Background
Triple-negative breast cancer (TNBC) is a notoriously aggressive, heterogeneous disease defined as lacking expression of oestrogen receptor (ER) and/or progesterone receptor (PR) as well as amplification of human epidermal growth factor receptor 2 (HER2), respectively [
1]. Although TNBC only constitutes approximately 15–20% of breast cancer cases, it is disproportionately responsible for breast cancer-associated deaths and carries a dismal prognosis, compared with hormone receptor-positive (HR+) breast cancers [
2‐
4]. For patients with HR+ breast cancer, endocrine therapy targeting ER is available in the form of aromatase inhibitors and selective oestrogen receptor modulators (e.g. tamoxifen) and other antagonists [
1]. Contrastingly, no effective targeted therapy which exploits the molecular properties of tumour cells exists for TNBC patients; clinical trials of targeted agents in TNBC have been disappointing [
5,
6]. Consequently, aggressive chemotherapy, radiotherapy and surgery remain the mainstay treatments [
7]. Furthermore, patients who develop resistance to treatment, or who do not respond to treatment whatsoever, follow an aggressive clinical course characterised by metastasis and a higher 5-year mortality post-diagnosis [
8]. Further complicating the treatment of TNBC is the degree of genetic heterogeneity observed in this disease. By analysing the gene expression profiles of TNBC cases, Lehmann et al. sub-classified TNBC into six different molecular subtypes: mesenchymal (M), mesenchymal stem-like (MSL), luminal androgen receptor-positive (LAR), immunomodulatory (IM), basal-like 1 (BL1) and basal-like 2 (BL2) [
9]. Most importantly, these subtypes exhibit dissimilar drug-sensitivity profiles, resulting in varied clinical responses [
9,
10]. The nature of TNBC clearly necessitates a more tailored approach to treatment, one which exploits the unique oncogenic addictions present. For chemotherapy-resistant TNBC patients, the development of targeted therapeutics which synergise with current treatment options to overcome resistance is therefore paramount.
Epidermal growth factor receptor (EGFR; also known as ERBB1/HER1) is often expressed at higher levels in triple-negative tumours than in HR+ tumours [
11], though expression levels vary, with up to ~ 80% of TNBC cases reported as being EGFR+ [
12]. EGFR amplification has also been reported to occur in a substantial proportion of TNBC cases [
13‐
15] with EGFR overexpression associated with a much poorer prognosis in general [
8]. Furthermore, EGF signalling is highly enriched in the basal and mesenchymal TNBC subtypes [
9]. EGFR therefore represents a bona fide drug target in triple-negative tumours. Various EGFR inhibitors have been developed, most notably anti-EGFR monoclonal antibodies (e.g. cetuximab) and EGFR-tyrosine kinase inhibitors (EGFR-TKIs) (e.g. erlotinib, gefitinib and lapatinib) [
16]. Despite these efforts, EGFR-TKI single treatment has performed poorly in clinical trials for TNBC patients with advanced or metastatic breast cancer, despite clear inhibition of EGFR [
17,
18] suggesting bypass inhibition of EGFR-related signalling in TNBC tumours [
19,
20]. Nonetheless, EGFR-TKIs have shown more promising results as combination therapies in TNBC [
21], perhaps indicating that monotherapeutic inhibition of EGFR is insufficient to shut down the myriad signalling pathways responsible for promoting aberrant proliferation and survival.
Here we demonstrate that multiple EGFR-TKIs synergise with the dual cdc7/CDK9 inhibitor PHA-767491 in a panel of TNBC cell lines resistant to EGFR-TKIs. This combination inhibited cell proliferation, induced apoptosis and G2-M arrest and downregulated components critical to cell cycle progression, DNA replication and transcription, thereby reversing resistance to EGFR-TKIs. Therefore, targeting deficiencies in regulation of the cell cycle and DNA replication in conjunction with transcriptional addiction downstream of growth factor pathways may constitute a powerful therapeutic opportunity for this difficult-to-treat breast cancer subtype.
