Differentiation‐related epigenomic changes define clinically distinct keratinocyte cancer subclasses
2022; Springer Nature; Volume: 18; Issue: 9 Linguagem: Inglês
10.15252/msb.202211073
ISSN1744-4292
AutoresLlorenç Solé‐Boldo, Günter Raddatz, Julian Gutekunst, Oliver Gilliam, Felix Bormann, Michelle S. Libério, Daniel Hasche, Wiebke Antonopoulos, Jan‐Philipp Mallm, Anke S. Lonsdorf, Manuel Rodríguez‐Paredes, Frank Lyko,
Tópico(s)Cancer Cells and Metastasis
ResumoArticle19 September 2022Open Access Transparent process Differentiation-related epigenomic changes define clinically distinct keratinocyte cancer subclasses Llorenç Solé-Boldo Llorenç Solé-Boldo orcid.org/0000-0002-6974-9066 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Writing - original draft, Writing - review & editing Search for more papers by this author Günter Raddatz Günter Raddatz orcid.org/0000-0002-9125-6890 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis, Validation Search for more papers by this author Julian Gutekunst Julian Gutekunst Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis Search for more papers by this author Oliver Gilliam Oliver Gilliam orcid.org/0000-0002-6277-4293 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis Search for more papers by this author Felix Bormann Felix Bormann orcid.org/0000-0002-5919-3375 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis Search for more papers by this author Michelle S Liberio Michelle S Liberio Single-cell Open Lab, German Cancer Research Center and Bioquant, Heidelberg, Germany Contribution: Methodology Search for more papers by this author Daniel Hasche Daniel Hasche orcid.org/0000-0001-9306-3059 Division of Viral Transformation Mechanisms, German Cancer Research Center, Heidelberg, Germany Contribution: Methodology Search for more papers by this author Wiebke Antonopoulos Wiebke Antonopoulos Tissue Bank of the National Center for Tumor Diseases (NCT), Heidelberg, Germany Institute of Pathology, Heidelberg University Hospital, Heidelberg, Germany Contribution: Resources, Methodology Search for more papers by this author Jan-Philipp Mallm Jan-Philipp Mallm Single-cell Open Lab, German Cancer Research Center and Bioquant, Heidelberg, Germany Division of Chromatin Networks, German Cancer Research Center and Bioquant, Heidelberg, Germany Contribution: Methodology Search for more papers by this author Anke S Lonsdorf Anke S Lonsdorf orcid.org/0000-0003-3582-7238 Department of Dermatology, University Hospital, Ruprecht-Karls University of Heidelberg, Heidelberg, Germany Contribution: Resources, Formal analysis, Investigation Search for more papers by this author Manuel Rodríguez-Paredes Corresponding Author Manuel Rodríguez-Paredes [email protected] orcid.org/0000-0001-5471-4277 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Institute of Toxicology, University Medical Center Mainz, Johannes Gutenberg University, Mainz, Germany Contribution: Conceptualization, Formal analysis, Supervision, Writing - original draft, Project administration, Writing - review & editing Search for more papers by this author Frank Lyko Corresponding Author Frank Lyko [email protected] orcid.org/0000-0002-4873-5431 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review & editing Search for more papers by this author Llorenç Solé-Boldo Llorenç Solé-Boldo orcid.org/0000-0002-6974-9066 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Writing - original draft, Writing - review & editing Search for more papers by this author Günter Raddatz Günter Raddatz orcid.org/0000-0002-9125-6890 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis, Validation Search for more papers by this author Julian Gutekunst Julian Gutekunst Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis Search for more papers by this author Oliver Gilliam Oliver Gilliam orcid.org/0000-0002-6277-4293 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis Search for more papers by this author Felix Bormann Felix Bormann orcid.org/0000-0002-5919-3375 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Software, Formal analysis Search for more papers by this author Michelle S Liberio Michelle S Liberio Single-cell Open Lab, German Cancer Research Center and Bioquant, Heidelberg, Germany Contribution: Methodology Search for more papers by this author Daniel Hasche Daniel Hasche orcid.org/0000-0001-9306-3059 Division of Viral Transformation Mechanisms, German Cancer Research Center, Heidelberg, Germany Contribution: Methodology Search for more papers by this author Wiebke Antonopoulos Wiebke Antonopoulos Tissue Bank of the National Center for Tumor Diseases (NCT), Heidelberg, Germany Institute of Pathology, Heidelberg