Background
Malignant mesotheliomas (MMs), aggressive tumors characterized by marked local invasiveness, are poorly responsive to current therapeutic approaches. Clinical outcomes for MM are poor, resulting in average patient survival times of 7 to 12 months from initial diagnosis. We hypothesized that chemotherapeutic agents used in the treatment of MM activate survival pathways governing drug resistance [
1]. For example, abnormal activation of the Raf/MEK/extracellular signal-regulated (ERK) pathway occurs in many human cancers, including MM [
2], due to mutations in upstream membrane receptors, Ras and B-Raf, as well as mutations in genes regulating Raf activity that reportedly induces chemoresistance to doxorubicin (Dox) and paclitaxel in breast cancer cells [
3]. Moreover, a phase II study in patients with MM shows activation of both ERK and PI3K/AKT pathways that are attributed to their resistance to erlotinib [
4].
ERK activation has been identified as a potential survival pathway in several tumor types [
5], and recent studies show that ERKs may also be activated in response to chemotherapeutic drugs [
6‐
8] or mTOR inhibitors [
9]. We focused here on whether ERK1 and 2 played critical roles in drug resistance and survival of MM, a generally incurable cancer exhibiting marked chemoresistance. To understand the mechanisms involved, we studied gene expression linked to drug resistance and metabolism, including ATP binding cassette (ABC transporters) genes. This large superfamily of membrane proteins is comprised of 48 members that are divided into 7 different families based on sequence similarities [
10].
We selected doxorubicin (Dox) (Adriamycin) for our studies as this drug has been widely used as the most successful drug of choice to treat MMs in single agent studies [
11,
12] and is used currently in treatment of MMs [
13,
14]. The goal of this study was to understand how Dox-induced resistance develops, and whether it can be overcome by combination therapy. In the present study we demonstrated that Dox treatment causes activation of survival signals (ERK1/2) in MM cells. Combined treatment with a MEK1/2 inhibitor (U0126) plus Dox increased MM cell death over levels observed with Dox alone. Furthermore, using human MM lines expressing shERK constructs, we show that both ERK1 and ERK2 contribute to Dox resistance in human MMs
in vitro and
in vivo. Microarray and qRT-PCR analyses of these cell lines revealed that ERK1 or 2 inhibition was linked to decreases in mRNA levels of ATP binding cassette (ABC) genes. Most importantly, we demonstrate that human shERK1 and shERK2 stable MM lines (in comparison to shControl lines) have a slower growth rate after treatment with Dox in a SCID mouse xenograft model. These data suggest that combined treatment using an ERK1/2 inhibitor or RNA interference approach with Dox (or other chemotherapeutic drug) may be more beneficial than single agent therapy in treatment of MMs.
Methods
Cell culture
None of the human malignant mesothelioma (MM) lines described in this manuscript are commercially available. However, they have been characterized previously by cell size, doubling time, immunohistochemical analyses, electron microscopy, and chromosomal karyotyping as reported (note that the names of these lines have changed since originally reported)[
15]. A sarcomatoid (MO) and epithelioid (ME-26) human pleural MM cell line were obtained from Drs. Luciano Mutti (Maugeri Foundation, Pavia, Italy) and Maurizio Bocchetta (Loyola University, Mayfield, IL), respectively. The HMESO MM line (epithelioid) was originally characterized by Reale et al [
16]. PPMMill, a sarcomatoid human MM cell line, was obtained from Dr. Harvey Pass (NYU School of Medicine, New York, NY). Human mesothelial LP9/TERT-1 (LP9) cells, an hTERT-immortalized cell line phenotypically and functionally resembling normal human mesothelial cells [
17], were obtained from Dr. James Rheinwald (Brigham and Women's Hospital, Harvard University, Boston, MA). Prior to initiating the studies described here, all isolates were confirmed as MM cells by immunohistochemistry using an antibody to calretinin and verified for lack of mycoplasma contamination using a polymerase chain reaction. Additionally, Hmeso tumor xenografts grown in SCID mice were resected and evaluated immunohistochemically by Dr. Michele Carbone and shown to be cytokeratin positive, indicating that they are mesothelial origin. Subsequent karyotype analysis of the Hmeso line by Dr. Joseph Testa demonstrated that the cells were human and possessed several deletions common in mesothelioma lines. These data support what was originally reported for this MM line [
16]. All cells were maintained in 50:50 DMEM/F12 medium containing 10% FBS and supplemented with penicillin (50 units/ml), streptomycin (100 μg/ml), hydrocortisone (100 μg/ml), insulin (2.5 μg/ml), transferrin (2.5 μg/ml), and selenium (2.5 μg/ml), incubated at 37°C in 5% CO
2 and grown to approximately 80-90% confluency [
18]. The synthetic MEK1/2 inhibitor, U0126, and its inactive analog, U0124, were obtained from Calbiochem (La Jolla, CA) and added to cells at 20 μM in medium containing ≤0.2% DMSO [
18]. Control cultures received medium without compounds but with vehicle (≤0.2% DMSO) alone and were treated identically. Doxorubicin (Dox) was obtained from Sigma (St Louis, MO).
