Background
The HER-2/neu oncogene is amplified two-fold to > 20-fold in approximately 25% of breast cancers. Overexpression or gene amplification of HER2 is associated with poor prognosis and an aggressive course of the disease, such as oncogenic transformation, tumorigenesis, and metastasis [
1]. HER2 belongs to the Type I receptor tyrosine kinase family, which includes four family members: EGFR (HER1), HER2 (neu or ERBB2), HER3, and HER4 [
2]. In spite of possessing no known ligand, HER2 is the preferred heterodimerization partner within the family and can form heterodimers with HER1 and HER3, leading to phosphorylation of tyrosine residues within the cytoplasmic domain [
3,
4]. Under certain circumstances, HER2 interacts with its binding partners, including MUC4, HSP90 or gp96, which may stabilize HER2 and make it endocytosis defective [
5‐
8]. As a consequence HER2 heterodimerization with other HER members is resistant to downregulation, and induces a variety of signal transduction pathways, such as the PI3K/AKT, Ras/MAPK, and JAK/STAT pathways, leading to cell transformation and cancer [
9,
10].
Due to its central role in aggressive tumor growth and metastases, HER2 serves as an ideal target for monoclonal antibody therapy, including the HER2 signaling inhibitor trastuzumab and pertuzumab, which can effectively treat tumors with HER2 gene amplification in 25% of patients as monotherapy and 50% when given with chemotherapy [
11]. The application of HER2-targeted therapy dramatically changes the clinical outcome for HER2-positive breast cancer patients, even providing a superior prognosis compared to HER2-negative cases. However, acquired and inherent resistance to anti-HER2 therapy in these patients poses a serious challenge, and a better knowledge of the underlying mechanisms of sensitivity to anti-HER2 therapies is of extreme importance for the development of effective strategies to overcome resistance [
12].
HER3 is overexpressed in 10–30% of breast cancer and is also associated with poor prognosis and worse survival [
13]. The most important and well-understood signaling activity of HER3 is its unique and potent ability to activate downstream PI3K and AKT pathway signaling, which subsequently controls many biological processes critical for tumorigenesis, including translation, survival, anti-apoptosis, metabolic regulation, and cell cycle control [
14]. In addition, HER3 is frequently co-expressed with HER2 in breast cancer, and high levels of HER2/HER3 dimerization are associated with poor survival prognosis in HER2-overexpressing breast cancer [
15]. Of note, inhibition of HER2 or the PI3K-AKT pathway in HER2-overexpressing cells is followed by feedback upregulation of HER3, leading to attenuation of the response to inhibition [
16,
17]. In HER2-positive metastatic breast cancer, high HER3 expression is linked to poor survival prognosis after anti-HER2 treatment [
18,
19]. All of these studies indicate that there exists crosstalk and co-operativity between HER2 and HER3 expression, and thus, simultaneously targeting HER2 and HER3 may provide a more efficient therapy for breast cancer.
MicroRNAs (miRNAs) are a large family of small noncoding RNA molecules of approximately 22 nt in length that inhibit target gene expression by affecting mRNA stability or/and translational efficacy [
20]. Mature miRNA duplexes are loaded onto the RNA-induced silencing complex (RISC), which contains a member of the RNA binding protein Argonaute family (Ago). They then pair with target sites (miRNA response elements, MREs) within the 3’untranslated regions (3’UTRs) of mRNAs to direct posttranscriptional downregulation [
21]. Long noncoding RNAs (lncRNAs), pseudogenes, and even viral RNAs have been demonstrated to function as competitive endogenous RNAs (ceRNAs) that elevate the expression of the corresponding protein-coding genes with shared miRNA binding sites via competing with miRNA binding and derepressing the expression of these genes [
22‐
24]. The relative abundance of ceRNAs vs. corresponding RNAs, levels of common miRNAs, and the number of MREs may all contribute to ceRNA interactions, according to a mathematical mass-action model for ceRNA networks [
25]. The mechanism is of particular relevance to HER2-positive breast cancer, where amplification of the HER2 gene leads to an enormous number of HER2 mRNA transcripts [
26].
