Longitudinal assessment of tumor development using cancer avatars derived from genetically engineered pluripotent stem cells

Abstract

Many cellular models aimed at elucidating cancer biology do not recapitulate pathobiology including tumor heterogeneity, an inherent feature of cancer that underlies treatment resistance. Here we introduce a cancer modeling paradigm using genetically engineered human pluripotent stem cells (hiPSCs) that captures authentic cancer pathobiology. Orthotopic engraftment of the neural progenitor cells derived from hiPSCs that have been genome-edited to contain tumor-associated genetic driver mutations revealed by The Cancer Genome Atlas project for glioblastoma (GBM) results in formation of high-grade gliomas. Similar to patient-derived GBM, these models harbor inter-tumor heterogeneity resembling different GBM molecular subtypes, intra-tumor heterogeneity, and extrachromosomal DNA amplification. Re-engraftment of these primary tumor neurospheres generates secondary tumors with features characteristic of patient samples and present mutation-dependent patterns of tumor evolution. These cancer avatar models provide a platform for comprehensive longitudinal assessment of human tumor development as governed by molecular subtype mutations and lineage-restricted differentiation.

Fig. 1. Different iHGG models derived from edited human iPSCs.

a Schema of iHGG generation. b Designs for gene editing indicating placement of sgRNAs. c Genotyping PCR and d Semi-quantitative RT-qPCR evaluating designated edits. Data are representative of three replicates, n = 3. Data are represented as mean ± SD. e RT-qPCR results of markers for iPSCs and NPCs. Data are representative of three replicates, n = 3. Data are represented as mean ± SD. f Kaplan–Meier curves showing survival of mice engrafted with (left) PTEN^−/− NPCs, PTEN^−/−;NF1^−/− NPCs, (right) TP53^−/− NPCs, and TP53^−/−;PDGFRA^Δ8–9 NPCs. Statistical significance was evaluated by the log-rank test. n = 4 animals for each arm for each model. Source data are provided as a Source Data file.

Fig. 2. Histology of iHGGs.

H&E staining of PTEN^−/−;NF1^−/− iHGGs showing a region of hypercellularity infiltrated by irregular, elongated to angulated tumor cells with occasional mitoses (a), scale bars, 50 μm (left) and 10 μm (right), biphasic dense (glial) and loose (mesenchymal/sarcoma) morphologies, typical of gliosarcoma (b), necrosis (central pink zone) with peripheral “pseudopalisading” of cells around the necrotic center (c), scale bars, 200 μm (b, c), vascular endothelial proliferation (d), scale bar, 100 μm, rupture through the pial surface, and consequently subarachnoid spread (upper right) (e), scale bar, 200 μm, and “secondary structures” typical of glioma, including perineuronal satellitosis and subpial accumulation of tumor cells (f), scale bar, 50 μm. GFAP (g), Olig2 (h), Ki-67 (i) staining of PTEN^−/−;NF1^−/− iHGGs, scale bars, 100 μm (g–i). H&E staining of TP53^−/−;PDGFRA^Δ8–9 iHGGs showing nodular growth of a primitive neuronal component (dark purple) intermingled with glial component (pink) (j), scale bar, 200 μm, rosettes with neuropil-like texture in a primitive neuronal component (k), scale bar, 50 μm, a serpiginous zone of pseudopalisading necrosis (l), scale bar, 200 μm, and a tumor rupture through ependyma illustrating intraventricular growth (m), scale bar, 500μm. GFAP (n), Olig2 (o), Ki-67 (p) staining of TP53^−/−;PDGFRA^Δ8–9 iHGGs, scale bars, 100 μm (n–p).

Fig. 3. Cells from iHGG models can be cultured in vitro and re-engrafted to form secondary tumors with different drug response.

a iHGG spheres obtained by maintaining iHGG tumor cells in neurosphere culture conditions, scale bars, 2 mm. b Extreme limiting dilution analysis of input NPCs and tumor-derived iHGG sphere cells. c H&E staining of secondary tumors generated from re-engraftment of primary iHGG spheres, scale bars, 5 mm, 250 μm, 5 mm, 250 μm, (left to right). d In vivo survival assays of mice orthotopically engrafted with primary iHGG sphere cells upon treatment either with vehicle or temozolomide. Data are representative of six replicates, n = 6 animals for each treatment arm for each model. Data were analyzed by the log-rank test. e MGMT expression levels in iHGG cells analyzed by semi-quantitative RT-qPCR. Data are representative of three replicates, n = 3. Data are represented as mean ± SD, analyzed by unpaired t-test. Source data are provided as a Source Data file.

Fig. 4. iHGG models present inter-tumor heterogeneities and divergent transcriptomes driven by molecular subtypes.

a Schema of scRNA-seq analysis of iGBMs. b Uniform Manifold Approximation and Projection (UMAP) analysis of all sequenced samples. c Principal component analysis of all sequenced samples (the color code is same as in b). d The heatmap of GBM molecular subtype analysis based on average gene expression of individual cells in each sample for a manually curated gene list based on ref. .

Fig. 5. Genetically distinct iHGG models present different patterns of longitudinal evolution.

Primary and secondary spheres of PTEN^−/−;NF1^−/− (a) and TP53^−/−;PDGFRA^Δ8–9 (b). UMAP plot color-coded by samples (top left) or by Louvain clustering (top right). Sample distribution found in each Louvain cluster, color-coded by sample identity (bottom left) and Louvain cluster distribution per sample, color-coded by cluster identity (bottom right). Clustered heatmaps of enriched GO terms extracted from differentially expressed genes of each Louvain cluster in PTEN^−/−;NF1^−/− (c) and TP53^−/−;PDGFRA^Δ8–9 (d). Color scale represents statistical significance. Gray color indicates a lack of significance.

Fig. 6. TP53^−/−;PDGFRA^Δ8–9 iHGG shows prominent karyotype abnormalities accompanied by extrachromosomal DNA.

a DAPI staining of PTEN^−/−;NF1^−/− primary iHGG cells, scale bar, 10 μm. b DAPI staining of TP53^−/−;PDGFRA^Δ8–9 primary iHGG cells. Red arrows indicate ecDNA, scale bars, 10 μm (left), 2 μm (right). c DAPI staining of TP53^−/−;PDGFRA^Δ8–9 secondary iHGG cells. Red arrows indicate ecDNA, scale bars, 10 μm (left), 2 μm (right). d EdU labeling of chromosomes and ecDNA in a metaphase spread of TP53^−/−;PDGFRA^Δ8–9 secondary iGBM, scale bar, 5 μm. e Spectral karyotyping analysis of TP53^−/−;PDGFRA^Δ8–9 iHGG cells.

Fig. 7. iHGG tumors confirm features characteristic of patient tumor samples.

PTEN^−/−;NF1^−/− (a) and TP53^−/−;PDGFRA^Δ8–9 (b) secondary tumors. UMAP plot color-coded by samples (top left) or by Louvain clustering (top right). Sample distribution found in each Louvain cluster, color-coded by sample identity (bottom left) and Louvain cluster distribution per sample, color-coded by cluster identity (bottom right). GBM molecular subtype analysis based on average gene expression of individual cells in each Louvain cluster for PTEN^−/−;NF1^−/− (c) and TP53^−/−;PDGFRA^Δ8–9 (d) tumors. For PTEN^−/−;NF1^−/− (e) and TP53^−/−;PDGFRA^Δ8–9 (f) tumors, stemness scores were calculated for each cell and results overlaid on a UMAP plot (top left) or summarized as violin plots for each cluster (top right). Cells were categorized based on their cell cycle status (G1, G2M or S), overlaid on a UMAP plot (bottom left) and their distribution was calculated for each Louvain cluster (bottom right).