Scientific Studies Detail Experimental Methods in Epigenetics, Skin Biology, and Cancer Research

Four separate scientific studies have published detailed methodologies covering DNA methylation in sea anemones, multi-species skin tissue analysis, in utero gene manipulation in mice, and human skin multi-omics integration.

The following summary synthesizes the factual procedures and analytical approaches from each study without interpretation or evaluation.

Study 1: DNA Methylation Inhibition in Nematostella vectensis

Research Overview

A study conducted at Queen Mary University of London, approved by the institutional ethics board, examined the role of DNA methylation in the sea anemone Nematostella vectensis. The research employed pharmacological and genetic methods to inhibit methylation during regeneration and growth, and assessed transgenerational inheritance.

Animal Husbandry and Treatment

• Laboratory-reared Nematostella were maintained in 15‰ artificial seawater at 19°C and fed Artemia nauplii.
• Spawning was induced using established protocols.
• Zygotes were treated with cytidine analogues (5-azacytidine, zebularine, decitabine) or GSK-3484862.
• Morpholino antisense oligonucleotides targeting DNMT1, DNMT3, and UHRF1 were microinjected into eggs.

Regeneration and Inheritance Studies

• Polyp oral halves were bisected repeatedly over five rounds; newly regenerated tissues were collected for methylome analysis.
• Gametes from demethylated F0 individuals were used for reciprocal crosses; offspring were sampled at gastrula or polyp stage.

Molecular Analyses

• DNA extraction: Genomic DNA was isolated from embryos, polyps, tissues, sperm, and eggs using commercial kits or phenol-chloroform extraction with mitochondrial depletion for oocytes.
• Oxford Nanopore sequencing: Libraries were prepared at low-passage (30–100 ng) and high-coverage (>30×) levels; methylation was called using Dorado or Guppy with modbam2bed or Modkit.
• Enzymatic methyl-seq (EM-seq): 100 ng gDNA was sheared, libraries were prepared with the NEBNext Enzymatic Methyl-seq Kit, and sequenced on Illumina NovaSeq. Reads were mapped with BS-Seeker2; methylation calling used CGmapTools.
• ATAC-seq: Nuclei from ~100 gastrulae were processed with Tn5 transposase. Libraries were sequenced 2×150 bp on NovaSeq. Reads were mapped with Chromap; peaks called with MACS3; differential accessibility computed with DEseq2.
• RNA-seq: Total RNA was extracted, polyA selected, libraries prepared with NEBNext Ultra II, and sequenced at ~50M paired-end reads. Reads were mapped with HISAT2; counts obtained with TElocal; differential expression with DEseq2.
• CAGE-seq: 4 μg RNA was used per replicate; libraries were prepared per published protocol; sequenced at 70 bp single-end on NextSeq 2000; processed with CAGEr.
• Histone modification analysis: Public ChIP-seq data were remapped; histone signal over DMRs was used in a linear regression model for methylation recovery.
• Haplotype phasing: Variants were called with Clair3 and phased with WhatsHap, using paternal and offspring samples to assign parental haplotypes.

Data Availability

Publicly available WGBS and EM-seq datasets were re-analyzed. Histone modification ChIP-seq data from NCBI GSE46488 and iChIP-seq data from GSE329959 were used.

Study 2: Multi-Species Skin Tissue Collection and Analysis

Tissue Sample Collection

Skin samples were collected from adult and neonatal specimens of the following species:

• Naked mole rats
• Rhesus macaques
• Common marmosets
• Bottlenose, long-beaked common, and short-beaked common dolphins
• North American grizzly bears
• Various mouse strains (WT C57BL/6, K14-Cre;Lef1fl/fl, K14-CreERT;Bmpr1afl/fl, K14-CreERT)
• Humans (gestational and adult)
• Multiple pig types (E90, P3, P10, 6mo, EDA-KO, Yucatan miniature hairless, Hanford miniature, Mangalitsa)

Samples were obtained from anatomical sites including back skin, trunk skin, dorsal rump, digits, and surgical discards. All animal studies followed approved IACUC protocols; human tissue collection followed IRB-approved protocols with informed consent.

Tissue was preserved in 4% PFA or 10% neutral buffered formalin, followed by storage in 70% ethanol or cryopreservation in OCT at -80°C.