Considering that HER2 gene amplification generates highly redundant mRNAs that harbor multiple miRNA binding sites, we speculate that HER2 mRNA acts as ceRNA and can sequester endogenous miRNAs within HER2 positive breast cancer cells, thus cross-regulating the stability and translational efficiency of other host mRNAs with shared miRNA response elements. In this study, we investigated the role of HER2 mRNA in breast cancer by analyzing potential HER2 mRNA-regulated miRNAs and the corresponding mRNA profiles. We further explored the impact of the interactions between HER2 and its corresponding mRNA on breast cancer growth and tumorigenesis. Our results provide valuable insights into the functional implications of HER2 mRNAs in anti-HER2 resistance.
Methods
Cell line
Human breast cancer cell lines AU565, BT474, T47D, and MCF7 and the human kidney 293 T cell line were obtained from Cell Resource Center, IBMS, CAMS/PUMC, China. Cell lines were passaged for < 6 months after receipt. Cell lines which were passed for > 6 months were identified by STR Classification. All cell lines were regularly tested negative for Mycoplasma. BT474.TtzmR and AU565.TtzmR sublines with acquired trastuzumab resistance were generated by continuous exposure of parental cells to increasing doses of trastuzumab (up to 10 μg/ml) for more than 6 months.
sgRNA-CRISPR/Cas9 system design and construction
Potential target sites were predicted using “
crispr.mit.edu” in the human genome, and two to three target sequences with low predicted scores for off-targets were chosen. Two complementary 20-bp oligonucleotides were annealed and cloned this into
BbsI-digested pSpCas9(BB)-2A-Puro (PX459). Then, cells were transfected with CRISPR/Cas9 sgRNA. Transfected cells were treated with puromycin at a concentration of 1 μg/ml. Surviving cells were sorted into 96-well plates by FACS. The genomic region encompassing the CRISPR/Cas9 target site was amplified and sequenced.
Antibodies and reagents
Trastuzumab (Herceptin) was purchased from a pharmacy. PE-anti-HER2, APC-anti-HER3 antibody was from BioLegend (San Diego, California). Ago2-antibody was purchased from Abcam. All other antibodies were purchased from Cell Signaling Technology. The chemically synthesized specific siRNA, miRNA mimics, miRNA inhibitors, and non-specific control, as well as cholesterol-conjugated siRNA and miRNA mimics and control mimics, were purchased from RiboBio Co., Ltd. (Guangzhou, China).
Real-time PCR
Total RNA was extracted with Trizol Reagent and quantified by real-time PCR using the SYBR Green Premix Reagent (Takara Bio Inc., Shiga, Japan) with an internal control for normalization.
TaqMan miRNA analysis
Real-time PCR analysis for miR-125a/b was performed using a TaqMan miRNA Kit (Applied Biosystems). The U6 endogenous control was used for normalization. Relative expression was quantified using the comparative threshold cycle (Ct) method.
Luciferase assay
To validate miRNA targeting, the 3’UTRs of HER2 and HER3 were cloned into the pGL3 firefly luciferase reporter plasmid. Cells were transfected in 48-well plates with 20 ng of pGL3 reporter, 2 ng of pRL-TK as the control, and 100 nM miRNA mimic. Firefly luciferase and Renilla luciferase activities were measured consecutively with the dual luciferase reporter system (Promega), and the firefly luciferase activity was normalized to that of Renilla luciferase after 48 h. To test the ceRNA activity of the HER2 3’UTR, 5× 104 293 T cells were transfected in 48-well plates with 20 ng of pGL3-HER3 3’UTR and 250 ng of pCDNA3.1-HER2 3’UTR or pCDNA3.1 as a control, as well as 1–10 nM miRNA mimic. Firefly luciferase and Renilla luciferase activities were measured consecutively with the dual luciferase reporter system (Promega), and the firefly luciferase activity was normalized to that of Renilla luciferase after 48 h.
Transcriptional profiling by microarray
Comparative microarray analysis of mRNAs from T47D cells transfected with pCDNA3.1-HER2 3’UTR and pCDNA3.1 as a control was performed on an Agilent Whole Human Genome Microarray at the Shanghai Biotechnology Corporation according to manufacturer’s instructions.
