Mapping Protein Occupancy on DNA with an Unnatural Cytosine Modification in Bio-orthogonal Contexts

The authors engineered non-CpG-specific DNA methyltransferases to introduce an unnatural 5-carboxymethylcytosine modification as a bio-orthogonal label, enabling the reliable, multimodal mapping of protein-DNA occupancy alongside native epigenetic marks without confounding readouts.

The Gist

Imagine your DNA is a massive, intricate instruction manual for building and running a living organism. But this manual isn't just static text; it has sticky notes, highlighters, and bookmarks that tell the cell which chapters to read and which to ignore. These "sticky notes" are chemical modifications to the DNA letters.

Now, imagine you want to know two things at the same time:

1. Where are the sticky notes? (Epigenetic marks)
2. Where are the readers? (Proteins sitting on the DNA, blocking access to certain instructions).

The problem is that most existing tools for finding these "readers" are like using a sledgehammer to find a bookmark. They often destroy the page, confuse the sticky notes with the text, or require expensive, specialized equipment to read the results.

The Big Idea: A "Ghost" Highlighter

The researchers in this paper came up with a clever solution: What if we could use a highlighter that doesn't exist in nature?

They created a special, "unnatural" chemical tag called 5-carboxymethylcytosine (let's call it the "Ghost Highlighter"). This tag is invisible to the cell's natural machinery and, crucially, it doesn't look like any of the natural sticky notes already on the DNA.

Here is how they made it work, step-by-step:

1. The Tool: A Mutated "Stapler"

Cells have natural enzymes (molecular machines) that act like staplers. They go around DNA and attach specific chemical tags (staples) to certain letters.

• The Problem: The natural staplers only work on specific patterns (like "CpG") and use a standard staple (methyl group).
• The Innovation: The scientists took a natural stapler (an enzyme called M.CviPI) and performed "surgery" on it. They swapped out one tiny part of the machine (a single amino acid) to make it a "neomorphic" enzyme.
• The Result: This mutated stapler now ignores the standard staples. Instead, it grabs a special, bulky "Ghost Staple" (carboxymethyl-SAM) and attaches it only to DNA letters that are not currently being held by a protein.

2. The Game of "Musical Chairs"

Think of the DNA as a long dance floor.

• The Proteins (Dancers): Some proteins, like the "LexA" repressor in bacteria, sit on the dance floor, blocking the view.
• The Enzyme (The Painter): The mutated enzyme runs around the dance floor with a can of "Ghost Paint."
• The Rule: The enzyme can only paint the spots on the floor that are empty. If a protein is sitting there, the enzyme bumps into it and can't paint that spot.

Because the "Ghost Paint" is unnatural, it doesn't get washed away or confused with the natural marks on the floor. It stays exactly where the enzyme put it.

3. Reading the Map

Once the painting is done, the scientists sequence the DNA.

• If a spot is painted: It means the spot was empty (no protein was there).
• If a spot is bare: It means a protein was sitting there, blocking the paint.

Because this "Ghost Paint" is so distinct, they can use standard, cheap, and common lab equipment to read the results. They don't need fancy, expensive machines. They can even see the natural sticky notes (like 5mC) and the Ghost Paint at the same time without them getting mixed up.

The Discovery: The "Double-Seat" Puzzle

To test this new tool, they looked at a specific bacterial gene called lexA. This gene has a "control panel" with two seats (binding sites) where the LexA protein can sit.

• The Mystery: Scientists weren't sure if the LexA protein could sit in both seats at the same time, or if it had to choose one. They also wondered if a natural "sticky note" (methylation) on the DNA would stop the protein from sitting down.
• The Result: Using their new "Ghost Highlighter," they saw that LexA can sit in both seats simultaneously, even when the natural sticky note is present. It's like finding out a person can sit in two chairs at once without the extra cushion on one chair stopping them.

Why This Matters

This paper is a breakthrough because it gives scientists a non-destructive, high-resolution camera for the genome.

• Old way: Like trying to figure out who is in a room by knocking down the walls (destructive) or using a flashlight that makes everything look the same color (confusing).
• New way: Like using a special paint that only appears on empty chairs. You can see exactly who is sitting where, what the room looks like, and you can do it with standard tools.

This opens the door to understanding how genes are turned on and off in much greater detail, potentially helping us understand diseases like cancer where these "sticky notes" and "readers" go wrong.

Technical Summary

1. The Problem

Understanding the epigenome requires the concurrent mapping of DNA base modifications (e.g., 5-methylcytosine, 5mC) and protein-DNA occupancy (transcription factor binding). Current multi-modal mapping methods face significant limitations:

• Chemical Overlap: Existing methods like NOMe-Seq use the natural enzyme M.CviPI to install 5mC in GpC contexts to mark accessible DNA. However, because 5mC is a natural modification, it is difficult to distinguish from endogenous 5mC (in CpG contexts) or 5-hydroxymethylcytosine (5hmC), leading to confounded readouts.
• Sequencing Constraints: Methods using 6-methyladenine (m6A) require third-generation sequencing, which increases input requirements and limits compatibility with established short-read protocols like Bisulfite Sequencing (BS-Seq).
• Destructive Methods: Chemical footprinting (e.g., hydroxyl radical) often requires large DNA inputs and causes biased fragmentation, hindering single-cell or rare-population analysis.

