Development of an osteosarcoma model with MYCN amplification and TP53 mutation in hiPS cell‐derived neural crest cells

Abstract Mesenchymal stem cell‐ or osteoblast‐derived osteosarcoma is the most common malignant bone tumor. Its highly metastatic malignant phenotypes, which are often associated with a poor prognosis, have been correlated with the modulation of TP53‐ and cell‐cycle‐related pathways. MYC, which regulates the transcription of cell‐cycle modulating genes, is used as a representative prognostic marker for osteosarcoma. Another member of the MYC oncoprotein family, MYCN, is highly expressed in a subset of osteosarcoma, however its roles in osteosarcoma have not been fully elucidated. Here, we attempted to create an in vitro tumorigenesis model using hiPSC‐derived neural crest cells, which are precursors of mesenchymal stem cells, by overexpressing MYCN on a heterozygous TP53 hotspot mutation (c.733G>A; p.G245S) background. MYCN‐expressing TP53 mutated transformed clones were isolated by soft agar colony formation, and administered subcutaneously into the periadrenal adipose tissue of immunodeficient mice, resulting in the development of chondroblastic osteosarcoma. MYCN suppression decreased the proliferation of MYCN‐induced osteosarcoma cells, suggesting MYCN as a potential target for a subset of osteosarcoma treatment. Further, comprehensive analysis of gene expression and exome sequencing of MYCN‐induced clones indicated osteosarcoma‐specific molecular features, such as the activation of TGF‐β signaling and DNA copy number amplification of GLI1. The model of MYCN‐expressing chondroblastic osteosarcoma was developed from hiPSC‐derived neural crest cells, providing a useful tool for the development of new tumor models using hiPSC‐derived progenitor cells with gene modifications and in vitro transformation.


| INTRODUC TI ON
Neural crest cells, progenitor cells of MSCs, have been efficiently differentiated from hiPSCs. 3 Moreover, NCCs are multipotent cells that are mainly divided into two subgroups, cNCCs and tNCCs. cNCCs differentiate into bone and cartilage, whereas tNCCs differentiate into sympathetic ganglia and melanocytes. 4 NCCs are of particular interest in cancer research as they constitute the origin of various tumors, including small-cell lung cancer, osteosarcoma, neuroblastoma, and glioblastoma. 5 MYC is involved in the process of differentiation of NCCs into osteoblasts, while MYCN is involved in the process of differentiation of NCCs into sympathetic ganglia; consistently, hyperregulation of these processes is considered to lead to tumorigenesis. 6,7 The MYC family members, MYC and MYCN, are representative oncogenes and have been widely used to generate mouse models of cancer, such as breast cancer and neuroblastoma. 8,9 Using an established cNCC differentiation protocol, we induced the generation of cNCCs from WT hiPSCs and LF syndrome patientderived hiPSCs, and subsequently introduced MYCN into these cNCCs using lentiviral transfection to study the MYCN-related oncogenic mechanism in the hiPSC-derived cNCCs. 3 Transfected NCCs were cultured in soft agar medium and TF clones were isolated. We administered these LF-derived/MYCN-transduced TF clones both subcutaneously and into the periadrenal adipose tissue of immunodeficient mice, resulting in the generation of high-grade chondroblastic osteosarcoma. Here, we investigated the molecular features of tumorigenesis of this high-grade osteosarcoma using comprehensive gene expression and whole-exome analyses compared with previous reports on osteosarcoma.

| ME THODS AND MATERIAL S
Expanded methods can be found in Appendix S1.

| Pluripotent stem cell maintenance
The 414C2 and LF human iPSC lines were provided by Dr. Kenji Osafune (Kyoto University, Kyoto, Japan). LF hiPSCs carry a heterozygous TP53 hotspot mutation (c.733G>A; p.G245S). Both cell lines were maintained on a feeder layer of mitomycin C-treated embryonic mouse fibroblasts in primate embryonic stem (ES) cell medium (REPROCELL, Tokyo, Japan) supplemented with 4 ng/mL recombinant human FGF2 (FUJIFILM, Tokyo, Japan). The 414C2 hiPSC cell line was generated from a healthy person. 10 The LF hiPSC cell line was generated by transducing episomal vectors (OCT3/4, SOX2, KLF4, L-MYC, LIN28, and p53shRNA) in dermal fibroblasts from a patient with LF syndrome.

