G-quadruplex structures act as a novel recognition motif for the meiosis-specific histone methyltransferase PRDM9

Technical Summary: G-quadruplex structures act as a novel recognition motif for the meiosis-specific histone methyltransferase PRDM9

Problem Statement
The histone methyltransferase PRDM9 is the primary determinant of meiotic recombination hotspot locations in most mammals, including humans. It functions by binding specific degenerate DNA sequence motifs via its C-terminal zinc-finger (ZnF) array and depositing H3K4me3 and H3K36me3 histone marks to recruit the double-strand break (DSB) machinery. However, a significant limitation in the current understanding of PRDM9 function is that the primary DNA sequence motif alone cannot fully explain its genome-wide occupancy. Many PRDM9-bound hotspots lack recognizable consensus sequences, and a subset of PRDM9 binding sites occurs in promoter-associated regions and CpG islands, which are typically depleted of recombination events. Furthermore, the rapid evolution of the PRDM9 ZnF array generates new alleles with distinct motif specificities, yet the mechanisms governing hotspot selection beyond simple sequence recognition remain unclear. The authors hypothesize that DNA secondary structures, specifically G-quadruplexes (G4s), which form in G-rich regions complementary to PRDM9's C-rich binding motifs, may serve as an additional recognition platform facilitating PRDM9 recruitment.

Methodology
The study employed a multi-faceted approach combining large-scale bioinformatics analysis with in vitro biochemical assays:

1. Computational and Genomic Analysis:

   • Data Integration: The authors integrated PRDM9 ChIP-seq data (from HEK293 cells overexpressing PRDM9-B) with BG4 antibody-based CUT&Tag data (specific for G4 structures) in the same cell line.
   • Enrichment Analysis: They developed a pipeline to compare PRDM9 binding sites against matched (controlling for length, GC content, and entropy) and global background genomic windows. Features analyzed included G4 motifs (predicted by pqsfinder and validated by BG4 CUT&Tag), histone marks (H3K4me3, H3K36me3), DNA methylation, chromatin accessibility (ATAC-seq), and proximity to known recombination hotspots.
   • Stratification: G4 motifs were stratified by stability scores (quartiles), loop length (short, medium, long), sequence architecture (canonical vs. non-canonical), and experimental support (validated vs. predicted).

2. Biochemical Assays (EMSA):

   • Protein Preparation: Full-length human PRDM9-A (the predominant European allele) fused to eYFP was transiently overexpressed in HEK293 cells. Whole-cell extracts were prepared to preserve the protein's native folding and ZnF array integrity.
   • DNA Substrates:
     • dsDNA: Biotinylated PCR fragments covering two well-characterized recombination hotspots (HSI in PCP4 and HSII in RBFOX1) containing PRDM9 motifs and overlapping G4s.
     • ssDNA G4s: Short biotinylated oligonucleotides representing G4 motifs from the hotspots, artificial G4 sequences (not found in the human genome), and mutated variants designed to disrupt G4 folding.
   • Electrophoretic Mobility Shift Assays (EMSA): Binding reactions were conducted in TKZN buffer with varying KCl concentrations (50 mM vs. 150 mM) and the addition of the G4-stabilizing ligand pyridostatin (PDS) to modulate G4 stability. Binding was visualized via chemiluminescence.
   • Structural Validation: Circular Dichroism (CD) spectroscopy and melting curve analyses were used to confirm the formation and thermal stability of the G4 structures under experimental conditions.

Key Contributions and Results

1. Genomic Enrichment of G4s at PRDM9 Sites:

   • Computational analysis revealed that G4 motifs are among the most enriched features at PRDM9 binding sites, comparable to the enrichment of the canonical PRDM9 sequence motif and histone marks.
   • Crucially, this enrichment was significantly stronger for experimentally validated G4s (detected by BG4 CUT&Tag) than for computationally predicted G4s lacking experimental support. This suggests the association is driven by the physical structure rather than just the G-rich sequence potential.
   • Stability and Canonical Structure: Highly stable G4s (top quartile of pqsfinder scores) and canonical G4 motifs showed the strongest enrichment. Non-canonical G4s and less stable structures exhibited substantially weaker enrichment.
   • Chromatin Context: PRDM9 binding sites overlapping with stable, validated G4s showed significantly higher pre-existing chromatin accessibility (ATAC-seq signal) compared to sites without G4s, suggesting G4s may mark or maintain accessible chromatin regions prior to PRDM9 binding.

2. Direct Binding of PRDM9 to G4 Structures:

   • dsDNA Binding: PRDM9 extracts successfully bound dsDNA fragments from recombination hotspots (HSI and HSII) containing PRDM9 motifs.
   • ssDNA G4 Binding: PRDM9 directly bound single-stranded G4-forming oligonucleotides. Binding affinity correlated with G4 stability:
     • The highly stable HSI-G4 (score 89) showed strong binding, whereas the less stable HSII-G4 (score 29) showed weak binding.
     • Increasing KCl concentration (stabilizing G4s) and adding PDS significantly enhanced PRDM9 binding to stable G4s (Q4 motif) but had minimal effect on unstable G4s (Q1 motif).
   • Artificial G4 Recognition: PRDM9 bound an artificial G4 sequence (not present in the human genome) that formed a stable hybrid topology. Mutating this sequence to prevent G4 formation abolished binding, confirming that PRDM9 recognizes the secondary structure itself, independent of the specific primary sequence context.

3. Competitive Binding:

   • In competition EMSAs, stable G4 oligonucleotides effectively competed with canonical dsDNA hotspot fragments for PRDM9 binding. Less stable G4s showed reduced competitive capacity, reinforcing the link between structural stability and recognition.

Significance and Claims
The authors propose a model where stable G-quadruplex structures act as a novel recognition motif for PRDM9, functioning alongside the canonical sequence-specific zinc-finger binding. The paper claims that:

• Structural Recognition: PRDM9 can recognize and bind G4 secondary structures directly, not just the primary DNA sequence.
• Stability Dependence: The efficiency of this recognition is strictly dependent on the thermodynamic stability of the G4 structure; highly stable G4s are preferentially bound.
• Mechanism of Hotspot Specification: Stable G4s likely facilitate PRDM9 recruitment by creating or maintaining discrete, accessible chromatin environments. This provides a mechanism for PRDM9 to target specific loci even when the primary sequence motif is suboptimal or when the site lacks a consensus motif.
• Evolutionary Implications: The study suggests that variations in G4 stability (e.g., due to SNPs like rs74149540) could modulate PRDM9 binding affinity and hotspot activity, contributing to the dynamic turnover of recombination hotspots and population-specific recombination landscapes.

The paper concludes that DNA secondary structure constitutes an additional layer of information regulating meiotic recombination, offering a framework to reconcile PRDM9 occupancy at sites lacking consensus motifs and explaining the enrichment of PRDM9 binding in promoter-associated regions. The authors acknowledge limitations, noting that the study relies on in vitro assays and computational predictions, and that future work is required to establish the causal role of G4 recognition in in vivo DSB formation.