Genes & Development

Polycomb Repressive Complexes PRC1 and PRC2 establish and maintain chromatin architectures critical for cell-type specific gene expression. In the developing spinal cord, PRC1 is essential for implementing Hox-dependent transcriptional programs that diversify motor neuron identities. Although PRC1 can assemble into subnuclear condensates, whether these structures contribute to the specification and maintenance of neuronal subtype fates remains unclear. Using chick and mouse models, we show that canonical PRC1 components - BMI1, PHC2, CBX2, RING1B - coalesce into subnuclear assemblies as neural progenitors differentiate into motor neurons. Disrupting PHC2 polymerization in either neural progenitors or postmitotic neurons dismantles BMI1-containing PRC1 assemblies, triggering derepression of caudal Hox genes and a cell autonomous loss of segment-specific motor neuron fates. Mouse mutants lacking the core PRC1 subunit Ring1 fail to form BMI1-containing assemblies, despite intact PHC2 and CBX2 polymers. In contrast, motor neuron-restricted deletion of core PRC2 subunits (Ezh1/2) depletes H3K27me3 histone marks but retains PRC1 assemblies and Hox boundaries. Together, these findings indicate that PHC2-dependent PRC1 assemblies stabilize Hox transcriptional boundaries and motor neuron segmental identities.

Cancer cell plasticity underlies tumor progression, metastasis and therapeutic resistance, yet how RNA-binding proteins regulate malignant state transitions remains poorly defined. Here, we identify the splicing regulator RBFOX2 as a suppressor of oncogenic cell plasticity in Group 3 medulloblastoma (MB). OTX2 antagonizes RBFOX2 by directing an opposing alternative splicing program that mirrors developmental origins and controls cytoskeletal dynamics and cell state transitions. Mechanistically, OTX2 alters the RBFOX2 RNA-binding landscape, limiting RBFOX2 engagement at cytoskeletal loci and suppressing exon inclusion. RBFOX2 loss rewires splicing and transcriptional outputs to promote highly plastic proliferative, mesenchymal-like and undifferentiated cellular states associated with enhanced oncogenic properties. Consistent with these findings, RBFOX2low states mark highly plastic cell populations in patient tumors and xenografts, positioning RBFOX2 as a determinant of malignant cell stability. Together, these findings establish RBFOX2 as a developmental gatekeeper that restrains malignant cell plasticity through coordinated control of cytoskeletal gene programs.

Ribonuclease Z enzymes remove 3' trailers from precursor tRNAs. In human cells, ELAC2/RNase ZL matures nuclear and mitochondrial pre-tRNAs, whereas the cytosolic short paralog ELAC1/RNase ZS repairs ANKZF1-cleaved tRNAs, but its broader substrate repertoire remains unclear. We investigated ELAC1 function using transcriptome profiling, small RNA sequencing, CLIP-seq, and biochemical reconstitution. ELAC1 was dispensable for canonical tRNA processing but associated with structured RNA polymerase III transcripts, prominently Y RNAs and vault RNAs (vtRNAs). Purified ELAC1 cleaved all four human vtRNAs near their 3' ends, with high-resolution mapping identifying a conserved cleavage position within the 3' extension. Efficient cleavage required structural features of the vtRNA terminal region. In contrast, ELAC2 showed little or no vtRNA cleavage under the conditions tested and displayed stricter substrate and cleavage-site selection on pre-tRNA. Loss and re-expression of ELAC1 altered structured ncRNA abundance and small-RNA end profiles. These findings identify vtRNAs as direct ELAC1 substrates and support functional specialization of human ELAC paralogs, with ELAC2 favoring precise pre-tRNA maturation and ELAC1 exhibiting broader RNA substrate specificity.

