|Home | About | Journals | Submit | Contact Us | Français|
The human epidermal growth factor receptor 2 (HER2)/receptor tyrosine-protein kinasebB-2 (ERBB2) is overexpressed in 20–30% of breast tumors leading to faster growing and more aggressive tumors. Alternative splicing generates a functionally distinct HER2 variant called Herstatin, which is produced by the inclusion of intron 8. Herstatin acts as a tumor suppressor by effectively blocking HER2 activity and cell proliferation, while promoting apoptosis. In the present study we investigated HER2 pre-mRNA regulatory sequences and splicing factors which regulate the alternative splicing of Herstatin. A Herstatin minigene, comprising exon 8/intron 8/exon 9 of HER2 was generated and subsequent in vitro splicing assays revealed that RNA secondary structure and somatic mutations did not impact on inclusion of intron 8. However, using RNase-assisted RNA chromatography, followed by mass spectrometry, we identified six RNA-binding proteins (splicing factors) that bind to RNA sequences surrounding exon 8/intron 8 and intron 8/exon 9 boundaries; these included hnRNP I, H1, D, A2/B1 and hnRNPA1 plus the SR protein SRSF1. Specifically, overexpression of hnRNP A1 significantly increased retention of intron 8 resulting in higher levels of Herstatin in SKBR3 breast cancer cells whereas SRSF1 only had a marginal effect in decreasing Herstatin but increased exogenous HER2 levels under these experimental conditions. In conclusion, we have identified the first splicing factors and regulatory sequences that are involved in the production of Herstatin.
Breast cancer is the most common cancer affecting females and has highest cancer-related mortality.1 The Human epidermal growth factor receptor 2 (HER2), also referred to as receptor tyrosine-protein kinase 2 (ERBB2), is overexpressed in up 30% of breast tumors (HER2+) and this is associated with poor prognosis, metastasis and reduced survival.2,3 HER2 is a tyrosine kinase receptor involved in cell proliferation, apoptosis regulation and differentiation.2-4 The full length HER2 protein shares a common structure with other HER family members which include HER1, HER3 and HER4. HER proteins comprise of three main domains: extracellular, transmembrane and intracellular (with tyrosine kinase activity).2,5 The extracellular domain includes two ligand-binding domains (LI and LII) and two cysteine-rich domains (CI and CII) which are required for establishing homo/hetero-dimers (Fig. 1A).5 Upon dimerization, the tyrosine kinase domains of both HER2 and its dimerization partner are activated; this trans-phosphorylates both monomers (Fig. 1A) which in turn mainly induces the oncogenic RAS/MAPK and PI3/AKT signaling pathways.6,7 Because of its unique conformation, no EGF ligand can directly bind and activate HER2.4,6 However this receptor is the preferred dimerization partner over other HER family members.8 HER2-containing dimers display delayed endocytosis as well as having a higher affinity for the ligand resulting in stronger signaling transduction.2,4
The humanized monoclonal antibody Trastuzumab (Herceptin®) and the tyrosine kinase inhibitor Lapatinib (Tyverb®) are clinically approved HER2-targeted therapies.9,10 It is important to note that a significant proportion of HER2+ breast cancer patients develop resistance to these therapies.11 The alternative splice variant HER2Δ16, arising from skipping of the exon 16 of HER2 gene, has been associated with resistance toward anti-HER2 therapies, for instance, via activation of other oncogenic pathways including Src kinase. Moreover, compared with the common HER2 protein, HER2Δ16 has been linked with more aggressive and advanced tumors.12,13 Conversely, the splice variant Herstatin has been investigated for its tumor suppressor potential and considered as alternative therapeutic options for treating diseases associated with HER1 and HER2 receptors.14,15 Herstatin arises from inclusion of intron 8 of HER2, which introduces a premature stop codon. The resulting protein is a 68kDa secreted form of HER2 which comprises the extracellular LI and CI domains and a novel 79 C-terminus.16 Herstatin can interact with the extracellular domain CI of HER2 blocking its transition toward the cell membrane as well as inhibiting HER2 dimers formation (Fig. 1B) and receptor phosphorylation.15,17,18 Thus Herstatin functions as endogenous inhibitor of HER2. Different studies have assessed Herstatin expression in breast cancer cell lines as well as heterogeneous tumor and normal breast tissues.16,19 To date, the mechanisms that switch splice site selection to generate Herstatin have not been described. In the present study we designed and utilized a HER2 minigene for in vitro splicing assays to investigate HER2 RNA secondary structure, somatic mutations as well as regulatory sequences which regulate the alternative splicing of Herstatin. Moreover, splicing factors that may affect Herstatin expression were identified; the role of hnRNP A1 and SRSF1 were confirmed through their impact on splicing of the HER2 minigene in SKBR3 breast cancer cells.
