Collectively, our data suggest that competency is synonymous with unbiased early primed pluripotency of the uncommitted anterior epiblast, whereas noncompetent lines were more similar to the posterior epiblast.

Finally, we asked whether CHR results in a sustained restoration of differentiation capacity. CHR cells cultured for at least 37 passages maintained a competent transcriptome, were able to produce brain organoids and retained the bivalent chromatin pattern distinct from parental line on most loci.

Differences in DNA methylation have also been implicated in varying hPSC differentiation potential. However, we did not detect any clear differences in the DNA methylation profiles over differentially expressed genes between non-competent vs competent & CHR lines.

But comparing our results to data from human stem cell embryo models (especially the pre-gastrulation epiblast), we found that non-competent lines exhibit a more posterior epiblast-like character, whereas competent and CHR lines display a more anterior epiblast-like transcriptome signature.

The state of pluripotent stem cells along the naive - primed - formative pluripotency axis can determine how they respond to developmental cues. However, we did not detect differential expression of primed or formative pluripotency marker genes between competent, non-competent and CHR lines.

Importantly, the erosion of bivalent chromatin is gene locus-specific. While there are some genome-wide differences in H3K27me3 occupancy between the hiPSC lines we profiled, a clear segregation between non-competent and competent & CHR lines is only evident over the differentially expressed genes.

At the gene expression and chromatin level, CHR re-silences aberrantly expressed developmental regulator genes and re-establishes a bivalent chromatin signature (gain of H3K27me3 and loss of H3K4me3).

Remarkably after undergoing CHR, previously non-competent hiPSC lines can now generate cell types representative of all three germ layers.

But can we go one step further and restore differentiation potential in non-competent lines? It takes someone exceptionally talented and brave to tackle this question, but luckily for us Magdalena Sutcliffe in @lancasterlab.bsky.social is both! Magda established chemical chromatin restoration (CHR)

Is this signature specific to human induced pluripotent stem cell lines (hiPSCs)? Surprisingly - no. We profiled a differentiation-compromised human embryonic stem cell (hESC) line and found that it had a very similar transcription and chromatin signature to the incompetent human iPSC lines.

Thus, we found that differentiation-compromised hiPSC lines exhibit a unique transcription & chromatin signature. Importantly, widely used pluripotency metrics such as PluriTest (www.nature.com/articles/nme...) suggested that these lines have a gene expression signature associated with pluripotency.

This is accompanied by an erosion of bivalent chromatin over most of these genes (loss of the repressive chromatin mark H3K27me3 & gain in H3K4me3). Surprisingly, the ATAC-seq profiles at DEGs are nearly indistinguishable between differentiation-capable and differentiation-compromised lines.

We found that differentiation-compromised hPSC lines indeed cluster together (PCA on RNA-seq data), and that they aberrantly express developmental transcription factor genes such as IRX4 and GBX2.

Jerber et al. found that out of 183 hiPSC lines tested, 48 (26%) failed to differentiate into dopaminergic neurons.

In theory, human pluripotent stem cells (PSCs) can generate all cell types of the human body. In practice however, many PSC lines – both induced pluripotent stem cell (iPSC) and embryonic stem cell (ESC) - have limited differentiation potential, and some fail to generate any specialised cell types

Now that we've completed our 2025 series, a massive thank you to all the brilliant speakers who have made the 'The 3D Regulatory Genome' seminars exciting, cutting-edge and hugely enjoyable!

Finally, we leverage naturally occurring genetic variation in a panel of human pluripotent stem cell lines to identify sequence determinants and candidate transcription factors that may play crucial roles in establishing the primed state at key enhancers for human neural development.

Surprisingly, we find that epigenetic priming also occurs at enhancers that become active at much later developmental stages. In some cases, epigenetic priming can already be detected in the epiblast at neural enhancers that are specific for mouse E11.5 and 7 weeks post conception in humans.

We focussed on ‘primed’ enhancers, which harbour specific chromatin signatures: ATAC-seq positive, DNA hypomethylation & H3K4me1 positive. But importantly, they lack the active enhancer mark H3K27ac, and are thought to become active only later in development during cell lineage diversification.

Finally, we asked whether these putative non-canonical enhancers are pluripotent stem cell specific. Mining publicly available genome-wide STARR-seq data sets, we found STARR-seq+ enhancers with similar chromatin profiles in human cancer cell lines, mouse ESCs and a Drosophila cell line.

But had we simply re-discovered non-canonical enhancers previously described in mouse embryonic stem cells discovered by the lab of ‪@wbickmor.bsky.social‬ (Pradeepa Nature Genetics 2026)? Our chromatin profiling suggests no (no enrichment for H3K122ac, but instead weak enrichment for H4K12ac).‬

In one case, remarkably CRISPRi perturbation of primed enhancer function led to changes of pluripotent stem cell characteristics, including loss of PSC colony morphology, reduced stem cell marker expression and aberrant expression of differentiation markers.

We tested this hypothesis directly using CRISPR epigenome engineering to interfere with enhancer function. Recruitment of dCas9-KRAB (CRISPRi) to primed enhancers resulted in reduced expression levels of target genes. Conversely, CRISPR activation led to increased expression levels of target genes.

Further, we found that inter-individual sequence variants in primed enhancers correlated with the expression levels of target genes, and with the chromatin accessibility of the primed enhancers we assessed. Very small n numbers admittedly but nonetheless intriguing?

In some cases – yes. But intriguingly, the majority of primed STARR-seq enhancers did NOT acquire the active enhancer mark H3K27ac in the more specialised cell types examined (neural progenitor cells (NPCs), mesendoderm cells (ME), mesenchymal stem cells (MSCs), trophoblast-like stem cells (TBL)).

Using Promoter Capture Hi-C, we linked STARR-seq enhancers to their putative target genes. This revealed that genes interacting with active enhancers had the highest expression levels, followed by genes interacting with primed enhancers, followed by genes interacting with poised enhancers.

Generating the first STARR-seq map in a non-cancer human cell line, we classified the identified STARR-seq+ enhancers into: - Active (ATAC-seq+; H3K4me1+; H3K27ac+) - Primed/Inactive (ATAC-seq+; H3K4me1+; H3K27ac-) - Poised (ATAC-seq+; H3K27me3+) - Chromatin repressed (ATAC-seq-)

I very much look forward to @mosterwalder.bsky.social's talk on "Functional architecture of cardiac TF regulatory landscapes in control of mammalian heart development" in 'The 3D Regulatory Genome', @enhancedgenomics.bsky.social's virtual seminar series. Join us at: us02web.zoom.us/webinar/regi...

I am very excited to host Dr Sarah Marzi @sj-marzi.bsky.social in our virtual seminar series 'The 3D Regulatory Genome' for her talk on "Epigenetic regulation of environmental and genetic risk in neurodegenerative diseases" 📅 March 13th 2025 ⏰ 4 pm GMT Join us at: us02web.zoom.us/webinar/regi...