The nuclear envelope contributes to chromatin compartmentalization in eukaryotes by anchoring heterochromatin to the lamina and active transcription sites to nuclear pore complexes (NPCs). During DNA replication, chromatin association with the nuclear envelope is thought to act as a topological barrier to DNA polymerases, potentially exacerbating transcription-replication conflicts (TRCs). While limited evidence suggests that the DNA replication checkpoint kinase Rad53/CHK1 promotes the release of genes from the NPC during replicative stress, the global scale and precise regulation of this process remain unknown. Through a candidate-based screen, we demonstrate that gene release from NPCs is a general feature of the cellular response to replication stress, tightly coordinated with replication fork progression. We show that the checkpoint kinases Mec1/ATR and Rad53/CHK1 are both essential for managing gene association with NPCs, but operate in a distinct, stepwise manner: Mec1/ATR triggers the initial dissociation by promoting the removal of RNAPII from the DNA, while Rad53/CHK1 prevents the premature reassociation of transcribed loci by phosphorylating the nucleoporin Mlp1. Finally, we demonstrate that maintaining this control over genes-NPC association is critical for efficient DNA replication under stress. Collectively, our findings reveal that the DNA replication checkpoint integrates genome duplication with chromosome spatial positioning to maintain genome integrity during replication stress.

Epistasis, the genetic interaction between loci, directs non-additive genotype-to-phenotype outcomes. Within a multicellular organism, how the tissue niche (E) shapes gene-by-gene epistasis (GxG) and its phenotypic consequences remains poorly studied. Using tumor lung metastasis as an experimental model, we mapped GxGxE through combinatorial CRISPR perturbation coupled with depletion of defined host myeloid populations in mice. The metastatic niche first selects disseminated tumor cells on GxG epistatic fitness, then drives transcriptome convergence regardless of genotype. Removing anatomically distinct myeloid populations reshaped the metastatic epistasis landscape despite identical tumor genotypes. Gene-by-gene-by-environment epistasis therefore operates at the organismal level, with interaction magnitude and sign set by niche cells. Beyond metastasis, our data suggest that the tissue niche conditions how genetic interactions translate into divergent phenotypes.

Genetic background can alter the latency and behavior of cancers initiated by the same oncogenic driver, but how inherited variation reshapes the regulatory landscape that permits tumor development remains poorly understood. The PWK genetic background accelerates mammary tumor onset in the NeuT model. We compared PWK NeuT and BALB NeuT tumors using tumor-level RNA-seq, histone ChIP-seq, and single-cell multiome profiling. PWK tumors exhibited broad transcriptional reprogramming characterized by reduced immune-associated expression and increased epithelial differentiation and neuronal-like programs. Chromatin remodeling was spatially redistributed rather than uniformly increased or decreased. PWK tumors showed expansion of H3K9me3- and H4K20me3-associated chromatin, while both marks were depleted across many expressed genes and concentrated within selected intergenic and repeat-rich territories, revealing coordinated redistribution of constitutive heterochromatin. Single-cell multiome profiling showed that broad epithelial representation remained similar between strains, yet specific malignant states were strongly strain biased. Integrated RNA and accessibility analyses identified concordant strain-selective regulatory programs and candidate loci, while trajectory analysis revealed distinct epithelial state transitions between PWK NeuT and BALB NeuT tumors. Together, these findings indicate that inherited genetic background reshapes transcriptional programs, heterochromatin organization, and malignant epithelial state trajectories in NeuT mammary tumors, providing a regulatory framework for understanding strain-dependent differences in tumor onset.

Mutations in the Ten-Eleven Translocation (TET) family of DNA dioxygenases are among the most common epigenetic alterations in clonal hematopoiesis and hematologic malignancies, yet how TET loss of function reshapes genome evolution during cellular expansion remains to be fully understood. Here we show that loss of Tet2/3 in invariant natural killer T (iNKT) cells drives aggressive lymphoid expansion that is consistently associated with recurrent chromosome 17 trisomy, a dominant aneuploid state observed across more than ten independently derived tumors following transplantation into immunocompetent recipient mice. Single-cell whole-genome sequencing of TET-deficient iNKT cells induced to proliferate in culture with antigen and IL-2 stimulation revealed that rare pre-existing cells with chromosome 17 trisomy appeared soon after Tet gene deletion in the thymus, together with cells harboring other copy number alterations, but only the cells with chromosome 17 trisomy were progressively enriched during further expansion. Similarly, inducibly Tet-deficient mouse embryonic stem cells (mESCs) showed selection for aneuploid states that were present prior to Tet gene deletion, including trisomy of chromosome 6. Reanalysis of independent datasets from Dnmt3a-deficient mouse lymphomas identified recurrent chromosome gains, including chromosome 17 trisomy. Our results establish TET and DNMT3A deficiency as drivers of karyotypic evolution, and suggest that disruption of DNA methylation/ demethylation pathways through either TET or DNMT3A deletion promotes chromosomal instability, in part by facilitating the emergence and selection of cryptic aneuploid subclones during clonal expansion.