Viability determination by cell counting
Viability of cells after Dox treatment was studied by plating cells at 1X105 per well in a 12 well plate. At confluence, cells were maintained in low serum containing medium (0.5% FBS) for 24 h before treating them with different concentrations of Dox (0-100 μM) for 24 h. Cells were trypsinized and counted using a hemocytometer.
MTS assay
Human MM cells (transfected or untransfected) were treated with different concentrations of Dox with and without U0126 or U0124 for 24 h, and cell viability was measured in cells using the colorimetric MTS Assay, CellTiter 96 Aqueous One Solution Cell Proliferation Assay (Promega) as per the manufacturer's recommendations. Absorbance was read at 490 nm on a spectrophotometer indicating MTS bioreduction to a colored formazan product by viable cells.
Western blot analysis
To verify activation of ERK1/2 in MM cells after Dox exposure with and without U0126 or U0124, Western blots were performed as described previously [
18] using antibodies specific to pERK1/2 (rabbit polyclonal anti-pERK1/2, 1:500, Cell Signaling Technology, Danvers, MA), total ERK1/2 (rabbit polyclonal anti-ERK1/2, 1:1000, Cell Signaling Technology, Danvers, MA), and total β-Actin 1:2000 (Abcam, Cambridge, MA). Western blots were quantitated by the Quantity One program [
18] and normalized to total ERK1/2 levels. Western blotting was also performed to validate the selective inhibition of ERK1 or 2 in sh MM lines.
Preparation of RNA and PCR array analyses
LP9 and MM cells were grown to confluence and treated with U0126 (20 μM for 24 h). RNA was prepared and purified using a Qiagen RNeasy plus kit (Valencia, CA). After quality assessment, 1 μg of RNA was employed for cDNA synthesis using the RT2 First Strand Kit (SABiosciences, Frederick, MD). Quantitative Real-Time PCR (qRT-PCR) was performed by the Vermont Cancer Center DNA Analysis Facility using RT2 Real-Time™ SYBR Green PCR Master Mix and Human drug resistance and metabolism template RT2 Profiler™ PCR Arrays (SABiosciences) (7900HT Sequence Detection System, Applied Biosystems). Data were analyzed using an on-line spreadsheet-based data analysis template (SABiosciences). qRT-PCR (TaqMan) was used to validate selected genes using Assay on Demand (AOD) Primers and Probes from Applied Biosystems.
Creation of shERK1 and shERK2 stable MM lines
HMESO cells were selected for these studies because these cells are well-characterized [
16] and form MMs reproducibly after injection into SCID mice. Confluent HMESO cells were transfected with either ERK1 or ERK2, or scrambled control Sure Silencing Plasmids (4 sh sequences for each gene were used) from SA Biosciences (Frederick, MD), using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). After selection for 14 days in G418 (400 μg/ml)-containing medium, clones were screened by qRT-PCR for inhibition of ERK mRNA levels as compared to scrambled control (shControl) transfected clones. Two clones from each shERK1 and shERK2 groups were processed by limited dilution to obtain clones in which individual ERKs were inhibited by more than 70% in comparison to shControl clones. Following this procedure, shERK1 and shERK2 clones exhibiting inhibition of >80% ERK expression were obtained. Similarly, shERK1/2 lines were also created from PPMMill lines to verify observations obtained with HMESO line. The experimentally verified shRNA design algorithm (SABiosciences) assures gene-specificity and efficacy. An advanced specificity search in addition to BLAST built into the algorithm helped to reduce potential off-target effects.
