ChromSMF: integrated profiling of histone modifications, protein-DNA interactions and DNA methylation on multi-kilobase DNA molecules

The paper introduces ChromSMF, a novel method that utilizes antibody-guided tethering and Nanopore sequencing to simultaneously profile histone modifications, protein-DNA interactions, nucleosome occupancy, and DNA methylation on individual multi-kilobase DNA molecules, thereby enabling the integrated analysis of combinatorial epigenetic regulation across entire cis-regulatory landscapes.

The Gist

Imagine your DNA as a massive, 3-billion-letter instruction manual for building and running a human (or a fly, in this case). For a long time, scientists have tried to read this manual, but they've been looking at it through a very blurry, fragmented lens.

Here's the problem: The manual isn't just text; it's a living, breathing document.

• The Text: The actual DNA sequence (the letters A, C, T, G).
• The Highlights: Chemical "sticky notes" called histone modifications that tell the cell, "Read this part!" or "Ignore this part!"
• The Bookmarks: Transcription factors (proteins) that sit on the text to open or close sections.
• The Locks: DNA methylation that can lock certain pages so they can't be read.

The Old Way:
Previously, scientists had to use different tools to read different parts of the manual. They would use one machine to find the "sticky notes," another to find the "bookmarks," and a third to check the "locks."

• The Analogy: Imagine trying to understand a movie by watching the sound, then stopping the tape to look at the subtitles, then rewinding to check the lighting. You know the movie has sound, text, and light, but you can never see how they work together at the exact same moment on the same scene. You lose the connection between the action and the dialogue.

The New Solution: ChromSMF
The researchers at EMBL Heidelberg have invented a new method called ChromSMF (Chromatin-informed Single Molecule Footprinting). Think of this as a high-tech, magical magnifying glass that can read the entire manual, page by page, while seeing all the highlights, bookmarks, and locks simultaneously on the same piece of paper.

Here is how they did it, using a creative analogy:

The "Tagging" Game

Imagine the DNA is a long, winding road.

1. The "Open Road" Detector (Chromatin Accessibility): First, they spray the road with a special "blue paint" (a chemical enzyme) that only sticks to the parts of the road that are open and clear. If a protein (a bookmark) is sitting on the road, the paint can't reach it. So, the blue paint maps out where the road is open and where it's blocked.
2. The "Highlighter" (Histone Modifications): Next, they use a team of "search-and-rescue" drones (antibodies) that are looking for a specific type of "sticky note" (a histone modification) on the side of the road. When a drone finds a sticky note, it drops a "green paint" (a different chemical enzyme) right next to it.
3. The "Super-Scanner" (Nanopore Sequencing): Finally, they run the entire road through a super-fast scanner (Nanopore sequencing). This scanner is so advanced it can read the DNA letters and see where the blue paint and green paint are sitting on the exact same stretch of road.

Why This is a Game-Changer

Because they are looking at single molecules (one long piece of DNA at a time), they can finally answer questions that were previously impossible:

• The "Co-Occurrence" Question: "Does the 'Read Me' sticky note (H3K4me3) always appear on the same DNA molecule that has an open road (accessible chromatin), or do they sometimes appear separately?"

  • Old way: "Hey, 50% of the time we see the note, and 50% of the time we see the open road. They must be related!"
  • ChromSMF way: "Ah, I see that on this specific molecule, the note and the open road are together. But on that other molecule, the note is there, but the road is blocked. They aren't always linked!"

• The "Long-Distance" Connection: DNA is long. A sticky note on an enhancer (a remote control) might be 10,000 letters away from the gene it controls.

  • ChromSMF can read a 5,000-letter chunk of DNA at once. It can see: "Oh, this green sticky note is on the left side of this molecule, and the road is open on the right side. They are connected on the same physical strand!" This reveals how distant parts of the genome talk to each other.

• The "Twin" Problem (Alleles): Humans have two copies of every gene (one from mom, one from dad). Sometimes, the "mom" copy is active and the "dad" copy is silent.