The control of gene expression is central to cellular physiology and to various biotechnological applications. In higher eukaryotes, synthesis of most messenger RNAs involves removal of introns during splicing. Intron-containing genes are often more highly expressed than intronless versions, despite generating the same mRNA transcripts, suggesting that the splicing of transcripts from a gene positively impacts its expression. Termed intron-mediated enhancement (IME), this phenomenon has been observed in animals, plants, and fungi, with enhancement ranging from 2- to 100-fold and the strongest effects observed for promoter-proximal introns1-3. Here, we assayed thousands of natural human introns, covering all distinct human 5' splice site (5SS) motifs and thousands of 3' splice sites (3SS), for their ability to enhance (or repress) expression of a reporter gene. The results confirmed previous observations that intronic UUUU motifs augment the magnitude of IME, and revealed two fundamental features of IME: (i) only introns spliced by the major spliceosome elicit IME, with minor introns often repressing expression; and (ii) the magnitude of IME increases with the strength of the 5SS motif, but is independent of 3SS strength. Our findings imply that genetic variants that alter the strength of the first 5SS of genes may alter their expression by modulating IME without necessarily impacting splicing. We confirm this expectation using human population genetic data and further show that disease-relevant perturbations of U1 snRNP components that enhance recognition of subsets of first 5SS up-regulate expression of the associated genes. These findings demonstrate that recognition of the first 5SS is vitally important to the expression of human genes.

X chromosome inactivation in mammals is driven by the long non-coding RNA Xist, which localizes and spreads across the chromosome from which it is transcribed. Although several factors required for Xist localization have been identified, the underlying mechanism remains unclear. Here, we combine Nuclear-BRITE, a method for visualizing and quantifying the dynamics of individual Xist RNA molecules, with acute depletion of key localization factors. We find that Xist molecules remain constrained throughout interphase, consistent with continuous association with chromatin. Building on this observation, we identify a contact-mediated transfer mechanism in which 3D chromosome dynamics promote handover of Xist molecules between chromosomal sites. We further show that the Xist-binding proteins SPEN and HNRNPU play distinct roles in this process by promoting Xist - Xist and Xist - chromatin interactions, respectively. Together, our findings define a mechanism for Xist RNA spreading and provide a framework for understanding its unique chromosome-wide localization.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive cancers, in part due to a dense and highly plastic tumor microenvironment dominated by cancer-associated fibroblasts (CAFs). In previous work, we identified a CAF population expressing a neural-associated NES+SOX4+NOTCH1+ program that increased with disease progression. The signals driving this state and the mechanisms regulating it remained unclear. Here, we investigated how tumor-derived signals activate this program and whether it represents a stable fibroblast identity or a reversible state. Using spatial transcriptomic data spanning normal pancreas to invasive carcinoma, we identified EGF, FGF2 and BDNF as candidate tumor-to-fibroblast signals associated with disease progression. Functional experiments in patient-derived CAFs and human fibroblast cell lines showed that EGF and FGF2 are sufficient to induce the NES+SOX4+NOTCH1+ program. Growth factor withdrawal reversed the NES+SOX4+NOTCH1+ program at both the protein and transcriptional levels, indicating that this state is dynamic and environment dependent. Single-cell transcriptomic analysis further revealed a multi-phase response involving early loss of fibroblast-associated features, transient activation of neural-associated genes, and a later stress-associated state. Regulatory analyses identified ETS-family transcription factors as key regulators of the program, while phosphoproteomic profiling and inhibitor studies showed that MAPK/ERK signaling plays a central role in its transcriptional activation. Finally, pan-cancer single-cell analysis showed that this state is enriched in CAFs and increases from normal to tumor tissue across multiple cancer types. Together, these findings define a reversible, growth factor-driven CAF program and identify signaling mechanisms that may provide opportunities to selectively modulate tumor-supportive fibroblast states.

Temporal patterning is essential for generation of neuronal diversity in the developing vertebrate nervous system. A shared sequence of temporal transcription factors (tTFs) partitions neurons by birth date across much of the central nervous system. The molecular mechanisms that control the timing of expression of these tTFs are incompletely understood. Here, we provide evidence that this process is under the control of microRNAs (miRNAs). Perturbing miRNA biogenesis by conditional deletion of Dgcr8 from the developing mouse nervous system extends the window during which neurons with early identities are generated, but abolishes late identities in the spinal cord, hindbrain and midbrain, and holds progenitors in an early transcriptional state. We further identify miRNAs of the let-7 and miR-9 families, which target transcriptional regulators of early progenitor and neuronal identities, as candidates for mediating this effect. Throughout the nervous system, levels of these miRNAs rise during the neurogenic period in progenitors, while levels in neurons correlate with their birth-date. Moreover, the sensitivity of tTF-encoding transcripts to miR-9 inversely correlates with their expression sequence, suggesting a model in which miR-9 controls the chronology of neuronal cell fate transitions by targeting specifiers of early progenitor and neuronal temporal identity in a level-dependent manner. Consistently, perturbations of miR-9 expression dynamics by gain and loss-of-function approaches perturb temporal patterning. Together, our results demonstrate a critical, evolutionary conserved role for miRNAs in temporal patterning and suggest that a temporal miR-9 expression gradient controls the timing of tTF expression in large parts of the developing vertebrate nervous system.

The epigenetic inheritance of mammalian DNA methylation requires DNMT1 and its E3 ligase cofactor UHRF1. At newly replicated chromatin, UHRF1 recognition of hemi-methylated DNA and histone H3 N-terminal tails directs catalysis of H3K14, H3K18, and/or H3K23 mono-ubiquitination to recruit DNMT1. While it is appreciated that UHRF1 can deposit multiple mono-ubiquitin marks on a single H3 tail and that DNMT1 recognizes this state through tandem ubiquitin interacting motifs, the mechanism that promotes successive ubiquitination and the biological function of multi-mono-ubiquitination are unknown. Here, we show that UHRF1 directly binds its mono-ubiquitinated H3 products through a previously uncharacterized LGDDSL loop in Tudor 2 of its tandem Tudor domain (TTD) to promote further ubiquitin deposition. Disruption of this ubiquitin reading activity impairs H3 multi-mono-ubiquitination and accelerates DNA methylation loss within late-replicating, CpG-sparse genomic regions that are characteristic of partially methylated domains (PMDs) in cancer and aging cells. These methylation defects overlap those observed by disruption of UHRF1 ubiquitin ligase activity, providing convergent evidence that both writing and reading of H3 ubiquitination support CpG-sparse DNA methylation maintenance. Together, these findings establish a feed-forward ubiquitin read-write mechanism that generates multi-mono-ubiquitinated H3 and safeguards DNMT1-dependent DNA methylation maintenance at vulnerable genomic regions of the mammalian methylome.

Tumorigenesis is a multi-step process shaped by the accumulation of somatic driver mutations, but identifying the earliest steps and cell of origin of tumorigenesis remains challenging. Single cell genomics in >1.7 million single cells across donors with the cancer predisposition syndrome neurofibromatosis type 1 (NF-1), shows that somatic 'second-hit' mutations in NF1 are pervasive and show positive clonal selection in oligodendrocytes (OLs) and OL precursor cells (OPCs), but are exceedingly rare in neurons and other lineages. Combined spatial genotyping and transcriptomic analysis enabled dissection of cell autonomous and non-autonomous effects of second-hit mutations including clonal expansion of mutant OPCs with strong bias for cerebral gray matter versus white matter, suggesting that gray matter influences OPC proliferation and differentiation via non-cell autonomous interactions between OPCs and neurons. NF1-null OL lineage cells upregulate CDKN2A--a tumor suppressor frequently lost in NF1-mutant tumors--whereas brain cancers from these same individuals show CDKN2A loss across multiple cancer cell states, suggesting CDKN2A loss in OPCs is a critical step in tumor progression. Our findings reveal developmentally early, exquisitely lineage-specific NF1 somatic mutations in premalignant NF-1 brain, suggest cell autonomous and non-autonomous roles in clonal selection in brain, and suggest that clones harboring cancer driver mutations can persist over a lifetime in phenotypically normal tissues.

