Does H3K27me3 erasure restore notochordal identity in degenerated disc cells?

The notochordal cell loss baseline that the new paper inherits

Earlier work established that the nucleus pulposus is the only intervertebral disc compartment derived from the embryonic notochord, and that large vacuolated notochordal cells are progressively replaced by smaller, non-vacuolated nucleus pulposus cells during maturation and degeneration [1][5]. Human fetal single-cell profiling identified SOX10+ and CTSK+ notochordal/nucleus pulposus clusters and confirmed TBXT as a canonical notochordal marker, while adult disc datasets showed that early notochordal markers such as SOX10, NPY and APOA1 are barely detectable in non-degenerative or degenerative nucleus pulposus [5]. Reviews of disc cell fate similarly describe the disappearance of the human notochordal phenotype in adolescence, accompanied by cell death and chondrocyte proliferation, with senescence and apoptosis increasing with age and degeneration severity [10]. This is the conceptual baseline: notochordal identity is lost, and its loss tracks with degeneration.

What remained unresolved was whether that loss is irreversible. The new paper's central question is whether the notochordal program is merely transcriptionally silent but still epigenetically retrievable, or whether the cells have permanently exited that state [1].

A single-cell H3K27me3 atlas across the notochordal-to-nucleus-pulposus transition

The anchor study used non-chondrodystrophic dogs as a naturally occurring model, sampling nucleus pulposus from three stillborn puppies, three healthy young adult dogs and three patient dogs undergoing discectomy for lumbosacral disc protrusion [1]. SORT-seq recovered 2,597 cells and defined juvenile notochordal clusters, mature notochordal clusters, a chondrogenic nucleus pulposus cell cluster, an IVD degeneration-associated fibrocyte cluster, annulus fibrosus subtypes and an inflammatory cluster; the fibrocyte, vascular and inflammatory populations were derived exclusively from patient discs, consistent with immune infiltration and vascular ingrowth [1]. The authors then applied single-cell T-ChIC to measure full-length transcripts and H3K27me3 in the same cells, retrieving 1,256 notochordal cells from 3,377 sequenced cells [1].

The key result is that H3K27me3 enrichment at TBXT, KRT18, CDH2 and SHH increases as notochordal cells differentiate into nucleus pulposus cells, and this enrichment pattern is conserved in aged human nucleus pulposus cells [1]. RNA velocity showed progressive reduction in unspliced mRNA for these genes across notochordal clusters, supporting transcriptional silencing rather than a purely post-transcriptional effect [1]. The authors also highlight GABBR1 and NCAM1 as genes actively transcribed in notochordal cells but selectively repressed in nucleus pulposus cells and in aged human nucleus pulposus cells [1]. This is the paper's distinctive contribution: it links a specific repressive histone mark to a specific cell-state transition at single-cell resolution, rather than inferring epigenetic change from bulk tissue.

Can erasing the mark restore TBXT? The proof-of-concept evidence and its limits

To test reversibility, the authors cultured human nucleus pulposus cells from degenerated aged discs with the EZH2 inhibitor GSK126 for one week, then used CRISPR activation with a gRNA targeting the TBXT promoter co-transfected with a PB-SAM-Dest activator [1]. The combination of H3K27me3 removal and CRISPR-mediated transactivation enhanced TBXT expression in these cells [1]. Because TBXT has been reported to bind the ACAN promoter and activate SMAD3 signaling with matrix-anabolic effects in human nucleus pulposus cells, the authors position TBXT de-repression as a route toward a healthier phenotype [1].

The boundary is explicit in the paper itself. The experiment was performed in vitro on human cells, not in vivo, and the authors state that global demethylation does not selectively target the gene of interest but broadly alters the epigenetic landscape, potentially disrupting essential regulatory networks [1]. They call for gene-specific epigenetic editing with CRISPR-dCas9 systems and note that the epigenetic landscape depends on disease stage and patient variability [1]. The supplied evidence therefore supports transcriptional de-repression as a proof of concept, not regenerative efficacy or long-term safety.

Competing evidence: notochordal-like cells may not fully disappear

A competing line of evidence complicates the assumption that notochordal identity is simply erased. In a cohort of pediatric and adult human nucleus pulposus samples spanning histological degeneration grades 0 to 12, CD24-positive cells were present in all samples, and CD24 positivity showed a significant linear trend to increase with degeneration grade and with age, while GD2 positivity was generally high and TIE2 positivity was low regardless of age or grade [8]. The authors concluded that CD24+ cells maintain a more notochordal phenotype than GD2+ cells and that healthy, notochordal-like profiles appeared reliant on CD24 rather than GD2 [8]. This matters for the anchor paper because it suggests that some notochordal-like cells persist in adult human discs, so the target of an epigenetic intervention may be a residual population rather than a fully converted one.

The two studies also differ in what they measure. The anchor paper measures H3K27me3 enrichment and TBXT transcription at single-cell resolution in dog tissue with human validation [1]; the CD24 study measures surface marker positivity by immunohistochemistry and flow cytometry in human tissue and cultured cells, and notes that marker expression became heterogeneous in culture [8]. These are complementary but not interchangeable readouts, and neither alone establishes that erasing H3K27me3 restores a functional notochordal cell in vivo.

Model choices, mechanism boundaries and what remains open

The anchor paper argues that non-chondrodystrophic dogs, unlike chondrodystrophic breeds with FGFR4-driven disease, better represent the complex age-related epigenetic remodeling seen in human patients, and it notes that the design used two healthy life stages plus naturally occurring disease rather than a continuous aging spectrum, so rare cell types or transient regulatory dynamics may be underrepresented [1]. A porcine annular injury model provides a competing large-animal perspective: injured pig nucleus pulposus cells were transcriptionally more similar to human low back pain discs than to human asymptomatic discs, and injured pig and human low back pain samples were enriched for cell stress, death and neural signaling pathways, while non-injured pig and human asymptomatic samples were enriched for metabolic and stem cell/developmental pathways [4]. That model also documents notochordal cell loss and emergence of pain-inducing nucleus pulposus cells after injury [4], reinforcing the phenotype transition but through an acute injury route rather than natural aging.

Other models and mechanisms widen the boundary further. Tree shrews show spontaneous age-related disc degeneration with gradual notochordal cell disappearance after skeletal maturity, and their growth plates close at maturity, unlike rodents [7]. Porcine N-glycome profiling of notochordal-cell-rich nucleus pulposus found age-related differences in outer-arm fucosylation, oligomannosidic N-glycans and α(2,3)-linked sialic acid, and the authors argue that degeneration and aging are not identical processes [9]. At the chromatin level, work on H3K27me3 spreading shows that the mark organizes canonical PRC1 chromatin architecture and that its confinement, not just its presence, controls target gene silencing during developmental transitions [3]. Finally, methodological reviews of single-cell epigenetics note that histone-modification-based single-cell technologies still face limitations in sensitivity, spatial implementation and standardization [6]. Together these papers indicate that H3K27me3 erasure is a plausible but incomplete lever: the mark is one layer in a broader regulatory and microenvironmental system, and the supplied evidence does not show that removing it restores a durable notochordal phenotype in living discs.

About These Sources

This research page is built on 10 peer-reviewed studies — published from 2022 to 2026, 6 from 2024 or later, collectively cited 143 times — selected as the most relevant from 11 studies that passed quality screening, drawn from 74 papers retrieved from a database of over 500 million.

Sources used in this answer