The lincRNA MIRAT binds to IQGAP1 and modulates the MAPK pathway in NRAS mutant melanoma

Despite major advances in targeted melanoma therapies, drug resistance limits their efficacy. Long noncoding RNAs (lncRNAs) are transcriptome elements that do not encode proteins but are important regulatory molecules. LncRNAs have been implicated in cancer development and response to different therapeutics and are thus potential treatment targets; however, the majority of their functions and molecular interactions remain unexplored. In this study, we identify a novel cytoplasmic intergenic lincRNA (MIRAT), which is upregulated following prolonged MAPK inhibition in NRAS mutant melanoma and modulates MAPK signaling by binding to the MEK scaffold protein IQGAP1. Collectively, our results present MIRAT’s direct modulatory effect on the MAPK pathway and highlight the relevance of cytoplasmic lncRNAs as potential targets in drug resistant cancer.

In this study, we performed de novo transcriptome analyses (RNA-Seq) to identify lncRNAs differentially expressed in melanoma cells with acquired resistance to small molecule inhibitors of the MAPK cascade. By comparing treatment naïve cells to their drug resistant counterparts, we identified MIRAT (MAPK Inhibitor Resistance Associated Transcript), an lncRNA located on chromosome 8, significantly overexpressed in resistant cells. Analyses in NRAS mutant melanoma cells revealed MIRAT's direct involvement in drug resistance to MAPK pathway inhibitors. Our results offer first evidence of a cytoplasmic lncRNA contributing to targeted inhibitor resistance in melanoma with the potential to open up new therapeutic avenues to further improve treatment.

Results
Acquired drug resistance to MAPK inhibitors in melanoma leads to the expression of the lin-cRNA MIRAT (MAPK Inhibitor Resistance Associated Transcript). To assess whether acquired drug resistance changes the melanoma transcriptome, we first established cell lines with acquired resistance to small molecule inhibitors of BRAF and MEK. We treated 5 melanoma cell lines (2 NRAS Q61 mutant: DO4, MM415; 3 patients derived BRAF V600E mutant: AV1, AV4, AV5) with increasing dosages of the MEK inhibitor trametinib and the BRAF inhibitor PLX4720. After six months of continuous drug incubation, we compared the drug sensitivities of treated clones with non-treated parental cells. Defined as a significant change in GI50 values (concentrations of drugs resulting in 50% growth inhibition relative to controls), 4 cell lines became resistant to the treatment (suffix RM; Resistant to MAPK inhibitor), while one cell line (AV5) remained sensitive to the treatment (Suffix ChrExp; Chronically Exposed (Supplementary Fig. 1; for definition of resistant cell lines see material and methods). Resistant cells MM415RM, DO4RM, AV1RM and AV4RM, but not the still drug sensitive AV5ChrExp, retained high signaling through the MAPK pathway despite increasing drug concentrations, as evidenced by high phospho-ERK levels ( Supplementary Fig. 1).
Next, we performed paired-end RNA sequencing from the above-mentioned cell lines and compared the transcriptomes of resistant cell lines to their parental counterparts. The initial bioinformatics analysis provided a list of differentially expressed genes (Supplementary Material and Methods). These results were further filtered with differentially expressed genes found in (i) acute (6 hours) trametinib treated cells (suffix A; Acutely treated) and (ii) chronically exposed cells (suffix ChrExp), to account for transcriptome changes after short-term and long-term drug exposures not leading to resistance (Supplementary Tables 1,2,3,4). Hierarchical clustering identified relationships among different conditions, and we found that resistant cell lines clustered together (Supplementary Fig. 2a; the heat-map illustrates the expression of differentially expressed genes). Surprisingly, after applying the different filters, only one lncRNA was differently expressed in all resistant clones (Fig. 1a,b). This lncRNA was intergenic (lincRNA) and mapped to the human chromosome 8q.24.12. We termed this transcript MIRAT: MAPK Inhibitor Resistance Associated Transcript ( Supplementary Fig. 2b).
Probing for other, already known and probable mechanism of resistance to the MEK inhibitor trametinib we did not detect any additional NRAS mutations, SOX10 and MITF were not downregulated, and COT, BRAF and RTKs were not significantly upregulated in both NRAS mutant resistant cell lines. We did not find mutations in the coding regions of MEK1/2 and MEK1/2 RNA expression values and protein levels were comparable between parental and resistant cells ( Supplementary Fig. 3) 26 . Due to low proliferation rates of patient derived avatar cell lines we performed all further experiments only in NRAS mutant DO4 and MM415 cell lines.
MIRAT is enriched in the cytoplasmatic compartment of MEK inhibitor resistant NRAS mutant melanoma cell lines and is found in patient samples. We focused our efforts on the further characterization of the newly identified transcript. We confirmed MIRAT's presence in both NRAS and BRAF mutant resistant cells by reverse transcription quantitative PCR (RT-qPCR) (Fig. 1c), and verified its exact sequence by Rapid Amplification of cDNA ends analysis (RACE). MIRAT has a 5′ cap and 3′ polyadenylated tail, indicating a transcription by RNA polymerase II. The predicted secondary structure of MIRAT is depictured in Supplementary  Fig. 4. 27 Subcellular RNA isolation showed that MIRAT is primarily localized in the cytoplasmic cell compartment, indicating transcript stability (Fig. 1d).

