A previously published cohort of thyroid resection specimens from Vanderbilt University Medical Center and the University Washington, including 32 ATCs and 20 MNGs, was used for RNA and histologic analysis [8].

Computational analyses were performed with R version 4.3.1. Raw RNA count data from Xu et al was subset to ATC (32) and MNG (20) and normalized using DESeq2 1.40.2 [15]. A Hallmark Wnt signaling score was calculated as the average mRNA Z-score of the 42 genes in the hallmark Wnt signaling gene set from the Molecular Signatures Database (MSigDB) [16]. Hallmark Wnt scores and Log2 expression of Wnt ligands were compared between ATC and MNG using Wilcoxon rank sum tests and plotted using ggplot2 3.4.4 [17].

Xenograft cell lines, THJ-11T, THJ-16T, THJ-21T, and THJ-29T, were obtained from Dr. John Copland (Mayo Clinic, Jacksonville, FL, USA) (refer to Supplemental Table 1 in Online Resource I). Cells were authenticated using STRS analysis and maintained and used experimentally at <20 passages from thaw. Cells were grown in complete media, which contains RPMI (VWR) containing 10% fetal bovine serum (FBS) (ThermoFisher Scientific), 1% penicillin-streptomycin (pen/strep) (Sigma), 1X MEM Non-Essential Amino Acids (VWR), and 1 mM sodium pyruvate (Vanderbilt Molecular Biology Resource).

300 cells in 30 μL of suspension culture were plated per well in black 384-well cell-repellant culture plates (Greiner Bio-One). Spheroids were allowed to form for 24 hours prior to treatment with DMSO, dabrafenib+trametinib, and/or pyrvinium pamoate (Selleck Chemicals) and Matrigel in complete media. Following 72 hours of treatment, wells were imaged using an ImageXpress Micro XL automated high-content microscope (Molecular Devices). To assess viability, CellTiter-Glo 3D (Promega) was added to wells and mixed with the Bravo liquid handler (Velocity 11/Agilent). Per the CellTiter-Glo protocol, plates were placed on a shaker for 25 minutes before luminescence was quantified using a Synergy NEO (BioTek multi-mode plate reader) [11].

Patient-derived organoids were collected and maintained as previously described [12, 13]. For drug treatment, organoids were centrifuged for 5 minutes at 340xg to form a loose pellet. This pellet was washed with cold PBS prior to dissociation for 30 minutes at 37 °C in 1X TrypLE (Gibco). The resulting cell suspension were washed and resuspended in complete media containing 5% Matrigel and DMSO, 30 nM dabrafenib+5 nM trametinib, and/or 300 nM pyrvinium prior to plating in a 24-well low-attachment culture plate. Organoids were incubated at 37 °C, 5% CO2 for 8 days prior to imaging with a Leica DMi1 MC170 inverted microscope with a 4X objective and processed on Leica Application Suite, version 4.10.0. All work with patient-derived cells was approved by the Vanderbilt Institutional Review Board.

All procedures were approved by the Institutional Animal Care and Use Committee prior to completion. NOD.PrkdcscidIl2rg-/- (NSG-Jackson Laboratories) were injected with 1 x 106 xenograft THJ-16T or xenograft THJ-21T cells subcutaneously in the flank using a 25G SubQ needle affixed to a 1 mL syringe. When tumors became palpable (approximately 2 weeks post-injection), intraperitoneal injection of vehicle or pyrvinium pamoate (5% DMSO in corn oil) was started. To decrease the side effects of Wnt inhibition (colitis and fractures) and increase tolerability, we performed a dose-escalation of 0.5 mg/kg, 1 mg/kg, 1.5 mg/kg, 2 mg/kg with injections every other day, maintaining at 2 mg/kg for the remainder of the experiment. Injections were performed with a 25G SubQ needle on 1 mL syringes. Tumors were measured every other day using digital calipers, and mice were weighed weekly to ensure that weight loss did not exceed 20% of body weight. When tumors reached 2 cm in any dimension or ulcerated, mice were humanely euthanized. Tumor volumes were transformed to the square root scale to address skewness. A linear mixed-effects model was employed to account for the correlation inherent in repeated measurements within individual mice over time. Mouse ID was treated as random effect. The estimated growth rates from the model were compared between two treatments using Wald test. The analysis of tumor volume was conducted using R version 4.3.2.

Stable Wnt reporter cell lines were generated using lentiviral transduction. Viral media was collected from HEK293FT cells transfected with the 7TFP lentiviral plasmid (Addgene #24308), along with the PAX2 (packaging) and pMD2G (envelope) plasmids. Thyroid cancer cell lines were cultured in lentiviral media with 8 mg/mL Polybrene for 24 hours. Antibiotic selection was performed with puromycin.

The TOPFLASH/7TFP reporter system was used to measure TCF/LEF transcriptional activity. Following Wnt treatment (16 hours), reporter cells were lysed in 1x Passive Lysis Buffer (Biotium) and incubated on a shaker for 15 minutes. Sample lysates were assayed for firefly luciferase reporter activity using Steady-Glo Luciferase Assay (Promega) and for cell viability using Cell Titer Glo (Promega). Luminescence was measured via FLUOstar Luminometer. Luciferase activity was normalized to cell viability. Assays were performed in triplicate and repeated independently at least three times (refer to Supplemental Figure I in Online Resource I).