University Hospital, Heidelberg, Germany Contribution: Resources, Methodology Search for more papers by this author Jan-Philipp Mallm Jan-Philipp Mallm Single-cell Open Lab, German Cancer Research Center and Bioquant, Heidelberg, Germany Division of Chromatin Networks, German Cancer Research Center and Bioquant, Heidelberg, Germany Contribution: Methodology Search for more papers by this author Anke S Lonsdorf Anke S Lonsdorf orcid.org/0000-0003-3582-7238 Department of Dermatology, University Hospital, Ruprecht-Karls University of Heidelberg, Heidelberg, Germany Contribution: Resources, Formal analysis, Investigation Search for more papers by this author Manuel Rodríguez-Paredes Corresponding Author Manuel Rodríguez-Paredes [email protected] orcid.org/0000-0001-5471-4277 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Institute of Toxicology, University Medical Center Mainz, Johannes Gutenberg University, Mainz, Germany Contribution: Conceptualization, Formal analysis, Supervision, Writing - original draft, Project administration, Writing - review & editing Search for more papers by this author Frank Lyko Corresponding Author Frank Lyko [email protected] orcid.org/0000-0002-4873-5431 Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review & editing Search for more papers by this author Author Information Llorenç Solé-Boldo1, Günter Raddatz1, Julian Gutekunst1, Oliver Gilliam1, Felix Bormann1, Michelle S Liberio2, Daniel Hasche3, Wiebke Antonopoulos4,5, Jan-Philipp Mallm2,6, Anke S Lonsdorf7, Manuel Rodríguez-Paredes *,1,8,† and Frank Lyko *,1,† 1Division of Epigenetics, DKFZ-ZMBH Alliance, German Cancer Research Center, Heidelberg, Germany 2Single-cell Open Lab, German Cancer Research Center and Bioquant, Heidelberg, Germany 3Division of Viral Transformation Mechanisms, German Cancer Research Center, Heidelberg, Germany 4Tissue Bank of the National Center for Tumor Diseases (NCT), Heidelberg, Germany 5Institute of Pathology, Heidelberg University Hospital, Heidelberg, Germany 6Division of Chromatin Networks, German Cancer Research Center and Bioquant, Heidelberg, Germany 7Department of Dermatology, University Hospital, Ruprecht-Karls University of Heidelberg, Heidelberg, Germany 8Institute of Toxicology, University Medical Center Mainz, Johannes Gutenberg University, Mainz, Germany † These authors contributed equally to this work *Corresponding author. Tel: +49 6221424631; E-mail: [email protected] *Corresponding author. Tel: +49 6221423800; E-mail: [email protected] Molecular Systems Biology (2022)18:e11073https://doi.org/10.15252/msb.202211073 PDFDownload PDF of article text and main figures.PDF PLUSDownload PDF of article text, main figures, expanded view figures and appendix. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions Figures & Info Abstract Keratinocyte cancers (KC) are the most prevalent malignancies in fair-skinned populations, posing a significant medical and economic burden to health systems. KC originate in the epidermis and mainly comprise basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC). Here, we combined single-cell multi-omics, transcriptomics, and methylomics to investigate the epigenomic dynamics during epidermal differentiation. We identified ~3,800 differentially accessible regions between undifferentiated and differentiated keratinocytes, corresponding to regulatory regions associated with key transcription factors. DNA methylation at these regions defined AK/cSCC subtypes with epidermal stem cell- or keratinocyte-like features. Using cell-type deconvolution tools and integration of bulk and single-cell methylomes, we demonstrate that these subclasses are consistent with distinct cells-of-origin. Further characterization of the phenotypic traits of the subclasses and the study of additional unstratified KC entities uncovered distinct clinical features for the subclasses, linking invasive and metastatic KC cases with undifferentiated cells-of-origin. Our study provides a thorough characterization of the epigenomic dynamics underlying human keratinocyte differentiation and uncovers novel links between KC cells-of-origin and their prognosis. Synopsis An integrated single-cell study reveals that DNA methylation patterns at differentially accessible chromatin regions can stratify a wide range of keratinocyte cancers (KC) into subtypes with distinct cell-of-origin and clinical features. Single-cell multi-omics defines new potential transcription factors involved in human epidermal differentiation, such as MEF2A. DNA methylation at differentially accessible regions between undifferentiated and differentiated keratinocytes defines EpSC-like and keratinocyte-like KC subtypes. Single-cell methylomics and cell type deconvolution of bulk methylomes identify the KC cells-of-origin. Cell-of-origin-based KC subclasses display distinct proliferative and invasive features as well as different metastatic potential. Introduction The epidermis constitutes the first line of defense of the human body against environmental damage. This stratified squamous epithelium is mainly composed of keratinocytes, which arise from epidermal stem cells (EpSCs) located at the basal layer of the epidermis (Gonzales & Fuchs, 2017; Moreci & Lechler, 2020). As EpSCs start differentiating, they detach from the basement membrane and migrate upwards, resulting in distinct differentiated keratinocyte populations (i.e., spinous, granular, and cornified) (Gonzales & Fuchs, 2017; Moreci & Lechler, 2020). Terminally differentiated keratinocytes are continuously desquamated. As such, the homeostatic epidermis is subjected to a constant turnover, which is regulated by a fine-tuned balance between self-renewal and differentiation (Blanpain & Fuchs, 2009). Keratinocyte cancers (KC), also known as non-melanoma skin cancers (NMSC), originate from epidermal keratinocytes. They represent the most common malignancies worldwide in the fair-skinned population, with an incidence 20 times higher than that of melanoma, the other major skin cancer (Apalla et al, 2017b; Fitzmaurice et al, 2019; Stang et al, 2019). The incidence of KC has alarmingly risen over the last decade, with an increase of ~33% in the total number of cases worldwide between 2007 and 2017 (Fitzmaurice et al, 2019). These numbers illustrate why, despite a lower mortality rate, KC are associated with significant morbidity and a heavy burden on public health systems (Mudigonda et al, 2010; Apalla et al, 2017b; Fitzmaurice et al, 2019). Two distinct malignancies account for 99% of all KC: basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC) (Apalla et al, 2017a; Bartoš & Kullová, 2018). Even though cSCC represents only 20% of KC cases, it accounts for the vast majority of deaths associated with such malignancies, as about 5% of the tumors metastasize, with a mortality rate exceeding 70% (Ratushny et al, 2012; Burton et al, 2016). In contrast, the estimated metastatic potential of BCC is less than 0.05% (Apalla et al, 2017a). Most invasive cSCCs arise either from a precancerous dysplasia known as actinic keratosis (AK) or from an in situ carcinoma known as Bowen's disease (BD), with a progression rate of 0.025-16% and 3-5% per year and event, respectively (Ratushny et al, 2012; Burton et al, 2016). However, the molecular mechanisms underlying their progression to invasive cSCC remain largely unknown. DNA methylation is a dynamic epigenetic modification that mainly occurs in the context of CpG dinucleotides at the carbon-5 position of cytosines (Lyko, 2018). Catalyzed by a set of three methyltransferases (DNMT1, DNMT3A, and DNMT3B), it has a strong influence on gene expression and other essential genetic functions (Lyko, 2018). Consequently, DNA methylation is essential for the establishment and maintenance of cellular identity (Lyko, 2018; Greenberg & Bourc'his, 2019). Disruption of normal DNA methylation patterns is currently considered a hallmark of cancer, which presents a characteristic genome-wide hypomethylation and regional hypermethylation (Jones & Baylin, 2007). Importantly, tumor methylomes not only include cancer-specific methylation changes, but also partially maintain the DNA methylation patterns of their tumor-initiating cell (Kulis et al, 2013; Moran et al, 2016). In fact, a systematic study concluded that cell-of-origin-related patterns are the main variable influencing tumor stratification in many tumor entities (Hoadley et al, 2018). Epidermal differentiation has been associated with dynamic changes in DNA methylation. In mice, keratinocyte differentiation was associated with a general loss of DNA methylation at lineage-specific regulatory elements, while methylation gains occurred at regulatory regions of other lineages (Bock et al, 2012). Similarly, a loss of DNA methylation in the promoter region of roughly 60% of the genes induced upon keratinocyte differentiation in vitro has been observed in humans (Sen et al, 2010). In agreement with these findings, we have previously identified two subclasses of AK and cSCC based on their methylation patterns and that we interpreted to arise from keratinocytes at two distinct epidermal differentiation stages: one more closely related to the EpSCs and one to a more differentiated keratinocyte (Rodríguez-Paredes et al, 2018a). However, direct proof for this interpretation has been lacking and the