MIRAT is upregulated in a time-and dose-dependent manner in response to MAPK inhibition.
Our RNA-Seq experiments detected MIRAT in drug-resistant melanoma cell lines but not in those treated for only 6 hours (DO4A, MM415A). To investigate the dynamics of MIRATs' expression during time on drug the, we incubated cells with the MEK inhibitor trametinib and quantified MIRAT by RT-qPCR at different time points.
MIRAT was upregulated as early as 24 hours after MEK inhibitor treatment, but at expression levels significantly lower than in resistant clones. Additionally, we found that MIRAT expression increased with higher trametinib doses (Fig. 2b). In parallel we tested the effect of the MEK inhibitor trametinib on the MAPK pathway and found a bimodal response to the drug with a p-ERK reduction during early drug incubation an its increase after 96 hours (Fig. 2a,b).
MIRAT could modulate the MAPK signaling pathway. Since MIRAT was expressed as early as 24 h after trametinib incubation and increased up to 500-fold in resistant clones (Fig. 2a), we investigated if MIRAT was involved in the modulation of the MAPK pathway signaling using transient gene over-expression and RNA-silencing assays. We transfected the DO4 melanoma cell line with the two most abundant MIRAT isoforms (isoforms 1 and 2). Forty-eight hours after transfection we measured ectopic MIRAT levels similar to those observed in DO4RM cells, and confirmed a prevalent cytoplasmatic location of the transcript ( Supplementary Fig. 5). The ectopic MIRAT overexpression led to a relative signaling resistance to the MEK inhibitor trametinib, evidenced by higher phospho-ERK levels after trametinib exposure in transfected cells (Fig. 3a,b).
To test whether MIRAT silencing might increase signaling sensitivity to MAPK pathway inhibition, we used two independent RNA interference methods: a pool of customized siRNA and the endoribonuclease-prepared siRNAs (esiRNA). esiRNA, known to have less off-target effects than siRNA and to be particularly efficient in silencing noncoding transcripts 29 , was superior in depleting MIRAT compared to siRNA and was thus used for all further knockdown experiments ( Supplementary Fig. 6). MIRAT depletion did not significantly affect cell viability in resistant cells ( Supplementary Fig. 7) but led to reduced p-ERK levels in both resistant cell lines tested (Fig. 3c), while the DO4 cells were only minimally affected ( Supplementary Fig. 8).
To assess signaling changes in response to MIRAT depletion in resistant cells, we performed gene expression analyses comparing cells treated with esiRNA targeting MIRAT to cells treated with control siRNA (Supplementary Dataset 1). 'Cell surface receptor linked signal transduction' was found to be the most affected biological process following MIRAT knock-down, supporting its involvement in MAPK related signaling events ( Supplementary Fig. 9).
The effect of MIRAT on the MAPK signaling cascade could be mediated by IQGAP1. After we established MIRAT's connection to receptor signaling and the MAPK pathway we aimed to identify its possible binding partners to elucidate molecular mechanisms by which MIRAT interconnects with the MAPK pathway. We used RNA Antisense Purification Mass-Spectrometry (RAP-MS) and RNA binding-protein immunoprecipitation (RIP) to search for possible binding partners. Among possible binding candidates we focused on IQGAP1 ( Supplementary Fig. 10, Supplementary Dataset 2). IQGAP1 is a scaffold protein that binds to BRAF, MEK and ERK 30 and is known to modulate their activity 31 . When we performed RIP experiments we found MIRAT enriched in the IQGAP1 immunoprecipitated products. As MIRAT was found in the negative control IgG-rabbit antibody conditions and in the SNRNP70 antibody (Rabbit polyclonal IgG) pull down ( Supplementary Fig. 11).
Binding of IQGAP1 to MIRAT was confirmed by enriched MIRAT levels in RNA-protein extracts from IQGAP1 pull down experiments, but not in the immunoglobulin G (IgG)-mouse bound sample used as negative control (Fig. 3c). However, MIRAT was also found in the negative control IgG-rabbit antibody conditions and in the SNRNP70 antibody (Rabbit polyclonal IgG) pull down ( Supplementary Fig. 11).
To verify if MIRAT and IQGAP1 are involved in modulating the MAPK pathway, we performed further gene silencing experiments. Knock-down of both, MIRAT and IQGAP1, led to decreased levels of p-ERK in both resistant cell lines (Fig. 3d).

