Recapitulating thyroid cancer histotypes through engineering embryonic stem cells

The study addresses which cell population in the thyroid lineage hierarchy serves as the cell of origin for distinct follicular cell-derived thyroid cancer (TC) histotypes once somatic mutations are acquired. Despite progress in understanding TC biology, effective therapies for metastatic, radioiodine-refractory disease are lacking due to loss of sodium/iodide symporter (NIS) expression. Models positing different cells of origin or a single target cell with differing mutations have not captured TC intertumoral heterogeneity. Adult stem/progenitor cells, due to longevity and self-renewal, are plausible targets for oncogenic transformation, as shown in hematologic malignancies. The authors hypothesize that human embryonic stem cell (hESC)-derived thyroid progenitor cells (TPCs) represent the susceptible target for TC-initiating mutations and that engineered mutations in these cells can recapitulate TC histotypes and progression, enabling mechanistic and therapeutic studies, particularly relevant for radioiodine-resistant disease.

The paper contextualizes TC histotypes—papillary (PTC), follicular (FTC), and anaplastic (ATC)—highlighting clinical behaviors and the frequent dedifferentiation of differentiated TC to ATC with accumulating mutations. Prior models of carcinogenesis propose either distinct cells-of-origin or a common target acquiring varied mutations, but they fail to reflect TC heterogeneity. Evidence from leukemia shows mutations accumulating in progenitors/stem cells, supporting a tumorigenesis model where long-lived stem/progenitor cells are primary targets. Known TC genetics converge on MAPK and PI3K/AKT/mTOR pathways. BRAF mutations are common in PTC and ATC, while RAS mutations characterize FTC and are present in ATC; TP53 mutations are rare in differentiated TC but frequent in ATC. The authors also reference prior identification of thyroid cancer stem cells and the role of CD44 variants and TIMP1/MMP9 signaling in cancer progression and metastasis, providing rationale to examine these pathways in TC initiation and progression.

• hESC differentiation: WA09 hESCs were sequentially differentiated following Longmire’s protocol to definitive endoderm (5 days), anterior foregut endoderm (day 6; serum-free medium with Noggin and SB431542), thyroid lineage specification to TPCs by day 22 (serum-free medium with FGF10, KGF, BMP4, EGF, high-dose FGF2, heparin), and maturation to thyrocytes by days 26–30 (FGF2, FGF10, heparin, TSH). Cells were cultured on hESC-qualified matrix to day 6 then Matrigel to day 30 with ROCK inhibitor after passages.
• Genetic engineering: At days 0/6/22/26/30, cells were transfected with CRISPR/Cas9 all-in-one vectors to introduce BRAF V600E, NRAS Q61R (or Q61K in vector description) and TP53 R248Q mutations; an OFP reporter enabled FACS enrichment. Nontargeting CRISPR served as control. Sanger and targeted NGS validated edits; off-target top candidates were sequenced.
• In vitro assays: Cell cycle profiling (PI staining), proliferation (trypan blue counts), invasion (Matrigel transwell), clonogenicity (ELDA), protein analyses (Western blot for β-catenin, TIMP1, CD44, pAKT/AKT, pERK/ERK, NIS), flow cytometry (CD44v6 and lineage markers), MMP9 ELISA, gene expression (qRT-PCR panels, ddPCR for KISS1R), and RNA-seq with GSEA/Reactome enrichment analyses.
• Modulation of pathways: Treatments with recombinant TIMP1, TIMP1 neutralizing inhibitor, KISS1R inhibitor; Wnt3A and R-spondin1 pre-treatment for rescue experiments. Lentiviral overexpression of TIMP1, MMP9, CD44, CD44v6; shRNA-mediated CD44v6 knockdown.
• Iodine uptake: Nonradioactive iodine assay and 125I uptake in engineered TPCs and thyroid cancer cell lines (BCPAP, TT2609co2, Cal62) following KISS1R/TIMP1 inhibition.
• In vivo models: Subcutaneous injections of engineered cells from different differentiation stages (D6/22/26/30) into NOD SCID mice to assess tumorigenicity and histotype formation; orthotopic thyroid injections (3×10^5 cells) to evaluate metastasis. Treatment study with intratumoral KISS1R inhibitor (daily) and anti-TIMP1 antibody (weekly), alone or combined, in D22 NRAS/TP53 tumors. Tumor growth monitored, survival analyzed; xenografts assessed by H&E, immunohistochemistry for lineage and pathway markers.
• Statistics and data sources: t-tests, Mann–Whitney, ANOVA, Kruskal–Wallis with appropriate datasets (GEO GSE33630, TCGA THCA, R2 platform, GEPIA) for expression and DFS analyses. Data deposited under BioProject PRJNA887246.