  • ChromSMF can tell the difference between the two copies (like distinguishing between identical twins by a tiny birthmark). It can show: "On the 'mom' copy, the road is locked and the sticky note is missing. On the 'dad' copy, the road is wide open and the sticky note is present." This helps explain why we get certain diseases or traits.

The Bottom Line

Before ChromSMF, scientists were like detectives trying to solve a crime by interviewing witnesses separately and hoping their stories matched up. They knew the facts, but they missed the context.

ChromSMF is like putting all the witnesses in the same room and watching the crime happen in real-time. It allows scientists to see the combinatorial function—how the genetic code, the chemical tags, and the physical structure of DNA work together as a single, coordinated team to control life.

This isn't just about reading a book; it's about understanding the story the book is telling, with all its plot twists, hidden clues, and character interactions, all at once.

Technical Summary

1. Problem Statement

Epigenetic regulation involves a complex interplay between histone modifications (HMs), chromatin accessibility, transcription factor (TF) binding, and DNA methylation. Traditionally, these features are measured using separate bulk assays (e.g., ChIP-seq for HMs, ATAC-seq/DNase-seq for accessibility, bisulfite sequencing for methylation).

• Limitations of Current Methods:
  • Loss of Co-occurrence: Bulk assays average signals across cell populations, losing the ability to determine if specific combinations of marks exist on the same DNA molecule (haplotype).
  • Correlation vs. Causation: It is difficult to establish direct functional links between a specific histone mark and chromatin opening or TF binding because they are not measured simultaneously on the same molecule.
  • Genetic Context: Existing methods struggle to link epigenetic states to underlying genetic variations (SNPs) or haplotypes without complex phasing or enrichment steps.
  • Resolution: Short-read methods cannot resolve long-range interactions (kilobases) between regulatory elements on a single molecule.

2. Methodology: ChromSMF

The authors developed Chromatin-informed Single Molecule Footprinting (ChromSMF), a long-read, single-molecule genomic approach that integrates five layers of information on the same DNA molecule:

1. Histone Modifications (via antibody-guided adenine methylation).
2. Chromatin Accessibility (via cytosine methylation footprinting).
3. Transcription Factor Binding (via cytosine methylation footprinting).
4. Endogenous DNA Methylation (5mC).
5. Genetic Variation (SNPs).

Experimental Workflow:

• Cell Permeabilization: Cells are immobilized on Concanavalin A beads and chemically permeabilized.
• Step 1: Chromatin Footprinting (mC): Cells are incubated with the exogenous GpC methyltransferase M.CviPI. This enzyme methylates cytosines in accessible (unprotected) GpC contexts. DNA protected by nucleosomes or TFs remains unmethylated.
• Step 2: Histone Modification Mapping (mA): Cells are sequentially incubated with a primary antibody targeting a specific histone modification (e.g., H3K4me3) and a Protein A-Hia5 fusion protein. Hia5 is an adenine methyltransferase that deposits 6mA (methylated adenine) locally at the site of the antibody-bound histone mark.
  • Optimization: The authors optimized a sequential protocol (mC first, then mA) to prevent enzyme competition and ensure high fidelity. They also trained a custom deep learning model (using Remora) to call adenine methylation in the presence of cytosine methylation, overcoming previous accuracy issues in Nanopore data.
• Sequencing: High-molecular-weight DNA is extracted and sequenced using Oxford Nanopore Technologies (ONT). The long reads (median ~5 kb) allow simultaneous detection of mC and mA signals, as well as endogenous 5mC and SNPs, on the same molecule.

Computational Framework:

• Single-Molecule Classification: Reads are classified as "marked" or "unmarked" for specific features based on signal thresholds (e.g., >10% mA for HM presence, >50% mC for accessibility).
• Co-occurrence Analysis: Statistical tests (Fisher's exact test) are used to determine if the presence of an HM correlates with changes in accessibility or nucleosome phasing on the same molecule.
• Haplotype Resolution: In mammalian systems, reads are phased using SNPs (either known reference SNPs or data-driven phasing) to assign epigenetic states to specific alleles.

