Analysis of long and short enhancers in melanoma cell states

  1. David Mauduit
  2. Ibrahim Ihsan Taskiran
  3. Liesbeth Minnoye
  4. Maxime de Waegeneer
  5. Valerie Christiaens
  6. Gert Hulselmans
  7. Jonas Demeulemeester
  8. Jasper Wouters
  9. Stein Aerts

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This is a well-executed study describing multiple new findings related to the association between enhancer activity as measured by massively parallel reporter assay (MPRA), chromatin features, and underlying TF binding profiles in the context of different transcriptional states of cutaneous melanoma. There are several technical and conceptual advances including the refinement of MPRA assays and the understanding of regulatory mechanisms controlling cell state-specific enhancer activity that will be valuable for other investigators to adapt the experimental design and data analysis strategies. This work will be of broad interest to those seeking to understand cell type- or cell identify-specific gene regulation at the level of transcription and epigenetic control using these and related approaches.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Understanding how enhancers drive cell-type specificity and efficiently identifying them is essential for the development of innovative therapeutic strategies. In melanoma, the melanocytic (MEL) and the mesenchymal-like (MES) states present themselves with different responses to therapy, making the identification of specific enhancers highly relevant. Using massively parallel reporter assays (MPRAs) in a panel of patient-derived melanoma lines (MM lines), we set to identify and decipher melanoma enhancers by first focusing on regions with state-specific H3K27 acetylation close to differentially expressed genes. An in-depth evaluation of those regions was then pursued by investigating the activity of overlapping ATAC-seq peaks along with a full tiling of the acetylated regions with 190 bp sequences. Activity was observed in more than 60% of the selected regions, and we were able to precisely locate the active enhancers within ATAC-seq peaks. Comparison of sequence content with activity, using the deep learning model DeepMEL2, revealed that AP-1 alone is responsible for the MES enhancer activity. In contrast, SOX10 and MITF both influence MEL enhancer function with SOX10 being required to achieve high levels of activity. Overall, our MPRAs shed light on the relationship between long and short sequences in terms of their sequence content, enhancer activity, and specificity across melanoma cell states.

Article activity feed

  1. Version published to 10.7554/elife.71735 on eLife
  2. eLife

    Evaluation Summary:

    This is a well-executed study describing multiple new findings related to the association between enhancer activity as measured by massively parallel reporter assay (MPRA), chromatin features, and underlying TF binding profiles in the context of different transcriptional states of cutaneous melanoma. There are several technical and conceptual advances including the refinement of MPRA assays and the understanding of regulatory mechanisms controlling cell state-specific enhancer activity that will be valuable for other investigators to adapt the experimental design and data analysis strategies. This work will be of broad interest to those seeking to understand cell type- or cell identify-specific gene regulation at the level of transcription and epigenetic control using these and related approaches.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  3. eLife

    Reviewer #1 (Public Review):

    In this manuscript, Mauduit et al described the comprehensive analysis of a panel of cell state-specific cis-regulatory elements (CREs) by massively parallel reporter assay (MPRA) using libraries with different resolution, including H3K27ac ChIP-seq, ATAC-seq and tiling based design. The authors first identified a set of differentially enriched CREs as candidate enhancers by comparing H3K27ac ChIP-seq, ATAC-seq, and gene expression in melanoma cells representing melanocytic (MEL) and mesenchymal-like (MES) states. They next evaluated enhancer activity of selected regions by MPRA in multiple MEL, MES and intermediate state melanoma cells, and uncovered that more than 60% of the selected regions harboring enhancer activity are within ATAC-seq peaks. They further used the deep learning model DeepMEL2 to discover candidate transcription factors (TFs) responsible for cell state-specific enhancer activity, and identified SOX-MITF and AP1 as the major TFs required for enhancer function in MEL and MES cells, respectively.

    Overall, this is an important and well-executed study describing multiple new findings related to the association between enhancer activity as measured by MPRA, chromatin features, and underlying TF binding signals. There are several technical advances in the refinement of MPRA assays that will be valuable to other investigators to adapt the experimental design and data analysis strategies. The results in this study not only confirmed several established concepts in enhancer biology (e.g. ATAC-seq peak is more predictive of enhancer activity than H3K27ac), but also uncovered several new insights into mechanisms controlling cell state-specific enhancer function. The integrative analysis of multiple data types and the inference of the regulatory logics therein are well performed, and the data are of high quality. There are several remaining questions, including how MES and MEL-specific regions were selected, the nomenclature, the correlation with TF binding signals, and discussion/clarification points, that if addressed, will further enhance the strength of conclusions and overall impact of this manuscript.

