Shedding light on the long-range interaction of the human epigenome using ultra-long nanopore sequencing

Roham began his presentation by providing an overview of the role of promoter-enhancer interactions in transcriptional regulation. Briefly, the gene promoter directs RNA polymerase to the transcription start site, whereas enhancers are gene expression regulatory regions that can be up to 1 Mb upstream or downstream from the transcription start site. Roham added that he is interested in the methylation status of these regulatory regions. He explained that methylation of the promoter confers gene silencing, whereas un unmethylated promoter confers transcriptional activity. Enhancer methylation, on the other hand, results in a more nuanced regulation of gene silencing. Roham noted that there currently a few ways of analysing promoter-enhancer interactions and methylation status simultaneously (e.g. methyl-HiC), however these methods are caveated with the fact they cannot quantitatively look at interactions and methylation status at the single-cell level. What’s more they are unable to distinguish between allele specific interactions, and Roham believes that ‘nanopore sequencing could address both of these concerns.’ To that end, Roham’s aims were to use ultra-long nanopore sequencing reads to capture these long-range promoter-enhancer interactions with methylation status on a single-read level.

Roham used the newly developed ultra-long kits from Circulomics and Oxford Nanopore and outlined the main steps to the protocol. Roham said ‘we can typically create more than 100 Gb of data with an N50 of around 100 kb on a single PromethION flow cell’. He then pointed out his data on the colorectal cancer cell line HCT-116, where he obtained 40X coverage of data with an N50 of 98 kb, and a read exceeding 2 Mb. Roham then moved on to discuss his analysis on allele specific methylation status. He used the tool megalodon for robust methylation detection and PEPPER to phase the methylation calls into their respective haplotypes. He demonstrated this by displaying the phased methylation status of the gene TBX18 – a putative driver of cancer progression. He pointed out the discrepancy in enhancer and promoter methylation between the two haplotypes culminating in their respective gene silencing and gene activation. Roham emphasised the utility of ultra-long nanopore reads, saying ‘we can look at methylation status across a whole gene body and their regulatory elements’.

Roham then moved onto capturing more than one interaction between the promoter and enhancers. He focused on the gene CCN1 (CYR61), which when upregulated correlates with poor prognosis in colorectal cancer. To get upregulated, the CCN1 promoter interacts with its 3 enhancers downstream of the gene body. Roham stated this ultimately leads to cell migration. He then delved deeper into the enhancer regions, pointing out that current literature asserts that enhancer 3 has the highest activity. He explained that enhancer 2 is inactive, shown by a lack of h3k27 acetylation. These observations led Roham and his team to develop two looping interaction models. In the first model, the promoter interacts with enhancer 3, which Roham noted was the most common interaction observed in these cell lines. In the second model, the promoter interacts with both enhancer 1 and enhancer 3, although this occurs at a much lower frequency than model 1. Roham wanted to emphasise the importance of nanopore ultra-long reads in this context, and to do so, he zoomed in on the methylation status of the enhancer 1 and enhancer 2 which showed clear heterogeneity. Roham stated that he had ‘not been able to capture this information with other methods’ and this suggests that he can use this methylation status heterogeneity to cluster reads based on methylation levels. Roham is optimistic that this will help refine the loop-interaction models and maybe discover some novel biology. Roham noted that he is currently working on clustering tools in order to phase reads based on their methylation status across these large regions.

Roham outlined the key take-home messages to conclude his talk. He hoped he conveyed the usefulness of ultra-long nanopore reads to: investigate allele-specific enhancer-promoter interactions, analyse the relationship between enhancer methylation status and looping interactions, and finally interrogate the heterogeneity of methylation status across regulatory elements. In the future, Roham intends to perform ultra-long pan-epigenome analysis on 30 samples from the human pangenome consortium as well as continue to develop an open-access computational pipeline.