Discussion
The studies of mechanisms of resistance to targeted small molecule inhibitors could lead to a better understanding of tumor biology and thus improve therapeutic strategies. Different mechanisms of resistance as a response to MAPK inhibition have already been reported 13,16,[32][33][34][35] , but only few studies have assessed the role of lncRNAs 20 . Here, we test how melanoma cell transcriptomes change following MAPK inhibitor treatment. We compare treatment naïve, short-term treated, chronically drug exposed and drug resistant cell lines and identify the lncRNA MIRAT which seems to be implicated in MAPK signaling after targeted pathway inhibition.
MIRAT mapped to the human chromosome 8q.24.12, a region characterized by only a few protein-coding and several lncRNA genes. This region is known for multiple SNPs and genomic alterations associated with cancer and has emerged as an important region for genetic susceptibility to various malignancies such as breast, prostate, bladder, colon, lung, ovaries, pancreas and brain cancer 36 . Our findings support the biological relevance and indicate that the region 8q.24 is associated with drug resistance in cancer. Although MIRAT has recently been annotated, its function and intracellular location has yet not been studied 19 .
We found that MIRAT is predominately present in the cytoplasm, which is in contrast to most other known lncRNAs 37 . MIRAT silencing and overexpression, as well as RNA-protein interaction results, indicate that this lncRNA modulates the MAPK signaling pathway, at least in the tested NRAS mutant melanoma cell lines. However, this modulation is not sufficient to account for large changes in cell viability and thus is only a part of multiple mechanisms of resistance following small molecule inhibitor treatment. The exact description of such mechanisms of resistance on genetic, epigenetic and protein levels is beyond the scope of this manuscript but further studies are warranted.
Our findings suggest that MIRAT functions by binding to the scaffold protein IQGAP1, which is an essential protein facilitating signaling from MEK to ERK 30,31 . IQGAP1 is known to be elevated in various cancers, and could serve as a potential therapeutic target, not only in melanoma but also in other cancers with hyper-activated MAPK signaling 31,[38][39][40] . MIRAT could bind to IQGAP1 and stabilize it, decelerate its decay and thus facilitate signaling through the MAPK pathway. Also the fact that MIRAT is elevated only 24 hours after MAPK inhibition, points towards a regulatory feedback mechanism.
Interestingly the microarray data shows mostly upregulated genes following MIRAT's knockdown in both cell lines tested. This may be due to a variety of other regulatory mechanisms which follow slight changes in MAPK signaling or due to further still not discovered functions of MIRAT. Furthermore, we have to consider that silencing RNA off-target effects cannot be fully excluded.
Taken together, we describe a noncoding RNA directly involved but not leading to drug resistance to small molecule inhibitors. Our results highlight the relevance of cytoplasmic lncRNA for oncogenic signaling pathways, and suggest that MIRAT and its protein-binding partner IQGAP1 may serve as potential therapeutic targets in melanoma through their modulation of the MAPK pathway. Future strategies of improving target therapies in cancer are likely to involve selective interference with lincRNAs and their protein-binding partners.

