Background
The incidence of melanoma is on the rise [
1], as is the number of individuals dying from metastatic melanoma [
2]. There are numerous genetically defined activating mutations in melanoma cells leading to enhanced activity of the RAF/MEK/ERK signaling cascade [
3‐
7]. Numerous recent reports focusing on BRAF-targeted therapy designed to interrupt the RAF/MEK/ERK mitogen activated protein kinase (MAPK) pathway in melanoma patients have not made any distinctions between ERK1 and ERK2 [
8‐
15]. To our knowledge no group has attempted to distinguish or target the different isoforms of ERK (e.g. ERK1 or ERK2) specifically in melanoma cells (reviewed in [
16]).
Over 20 years ago, it was discovered that a prominent response to addition of extracellular mitogen to fibroblasts triggered a series of intracellular biochemical events including several kinases such as MEK and p44
MAPK/ERK1 [
17‐
20] and p42
MAPK/ERK2 [
20]. While ERK1 and ERK2 share 84% amino acid sequence homology, knocking out ERK1 vs. ERK2 in mice produces different phenotypes supporting distinct functions for these isoforms [
21,
22]. Many components of RAF/MEK/ERK signaling cascades are mutated or aberrantly expressed in human cancer cells responsible for transformation accompanied by altered proliferation, survival and resistance to treatment [
23]. As clinicians have refocused their therapeutic strategies including targeting mutated BRAF, and downstream molecules such as MEK, the potential efficacy of targeting ERK1 and/or ERK2 has not been tested [
24].
To fill the experimental and therapeutic void regarding the roles for ERK1 and/or ERK2 in human melanoma, a cell line containing mutated BRAF (e.g. A375 cells) was studied in detail using shRNAs selective for each isoform. After confirming effective and selective silencing of ERK1 and ERK2, a series of experiments was conducted to evaluate these kinases in melanoma. While functional differences between ERK1 and ERK2 are controversial depending on the cell type examined [
25], we observed both similar as well as distinct effects such as differentially involving specific pro-apoptotic proteins (i.e. Noxa) in A375 cells upon silencing of ERK1 and ERK2. Given that activation of the ERK pathway is important in melanoma progression [
26], these findings lay the groundwork for new approaches in metastatic melanoma using a molecularly-based targeted approach [
27].
Such novel approaches are urgently needed as it is clear that melanoma cells possess multiple mechanisms to bypass, or overcome drug resistance to agents with clinical success such as PLX4032 (Vemurafenib), a drug targeting mutant BRAF [
9,
28]. An interesting and relevant common intersection point for the various roads to PLX4032 resistance is ERK signaling (ibid). Thus, we decided to expand our studies to not only include silencing of ERK1 and/or ERK2, but to compare and contrast the biological responses and bypass mechanisms triggered by exposing A375 melanoma cells to PLX4032, as well as a MEK inhibitor (PD0325901). The results clearly demonstrate that not only is a combination of ERK1 and ERK2 superior in triggering a caspase-dependent mode of killing A375 melanoma cells compared to PLX4032 or PD0325901, but drug resistant clones infrequently appear by directly targeting ERK. The ability of using ERK shRNAs to not only kill melanoma cells, but to block emergence of treatment resistant clones likely involves not only reductions in levels of phospho-ERKs, but also in upstream reductions in BRAF, CRAF and phospho-MEK thereby interrupting a feedback loop critical to melanoma survival. ERK shRNAs were also shown to increase the sensitivity of melanoma cells to killing by PLX4032 paving the way for combination therapeutic approaches in melanoma. These results demonstrate that targeting ERK in melanoma can overcome the apoptotic resistance of this highly aggressive and difficult to cure tumor.