  4. eLife

    Reviewer #2 (Public Review):

    Mauduit et al aimed to gain mechanistic insights into the enhancer regulatory circuits underlying different cellular states of melanoma cells. To this aim, they utilized high-throughput enhancer reporter assays in multiple primary melanoma cell lines to identify cell state specific enhancers, and by transcription factor (TF) binding sites discovery, they further explored the regulator mechanism behind the activity and specificity of cell state specific enhancers. Their results indicated that differentially expressed TFs, specifically, AP-1 in the mesenchymal-like cancer cell and SOX and MITF in the melanocytic cell, were controlling cell state specific enhancers. These massive and systematic efforts could be of great interest to understanding pathological mechanisms underlying melanoma progression and provide a rich resource of the design, experimental setup, and interpretation of enhancer reporter assay to the general study of enhancer biology.

  5. eLife

    Reviewer #3 (Public Review):

    Mauduit et al. apply leading edge analysis with multiple levels of MPRA analysis of melanoma-state specific enhancer regions guided by both ChIP- and ATAC-seq data and melded with AI-type predictions of transcription factor function. Overall, they find that chromatin regions marked by H3K27Ac (putatively enhancers) and open chromatin regions (less predictive) function as enhancers in MPRAs using large (1.2 to 2.9 kb sequences), middle-sized (501 bp), and small tiled 190 bp analyses. Broadly speaking, MEL-state specific enhancers report stronger activity in more MEL-like cell lines, with correspondingly similar findings for MES-state enhancers and MES-like cells, albeit with lower MES activity in melanoma lines with less-defined states. Particularly interesting is the relationship of increased SOX motifs in synthetic contexts to increased activity in MEL lines, and the central role of AP-1 sites in MES-type enhancers.

    Overall, the data are high quality, the conclusions are well-supported, and the study contributes to a better understanding of how to use and interpret approaches studying specific enhancers, or parts of enhancers, or even specific TF binding sites.

    Strengths:

    1. The use of multiple sizes of MPRA test sequences including the use of overlapping, tiled short sequences is particularly interesting and valuable. In addition, the testing of multiple MRPA backbones adds to generalizability of the findings, as does the analysis across different cell lines. Harkening back even to gel-shift assays/EMSAs, there is always that tension of focused analysis on a defined, short sequence vs the more technically hard-to-produce/expensive (but potentially more "chromatin-like") larger sequence, and this study will provide a nice guide for how we think about these competing aspects. The example of the proposed ZEB-1 repressor interaction with AP-1 activator activity (page 16, line 427) is particularly interesting, and again highlights the importance of context for multiple TF binding sites.

    2. The study also adds additional context to the use of TF prediction algorithms with the incorporation of the DeepMEL2 tool previously developed by the authors, which indicates such tools can be useful in highlighting regions of interest as candidate gene regulatory elements.

    3. The type of data presented is necessarily very dense, and overall, the data are well-displayed to incorporate the multiple layers of -omics data from gene structure, ChIP-seq, ATAC-Seq, MPRA design, etc., with some minor readability issues.

    Weaknesses:

    1. The major weakness of the study is one shared by all approaches using isolated stretches of putative enhancer DNA in a plasmid or reporter vector. Even if one posits some level of chromatinization of the vectors (as noted in the Discussion, page 20), the reporter assay relies on a somewhat artificial context.

    2. While the incorporation of the DeepMEL2 analysis is interesting, it seems hard to fully know the generalizability of such a deep learning tool in the sense that it was trained/developed on melanoma data (melanoma cells). Perhaps this is a strength in its specificity/ability to work in predicting melanoma-specific regulatory networks (or look at how are MES vs MEL states are regulated), but developing or applying an in silico tool with good predictive utility for enhancer function in different cell state contexts is not addressed.

  6. Version published to 10.1101/2021.07.27.453936v1 on bioRxiv