The therapeutically actionable long non-coding RNA ‘T-RECS’ is essential to cancer cells’ survival in NRAS/MAPK-driven melanoma

Finding effective therapeutic targets to treat NRAS-mutated melanoma remains a challenge. Long non-coding RNAs (lncRNAs) recently emerged as essential regulators of tumorigenesis. Using a discovery approach combining experimental models and unbiased computational analysis complemented by validation in patient biospecimens, we identified a nuclear-enriched lncRNA (AC004540.4) that is upregulated in NRAS/MAPK-dependent melanoma, and that we named T-RECS. Considering potential innovative treatment strategies, we designed antisense oligonucleotides (ASOs) to target T-RECS. T-RECS ASOs reduced the growth of melanoma cells and induced apoptotic cell death, while having minimal impacton normal primary melanocytes. Mechanistically, treatment with T-RECS ASOs downregulated the activity of pro-survival kinases and reduced the protein stability of hnRNPA2/B1, a pro-oncogenic regulator of MAPK signaling. Using patient- and cell line- derived tumor xenograft mouse models, we demonstrated that systemic treatment with T-RECS ASOs significantly suppressed the growth of melanoma tumors, with no noticeable toxicity. ASO-mediated T-RECS inhibition represents a promising RNA-targeting approach to improve the outcome of MAPK pathway-activated melanoma.


Introduction
Melanoma is the deadliest form of skin cancer and its incidence is rising. 1 Most melanoma tumors harbor oncogenic mutations that activate the MAPK signaling pathway, which regulates cancer cell proliferation and survival. 2Even though blocking the MAPK pathway with targeted drugs that inhibit BRAF or MEK kinases has been effective in treating melanoma, 2,3 acquired or up-front therapeutic resistance is commonly observed in patients. 2Discovering molecular mechanisms that may be targeted to reinforce the inhibition of the MAPK pathway or to block parallel pathways that cause melanoma tumors to bypass the effects of MAPK-therapy, is a biomedical priority. 2,4signi cant proportion of the transcriptome in both normal and diseased cells remains untranslated.The majority of these transcripts are longer than 200 nucleotides and belong to the group of long non-coding RNAs (lncRNAs). 5,6LncRNAs are RNA sequences that span ≥200 nucleotides. 6Although they do not code for proteins, their expression is tissue speci c and is often altered in cancer.LncRNAs can play a role in oncogenesis through various mechanisms, for instance regulating the expression of cancerspeci c genes, changing the epigenetic landscape via histone interaction, or serving as splicing factors. 6me lncRNAs can also activate or stabilize proteins through direct binding, which can promote malignancy and in uence the response or resistance to therapies. 7,8Beyond commonly known lncRNAs such as MALAT1, H19, HOTAIR or SAMMSON, 6,[9][10][11] more lncRNA transcripts keep being identi ed as important regulators of cancer.Whether and how lncRNAs may be involved in melanoma progression or in NRAS-mutated/MAPK-dependent cancers, remains largely unknown. 12ong recent drug development innovations, RNA-targeting therapies are considered to be particularly promising due to the very broad spectrum of actionable targets they give access to, as well as the highly speci c and strong inhibitory effect they can exhibit. 13,14Multiple RNA-targeting strategies have been shown to induce apoptosis and eliminate tumor cells. 13,14A number of RNA-targeting drugs, including Antisense Oligonucleotides (ASOs), are already FDA-approved. 13,14Several ASO drug modalities are currently being tested in clinical trials for cancer treatment. 14re, we report on the discovery of a melanoma-associated lncRNA that we named 'T-RECS', and that is signi cantly upregulated in NRAS-/BRAF-mutated cell lines and patient tumors compared to normal/nonmalignant cells or tissues.T-RECS, which stands for 'Transcript REgulating Cell Survival', is coded by the gene AC004540.4(ENSG00000225792), a lncRNA observed in non-alcoholic fatty liver disease and hepato-carcinoma.15,16 Our study shows that reducing the levels of T-RECS with ASOs induces apoptosis in cancer cells, restores sensitivity to MAPK-therapy, and suppresses tumor growth in mouse models.
Mechanistically, we demonstrate that, unlike other lncRNA inhibiting ASO treatments, T-RECS-targeting ASOs speci cally and signi cantly suppress the activity of pro-survival kinases and downregulates the levels of the pro-oncogenic hnRNPA2/B1 protein.Our results establish that the lncRNA T-RECS is a novel melanoma vulnerability, and that ASO-based RNA-targeting strategies can be deployed to inhibit such lncRNA dependency.Altogether, ASO-mediated T-RECS inhibition represents an unprecedented therapeutic opportunity to treat NRAS-/BRAF-mutated melanoma.

Results
MAPK-pathway activated melanoma cells and tumors express high levels of the lncRNA AC004540.4   To characterize the expression of lncRNAs in MAPK-driven melanoma, we used D04 and MM415 cell lines, two commonly studied NRAS Q61 -mutated melanoma models.We also analyzed the transcriptional pro le of patient-derived primary melanocytes that were either not transduced or modi ed to express NRAS Q61 or an empty control vector (i.e., primary human melanocytes abbreviated as PHM, PHM Q61 , PHM E ).A schematic of the computational process that we used to unbiasedly identify differentially expressed genes including potential lncRNAs, is presented in Fig. 1a.Validation of NRAS Q61 expression and downstream activation of the MAPK pathway in PHM Q61 cells are provided in Suppl.Fig. 1a.We compared the paired-end non-poly A enriched 101-bp RNA-Seq data from PHM Q61 , D04 and MM415 to the pro les of PHM E or PHM (Fig. 1b; PHM Q61 vs. PHM E , D04 vs. PHM, M415 vs. PHM; detected genes were classi ed as coding for a protein or a non-coding RNA or a transcript of unknown coding potential (TUCP)).Out of the >15,000 differently expressed genes we detected, 197 genes were conserved in NRAS Q61 expressing PHM Q61 , D04 and MM415 cells (Fig. 1c, Suppl.Fig. 1b).Among these 197 genes, 81 were upregulated lncRNAs.
To hone in on lncRNAs that may be most clinically relevant, we computationally mined the Cancer Genome Atlas -Skin Cutaneous Melanoma (TCGA-SKCM) dataset and found that 24 of these 81 lncRNAs were also expressed in more than 90% of patients' melanomas harboring an NRAS-mutation (Fig. 1d; FPKM-values > 0.2).In particular, a lncRNA identi ed as AC004540.4(Ensembl Gene ID: ENSG00000225792) was expressed in >97% of patients' melanomas, with an average FPKM of 11.28 (Fig. 1d; red highlight).AC004540.4 is located on chromosome 7, expressed as two isoforms, and detectable in some normal tissues (Suppl.Fig. 2a-b; Genotype-Tissue Expression (GTEx) database); AC004540.4 is not conserved in other species.We compared the expression level of AC004540.4 in biospecimens collected from healthy, non-cancerous skin samples banked in the GTEx database (n=1,305 patient tissues), versus the collection of NRAS-or BRAF-mutated melanoma tumors available in the TCGA database (n=366 patient tissues).We found that the lncRNA AC004540.4was signi cantly more expressed in MAPK pathway-mutated melanomas than in normal skin tissues (Fig. 1e; p < 0.001).

AC004540.4-inhibition reduces melanoma cell growth
To assess whether the lncRNA AC004540.4may confer survival and growth advantage to melanoma cells, we initially used endoribonuclease-prepared siRNA (esiRNA), which is a highly effective method to degrade speci c RNA targets. 17We found that esiRNA-mediated silencing of AC004540.4greatly reduced the growth of D04 and MM415 cell lines (Fig. 2a).
Next, we tested the effects of two RNA-targeting approaches that are clinically relevant: siRNA and antisense oligonucleotides (ASOs), which have received FDA and/or EMA approval for therapy and are commonly used in animal models. 18We designed three siRNA sequences and three ASO constructs of the GapmeR type to speci cally target AC004540.4(respectively AC004540.4siRNA-1/-2/-3 and ASO-1/-2/-3 used in Fig. 2b-d and 2e-g; see Methods for details), and tested their effects on cell viability in comparison to non-targeting siRNA or ASO sequences, which served as control for non-speci c effects of oligonucleotide treatment (Fig. 2b-g; light vs. dark grey bars).Using a panel of 8 melanoma cell lines (D04, MM415, WM1366, VMM39, Sk-Mel-2, WM3629, Sk-Mel-28, WM3211) and 2 primary melanoma cells (Hs852.T, AV5), we found that these RNA-targeting interventions signi cantly inhibited cell growth (Fig. 2b-g and Suppl.Tables 1-2).While siRNAs targeting AC004540.4reduced cell growth by 39% compared to Control siRNA on average (Fig. 2b-d), ASOs had a signi cantly greater effect, inhibiting cell growth by 80% on average and up to 95% in some instances (Fig. 2e-g).Notably, both siRNAs and ASOs had minimal impact on normal, primary human melanocytes (PHM; far left bar pairs in each graph of Fig. 2bg), which underlines the speci c, anti-melanoma effect of these RNA-targeting interventions while sparing normal melanocytes.
To further determine the e cacy of ASO intervention, we performed clonogenic assays.Compared to Control ASO treatment at equal dose and incubation time, AC004540.4-targetingASO drastically reduced the ability of D04, MM415 and Sk-Mel-28 cells to form colonies (Fig. 2h; p-val < 0.004).
The lncRNA AC004540.4 is enriched in the nucleus of melanoma cells LncRNAs display diverse subcellular distributions, with some primarily located in the nucleus, others in the cytoplasm. 19To test whether AC004540.4 is a predominantly nuclear or cytoplasmic lncRNA, we applied RNA-FISH staining on formalin xed and para n embedded D04 and MM415 cell pellets.Analysis of ISH-stained cell pellets showed that AC004540.4 is mainly located in the nucleus (Fig. 3a-b).To validate these ndings, we performed subcellular fractionation of D04 and MM415 cells followed by RNA extraction and qRT-PCR to compare the ratio of nuclear versus cytoplasmic RNA levels.As control of RNA subcellular location, we used the GAPDH mRNA and H19 lncRNA, which are mainly cytoplasmic 20,21 , and the MALAT1 lncRNA, which is mainly nuclear. 19The nuclear-to-cytoplasmic ratios of GAPDH, H19 and AC004540.4were further normalized to the nuclear-to-cytoplasmic ratio of MALAT1.We found that AC004540.4was mainly enriched in cells' nucleus, and signi cantly more nuclear compared to MALAT1 (Fig. 3c; 6.7-and 5.2-fold respectively in D04 and MM415).

