Here we have sought to identify differentially expressed miRNAs in ES xenografts and to investigate the underlying molecular changes by integration of these results with aCGH analysis of the same samples.
MiRNA expression profile of ES xenografts
Xenografts displayed 60 differentially expressed miRNAs that distinguished them from control samples (Human mesenchymal stem cells). Of these, 46 miRNAs were exclusively expressed in xenografts while 2 (miR-31 and miR-31*) miRNAs were exclusively expressed in controls. The remaining 5 miRNAs (miR-106b, miR-93, miR-181b, miR-101, miR-30b) were significantly over-expressed while 6 miRNAs (miR-145, miR-193a-3p, miR-100, miR-22, miR-21, miR-574-3p) were significantly under-expressed in xenografts. The expression profiles of 4 miRNAs (miR-31, miR-31*, miR-106b, miR-145) were confirmed by RT-PCR.
To evaluate the potential role of the differentially expressed miRNAs, three databases were searched for the known ES-associated genes targeted by these miRNAs, by applying target prediction algorithms. The targets included EWSR1 (GeneID: 2130), FLI1 (GeneID: 2313), SOX2 (GeneID: 6657), p53 (GeneID: 7157), IGFBP3 (GeneID: 3486), IGF1 (GeneID: 3479) and IGF1R (GeneID: 3480). The differential expression of the miRNAs regulating these genes may play a role in the tumorigenesis and tumor progression of ES.
Interestingly, miR-150, which targets the tumor suppressor gene
TP53, was expressed in all xenograft samples but in none of the control samples. This is in accordance with the study of Fabbri and colleagues [
22] who have included TSGs in their investigation of likely over-expressed miRNA target genes. In addition, one of our xenograft series (Case number 451) showed losses at 17p, containing
TP53, that appeared in later passages. Previous ES studies have shown that, despite the low frequency of mutations in
TP53, an alteration of
TP53, in conjunction with the deletion of
CDKN2A, is associated with a poor clinical outcome [
23,
24]. Moreover, the homozygous deletion of this gene has been reported in a small subset of ES patients [
25,
26].
The IGF-1 pathway, whose genes
IGF1R, IGF-1 and
IGFBP-3 are among the target genes of the differentially expressed miRNAs, plays a critical role in cancer development, including ES [
26‐
28].
IGF1R is targeted by miR-145 and miR-31*, and previous studies have shown
IGF1R to be a direct target of miR-145 [
29] as well as to be over-expressed in Ewing tumors [
27,
28]. As for
IGF-1, it is the target of 11 miRNAs including miR-21, miR-31, miR-145, miR-150, miR-194, miR-215, miR-421, miR-486-5p, 548c-5p, and miR-873. Interestingly,
IGFBP3, which is among the target genes of miR-150*, was, in our study, expressed in all xenografts but not in control samples.
IGFBP-3, which is a major regulator of cell proliferation and apoptosis, inhibits the interaction of IGF-1 with its receptor (IGF1R) [
30‐
33]. Indeed, it has been reported that high IGF-1 and low IGFBP-3 levels in serum increase the risk of cancer [
26]. IGFBP3 is strongly down-regulated by the
EWS/FLI-1 fusion gene [
34], which is able to induce expression of embryonic stem cell gene
SOX2. Consequently,
SOX2 participates in ES cell proliferation and tumorigenesis and might play a central role in ES pathogenesis [
35]. As for our study,
SOX2 was among the target genes of miRNA-21 that showed under-expression in xenografts. Another under-expressed miRNA, miR-145, was previously found to target
FLI1 and its increased expression leads to a decreased migration of microvascular cells in response to the growth factor gradients
in vitro [
36].
Finally, miR-106b targets EWSR1, which undergoes a chromosomal translocation to produce the EWS-FLI fusion gene in a majority of ES cases, where it is commonly considered to trigger the condition. The action of miR-106b is, thus, likely to only impact on the original/unmodified locus for EWSRI since the EWS-FLI lacks the 3' portion of EWSR1. Further studies would, naturally, be required to confirm this hypothesis.