The authors sought a bio-orthogonal solution: an unnatural DNA modification that can be enzymatically installed in non-CpG contexts, is resistant to standard chemical/enzymatic conversion, and allows for simultaneous profiling of native modifications and protein occupancy.

2. Methodology

The study employed a structure-guided protein engineering approach to create "neomorphic" (new function) DNA methyltransferases (MTases).

• Target Selection: The authors focused on two non-CpG specific cytosine MTases: M.CviPI (recognizes GpC) and M.CviQIX (recognizes CpC).
• Rational Design:
  • They utilized AlphaFold3 to predict the structures of M.CviPI and M.CviQIX bound to DNA.
  • These were aligned with the crystal structure of M.MpeI (a CpG-specific MTase previously engineered to accept carboxy-S-adenosyl-L-methionine, CxSAM).
  • They identified homologous residues to the critical N374 site in M.MpeI: Y205 in M.CviPI and N218 in M.CviQIX.
  • Based on prior findings that cationic residues facilitate CxSAM usage, they generated Y205K and Y205R mutants for M.CviPI, and N218K/R mutants for M.CviQIX.
• Screening & Validation:
  • In Vivo: Expressed variants in E. coli and assessed protection against HaeIII restriction digestion (which cuts GGCC unless modified).
  • In Vitro: Purified enzymes and tested activity on pUC19 plasmids and synthetic oligonucleotides using either SAM (natural cofactor) or CxSAM (unnatural cofactor).
  • Mass Spectrometry: Used LC/MS on oligonucleotides to definitively confirm the mass shift corresponding to 5-carboxymethylcytosine (5cxmC).
• Sequencing Compatibility:
  • Applied the engineered enzymes to $\lambda$ phage genomic DNA.
  • Sequenced the modified DNA using BS-Seq (chemical) and ACE-Seq (enzymatic) to determine if 5cxmC resists both C-to-T conversion methods.
• Biological Application:
  • Used the GpC-specific CxMTase (M.CviPI Y205K + CxSAM) to map LexA binding on the E. coli lexA promoter.
  • The lexA promoter contains two SOS boxes and an endogenous Dcm methylation site (CCWGG). The assay tested if LexA could bind both sites simultaneously and if endogenous methylation affected this binding.

3. Key Contributions

• Engineering of Non-CpG CxMTases: Successfully engineered M.CviPI variants (specifically Y205K and Y205R) to utilize CxSAM and install 5cxmC at GpC sites.
• Bio-orthogonal Labeling: Demonstrated that 5cxmC is chemically distinct from natural 5mC and 5hmC, resisting both bisulfite conversion and enzymatic deamination (A3A).
• Symmetric Modification: Unlike the previously engineered CpG-specific CxMTase (which showed hemi-methylation bias), the GpC-specific CxMTase showed symmetric modification on both DNA strands, allowing for higher global modification efficiency.
• Improved Specificity: The engineered mutants (Y205K) exhibited higher fidelity for GpC sites compared to the wild-type M.CviPI, which is known to have off-target CpC activity.

4. Key Results

• Enzyme Activity: The M.CviPI Y205K and Y205R mutants efficiently converted GpC cytosines to 5cxmC using CxSAM. LC/MS confirmed a mass shift of ~57.8 Da, consistent with 5cxmC.
• Sequencing Compatibility:
  • BS-Seq: 5cxmC protected GpC sites from C-to-T conversion (similar to 5mC).
  • ACE-Seq: Crucially, 5cxmC also protected GpC sites from enzymatic deamination, whereas 5mC (from SAM) did not. This allows for the unambiguous detection of the unnatural mark using all-enzymatic or chemical pipelines.
  • Context Specificity: The mutants showed strong preference for GpC contexts with minimal off-target activity, even when using SAM.
• LexA Footprinting:
  • The GpC-CxMTase successfully blocked modification at GpC sites occupied by LexA, creating a clear "footprint."
  • Simultaneous Binding: Single-molecule analysis of the lexA promoter revealed that LexA can concurrently bind to both SOS boxes in the native promoter.
  • Methylation Independence: The presence of endogenous Dcm methylation at the embedded CCWGG site did not alter LexA occupancy or binding affinity.

5. Significance

• Novel Multimodal Profiling: This work establishes a fully enzymatic pipeline for mapping protein occupancy that is compatible with standard short-read sequencing (BS-Seq) and avoids the input limitations of third-generation sequencing.
• Decoupling Native and Exogenous Marks: By using an unnatural base (5cxmC) in non-native contexts (GpC), researchers can now map protein-DNA interactions without confounding signals from the cell's native epigenetic landscape (5mC/5hmC).
• Biological Insights: The method provided the first single-molecule evidence that LexA binds both SOS boxes simultaneously in the lexA promoter, a finding previously difficult to resolve due to the limitations of synthetic oligo studies.
• Future Applications: The bio-orthogonal nature of 5cxmC opens the door for complex multi-modal profiling, such as simultaneously mapping 5hmC (via ACE-Seq) and open chromatin (via 5cxmC footprinting) in the same sample, a task currently infeasible with natural modification labels.

In summary, the authors successfully expanded the toolkit of epigenetic engineering by creating a non-CpG specific enzyme that installs a bio-orthogonal "barcode" (5cxmC), enabling high-resolution, non-destructive mapping of protein-DNA interactions alongside native epigenetic marks.