AC004540.4 promotes cell survival pathways and can be targeted to induce apoptosis
To start uncovering the molecular mechanisms underlying the effect of AC004540.4-inhibitionand the potential role of AC004540.4 in MAPK-signaling, we performed gene expression pro ling of D04 cells treated with either non-targeting Control ASO or AC004540.4-targetingASO.This comparison revealed that, 72h after treatment, AC004540.4-inhibitioninduced signi cant changes in 1,067 genes (Fig. 4a; cut off for DE genes: FDR p-values: <0.05 (y-axis); cut off for log2 fold change: > +1.5 or < -1.5 (x-axis)).Further functional annotation and clustering analysis was done using DAVID to search for biologically relevant GO terms and pathways.Terms grouped with "protein tyrosine kinase activity", "Ras guanylnucleotide exchange factor activity", and "PI3K-AKT signaling pathway" were found among the highest ranked terms (Fig. 4a; see Suppl.Table 3-5 for details).
To demonstrate that these effects are speci c to AC004540.4-targeting,and not a general effect of inhibiting pro-oncogenic lncRNAs, we treated the same three cell lines with an ASO that targets MALAT1, which is a known pro-oncogenic lncRNA 10,11,19 , and measured changes in kinase activity using HT-KAM.We found that after 24h, the kinome signatures of cells treated with MALAT1 ASO were different from AC004540.4ASO treated cells (Fig. 4e; average Pearson correlation r < 0.3; all data normalized to non-targeting Control ASO).More speci cally, and unlike AC004540.4inhibition, we found that targeting MALAT1 did not result in a signi cant decrease in the activity of pro-survival kinases AKT1, CDK1, LYN, YES1, CHEK1, PKA, LKB1, PKCa and PIM1 (Fig. 4d, right panel).The difference between the activity of these kinases in D04, D04RM and MM415 cells treated with AC004540.4ASO versus MALAT1 ASO, was signi cant (Fig. 4d, left vs. right panel; p-val < 0.00007).

T-RECS regulates hnRNPA2/B1 protein stability
Considering that lncRNAs can regulate protein-coding genes located nearby their genomic location 31 , we tested whether T-RECS may be involved in the expression of HnRNPA2/B1, an oncogene associated with many cancer types including melanoma. 32,33We found that T-RECS ASO treatment induced a sharp decrease in HnRNPA2/B1 protein levels in D04 cells, as shown by immunoblot and immuno-uorescence (Fig. 5a,b; up to 75% reduction in HnRNPA2/B1 protein).Since the protein half-life of the HnRNPA2/B1 protein is known to range from several days to up-to-4 weeks in primary human cells 34 , and since we found that directly targeting hnRNPA2/B1 with an ASO induced >50% reduction in RNA level but was accompanied with <20% reduction in protein level of HnRNPA2/B1 after 1 day (Suppl.Fig. 3a-b), our results suggest that T-RECS may stabilize HnRNPA2/B1 protein through physical binding.To test this, we pulled down HnRNPA2/B1 protein and measured the levels of RNA bound to it.To account for potential unspeci c interactions with the RNA-binding protein HnRNPA2/B1 33 , we compared T-RECS RNA levels to two other lncRNA transcripts, MALAT1 and HOTAIR 11 .D04 melanoma cell lysates were immunoprecipitated using either IgG control or HnRNPA2/B1 antibodies (Fig. 5c).We found T-RECS to be strongly and speci cally enriched in the HnRNPA2/B1 pull down lysate (Fig. 5d; left bar).Conversely, HOTAIR and MALAT1 were signi cantly less co-precipitated with HnRNPA2/B1 (Fig. 5d; two right bars), even though their baseline expression levels were either similar to T-RECS (HOTAIR; 1.5-fold), or far more elevated than T-RECS (MALAT1; > 500-fold) (Fig. 5e).Next, we used RNAscope combined with immuno-uorescence to determine whether T-RECS and HnRNPA2/B1 colocalized.We found that T-RECS and HnRNPA2/B1 signals overlapped in the nuclei of D04 and MM415 cells (Fig. 5f-g).Together, our results indicate that the lncRNA T-RECS may directly interact with and stabilize the HnRNPA2/B1 protein (Fig. 5a-g).
T-RECS and hnRNPA2/B1 are co-expressed in melanoma tumors Since HnRNPA2/B1 can promote tumorigenesis and modulate MAPK-pathway signaling 33 , we asked whether HnRNPA2/B1 may also be upregulated in MAPK pathway-mutated melanoma tumors, and whether it may be correlated with T-RECS expression.Comparing expression levels of HnRNPA2/B1 in non-cancerous, patient-derived skin samples from the GTEx database (n=1,305 patient tissues) versus in NRAS-or BRAF-mutated melanoma tumors from the TCGA database (n=366 patient tissues), we found that HnRNPA2/B1 is signi cantly more expressed in MAPK-driven melanoma (Fig. 5h; p < 0.001).Given that both T-RECS and HnRNPA2/B1 genes are signi cantly upregulated in melanoma, we tested whether their expression is correlated in the GTEx and TCGA datasets.We calculated expression correlation between T-RECS and hnRNPA2/B1 and compared it to the correlation of each of the two genes to 10 sets of 200 randomly chosen genes (see Methods for computational details).We found that in non-malignant skin samples (GTEx), there was no signi cant difference in the correlation of T-RECS and hnRNPA2/B1 compared to the correlation of each of them to any 200 random gene set (Fig. 5i; left panels; p > 0.05; only one of the ten different 200-gene sets is shown; all sets are provided in Suppl.Fig. 3c-d).However, in melanoma (TCGA), the correlation between T-RECS and hnRNPA2/B1 was systematically and signi cantly higher when compared to the random gene sets (Fig. 5j; right panels).Ranking of expression correlations showed that the correlation of T-RECS with hnRNPA2/B1 is signi cantly higher ranked in TCGA than GTEx when compared to the ten different 200-gene sets (Suppl.Fig. 3e-f).These results show that T-RECS and hnRNPA2/B1 are not co-expressed in healthy skin, but they are jointly and non-randomly upregulated in melanoma.