Flow cytometry
To quantitate Dox fluorescence shControl (shCon), shERK1 and shERK2 HMESO cells were grown to confluence and then treated with Dox (0.5 or 5.0 μM) for 1 h or 5 h. Negative controls had no drug added. Cells were washed 3X with phosphate-buffered saline (PBS), trypsinized, counted, suspended in PBS, and Dox fluorescence was examined by flow cytometry using an LSRII flow cytometer (BD Biosciences, MA). A 695/40-nm-band-pass filter with a 685-nm long pass was used to measure Dox fluorescence.
Fluorescence microscopy for Dox fluorescence
shControl, shERK1 and shERK2 cells were grown to confluence in 4-chambered CultureSlides (BD Falcon, Bedford, MA) in medium containing 10% FBS. Media was replaced with that containing 0.5% FBS 24 h before treatment. Cells were either untreated or treated with 0.5 or 5 μM Dox for 1 h or 5 h at 37°C. Slides with attached cells were then washed in PBS and fixed in 100% methanol for 20 min at -20°C. Slides were washed in PBS and water, allowed to dry, and coverslipped with Aqua Poly/Mount (Polysciences Inc., Warrington, PA). Slides were then stored at 4°C until fluorescent images were acquired using an Olympus BX50 Light Microscope with attached mercury epi-fluorescence illumination.
Affymetrix gene profiling
Microarrays were performed on MM cell samples from 3 independent experiments as described previously [
19]. Each of the samples was analyzed on a separate array, i.e., N = 3 arrays per MM line (3 independent biological replicates). A Human U133A 2.0 array (Affymetrix, Santa Clara, CA) was scanned twice (Hewlett-Packard GeneArray Scanner), the images overlaid, and the average intensities of each probe cell compiled. Microarray data were analyzed using GeneSifter software (VizX Labs, Seattle, WA). This program used a "t test" for pairwise comparison and a Benjamini-Hochberg test for false discovery rate (FDR 5%) to adjust for multiple comparisons. A 2-fold cut-off limit was used to assess statistical significance.
Quantitative real time PCR (qRT-PCR)
To validate microarray profiles and PCR Array profiles of genes, qRT-PCR (TaqMan) was performed as described previously [
19]. Triplicate assays were performed with RNA samples isolated from at least 3 independent experiments. Fold changes in gene expression were calculated using the delta-delta Ct method using hypoxanthine phosphoribosyl transferase (
HPRT1) as the normalization control. The Assay on Demand primers and probes used were purchased from Applied Biosystems (Foster City, CA).
Assessment of Dox effects on human tumors appearing after injection of stably transfected shERK1, shERK2 and shControl MM lines in a mouse xenograft model
HMESO cells stably transfected with either shERK1, shERK2 or shControl were injected into 4 subcutaneous sites (5 × 10
6 cells per injection site) on the dorsa of 6 wk old Fox Chase Severe Combined Immunodeficient (SCID) mice (Jackson Laboratories, Bar Harbor, ME). All mice (N = 6/group) were weighed weekly and examined every other day for morbidity and tumor growth (measured using a digital caliper). Immediately after tumor appearance (1 wk post cell injection) each group was divided in two subgroups, each containing 3 mice. Three mice in each group were given Dox (6 μg/50 μl saline/tumor site) 3X weekly. The remaining 3 mice from each group received saline (50 μl/tumor site) 3X weekly. The Dox dose and frequency were previously determined to cause no toxicity to mice. After 6 wk of treatment, all mice were weighed and euthanized by intraperitoneal (ip) injection of sodium pentobarbital, necropsied to determine possible gross metastases, and major organs removed and stored in 4% paraformaldehyde before processing for histopathology. Tumors were characterized using previously described histochemical criteria [
20] and karyotyped to prove that they were human in origin. Tumor volumes were calculated using formula (π × long axis × short axis × short axis)/6. All experiments using mice were approved by the Institutional Animal Care and Use Committee at the University of Vermont College of Medicine.