Methods
Cell culture and chemicals
The human melanoma cell line A375 was purchased from American Type Culture Collection (Manassas, VA, USA) and maintained in DMEM (Lonza, Walkersville, MD, USA) plus 10% FCS (Gemini Bio-Products, Woodland, CA) in a humidified incubator (37°C, 5% CO2). Annexin-FITC was purchased from Biovision Research Products (Mountain View, CA, USA), and tetra methyl rhodamine ethyl ester (TMRE) was purchased from Invitrogen Molecular Probes (Eugene, OR, USA). Pan-caspase inhibitor ZVAD was purchased from BD Biosciences (San Jose, CA, USA). Propidium iodide (PI) was purchased from Sigma Chemical Co (St Louis, MO, USA). PD0325901 and PLX4032 were purchased from Biovision Research Products (Mountain View, CA, USA) and Selleck Chemicals (Houston, TX, USA), respectively. Abs against ERK1, ERK2, pERK1, pERK2, MEK, pMEK, p-Bad, Bak, Bim, PUMA were purchased from Cell Signaling Technology (Beverly, MA, USA); whereas Bcl-XL, Mcl-1, Bad, PARP, caspase 3, Raf-1, Raf-B and GAPDH were purchased from Santa Cruz (Santa Cruz, CA). Ab against Bcl-2 was from DAKO (Glostrup, Denmark), ab against actin from Chemicon Int. (Billerica, MA, USA); Ab against Bax from Calbiochem (San Diego, CA, USA), and against XIAP and activated Bax from BD Transduction Lab (Franklin Lakes, NJ, USA). Primary Abs incubated overnight at 4°C, and secondary Abs were incubated at room temperature for 1 hr.
Production of lentiviral supernatants
Mission TCR shRNAs targeting human ERK1 (NM_002746) and ERK2 (NM_138957) were purchased from Sigma Chemical Co. pLKO.1 Scramble control shRNA plasmid, psPAX2 packaging plasmid and pMD2.G envelope plasmid were provided by Addgene. To make lentiviral particles, HEK-293 T cells were plated into 10 cm plates, 2 × 106 cell/plate, with 8 ml of DMEM plus 10% FBS and no antibiotics. On the next day, for each plate 3 ug of pLKO.1 shRNA plasmid together with 2.25 ug of psPAX2 and 0.75 ug of pMD2.G plasmid were transfected with FuGen 6 reagents (Roche, New Jersey) according to the manufacture's instruction. The transfection reagent was removed by replacing the medium with fresh DMEM containing FBS and penicillin/streptomycin on the following day. The cells were incubated at 37°C, 5% CO2 for 24 hr for another 2 days. Supernatants from 24 hr and 48 hr incubations were harvested and combined followed by centrifugation to remove cell debris and stored at -80°C.
Gene transduction with lentivirus based shRNA
A375 melanoma cells were plated onto 6 well plates at 3 × 105 cells/well and incubated at 37°C, 5% CO2 overnight. Cells were washed 1x with PBS and 1 ml of lentiviral supernatants containing shRNA for either ERK1 or ERK2 or scramble control was added in each well. For ERK1 and ERK2 double knockdown, both supernatants (1 ml of each) were added into one well. All viral supernatants were added with hexadimethrine bromide (Sigma Chemical Co.), final concentration 8 μg/ml) before use. After 4-6 hr incubation, supernatants were changed with fresh medium and cells were incubated for another 1 or 2 days before being split for experiments.
Quantitation of cell viability
Cell death was measured by flow cytometry after staining cells with Annexin-V-FITC and I mg/ml of PI. To investigate DNA degradation, one of the hallmarks of apoptosis, cells were fixed with 70% ethanol and PI stained in the presence of RNAse (10 μg/ml). The relative percentage of cells with hypo-diploid DNA content (sub-G0) was determined by FACS analysis and use of Excel software.
For anchorage independent colony formation assay, A375 cells were transduced with shRNAs for 2 days and then suspended in DMEM with 0.5% agarose solution, and plated onto solidified 1% agarose in 6 well plate at a density of 2500 cells per well in triplicate. A375 cells were maintained in culture by feeding with 0.5 ml fresh DMEM plus 10% FBS medium twice a week, for a total 3 weeks. Two independent assays were carried out and the number of colonies were counted after staining with 0.1% crystal violet solution at the end of each experiment. For anchorage dependent colony assay, A375 cells infected with shRNAs for 2 days were seeded into 6 well plates at a density of 2000 cells per well in triplicate. The cells were cultured with complete DMEM for another 9 days and colony number counted after being stained with 0.1% crystal violet solution.
Mitochondrial membrane potential assay
Assessment of mitochondrial membrane potential was determined by addition of 100 nM of TMRE dye that accumulates in mitochondria of living cells [
29]. Reduction in TMRE retention is indicative of loss of mitochondrial membrane potential.