RNA immunoprecipitation
Cells were washed and lysed. Ago2-antibody or control IgG was incubated with ProteinG Sepharose beads (GE Healthcare). The beads were pelleted and washed and then subsequently incubated with cell lysate. After incubation, the beads were washed. RNA was isolated from the immunoprecipitated pellet by adding Trizol reagent. Total RNA was used for reverse transcription and real-time PCR analysis. The following primers were used: Ago-HER2 forward, 5’-AGCCGCGAGCACCCAAGT-3′ and reverse, 5′-TTGGTG GGCAGGTAGGTGAGTT’; and Ago-HER3 forward, 5’-GGGTTAGAGGAAGAGGATGTCAAC-3′ and reverse, 5′- GGGAGGAGGGAGTACCTTTGAG’.
Transcript copy-number analysis
RNA was extracted from 1 × 105 cells using Trizol Reagent. Absolute quantification of total HER2 and HER3 mRNAs and miR-125a/b was performed by real-time PCR. For a standardized evaluation, threshold cycle (CT) values were compared to a 10-fold serial dilution of either in vitro-transcribed HER2 or HER3 mRNAs or synthetic miR-125a/b mimics (RiboBio Co., Ltd). Then, the copies of transcript per cell were calculated with standard stoichiometric methods.
Animal studies
Tumors were established by orthotopic injection of 1 × 107 cells/mouse in 200 μl of PBS into the flanks of six-week-old female nude mice. The mice were randomly divided into groups at 15 d after inoculation. Tumor growth was measured every 3 days. Trastuzumab (10 mg/kg) was given intraperitoneally (i.p.) once a week. For delivery of cholesterol-conjugated siRNA, 10 nM siRNA in 0.1 ml PBS was locally injected into the tumor mass every 3 days for 2 weeks. Tumor growth was measured twice weekly, and the volume of the tumors was calculated as volume = length× width2 / 2.
Immunohistochemistry analysis
Tissues were fixed in 4% paraformaldehyde overnight and embedded in paraffin according to standard procedures. Tissue sections were stained with the following procedures. Briefly, after deparaffinization and rehydration, antigen retrieval was performed using antigen retrieval butter in an autoclave at 121 °C for 100 s. Slides were then incubated with primary antibodies at room temperature for 40 min. Slides were washed with PBS and stained with fluorescence-conjugated secondary antibodies. Images were acquired using a Leica TCS SP8 confocal laser-scanning microscope (Leica Microsystems, Heidelberg, Germany).
Breast primary tumor cells isolation and transfection
Informed written consents were obtained from breast cancer patients according to the General Hospital of People’s Liberation Army. Tumor biopsy was washed with PBS containing 100 U/ml penicillin and 100 μg/ml streptomycin, and was cut up into small pieces. Small tissues were minced in RPMI1640 medium. To obtain single cell suspensions, samples were treated with 1 mg/ml collagenase type IV and 1 mg/ml hyaluronidase followed by digestion with trypsin-EDTA at 37 °C for about 4 h with periodic agitation. After digestion, cells were centrifuged and cultured in RPMI1640. Fibroblast like elongated cells were visible firstly. Later, epithelial like cells started to make colonies in dome like shapes. These cell colonies were isolated and transferred to cell plates. These cells were passaged and expanded for transfection experiments.
Transfections were carried out using electroporation with Bio-rad transfection system according to the manufacturer’s instructions. Briefly, 1 × 106 cells were transfected with 10 μg plasmids using a BioRad Gene Pulser II at 250 V and 950 μF. Each treatment was performed for at least three times.
Expression data from the cancer genome atlas
The transcriptome expression profiles of breast cancer were downloaded from The Cancer Genome Atlas (TCGA) data portal (
https://cancergenome.nih.gov). In this study, the transcriptome profiles of 887 cases and 101 HER2-positive breast tumors were included in the coexpression analysis. Level 3 Illumina miRNASeq was used to analyze miRNA expression. For the miRNASeq data, “reads_per_million_miRNA_mapped” values were used to calculate miRNAs.
Statistical analyses
Data are expressed as the mean ± SD (standard deviation) from three independent experiments. The statistical significance between two and more than two groups was measured using the two-tailed Student’s t-test. P values < 0.05 were considered significant.