Histological Analysis

• Paraffin sections were cut at 5 μm or 10 μm; cryo-sections at 10-15 μm.
• Staining was performed with Hematoxylin and Eosin (H&E) or Herovici’s polychrome.
• Coverslips were mounted using Permount or DPX.
• Imaging used Nikon Eclipse E600, Keyence BZ-X810, Leica cryostat, Olympus FV3000, Leica SP5/SP8 confocal, and Leica DMI8 systems.

Specialized Porcine Studies

• Wound Healing: Neonatal pigs received 2.5 × 2.5 cm full-thickness wounds. Wound size was assessed periodically; tissue was collected at 28, 43, and 58 days post-wounding (dpw).
• BrdU Labeling: Neonatal pigs received intraperitoneal injections of 50 mg/kg BrdU from P5 to P7. Tissues were collected at P8, P12, and P16.

Immunofluorescence Analysis

Cryo-preserved tissues were sectioned at 60 μm or 10 μm; paraffin sections at 10 μm. Standard immunofluorescence protocols were followed, including antigen retrieval for paraffin sections.

Primary antibodies targeted ITGA6, LEF1, αSMA, PDGFRA, KRT10, MKI67, BrdU, SMAD1/5, PECAM1, and PDGFC. Secondary antibodies included Alexa Fluor conjugates; DAPI was used for nuclear counterstaining. Imaging used Leica SP5/SP8 confocal, Leica DMI8, and Olympus FV3000 microscopes.

Image Analysis and Quantification

Quantifications were performed using Fiji ImageJ (v.1.53c) and ggplot (v.3.4.0) in R (v.4.2.2). Metrics included epidermal thickness, rete ridge density, apical ridge length, hair density, scar size, and MKI67+/BrdU+ cell counts. Correlations between hair density and epidermal measurements used linear regression.

Single-Cell RNA Sequencing (scRNA-seq)

Single-cell suspensions were generated from pig skin (E90, P3, P10, and 6 mo) via enzymatic digestion with elastase, hyaluronidase, and collagenase IV. Libraries were prepared with the 10x Genomics scRNA-seq 3′ V3 Kit and sequenced on Illumina NovaSeq PE150.

Data were processed using 10x Genomics Cell Ranger, aligned to the Sscrofa.11.1 genome, and analyzed with the Seurat package in R. Analysis included quality control, normalization (SCTransform), dimensional reduction (UMAP), clustering (SLM algorithm), and cell type annotation based on canonical markers. Pseudotime analysis used Monocle3. Previously published human and mouse scRNA-seq datasets were also reanalyzed.

Stereo-seq Analysis

PFA-fixed frozen cryo skin samples from P3, P10, and 6 mo pigs were sectioned at 10 μm. Stereo-seq used a Complete Genomics T FF v.1.2 kit and DNBSEQ-T7 sequencing. Data processing involved SAW for gene expression matrix generation and alignment. Analysis used Stereopy in Python for quality control, normalization (sctransform), clustering (Leiden algorithm), and spatial cell type assignment.

Cell-Cell Communication Analyses

CellChat and Spatial CellChat were used to infer pathway and ligand–receptor interactions from scRNA-seq and Stereo-seq datasets, respectively, using the human ligand–receptor database. Spatial CellChat constrained interactions to biologically realistic distances.

Statistical Analysis

Analyses were conducted in R (v.4.2.2) using one-way ANOVA with post hoc Tukey’s HSD, Welch’s two-sample t-test, and linear regression. A P-value of < 0.05 was considered statistically significant. No data points were excluded.

Study 3: In Utero Electroporation, Cell Isolation, and Genomic Analyses

In Utero Electroporation (IUE)

IUE procedures were conducted on pregnant CD1 mice at E16.5 following institutional approval. A DNA plasmid cocktail was injected into the lateral ventricles of embryos, followed by six 33-V electric pulses for 55 ms each at 100-ms intervals. Pups were monitored for tumor growth via clinical signs and bi-weekly MRI. End point was determined by neurological symptoms; tissues were collected for nuclei isolation or immunofluorescence.

Cell Isolation Techniques

• RGC Isolation: Retinal ganglion cells were isolated from Ink4a-knockout mice. Cells were cultured in neural basal medium and treated with lentivirus to induce ZR positivity. Single-guide RNAs for Plagl1 were introduced; knockout efficiency was confirmed by deep sequencing.
• OPC Isolation: Primary oligodendrocyte progenitor cell cultures were established from E14.5 mouse cortical tissues. Neural stem cells were cultured as neurospheres, then dissociated and plated. Cells were infected with ZR-Lenti-Cherry or control virus before collection for ATAC-seq and RNA-seq.

Genomic and Sequencing Analyses

• RNA-seq and ATAC-seq: Both datasets were processed using Genialis Expression software. RNA-seq involved read filtering, mapping to Ensembl 92 with STAR, gene expression quantification with FeatureCounts, and differential expression with DESeq2. ATAC-seq involved trimming reads, mapping to mm10 with Bowtie2, and peak calling with MACS2.
• TrackerSeq Library Generation and Validation: An 8-bit barcode was cloned into the pCAG-SacB plasmid. Purified plasmids were electroporated into Endura DUOs; library plasmids were purified and validated by amplification and sequencing.
• CSI DNA Binding Assay: HEK293T cells were transfected with HA-tagged ZR fusion plasmids. Cell lysates were incubated with a double-stranded DNA library containing randomized central regions. The mixture was incubated with anti-HA beads, washed, and amplified via PCR. After three rounds, purified library fragments were barcoded, adapted, and sequenced. Results were sorted and ranked by enrichment.
• Nuclei Isolation: Nuclei were isolated from human tumors and mouse embryonic forebrain using specific protocols with adjusted lysis buffer incubation times. For IUE tumors, a Dounce homogenizer and iodixanol gradient were used. Nuclei were counted and loaded onto a 10x Chromium Chip.

Imaging and Immunological Methods

• RNAscope: End-point ZR tumor-bearing mouse brains were flash-frozen and sectioned. The ACDBio Multiplex Fluorescent v.2 fresh frozen RNAscope protocol was followed with modifications. Specific probes for Zfta, Cdk1, Pax6 were used; images were acquired on a Zeiss LSM 780 microscope.
• Immunofluorescence: IUE surgery was performed with PBCAG-GFP. P2 pup brains were fixed, cryoprotected, sectioned, permeabilized, blocked, and incubated with primary (GFP) and secondary antibodies. DAPI was used for nuclear staining. Immunofluorescence for primary OPCs used antibodies for OLIG2 and HA.

Single-Cell Data Processing and Analysis

• snMultiome Data Processing: The ‘cellranger-arc count’ pipeline was used for human and mouse datasets, followed by quality control with SoupX. High-quality nuclei were retained using Seurat and Signac. Doublets and empty droplets were removed using DoubletFinder and DropletQC.
• snMultiome Data Integration: Open chromatin region peaks were called using MACS2. TF-IDF normalization and LSI dimensionality reduction were performed. Datasets were merged and integrated using FindIntegrationAnchors and IntegrateEmbeddings. For snRNA data, normalization (SCTransform v.2), PCA, and UMAP were performed. Weighted nearest-neighbor (WNN) analysis was used for combined data integration.
• Motif Activity Estimation: Motif/TF chromatin accessibility was calculated for 841 TFs from the JASPAR 2022 database using RunChromVAR in Signac.
• Cell Type Annotation: Cell type annotation used CellTypeEstimate, adapted from ScType, incorporating custom code and a marker database from over 40 published studies. Annotated labels were manually reviewed.
• TrackerSeq Data Processing: Barcode reads were processed to trim flanking sequences; whitelists were generated using UMI-tools. High-confidence barcodes were matched to cells from the Seurat object. Cells were classified as single or multiple LB; single-LB cells were retained.