Detection of intracellular levels of activated BH3-multidomain proapoptotic bax protein
A375 cells were fixed with 2.5% paraformaldehyde (10 min, room temperature), washed and incubated with a primary ab detecting the activated configuration of Bax (BD Pharmingen Inc, San Diego, CA, USA) in FACS buffer with 0.3% saponin as previously described [
30]. The percentage of melanoma cells with activated Bax was measured by fluorescence intensity greater than control ab levels.
Immunoblotting
Western blot analysis was performed as previously described [
31]. Briefly, cells were harvested by scraping and lysed with M-Per mammalian protein extraction reagent (Thermo Scientific, Rockville, IL, USA) supplemented with protease inhibitor cocktail (Roche Diagnostics GmbH, Germany) and phosphatase inhibitor cocktail set II, (Calbiochem, Los Angeles, CA, USA) followed by shaking and centrifugation at 4°C. Supernatants were collected as whole cell extracts and protein concentrations were measured using Bradford reagents (Bio-Rad laboratories, Hercules, CA, USA). 30 ug of proteins were resolved by SDS- PAGE and transferred to PVDF membrane followed by 1 hr blocking with buffer supplied by LI-COR Biosciences (Lincoln, NE, USA). Blots were probed with primary Abs overnight at 4°C, washed and incubated with corresponding fluorescence-labeled secondary Ab for 1 hr at room temperature in dark. Protein levels were visualized with LI-COR Infrared Imaging System.
Reverse transcriptase-real time PCR
Total RNA was extracted from A375 cells using Trizol (Invitrogen Life Technologies, Inc.). One microgram of RNA was reverse transcribed to cDNA using TaqMan reverse transcription reagents (Applied Biosystems, Foster City, CA). The following specific primer pairs were used: BRAF: 5'-CTC GAG TGA TTG GGA GAT TCC TGC-3', (forward), 5'-CTG CTG AGG TGT AGG TGC TGT CAC-3' (reverse); 18sRNA: 5'-GGC GCC CCC TCG ATG CTC TTA G-3', (forward), 5'-GCT CGG GCC TGC TTT GAA CAC TCT-3', (reverse). PCR reaction was performed by adding 25 μl of 2X SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA), 19 μl DEPC treated H2O, 2 μl of each primer, 2 μl of diluted cDNA template. DNA amplification was completed in ABI prism model 7300 thermal cycler. All reactions were run in triplicate and two independent assays were performed. The comparative expression level was determined by applying the calculation of 2(Δ Ct B-raf-ΔCt 18s).
Statistical analysis
All statistical analyses were performed using the unpaired, two sided Student's t-test, and results considered significant when P values were less than 0.05.
Discussion and conclusions
Despite an impressive array of exciting recent results highlighting the importance of the RAF/MEK/ERK pathway in melanoma patients, surprisingly these reports did not make any distinctions between ERK1 and ERK2. By contrast, we focused on dissecting important similarities and differences in the biology and therapeutic targeting efficacy between ERK1 and ERK2 in A375 melanoma cells. Taken together, the detailed characterization of the cellular and molecular events in A375 melanoma cells following silencing of ERK1 and/or ERK2 revealed many insights of potential therapeutic significance. First, it is clear that ERK1 and ERK2 control similar, but not identical, signaling events in melanoma cells. For example, reducing ERK2, but not ERK1 increased levels of Noxa; yet combining ERK1 plus ERK2 shRNAs reduced Noxa below constitutive levels; whereas reducing ERK1 but not ERK2 increased XIAP levels (Figure
4B). Also, while both isoforms compete for MEK, reduction in one isoform did not lead to increased phosphorylation of the other isoform in A375 cells (Figure
1A). Second, targeting ERK1 and/or ERK2 trigger greater cell death in A375 cells compared to chemical inhibitors of mutant BRAF (PLX4032) or MEK (PD0325901) that possessed similar rapid reductions in phospho-ERK levels (Figures
1A, B, and
5A, B).