| Differentiation of 414C2 and LF hiPSCs into cNCCs and preparation of MYCN-overexpressing cells
We subjected 414C2 and LF hiPSCs to differentiation into cNCCs according to the reported protocol. 3 Using the neurotrophin receptor (p75, a cell surface marker of NCCs), we collected the p75-positive cells by cell sorting ( Figure S1). We then subjected the sorted cNCCs to gene expression analysis of stem cell and NCC markers ( Figure 1A).
We found that the sorted cNCCs expressed representative markers of NCCs, such as PAX3, p75, SOX10, and ETS1. 3 In contrast, we detected that the expression of OCT4 was clearly downregulated in cNCCs compared with that in hiPSCs. We infected these cNCCs with either control (mock) or MYCN-expressing (MYCN) lentiviruses.
Subsequently, we analyzed the infected cNCCs by RT-PCR and western blotting to confirm the induction of the gene and protein expression of MYCN. Notably, we observed that the expression of MYCN was increased in LF-derived cNCCs compared with that in 414C2-derived cNCCs ( Figure 1B,C).

| Transformation using the soft agar colony formation assay
We performed the soft agar colony formation assay using mock-or MYCN-expressing 414C2-and LF cNCCs. We found that for 414C2

| Tumorigenesis experiments by subcutaneous administration
We selected the well-grown clones ( Figure 2C: upper panel) and subcutaneously administered 414C2 (MYCN TF-1 to TF-3) and LF (mock TF-1 to TF-3 and MYCN TF-1 to TF-4) cNCC clones to BALB/c Ajcl nu/ nu mice. We observed that 414C2 MYCN TF clones did not generate any tumor masses (three out of three). However, we detected the generation of tumor masses in two out of four LF mock TF-3, one out of four LF MYCN TF-1, and two out of four LF MYCN TF-4 clones ( Figure 3A).
In contrast, LF mock TF-1, and -2, as well as LF MYCN TF-2 and -3 did not generate any tumor masses. We found the apparent development of bone and cartilage with atypical spindle cells in all masses and identified all generated tumors as chondroblastic osteosarcoma according to H&E staining ( Figure 3B). PCNA (proliferation marker) immunostaining showed cell proliferation in all masses ( Figure 3B). We also detected a high PCNA labeling index (98.3%) in the tumor derived from LF MYCN TF-4 ( Figure 3C). According to the percentage of PCNA, we defined more than 70% as high grade (LF MYCN TF-4), between 30% and 70% as intermediate, and less than 30% as low grade. In addition, to examine the effects of tumorigenesis in TP53 and MYCN, lentivirus-infected mockor MYCN-expressing 414C2-and LF cNCCs were administered subcutaneously to SCID/Beige mice, which are more severely immunodeficient than BALB/c Ajcl nu/nu mice. We found that LF mock-and LF MYCNexpressing cNCCs generated masses that were larger than the 414C2 mock-and 414C2 MYCN-expressing cNCC masses, respectively, suggesting a positive effect of the TP53 mutation on osteosarcoma tumorigenesis ( Figures S3 and S4). H&E staining showed that all tumor sections contained bone tissues, which were assumed to be due to differentiation into cNCCs ( Figure 3D). Also, the mutation of TP53 and expression of MYCN induced the formation of nuclear atypia (H&E; Figure 3B  The H&E staining showed a higher tumor cell density, more prominent cell division, and higher necrosis in LF MYCN CO-3 compared with LF MYCN CO-2 ( Figure 4A). We further observed that all F I G U R E 2 Isolation of transformed cells using the soft agar colony formation assay. (A) Anchorage-independent growth ability was evaluated by soft agar colony formation assay. Colonies were counted using the Clono-Counter software. Results were represented as the mean ± SD of triplicate samples. (B) Protocol for the isolation of colonies from soft agar dishes. A single colony was isolated in a 1.5 mL tube containing 50 μL accutase and pipetted with CDMi medium. Pipetted clones were cultured in 24-well plates. (C) Western blotting and semiquantitative RT-PCR analysis (upper panel) of the expression of MYCN in isolated 414C2 MYCN TF 1-3, LF mock TF 1-4, and LF MYCN TF 1-4 clones. ACTIN was used as loading control. The lower panel shows anchorage-independent growth ability evaluated by soft agar colony formation assay. Results are represented as the mean ± SD of four replicates.