Advanced Cell Classification and Trajectory Analysis

• Malignant Cell Classification: Genome-wide CNVs were inferred using InferCNV. CNVs were defined as loss (≤0.8) or gain (≥1.2). Malignant cells were estimated as those with large CNV regions (≥100 genes) or a CNV ratio greater than 0.5 per sample.
• ZR Fusion Signal Signatures: A single-cell ZR fusion signature was calculated based on 93 ZR driver genes using AddModuleScore in Seurat. Scores >0.2 indicated confidence; 0.1–0.2 uncertainty; <0.1 non-signal.
• Cell Cycling Signatures: Cell cycle phase was scored using Seurat's CellCycleScoring. A recalculated cell cycle score based on updated markers defined cycling cells (cc.score, S.score, or G2M.score > 0.2) and non-cycling cells (all scores < 0.1).
• Pseudotime Trajectories: Pseudotime trajectories were constructed using Monocle3 and Slingshot. Root nodes were selected based on known progenitor cell types.
• Progenitor and Lineage Scores: Progenitor and lineage scores were calculated for malignant cells. RGC-like or cycling progenitor-like cell types were designated as progenitors.

Study 4: Human Skin Tissue Analysis and Multi-Omics Integration

Study Participant Enrollment and Tissue Collection

De-identified autopsy samples were collected under an approved Mount Sinai IRB protocol (12-00145). Surgical discards for MERFISH profiling from de-identified donors were exempted by the Mount Sinai IRB.

Autopsy donors had a post-mortem interval (PMI) of less than 20 hours and no skin disease history. Additional de-identified skin was obtained from surgical discards during Mohs reconstruction and panniculectomy procedures.

For rapid autopsies, 4-mm punch biopsies were collected from up to 12 body sites per donor and fixed in 10% buffered formalin for 18–24 hours, then transferred to 70% ethanol before paraffin embedding. Tumor-free tissues from Mohs reconstruction were placed in 10 nM Ribonucleoside-Vanadyl Complex in PBS on ice, then sectioned for formalin fixation and paraffin embedding, or embedded in OCT.

Histology

Formalin-fixed paraffin-embedded skin tissue blocks were sectioned at 5 μm, stained with hematoxylin and eosin, and imaged at ×20 or ×40 magnification using a Hamamatsu Nanozoomer S210. Images were analyzed with QuPath (v0.4.0). Epidermal thickness was measured from the basement membrane to the upper stratum corneum boundary; stratum corneum thickness from its basal to superficial boundary.

MERFISH Sample Preparation

Samples were prepared following Vizgen's MERSCOPE user guide (91600112) with modifications for skin tissue. Drying, decrosslinking, and digestion times were adjusted. A concentrated clearing step was added. Photobleaching was extended to 5 hours. A weighted gel embedding step was added before imaging. DAPI staining was extended to 20 minutes.

Human Skin Explants

Full-thickness skin samples were collected from the abdomen of de-identified panniculectomy surgical discard donors (five female donors, ages 32-60). After trimming subcutaneous fat, 8-mm punch biopsies were cultured in triplicates on Surgifoam in 24-well plates with DMEM supplemented with 1% FBS and 1% penicillin–streptomycin, maintaining an air–liquid interface. After 24 hours, media was replaced with recombinant human TNF (10 ng ml−1) or PBS + 0.1% BSA vehicle control for another 24 hours.

RNA FISH

Tissues were prepared using the RNAscope (ACDBio, 323100) protocol for fixed frozen tissue. Fresh skin or explants from two donors were fixed in 4% PFA for 24 hours, subjected to sucrose gradients, and frozen in OCT. Cryostat sections (10 μm) were used. Target RNA transcripts (Hs-CCL19-C1 and Hs-PDGFRA-C2) were detected using RNAscope VIVID dyes. Quantification involved manually counting PDGFRA+/CCL19+ cells per ×20 high-powered field.

Immunofluorescence

Fresh human skin or explants were cryosectioned at 10 μm. Sections were postfixed, washed, and permeabilized. Blocking used 5.0% normal donkey serum, 1% BSA, and 0.3% Triton X in PBS. Sections were incubated overnight at 4°C with anti-activated Caspase-3 (1:100), followed by donkey antirabbit Alexa Fluor 647 secondary antibody (1:400). DAPI was used for nuclear counterstaining.

qPCR

Fresh human abdominal skin and explants (four donors, n=3 replicates) were embedded in OCT and snap-frozen. Approximately 40 frozen sections (16 μm each) were used for RNA isolation using QIAzol Lysis Reagent and purified with the Quick-RNA Microprep Kit. One microgram of total RNA was reverse-transcribed using the iScript cDNA Synthesis Kit. Quantitative PCR used Luna Universal qPCR Master Mix in technical quadruplicates on a LightCycler 480 System. Relative gene expression was determined using the 2−ΔΔCt method.