Faithful genome duplication requires coordination of DNA replication, chromatin assembly, and checkpoint signaling to prevent inappropriate mitotic progression. Using a targeted siRNA screen for factors modulating CHK1 inhibitor-induced DNA damage and premature mitotic entry, we identified replication-coupled chromatin assembly as a determinant of checkpoint dependence. Acute depletion of the CAF-1 subunit CHAF1A modestly delayed S-phase progression but caused little DNA damage, whereas inhibition of CHK1 or ATR triggered extensive DNA damage, premature mitotic entry, and loss of viability. Transcriptomic analyses revealed that CHK1 inhibition induced a transcriptional response enriched for B-MYB/Myb-MuvB (MMB) target genes in CHAF1A-deficient cells. Notably, induction of these genes occurred without corresponding increases in promoter accessibility, suggesting that enhanced MMB-dependent transcription is largely independent of widespread promoter opening. B-MYB depletion attenuated MMB target-gene induction, premature mitotic entry, and DNA damage. Depletion of ASF1 and NAP1L1 similarly enhanced checkpoint inhibitor sensitivity, indicating that increased dependence on ATR-CHK1 signaling is a broader consequence of impaired histone chaperone function. Together, our findings establish a functional link between replication-coupled chromatin assembly, ATR-CHK1 signaling, and MMB-dependent mitotic transcription.

Pathogenic adipose tissue remodeling promotes metabolic disease in obesity, but the mechanisms that establish this unhealthy tissue state remain poorly understood. Here, we show that obesity drives SOX4-dependent reprogramming of mesenchymal stromal cells (MSCs) into cancer-associated fibroblast-like (CAF-like) cells that promote adipose tissue dysfunction. TGF{beta} signaling is elevated in obesity and activates SOX4 in mouse and human MSCs, inducing their conversion to a CAF-like state. In mice, MSC-specific SOX4 activation induces the CAF-like program and exacerbates adipose tissue inflammation and glucose intolerance, whereas Sox4 deletion attenuates inflammation and improves glucose homeostasis during obesity. We further identify the growth factor Midkine (MDK) as a SOX4-regulated paracrine effector produced by CAF-like cells. MDK inhibition in obese mice reduces adipose tissue inflammation and improves metabolic function. Together, these findings define a TGF{beta}-SOX4-MDK stromal signaling axis that drives pathological adipose tissue remodeling in obesity and highlight this pathway as a potential therapeutic target for improving metabolic health.

Leiomyosarcoma is an aggressive soft tissue sarcoma that metastasizes preferentially to lung and liver. Because matched primary and metastatic specimens are rarely available, the cellular changes that accompany metastatic progression remain poorly defined. Here we combined single-nucleus RNA sequencing of archival FFPE tissue from four matched primary-metastasis pairs (two lung, two liver metastases; 48,941 nuclei) with imaging-based spatial transcriptomics of an extended tissue microarray cohort of 21 patients (69 cores, 313,218 cells). Malignant cells, not assignable from transcriptomic references alone, were defined by inferred copy number profiles that were concordant with matched DNA methylation arrays and confirmed the clonal origin of each metastasis. Malignant metaprograms were organ-associated, with a stress- and hypoxia-associated program enriched in lung metastases and a proliferative, matrix-remodeling program enriched in liver metastases. Antigen-presenting cancer-associated fibroblasts (apCAFs) were enriched in metastases relative to matched primaries. This shift was reproduced in the spatial cohort (p = 0.001), where apCAFs formed a discrete band at the tumor-organ interface between malignant cells and adjacent parenchyma, whereas they were sparse and randomly distributed in primary tumors. Spatial transcriptomics further resolved the malignant compartment into three histomolecular phenotypes: (i) smooth-muscle-like with myxoid features, (ii) GPNMB with anaplastic features, and (iii) PDGFRB with vessel-rich stroma, each with a concordant immunohistochemical (-SMA, GPNMB, PDGFRB) and morphological correlate. Together, our data provide a matched primary-metastasis reference for leiomyosarcoma, identify apCAFs as a spatially organized, metastasis-associated stromal population, and delineate histomolecular phenotypes recognizable by routine immunohistochemistry, including a GPNMB phenotype potentially amenable to GPNMB-directed therapies.