F I G U R E 3 Subcutaneous administration with mock and MYCN cNCCs/TF clones of 414C2 and LF into BALB/c AJcl-nu/nu mice. (A)
BALB/c AJcl-nu/nu mice were subjected to subcutaneous xenografts using 414C2 MYCN, LF mock, or LF MYCN TF clones. Respective TF clones were administered to four independent parts, as described in Methods. Proliferating TF clones are shown. Mice whose tumor mass volume did not reach 50 mm 3 Figure 4C). In particular, we detected that tumorigenesis further increased the expression of MYCN compared with that in the administered clones. We found that the levels of osteosarcoma-related genes were upregulated in LF MYCN CO-1, -2, and -3, and in LF mock CO tumors, indicating that all generated tissues had genetic features of osteoblasts or osteosarcoma, including chondroblastic osteosarcoma. In addition, we detected that RUNX2 and its downstream genes (FGFR2, FGFR3, SP7, and SMOC2) were downregulated in the highly aggressive LF MYCN CO-3 clone ( Figure 4C). RUNX2 is reported to regulate osteoblast differentiation 13,14 and suppress osteosarcoma cell proliferation. 15 We also found that LF MYCN CO-3 had the same PCNA rate as LF MYCN CO-1, but exhibited higher grade features in terms of H&E staining and lower expression of RUNX2, suggesting the tumor suppressor function of RUNX2 in cNCC-derived chondroblastic osteosarcomas.

| Genetic changes during the transition from cNCCs to tumorigenic clones
To uncover the mechanisms underlying tumorigenesis, we performed gene expression profiling using microarray analyses of LF mock cNCCs, LF MYCN cNCCs, LF MYCN TF-4 clones, and LF MYCN CO-3 tumors.
Our principal component analysis separated distinctly the four different groups of samples ( Figure 5A). Previous studies have reported the activation of pathways such as TGFβ, NF-kappa B, and WNT signaling in osteosarcoma. 16,17 Therefore, we analyzed the upregulated genes (moderated t-test of <0.01 and fold changes of >2) using the KEGG pathway analysis program from DAVID v6.8. Functional annotation of upregulated genes suggested that the NF-kappa B signaling pathway, basal cell carcinoma, TGFβ signaling pathway, and cytokine-cytokine receptor interaction were significantly enriched in the process of tumorigenesis ( Figure 5B). Subsequently, among these pathways, we focused on the genes in the TGFβ signaling pathway and basal cell carcinoma, including bone morphogenetic proteins, and compared the expression of representative genes among cells, clones, and tumors using RT-PCR ( Figure 5C). We found that LF MYCN CO-3 showed significantly higher expression of genes involved in the TGFβ signaling pathway affecting osteosarcoma tumor growth and basal cell carcinoma, suggesting the increased expression of cell proliferation-related genes during the process of transition to becoming a high-grade tumor ( Figure 5C; Table S1).
In addition, GSEA suggested that LF MYCN CO-3 displayed tumor characteristics that were more associated with MSC-derived sarcomas obtained from soft tissue and bone compared with LF mock cNCCs ( Figure 5D). [18][19][20][21][22] These results indicated that LF MYCN CO-3 exhibited features characteristic of osteosarcoma, as indicated in Figures 4 and 5.

| Suppression of MYCN in osteosarcoma cell lines
Although the malignant phenotypes of osteosarcoma are known to be related to the high expression of c-MYC, little has been reported on the relationship between MYCN and osteosarcoma. 23 Interestingly, the expression of MYCN has been shown to be higher in osteosarcoma than that in MSC and this higher expression has been related to a lower overall survival probability in osteosarcoma ( Figure S2  Abbreviation: ND, not detected due to post-freezing.

| Analysis of genomic changes using wholeexome sequencing
Although  Table S2. Among the 29 extracted mutations collated using dbSNP and ClinVar, we detected three mutations in AQP7, DYX1C1, and NOL12, which mutations are reported as benign or unknown (Table S2). Annotation was performed for the pathogenicity of all extracted mutations using   (Figure 7). We accordingly identified 13 candidate TSGs and six candidate oncogenes, and especially found that GLI1 was highly upregulated in the LF MYCN CO-3 tumor ( Figure 5C; Table S3).

| DISCUSS ION
In recent years, there has been an increased interest in exploring It has been previously suggested that p53 dysfunction leads to osteosarcoma tumorigenesis, which has been shown in vitro F I G U R E 7 Expression profiles of downregulated and upregulated genes in LF MYCN CO-3. Clustering analyses using average linkage and Pearson's correlation distance were performed for downregulated genes in focal deletions (FC < 0.5) and upregulated genes in focal amplifications (FC >2) in LF MYCN CO-3 compared with those in LF mock or LF MYCN cNCCs.
using LF hiPS cell-derived osteoblasts. 29 MSCs and osteoblasts are predicted as the origin of osteosarcoma, 30 and in this study, osteosarcoma was generated from NCCs, which is a further undifferentiated state. Amplification of MYC is frequently observed in osteosarcoma, and thus c-MYC is used as a prognostic biomarker. 16 The MYC family also includes MYCN and MYCL. A previous study showed the relationship between the expression of MYCN and osteosarcoma, although it was considerably less than that of C-MYC. 31 In our study, all generated tumors were pathologically diagnosed as chondroblastic osteosarcoma. Microarray analysis of the molecular profile revealed the activation of pathways such as cytokine-cytokine receptor interaction, NF-kappa B signaling, and TGFβ signaling, which are known to be frequently activated in osteosarcoma ( Figure 5B). 16 BMI1 is known to function as an oncogene in osteosarcoma, promoting tumorigenesis. Interestingly, tumorigenesis has been shown to upregulate BMI1, 32 which is the same finding in the expression of MYCN in our study. In addition, the expression of GAL-1 33 and EZR, 34 which have been reported as chondroblastic osteosarcoma markers, was increased during tumorigenesis ( Figure 4C). RUNX2 is the earliest determinant of osteoblast differentiation, and its forced expression has been shown to suppress the proliferation of osteosarcoma cell lines, suggesting that LF MYCN CO-3 is high grade compared with LF MYCN CO-1. 15 Although the relationship between chondroblastic osteosarcoma and TP53 mutation is known, its relationship to MYCN status has not been reported. 30,35 Our study showed that MYCN  GLI1 has been reported to be amplified in osteosarcoma; GLI1 copy number was similarly amplified in LF MYCN CO-3 (Tables S3 and   S4). 42 In addition, GLI1 inhibition by shRNA decreased colony and sphere formation in cultured osteosarcoma cells, whereas upregulation of GLI1 has been shown in other tumor types, 38

ACK N OWLED G M ENTS
We thank Editage (www.edita ge.jp) for English language editing, and Ms. Kimie Nomura for technical assistance.

FU N D I N G I N FO R M ATI O N
This study was supported by grants from the Japan Society for the Promotion of Science (JSPS) KAKENHI grant nos. JP 19K16759, JP 19H03625, and JP 22H04922 (AdAMS).

CO N FLI C T O F I NTER E S T S TATEM ENT
Takehiko Kamijo, Miki Ohira, and Junya Toguchida are Editors of Cancer Science. The other authors have no conflict of interest.

E TH I C S S TATEM ENT
The authors declare: