The team, led by Dania Camila Pulido-Barrera and Susan E. Mango at the Biozentrum of the University of Basel, used a technique called chromatin tracing, a DNA fluorescence in situ hybridization method that labels consecutive 100-kilobase segments along a chromosome and reconstructs their three-dimensional positions in single nuclei. Where population-averaged assays such as Hi-C blend thousands of cells into one consensus map, chromatin tracing preserves the individuality of each chromosome, allowing the researchers to sort thousands of traces into structural clusters. When they averaged their data, the result looked just like published Hi-C maps of the worm genome: smooth, featureless, and devoid of clear domain boundaries. But when they clustered the single-molecule traces, sharp, well-insulated domains popped into view.

Unsupervised clustering of more than five thousand chromosome traces from early embryos revealed seven distinct conformational groups. Three of them harbored roughly one-megabase domains with crisp boundaries and insulation scores nearly fourfold higher than the population average. The two strongest of these clusters accounted for about twenty percent of all traces and never appeared together in the same chromosome; instead, they alternated in an either-or fashion or merged into larger structures. This structural heterogeneity, the authors argue, is precisely why population-level assays have failed to detect such domains in C. elegans, a species long thought to lack canonical TADs because it has no CTCF protein, the insulator factor that anchors most TAD boundaries in vertebrates.

The researchers then asked which molecular machines build these domains. Their prime suspect was condensin I, a ring-shaped complex of the structural maintenance of chromosomes family that had previously been shown to fold the worm’s X chromosome into megabase-scale TADs during dosage compensation. Using a strong loss-of-function allele of dpy-28, which encodes a condensin I subunit, they traced chromosomes in condensin-depleted embryos. The effect was dramatic and selective. One of the two prominent wild-type domains, located at 19.7 to 20.9 megabases on chromosome V, vanished entirely in the mutant, while the neighboring domain at 18.2 to 19.6 megabases persisted, albeit slightly loosened. The condensin-dependent structure, with its sharp boundaries and TAD-like behavior, was christened a TADL, for TAD-like domain.

The surviving domain received a different name: elCID, short for elegans condensin-independent domain. Further experiments showed that elCIDs are also indifferent to histone H3K9 methylation, the classic chemical mark of heterochromatin. Embryos lacking virtually all H3K9 methylation, due to mutations in the met-2 and set-25 methyltransferase genes, still formed both TADL and elCID domains. H3K9 methylation did matter, however, for compaction: mutant chromosomes were significantly larger and less densely packed, with radius-of-gyration measurements matching those of condensin mutants. In other words, the epigenetic mark controls how tightly the genome is squeezed, but not where its domain walls stand. Cohesin, the other major SMC complex, proved irrelevant at this scale, organizing chromatin only in loops of twenty to forty kilobases that are averaged away at the tracing resolution.

Extending their analysis upstream, the team discovered a second elCID spanning 17.0 to 17.9 megabases, a region containing the 5S ribosomal RNA genes and the spliced leader repeat locus, hence its name, the 5SL elCID. This domain kept its sharp boundary even when condensin or H3K9me3 was abolished. Curiously, although the 5S locus is central to ribosome biogenesis, it showed no spatial association with the nucleolus; the median distance between the two exceeded 1.3 micrometers, with 98.6 percent of measurements above 300 nanometers. RNA polymerase III foci marking the 5S genes instead occupied transcriptionally permissive microenvironments, with H3K9me3 depleted inside the foci and enriched in the surrounding chromatin, suggesting the domain packages its precious cargo in a protected but not heterochromatin-embedded pocket.

The boundaries of the two domain types bore entirely different molecular fingerprints. The TADL boundary at 19.7 megabases was marked by peaks of all five condensin I subunits, a high-occupancy transcription factor site, and a break in the repressive H3K9me3 and H3K27me3 marks. The elCID boundaries, by contrast, coincided with active H3K36me3 marks, depletion of repressive marks, and robust maternal messenger RNA, with only weak zygotic transcription. This last observation points to a surprising conclusion: the boundaries of condensin-independent domains may be inherited from the mother’s transcriptional activity, consistent with the maternal inheritance of H3K36me3. Repeat elements, including long terminal repeats and satellite sequences, were also enriched at all three boundaries, and a zinc finger factor called ZTF-6 showed modest enrichment near the 19.0-megabase boundary.

Developmental timing added another layer of distinction. The TADL domain was already detectable in roughly thirty percent of embryos before zygotic genome activation, when no transcription occurs in the region, indicating that it assembles without any help from gene expression. The 5SL elCID appeared at the minor wave of zygotic activation, while the second elCID emerged only at the nine-to-forty-cell stage, coinciding with the onset of robust zygotic transcription, hinting that this domain class may depend on transcriptional activity. Power-law analysis of spatial scaling showed that wild-type chromosomes progressively compacted over development, with the scaling exponent falling from 0.19 to 0.13, whereas H3K9me mutants remained decompacted throughout and condensin mutants showed a unique defect in local folding at the 100-kilobase scale.

Perhaps the most conceptually striking result came from statistical polymer modeling. Even a free polymer with only nearest-neighbor interactions produced apparent boundaries and insulation in simulated traces, showing that generic molecular jitter can masquerade as domain structure. But such simple models failed to reproduce the measured boundary probabilities and the excess variability seen in real embryos. Only when structured, locus-specific interactions were included did the simulations match the data. Notably, condensin loss reduced this structured variability while H3K9me loss did not, identifying condensin I as a major source of organized chromatin fluctuation. The modeling also revealed that closely spaced domains cooperatively reinforce one another’s boundaries, explaining why the condensin-independent elCID loses some boundary strength when its condensin-dependent TADL neighbor is dismantled.

The study forces a reinterpretation of earlier claims that C. elegans lacks TADs. That absence, the authors conclude, reflects conformational variability rather than a genuine lack of domains: structures present in only a minority of chromosomes dissolve into the average. The finding resonates with live-imaging work in mammalian cells showing that TADs occupy only a small fraction of chromosome conformations at any moment. Whether the same diversity of heterochromatin domains extends to euchromatic regions and other chromosomes remains to be tested, and the 100-kilobase probe spacing resolves only megabase-scale architecture, leaving finer features unexplored. Still, the message is clear: to understand how genomes fold, scientists must stop averaging and start watching chromosomes one molecule at a time.

Subject of Research: Single-molecule chromatin tracing of heterochromatin domain architecture in C. elegans embryos

Article Title: Single-molecule chromatin tracing reveals a diversity of megabase heterochromatin domains

Article References: Pulido-Barrera, D. C., You, J. E., Angonezi, A. L., Kuznetsov, A., Xu, F., Verheijen, T., Saw, A. N., Kos, P., Giorgetti, L., Molina, N., Brückner, D., & Mango, S. E. (2026). Single-molecule chromatin tracing reveals a diversity of megabase heterochromatin domains. Nature Structural & Molecular Biology. https://doi.org/10.1038/s41594-026-01895-4

Image Credits: AI Generated

DOI: 10.1038/s41594-026-01895-4

Keywords: chromatin tracing, heterochromatin, TADs, condensin I, C. elegans, H3K9 methylation, elCID, genome organization, 3D genomics, embryogenesis, loop extrusion, single-molecule imaging