The alteration of 41 miRNAs was observed in xenograft passages derived from lung metastatic, which may play a crucial role in triggering tumor metastasis. Eight of these miRNAs, all located at the 14q32 imprinted domain (miR-154*, miR-337-3P, miR-369-5p, miR-409-5p, miR-411, miR-485-3p, miR-487a, miR-770-5p) were not expressed in metastasis xenografts but in control samples, thus suggesting a tumor suppressor function. Interestingly, gastrointestinal stromal tumors (GISTs) have displayed 44 expressed miRNAs originatingfrom the 14q32 chromosomal region, for which the low expression of miRNAs was related to tumor progression [
37]. A report by Saito and colleagues [
38] suggests that miRNAs located in this region function as tumor repressor genes and changes in the methylation status of their promoters could trigger cancer development. This evidence suggests that the miRNAs identified in our study may act as tumor repressors and their absence could increase the risk of metastasis and tumor progression in ES.
Copy number aberrations in ES xenografts
The most recurrent copy number alterations detected in our CGH analysis (gains at chromosome 8, 1q and losses at 9p21.3 and 16q) are in agreement with other findings on ES patients [
1,
39‐
46]. The crucial role of these changes, gains in 1q, 8 and losses of 9p21.3 (including loss of
CDKN2A) and 16q, has been clarified by notable tumor development and adverse clinical outcome [
42,
47,
48]. These copy number changes were seen throughout the whole xenograft series. In all passages of lung metastasis, losses were observed at 1p36.12-pter/1p36.21-pter. Of note, deletion of this site (1p36) has been found to be related to a poor clinical outcome in ES[
43,
47]. The loss of 1p36.12-pter in the first two passages originating from lung metastasis (1 and 4) changed to loss of 1p36.21-pter in the last three passages (14, 21 and 30). The lung metastasis xenografts showed 9 copy number changes, whereas only 3 of these aberrations were observable in the xenograft passages from its primary tumor. Likewise in many tumors during the disease progression, the increase of genomic instability is also seen here. This instability most probably explains the variation of the size of 1p deletion. The fact that the terminal part is retained in the deletion emphasizes the importance of 1p36.21-pter region in the selection and in the disease progression.
Somatic mosaicism/heterogeneity occurs commonly in tumors and plays an important role in the progression of the tumor and, thereby, can also explain why some xenograft passages show copy number changes and others do not.
Integration of miRNA expression profiles and DNA copy number changes
DNA copy number abnormalities can have a direct impact on miRNA expression [
49]. In the current study, 20 differentially expressed miRNAs were located in the copy number altered regions. These findings are in accordance with Calin et al. (2004) who observed that half of the miRNAs are located in cancer-associated regions of chromosomes as well as in genomic regions frequently amplified or lost in cancer [
49]. The target genes for many of the changes are still unknown and, therefore, miRNAs could well be considered to be the drivers of the underlying changes.
MiR-31 and miR-31*, targeting
IGF1 and
IGF1R, are located at the frequently deleted region of 9p21.3, containing the
CDKN2A gene, which was frequently lost in our samples. Under-expression of miR-31 or deletion of the miR-31 genomic locus is also found in human breast cancers. This miRNA regulates metastasis by opposing local invasion and metastatic colonization in this cancer [
50‐
52]. Chromosome 1 gain is a frequent gain that contains the whole chromosome and seems to be poor prognostic sign [
53]. Interestingly, in our study five overexpressed miRNAs (miR-765, miR-135b, miR29c, miR-215, and miR-557) (Table
6) were associated to 1q gain. These candidate miRNAs have an important role and could explain the underlying mechanism in ES. Nevertheless, functional validations of the predicted target genes are needed to better understand their role in the ES tumorgnesis.