Comparative Single-Cell Transcriptomics of Human Neuroblastoma and Preclinical Models Reveals Conservation of an Adrenergic Cell State

Abstract

Transgenic mice and organoid models, such as three-dimensional tumoroid cultures, have emerged as powerful tools for investigating cancer development and targeted therapies. Yet, the extent to which these preclinical models recapitulate the cellular identity of heterogeneous malignancies, like neuroblastoma, remains to be validated. In this study, we characterized the transcriptional landscape of TH-MYCN tumors by single-cell RNA sequencing and developed ex vivo tumoroids. Integrated analysis with murine fetal adrenal samples confirmed that both TH-MYCN tumors and tumoroids closely mirror the cellular profiles of normal embryonic sympathoblasts and chromaffin cells. Comprehensive comparison between tumors from patients with neuroblastoma and TH-MYCN mice demonstrated similarities in adrenergic tumor cell composition. Ex vivo tumoroid cultures displayed histologic resemblance and shared transcriptional profiles with the originating TH-MYCN tumors and human neuroblastoma tumors. Importantly, subpopulations within tumoroids exhibited gene expression associated with poor survival of patients with neuroblastoma. Notably, recurrent observations of a low-proliferative chromaffin phenotype connected to the highly proliferative sympathetic phenotype suggested that pushing sympathoblasts into a chromaffin-like state may offer an interesting therapeutic strategy for neuroblastoma. Together, this study not only deepens our understanding of a widely used transgenic mouse neuroblastoma model but also introduces an ex vivo model that maintains critical adrenergic cell state identity, thereby enhancing its translational potential for neuroblastoma research. Significance: Transgenic mouse models and ex vivo tumoroids, characterized through single-cell RNA sequencing, faithfully recapitulate neuroblastoma cellular identity, offering a useful platform for investigating potential therapeutic strategies.

Graphical abstract

Figure 1.

A single-cell transcriptomic atlas of TH-MYCN tumors. A, Schematic overview of study design and scRNA-seq data acquisition. AAA, polyadenylated mRNA; TTT, poly(dT) sequence. B, TH-MYCN tumor samples used for scRNA-seq, ex vivo tumoroid cultures, and histologic validation experiments. Colors indicate mouse information, including genotype, sex, and age of mice, which is indicated in the bar graph at the bottom. C, UMAP embedding of tumor cells derived from five homozygous TH-MYCN mice (left) or five hemizygous TH-MYCN mice (right). Major cell populations are colored and labeled. D, Cell-cycle and adrenergic tumor marker gene expression shown in feature plots; homozygous tumors (top) or hemizygous tumors (bottom). E, Dot plot of marker genes associated with major cell types in TH-MYCN tumors from homozygous mice (left) or hemizygous mice (right). F, Quantification of immunofluorescence images in three TH-MYCN tumor samples shown as a bar plot, representing the percentage of N-MYC⁺, PHOX2B⁺, and TH⁺ cells relative to DAPI⁺ nuclei (homozygous 2, n = 1,790; homozygous 4, n = 1,461; hemizygous 2, n = 1,111; left); PHOX2B⁺ and Ki67⁺ cells relative to DAPI⁺ nuclei (homozygous 2, n = 1,201; homozygous 4, n = 733; hemizygous 2, n = 1,069; right). G, Immunofluorescence staining of TH-MYCN tumors. Left, immunostaining for TH (green), PHOX2B (magenta), and N-MYC (yellow); middle, Ki67 (yellow). Arrows, TH⁺PHOX2B⁺N-MYC⁺ triple-positive cells. Right, immunostaining for TH (green) and CART (magenta). Arrows, TH⁺CART⁺ double-positive cells. mDC, myeloid dendritic cells.

Figure 2.

Gene signature association between tumor cells and normal murine fetal development. A, UMAP of TH-MYCN tumor clusters (left, homozygous; right, hemizygous) illustrating the signature score of genes characterizing mouse adrenal anlagen cell clusters from Supplementary Table S2: row 1, at E13.5 from Furlan and colleagues (15); row 2, from E13.5 to postnatal day 5 (P5) from Hanemaaijer and colleagues (17); row 3, at E13.5 from Kameneva and colleagues (18); row 4, human developing adrenal gland signature score from Jansky and colleagues (19). B, Joint alignment of adrenal cell populations in TH-MYCN tumor samples and mouse embryo trunk at E13.5 (n = 3) from Kameneva and colleagues (18) of adrenal clusters, colored by sample type (left), normal mouse embryo cluster annotation (middle), and TH-MYCN tumor cell annotation (right). The locations of SCP, bridge, transitioning adrenal, chromaffin, and sympathoblast clusters are outlined and labeled. C, Schematic model summarizing the transitional order within developing mouse sympathoadrenal system. In the TH-MYCN model, MYCN is overexpressed under the Th promoter. D, Heatmap demonstrating differentially expressed genes in adrenergic chromaffin cells and sympathoblasts between normal embryonic and TH-MYCN tumor cells.

Figure 3.

Comprehensive comparison of cell state composition of tumors from patients with neuroblastoma and TH-MYCN mice. A, UMAP embedding of cells derived from MYCN-amplified neuroblastoma patient tumors from two datasets, n = 6 [Dong and colleagues (23), T162, T200, and T230; Olsen and colleagues (25), NB16, NB21, and NB22; left]; homozygous TH-MYCN tumors, n = 5 (middle); hemizygous TH-MYCN tumors, n = 5 (right). B, Cell proportion comparison of cell states across neuroblastoma tumors from patients (human) and homozygous and hemizygous TH-MYCN mice. C, Expression of Molecular Signatures Database Hallmark gene set signaling pathways and MYCN target genes across neuroblastoma patient (left), homozygous (middle), and hemizygous (right) tumors in feature plots. KRAS down, genes downregulated by KRAS activation. D, Dot plot of human tumor clusters top differentially expressed genes, presented across tumors from patients with neuroblastoma (left) and homozygous (middle) and hemizygous (right) TH-MYCN mice. mDC, myeloid dendritic cells.

Figure 4.

Ex vivo tumoroid cultures recapitulate tumor single-cell transcriptomic heterogeneity. A, Schematic overview of TH-MYCN tumoroid culturing, passaging, and harvesting for subsequent scRNA-seq data acquisition and histologic validation of cell states. AAA, polyadenylated mRNA; TTT, poly(dT) sequence. B, Immunofluorescence staining of TH-MYCN tumor (left) and matched ex vivo tumoroids at passages 0 (middle) and 5 (right) for PHOX2B (cyan) and N-MYC (magenta). Quantification representing the percentage of N-MYC⁺ and copositive N-MYC⁺PHOX2B⁺ cells relative to DAPI⁺ nuclei. C, UMAP embedding of TH-MYCN tumors and matched ex vivo tumoroids at passages 0 and 5. Left, major cell populations are colored and labeled. Right, distribution of cells from each sample shown on the UMAP embedding. D, Dot plot of marker genes associated with major cell types in TH-MYCN tumors and tumoroids. E, Left, cell-cycle phase classified by ccSeurat: G₁, S, and G₂/M. Right, cell-cycle and subset marker gene expression distinguishing cell-cycle phases and tumor cell clusters, shown in feature plots.

Figure 5.

Distinct ex vivo tumoroid-enriched sympathoblast signatures association with neuroblastoma survival. A, UMAP of TH-MYCN tumor and tumoroid clusters illustrating the signature score of genes characterizing mouse adrenal anlagen cell clusters: top row, at E13.5 from Furlan and colleagues (15); middle row, from E13.5 to postnatal day 5 (P5) from Hanemaaijer and colleagues (17); bottom row, at E13.5 from Kameneva and colleagues (18). B, Immunofluorescence staining of TH-MYCN tumor (top row) and tumoroid sections (bottom row) for TH (green), PHOX2B (cyan), and Ki67 (magenta). C, Distribution of cells from in vivo tumors (gray) or ex vivo tumoroids (pink) highlighted on the UMAP embedding. Ex vivo enriched clusters C1, C2, C3, and C4 are outlined and labeled. D, Expression of C1, C2, C3, and C4 signature genes shown in feature plots: ex vivo (left) and in vivo (right). E, Dot plot of gene signatures of key marker in clusters C1, C2, C3, and C4. F, Overall survival analysis on bulk RNA-seq data SEQC-498 (GSE49711). Patients with neuroblastoma were stratified into groups based on the average expression of signatures as defined in E.