T-RECS ASO treatment synergizes with MEKi to inhibit the growth of melanoma cells
Having demonstrated that MAPK-driven melanoma tumors and cell lines depend on the expression of T-RECS to grow, we hypothesized that melanoma cells may respond to MAPK-inhibition and overcome therapeutic stress by upregulating T-RECS to survive.To test this, we rst measured T-RECS levels in D04 cell treated for 72 hours with a MEK-inhibitor (MEKi; trametinib) at two different concentrations using qRT-PCR.We found that T-RECS was upregulated in MEKi-treated cells compared to control (Fig. 6a; 4-to-6 fold).These results suggest that, as a therapeutic strategy, inhibiting T-RECS may augment the response of melanoma cells to MEKi.Using a cell survival assay, we found that combining trametinib with T-RECS ASO inhibited the growth of melanoma cell lines and primary melanoma cells more than trametinib alone (i.e., D04, MM415 and AV5), and that this combination was synergistic (Fig. 6b).Furthermore, we established that melanoma cell lines that we induced to become MEKi-resistant (i.e., D04RM, MM415RM, WM3629RM, and Sk-Mel-2RM cells chronically exposed to increasing MEKi doses; see Methods), remained sensitive to T-RECS ASO treatment (Fig. 6c).Finally, we found that upregulation of T-RECS upon MEKi was accompanied by the stabilization and increased protein levels of hnRNPA2/B1 (Fig. 6d; 3.5-to-6 fold).Together, our results indicate that melanoma cells upregulate T-RECS in response to MEK-targeting therapy, and that T-RECS is a druggable vulnerability to either augment the anti-tumor response to MEK inhibition, or to restore therapeutic response in melanoma cells that have become resistant to MEK inhibition.

T-RECS ASO treatment signi cantly suppresses melanoma tumor growth in vivo
To translate our ndings in vivo and assess the potential clinical value of ASO intervention for melanoma, we tested the e cacy of T-RECS ASO treatment using melanoma cell line-derived xenograft and patientderived xenograft (PDX) mouse models.We initially treated D04 tumor-bearing mice with 200µg subcutaneous ASO injection, three times a week (i.e., total: 600µg ASO/week), over the course of three weeks.We observed that the average tumor size was signi cantly smaller in the T-RECS ASO treatment group compared to the Control ASO group (Fig. 7a; p < 0.005).
Next, in order to reduce the dosage of ASO and follow the most recent advances in ASO-based therapy, we co-administered Control ASO or T-RECS ASO with an in vivo transfection reagent that is under clinical development (i.e., JetPEI®).We adjusted the treatment protocol down to 60µg subcutaneous ASO injection and twice a week (i.e., total: 120µg ASO/week), for three weeks.We tested this strategy using three tumor xenograft models, established from a melanoma cell line (D04), a primary melanoma cell line (AV5), and a NRAS Q61 melanoma PDX (TM01341).We found that this regimen induced signi cant tumor growth inhibition in tumors treated with T-RECS ASO in comparison to Control ASO across all models (Fig. 7b-d; p < 0.005, p < 0.04 and p < 0.003 in respectively D04, AV5, and TM01341).
Mouse weight, a surrogate to assess potential treatment toxicity, remained stable over the course of therapy for Control ASO and T-RECS ASO in all cohorts (Fig. 7e-h).We measured T-RECS lncRNA expression levels in tumors harvested from mice treated with T-RECS ASO versus Control ASO and coadministered with JetPEI, and we con rmed that T-RECS was signi cantly reduced in T-RECS ASO-treated D04, AV5 and TM01341 tumors (Fig. 7i).To test whether apoptosis was induced in vivo by T-RECStargeting ASO treatment, we used immunohistochemistry to detect cleaved-caspase-3, a marker of apoptosis.We found that D04 xenograft tumors treated with T-RECS ASO systematically displayed greater levels of cleaved-caspase-3 staining in comparison to Control ASO treated tumors (Fig. 7j).Since ASOs have been reported to accumulate in the liver and cause toxic side effects, we also analyzed liver tissue for potential histopathologic changes by H&E staining and found no detectable hepatotoxicity across ASO-treated animals (Fig. 7k).Altogether, these in vivo results demonstrate that systemic treatment with an ASO that speci cally targets the lncRNA T-RECS, is a well-tolerated and effective strategy to suppress the growth of MAPK pathway-mutated melanoma tumors.