Material and Methods
Cell culture and establishment of resistant clones. Human NRAS mutant DO4 and MM415 were provided by Boris Bastian, University of San Francisco, California. DO4 and MM415 harbor NRASQ61L mutation. Cell lines were maintained in RPMI 1640 media (UCSF Cell Culture Facility) supplemented with 10% (vol/vol) heat inactivated fetal bovine serum. Primary avatar cell lines (AV1, AV4, AV5) were obtained from metastases of BRAF mutant patients. All experimental protocols were approved by UCSF Human Research Protection Program Institutional Review Board (IRB# 12-0948), all patients signed informed consent, and methods were carried out in accordance with relevant guidelines and regulations. Cell lines were incubated at 37 °C under 5% CO 2 .
To establish resistant cell lines, NRAS and BRAF mutant cell lines were continuously exposed to increasing concentrations of respectively MEK inhibitor (GSK1120212) (Selleckchem) and BRAF inhibitor PLX-4720 (Selleckchem). Resistant clones were selected, maintained and propagated in a complete medium supplemented with drugs for 6 months. At the end of this time period, the drug sensitivity of the resulting clones was tested seeding cells in 96 well plates with a density of 3,000-5,000 cells per well, treating them with the respective inhibitor increasing concentrations. Cell viability was measured using CellTiter-Glo Luminiscent Cell Viability Assay (Promega; G7570), according to the manufacturer's protocol. Luminescence was measured on the SynergyHT plate reader (BioTek) using Gen5 software (Version 1.11.5). All experiments were performed at least in triplicates. Cell lines were defined resistant to a drug if GI50 values (concentrations of drugs resulting in 50% growth inhibition relative to controls) were increased by at least two-fold (calculated by CalcuSyn software (Biosoft, Cambridge, UK; Version 2.1)). Resistant clones were termed RM (Resistant to MAPKi): DO4-RM, MM415-RM, AV1-RM, and AV4-RM. AV5 cells did not acquire resistance, therefore the resulting clone after 6 months' exposure to PLX-4720 was termed ChrExp (Chronical Exposed) Immunoblotting. Cell lysates were prepared in RIPA buffer supplemented with Halt protease and phosphatase inhibitor cocktail (Thermo Scientific). Equal amounts of protein, as measured by BCA protein assay, were resolved in 4-12% Bis-Tris NuPage gradient gels (Life Technologies) and transferred electrophoretically on a polyvinylidene difluoride 0.45-m membrane. Membranes were blocked for 1 h at room temperature in 5% bovine serum albumin (BSA) or non-fat milk in TBST before being incubated overnight at 4 °C with the primary antibodies. All primary antibodies were diluted 1:1,000 in 5% BSA or non-fat milk in TBST. After three washes of 5 min in TBST, secondary antibodies were diluted 1:3,000 in 5% non-fat milk in TBST and incubated for 1 h at room temperature. After another three washes in TBST, detection of the signal was achieved by incubating the membrane and exposing it on autoradiography films from Denville Scientific (Metuchen, NJ, USA). The primary antibodies used were: anti-phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (no. 9101), anti-AKT (no. 9272), anti-phospho-AKT (Ser473) (no. 9271). The secondary antibodies: Amersham ECL anti-rabbit IgG, HRP-linked whole ab from donkey (no. NA934-1ML) and anti-mouse IgG, HRP-linked whole ab from donkey (no. NA931-1ML) were purchased from GE Healthcare Life Sciences. Anti-ERK2 (C-14) (sc-154) was purchased from Santa Cruz Biotechnology. Anti-IQGAP1 (clone AF4) (no. 05-504) was purchased from EMD Millipore. Anti-MITF (C5) was purchased from ThermoFisher Scientific (no. MA5-14146). Patient sample collection and RNA extraction. We selected a set of patient melanoma tumors formalin-fixed paraffin-embedded (FFPE) blocks from the tissue bank of our institution (University of California, San Francisco). All experimental protocols were approved by UCSF Human Research Protection Program Institutional Review Board (IRB# 12-0948), patients signed informed consent, and procedures were carried out in accordance with relevant guidelines and regulations. RNA was extracted using RNeasy FFPE kit (Qiagen, Valencia, CA). RNA yield and A260/A280 ratio were monitored with a NanoDrop ND-1000 spectrometer (NanoDrop Technologies). We pre-amplified the transcripts and performed TaqMan-qRT-PCR. GAPDH was used as an endogenous control, and to express the relative abundance of the transcripts making a comparison between samples possible. Patient information was extracted retrospectively from the electronic patient charts into an excel file, following patient anonymization. The TANRIC web application was used to test MIRAT expression among available TCGA melanoma samples accordingly to described methods 28  Cloning procedures. MIRAT  For esiRNA, we designed esiRNA libraries as previously described 42,43 . Briefly, esiRNA library was designed using DEQOR (a web based tool for design of siRNAs 44 ) choosing the optimal gene-silencing region for PCR (400bp-600bp), primer design, attaching the T7 RNA polymerase promotor to the 5′ end of each primer. esiRNA were synthesized with the following steps: template generation, in vitro transcription to generate long dsRNA, digestion to generate a pool of short dsRNAs, purification of esiRNA. Transfection was performed with Lipofectamine.

RIP assay and RNA antisense purification-mass spectrometry (RAP-MS). Magna RIP TM RNA-Binding
Protein Immunoprecipitation Kit (no. 17-700) was performed to identify and validate the protein-binding partner of MIRAT, accordingly to the manufacturer's protocol. RNA antisense purification and mass spectrometry were performed as previously described 20,45-48 . Subcellular fractionation. Subcellular fractionation and RNA purification for the localization of MIRAT were performed using the Cytoplasmic and nuclear RNA purification Kit (#2100,34700) according to the manufacturer's protocol.
Oligonucleotide sequences 5′-3′, probes and primers. MIRAT  Statistical analyses. The statistical analyses were performed using Graphpad Prism V.7.0c and Stata 12.0 statistical software. Comparisons between two independent non-normally distributed groups were performed using the nonparametric Wilcoxon rank-sum test; P values less than 0.05 were considered statistically significant.