Discussion
HER2 heterodimerization with other HER members leads to the most potent of receptor combinations for causing continual downstream PI3K/Akt, Ras/MAPK, and JAK/STAT signaling, which drives oncogenic transformation and breast tumor growth. In the present study, we investigated whether highly expressed HER2 mRNAs in HER2 gene-amplified breast cancer could directly affect the expression of other HER members via shared MREs. We focused on HER3 because, as a specialized allosteric activator, it has a unique and potent ability to activate the downstream PI3K and AKT pathways. We showed herein that the HER2 3’UTR derepresses HER3 by sequestering cellular miR-125a/b in breast cancer cells and tumors. The tumor growth promotion effect of the HER2 3’UTR was nearly comparable to the HER2 CDS (Fig.
4). Our study establishes the HER2 3’UTR as a potent oncogenic transcript that elevates the expression of HER3 and activates downstream AKT and PI3K pathways in a MRE-dependent and coding-independent manner. These results are consistent with the observation of the positive correlation between HER2 and HER3 mRNA levels in breast cancer patients. Remarkably, we demonstrated that inhibition of HER2 with trastuzumab results in time- and dose-dependent upregulation of HER3 mRNA and protein by the HER2 3’UTR, which confers trastuzumab resistance. Our results may therefore provide further dissection of the impact of HER2 overexpression on both HER3 mRNA regulation and anti-HER2 resistance, as shown in Fig.
6g. In this model, HER2 mRNA-mediated HER3 up-regulation plays an important role in breast cell transformation, tumor growth, and resistance to anti-HER2 therapy. These data imply that targeting HER2/HER3 mRNAs may provide an effective strategy for the treatment of HER2-positive breast cancer.
In HER2-amplified cancer, activation of HER3 may occur through high-level expression of heterodimerization with HER2 [
30]. HER3 is classified as a pseudokinase which lacks intrinsic kinase activity. HER2 has no known ligands but can dimerize with other HER family members, especially HER3. HER3 can be phosphorylated and activated by residual HER2 activity. Phosphorylated HER3 in turn activates PI3K via its six docking sites for the p85 adaptor subunit of PI3K. Due to HER3-mediated compensation, current clinical therapy against HER2 will not block PI3K pathway completely, resulting in unrestrained PI3K-AKT-mTOR signaling that is essential for tumorigenesis and drug resistance [
14,
31,
32]. In breast carcinoma, HER3 levels can significantly increase due to overexpression instigated via gene amplification [
33], transcription, and protein translation [
15,
17,
34] or via enhanced molecular stability by dimerization [
27]. miRNAs affect HER3 expression at the post-transcription level [
29,
35‐
38]. Both HER2 and HER3 are validated targets of miR-125a and miR-125b [
28].
HER2/HER3 co-overexpression is significantly associated with metastasis and shorter survival in breast cancer [
13,
39]. All of these studies indicate that there may be crosstalk and cooperativity between HER2 and HER3. Our current work suggests that besides by HER2 protein heterodimerization, HER2 overexpression in breast cancer up-regulates HER3 via the HER2 3’UTR, which acts as a sponge to bind and sequester endogenous miR-125a/b. The HER2 3’UTR increases HER3 mRNA levels in a miR-125a/b response element- and miR-125a/b-dependent manner. HER2 3’UTR-induced miR-125a/b sequestration results in reduced Ago2 binding and HER3 mRNA derepression. In addition, a low ratio of miR-125a/b: HER2mRNAs and high ratios of miR-125a/b: HER3 mRNA and HER2:HER3 mRNAs were displayed in HER2 gene-amplified and protein-overexpressing breast cancer cells (Fig.
5a), which further indicates that the interaction between HER2 and HER3 mRNA may occur under such a circumstance. We consider that HER2 3’UTR preferentially regulates HER3 mRNA level but not vice versa, as the binding affinity between HER2 3’UTR and miR-125a/b was relatively higher than that between HER3 3’UTR and miR-125a/b by the maximum free energy predicted hybridization configurations (data not shown) and luciferase reporter assay (Fig.
3a and Additional file
9: Figure S5), and a high ratio of HER2:HER3 mRNAs were found in HER2 gene amplified cells (Fig.
5a). Besides HER3, expression levels of HER4 in HER2 3’UTR-transfected cells were also increased (Fig.
2a, k). MiRNA prediction indicated that 46 miRNAs have binding sites in both human HER2 and HER4 3’UTRs (Additional file
3: Table S3). These data suggested that HER2 may also regulate HER4 expression through ceRNA crosstalk.
