Uncovering cancer mutations in the “Dark Matter” of the human genome

The human genome contains approximately 100,000 long noncoding RNAs (lncRNAs), which is five times more than protein-coding genes. The function of most remains unknown, earning them the name as “Dark Matter of the Genome”. Understanding this Dark Matter, and particularly its role in disease, is one of the great challenges in Biology today. Furthermore, the advent of technologies like antisense oligonucleotides (ASOs) offers a potential route to drugging these lncRNAs for therapy.

To shed light on the role of lncRNAs in cancer, the team of Rory Johnson (University of Bern and University College Dublin, Ireland), along with collaborators, has developed two complementary strategies.

In the first project, the team mined the DNA sequences of thousands of tumours to identify those lncRNAs that carry genetic mutations and hence, may be contributing to cancer development.

The identification of such “driver mutations” is complicated by the vast number of additional “passenger” mutations in the DNA of a typical tumour. To sensitively identify true cancer promoting mutations, researchers developed and validated ExInAtor2, a new analysis pipeline for precisely identifying driver mutations in lncRNAs. Applying this pipeline to data from over six thousand primary and metastatic tumors identified fifty-four lncRNAs harboring potential cancer driver mutations. 

Next, the researchers went to the laboratory to experimentally test the cancer promoting activity of these mutations and their impacted lncRNAs. Overexpression in in vitro tumour models of several lncRNAs promoted cancer cell division. For a liver-cancer-associated lncRNA, downregulation by antisense oligonucleotides (ASOs) decreased cell growth. These results confirm that the driver lncRNAs are capable of promoting cancer growth.

For the same lncRNA, it was found that introducing a single computationally identified driver mutation did not promote cell growth significantly in the model used compared to the unmutated one. However, when four of these stemming from two patients were combined, tumor growth was faster than when expressing the non-mutated lncRNA from plasmids. 

Turning to the lncRNA NEAT1, whose role in cancer pathogenesis was reported in previous publications, the researchers used CRISPR-Cas genome editing to introduce tumour-like mutations into the DNA of living cells. In several models, including a mouse in vivo cancer model, tumor cells and tumors containing NEAT1 mutations outgrew those without mutations in this lncRNA. Further experiments suggest that these mutations act by altering the proteins that interact with NEAT1. This study, published in Nature Communications, represents the first demonstration that cancer-promoting mutations can function by altering the activity of lncRNAs.

In a second study, the researchers have developed high-throughput methods to search for cancer promoting lncRNAs in non-small cell lung cancer (NSCLC), which is the largest cancer killer and remains largely incurable. Starting from cancer genome data, eight hundred lncRNAs were identified as potentially playing a role in NSCLC pathogenesis. A CRISPR-Cas9-based screening system was established to delete transcription start sites of the selected lncRNAs in cell lines derived from two patients. 

The readout of the screen were three cancer cell hallmarks, namely proliferation, metastasis and drug resistance, assessed through five different experimental assays. "The advantage of assessing three different cancer hallmarks is that we have a comprehensive view. However, it also creates the challenge of integrating these data types to come up with a single list of long noncoding RNAs important for non-small cell lung cancer," says Rory Johnson 

To integrate the data from these different readouts and rank the effect of the lncRNAs, an algorithm was developed that identified a list of 80 high-confidence candidates relevant to NSCLC. Out of these, nine were selected for follow-up experiments using two independent ASOs to knock down the respective lncRNA levels. For five out of these nine, loss of cancer cell proliferation with either one of the two ASOs was observed. 

Non-cancerous lung cells, which should not be affected by treatment, either responded barely or not at all to the knockdown of these five lncRNAs. The researchers then focused on the two lncRNAs for which knockdown had the least effect. These were GCAWKR, previously associated with poor colon cancer prognosis, and a second, never implicated in cancer. This second one was renamed by the researchers from its previous systematic name to Cancer Hallmark in Lung LnCNRA 1 (CHiLL1). 

Moving to 3D cell culture models, ASO-mediated knockdown of either lncRNA in lung cancer cells decreased viability, similar to the positive control, mTOR. Likewise, did a CHiLL1 knockdown in patient-cell-derived organoids. "We were amazed and happy to see how well the antisense oligonucleotides could restrain tumor growth in different models," states Taisia Polidori, co-first author of the publication in Cell Genomics. 

Interestingly, when using cocktails of ASOs targeting two or five lncRNAs (keeping the same total ASO as for a single knockdown) resulted in a bigger effect at no increased toxicity. Nearly 80% of KRAS-positive NSCLC tumors analyzed showed expression of either GCAWKR or CHiLL1. Additional experiments showed that GCAWKR and CHiLL1 achieved their effect via different mechanisms. 

These findings led to a patent application and the founding of LiNN Therapeutics in collaboration with venture builder NLC Health. Regarding other cancers, Roberta Esposito, co-first author on both publications says, "Like a telescope that can be quite easily repositioned to study a different part of space, our approach should be easily adaptable to reveal new potential treatment targets for other cancer types." 

Together these studies shed light on fundamental disease mechanisms and open a pathway towards developing new RNA therapeutics targeting the genomic Dark Matter.

Publications:

Esposito, Polidori et al. (2022) Cell Genomics 2(9):100171 (Open Access)
Esposito, Lanzós et al. (2023) Nat Commun 14(1):3342 (Open Access)

Text: Dominik Theler