Statistical analyses
In all in vitro assays, at least 3 independent samples were examined at each time point per group in duplicate or triplicate experiments. Data were evaluated by ANOVA using the Student Neuman-Keul's procedure for adjustment of multiple pairwise comparisons between treatment groups or using the non-parametric Kruskal-Wallis and Mann-Whitney tests. Differences with p values ≤0.05 were considered statistically significant. The difference in tumor growth rates between different groups in in vivo studies was assessed using a hierarchical regression model to take into account the correlation between repeated measurements on the same tumor and multiple tumors in the same animal. In this analysis, the regression coefficient describing tumor growth is modeled as a function of treatment group as well as random variation due to differences between animals and tumors on the same animal.
Discussion
Treatment of various MM lines with doses of Dox much lower than LD
50 concentrations resulted in phosphorylation of ERK1 and 2, the most abundant ERKs in mammalian cells. In addition to Dox, many other anti-cancer drugs such as paclitaxel and cisplatin induce activation of ERKs in different tumor types [
21‐
23]. However, taxol inhibits ERK activation in different cell types depending upon experimental conditions [
23]. In our study, Dox-induced ERK1/2 activation protected MM cells from Dox-induced cell death, as shown when MM lines were pretreated with the MEK1/2 inhibitor, U0126, prior to Dox exposure (Figure
2). In support of our findings, it has been reported that, in most cases, ERK activation protects cells from drug-induced cell death [
23,
24], while in some tumor cells, ERK activation contributes to cell death [
21,
22]. These different effects may be explained by differences in subcellular distribution of specific ERKs, the longevity of ERK signaling, or phosphorylation of different substrates which may dictate death or survival [
25].
We studied 4 different MM lines for Dox responses after ERK1/2 manipulation either with an inhibitor (U0126) or by shRNA approaches. With the use of the ERK1/2 inhibitor (inhibits both ERK1 and 2), HMESO cells were the best responders (most susceptible) as compared to MO and ME-26 (both lines showed the same susceptibility) (Figure
2A). A shRNA approach to inhibit either ERK1 or ERK2 was studied in 2 MM lines (HMESO and PPMMill). Of the two lines studied by this approach, HMESO again showed more sensitivity to Dox-induced killing after ERK1 or ERK2 inhibition as compared to PPMMill (Figure
2B). In addition, in both cell lines, ERK2 inhibition was more effective than ERK1 inhibition in Dox-induced cell killing (Figure
2B).
Although regulation of apoptotic pathways has been implicated in resistance of many cancers to chemotherapy, we show that human MM lines endogenously overexpress many prosurvival genes (
BCL2, cFOS, MET, etc.) in comparison to nontransformed mesothelial cells. The increased levels of these commonly upregulated genes, as reported by our lab and others [
18,
26‐
29] may in part be responsible for drug resistance in MM cell lines. For example,
BCL2 and
BCL-xL antisense treatment facilitates apoptosis in mesothelioma cells, suggesting
BCL2/BCL-xL bispecific antisense treatment in combination with cisplatin or gecitabine may result in a more effective therapy of MM [
30]. Consistent with our findings, ERK1/2 activation has been linked to expression and activation of
BCL2 in various systems [
3,
31] resulting in an anti-apoptotic or survival outcome.
cFOS, a protooncogene and component of activator protein-1 (AP-1), is upregulated by crocidolite asbestos in rat pleural mesothelial cells [
32], and endogenously upregulated in human mesothelioma cell lines and tumors [
18,
28]. We show for the first time that
BRCA1 and
BRCA2 are endogenously overexpressed in MM cells, and are pursuing their mutation and functional status in various MMs. ERK1/2 has been linked to feedback regulation of the tumor suppressor/DNA repair gene
BRCA1 in irradiation induced DNA damage checkpoint activation [
33,
34].