subclasses were not characterized in detail. Here, we performed an integrated analysis of the chromatin dynamics associated with human epidermal differentiation using single-cell multi-omics and transcriptomics approaches. We identified more than 3,800 differentially accessible regions between undifferentiated and terminally differentiated keratinocytes. Further characterization of these regions revealed that they comprised regulatory regions associated with known but also novel epidermal differentiation transcription factors. Tumor stratification based on the DNA methylation patterns found at these differentially accessible regions identified two subtypes of AK and cSCC with EpSC-like and keratinocyte-like features. Importantly, we also show for the first time DNA methylation dynamics in the human epidermis at single-cell resolution, which we studied with single-cell combinatorial indexing for methylation analysis (sci-MET) (Mulqueen et al, 2018), after addressing important shortcomings of the original protocol. The integrative analysis of bulk and single-cell methylation datasets, as well as the use of deconvolution tools based on scRNA-seq, provided direct evidence of the cell-of-origin interpretation of the AK/cSCC subtypes. Furthermore, epigenomic data analyses using a mitotic-like clock and the stratification of an expanded dataset, which included BCC and other yet unstratified epidermal entities, suggested a more invasive phenotype and higher metastatic potential for tumors arising from undifferentiated keratinocytes. All in all, our DNA methylation-based tumor stratification strategy may represent an important advance in the risk assessment of KC patients. Results Single-cell multi-omics analysis of healthy human epidermis To investigate differentiation-related epigenomic changes in the human epidermis at the single-cell level, we used a combination of single-cell multi-omic and transcriptomic approaches. First of all, we generated a single-cell multi-omics (scATAC-seq + scRNA-seq) dataset from two sun-protected healthy epidermis samples (55 and 72 y/o, male). A total of 5,565 cells passed the quality control for both genomic layers and were integrated into a common dataset to avoid batch effects (Fig EV1A). Unsupervised clustering identified 10 cell clusters, which were visualized using a joint uniform manifold approximation and projection (UMAP) representing both gene expression and chromatin accessibility (Fig EV1B). To identify the cell identity of each cluster, we also generated a reference scRNA-seq dataset by combining our own data generated from a sun-protected healthy epidermis sample from a 30 y/o male donor (Appendix Fig S1A and B), with a matching subset of sun-protected healthy epidermis from three donors (Cheng et al, 2018a; Data ref: Cheng et al, 2018b). All four samples were obtained from the trunk area and did not display significant differences. The integrated dataset contained 32,272 high-quality cells and their unsupervised clustering defined 13 cell clusters, which comprised cells from all donors (Appendix Fig S1C and D). These included six archetypical keratinocyte populations: two basal undifferentiated populations, two mitotic clusters, and the well-differentiated spinous and granular keratinocytes (Fig EV2A and B, and Dataset EV1). Highly specialized keratinocyte populations such as channel or pro-inflammatory keratinocytes were also detected (Fig EV2A and B, and Dataset EV1). Lineage inference using RNA velocity analysis was possible with our own dataset and placed the Basal 1 population at the beginning of the differentiation process (Fig EV2C and Appendix Fig S2). The trajectory then progressed to the mitotic keratinocytes and, lastly, to the well-differentiated spinous population (Fig EV2C). Hence, these results suggest that the main EpSC population is contained in the Basal 1 cluster. Figure 1. Single-cell multi-omics characterization of human epidermal differentiation A. Joint UMAP plot depicting both scATAC-seq and scRNA-seq data from 5,355 keratinocytes from sun-protected human epidermis (n = 2). Color depicts the unsupervised clustering (left) as well as cell-type annotation based on the reference scRNA-seq dataset (right). B. Representative examples of chromatin accessibility and gene expression of KRT5 and KRT10, two key epidermal differentiation-related genes, in the main keratinocyte populations. C. Co-accessibility at the human epidermal differentiation complex (EDC) in undifferentiated (cluster 2) and differentiated (cluster 3) keratinocytes. Only connections (arcs) with a co-accessibility score above 0.25 are plotted. Gray boxes below tracks represent scATAC-seq