Melanoma Addiction to the Long Non-Coding RNA SAMMSON

Prof. Jean-Christophe Marine and Dr. Eleonora Leucci work at the Laboratory for Molecular Cancer Biology, K.U. Leuven in Belgium. They recently published a Nature paper showing that melanoma cancer cells are addicted to the lncRNA SAMSSON. They have been using Antisense LNA GapmeRs to silence the lncRNA SAMSSON and suppress tumor growth.

You recently published a spectacular paper in Nature about a lncRNA expressed specifically in melanoma. Can you briefly summarize your findings?

In collaboration with the group of P. Mestdagh (UGent), we have established an exciting link between malignant melanoma and a non-coding RNA gene called SAMMSON for Survival Associated Mitochondrial Melanoma-Specific Oncogenic Noncoding RNA. SAMMSON is specifically expressed in human melanoma and, strikingly, the growth/survival of melanoma is highly dependent on this gene. The conclusions could pave the way for improved diagnostic tools and melanoma treatment.

To assess the importance of specific long non-coding RNA genes in melanomagenesis, P. Mestadgh’s team had performed a large screening to study the expression of numerous lncRNAs across different cancer types. This screening identified SAMMSON as a melanoma-specific lncRNA.

SAMMSON is a melanoma-specific lncRNA.
We confirmed that SAMMSON is expressed specifically in more than 90% of human melanoma clinical samples, but not in benign lesions. Strikingly, when reducing SAMMSON's expression, melanoma cells rapidly and massively died off, irrespective of their mutational status. We have also discovered that SAMMSON is recruited to mitochondria, an organelle that provides energy to the cancer cells, and provided evidence that silencing SAMMSON disrupts vital mitochondrial activities.
Strikingly, when reducing SAMMSON’s expression [using Antisense LNA GapmeRs], melanoma cells rapidly and massively died off.

What made you focus on SAMMSON lncRNA in connection with melanoma?

The observation that SAMMSON is expressed specifically in melanoma was a key initial observation. Moreover, the chromosomal location of the SAMMSON gene attracted our attention. Indeed, the gene is located immediately downstream of MITF, which encodes a transcription factor contributing to the establishment of the melanocytic lineage and known to be overexpressed in melanoma. In fact, an in silico analysis revealed that both genes are invariably co-amplified in about 10–15% of human melanoma. Since our lab has a long-standing interest in dissecting the contribution of the non-coding portion of the genome in the etiology of melanoma we decided to initiate a project aimed addressing the role of this particular lncRNA in the etiology of the disease.

You show that expression of SAMMSON in cancer is exclusive to melanoma. How is this cell type specific expression pattern achieved, and do you know if SAMMSON is expressed during development or in certain types of adult normal tissues?

The lineage-specific expression profile of SAMMSON is indeed one interesting feature that attracted our attention (see above). Note that many lncRNA genes, just like SAMMSON, have very restricted expression profiles. We observed that during melanomagenesis SAMMSON expression was specifically upregulated during the benign to malignant transition: intriguingly SAMMSON was undetectable in normal melanocytes and in all of the human adult tissues we looked at. Interestingly, SAMMSON is expressed in embryonic/fetal melanoblasts, raising the possibility that its primary physiological function is linked to the biology of these cells. Mechanistically, SAMMSON expression in human melanoma is a consequence of gene amplification in about 10–15% of the cases. It is still unclear what drives its expression in the remaining cases and in melanoblasts. Although more work is clearly needed to elucidate this enigma, a first hint came from our observation that the transcription factor SOX10 is a necessary, but not sufficient, driver of SAMMSON expression in melanoma.

What were the first experiments that suggested to you that SAMMSON plays a pivotal role in melanoma?

The observation that knocking-down SAMMSON using Antisense LNA GapmeRs kills virtually all of the melanoma cell lines and primary cultures expressing SAMMSON really triggered our interest and excitement about this project. This experiment established that melanoma cells are addicted to SAMMSON’s expression for the In vitro growth and survival. We later confirmed that SAMMSON targeting decreases melanoma growth in an in vivo setting, i.e. melanoma patient-derived xenograft models.

You performed experiments where you ectopically expressed versions of SAMMSON in which the site targeted by the Antisense LNA GapmeR was mutated to make SAMMSON resistant to silencing by the LNA GapmeR. Could you explain the logic behind these experiments and what they achieved?

We indeed believe that this is one of the critical experiments required to ascertain that the observed anti-tumor effect is not a consequence of an off-target effect. This experiment nicely complements other control experiments we also performed, such as the ones showing no growth inhibition upon addition of Antisense LNA GapmeR to cells that do not express SAMMSON, and the use of multiple Antisense LNA GapmeRs targeting distinct target sites.

So after loss of function analysis, how did you go about characterizing SAMMSON function?

