Super-enhancer analysis reveals subtypes in AML and MDS

Share :
Published: 12 Jun 2016
Views: 2734
Rating:
Save
Dr Michael McKeown - Syros Pharmaceuticals, Cambridge, USA

Dr McKeown talks to ecancertv at EHA 2016 about super-enhancers, gene regulators found within non-coding DNA, which can be used to identify acute myeloid leukaemia and myeloid dysplastic syndrome patient subgroups based on retanoic acid receptor (RARα) expression.

He describes the process of sequencing and interrogating patient genomes to find these markers, the value of primary samples, and how these results have informed the design and approval of SY-1425, a super-enhancer associated therapy.

ecancer's filming at EHA 2016 has been kindly supported by Amgen through the ECMS Foundation. ecancer is editorially independent and there is no influence over content.

 

EHA 2016

Super-enhancer analysis reveals subtypes in AML and MDS

Dr Michael McKeown - Syros Pharmaceuticals, Cambridge, USA


Our company does research on defining the positions of these things called enhancers which are gene regulatory elements in the non-coding regions of the DNA. So you’ve got so many genes but then there’s a huge amount of space that’s not the coding region and so there’s these regulatory elements that are called enhancers. But among these enhancers there’s a small subset of these, about 1-5% that are quite a bit larger than the rest. So if you imagine a plot of it, it’s like an exponential increase for these very large ones and we call those super-enhancers. So the position of these super-enhancers near genes that they’re driving tells us what’s extremely important for the identity of that particular cell. Now if that’s a cancer cell that’s then flagging some of the key oncogenes, classics like Myc, BCL-2, oestrogen receptor, HER2, MIB, so the who’s who list of classic oncogenes. So for my research we were performing this analysis in a cohort of 66 AML primary patient samples in collaboration with the Majeti Lab at Stanford. So when we performed this profile and we found that a subset of these had a very large enhancer driving RARα so that flagged that particular gene as potentially very important for those cells and for this non-APL AML.

When it comes to these subtypes you’ve got this alpha there, are there any others similarly associated?

It’s called RARα, it’s actually a family of proteins so there’s a RARβ and a RARγ. So RARβ tends to be silenced or very lowly expressed in cancer and RARγ has some potential skin toxicity, it’s been implicated in skin tox or there are some publications saying it might help stem cell renewal. So in fact we’re trying not to target those and so the compound that we’re developing, SY-1425, is actually a potent and selective RARα agonist as opposed to the other family members.

When it comes to that potency what kind of sensitivity are you finding?

Retinoic acid receptors naturally bind to something called retinoic acid but our synthetic compound, SY-1425, is about tenfold more potent than that. So we are getting into the sub-nanomolar range and then really the presence of this superenhancer that I was talking about is predictive for the sensitivity in our preclinical models which are cell lines or patient derived xenograft mouse models. So in the cell line, for instance, where we can get an EC50, which is the 50% concentration of an anti-proliferative effect, we see that about a single digit nanomolar range for the ones that have the biomarker but not for the ones that don’t.

Could you just talk us through how you are generating the sequencing to identify the markers?

We perform primarily ChIP-seq which is a chromatin immunoprecipitation with high throughput sequencing. Typically we look for the histone-3 lysine-27 acetylation mark which is a mark for these active enhancer areas. So we do this analysis genome wide and we get the positions of all these enhancers across the entire genome. We then perform an analysis to find out which ones are super-enhancers and then we have a bioinformatics scene that links those to the target genes that are then implicated as potential targets or the critical regulatory system for that cell. So ChIP-seq. Then we usually do a correlated RNA-seq or array based expression analysis.

Taking a step back, big data as a concept has been incorporated into lots of therapies and sequencing is playing a part in that. Do you see the two of those coming together in future trials?

In future drugs? Yes, our genomics platform is that we’re starting with the primary patient samples because cell lines are good models for testing hypotheses in the lab but they just aren’t quite the same as the primary disease after they’ve been in a plastic dish for decades, you can imagine. So we try to start with the primary samples so the idea is that if we get a collection of those, like the 66 I mentioned in AML, we can do this ChIP-seq genomic mapping across those, pair that with expression and then get a much better understanding of what the real critical targets are. So there are thousands and thousands of genes you could go after but we want to be able to know which ones are critical and in particular in subsets. So this serves as both a discovery tool and a potential biomarker for when it’s important contextually.

Personalised therapy from an actual person.

That’s right and so for this SY-1425 drug that we’re developing we’ve actually just gotten approval in the US, IND approval, to proceed into a phase II clinical study. So we’re using a biomarker that was developed to be highly correlated with the super-enhancer, so a biomarker that reports on the presence of the super-enhancer, to select the patients that we believe may be most likely to benefit from the treatment. So very, very much genomics coupled with personalised medicine.