Discussion
][4] Here, we investigated whether particular long non-coding RNAs (lncRNAs) may be speci cally associated with the tumorigenicity of NRAS-mutated/MAPK-driven melanoma, and whether such lncRNAs may be amenable for therapy using cutting-edge RNA-targeting interventions (ASO).Leveraging advanced genomic and proteomic methods, we found that MAPK-pathway activated melanomas display increased levels of expression of the lncRNA gene AC004540.4,which we named T-RECS for "Transcript REgulating Cell Survival".We showed that T-RECS plays a pivotal role in repressing apoptosis of melanoma cells and tumors via regulation of pro-survival kinases and hnRNPA2/B1 functions.Our characterization of T-RECS as an actionable vulnerability shows that lncRNAs are therapeutically tractable, and that ASO-based RNA-targeting strategies can suppress the growth of NRAS/MAPK-driven melanoma tumors.
Therapies that target RNAs have opened a new clinical era.RNA targeting is a transformative approach because in principle it allows to target the product of any transcripted gene, whatever it may code for, thus immensely increasing the pool of druggable targets in any cell.The utility of ASO therapy is wellrecognized to treat non-cancerous diseases and it is currently undergoing rigorous evaluation for cancer treatment in many clinical trials. 13,14Previous studies showed that ASOs can be more e cacious than siRNAs, especially to target nuclear RNAs. 19,35We found that the lncRNA T-RECS was signi cantly enriched in the nucleus of melanoma cells, and that targeting T-RECS with ASOs was highly effective to reduce the levels of T-RECS RNA and to inhibit melanoma cell growth and survival.T-RECS ASOs also performed systematically better than T-RECS siRNAs.Prior reports show that systemic ASO treatment can pose challenges, including limited ASO delivery and effectiveness 35 as well as liver toxicity. 36To augment the binding a nity, stability, and target speci city, we modi ed the T-RECS-targeting GapmeR ASOs to incorporate LNAs and employed a fully modi ed phosphorothioate (PS) backbone. 37To reduce the risk of hepatotoxicity and increase delivery, we conducted in vivo e cacy experiments using unassisted and assisted delivery methods, the latter involving the transfection reagent JetPEI®.JetPEI is a reagent used in clinical trials for melanoma gene therapy that ensures robust, effective and reproducible RNA transfection (ClinicalTrials.govID: NCT04925713, NCT04853602, and NCT04160065).Both delivery methods signi cantly reduced melanoma tumor growth while causing no noticeable toxic side effects in mice, including in the liver.Importantly, co-administering JetPEI with T-RECS ASOs enabled us to greatly reduce the amount of ASO while effectively suppressing tumor growth (5-fold less ASO amount compared to unassisted delivery and at greater time intervals), thus further improving on the potential for treatment tolerability.In the perspective of additional clinical development, we envision that the versatility of ASOs to accommodate chemical modi cations increasing drug bioavailability and pharmacokinetics 35,37 , will allow to further enhance the anti-tumor effects of T-RECS ASOs while ensuring minimal toxicity.
To understand the mechanism of action of the T-RECS-targeting ASO treatment and how T-RECS may regulate the growth and survival of NRAS-mutated/MAPK-pathway driven melanoma tumor cells, we pro led the activity of kinase enzymes in melanoma samples.HT-KAM-derived kinome signatures revealed that treatment with T-RECS-targeting ASO speci cally and signi cantly suppressed the activity of several kinases that promote cell survival and counteract apoptosis, such as AKT1, PIM1, CDK1, SRCs, PKA or PKC.This kinome reprogramming effect was not observed when treating melanomas with other ASOs that target other lncRNAs, including the known pro-oncogenic lncRNA MALAT1.Since these kinases play pivotal roles in orchestrating cell survival mechanisms, [26][27][28][29][30] our results suggest that the lncRNA T-RECS may act as an central hub that regulates the intricate networks of pro-survival pathways, and confers a unique growth advantage to melanoma cells.This also suggests that, in order to blunt the complex signaling responses, that cancer cells engage and depend on to survive and keep proliferating, it may be therapeutically more effective to target T-RECS than any single one of these individual kinases or kinase-regulated pathways.
3][4] These MAPK pathway-targeting therapies are often administered in combination regimens to reinforce the inhibition of the MAPK pathway, whether to induce a profound, upfront therapeutic response or to restore therapeutic sensitivity in relapsing melanomas. 4Our results suggest that such combinatorial targeted therapy strategy could be applied in the context of T-RECS dependency.Indeed, we found that treatment with T-RECS ASOs ampli ed the therapeutic e cacy of -and synergized with-the MEKi trametinib in NRAS-mutated melanoma tumor cells.As well, melanoma cells responded to MEKi treatment by increasing T-RECS expression, and MEKi-resistant melanoma cells remained sensitive to T-RECS ASO.Based on the kinome reprogramming triggered by T-RECS ASO and its signi cant impact on AKT1 and PIM1 activities, it is possible that jointly targeting MEK and T-RECS directly inhibits both the MAPK and AKT pathways, which are the two main drivers of therapeutic resistance and melanoma progression. 2,24r study also revealed intricate regulatory connections between the lncRNA T-RECS and hnRNPA2/B1, a pro-oncogenic protein whose interaction networks includes nuclear-enriched lncRNAs. 38Treatment with T-RECS ASOs produced a rapid and striking reduction in hnRNPA2/B1 protein levels.We found a direct interaction between T-RECS and hnRNPA2/B1 proteins.In agreement with other studies that showed how lncRNAs can act as stabilizers of proteins, 39 our ndings suggest that T-RECS may stabilize hnRNPA2/B1 through their association in the nucleus.T-RECS and hnRNPA2/B1 exhibited elevated expression levels in melanoma, and their expression was signi cantly correlated in melanoma tissue specimens, while this was not observed in skin biopsies from healthy patients.Additionally, the upregulation of T-RECS coincided with elevated hnRNPA2/B1 protein levels, suggesting a potential reciprocal interaction between these molecules.Our ndings suggest that the interactions between T-RECS and hnRNPA2/B1 may promote their pro-oncogenic effects in melanoma.
Most oncogenic lncRNAs, including T-RECS, exhibit strong tissue-speci c expression patterns. 6While our study focuses on the role of T-RECS in NRAS-mutated/MAPK-driven melanoma cell survival, it likely has broader functions in other types of melanomas and cancers, as well as in pre-and non-malignant tissues.
Our data contribute novel insights into the previously uncharted territory of the lncRNA T-RECS.Our ndings are the rst to demonstrate an oncogenic function for T-RECS.Our results indicate that targeting T-RECS represents an untapped therapeutic opportunity to treat NRAS-mutated/MAPK-driven melanoma.Along with other recent studies, our work demonstrates that lncRNAs are promising targets for nextgeneration cancer therapy, and that ASOs can be used as emerging therapeutic tools to precisely target lncRNA molecules, opening a new window of therapeutic opportunities to treat melanoma.