The first classical experiment in the field is to determine the cellular localization of the lncRNA of interest. We performed multiplexed RNA-FISH and cell fractionation experiments and provided evidence that SAMMSON is predominantly located in the cytoplasm, a large fraction of which being associated with mitochondria.

The other, more challenging, experiment consisted of pulling-down endogenous SAMMSON using a pool of antisense biotinylated probes, and identifying putative SAMMSON’s protein partners by mass spectrometry. Among the top candidates, p32 attracted our attention because this protein is a well-known master regulator of mitochondria metabolism.

What is the significance of the SAMMSON–p32 interaction?

We believe that the ability of SAMMSON to regulate p32 cellular localization is key to the addiction of melanoma cells to SAMMSON’s expression. Our data points to a model in which melanoma cells reprogram their mitochondrial metabolism in order to cope with an increasing demand in energy and/or oncogenic-induced stress. They do so partly by upregulating p32 levels and evolving mechanisms that increase the efficiency of p32 mitochondrial import. SAMMSON's upregulation is one such mechanism. Knockdown of SAMMSON significantly decreases p32 mitochondrial abundance and ultimately affects mitochondrial metabolism and biogenesis.

You tested the therapeutic potential of SAMMSON targeting in a mouse model. Can briefly describe these experiments and the results?

We used melanoma patient-derived xenograft models established in house. We injected intra-tumorously or intravenously SAMMSON-targeting Antisense LNA GapmeRs (or control/non-targeting LNA GapmeRs) in cohorts of mice, and measured tumor growth over time. Tumor growth was significantly inhibited in mice injected with the SAMMSON-targeting Antisense LNA GapmeRs compared to the controls. Importantly, we showed that this decrease in tumor growth was accompanied by a decrease in SAMMSON’s expression, an increase in apoptotic cell death and gene expression signatures compatible with induction of mitochondrial defects. We also provided evidence that SAMMSON-targeting Antisense LNA GapmeRs very significantly enhanced the anti-tumor effects of a clinically-relevant BRAFV600E-inhibitor. Together these data identify SAMMSON as an attractive target for combination melanoma targeted therapy.

Tumor growth was significantly inhibited in mice injected with the SAMMSON-targeting Antisense LNA GapmeRs.

You used Antisense LNA GapmeRs to achieve knockdown of SAMMSON in cell cultures and in mice. What has been your experience so far with Antisense LNA GapmeRs?

Our experience is very positive indeed. We have been able to achieve robust knockdown efficiencies with these tools. In the context of the SAMMSON project, the knockdown efficiency was in fact far more impressive with Antisense LNA GapmeRs than when using an siRNA-based approach. We have been particularly impressed by the efficiency of targeting using Antisense LNA GapmeRs in an in vivo context (see above) and therefore would recommend to use these tools for preclinical studies such as the one described in our recent paper.

The knockdown efficiency was in fact far more impressive with Antisense LNA GapmeRs than when using siRNA.

In your opinion what are the advantages of Antisense LNA GapmeRs over RNAi-based methods and Crispr-Cas9 technology when it comes to loss of function analysis of lncRNAs, and how might these technologies complement each other?

We do believe that these approaches complement each other well, and whenever possible they should be used in parallel. All three approaches have their pro and cons. Crispr-Cas9-mediated knockout approaches allow the generation of full knockout/loss of function, which can be advantageous in specific cellular contexts. In the case of the SAMMSON project knocking out SAMMSON was simply not an option, as cells rely on SAMMSON for survival. It should also be mentioned that knocking out the function of a lncRNA is not as simply as knocking out the function of a protein-coding gene using the Crispr-Cas9 approach for the following reasons.

Targeting the locus of a lncRNA with a single gRNA is not an option, since small insertion/deletions are unlikely to have much effect on lncRNA function. Deletion of the entire lncRNA locus is a possibility, but this approach is not always advisable, as it may for instance affect the expression of neighboring genes. Introducing a transcriptional STOP cassette in front of the lncRNA locus is one of the preferred options, but this is still a relatively technically challenging and time-consuming one. Knocking down lncRNA expression using siRNA-approaches is an alternative and valid option, however the knockdown efficiency is not always good enough, especially for nuclear lncRNAs, most of which are poorly targeted by this approach. Antisense LNA GapmeRs instead often promote efficient silencing of lncRNAs, including those that are exclusively nuclear. No matter what approach is chosen, many controls should be performed to rule out potential confounding off-target effects.

We have been particularly impressed by the efficiency of targeting using Antisense LNA GapmeRs in vivo and would recommend them for pre-clinical studies.
Reference
Leucci et al., Melanoma addiction to the long non-coding RNA SAMMSON. Nature. 2016. 531(7595): 518-22. PMID: 27008969.