Proteomics-Based Discovery of Symmetry-Specific Readers and Antireaders of 5-Formylcytosine in Mammalian DNA

This study presents the first proteome-wide analysis revealing that mammalian proteins, including transcription factors and DNA repair enzymes, exhibit diverse and symmetry-specific recognition patterns for 5-formylcytosine modifications in CpG dyads, thereby modulating target selection and chromatin regulation.

The Gist

Imagine your DNA as a massive, complex instruction manual for building and running a human body. For a long time, scientists thought this manual was written in just two main "letters": Cytosine (C) and its methylated version, 5-methylcytosine (mC). Think of mC like a sticky note that says, "Don't read this part," effectively silencing genes.

However, nature has a way of editing these notes. Enzymes can take that "methyl" sticky note and turn it into something else, like 5-formylcytosine (fC). You can think of fC as a special, glowing highlighter. It doesn't just silence genes; it seems to send a different, more complex signal to the cell's machinery.

The Big Mystery: The "Symmetry" Puzzle

Here is the tricky part: DNA is a double helix, meaning it has two strands that mirror each other. Usually, these strands are identical (symmetric). But sometimes, one strand might have the glowing highlighter (fC) while the other has a regular letter (C), or a methyl note (mC).

This creates different "combinations" or symmetries:

• Symmetric: Both strands have the highlighter (fC/fC).
• Asymmetric: One strand has the highlighter, the other has a regular letter (fC/C) or a methyl note (fC/mC).

For years, scientists only looked at the symmetric version (fC/fC). They knew proteins existed that could "read" these highlighters, but they didn't know if the cell's machinery cared about the other strand. Did a protein care if the highlighter was alone or paired with a methyl note?

The Experiment: A Massive "Speed Dating" Event

The researchers in this paper decided to throw a massive "speed dating" event for proteins.

1. The Setup: They created three different types of "date cards" (DNA probes) based on real human and mouse gene promoters. Each card had the glowing highlighter (fC) in different combinations: paired with itself, paired with a regular letter, or paired with a methyl note.
2. The Guests: They took the "nuclei" (the control centers) from human cells (like HEK293T and HeLa) and mouse stem cells.
3. The Matchmaking: They dropped these DNA cards into a bowl of proteins and asked: "Who wants to hold hands with this specific card?"
4. The Reveal: Using a high-tech scanner (Mass Spectrometry), they identified exactly which proteins grabbed which cards.

The Surprising Discoveries

1. The Highlighter is a Magnet
They found that the glowing highlighter (fC) attracts a huge crowd of proteins—more than the standard "silencing" notes (mC) or regular letters (C). It's like fC is a VIP section at a club that everyone wants to get into.

2. The "Symmetry" Matters
The most exciting finding is that proteins are picky about who the highlighter is paired with.

• Some proteins are Symmetry Lovers: They only want to hold hands if both strands have the highlighter (fC/fC).
• Some are Asymmetry Fans: They prefer the highlighter to be paired with a regular letter (fC/C).
• Some are Anti-Readers: They actually run away from the highlighter if it's paired with a methyl note (fC/mC).

It's as if the protein is saying, "I love this glowing note, but only if it's sitting next to a normal letter, not a methyl note!"

3. The "Context" is King
The study found that a protein's preference changes depending on the surrounding DNA sequence.

• Example: The protein MAX (a key player in cell growth) usually ignores methyl notes. But when it sees the glowing highlighter (fC), it grabs it enthusiastically—unless the highlighter is paired with a methyl note on the other strand, in which case it ignores it.
• Analogy: Imagine a security guard (MAX) who usually lets people with red hats (mC) in, but ignores them. But if someone wears a glowing red hat (fC), the guard lets them in immediately. However, if the glowing hat is paired with a specific badge (mC) on the other side, the guard suddenly stops them. The "context" changes the rule.

4. The Repair Crew is Busy
They also found that the cell's "repair crew" (proteins like TDG and MPG) are very interested in these highlighters. In fact, some of these repair proteins act like "readers" that specifically look for the glowing highlighter to fix the DNA. This suggests fC isn't just a signal; it might be a "flag" telling the repair crew, "Hey, fix this spot!"

Why Does This Matter?

Think of the cell's genome as a city.

