helicase with oncogenic properties, which has been found to promote metastasis in functional studies. However, DDX3 expression had never been specifically evaluated in metastatic cancer patient samples. In this study, we therefore compared DDX3 expression in primary breast cancers to that in corresponding distant metastatic lesions. Cytoplasmic DDX3 expression was significantly higher in metastatic cancer samples, especially in brain metastases and triple negative cases. In addition, there is a correlation between DDX3 expression in the metastasis and worse overall survival in patients with metastatic breast cancer.
Previous studies have indicated that DDX3 overexpression facilitates dissemination of cancer cells through induction of an EMT phenotype [
10‐
12]. Increased motility and anchorage independent growth have also been linked to the role DDX3 has in mRNA translation. Chen, et al. found DDX3 to increase invasive properties through a direct role in Rac1 translation, which in its turn stabilizes β-catenin expression resulting in activated Wnt-signaling [
14]. Furthermore, Hagerstrand, et al. found that DDX3 mediates IRES-dependent translation, resulting in increased anchorage independent growth in cancers with 3q26 amplification. In addition to promoting the dissemination process, our finding that among patients with established metastases, those with DDX3 expression have worse overall survival is suggestive of DDX3 also providing a survival benefit to cancer cells after colonization of the metastatic niche. However, this difference can also be partly attributed to the frequent triple negative phenotype and brain localization of metastases with high cytoplasmic DDX3 expression. Notably, there are some contradictory reports in literature pointing towards DDX3 functioning as a tumor suppressor [
28,
29]. It is possible that the role of DDX3 in oncogenesis differs between genetic backgrounds and cancer types [
30].
The mechanisms behind cytoplasmic overexpression and nuclear retention of DDX3 remain largely to be elucidated. Mutations in DDX3 have been detected in medulloblastomas [
31], head and neck cancers [
32] and hematological malignancies [
33,
34], but were not identified in breast cancers [
35]. In addition, there is no amplification of the DDX3 locus in DDX3 overexpressing breast cancer cell lines [
10]. With regard to nuclear retention of DDX3, we know that DDX3 is exported out of the nucleus as part of messenger ribonucleoprotein complexes [
2,
26,
36]. In the nucleus, DDX3 was previously found to localize to the nucleolus [
37] where ribosomal assembly takes place, suggesting that nuclear DDX3 retention in metastases possibly reflects increased demand in protein synthesis. More research to further clarify the mechanisms of DDX3 overexpression and nuclear retention is needed.
We found a particularly large increase in cytoplasmic DDX3 expression rates in brain metastases. Brain metastases are more common in patients with triple negative or HER2 amplified primary breast cancers [
38], which have relatively high DDX3 expression. However, discordance rates for DDX3 were much higher (48% upregulation) than for HER2 (2%) and estrogen receptor (13%) [
21]. It is therefore hard to explain the DDX3 upregulation in brain metastases rates solely by an association with these molecular subtypes. Several other biological signatures have been found to characterize brain metastases. Wnt signaling mediates metastasis to the brain in both lung [
39] and breast cancer [
40]. DDX3 is a multilevel activator of the Wnt-signaling pathway [
5,
6,
14,
41] and therefore potentially facilitates brain colonization in a Wnt-mediated fashion. Another feature of brain metastases is the expression of DNA repair genes [
42,
43]. Inhibition of DDX3 reduced non-homologous end joining, a double strand break repair mechanism [
5], implying that the high DDX3 levels in brain metastases could reflect a DNA damage response. Furthermore, overexpression of hypoxia-inducible factor 1α is common in brain metastases [
44] and also associated with DDX3 expression in breast cancer [
45]. However, metastatic DDX3 expression did not correlate with expression of the HIF-1α target genes carbonic anhydrase IX (CAIX) and Glucose transporter 1 (GLUT-1; data not shown), making it unlikely that high DDX3 expression in brain metastases is hypoxia-mediated. Last, metastatic niches differ also by the bioenergetic profile they impose on cells [
46]. Brain metastases were demonstrated to upregulate glycolysis and oxidative phosphorylation capacity [
47] and to have increased hexokinase 2 expression [
48]. An additional reason for brain metastases to elevate DDX3 expression could be that DDX3 supports metabolic adaptation of cancer cells to the microenvironment of the brain. Although liver and bone metastases are also common in breast cancer patients, limited availability of tissue from these sites did not allow for subgroup analysis.
Besides biological relevance, high DDX3 expression in breast cancer metastases has potential clinical applications. Metastatic breast cancer, especially localized in the brain, is associated with short patient survival and severely impaired quality of life. Cerebral metastases occur early in triple negative cases [
49], where the systemic therapeutic arsenal is particularly lacking. High DDX3 expression could serve as a therapeutic target in these patients. There are several small molecule inhibitors of DDX3 currently under development [
50]. Although diffusion of these compounds over an intact blood brain barrier is limited [
5], the small size of the inhibitors and the compromised blood brain barrier in brain metastases [
51] potentially do allow for therapeutic levels to be reached. The DDX3 inhibitor RK-33 has potent radiosensitizing abilities [
10], which could enhance the effect of whole brain radiation to treat brain metastases. Furthermore, given the role of DDX3 early in the metastatic process, DDX3 inhibitors could potentially also be used to prevent the emergence of metastases. At last, evaluation of DDX3 expression in patient samples could serve both as a prognostic biomarker and facilitate selection of those patients benefiting most from DDX3 inhibitors.
Acknowledgements
The Dutch Breast Cancer Consortium included Department of Pathology, University Medical Center Utrecht, Department of Pathology, Academic Medical Center, Amsterdam; Department of Pathology, Medical Center Alkmaar; Department of Pathology, Medical Center Zaandam; Department of Pathology, Radboud University Nijmegen Medical Center, Nijmegen; Department of Pathology, Canisius Wilhelmina Hospital, Nijmegen; Department of Pathology, VU University Medical Center, Amsterdam; Department of Pathology, The Netherlands Cancer Institute, Amsterdam; Laboratory for Pathology, Dordrecht; Department of Pathology, University Medical Center Groningen; Department of Pathology, St Antonius Hospital, Nieuwegein; Department of Pathology, Diakonessenhuis, Utrecht, Isala klinieken, Zwolle; Erasmus Medical Center Rotterdam; Gelre Hospital Apeldoorn, Laboratory Sazinon, Hoogeveen, and the Laboratory for Pathology Oost Nederland, Enschede, all in The Netherlands.