TL;DR
HiPore-C is an optimized nanopore-based sequencing workflow that dramatically increases the throughput of "multiway" chromatin contact mapping. By solving the technical hurdle of nanopore clogging, the researchers were able to look past the "average" 3D genome structure to see how individual alleles fold, revealing that even highly conserved genomic domains possess cell-type-specific internal architectures.
What Problem Does This Paper Address?
For decades, our understanding of the 3D genome—how two meters of DNA is packed into a microscopic nucleus—has relied on Hi-C. While revolutionary, Hi-C has a major limitation: it only captures pairwise interactions (Point A touches Point B). Because Hi-C results are an "ensemble average" of millions of cells, we lose the context of multiway interactions (Point A, B, and C touching simultaneously) and the single-allele topology (how one specific copy of a chromosome is folded at a single moment).
Existing "Pore-C" methods attempted to solve this using long-read nanopore sequencing to capture entire concatemers of interacting DNA. However, these methods were plagued by low throughput. Residual proteins crosslinked to the DNA would clog the nanopores, making the process expensive and data-poor.
Methods
The authors introduce HiPore-C, which features two primary innovations:
- Biochemical Optimization: The team discovered that incomplete protein removal was the culprit behind pore clogging. They optimized the protocol using multiple rounds of Protease K digestion or a combination of Protease K and Pronase (a powerful non-specific protease mixture). This "deep cleaning" of the DNA concatemers increased sequencing yield by approximately 80%.
- MapPore-C Pipeline: They developed a bioinformatics pipeline that integrates the
NGMLRandMinimap2aligners to accurately map long, multi-fragment reads to the reference genome, allowing them to reconstruct both virtual pairwise maps (for comparison with Hi-C) and complex multiway interaction networks.
Crucially, because they used nanopore sequencing, they were able to detect DNA methylation (5mC) directly from the raw signal, allowing them to overlay epigenetic data onto the 3D structural data without additional bisulfite treatment.
Datasets
The study evaluated HiPore-C using:
- Human Cell Lines: GM12878 (B-lymphocytes) and K562 (chronic myeloid leukemia).
- Validation Benchmarks: Standard Hi-C datasets, previous Pore-C data (Deshpande et al., 2022), and ENCODE data for RNA-seq, ChIP-seq, and Whole Genome Bisulfite Sequencing (WGBS).
- Yield: They achieved ~128–144 Gbase per ONT PromethION flow cell, a significant jump over previous standards.
Key Insights and Ideas
- TADs are Ensembles of Diverse Topologies: Topologically Associating Domains (TADs) are often thought of as stable "bricks" of the genome. HiPore-C shows they are actually statistical averages of many different single-allele folding patterns.
- Hidden Cell-Type Specificity: The most striking finding was that even when a TAD's boundaries are identical between two cell types (conserved), the internal clustering of single-allele topologies is highly cell-type specific. This suggests that the way DNA moves inside its "container" is what drives specific gene expressions.
- Rare Synergistic Looping: While multiway reads are common, "synergistic" hubs (where three or more specific distal sites consistently meet) are rarer than simple pairwise loops. Most multiway contacts resemble "C-walks"—sequential proximities along a chromatin fiber.
- Interchromosomal Hubs: The study identified two distinct types of interchromosomal hubs: an "active hub" (enriched with A-compartment/active chromatin) and an "inactive hub" (enriched with B-compartment/repressive chromatin), providing a clearer picture of how different chromosomes talk to each other.
Key Results
- Throughput & Cost: HiPore-C produced significantly more virtual pairwise contacts at a lower cost per million contacts compared to previous Pore-C methods.
- Structural Fidelity: HiPore-C faithfully reproduced canonical 3D structures (A/B compartments, TADs, and loops) with high correlation to Hi-C (Pearson’s $r \approx 0.96$ for compartments).
- Methylation Accuracy: The simultaneous 5mC detection showed a high correlation ($r = 0.80$) with gold-standard WGBS, proving that 3D structure and epigenetics can be mapped in a single "one-pot" experiment.
- Single-Allele Clustering: By clustering multiway reads, they identified specific "topology clusters" within the $\beta$-globin locus that correlate with the activation of embryonic vs. adult globin genes.
Limitations and Future Work
While HiPore-C provides unprecedented depth, the authors acknowledge that it still represents a population of cells. A true "single-cell multiway" approach would be the next frontier. Additionally, while they show that internal TAD topologies are cell-type specific, the causal link—whether the topology drives the gene expression or vice versa—remains a "chicken and egg" problem that requires further functional studies using tools like CRISPR-mediated structural disruption.
Finally, while throughput is much higher, the computational cost of processing multi-fragment long reads is significantly higher than standard short-read Hi-C, necessitating further optimization of the bioinformatics pipeline.
Summary updated