• C is the standard street sign.
• mC is a "Road Closed" sign.
• fC is a flashing neon sign that says "Construction Ahead" or "Special Event."

This paper tells us that the city's workers (proteins) don't just see the neon sign; they look at the whole intersection. They check if the neon sign is alone, paired with a "Road Closed" sign, or paired with another neon sign. Depending on that combination, they decide whether to build a new road, close a building, or send a repair crew.

The Takeaway:
This research gives us the first "phone book" of who talks to these glowing DNA signals. It shows that the cell uses the symmetry of these signals to send incredibly precise instructions. This is crucial for understanding how cells decide to grow, differentiate, or become cancerous. If we can understand exactly which proteins read which "highlighter combinations," we might be able to design new drugs to fix broken instructions in diseases like cancer.

Technical Summary

1. Problem Statement

5-methylcytosine (mC) and its oxidized derivatives (5-hydroxymethylcytosine [hmC], 5-formylcytosine [fC], and 5-carboxylcytosine [caC]) are critical epigenetic marks regulating transcription, differentiation, and cancer. While previous proteomic studies have identified protein "readers" for these modifications, they have largely focused on symmetric CpG dyads (e.g., fC/fC).

However, in mammalian genomes, CpG dyads often exist in asymmetric states (e.g., fC/C or fC/mC) due to the co-existence of different cytosine forms on opposing strands. This creates 15 possible symmetric and asymmetric dyad states, each presenting a unique physicochemical mark in the DNA major groove. It remains poorly understood whether nuclear proteins can distinguish between these specific symmetry states or if they recognize fC indiscriminately. There is a lack of systematic data regarding how protein binding preferences change based on the symmetry (symmetric vs. asymmetric) and sequence context of fC modifications.

2. Methodology

The authors employed a quantitative proteomics approach coupled with DNA enrichment to map protein interactions with specific fC dyad states across different cell types and sequence contexts.

• Cell Lines: Nuclear extracts were prepared from three distinct sources:
  • HEK293T (Human embryonic kidney cells)
  • HeLa (Human cervical adenocarcinoma cells)
  • mESC (Mouse embryonic stem cells)
• DNA Probe Design: Three distinct DNA probes were generated to cover different promoter regions and modification patterns:
  1. VEGFA Probe: A ~200 nt PCR-generated probe from the human VEGF-A promoter.
  2. hSP1 Probe: A PCR-generated probe from the human SP1 promoter.
  3. mSP1 Probe: A 50 nt synthetic probe from the mouse Sp1 promoter (allowing precise modification at CpG sites only).
• Modification States: Each probe was synthesized in five variants representing different CpG dyad states:
  • C/C (Unmodified)
  • mC/mC (Symmetric methylation)
  • fC/fC (Symmetric formylation)
  • fC/C (Asymmetric: fC on one strand, C on the other)
  • fC/mC (Asymmetric: fC on one strand, mC on the other)
• Enrichment & Mass Spectrometry:
  • Biotinylated probes were immobilized on streptavidin beads.
  • Beads were incubated with nuclear extracts (with competitor DNA to reduce non-specific binding).
  • Bound proteins were eluted, digested (Lys-C/Trypsin), and analyzed via Data-Independent Acquisition (DIA) Mass Spectrometry (Orbitrap Exploris 480).
  • Data analysis involved statistical filtering (t-tests, ANOVA) to identify significantly enriched proteins (Readers) or depleted proteins (Antireaders).
• Validation: Selected candidates (Transcription Factors and DNA Repair proteins) were cloned, expressed as MBP-fusion proteins in E. coli, purified, and validated using Electrophoretic Mobility Shift Assays (EMSA) to measure binding affinities ($K_D$) for specific dyad states. Structural modeling was performed using AlphaFold 3.

3. Key Contributions

• First Proteome-Wide Symmetry Profiles: This study provides the first comprehensive interaction profiles for both symmetric and asymmetric fC modifications in mammalian DNA.
• Discovery of Symmetry-Specificity: The authors demonstrate that protein recognition of fC is highly dependent on the symmetry of the CpG dyad. Many proteins act as "readers" for one symmetry state but "antireaders" (or weak binders) for another.
• Context-Dependent Binding: The study reveals that binding preferences are not only symmetry-dependent but also highly sensitive to the local DNA sequence (probe design) and the tissue/cell type (HEK293T vs. HeLa vs. mESC).
• Identification of Novel Readers: The study identifies a diverse set of fC readers, including specific Transcription Factors (TFs) and Base Excision Repair (BER) proteins that were not previously characterized for fC symmetry specificity.