The HER2+/ER- breast cancer cell line SKBR3 (ATCC) was cultured at 37C with 5% CO2 using McCoy's 5a GlutaMAX medium (GIBCO by Life Technologies) supplemented with 10% Foetal Calf Serum (Sigma-Aldrich) and 1U/ml Penicillin/Streptomycin (Sigma-Aldrich).
The HER2 minigene comprises the last 300bp of intron 7, full exon 8 (120bp), full intron 8 (274bp), full exon 9 (127bp) and the first 300bp of intron 9 of the HER2 gene; this sequence was cloned into a pcDNA 3.1(-) vector (Invitrogen) using XhoI and EcoRI restriction sites (Fig. S1; for sequence details see Table S1). Site-directed mutagenesis as well as deletion D1 were applied to the minigene using a two-step PCR-based protocol as previously described (for sequence details see Table S2 and 3).20 An additional expression plasmid carrying deletion D2 (Fig. S3; for sequence details see Table S3) was generated and purchased form 96 Proteins (San Francisco, USA).
SKBR3 breast cancer cells were transfected or co-transfected at 70–80% confluency using 0.5 µg of plasmid DNA with 1.5 µl of FuGENE® HD (Promega). In co-transfections the mini genes were also transfected with expression plasmids to overexpress hnRNPA1 (pCG-A1) or SRSF1 (pCG-SF2) or pCG (no insert) control plasmid (all plasmids were kindly provided by Professor Javier Caceres). RNA was isolated after 24 hours post transfection.
RNA was isolated using the ReliaPrep™ RNA Cell Miniprep System (Promega). Genomic and plasmid DNA degraded with 2U of TURBO™ DNase (Ambion); RNA was then purified by phenol-chloroform extraction. Reverse transcription was performed using 500ng of RNA and Superscript II Reverse Transcriptase (Life Technologies). No reverse transcriptase controls (-RT) were included in all experiments (Fig. S5). PCR was performed using PCR Master Mix (Promega). For exclusively detecting HER2 minigene transcripts, reverse transcription and PCR were performed employing reverse primers, MGR2 and MGR1 respectively, which bind the pcDNA 3.1(-) vector, downstream from the HER2 minigene. PCR products were analyzed by standard agarose gel electrophoresis and by high-resolution capillary gel electrophoresis (QIAxcel).
For the RNase-assisted chromatography, RNA oligonucleotides were synthesized via T7 Polymerase (NEB) in vitro transcription using DNA oligonucleotides (Invitrogen) as template. DNA oligonucleotides, carrying a T7 promoter sequence, and a T7 promoter oligonucleotide were diluted in Annealing Buffer (10mM Tris, pH 7.5–8.0, 50mM NaCl, 1mM EDTA) and hybridized by heating to 95°C and cooling slowly to room temperature. SKBR3 nuclear extract was prepared as previously described.21 RNase-assisted chromatography was based on the RNase-assisted RNA chromatography by Michlewski and Caceres22 and performed as previously described.21 A negative control (CT), in absence of RNA template, was also included to ensure accuracy of the assay. RNA-binding proteins were separated on a NuPage pre-cast gel (Invitrogen, Life Technologies) and visualised with SimplyBlue Safestain (Invitrogen, Life Technologies). Individual proteins were identified by MALDI-TOF mass spectrometry (Northumbria University) or by immunoblotting.