peaks. D. Left: Heatmap displaying the differentially accessible peaks between undifferentiated (cluster 2) and differentiated (cluster 3) keratinocytes. Number of Tn5 insertion sites in each region was scaled by row. Right: DNA sequence motifs for the top six overrepresented transcription factor (TF) motifs in undifferentiated (cluster 2) and differentiated (cluster 3) keratinocyte-specific accessible peaks. E. Heatmaps displaying the predicted top five transcription factors in each cell cluster using chromVAR motif activity (left) and gene expression (right). F. chromVAR deviations (in quantiles) and gene expression for representative enriched TF projected onto the joint UMAP plot of keratinocytes from the single-cell multi-omics dataset. G. Heatmap showing the top five active regulons in each cell cluster, based on the expression data of the multi-omics dataset. TF regulons also identified by other approaches are highlighted. Download figure Download PowerPoint Click here to expand this figure. Figure EV1. Unsupervised clustering of the integrated single-cell multi-omics dataset A–C. Joint UMAP plot depicting both scATAC-seq and scRNA-seq data from 5,565 cells from sun-protected human epidermis (n = 2) after data integration. Coloring is according to donor (A), unsupervised clustering based on gene expression (B), and cell-type annotation based on the reference scRNA-seq dataset (C). Download figure Download PowerPoint Click here to expand this figure. Figure EV2. Single-cell RNA sequencing analysis of the human epidermis A. Uniform manifold approximation and projection (UMAP) plot depicting single-cell transcriptomes from healthy sun-protected human epidermis (n = 4). Each dot represents a single cell (n = 32,272). Colors depict the six archetypical keratinocyte populations described in the text, as well as other minority cell types (Cheng et al, 2018a; Data ref: Cheng et al, 2018b). B. Average expression of the top five gene markers defining each cell population projected on the UMAP plot. Red indicates maximum average gene expression, while blue indicates low or no expression of a particular set of genes in log-normalized UMI counts. C. Left: RNA velocities calculated using the 7,068 keratinocytes from the in-house generated dataset of healthy human epidermis, projected onto the UMAP embedding. Right: UMAP plot displaying the latent time calculated by scVelo. Download figure Download PowerPoint After cell annotation based on the reference scRNA-seq dataset, most keratinocyte populations identified in the scRNA-seq experiment were also detected in the multi-omics dataset (Fig 1A). Of note, our multi-omics analysis showed the expected chromatin accessibility and gene expression dynamics for several established epidermal differentiation markers. For instance, ATAC peaks associated with either the basal keratinocyte gene marker KRT5 or the suprabasal differentiated keratinocyte gene marker KRT10, lost or gained accessibility as they became less or more expressed along the differentiation trajectory, respectively (Fig 1B). Consistently, we observed an increase in co-accessibility in the epidermal differentiation complex (EDC), a genomic region containing multiple genes related to terminal differentiation and cornification (Kypriotou et al, 2012), in terminally differentiated keratinocytes (Spinous and Granular, cluster 3) compared with basal undifferentiated keratinocytes (Basal 1, cluster 2, Fig 1C). To further characterize differentiation-related changes occurring at the chromatin level, we compared the genome accessibility in basal undifferentiated keratinocytes (Basal 1, cluster 2) and in terminally differentiated keratinocytes (Spinous and Granular, cluster 3). This comparison identified 3,838 differentially accessible peaks, of which 1,659 were only accessible in undifferentiated keratinocytes and 2,179 were only accessible in differentiated keratinocytes (Fig 1D, Dataset EV2). Motif enrichment analysis for each set of accessible peaks identified cell-type-specific overrepresentation of transcription factor (TF) binding motifs (Fig 1D). For example, TF-binding motifs associated with key regulators of epidermal stem cell proliferation and differentiation such as TP63 (Soares & Zhou, 2018) and OVOL1 (Lee et al, 2014) were enriched in peaks that were specific to undifferentiated keratinocytes (Fig 1D). In contrast, TF-binding motifs from members of the CEBP family, which are associated with terminal differentiation in keratinocytes (Borrelli et al, 2007; Lopez et al, 2009), were enriched in peaks specific to differentiated keratinocytes (Fig 1D). To refine our multimodal analysis, we then combined motif activity