BRCA2 was also endogenously upregulated in MM cells and ERK1/2 inhibition decreased expression of this gene (Table
1), consistent with already published work that ERK1/2 activation inhibits replication of prostate cells via upregulation of
BRCA2 [
35]. Another gene, PPARγ, which was upregulated only in ME-26 and was significantly inhibited by the U0126 MEK1/2 inhibitor is activated via an ERK1/2 dependent COX-2 pathway in macrophages [
36]. Inflammatory pathways involving PPARγ or COX-2 are promising therapeutic targets in a number of cancers [
37]. We also report for the first time the upregulation of a cytochrome P450 enzyme gene, CYP3A4, related to drug metabolism in the ME-26 epithelioid cell line that was decreased 3-fold after addition of U0126. The presence of the androgen receptor and its endogenous expression in sarcomatoid MM cells is also a novel finding, and both AR and ESR2 have been linked to the ERK pathway [
38‐
40] as shown in Table
1 in MO cells. A recent study suggests that ER-β affects the prognosis of MM by acting as a tumor suppressor [
41].
ATP-binding cassette (ABC) transporters transport various molecules, including chemotherapeutic drugs, across extra- and intracellular membranes. Increased expression of one or more of these proteins is seen in almost all resistant cancers and is considered responsible fully or in part for the observed drug resistance in most cancer cell lines. In a previous study using MM cell lines, coordinated overexpression of the multi drug resistance pump (MRP) and gamma-glutamylcysteine synthetase genes, but not MDR1, correlated with Dox resistance [
42]. In the 3 MM lines we studied by PCR array or microarray analysis, different types of ABC transporter genes were endogenously overexpressed as compared to untransformed LP9/TERT-1 mesothelial cells (Table
1, Table
3). The overexpression of different types of ABC genes in different MM cells further confirms the highly heterogenic nature of MM tumors that vary widely in their prognosis and response to therapy. Inhibition of ABC genes by ERK1 or 2 inhibition may be responsible for the increased accumulation of Dox observed in shERK1 and shERK2 MM cells (Figure
3). Among ABC genes inhibited by shERK2 in HMESO cells,
ABCA8 is a relatively uncharacterized new transporter [
43] whereas Dox is a known substrate for
ABCC2,
ABCA2 and
MDR/TAP [
44‐
46]. Our data suggest that different ERKs regulate distinct ABC genes, and a detailed study is needed to understand the roles of different ERKs, including ERK5 that has been linked to chemoresistance in breast cancers [
47], in ABC gene regulation. Consistent with our studies, ERK1 and 2 are linked to regulation of many ABC genes, including
ABCG1,
ABCA1,
MDR1, and
MRP1 in various cancer and non-cancer cells [
6,
48‐
50].
Acknowledgements
We thank Jennifer Díaz for manuscript preparation. We acknowledge the help from the Vermont Cancer Center (VCC) DNA Analysis Facility for qRT-PCR and the INBRE/VGN Microarray Facility for microarrays. We also acknowledge the help of Colette Charland (University of Vermont) for help with Flow Cytometry.
Grant Support: NCI (P01CA114047; M. Carbone, B.T. Mossman, J.R. Testa, H.I. Pass, and A. Shukla), Mesothelioma Applied Research Foundation (A. Shukla and B.T. Mossman), NIEHS T32ES077122 (B.T. Mossman and J.M. Hillegass), NIH (P30CA006927) an appropriation from the Commonwealth of Pennsylvania (J.R. Testa), and the Vermont Cancer Center/Lake Champlain Cancer Research Organization (A. Shukla).
Microarray data accession number: GSE21750
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AS and BTM designed research; AS, JMH, MBM and SLB performed research; HIP contributed new reagents and tools; AS, BTM and PMV analyzed data; and AS and BTM wrote the paper. AS, JMH, MBM, SLB, BTM, PMV, MC, HIP and JRT critically reviewed the manuscript and given final approval of the version to be published.