4. Key Results

A. Proteomic Landscape

• Abundance: The number of proteins binding to fC-modified probes (all three states combined) exceeds the number of proteins binding to canonical C/C and mC/mC probes combined.
• Distinct Groups: The three fC states (fC/fC, fC/C, fC/mC) attract largely distinct groups of proteins with minimal overlap, suggesting unique regulatory roles for each state.
• Tissue and Sequence Specificity: While some "common readers" exist across cell lines, a significant portion of the interactome is tissue-specific (e.g., HeLa vs. HEK293T) and sequence-specific (VEGFA vs. SP1 probes).

B. Specific Protein Findings (Validated by EMSA)

The study validated 11 candidate proteins, revealing distinct symmetry preferences:

• Transcription Factors (TFs):

  • MAX: A bHLH TF that reads C/C and fC/C with high affinity but is repelled by mC/mC and fC/mC. It treats fC similarly to C in specific E-box contexts, suggesting fC can be an "activating" mark compared to mC.
  • HEY1: Similar to MAX, binds fC/C, fC/fC, and C/C with high affinity, but is repelled by mC/mC and fC/mC.
  • RFX5: An mC reader that tolerates fC. It binds fC/mC and fC/fC with higher affinity than C/C, but acts as an antireader for fC in the fC/mC context compared to hmC.
  • SIX1 & SIX2: Homeobox TFs that show a strong preference for symmetric fC/fC over asymmetric states (fC/C, fC/mC) and unmodified C/C. They exhibit strand-specific recognition.
  • FOXJ3: Prefers fC/fC > fC/mC > fC/C, showing strand-specificity.
  • TP53: Shows enrichment for fC/mC and fC/C over C/C in specific contexts, indicating context-dependent reading.

• DNA Repair Proteins (BER):

  • TDG (Thymine-DNA Glycosylase): Confirmed as a universal fC reader but shows the highest affinity for symmetric fC/fC, with reduced affinity for asymmetric states (fC/C, fC/mC).
  • MPG (N-methylpurine DNA glycosylase): Acts as a reader for fC but exhibits complex behavior. In the VEGFA probe, it prefers fC/fC; however, in the mSP1 probe, it shows high affinity for mC/mC and variable affinity for fC depending on the sequence context and strand orientation. This suggests MPG's interaction with fC is highly sequence-dependent and may involve inhibition mechanisms.

C. Functional Clustering

• C/C Readers: Primarily chromatin regulators and Polycomb repressive complex (PRC) members.
• mC/mC Readers: Canonical methyl-CpG binding proteins (MBDs, MeCP2).
• fC Readers: Enriched for Base Excision Repair (BER) factors (TDG, MPG, APEX1, XRCC1) and specific TFs. Notably, fC/fC readers include a cluster of TFIID members, suggesting a role in transcription initiation.

5. Significance

• Mechanistic Insight: The study establishes that fC is not a monolithic epigenetic mark. Its function is dictated by the symmetry of the CpG dyad, allowing for a "code" where fC/C, fC/mC, and fC/fC recruit distinct protein machineries.
• Regulatory Complexity: The discovery that proteins can act as "readers" or "antireaders" depending on the symmetry state (e.g., MAX reading fC/C but not fC/mC) implies a sophisticated layer of gene regulation where the cell can fine-tune protein recruitment by modulating the symmetry of oxidation.
• Disease Relevance: Given the roles of fC in early embryogenesis (paternal pronuclei) and its dysregulation in cancers (TET mutations, IDH mutations), this reader catalogue provides a resource to understand how altered fC symmetry contributes to malignant processes.
• Future Directions: The authors highlight the need for high-resolution genomic mapping of fC symmetry states (currently lacking compared to hmC) to fully correlate these proteomic findings with physiological outcomes.

In summary, this paper fundamentally shifts the understanding of 5-formylcytosine from a simple intermediate in demethylation to a complex, symmetry-dependent regulatory signal that recruits a diverse and specific set of nuclear proteins.