Besides the aforementioned attributes of targeting ERK1 and/or ERK2, there are several other advantages to this strategy. First, since drug resistance to PLX4032 is linked to recovery of pERK activity [
34], it was therapeutically beneficial that sustained and near complete reduction in both pERK1 and pERK2 can be accomplished using the shRNAs to obviate this problem. Second, along the same line of inquiry, unlike PLX4032 which is preferentially active in tumor cells bearing the V600E mutation of BRAF, reducing levels of pERK1 and/or 2 does not depend on the mutation status of the melanoma cell. Indeed, it has been observed that within melanoma lesions, there can be both clones containing both BRAF mutant alleles as well as wild-type BRAF [
35]. Third, by silencing ERK1 and/or ERK2, the feedback loop between ERK and RAF and MEK is interrupted as clearly observed by comparing treatment of cells with MEK inhibitor which led to increased phospho-MEK as reported by others [
36], in contrast to the decrease in phospho-MEK after our shRNA approach (see below for more discussion).
Additional evidence for feedback loops, and significant molecular complexities became apparent by probing for changes in BRAF and CRAF with these various treatments. It should be noted that phosphorylation events amongst the components of this pathway can be either activating or inhibitory [
37]. A simple negative feedback loop has been suggested whereby activated ERK would phosphorylate and thereby inhibit further MEK activity [
38]. While blocking ERK with the chemical inhibitors or the ERK shRNAs would therefore be expected to enhance phospho-MEK levels, this was only observed using PD0325901 (Figure
6A). Thus, feedback loops are more complicated as regards ERK and MEK. In another scenario, downstream from ERK are phosphatases (e.g. DUSP) that serve as negative feedback components, and decreased ERK levels could lead to reduced phosphatases thereby facilitating accumulation of phospho-MEK [
39].
Perhaps even greater complexity was uncovered between ERK and more upstream kinases such as BRAF and CRAF. In this situation, markedly diminished levels of BRAF and CRAF accompanied using shRNAs, reducing ERK1 and or ERK2 levels. ERK is known to be able to hyperphosphorylate members of the RAF serine/threonine kinase family, leading to decreased signaling, and it is possible the altered phosphorylation triggered by decreased ERK levels influenced the ability of the antibody to recognize BRAF or CRAF due to conformational changes [
38]. Support for this posttranscriptional modification was provided by the RT-PCR results in which the shRNAs did not influence relative mRNA levels for BRAF (Figure
6B). A feedback loop could also be observed using the MEK inhibitor as regards increased BRAF levels and decreased electrophoretic motility possibly due to hyperphosphorylation [
40,
41]. While undoubtedly complex, elements of the feedback loops (both positive and negative) will require exploration beyond the scope of this current work, but are likely contributors to the lack of emergence of treatment resistant clones using the ERK shRNAs.
Despite the remaining challenges required to more fully understand the biology of silencing ERK1 and or ERK2, it is clear that killing of melanoma cells by silencing ERK1 and/or ERK2 is caspase dependent (Figure
3B), provoking altered mitochondrial function (Figure
4A), and highlights a key role for the ERK signaling pathway in tumor cell survival [
42]. The higher level of cytotoxicity exhibited by the ERK shRNAs and activation of caspase cascades also are likely to contribute to the paucity of treatment resistant clones (Figure
2B) with our novel therapeutic approach. Furthermore the ability of PLX4032 to kill A375 melanoma cells was greatly increased when combined with ERK1 and/or ERK2 shRNAs (Figure
7), suggesting the possibility of such combination therapies worthy of a additional study.
Based on the current in-vitro findings, it may be worth moving forward in the clinic with several drug combinations to prevent drug resistance to PLX4032, which would include drugs targeting ERK, or with drugs targeting MEK. While these findings were consistently observed in the A375 cell line, future studies are indicated to explore other V600E BRAF mutant bearing cell lines to compare and contrast with the results for A375 cells. In conclusion, targeting the BRAF-MEK-ERK pathway for melanoma patients is rapidly accelerating in the clinic [
43,
44], and further studies using agents that silence ERK1 and/or ERK2 should be seriously considered for future lines of inquiry to overcome the notorious apoptotic resistance and treatment bypass repertoire of melanoma cells either alone or in combination with PLX4032 [
45].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JQ and HX carried out all experimental procedures. BJN conceived and designed the study, wrote and guided the editing of the manuscript. All authors read and approved the final manuscript.