RIPA Buffer (25mM Tris-HCl pH7.6, 0.1% SDS, 150mM NaCl, 1% NP-40, 0.1% sodium deoxycholate, 20µl/ml protease inhibitor cocktail) was used for cell lysis and protein extraction. 1X Laemmli buffer (125mM Tris-HCl pH6.8, 2% SDS, 10% glycerol, 10% β mercapthoethanol, 0.1% bromophenol blue) was added to each samples before loading to 12% SDS-PAGE. Proteins were transferred to Immun-blot® PDVF (BioRad) followed blocking with 1X PBS+10% milk for 1 hour and overnight incubation with primary antibody (Table S4) diluted in 1X PBS + 1% milk.
In silico software SpliceAid,23 available at http://www.introni.it/splicing.html, was used to predict binding sites for RNA-binding proteins (splicing factors) (Fig. S2). The prediction of RNA secondary structures was performed using RNAstructure software from University of Rochester Medical Center (available at http://rna.urmc.rochester.edu/RNAstructureWeb/Information/Contributors.html).
The HER2 minigene comprised of the sequence of HER2 including the exon 8/intron 8 and intron 8/ exon 9 boundaries (Fig. S1). Splice variants generated from HER2 minigene expression were detected employing two reverse primers, MGR2 or MGR1 respectively, priming the plasmid vector downstream of the HER2 minigene. MGR1 was used in combination with a forward primer (Exon 8 Fw) which primed the exon 8 of HER2. Three amplicons, resulting from the HER2 minigene expression, were detected (Fig. 2A): a 710bp amplicon included exon 8, intron 8, exon 9 and the beginning of intron 9 of HER2 minigene; this transcript is referred as exogenous Herstatin (eHST); a 436bp amplicon comprising exon 8, exon 9 and the beginning of intron 9 is referred as exogenous HER2 (eHER2); a 232bp amplicon included exon 8, exon 9 and a shorter portion of intron 9; this variant resulted from the selection of an alternative 3′splice site within the intron 9 of the minigene and which is not commonly recognized in the endogenous HER2. The latter amplicon is referred as ‘short’ exogenous HER2 (sh-eHER2). Subsequently, in order to assess the HER2 minigene splicing efficiency, the 5′splice site of intron 8 was mutated; the AG/GT motif at the exon 8/ intron 8 boundary was replaced with CG/GG (Fig. 2B and Fig. S2). The AG/GT>CG/GG mutation abolished intron 8 splicing leading to exclusive production of exogenous Herstatin and ‘short’ exogenous Herstatin (sh-eHST) which arises from the selection of the alternative 3′splice site within intron 9 (Fig. 2B). These results indicate that the HER2 minigene consistently reproduced the alternative splicing of intron 8 of HER2, therefore, it is a good model to study the alternative splicing of Herstatin.
In silico analysis using the RNAstructure software (Fig. S3A), predicted that the AG/GT>CG/GG mutation at the 5′splice site of intron 8, which previously abolished the splicing of that intron, may affect the RNA secondary structure of the exon 8/intron 8 boundary of HER2 (Fig. S3B). More precisely, this mutation might influence the size of the hairpin characterizing the beginning of intron 8 (Fig. S3B).
To assess whether the RNA secondary structure of the HER2 exon 8/intron 8 boundary play a role in regulating Herstatin, a C>T mutation was generated downstream of the 5′splice site of the intron 8 (Fig. S3C). This mutation was predicted to exclusively effect the hairpin size, while preserving the surrounding RNA structure. SKBR3 cells were then individually transfected with HER2 minigene (wild type) or minigene carrying the C>T mutation and their splicing pattern. Results indicate that the C>T mutation does not affect the alternative splicing of the HER2 minigene (Fig. S3D).
Four somatic mutations, lying within exon 8 of HER2, have been reported in patients with breast cancer or glioblastoma and these have been shown to affect HER2 phosphorylation, cell proliferation and response to drugs in vitro,24,25 They are single nucleotide substitutions: 926 G>C, 926 G>A, 929 C>T and 1000 T>A.
In order to investigate the role of these somatic mutations in the splicing of intron 8 of HER2, four HER2 minigenes, each carrying one mutation, were generated (Fig. 3). SKBR3 cells were transfected with individual minigenes and the splicing pattern compared with the wild type HER2 minigene. Results indicate that none of the investigated somatic mutations affected the alternative splicing of the intron 8 (Fig. 3).