scores calculated using chromVAR and gene expression data in order to identify the transcription factors with specifically enriched expression and motif accessibility in each cell cluster. This identified key regulators of the basal undifferentiated keratinocytes, including MEF2A, TEAD1, IRF1, TP63, and NFKB and key transcription factors of terminally differentiated keratinocytes, including GRHL1, RORA, CEBPA, NR1D1, or SREBF2 (Fig 1E and F). While most of these transcription factors have been previously found to play important roles in epidermal differentiation (Truong et al, 2006; Dai et al, 2013; Gulati et al, 2013; Mlacki et al, 2014; Yuan et al, 2020), MEF2A has not yet been associated with this process. Gene regulatory networks analysis using single-cell regulatory network inference and clustering (SCENIC) (Aibar et al, 2017) on the transcriptomics data of our multi-omics dataset also identified MEF2A and TEAD1 as key transcription factors for undifferentiated keratinocytes, and SREBF2, CEBPA, GRHL1, and NR1D1 for differentiated keratinocytes (Fig 1G). Altogether, our multi-omics data recapitulated known accessibility and gene expression dynamics during epidermal differentiation and identified potential new key regulators, such as MEF2A. DNA methylation at differentially accessible regions defines AK/cSCC subtypes To investigate whether the differentially accessible peaks detected during epidermal differentiation corresponded to regulatory regions, such as gene promoters or enhancers, we made use of published ChIP-seq data for several histone marks generated on normal human epidermal keratinocytes (NHEK). Accessible peaks from basal and differentiated keratinocytes showed no enrichment for the repressive chromatin mark H3K27me3, in agreement with their open state (Fig 2A). On the contrary, differentiated keratinocyte-specific peaks showed a strong correlation with H3K27ac and H3K4me1, two histone marks that are associated with active enhancers (Creyghton et al, 2010) (Fig 2A). Furthermore, undifferentiated keratinocyte-specific peaks were enriched for H3K27ac and H3K4me2/me3, histone marks that are associated with gene promoters and actively transcribed regions (Bernstein et al, 2005; Orford et al, 2008) (Fig 2A). Altogether, our analyses indicate that the differentially accessible regions identified in our scATAC-seq data correspond to regulatory regions associated with key regulators of epidermal differentiation. Figure 2. Epidermal differentiation-specific accessible regions define AK/cSCC subclasses A. Average histone modification profiles of undifferentiated and differentiated keratinocyte-specific peaks using previously published data generated on NHEK cells (ENCODE). The normalized signal of H3K27ac, H3K4me1/me2/me3, and H3K27me3 were measured in a window of ± 10,000 base pairs (bp). B. Fractions of CpGs located within epigenomic substructures for the 4,351 InfiniumEPIC CpG probes found within undifferentiated and differentiated keratinocyte-specific peaks. C. Unsupervised hierarchical clustering of 12 healthy, 20 AK, and 35 cSCC epidermal samples based on the methylation status at undifferentiated and differentiated keratinocyte-specific peaks. Each row represents the average methylation value of all CpGs contained in a particular peak. D. Principal Component Analysis (PCA) of 67 AK/cSCC and healthy controls performed with all detected CpGs after filtering (n = 632,778). Coloring is according to sample type and shape is according to cell-of-origin-related subclass. Data information: AK: actinic keratosis, cSCC: cutaneous squamous cell carcinoma, DK: Differentiated keratinocytes, UK: undifferentiated keratinocytes. Download figure Download PowerPoint DNA methylation cooperates with chromatin accessibility to establish and maintain cellular identity (Guo et al, 2016; Li et al, 2021). Furthermore, DNA methylation patterns at regulatory regions have been used to define the cellular origin of several human cancer types (Kulis et al, 2013; Moran et al, 2016; Hoadley et al, 2018). To assess whether the methylation patterns at the differentially accessible regions between undifferentiated and differentiated keratinocytes would be informative for identifying the cellular origin of epidermal tumors, we extracted the CpGs located in the 3,838 differentially accessible peaks and that can be interrogated with probes on the Infinium EPIC array. This identified 2,925 CpG probes located in undifferentiated keratinocyte-specific peaks and 1,426 CpG probes located in differentiated keratinocyte-specific peaks. These probes covered 914 and 864 peaks accessible exclusively in undifferentiated or differentiated
Referência(s)