Trastuzumab and pertuzumab, monoclonal antibodies targeting HER2 homodimerization and heterodimerization, as well as the HER2 tyrosine kinase inhibitor (TKI), display considerable clinical efficiency. However, the high prevalence of drug resistance after continuous treatment is a major concern [
40,
41]. Several lines of evidence demonstrate that upregulation of HER3 is one of the main roads to resistance to anti-HER2 therapies [
42,
43]. In current study we found that trastuzumab treatment upregulated HER3 expression via elevated HER2 3’UTR. Although anti-HER2 therapies lead to substantially decreased HER2 level, upregulated HER3 may form dimmers with other kinase including EGFR, FGFR, Met, Src that could also phosphorylate and activate HER3 [
16,
44‐
46]. In addition, even the maximal doses of trastuzumab did not totally deplete HER2 expression (see Additional file
10: Figure S6), it is possible that the residual HER2 kinase activity was enough to partly maintain HER3 phosphorylation. All these could lead to resistance to HER2-targeted therapies. Breast cancer patients with high levels of HER2/HER3 dimerization have poor survival prognosis under treatment with adjuvant trastuzumab [
15]. Moreover, a phase II clinical trial study shows that that low HER3 mRNA may represent a pertuzumab-sensitive phenotype in an enriched ovarian cancer patient [
47]. Besides, elevated expression of HER3 and MUC4 and their interactions that possibly induced by increased phosphorylation of ERK and expression of PI3K and c-Myc were observed in HER2 knockdown pancreatic cancer cells, leading to increased cell proliferation, motility and tumorigenicity [
48]. The feedback mechanisms of inhibition of HER2 or the PI3K/AKT pathway lead to a rebound in HER3 expression and signaling, which provides a rationale for the combination of targeted-therapies and drugs targeting HER2 and HER3. Indeed, dual targeting of HER2 and HER2/HER3 dimerization with trastuzumab and pertuzumab results in enhanced tumor inhibitory effects in mouse xenograft models, and significantly prolongs progression-free survival in HER2-overexpressing breast cancer [
11,
49,
50].
Consistent with these previous studies, our findings show that continuous treatment with trastuzumab leads to a substantial down-regulation of HER2 protein levels but a compensatory upregulation of HER2 mRNA, thereby leading to increased HER3 mRNA levels through ceRNA crosstalk. Trastuzumab upregulated HER3 in a miR125a/b-dependent manner as elevated HER2 mRNA by trastuzumab treatment sequestered endogenous miR-125a/b by its 3’UTR and subsequently derepressed HER3 mRNA. Targeting of HER3 with siRNA and/or mutation of the miR-125a/b responsive element within the HER2 3’UTR sensitized HER2-overexpressing breast cancer cells and xenografts to trastuzumab. Our study therefore presents an effort to address the mechanism of regulation of HER3 mediated under anti-HER2 treatment, which further supports the notion that inhibition of HER3, especially targeting HER2 mRNA might provide a novel therapeutic approach for HER2-targeted therapies in breast cancer.
The sensitivity of breast cancer to trastuzumab is directly related to HER2 expression levels, and elevated HER3 is involved in trastuzumab resistance. Although transfection of HER2 CDS and full-length gene both led to increased HER3 expression, they highly elevated HER2 expression in the meantime (see Fig.
4c and g), transforming T47D cells into HER2-positive cells and therefore making these cells more sensitive to trastuzumab. In contrast, HER2 3’UTR only moderately increased HER2 level. This may explain why cells stably expressing the HER2 3’UTR displayed more resistance to trastuzumab treatment than those expressing HER2 CDS or full-length gene (see Fig.
5j). As trastuzumab-induced upregulation of HER3 mRNA was largely dependent on miR125a/b and the HER2 3’UTR (see Fig.
5), we believe that HER2 3’UTR-mediated HER3 upregulation plays an important role in the increased HER3 expression under trastuzumab treatment. In addition, it is worthwhile to fully dissect the cross-regulatory function of HER2 mRNAs on other pathways and their casual role in breast cancer development and drug resistance.
Acknowledgments
We thank Fulian Liao and Lihong Zhou for providing technical help in culturing cells. We thank Tong Zhao for FACS sorting. We also thank Kuan Li and Xinqi Jiang for data analysis.