3. Key Contributions

• Multi-Layer Integration: First method to simultaneously profile histone modifications, chromatin accessibility, TF binding, DNA methylation, and genetic variation on single, multi-kilobase DNA molecules.
• Technical Innovation: Development of a robust sequential enzymatic protocol and a custom computational model to accurately detect 6mA in the presence of 5mC/5hmC on Nanopore reads.
• Haplotype-Resolved Epigenomics: Demonstrated the ability to resolve allele-specific epigenetic regulation in complex mammalian genomes without prior knowledge of the haplotype structure (reference-free phasing).
• Long-Range Analysis: Enabled the study of regulatory interactions over distances of several kilobases, linking distal enhancers to promoters on the same molecule.

4. Key Results

• Benchmarking in Drosophila:
  • ChromSMF was validated in Drosophila S2 cells against H3K4me3, showing strong correlation ($R=0.79$) between single-molecule marking frequency and bulk ChIP-seq enrichment.
  • It successfully distinguished active promoters (high H3K4me3, high accessibility) from inactive ones.
  • Molecular Heterogeneity: Revealed that even at active promoters, only a fraction of molecules (median ~60%) carry the H3K4me3 mark, highlighting cell-to-cell heterogeneity.
• Histone Modification Specificity:
  • Successfully profiled three active marks (H3K4me3, H3K27ac, H3K4me1) and two repressive marks (H3K27me3, H3K9me3).
  • Context-Dependent Effects: Found that H3K4me3 presence correlates with increased chromatin accessibility and specific nucleosome phasing at only ~14% of promoters, indicating locus-specific regulation.
  • Enhancer Logic: H3K27ac-marked molecules showed significantly higher accessibility than unmarked molecules at specific enhancers, but this association was not universal across all enhancers.
• Long-Range Associations:
  • Identified that H3K27ac at an enhancer can be associated with increased chromatin accessibility at distal promoters and other enhancers up to 5 kb away on the same molecule.
  • In contrast, H3K4me3 effects were largely restricted to the immediate promoter region.
• Mammalian Allele-Specific Analysis:
  • Applied to F1 hybrid mouse embryonic stem cells (Bl6 x Cast).
  • Imprinted Control Regions (ICRs): Recapitulated known monoallelic expression patterns at the Kcnq1ot1 locus, showing differential 5mC, accessibility, and H3K4me3 between maternal and paternal alleles.
  • TF Footprinting: Detected allele-specific binding of the repressor Zfp57 on the repressed maternal allele, a detail missed by bulk assays.
  • Causal Inference: At the Tal2 gene, allelic imbalance in expression was driven primarily by differences in H3K4me3 frequency (36% difference) rather than changes in chromatin accessibility or DNA methylation, demonstrating the necessity of multi-layer profiling to explain gene regulation.

5. Significance

• Mechanistic Insight: ChromSMF moves beyond correlation to provide direct evidence of how specific epigenetic marks functionally influence chromatin structure and gene regulation on a single-molecule level.
• Disease Relevance: By linking genetic variation (SNPs) directly to epigenetic states and regulatory outcomes on the same molecule, the method offers a powerful framework for understanding the mechanistic basis of disease-associated mutations and allele-specific expression in complex traits.
• Clinical Potential: The ability to perform haplotype-resolved analysis without a reference genome makes this approach applicable to clinical samples where personalized genome assemblies are unavailable.
• Future Directions: The authors suggest the method can be expanded to profile other chromatin-associated proteins (e.g., RNA Pol II, histone variants) and is adaptable to PacBio HiFi sequencing for higher base-calling accuracy.

In summary, ChromSMF represents a paradigm shift in epigenomics, providing a unified, quantitative, and haplotype-resolved view of the combinatorial logic governing gene regulation.