Reference Annotation
A custom reference annotation of total 75,506 transcripts, referring to 35,101 genes, of which 16,405 were classi ed as non-coding, was built by integrating 13,870 lncRNA genes from the GENCODE 40

Assembly and identi cation of previously unidenti ed lncRNAs
After alignment to the human genome with TopHat (version 2.0.11), the reads were assembled into transcripts with Cu inks (version 2.1.1).To discover novel lncRNAs, we excluded all transcript IDs that overlapped with any gene IDs from our initial reference annotation.To lter out transcriptional noise, we kept only multi-exonic transcript IDs which were > 200bp and had at least one intron region > 10bp.Next, isoforms were merged with Cuffcompare.

Coding Potential Assessment of Transcripts
To identify transcript IDs with a coding potential, we ran (a) the HMMER3 algorithm (considering all 6 open reading frames) to identify any protein family domain as noted in the Pfam database (release 27.0, UAACUAUUAGCUUCAUGUUUUUACCCA) and AC004540.4siRNA-3 (guide strand sequence: AUCACUGAAUUGACAUGCUGUUGGCAG) were designed by and purchased from Integrated DNA Technologies, Inc (IDT).For pooled non-targeting Control siRNA design, the guide strand sequences UGGUUUACAUGUCGACUAA, UGGUUUACAUGUUGUGUGA, UGGUUUACAUGUUUUCUGA and UGGUUUACAUGUUUUCCUA were used.The AC004540.4ASO-1 (sequence: GACTGGAGATAATTAA), AC004540.4ASO-2 (sequence: TGCGCGGCGGAAAGAA), MALAT1 ASO (sequence: TAAAGCCTAGTTAACG) and hnRNPA2/B1 ASO (sequence: GACCGTAGTTAGAGG) GapmeRs were purchased from QIAGEN N.V. and designed using the GeneGlobe design and analysis hub.The AC004540.4ASO-3 (sequence: CTCATGAGCTGTCGTA) GapmeR was designed by and purchased from IDT.For non-targeting Control ASO design, the QIAGEN N.V. standard sequence AACACGTCTATACGC was used.In all experimental procedures, a consistent antisense oligonucleotide concentration of 50nM was used, unless speci ed otherwise.The transfection reagent Lipofectamine TM 3000 (2ul/ml) was added according to the manufacturer's instructions.

Cell growth analysis
Dependent on cell doubling time, 0.7-2 x 10^3 cells were seeded in 96 well-plates one day prior to treatment.Cells were treated with ASOs for ve days unless speci ed otherwise.Total luminescence was measured on the Synergy™ HT (Agilent Technologies Inc) plate reader using Promega® CellTiter-Glo® and Gen5 software.Cell growth was normalized to Control ASO treatment.

Colony formation Assay
Dependent on cell doubling time, 1-2 x 10^3 cells were seeded in six well-plates one day prior to treatment.Six days after transfection, cells were washed with PBS, xed with 10% neutral buffered formalin, and stained with 0.1% crystal violet solution.Colonies were de ned as cell conglomerates with > 50 cells.
Digital images of plates were evaluated by two independent reviewers for colony counts.The nal counts were calculated as the average count of both reviewers for all triplicates.

RNA extraction and quantitative real-time PCR (qRT-PCR)
TRIzol™ Solution (Thermo Fisher Scienti c®), Phenol:chloroform:isoamyl alcohol (MilliporeSigma®) or NucleoSpin® RNA kit (Takara Bio USA, Inc.) were used for extracting Total RNA from cells and tissues according to the manufacturer's instructions.Total RNA was quanti ed by NanoDrop™ ND-1000 (Thermo Fisher Scienti c®) or Quibit™ 4 (Thermo Fisher Scienti c®).50ng or RNA was reverse transcribed using the cDNA synthesis and gDNA removal QuantiTect® Reverse Transcription Kit (Thermo Fisher Scienti c®).Real time PCR was performed using the iTaq TM Universal SYBR® Green Supermix (Bio-Rad Laboratories, Inc.), 20ng of cDNA and on a QuantStudio TM 5 Real-Time PCR System or a 7500 fast real time PCR system (both from Thermo Fisher Scienti c®).Relative gene expression was calculated using the comparative Ct method, normalized to GAPDH or β-actin.Primers are listed in Suppl.Table 7. Primers were obtained from IDT.

In situ hybridization and Immuno uorescence
In situ hybridization analyses were conducted using the RNAscope Multiplex Fluorescent Reagent Kit version 2 system, using custom-made probes to AC004540.4(T-RECS).Following RNAscope, sections were immediately processed for immuno uorescent staining with hnRNPA2/B1.The Multiplex Fluorescent Reagent Kit version 2 was used according to the manufacturer's instructions.Immuno uorescent staining was performed using primary antibody dilutions of 1:200 for hnRNPA2/B1 (ProteinTech 14813-1-AP), incubating overnight at room temperature.Sections were treated with secondary antibodies of donkey anti-rabbit AlexaFluor-Plus 555 (1:400; Invitrogen) and DAPI for 2 hours prior to imaging.TSA Cy5 uorophores were used to amplify the uorescent signals for T-RECS, in accordance with the instructions in the Multiplex Fluorescent Reagent Kit version 2. 100nM nal ASO concentration for 1 day was used for cells that received treatment.

