Comparison of human pluripotent stem cell differentiation protocols to generate neuroblastoma tumors

Abstract

Neuroblastoma is the most common pediatric extracranial solid tumor and is derived from trunk neural crest cells (tNCC) and its progenitor sympathoadrenal (SA) cells. While human pluripotent stem cell (PSC) models of neuroblastoma have been described, the PSC were differentiated using protocols that made neural crest cells, but not specifically the trunk subtype. Here, we compared four recent protocols to differentiate pluripotent stem cells (PSC) toward SA cells and examined their efficiency at generating SA cells along with earlier cell states (neuromesodermal progenitors [NMP], tNCC), as well as generating MYCN-driven tumors. Interestingly, the protocols that created cells with the highest level of NMP markers did not produce cells with the highest tNCC or SA cell markers. We identified a protocol that consistently produced cells with the highest level of SA markers using two PSC lines of different genders. This protocol also generated tumors with the highest level of PHOX2B, a marker of neuroblastoma. Transcriptionally, however, each protocol generates tumors that resemble neuroblastoma. Two of the protocols repeatedly produced adrenergic neuroblastoma whereas the other two protocols were ambiguous. Thus, we identified a protocol that reliably generates adrenergic neuroblastoma.

Keywords: Human pluripotent stem cells; MYCN; Neuroblastoma; Sympathoadrenal cell.

Fig. 1

Protocols #2 and 3 generate cells with the highest NMP markers at day 3 of differentiation. (A) Schematic of analyzing cells differentiated from iPSC using four different protocols toward NMPs at day 3. (B) RT-qPCR analysis of NMP markers in (left) EDi27 and (right) EDi28 cells. n = 3, error bars represent standard error of mean. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (C) Immunofluorescence analysis of each protocol at day 3 for CDX2, T (Brachyury) and SOX2 co-stained with DAPI. Scale bar = 180 um.

Fig. 2

Protocol #4 generates cells that best resemble NCC at day 8 of differentiation. (A) Schematic of analyzing cells differentiated from iPSC using four different protocols toward tNCC at day 8. (B) RT-qPCR analysis of tNCC markers in (left) EDi27 and (right) EDi28 cells. n = 3, error bars represent standard error of mean. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (C) Immunofluorescence analysis of each protocol at day 8 for HNK1, p75, and AP2a co-stained with DAPI. Scale bar = 180um.

Fig. 3

Protocol #4 generates cells that best resemble SA cells at day 12 of differentiation. (A) Schematic of analyzing cells differentiated from iPSC using four different protocols toward SA cells at day 12. (B) RT-qPCR analysis of tNCC markers in (left) EDi27 and (right) EDi28 cells. n = 3, error bars represent standard error of mean. *p < 0.05,**p < 0.01,***p < 0.001,****p < 0.0001. (C) Immunofluorescence analysis of each protocol at day 12 for PHOX2B, HAND2, and DBH co-stained with DAPI. Scale bar = 180 um.

Fig. 4

Protocol #4 generates tumors with the highest level of PHOX2B. (A) Schematic of generating tumors using each differentiation protocol in TRE-FLAG-MYCN transduced iPSC. (B) Western blot analysis of EDi27 and EDi28 TRE-FLAG-MYCN iPSC untreated and treated with doxycycline (100 ng/mL) for 24 h. (C) Kaplan Meier survival curves of mice implanted with (left) EDi27 (n = 4 per group) and (right) EDi28 (n = 4 per group) TRE-MYCN SA cells differentiated using each of the four protocols. (D) Western blot for expression of FLAG-MYCN in two tumors for each protocol compared to EDi27 iPSC. (E) