DeepVelo: deep learning extends RNA velocity to multi-lineage systems with cell-specific kinetics.

Existing RNA velocity estimation methods strongly rely on predefined dynamics and cell-agnostic constant transcriptional kinetic rates, assumptions often violated in complex and heterogeneous single-cell RNA sequencing (scRNA-seq) data. Using a graph convolution network, DeepVelo overcomes these limitations by generalizing RNA velocity to cell populations containing time-dependent kinetics and multiple lineages. DeepVelo infers time-varying cellular rates of transcription, splicing, and degradation, recovers each cell's stage in the differentiation process, and detects functionally relevant driver genes regulating these processes. Application to various developmental and pathogenic processes demonstrates DeepVelo's capacity to study complex differentiation and lineage decision events in heterogeneous scRNA-seq data.

Developmental basis of SHH medulloblastoma heterogeneity.

Many genes that drive normal cellular development also contribute to oncogenesis. Medulloblastoma (MB) tumors likely arise from neuronal progenitors in the cerebellum, and we hypothesized that the heterogeneity observed in MBs with sonic hedgehog (SHH) activation could be due to differences in developmental pathways. To investigate this question, here we perform single-nucleus RNA sequencing on highly differentiated SHH MBs with extensively nodular histology and observed malignant cells resembling each stage of canonical granule neuron development. Through innovative computational approaches, we connect these results to published datasets and find that some established molecular subtypes of SHH MB appear arrested at different developmental stages. Additionally, using multiplexed proteomic imaging and MALDI imaging mass spectrometry, we identify distinct histological and metabolic profiles for highly differentiated tumors. Our approaches are applicable to understanding the interplay between heterogeneity and differentiation in other cancers and can provide important insights for the design of targeted therapies.

K27M in canonical and noncanonical H3 variants occurs in distinct oligodendroglial cell lineages in brain midline gliomas.

Canonical (H3.1/H3.2) and noncanonical (H3.3) histone 3 K27M-mutant gliomas have unique spatiotemporal distributions, partner alterations and molecular profiles. The contribution of the cell of origin to these differences has been challenging to uncouple from the oncogenic reprogramming induced by the mutation. Here, we perform an integrated analysis of 116 tumors, including single-cell transcriptome and chromatin accessibility, 3D chromatin architecture and epigenomic profiles, and show that K27M-mutant gliomas faithfully maintain chromatin configuration at developmental genes consistent with anatomically distinct oligodendrocyte precursor cells (OPCs). H3.3K27M thalamic gliomas map to prosomere 2-derived lineages. In turn, H3.1K27M ACVR1-mutant pontine gliomas uniformly mirror early ventral NKX6-1+/SHH-dependent brainstem OPCs, whereas H3.3K27M gliomas frequently resemble dorsal PAX3+/BMP-dependent progenitors. Our data suggest a context-specific vulnerability in H3.1K27M-mutant SHH-dependent ventral OPCs, which rely on acquisition of ACVR1 mutations to drive aberrant BMP signaling required for oncogenesis. The unifying action of K27M mutations is to restrict H3K27me3 at PRC2 landing sites, whereas other epigenetic changes are mainly contingent on the cell of origin chromatin state and cycling rate.

Failure of human rhombic lip differentiation underlies medulloblastoma formation.

Medulloblastoma (MB) comprises a group of heterogeneous paediatric embryonal neoplasms of the hindbrain with strong links to early development of the hindbrain1-4. Mutations that activate Sonic hedgehog signalling lead to Sonic hedgehog MB in the upper rhombic lip (RL) granule cell lineage5-8. By contrast, mutations that activate WNT signalling lead to WNT MB in the lower RL9,10. However, little is known about the more commonly occurring group 4 (G4) MB, which is thought to arise in the unipolar brush cell lineage3,4. Here we demonstrate that somatic mutations that cause G4 MB converge on the core binding factor alpha (CBFA) complex and mutually exclusive alterations that affect CBFA2T2, CBFA2T3, PRDM6, UTX and OTX2. CBFA2T2 is expressed early in the progenitor cells of the cerebellar RL subventricular zone in Homo sapiens, and G4 MB transcriptionally resembles these progenitors but are stalled in developmental time. Knockdown of OTX2 in model systems relieves this differentiation blockade, which allows MB cells to spontaneously proceed along normal developmental differentiation trajectories. The specific nature of the split human RL, which is destined to generate most of the neurons in the human brain, and its high level of susceptible EOMES+KI67+ unipolar brush cell progenitor cells probably predisposes our species to the development of G4 MB.

A brain precursor atlas reveals the acquisition of developmental-like states in adult cerebral tumours.

Human cerebral cancers are known to contain cell types resembling the varying stages of neural development. However, the basis of this association remains unclear. Here, we map the development of mouse cerebrum across the developmental time-course, from embryonic day 12.5 to postnatal day 365, performing single-cell transcriptomics on >100,000 cells. By comparing this reference atlas to single-cell data from >100 glial tumours of the adult and paediatric human cerebrum, we find that tumour cells have an expression signature that overlaps with temporally restricted, embryonic radial glial precursors (RGPs) and their immediate sublineages. Further, we demonstrate that prenatal transformation of RGPs in a genetic mouse model gives rise to adult cerebral tumours that show an embryonic/juvenile RGP identity. Together, these findings implicate the acquisition of embryonic-like states in the genesis of adult glioma, providing insight into the origins of human glioma, and identifying specific developmental cell types for therapeutic targeting.

The transcriptional landscape of Shh medulloblastoma.

Sonic hedgehog medulloblastoma encompasses a clinically and molecularly diverse group of cancers of the developing central nervous system. Here, we use unbiased sequencing of the transcriptome across a large cohort of 250 tumors to reveal differences among molecular subtypes of the disease, and demonstrate the previously unappreciated importance of non-coding RNA transcripts. We identify alterations within the cAMP dependent pathway (GNAS, PRKAR1A) which converge on GLI2 activity and show that 18% of tumors have a genetic event that directly targets the abundance and/or stability of MYCN. Furthermore, we discover an extensive network of fusions in focally amplified regions encompassing GLI2, and several loss-of-function fusions in tumor suppressor genes PTCH1, SUFU and NCOR1. Molecular convergence on a subset of genes by nucleotide variants, copy number aberrations, and gene fusions highlight the key roles of specific pathways in the pathogenesis of Sonic hedgehog medulloblastoma and open up opportunities for therapeutic intervention.

An OTX2-PAX3 signaling axis regulates Group 3 medulloblastoma cell fate.

OTX2 is a potent oncogene that promotes tumor growth in Group 3 medulloblastoma. However, the mechanisms by which OTX2 represses neural differentiation are not well characterized. Here, we perform extensive multiomic analyses to identify an OTX2 regulatory network that controls Group 3 medulloblastoma cell fate. OTX2 silencing modulates the repressive chromatin landscape, decreases levels of PRC2 complex genes and increases the expression of neurodevelopmental transcription factors including PAX3 and PAX6. Expression of PAX3 and PAX6 is significantly lower in Group 3 medulloblastoma patients and is correlated with reduced survival, yet only PAX3 inhibits self-renewal in vitro and increases survival in vivo. Single cell RNA sequencing of Group 3 medulloblastoma tumorspheres demonstrates expression of an undifferentiated progenitor program observed in primary tumors and characterized by translation/elongation factor genes. Identification of mTORC1 signaling as a downstream effector of OTX2-PAX3 reveals roles for protein synthesis pathways in regulating Group 3 medulloblastoma pathogenesis.

Metabolic Regulation of the Epigenome Drives Lethal Infantile Ependymoma.

Posterior fossa A (PFA) ependymomas are lethal malignancies of the hindbrain in infants and toddlers. Lacking highly recurrent somatic mutations, PFA ependymomas are proposed to be epigenetically driven tumors for which model systems are lacking. Here we demonstrate that PFA ependymomas are maintained under hypoxia, associated with restricted availability of specific metabolites to diminish histone methylation, and increase histone demethylation and acetylation at histone 3 lysine 27 (H3K27). PFA ependymomas initiate from a cell lineage in the first trimester of human development that resides in restricted oxygen. Unlike other ependymomas, transient exposure of PFA cells to ambient oxygen induces irreversible cellular toxicity. PFA tumors exhibit a low basal level of H3K27me3, and, paradoxically, inhibition of H3K27 methylation specifically disrupts PFA tumor growth. Targeting metabolism and/or the epigenome presents a unique opportunity for rational therapy for infants with PFA ependymoma.

Medulloblastoma Arises from the Persistence of a Rare and Transient Sox2+ Granule Neuron Precursor.

Medulloblastoma (MB) is a neoplasm linked to dysregulated cerebellar development. Previously, we demonstrated that the Sonic Hedgehog (SHH) subgroup grows hierarchically, with Sox2+ cells at the apex of tumor progression and relapse. To test whether this mechanism is rooted in a normal developmental process, we studied the role of Sox2 in cerebellar development. We find that the external germinal layer (EGL) is derived from embryonic Sox2+ precursors and that the EGL maintains a rare fraction of Sox2+ cells during the first postnatal week. Through lineage tracing and single-cell analysis, we demonstrate that these Sox2+ cells are within the Atoh1+ lineage, contribute extensively to adult granule neurons, and resemble Sox2+ tumor cells. Critically, constitutive activation of the SHH pathway leads to their aberrant persistence in the EGL and rapid tumor onset. We propose that failure to eliminate this rare but potent developmental population is the tumor initiation mechanism in SHH-subgroup MB.

Locoregional delivery of CAR T cells to the cerebrospinal fluid for treatment of metastatic medulloblastoma and ependymoma.

Recurrent medulloblastoma and ependymoma are universally lethal, with no approved targeted therapies and few candidates presently under clinical evaluation. Nearly all recurrent medulloblastomas and posterior fossa group A (PFA) ependymomas are located adjacent to and bathed by the cerebrospinal fluid, presenting an opportunity for locoregional therapy, bypassing the blood-brain barrier. We identify three cell-surface targets, EPHA2, HER2 and interleukin 13 receptor α2, expressed on medulloblastomas and ependymomas, but not expressed in the normal developing brain. We validate intrathecal delivery of EPHA2, HER2 and interleukin 13 receptor α2 chimeric antigen receptor T cells as an effective treatment for primary, metastatic and recurrent group 3 medulloblastoma and PFA ependymoma xenografts in mouse models. Finally, we demonstrate that administration of these chimeric antigen receptor T cells into the cerebrospinal fluid, alone or in combination with azacytidine, is a highly effective therapy for multiple metastatic mouse models of group 3 medulloblastoma and PFA ependymoma, thereby providing a rationale for clinical trials of these approaches in humans.

Recurrent noncoding U1 snRNA mutations drive cryptic splicing in SHH medulloblastoma.

In cancer, recurrent somatic single-nucleotide variants-which are rare in most paediatric cancers-are confined largely to protein-coding genes1-3. Here we report highly recurrent hotspot mutations (r.3A>G) of U1 spliceosomal small nuclear RNAs (snRNAs) in about 50% of Sonic hedgehog (SHH) medulloblastomas. These mutations were not present across other subgroups of medulloblastoma, and we identified these hotspot mutations in U1 snRNA in only <0.1% of 2,442 cancers, across 36 other tumour types. The mutations occur in 97% of adults (subtype SHHδ) and 25% of adolescents (subtype SHHα) with SHH medulloblastoma, but are largely absent from SHH medulloblastoma in infants. The U1 snRNA mutations occur in the 5' splice-site binding region, and snRNA-mutant tumours have significantly disrupted RNA splicing and an excess of 5' cryptic splicing events. Alternative splicing mediated by mutant U1 snRNA inactivates tumour-suppressor genes (PTCH1) and activates oncogenes (GLI2 and CCND2), and represents a target for therapy. These U1 snRNA mutations provide an example of highly recurrent and tissue-specific mutations of a non-protein-coding gene in cancer.

Single-Cell Transcriptomics in Medulloblastoma Reveals Tumor-Initiating Progenitors and Oncogenic Cascades during Tumorigenesis and Relapse.

Progenitor heterogeneity and identities underlying tumor initiation and relapse in medulloblastomas remain elusive. Utilizing single-cell transcriptomic analysis, we demonstrated a developmental hierarchy of progenitor pools in Sonic Hedgehog (SHH) medulloblastomas, and identified OLIG2-expressing glial progenitors as transit-amplifying cells at the tumorigenic onset. Although OLIG2+ progenitors become quiescent stem-like cells in full-blown tumors, they are highly enriched in therapy-resistant and recurrent medulloblastomas. Depletion of mitotic Olig2+ progenitors or Olig2 ablation impeded tumor initiation. Genomic profiling revealed that OLIG2 modulates chromatin landscapes and activates oncogenic networks including HIPPO-YAP/TAZ and AURORA-A/MYCN pathways. Co-targeting these oncogenic pathways induced tumor growth arrest. Together, our results indicate that glial lineage-associated OLIG2+ progenitors are tumor-initiating cells during medulloblastoma tumorigenesis and relapse, suggesting OLIG2-driven oncogenic networks as potential therapeutic targets.

Childhood cerebellar tumours mirror conserved fetal transcriptional programs.

Study of the origin and development of cerebellar tumours has been hampered by the complexity and heterogeneity of cerebellar cells that change over the course of development. Here we use single-cell transcriptomics to study more than 60,000 cells from the developing mouse cerebellum and show that different molecular subgroups of childhood cerebellar tumours mirror the transcription of cells from distinct, temporally restricted cerebellar lineages. The Sonic Hedgehog medulloblastoma subgroup transcriptionally mirrors the granule cell hierarchy as expected, while group 3 medulloblastoma resembles Nestin+ stem cells, group 4 medulloblastoma resembles unipolar brush cells, and PFA/PFB ependymoma and cerebellar pilocytic astrocytoma resemble the prenatal gliogenic progenitor cells. Furthermore, single-cell transcriptomics of human childhood cerebellar tumours demonstrates that many bulk tumours contain a mixed population of cells with divergent differentiation. Our data highlight cerebellar tumours as a disorder of early brain development and provide a proximate explanation for the peak incidence of cerebellar tumours in early childhood.

Intracellular galectins in cancer cells: potential new targets for therapy (Review).

Dysregulation of galectin expression is frequently observed in cancer tissues. Such an abnormal expression pattern often correlates with aggressiveness and relapse in many types of cancer. Because galectins have the ability to modulate functions that are important for cell survival, migration and metastasis, they also represent attractive targets for cancer therapy. This has been well-exploited for extracellular galectins, which bind glycoconjugates expressed on the surface of cancer cells. Although the existence of intracellular functions of galectins has been known for many years, an increasing number of studies indicate that these proteins can also alter tumor progression through their interaction with intracellular ligands. In fact, in some instances, the interactions of galectins with their intracellular ligands seem to occur independently of their carbohydrate recognition domain. Such findings call for a change in the basic assumptions, or paradigms, concerning the activity of galectins in cancer and may force us to revisit our strategies to develop galectin antagonists for the treatment of cancer.