Fluorescence Imaging
Fluorescence imaging was performed using a Zeiss Axio Observer Z1 with a 20X objective.All images for assessment of each outcome measure were captured at a constant exposure, using identical microscope settings.

Puri cation of Nuclear and Cytoplasmic RNA
Total nuclear and cytoplasmic extracts were obtained using the SurePrep™ Nuclear/Cytoplasmic RNA puri cation kit (Thermo Fisher Scienti c®) according to the manufacturer's instructions.RNA extraction and qRT-PCR were performed as described above.

RNA-sequencing
Total RNA was isolated using the RNeasy® mini-Kit (QIAGEN N.V.) following the manufacturer's protocol.Quality check for extracted RNA was done using 2100 Bioanalyzer (Agilent Technologies Inc.) or Tapestation 4200 System (Agilent Technologies Inc.).All samples had a RIN score > 8.For samples used for identi cation of MAPK-responsive lncRNAs, cDNA-sequencing libraries were prepared using the Illumina® TruSeq® Total RNA Sample kit and paired-end, 101-bp sequencing was performed by Centrillion Genomic Services (Centrillion Tech.) on an Illumina® HiSeq® 2000.Sequence reads were aligned to the human genome (hg19) using TopHat (Version 2.0.11).For DE gene analysis of ASOtransfected D04 samples (3 days of either Control ASO or AC004540.4ASO-1 incubation), NEBNext® ultra-RNA library prep kit (New England Biolabs® Inc.) with rRNA depletion and paired-end, 2x150-bp sequencing was performed by Genewiz® on an Illumina® HiSeq® 4000.Sequence reads were aligned to the GRCh38 reference genome.

Analysis of T-RECS ASO treatment induced DE genes
Differential expression (DE) analysis was done using DESeq2.Differentially expressed genes were de ned by more than 1.5-fold changes (log 2 > 0.58 or < -0.58) in expression with FDR < 0.05.Pathway enrichment analysis was done using DAVID Functional Annotation Clustering analysis (version 6.8). 44,45nase activity mapping technology For samples to be analyzed with the HT-KAM platform, cells were treated with ASOs for 24 hours and at ~85% con uency cells were washed three times with cold PBS and lysed with freshly prepared 1X cell lysis buffer (1ml per 2.5x10 6 cells) (10x Cell lysis buffer, Cell Signaling Technology®, Cat.No.9803), complemented with 1x Halt Protease and Phosphatase (Thermo Fisher Scienti c® Cat.No 1861281).Lysates were scraped off, spun down at 14,000 rpm. at 4°C for 15 min and supernatants were stored at −80°C.The high throughput kinase activity mapping (HT-KAM) platform uses arrays of peptides that act as sensors of phosphorylation activity. 24The phospho-catalytic signature of samples is established from simultaneously occurring ATP-consumption tests measured in the presence of individual peptides that are experimentally isolated from each other.Assays were run in 384 well-plates, where each experimental well contains one peptide.The nal 8µL reaction mixtures per well contain: (a) kinase assay buffer (1X KaB: 2.5mM Tris-HCl (pH7.5),1mM MgCl 2 , 0.01mM Na 3 VO 4 , 0.5mMglycerophosphate, 0.2mM dithiothreitol (DTT), prepared daily; (10X KaB Cell Signaling Technology®, Cat.No.9802), (b) 250nM ATP (prepared daily with 1X KaB; Cell Signaling Technology® Cat.No.9804), (c) 200µg/ml 11-mer peptide (lyophilized stocks originally prepared as 1mg/ml in 1X KaB, 5% DMSO), and (d) samples made from cell at ~10µg/ml total protein extract.Samples are kept on ice and diluted in 1X KaB < 30min before being used.Controls with no-ATP, or no-peptide, or no-sample as well as ATP standards are run side-by-side within each 384 well-plate.High-throughput liquid dispensing of all reagents is performed using a Biomek® FX Laboratory Automation Workstation from Beckman Coulter.
All reagents are kept on ice and plates on cold blocks until enzymatic reactions are started.Once the dispensing of the reaction mixtures is complete, the plates are incubated for 1h at 30ºC.ATP is detected using Kinase-Glo revealing reagent (Promega®; Cat.No V3772), which stops the activity of the kinases and produces a luminescent signal that directly correlates with the amount of remaining ATP in the samples.Luminescence is acquired using the Synergy 2 Multi-Mode Microplate Reader from BioTek.
Luminescence data are inversely correlated with the amount of kinase activity.For a more detailed description of the peptide sensors design, sequence and connectivity between peptides and kinases, as well as data normalization steps and analysis, refer to: 22,24,25 .The activity of kinase enzymes is derived from their respective subset of biological peptide targets included in the assay.