To predict splicing factors which may regulate the alternative splicing of Herstatin, in silico analyses were performed which predicted binding sites for SR proteins at ~60bp, upstream and downstream, from the 5′splice site of intron 8; whereas the splicing inhibitors hnRNPs, ETR-3 and YB-1, were predicted to bind at ~100bp upstream of the 3′splice site of the intron (Fig. S2).
Two 100bp RNA oligonucleotides (RNAnts) were designed for the exon 8/intron 8 boundaries (RNAnt-H1 and H2) and two 100bp RNAnts for the intron 8/exon 9 boundaries (RNAnt-H3 and H4) (Fig. 5A and Table S3). T RNAnts were then used in RNase-assisted RNA chromatography to identify RNA-binding proteins/splicing factors that bind to these specific regions of HER2. Mass spectrometry identified five hnRNP proteins including hnRNP I (or PTB), hnRNP H, hnRNP D, hnRNP A2/B1and hnRNP A1 (Fig. 4B and Table S5). To validate these results, RNA-chromatography was repeated with subsequent immunoblotting experiments which confirmed binding of hnRNP H and hnRNP A2/B1 exclusively to the RNAnt-H1; hnRNP D bound to all four RNAnts although to a lesser degree RNAnts H2 and H3. Results also show strong binding for hnRNP A1 to all four RNAnts-H1-H4. Note the doublet represents the two splice variants hnRNP A1B (upper band) and hnRNP A1 (lower band). hnRNP I binding exclusively to the RNAnt-H3 (Fig. 4C). Additionally, SRSF1 was identified binding exclusively the RNAnt-H1 and H4 (Fig. 4C).
RNA-chromatography identified hnRNPA1 as a potential candidate regulator. In order to test the role of hnRNPA1 SKBRS cells were simultaneously transfected with the HER2 minigene and pCG-A1 to overexpress hnRNPA1 together with an empty pCG control plasmid. The effect of over-expressing hnRNPA1 in SKBR3 cells influenced inclusion of intron 8 resulting in a significant increase in levels of Herstatin at the expense of exogenous and short HER2 variants (Fig. 5 A-B).
According to the results obtained with RNA-chromatography, hnRNP A1 and SRSF1 can bind to the same HER2 sequence covered by the RNAnt-H1 (Fig. S2). Since the antagonising roles between these two splicing factors is well known.26-28 SRSF1 was also overexpressed in SKBR3 cells to assess its role on the splicing of the HER2 minigene. Results show that overexpression of SRSF1 in SKBR3 cells does appear to increase splicing of the intron 8 of HER2 minigene by increasing exogenous HER2 mRNA levels at the expense of Herstatin, although this effect is not statistically significant (Fig. 5 C-D).
To assess whether potential SRSF1 binding sites affected the splicing of intron 8 of HER2, two deletions, D1 and D2, were generated on HER2 minigene. D1 targeted 17 nucleotides right upstream the 5′splce site of the intron 8; this sequence was covered by the oligo RNAnt-H1, at which SRSF1 bound during RNA-chromatography (Fig. 6). Deletion D2 targeted 13 nucleotides, comprising a GGAGG motif, located within the intron 8 at ~50 nucleotides upstream the predicted branch point site and polypyrimidine tract (Fig. 6).
Deletions D1 and D2 were individually generated on distinct HER2 minigenes and, additionally, a simultaneous D1+D2 deletion was performed in-order to assess their potential combinatory effect on the splicing of intron 8. Subsequently, SKBR3 cells were transfected with HER2 minigene wild type or with minigenes carrying deletions D1 or D2, or D1+D2. Compared to the HER2 minigene wild type, deletions D1 and D2, individually, as well as D1+D2 had little effect on the production of exogenous Herstatin (Fig. 6).
Alterations of the alternative splicing of proto-oncogenes are not uncommon in breast cancer and they can result from deregulation in the expression of splicing factors as well as mutations of their binding sites.29,30 Changes in the alternative splicing of the HER2 oncogene can promote the progression of breast cancer.31,32 Conversely, the tumor suppressor activity of the HER2 splice variant Herstatin has been previously described.15,31
We have recently identified splicing factors that regulate a splicing hot-spot around exons 15-17 of the HER2 gene,21 however, to date the mechanisms that regulate expression Herstatin remain still unknown. Factors that could potentially include mutations, or SNPs and changes in Herstatin expression may derive from deregulation of splicing factors which either promote or prevent the splicing of intron 8.
This is the first study which focuses on the regulation of the alternative splicing of Herstatin in breast cancer. Using the HER2 minigene, mutation of the 5′splice site of intron 8 of the minigene, abolished splicing of that intron as well as modified the secondary structure of exon 8/intron 8 boundary. However, subsequent experiments, have excluded any role of RNA secondary structures on Herstatin regulation. Previous studies have described four distinct somatic mutations within the exon 8 of the HER2 gene in patients with breast or glioblastoma tumors; using the HER2 minigene, none of these mutations showed to affect the splicing of the intron 8.
RNA-chromatography, followed by mass spectrometry, identified hnRNPs A1, A2/B1, H, I and D, as well as the SR protein SRSF1, as potential regulators of the alternative splicing of the intron 8 of HER2. Among these, the role of hnRNP A1 on Herstatin expression was confirmed after observing its effect on the HER2 minigene splicing pattern post overexpression in SKBR3 breast cancer cells. Indeed, hnRNP A1 overexpression promoted inclusion of the intron 8 of the minigene while overexpression of SRSF1 promoted its splicing and increased levels of endogenous WT HER2. These results are supported by previous studies which describe the antagonising roles between hnRNPs A1 and SRSF1.26,28,33 Furthermore, SRSF1 has previously been classified as a proto-oncogene,34 which is consistent with its role in favoring expression of the oncogenic HER2.
Using RNA-chromatography, SRSF1 mainly bound to the exon 8/intron 8 boundary of HER2, which was covered by the oligo RNAnt H1. SRSF1 commonly induces splicing by interacting with degenerate purine-rich RNA sequences, including GGAGG and AAGA motifs.35-37 SRSF1 can recruit snRNP U1 to the 5′ splice site as well as induce the recognition of branch point sites.38,39,40 Using the HER2 minigene, deletions D1, targeting a potential SRSF1 binding motif adjacent the 5′splice site of the intron 8, but had little effect on exogenous Herstatin or HER2 levels. In the same way, deletion D2, affecting a GGAGG motif proximal to the predicted branch point site of the intron 8, also didn't significantly affect exogenous Herstatin or HER2 levels nor did the simultaneous deletions D1+D2. It is also possible that other SRSF1-binding sites within the HER2 pre-mRNA transcript exist but have yet to be identified.
In conclusion, here we speculate a potential working model, partially explaining the regulation of the alternative splicing of the intron 8 of HER2 in SKBR3 breast cancer cells. In the most common scenario, SRSF1 can bind exonic splicing enhancer (ESE), adjacent the 5′splice site of the intron 8 of HER2 pre-mRNA; this would recruit the snRNP U1 to this splice site. Additionally, SRSF1 can also interact with the GGAGG motif proximal to the branch point site of the intron 8 which could reinforce recognition of its 3′splice site of that intron. As result, splicing of the intron 8 is promoted while preventing the formation of Herstatin.
In a less common scenario hnRNP A1 could bind GAGG motifs upstream and downstream the 5′splice site of the intron 8 of HER2 pre-mRNA. In doing so, hnRNP A1 could counteract the role of SRSF1 by weakening the recognition of the 5′ and the 3′ splice sites of intron 8. As a consequence, this would promote Herstatin formation at the expense of oncogenic HER2.
In summary, the present study has now started to unravel how the protective and non-oncogenic splice variant of HER2, Herstatin, is regulated in breast cancer cells. Future studies are now underway to assess the coexistence/ratio of Herstatin versus full length HER2 and the nasty highly oncogenic HER2Δ16 splice variant in a large cohort of breast cancer patients. The long-term goal of this research program is to provide insights into HER2 biology that could pave the way to the development of individualised therapeutic strategies for treating patients who are classified as HER2 positive.
No potential conflicts of interest were disclosed.
We would like to acknowledge Prof Gary Black and Dr Andrew Porter (Northumbria University) for performing the Mass spectrometry experiments.
This project was funded by the Royal Victoria Infirmary Breast Cancer Appeal and supported by the Breast Cancer Now Tissue Bank.