Methods
Cell culture
All cell lines were maintained in RPMI-1640 medium (Gibco, ThermoFisher Scientific, Breda, The Netherlands) supplemented with 10% FBS (Thermo Fisher Scientific; 10270106) and 25 IU/ml penicillin and 25 μg/ml streptomycin (ThermoFisher Scientific; 15070-063). Cells were cultured in a humidified incubator at 37 °C, 5% CO2. Cell lines were provided by Erasmus MC Rotterdam and tested monthly for mycoplasma using PCR.
Antibodies and kinase inhibitors
The primary antibodies against pEGFR (Y1173, #4407), pERK1/2 (T202/Y204, #9101), ERK1/2 (#4695), pAKT (S473, #9271), AKT (#9272), MCM2 (#3619), pRNA-II (S2/5; #4735), RNA-II (#2629), and p-pRb (S780, # 9307) were commercially supplied from Cell Signaling TECHNOLOGY®, EGFR (sc-03), pRb (sc-102), CDK4 (sc-601), Cyclin D1 (sc-20,044) from Santa Cruz BIOTECHNOLOGY®, cdc7 (ab10535) and pMCM2 (S40/41; ab70371) from Abcam® and Tubulin (T-9026) from Sigma®. Secondary antibodies included Cy5-conjugated anti-mouse and horseradish peroxidase (HRP) anti-mouse or anti-rabbit (Jackson ImmunoResearch). Individual kinase inhibitors lapatinib (S2111), gefitinib (S1025), erlotinib (S7786) and PHA-767491 (S2742), plus the previously described 273-kinase inhibitor library (L1200), were purchased from Selleckchem® (Munich, Germany) and dissolved in DMSO solution at 10 mM [
22]. TAK-931 and BAY-1143572 were purchased from MedChemExpress (Sollentuna, Sweden).
Kinase inhibitor treatment and drug combination screen
Cells were seeded into 96-well plates at the appropriate densities (Additional file
1: Table S1). The following day, cells were treated with individual kinase inhibitors in dose range as indicated. Vehicle DMSO (1:1000) was used as control. For the kinase inhibitor library screen, cells were screened in duplicate against the kinase inhibitor library containing 273 kinase inhibitors at concentration of 1 μM alone, or the 1 μM library inhibitors in combination with lapatinib at 3.16 μM, since this concentration effectively inhibited EGFR phosphorylation in all cell lines tested and since studies have shown that levels of lapatinib in patient tumours vary between 1 and 12 μM depending on dosing schedule [
23]. After 4-day treatment, proliferation was evaluated by sulphorhodamine B (SRB) colorimetric assay [
24] and analysed by % of control cell growth = (mean sample OD − mean 0-day OD)/(mean control OD − mean 0-day OD) × 100. To assess synergistic interaction of combined drugs, combination index (CI) analysis [
25,
26] was performed, using the formula ‘CI = (D)1/(Dx)1 + (D)2/(Dx)2’. (D)1 and (D)2 are respective combination doses of two compounds that yield an effect of 50% of proliferation inhibition, with (Dx)1 and (Dx)2 being the corresponding single doses for either compound that results in the same effect, which is by definition the IC50 of each compound. CI values less than 1 (CI < 1), equal to 1 (CI = 1) or greater than 1 (CI > 1), indicate synergy, additivity or antagonism, respectively.
Western blotting
Cells were seeded in 6-well plates at the appropriate density. For stimulation/starvation assays, medium was refreshed with serum-free medium (SFM) the following day and cells were starved overnight. Thereafter, cells were pre-treated with drug solutions for 4 h, then stimulated with 100 ng/ml EGF (Sigma; E9644) for 5 min in SFM. For time-course exposures to drugs, cells were treated with drug solutions prepared in complete medium. Cell lysates were harvested at the indicated time points in RIPA lysis buffer with 1:100 Protease Inhibitor Cocktail (Sigma; P8340). Cellular proteins were denatured in sample buffer containing 10% β-mercaptoethanol, loaded with 30 μg/lane into 7.5% polyacrylamide gels, resolved using SDS-PAGE and subsequently transferred to PVDF membranes (Merck Chemicals; IPVH00010) overnight. PVDF membranes were then blocked with 5% BSA-TBST (Tris-buffered saline 0.05% Tween-20) and subsequently incubated at 4 °C overnight with appropriate primary antibodies. The following day, membranes were incubated for 1 h with HRP- or Cy5-conjugated secondary antibodies and chemiluminescence or fluorescence was detected using the Las4000 (GE Healthcare).