Caspase Glo 3 & 7 assay
Dependent on cell doubling time, 2-3 x 10^3 cells were seeded in 96 well-plates one day prior to transfection.One day after transfection Total luminescence was measured on the the Synergy™ HT (Agilent Technologies Inc) plate reader using The Promega® Caspase-Glo® 3/7 Assay and Gen5 software.Experiments were performed in quadruplicates.Annexin V assay 1 x 10^5 D04 cells were seeded in six well-plates one day prior to transfection.One day after transfection live, dead, and apoptotic cells were differentiated using the Invitrogen™ Dead Cell Apoptosis Kits with Annexin V (Cat.no: V13241), following manufacturers protocol.Cells were sorted using a BD® LSR II Flow Cytometer.
Protein extraction and immunoblotting 1 x 10^5 D04 cells were seeded in six well-plates one day prior to transfection.Total protein lysates were homogenized in 1x RIPA buffer and Halt protease and phosphatase inhibitor cocktail (Thermo Fisher Scienti c®) followed by centrifugation at 14,000 RPM/minute at 4°C.Protein concentration was quanti ed using the Pierce™ BCA Assay Kit (ThermoFisher Scienti c®).Linear absorbance was measured using the Synergy™ HT (Agilent Technologies Inc) plate reader.Total protein in 1× Laemmli buffer with 10% 2-mercaptoethanol was separated by SDS/PAGE, transferred for 15 h to a PVDF membrane (IPVH00010; MilliporeSigma®) by electroblotting with 20% (vol/vol) methanol, and blocked for 1 h in in Intercept (TBS) blocking buffer (LI-COR®).Membranes were incubated overnight at 4 °C with primary antiserum for hnRNPA2/B1 (Abcam®, cat.no.: ab31645, dilution 1:750) and b-Actin (Cell Signaling Technology®, cat.no.: 8457, dilution 1:2500) following incubation with secondary Goat Anti-Rabbit serum (LI-COR®, dilution 1:5000) for 1 h and scanned using the Li-COR® Odyssey® Imaging system.Protein expression was quanti ed using Image Studio Lite Version 5.2.5

Calculation of Combinational Index (CI)
The effects of drug combinations on cell growth assessed by calculating the combination index (CI) , where E a and E b correspond to the effects of drugs A and B alone at a given concentration, and E a,b corresponds to the combined effects of drugs A and B at the same concentration, and a combination index of < 0 indicates synergy while a combination index of > 0 relates to an antagonistic effect.The individual combination indices per drug combination were then averaged.Cells were treated for three days.
Animal models experimental procedures were approved by the O ce of Research institutional Animal Care and Use Program (IACUC, Chair: Jeremy Lieberman, MD) at the University of San Francisco (UCSF).All in vivo studies were conducted under the authorized protocol number AN174613-03.Mice were maintained in a pathogen free environment and had free access to food and water.For PDX tumor models, the PDX type TM01341, derived from liver metastasis of a male melanoma patient, was engrafted on 4-to 6-week-old NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ mice (Stock.no005557) on the right posterior dorsal ank (n = 4/group).
For cell-line models, 2x10^6 D04 (n = 5/group) and AV5 (n = 3/group) cells in 150µl of PBS and 50µl of Matrigel were subcutaneously injected on the posterior dorsal anks of 4-to 6-week-old homozygous nude Foxn1 nu /Foxn1 nu mice (Stock.no007850).Mice and PDX tissue were obtained from JAX®.Tumor size was measured using a digital caliper and the formula 0.5 x (length x (width^2)) was used to calculate tumor volume.For unassisted delivery, mice were treated 3x/week with 200µg of ASOs diluted in an overall amount of 100µl PBS.For assisted delivery, mice were treated twice a week with 60µg of ASOs and 9.6µl of in vivo JetPEI® diluted in an overall amount of 200µl 5% glucose.ASO injections were applied subcutaneously.Mice were weighted at the days of treatment and observed for signs of distress or disorder.Mice were euthanized after three weeks of ASO application or when tumors reached a diameter of > 2cm.All experiments were performed in accordance with the UCSF Laboratory Animal Resource Center (LARC) guidelines.After euthanasia, tumor samples and liver tissue were excised and xed in formalin solution, followed by storing in 70% ethanol and Immunohistochemistry staining.Tumor samples were also placed in RNAlater™ Stabilization Solution (Thermo Fisher Scienti c®) and stored at -20°C.Invitrogen™ TRIzol™ Solution (Thermo Fisher Scienti c®) was used to extract RNA from tissue and qRT-PCR was performed to analyze gene expression.

Immmunohistochemistry
Tumor tissues were extracted from mice immediately after euthanasia and xed in 10% neutral buffered formalin for 24 hours, followed by storage in 70% EtOH.Histopathology was conducted by the UCSF Histology and Biomarker Core.FFPE sections were collected at 4-micron thickness and mounted on center of positively charged glass slides (Superfrost Plus Slides Fisher Item# 12-550-15) following lab standard procedures for the sectioning of para n blocks.Slides where air-dried overnight for 8 to 24 hours then baked and depara nized.For hematoxylin and eosin (H&E) staining, slides were baked at     1+2.

Supplementary Files
This is a list of supplementary les associated with this preprint.Click to download.

Figures
Figures

Figure 1 A
Figure 1

Figure 5 T
Figure 5

Figure 6 T
Figure 6

Figure 7 T
Figure 7 for 15 minutes prior to staining; for immunohistochemistry (IHC) staining, slides were baked overnight at 60°C prior to staining.H&E staining was performed on the Leica XL Autostainer using hematoxylin (part no.6765015 from Thermo Scienti c) for seven minutes and alcoholic eosin (Part No. 6765040 from Thermo Scienti c) for 20 seconds.IHC was done on the Roche Ventana Discovery Ultra Autostainer using antibodies for Cleaved-Caspase-3 (Cell Signaling Technology®, cat.no.: 9664S, dilution 1:600).Ventana Protocol Summary: CC1 cell conditioning at 95°C for 24 minutes, Inhibitor time 12 minutes, primary antibody incubation at 36°C for 60 minutes, anti-Rabbit HQ detection at 37°C for 12 minutes, Chromagen: DAB.The Zeiss Axio scanner was used for digital slide scanning with bright eld images collected at 20X magni cation using Zeiss Zen software.Additionally, we thank William Chou at UCSF for technical support and Ernesto Llamado for assistance throughout the project.