Region-specific transcription factors (TFs) establish small intestinal and colonic identity, yet what activates their region-specific expression during development is unresolved. Moreover, discrete fetal gene sets reactivate during post-injury regeneration, but their full scope, and what reactivates the programs required to restore regional identity and tissue structure, remain unclear. By determining CDX2 and partner TF binding, enhancer activity, and gene expression dynamics across intestinal regions, in stem and differentiating cells, and through development and adult regeneration, here we show that CDX2, despite uniform expression across the fetal and adult intestinal epithelium, organizes a hierarchical enhancer network. This network cooperates with partner factors such as CTCF to establish a shared, plastic fetal enhancer landscape, which resolves into region-specific programs upstream of regional TFs. During regeneration, rather than reactivating a general fetal state, CDX2 selectively rebinds many region-specific and common enhancers for genes associated with fetal cell-motility and tissue-morphogenesis signatures along with a more recently active adult stem-cell program, each deployed across distinct crypt tiers. CDX2 thus functions as a central, context-responsive regulator, deploying fetal, adult stem-cell differentiation, and regenerative programs as tissue context requires.

Small RNAs represent important carriers of epigenetic information in nematodes, and factors involved in their biogenesis reside within perinuclear germ granules. Among these condensates, Z granules are essential for the inheritance of RNA interference (RNAi). The mechanisms by which Z granules form and promote this process remain unclear. We show that Casein kinase I alpha (KIN-19) is present in Z granules and associated with DEPS-1 and HERD-1 phosphorylation. In the absence of DEPS-1 and HERD-1 phosphorylation, animals have enlarged Z granules and enhanced RNAi inheritance. We propose that DEPS-1 acts as a phosphorylation-regulated scaffold that bridges ZNFX-1 and HERD-1 proteins. Our study demonstrates that Casein kinase I alpha-mediated phosphorylation is an important factor in germ granule modulation and RNAi inheritance, and additionally provides a novel molecular scaffold for understanding germ granule complexity in C. elegans.

Nucleus-forming jumbo phages import a phage-encoded non-virion RNA polymerase (nvRNAP) into a proteinaceous nucleus, but how it is selected for import is unknown. Here, we identify an import pathway that couples nvRNAP subunit interactions with phage-encoded import factors. Maximal nvRNAP import requires a novel factor Imp7 (gp166) and the essential import factor Imp1. Imp7 is essential at environmental temperatures (20 C), where impaired nvRNAP import causes a profound loss of middle and late transcription and blocks phage DNA replication. Import of individual nvRNAP subunits also depends on other subunits, consistent with import licensing at the level of an assembled complex. In the absence of middle transcription and DNA replication, the phage nucleus surprisingly still assembles and segregates from the cytoplasm, demonstrating that nvRNAP nuclear localization and activity are not a critical checkpoint for nucleus assembly. Together, these findings define a specialized pathway for nuclear import of the multi-subunit RNAP and reveal that nuclear compartment assembly precedes the transcriptional and replicative programs required for its maturation.

Intrinsically disordered regions (IDRs) mediate molecular interactions but can also promote aberrant self-association. Using donors embedded within IDR-coding sequences, exitron-type splicing can remove these sequences, giving them both protein-coding and splice-regulatory roles. Yet IDR amino-acid sequences often diverge rapidly, raising the question of whether and how their coding sequences support IDR-removing splicing across species. Using RNA-seq data from jawed vertebrates, we identified events predicted to remove most of an IDR while preserving annotated structured regions. Across species, donor-flanking nucleotide identity exceeded expectations from synonymous alternatives encoding the same amino-acid sequence. TARDBP, which encodes TDP-43, provided a model for examining how coding sequences both encode IDRs and regulate exitron-type splicing. Sequence-divergent vertebrate TARDBP terminal regions supported this splicing in human cells. In teleosts, TARDBP ohnologs (tardbpa and tardbpb) encoded compositionally similar disordered tails but differed markedly in IDR-removing exitron-type splicing. Reciprocal donor-proximal substitutions in zebrafish minigenes redirected donor choice and altered the balance between IDR-retaining and IDR-removing transcripts. In corresponding reporters, these substitutions altered protein-product balance and cytoplasmic puncta formation. These findings show that IDR-removing exitron-type splicing can recur despite sequence divergence and that donor-proximal TARDBP sequences both encode IDR segments and regulate IDR inclusion, linking RNA processing to protein behavior.