Annexin-V staining
Cells were seeded overnight in 96-well μCLEAR plates (Corning) at appropriate densities, then treated with drug solutions at indicated concentrations. At 24, 48 or 72 h post-treatment, cells were stained with Hoechst 33258 (1:10,000) and Annexin-V (1:1000) for 45 min at 37 °C, 5% CO2 before being imaged using BD Pathway 855 Microscope (BD Biosciences). Annexin-V staining was quantified using Cell Profiler software.
Cell cycle flow cytometry analysis
Cells were seeded in 6-well plates at the appropriate density. Twenty-four or 48 h post-treatment, all cells were harvested, re-suspended in ice-cold 200 μl 1 mM EDTA-PBS and 800 μl 100% ethanol, and stored at − 20 °C before being centrifuged at 1000 rpm at 4 °C. Cells were then re-suspended in 1 ml PBS and rehydrated for 15 min. After being spun at 1000 rpm for 5 min at room temperature, the pellet was re-suspended in 250 μl 3 mM DAPI (Sigma, 10236276001) staining buffer (100 μM Tris pH 7.4, 150 mM NaCl, 1 mM CaCl2, 0.5 mM MgCl2), incubated for 15 min at room temperature in the dark, followed by filtration through 70-μm EASYstrainer filters and analysed using FACS Conto II (BD Biosciences). Data were analysed using FlowJo V10.
siRNA transfection
Cells were seeded in 96-well plates at the appropriate density. For each siRNA transfection, 50 nM siGENOME siRNAs (Dharmacon) were transfected into cells per 96-well using INTERFERin transfection reagent (Polyplus; 409-50). The following day, the medium was refreshed. Forty-eight hours post-transfection, cells were either lysed for western blot to confirm knockdown or treated with drugs for the appropriate duration as described, then fixed for SRB proliferation assay.
Clinical evaluation of candidate target genes
The clinical relevance of cdc7, POLR2A and CDK9 was evaluated using in-house gene expression and metastasis-free survival data of 123 lymph node-negative, non-(neo) adjuvantly treated, oestrogen receptor-negative (ER-neg) primary breast cancer patients. The composition of this cohort is described in Additional file
2: Table S2. The clinical relevance of synergy-related candidate genes was evaluated using the previously described in-house as well as publicly available gene expression and MFS data of lymph node-negative, non-(neo) adjuvantly treated primary breast cancer patients, leading to a cohort of 142 triple-negative patients. Data were gathered from Gene Expression Omnibus (
http://www.ncbi.nlm.nih.gov/geo/) entries GSE2034, GSE5327, GSE2990, GE7390 and GSE11121, with all data available on Affymetrix U133A chip. Raw.cel files were processed using fRMA parameters (median polish) [
27] after which batch effects were corrected using ComBat [
28].
Transcriptome RNA sequencing and pathway integration analysis
Cells were seeded overnight in 6-well plates and treated in triplicate for 6 h with individual or combined kinase inhibitors at indicated concentrations, or vehicle. RNA was isolated with RNeasy Plus Mini Kit as described by the manufacturer (QIAGEN, Cat. 74136). Transcriptome RNA sequencing (RNA-Seq) was performed using Illumina high-throughput RNA sequencing. DNA libraries were prepared from the samples with the TruSeq Stranded mRNA Library Prep Kit. The DNA libraries were sequenced according to the Illumina TruSeq v3 protocol on an Illumina HiSeq2500 sequencer. Paired-end reads of 100bp in length were generated. Alignment was performed against the human GRCh38 reference genome using the STAR aligner (version 2.4.2a). Marking duplicates, sorting and indexing were performed using sambamba. Gene expression was quantified using the FeatureCounts software (version 1.4.6) based on the ENSEMBL gene annotation for GRCH38 (release 84). RNA-Seq data was normalised by TMM using EdgeR’s normalisation factor [
29], followed by quantile normalisation and presented in Log2 fold change (Log2 FC) scales. Genes with significant down- or upregulation (Log2 FC ≥ |0.5|) under indicated conditions were analysed by web-based functional analysis tool Ingenuity pathway Analysis (IPA) to visualise and annotate their biological functions and pathways.
