Introduction
Breast cancer is a heterogeneous disease encompassing a number of distinct entities, characterised by distinct biological features and clinical behaviour. Recent evidence has suggested that this heterogeneity is underpinned by distinct patterns of genomic aberrations, which may contribute to the different transcriptomic profiles and clinical phenotypes [
1‐
4]. Importantly, it has been shown that oestrogen receptor (ER)-positive and -negative breast cancers are fundamentally different diseases, with distinct transcriptomic profiles, gene copy number aberrations and somatic structural rearrangements [
5‐
8].
Based on the concept of oncogene addiction, we [
9,
10] and others [
11,
12] have demonstrated that the constellation of genes that are consistently overexpressed when amplified is enriched for genes selectively required for the survival of cancer cells harbouring their amplification and can be exploited as potential therapeutic targets. Using a combination of microarray-based comparative genomic hybridisation (aCGH) and gene expression profiling [
11‐
19], previous studies have identified genes which are consistently overexpressed when amplified and suggested potential "amplicon drivers" (for example,
FGFR1,
FGFR2,
GAB2,
PPAPDC1B and
ZNF703). It should be noted, however, that whilst many potential targets have been postulated, critical molecular drivers of several amplicons remain elusive.
It has now become evident that not all genes within an amplicon are overexpressed when amplified. For example, in the
HER2 amplicon, only 7 of the 13 genes that map to the smallest region of amplification are expressed at significantly higher levels when amplified [
20‐
22]. Conversely, evidence now suggests that an amplicon may harbour more than one driver [
10,
11,
17,
18,
23,
24]. For instance, within the 8p11.2 amplicon, the expression of FGFR1, PPAPDC1B, WHSC1L1, LSM1 and ZNF103 has been shown to be selectively required for the survival of cancer cells harbouring the amplification of these genes [
10,
11,
17‐
19,
25].
Amplification of the 19q12 locus has been reported to be found in up to 15% of ER-negative breast cancers [
9,
26]. This amplicon often encompasses the cell cycle regulatory gene
CCNE1, which has been shown to be overexpressed in a subgroup of ER-negative cancers. Although mRNA and protein expression are more prevalent than gene amplification,
CCNE1 has been postulated as the driver gene of this amplicon [
9,
26‐
28]. There is evidence, however, that genes within this amplicon other than
CCNE1 are consistently overexpressed when amplified [
29], including
POP4 and
C19ORF2.
The aims of this study were (i) to characterise the 19q12 amplicon in breast cancer, (ii) to determine the genes that are overexpressed when amplified in this amplicon, (iii) to investigate which of the genes mapping to this amplicon are selectively required for the survival of cells harbouring their amplification, and (iv) to determine if cancer cells with CCNE1 gene amplification are dependent on CCNE1 cell cycle-related functions for their survival.
Discussion
Here we demonstrate that the 19q12 locus is amplified in 5.1% of all invasive breast cancers and that this amplification is preferentially found in grade III ER-negative breast cancers. This is consistent with previous observations that suggested that 19q12 amplification is found in 5 to 15% of breast cancers and is associated with ER-negative disease [
9,
26,
27,
29]. Furthermore, through a combination of genomic profiling of primary breast cancers and breast cancer cell lines and RNAi experiments, we have demonstrated that the 19q12 amplicon may contain more than one 'driver'.
Previous studies have suggested that
CCNE1 is the likeliest driver of the 19q12 amplicon [
26,
27]. Although our results support the contention that
CCNE1 is one of the drivers of this amplicon, in this study
CCNE1 was amplified only in 5 out of 16 breast cancers harbouring 19q12 amplification. These observations are in agreement with those from studies of the 19q12 amplicon in gastric cancer, where
CCNE1 was amplified only in a subset of cases harbouring 19q12 amplification [
59], and with those of previous studies that have found a prevalence of
CCNE1 amplification in 1.2% to 1.4% of primary breast cancers [
14,
60]. Taken together, these lines of evidence suggest that
CCNE1 may not be the sole driver of this amplicon and that other genes within the 19q12 amplicon may also constitute drivers. Consistent with this hypothesis, we observed that cancer cells harbouring 19q12 amplification require the expression of not only CCNE1, but also PLEKHF1, POP4 and TSHZ3 for their survival.
PLEKHF1 (pleckstrin homology domain containing, family F (with FYVE domain) member 1) encodes a protein that is known to induce apoptosis through the lysosomal-mitochondrial pathway and triggers caspase-independent apoptosis [
61]. Recent evidence suggests that this process involves the recruitment of phosphorylated p53 and that silencing of endogenous p53 impairs its function [
62]. Despite its reported role in apoptosis, PLEKHF1 siRNA-mediated silencing in cancer cells harbouring its amplification did not lead to an increase in the sub-G0 proportion of cells. It could be speculated that in a
TP53 mutant background (as was the case of HCC1569 and MDA-MB-157; Additional file
2 Table S2),
PLEKHF1 gene amplification and overexpression may confer a selective advantage through mechanisms other than through the recruitment of phosphorylated p53. Regrettably, no breast cancer cell lines harbouring
PLEKHF1 gene amplification and wild-type
TP53 were available to test this hypothesis. All primary breast cancers with
PLEKHF1 amplification, however, were found to harbour
TP53 mutations, suggesting that
PLEKHF1 amplification and
TP53 gene mutations may have epistatic interactions and that
TP53 mutational status should be taken into account in the evaluation of the potential role of PLEKHF1 as a therapeutic target.