• Lineage and tumorigenicity: hESCs differentiated to a thyroid lineage with TPCs at day 22 (D22) expressing PAX8/TTF1 and progenitor markers (CD133, ABCG2, Nestin, HNF-4α, HHEX, FOXE1). Mature thyrocyte markers (TSHR, TPO, Tg, NIS) increased by D26–D30. D26/D30 cells showed negligible TIMP1/MMP9/CD44 and lacked tumorigenicity.
• Histotype recapitulation: CRISPR-engineered D22 TPCs harboring BRAF V600E formed PTC-like tumors; NRAS Q61R yielded FTC-like tumors. Adding TP53 R248Q to BRAF or NRAS backgrounds generated undifferentiated, ATC-like tumors with pleomorphic morphology and low Tg/CK19/NIS. TP53 mutation alone was insufficient for tumor initiation. Early-stage (D6) edited cells formed teratocarcinomas, not thyroid tumors. Mature thyrocytes (D26/D30) had very limited or no tumorigenic capacity.
• Genomics and signaling: High-efficiency genome editing confirmed by Sanger and NGS; engineered D22 cells showed increased expression of CSC-, EMT-, metastasis-related genes, and β-catenin accumulation in ATC-like contexts. Transcriptomics identified common upregulation of HGF, TIMP1, MMP9, CD44 in D22 versus D30.
• TIMP1/MMP9/CD44 ternary complex: Expression peaks at D22 and declines by D30; complex sustains PI3K/AKT activation and tumor initiation capability. Overexpressing TIMP1+MMP9+CD44 endowed otherwise weakly tumorigenic D30 NRAS/TP53 cells with increased proliferation and small tumor formation. Targeting TIMP1 reduced pAKT, proliferation, invasion, and CD44v6 expression across mutation backgrounds.
• CD44v6 role: CD44v6 was expressed in engineered D22 TPCs and matched patient tumors (PTC, FTC, ATC). shRNA knockdown of CD44v6 lowered pAKT, proliferation, and invasion; overexpression increased pAKT, which was reversed by TIMP1 inhibition. TIMP1 blockade downregulated Wnt/EMT/stemness genes; Wnt3A/R-spondin1 rescued β-catenin targets and growth.
• Metastasis and KISS1R: Orthotopic D22 BRAF/TP53 and NRAS/TP53 TPCs produced primary and metastatic lesions with increased EMT markers (TWIST, SNAIL) and high KISS1/KISS1R expression. Patient analyses showed KISS1/KISS1R upregulation in aggressive PTC metastases and ATC, and high KISS1R associated with higher metastatic risk and poorer DFS.
• Therapeutic implications: KISS1R inhibition reduced proliferation/invasion and PI3K/AKT and ERK activation. Combined TIMP1 and KISS1R inhibition increased thyroid differentiation markers (PAX8, TG, TSHR, TPO, TTF1), upregulated NIS protein, and restored functional 125I uptake in BRAF/TP53 and NRAS/TP53 D22 TPCs to levels comparable to PTC cell lines. In vivo, combined treatment delayed tumor growth, increased NIS expression in tumors, and prolonged mouse survival (n=6/group).

The findings demonstrate that thyroid progenitor cells (TPCs) at a defined differentiation stage (D22) are the preferential targets for oncogenic mutations that specify TC histotypes, supporting a genetic mutation model of carcinogenesis in long-lived progenitor cells. BRAF V600E and NRAS Q61R edits respectively drive PTC and FTC phenotypes, while additional TP53 loss-of-function cooperates to produce aggressive, undifferentiated ATC, consistent with clinical mutation patterns. The TIMP1/MMP9/CD44 axis, particularly CD44v6, sustains tumor initiation and growth via PI3K/AKT and Wnt/β-catenin signaling, while KISS1R supports metastatic progression and correlates with poor prognosis. Therapeutically, targeting TIMP1 and KISS1R reprograms aggressive cells toward a differentiated state, restores NIS function, enhances radioiodine uptake, slows tumor growth, and improves survival in vivo, suggesting an adjuvant strategy to sensitize undifferentiated or metastatic TCs to standard radiometabolic therapy. The engineered hESC-derived model faithfully mirrors patient tumor transcriptomic and phenotypic features and provides a platform to dissect oncogenic mechanisms and therapy resistance.

This study establishes a human hESC-derived thyroid progenitor cell model that recapitulates TC histotypes by introducing clinically relevant mutations. BRAF V600E and NRAS Q61R edits in TPCs generate PTC and FTC, respectively, while addition of TP53 R248Q yields undifferentiated ATC-like tumors. The TIMP1/MMP9/CD44 (CD44v6) complex is essential for tumor initiation via PI3K/AKT and Wnt signaling, and KISS1R promotes metastatic progression and portends poor prognosis. Dual inhibition of TIMP1 and KISS1R restores thyroid differentiation markers and NIS, enabling radioiodine uptake and improving tumor control and survival in vivo. The engineered system provides a versatile platform to model transformation in endoderm-derived tissues and to interrogate mechanisms of therapy resistance. Future work should validate combinatorial targeting strategies clinically and extend this progenitor-focused engineering approach to other organ lineages and mutational contexts.

The model relies on engineered human embryonic stem cell-derived progenitors and xenografts, which, while recapitulating key histotypes and signaling features, represent an artificial system that may not capture all aspects of human tumor microenvironment and disease heterogeneity. TP53 was modeled primarily as R248Q; other TP53 alterations and broader mutational spectra were not exhaustively assessed. Mature thyrocytes showed minimal tumorigenicity, but plasticity under varied conditions was not fully explored.