Statistical analyses
All statistical analyses, where appropriate, were performed in GraphPad Prism software version 7.0. One-way ANOVA multiple comparison test with Tukey’s post hoc test was applied with p values less than 0.05 considered as statistically significant.
Discussion
EGFR is highly expressed in both TNBC tumours and cell lines, supporting a role for EGFR as an oncogenic driver in TNBC. However, clinical trials suggest single inhibition of EGFR signalling is incapable of eliminating TNBC cells [
17,
18,
21,
31]. Consistently, our results demonstrated that targeting EGFR kinase activity by EGFR-TKIs, including lapatinib, erlotinib and gefitinib, insufficiently inhibits TNBC cell proliferation, despite inhibition of EGFR phosphorylation. Our kinase inhibitor combination screen demonstrated that the dual cdc7/CDK9 inhibitor PHA-767491 enables EGFR-TKIs to inhibit proliferation, induce G2-M cell cycle arrest and promote apoptosis in various TNBC cell lines expressing high levels of EGFR. This synergistic drug interaction downregulates the activity of components of the transcription apparatus and the DNA replication programme, including cdc7, CDK9, pMCM2 (S40/41), p-RNAII (S2/5), CDK4, cyclin D1 and Rb, making the combination of EGFR and cdc7/CDK9 molecular-targeted therapies promising for this subgroup of breast cancer.
CDK9 is a member of positive elongation factor P-TEFb and together with CDK7 is vital for gene transcription since CDK7 and CDK9 sequentially phosphorylate the C-terminal domain (CTD) of RNA Polymerase II (RNA II) at Ser5/Ser7 and Ser2, respectively, allowing dissociation of negative elongation factors and subsequent elongation of mRNA transcripts [
32‐
34]. Blockage of this critical elongation step results in stalled transcription which triggers ubiquitination of RNAII at active gene promoters and its subsequent proteasome-mediated degradation [
35]. Given that co-treatment with higher concentrations of PHA-767491 reduced total levels of RNAII, investigating whether CDK9 inhibition-mediated transcriptional stalling is responsible for proteasome-dependent depletion of this protein is prudent. Previous studies have demonstrated the potential of inhibiting CDK9 in in vitro and in vivo PDX models of TNBC using pan-CDK inhibitor dinaciclib, leading to G2/M cell cycle arrest and apoptosis, consistent with the results presented herein [
36]. CDK9 is essential for the growth of both HR+ and TNBC cell lines, whilst EGFR is one of many “Achilles’cluster” genes sensitive to CDK7 inhibition and vital for TNBC survival [
37]. Consistently, we showed that despite being resistant to inhibition of EGFR kinase activity by various EGFR-TKIs, complete silencing of EGFR is detrimental to TNBC cell growth. In addition, EGFR is capable of acting as a transcription factor [
38,
39]. The nuclear translocation of EGFR is associated with resistance to chemotherapeutics in TNBC [
38,
40‐
42] and shields the RTK from the effects of TKIs limited to the cell membrane, permitting EGFR to enhance transcription of genes which govern cell cycle progression, such as Cyclin D1 and Aurora Kinase [
39,
43]. Cdc7 kinase is itself indispensable for correct regulation of cell cycle progression, exerting control over both initiation of DNA replication and the DNA damage response [
44,
45]. By phosphorylating mini-chromosome maintenance proteins (MCM2-7) present in pre-replicative complexes formed during G1 phase, cdc7 activates the helicase activity of these proteins, leading to unwinding of DNA strands and thereby initiating DNA replication at the G1-S phase checkpoint [
46,
47]. It also has been reported that EGFR indirectly influences the initiation of DNA replication by eliciting phosphorylation of MCM7 in a Lyn kinase-dependent fashion, thereby delineating possible functional overlap between cdc7 and EGFR [
48]. TNBC cells often possess p53-inactivating mutations which abolish the DNA replication origin activation checkpoint, rendering them susceptible to the induction of replicative stress [
49]. Induction of G2-M arrest in our TNBC cell lines after combined inhibition of EGFR and cdc7/CDK9 is consistent with data from other studies which demonstrated that a p53-dependent checkpoint is critical for mitigating aberrant cell cycle progression after cdc7 depletion [
50,
51].