MERFISH Gene Panel Selection

Gene panels were designed via the Vizgen portal using sun-exposed NS as a reference. Panels included 5-10 canonical cell-type markers per cell type (304 genes, 54.1%), highly expressed L–R pairs (327 genes, 58.2% overlapping with CellChatDB or Omnipath), and 434 overlapping genes (86.2%) for integration.

MERFISH Cell Segmentation

Segmentation was performed with Vizgen postprocessing tool (VPT; v1.3) using Cellpose with DAPI and Cellbound3 stain or DAPI alone. Cellpose/Cellbound3 segmentations were fed into Baysor (v0.7.1) with optimized parameters. Cells with volume <100 μm3 or <10 detected transcripts were removed.

MERFISH Tissue and Compartment Area Calculation

Tissue compartments (dermis, epidermis, subcutis) were manually annotated based on canonical markers (KRT5, KLF5 for epidermis; SOX9 for dermis transition; ADIPOQ for subcutis/adipose tissue). Area calculations used alphashape of cellular centroid coordinates.

MERFISH Clustering and Cell-Type Identification

Global integration used scVI (v1.3.0) with sample barcode as batch variable. Leiden clustering (resolution 1) identified 20 clusters, curated into 18 broad cell types. Cells were grouped into epithelial, stromal, and immune compartments, then subset and reprocessed using Harmony. This resulted in 45 distinct cell types.

Integration of Public scRNA-seq Data

Public scRNA-seq data from 14 studies (93 samples, 85 donors) were downloaded. Cells passing quality control were retained, resulting in 285,887 cells. Data were processed using Seurat (v5) and Harmony for integration. Cells were annotated using a strategy similar to MERFISH.

Neighborhood Identification

Integrated latent space embeddings of the MERFISH dataset were adjusted with scANVI (v1.3.0) before running CellCharter (v0.3.4). The k number of neighborhoods was selected based on local maxima of the stability metric; k=10 was chosen.

Differential Abundance

Cell counts for each cell type or neighborhood were input into crumblr (v0.99.11), a mixed linear model framework for compositional data. Dream (v1.35.5) determined differentially abundant cell types. Donor sex, donor ID, and imaging batch were random effects; donor age and compartment areas were fixed effects.

Spatial Proximity Analysis

CellCharter and squidpy (v1.6.5) nhood_enrichment functions were used to determine spatially proximal cell types.

MERFISH Missing Gene Imputation

Tangram (v1.0.4) imputed missing gene expression in the MERFISH panel using the scRNA-seq dataset as a reference in cluster mode. A joint embedding was calculated using 434 overlapping genes.

L–R Analysis

CellChat (v2.1.2) was applied to the MERFISH dataset per neighborhood to identify spatially prioritized ligand-receptor interactions, restricted to cell pairs within 200 μm. CellChat was also run on scRNA-seq samples from full-thickness biopsies. Overlapping hits were cross-validated.

MERFISH Spatial L–R Coexpression Scoring

L–R expression was visualized using a k-nearest-neighbors smoothing approach. For each cell, the geometric mean of L–R expression was calculated for the centroid cell and its five nearest neighbors within 200 µm spatial distance.

Integration of Public Visium Data

Publicly available Visium data from six studies (81 samples, 63 donors) covering normal/nonlesional skin, BCC, SCC, HS, AD, and psoriasis were downloaded. Data were processed with the Seurat workflow and integrated with Harmony.

Visium cell2location

The scRNA-seq object was filtered for genes among the 10,000 most highly variable in at least ten donors. This object trained the cell2location model. Cell2location (v0.1.4) was run on each Visium tissue sample separately.

Visium Neighborhood Mapping

Differential expression analysis of MERFISH data identified top genes per neighborhood. Seurat's AddModuleScore() calculated a neighborhood score for each Visium spot.

Statistics and Reproducibility

Statistical analyses used R (v4.4.0) and Python (v3.12.9). Reporting includes number of replicates, independent experiments, measures of center/dispersion/precision, statistical tests, and significance. No statistical method predetermined sample size. All computational analyses are provided as code vignettes in the Zenodo repository.