POP4 (processing of precursor 4, ribonuclease P/MRP subunit (S. cerevisiae)) encodes one of the protein subunits of the small nucleolar ribonucleoprotein complexes and is involved in the processing of precursor RNAs. Further studies investigating the mechanisms that lead to a survival advantage in cancer cells harbouring amplification and overexpression of
PLEKHF1 and
POP4 are warranted.
TSHZ3 (teashirt zinc finger homeobox 3) is a zinc finger transcription factor and has been shown to be involved in muscle cell differentiation [
63,
64]. Given that the
TSHZ3 gene promoter is reported to be frequently methylated in primary breast cancers and breast cancer cell lines [
65], this gene has been suggested to display tumour suppressor function. Our results do not corroborate this hypothesis and suggest that
THSZ3 is one of the drivers of the 19q12 amplicon, given that we demonstrate here that
THSZ3 is amplified in 2.6% of breast cancers, and that its silencing is selectively lethal in cancer cells harbouring its amplification. Although protein expression is not directly correlated with amplification, the breast cancer cell lines harbouring 19q12 amplification displayed either the highest levels of THSZ3 protein expression (that is, MDA-MB-157) or the presence of a THSZ3 isoform (HCC1569, Figure
4).
Although
CCNE1 amplification was restricted to a subset of cancers harbouring 19q12 amplification, we set out to investigate if breast cancer cells harbouring
CCNE1 amplification would be selectively dependent on the expression of this gene for their survival. siRNA-mediated silencing of
CCNE1 had a significantly greater effect on the survival of cancer cells harbouring
CCNE1 amplification than in those lacking its amplification. This is in agreement with a recent study in ovarian cancer that demonstrated that reduction of CCNE1 expression significantly inhibited cell growth in CCNE1 expressing cells, with a more profound effect in ovarian cancer cells harbouring
CCNE1 gene amplification [
66]. Moreover, forced expression of CCNE1 in cells with low expression has been previously shown to result in increased cell proliferation [
66]. Here we demonstrate that
CCNE1 siRNA silencing in breast cancer cells harbouring
CCNE1 gene amplification, but not in those lacking this amplification, resulted in a significant arrest in G1 (Figure
5 and Additional file
4 Figure S2). These findings provide a rationale for the apparent selective reduction in sensitivity to chemotherapy agents caused by
CCNE1 siRNA silencing reported in ovarian cancer cells harbouring
CCNE1 gene amplification [
67], and described here in breast cancer cells harbouring
CCNE1 gene amplification (Additional file
5 Figure S3).
Progression through the G1/S phase of the cell cycle is regulated through the partnership of CDK2 with its regulatory subunit CCNE1 [
55]. Given that
CCNE1 silencing in cancer cells harbouring
CCNE1 gene amplification leads to G1 arrest, we tested whether these cells would be dependent on CDK2 expression and kinase activity for their survival.
CDK2 siRNA silencing and inhibition of CDK2 kinase activity using a CDK1, CDK2 and CDK9 inhibitor (AZD5438) resulted in significantly higher reduction in survival of cancer cells harbouring
CCNE1 gene amplification. This is in agreement with a recent study based on the analysis of the conditional mouse models MMTV-Low Molecular Weight (LMW)-
Ccne1;
Tp53+/-;
Cdk2+/+, MMTV-LMW-
Ccne1;
Tp53+/-;
Cdk2+/- and MMTV-LMW-
Ccne1;
Tp53+/-;
Cdk2-/-. While mice with at least one functional copy of
Cdk2 consistently developed mammary gland tumours,
Cdk2-/- mice did not develop tumours through 24 months. Furthermore, administration of two Cdk inhibitors delayed the progression of mammary gland tumours in MMTV-LMW-
Ccne1;
Tp53+/-;
Cdk2+/+ mice [
68]. It should be noted that although MCF7 cells showed a similar sensitivity to the CDK1, CDK2 and CKD9 inhibitor AZD5438,
CDK2 siRNA silencing had a significantly more limited impact on the viability of these cells, suggesting that the sensitivity of MCF7 cells to AZD5438 is unlikely to be caused by inhibition of CDK2. Taken together, these results demonstrate that breast cancer cells harbouring
CCNE1 gene amplification are dependent on CDK2 expression and kinase activity for their survival and suggest that
CCNE1 amplification may constitute a potential biomarker of sensitivity to CDK2 inhibitors. It should be noted, however, that CDK1, CDK2 and CDK9 inhibitors may also be efficacious in a subgroup of ER-positive breast cancers, given that MCF7 cells also show sensitivity to AZD5438. Analysis of the results of clinical trials testing CDK inhibitors in breast cancer patients (for example, "Maximum Tolerated Dose (MTD) of Liposomal Doxorubicin in Combination With Seliciclib for Patients With Metastatic Triple Negative Breast Cancer" trial;
http://clinicaltrial.gov identifier NCT01333423) are warranted.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
RN and JRF conceived the study. RN, DW, MBK, MA, KK, FCG and AL carried out the experiments. RN and AM analysed the data. HH, MVV, BK, FR and JP provided samples. RN, PW, BW and JRF wrote the manuscript. All authors read and edited the manuscript, and approved its final version.