The CDK4/Cyclin D1 complex phosphorylates Rb at Ser780/795 thereby inactivating Rb in G1/S checkpoint regulation [
52,
53]. Consistently, co-inhibition of cdc7/CDK9 and EGFR signalling in our TNBC cell lines reduces CDK4 and Cyclin D1 levels accompanied by reduced phosphorylation of Rb, thereby resulting in G2-M arrest and ultimately apoptosis. Whether the downregulation of Cyclin D1 and CDK4 by EGFR-TKIs and PHA-767491 represents a global decrease in the transcription of rapidly expressed, immediate response genes due to inhibition of CDK9-mediated transcriptional elongation, a decrease in the transcriptional activity of EGFR, or a by-product of the cell cycle arrest induced by cdc7 depletion, merits further investigation. Taken together, these results identify possible functional links between signalling downstream of EGFR and the function of both cdc7 and CDK9, which may to some extent explain the observed synergy between EGFR-TKIs and PHA-767491. Additionally, silencing of cell cycle-regulatory or transcriptional CDKs in combination with a cdc7-specific inhibitor (XL413) in breast cancer cells has been shown to mimic the cell cycle disruption caused by PHA-767491 [
54]. Silencing of CDK9 led to negligible impact on progression of MCF10A cells through S-phase, whilst CDK9-depleted cells treated with XL413 accumulated in late S-phase, suggesting that the profound cell cycle arrest in TNBC cells caused by PHA-767491 or cdc7 depletion may be somewhat dependent on CDK9 or can at least be augmented by inhibiting CDK9. Nonetheless, using RNAi-mediated silencing of cdc7 and CDK9, we were unable to fully recapitulate the observed synergy between EGFR-TKIs and PHA-767491 in selected TNBC cell lines. Although PHA-767491 has off-target effects on CDK1, CDK2 and GSK-3β which could contribute to sensitisation of TNBC cells to EGFR-TKIs, also knockdown of these genes had limited effects on the response to lapatinib. Despite that triple combination of the two other selective cdc7 and CDK9 inhibitors together with lapatinib strongly affected proliferation in the TNBC cell lines, this effect was not as pronounced as the combination of lapatinib with PHA-767491. This suggests that besides cdc7/CDK9 blockage by PHA-767491, also inhibition of other kinases likely contributes to the observed synergy. Broad spectrum kinase inhibition is not uncommon for highly effective anticancer therapeutics used in the clinic. Here we have only tested the PHA-767491/Lap combination in TNBC cell lines. Given the broader anti-kinase activity of PHA-767491 and the side effects of lapatinib and other EGFR inhibitors on the liver and/or heart, further assessment of the safety of such a combination treatment will be essential.
RNA-Seq transcriptomics identified genes specifically downregulated by co-treatment with EGFR-TKIs and PHA-767491, which were involved in pathways regulating survival, transcription and cell cycle progression. The decreased expression of these genes (a number of them associated with poorer MFS in TNBC) by PHA-767491 combined with inhibition of EGFR leads to apoptosis and downregulation of transcription and proliferation. Interestingly, the transcription factor MITF (microphthalamia-associated transcription factor), a major upstream regulator of pathways governing apoptosis, proliferation and transcription, was decreased together with its downstream targets pro-survival BCL-2 and cell cycle-regulatory CDC25B as a result of combining lapatinib and PHA-767491. With regards to TNBC, little is known about MITF’s contribution to EGFRi-resistant phenotypes. Further research is therefore required to validate whether targeting of MITF function constitutes a logical therapeutic avenue in TNBC, or whether MITF inhibition is sufficient to reverse the resistance of TNBC cells to EGFR-targeted therapies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.