Read on Twitter
Our manuscript ‘Epigenetic therapy induces transcription of inverted SINEs and ADAR1 dependency’ is out today @nature. Congratulations to first authors @Pari_Mehdipour @sajidMarhon and @IliasEttayebi for this amazing work! https://www.nature.com/articles/s41586-020-2844-1
We previously identified that DNA demethylating drugs are able to induce endogenous dsRNAs and activation of dsRNA pattern recognition receptors, leading to an anti-viral-like interferon response. We named this ‘viral mimicry’.
We always suspected these dsRNAs originated from transposable elements (TEs), but we didn’t know the exactly nature of those drug induced immunogenic dsRNA. @CellCellPress https://www.cell.com/fulltext/S0092-8674(15)00972-1
Initially we thought this response was unique to DNA demethylating drugs, but them in collaboration with other groups we found that targeting many other chromatin regulators leads to the same dsRNA response and loss of cancer cell fitness and increase anti-tumor immunity
For example, targeting LSD1 induce viral mimicry. @CellCellPress https://www.cell.com/cell/fulltext/S0092-8674(18)30715-3 Similarly, targeting EZH2 in chemotherapy resistant breast cancer cells leads to TE derepression, dsRNA formation and viral mimicry response. @CD_AACR https://cancerdiscovery.aacrjournals.org/content/10/9/1312.abstract
Many other groups also found that other cancer therapies lead to TE derepression, dsRNA formation and viral mimicry response. For example, CDK4/6i. @nature https://www.nature.com/articles/nature23465
Mutations in chromatin regulators also leads to a baseline state of viral mimicry and create vulnerabilities in cancer cells. For example, H3K27M gliomas have pervasive H3K27 acetylation and TE derepression and dsRNA formation. @Cancer_Cell https://www.cell.com/cancer-cell/pdf/S1535-6108(19)30199-0.pdf
Similar results were observed in SMARCB1 mutant in Rhabdoid Tumors @Cancer_Cell https://www.cell.com/cancer-cell/pdf/S1535-6108(19)30482-9.pdf and early this week it was shown that some PDAC subgroups have hypomethylated TEs at baseline, leading to dsRNA and interferon response @CD_AACR https://cancerdiscovery.aacrjournals.org/content/early/2020/10/09/2159-8290.CD-20-1202
Altogether, these papers suggest a broad role to TEs to alert the cell of chromatin dysregulation. This can be caused by therapy targeting chromatin regulators (‘epigenetic therapy’), mutations in chromatin regulators or other process during tumorigenesis. @CharlesIshak
In general, we believe TE derepression and viral mimicry response acts as an ‘epigenetic checkpoint’ or ‘proof-reading’ mechanism to prevent chromatin dysregulation. @trendscancer @AnnualReviews https://www.cell.com/trends/cancer/fulltext/S2405-8033(18)30130-4 https://www.annualreviews.org/doi/abs/10.1146/annurev-cancerbio-030419-033525
However, none of these previous publications identified the exactly source of endogenous dsRNAs able to activate the pattern recognition receptor.
To solve this, we used a MDA5 protection assay and found that most endogenous MDA5 agonists induced by epigenetic therapy were inverted repeat Alu (IR-Alu) retroelements. These are 2 copies of Alus close together in an inverted orientation.
Most IR-alus were unidirectionally transcribed (either sense or anti-sense), suggesting an intramolecular stem–loop model for dsRNA formation, rather than the previously suggested bi-directional transcription.
Many of those immunogenic IR-Alus were downstream to orphan CpG Islands and gain activation marks after demethylation, suggesting a mechanism of regulation for derepression. It also suggests a potential selective pressure for these orphan CGIs to be preserved during evolution.
IR-Alu stem-loops are known substrates of A-to-I editing by ADAR1, disrupting the dsRNA conformation. Also, ADAR1p150 is a known interferon induced gene. We hypothesized that ADAR1 induction upon viral mimicry would act as a negative feedback loop, restricting the response
Indeed, when we depleted ADAR1, cancer cells become markedly more vulnerable to epigenetic therapy, both in vitro and in vivo. Similarly, ADAR1 depletion allows stronger interferon response by CDK4/6 inhibition.
You can read the full article for free here: https://rdcu.be/b8PT0
Late last year, a couple of papers showed that some cancer cells are ADAR1 dependent, and this dependency correlated with intrinsic IFN signalling. Our results suggest a potential underlying mechanism for ADAR dependency. https://www.nature.com/articles/s41591-018-0302-5 and https://www.nature.com/articles/s41467-018-07824-4
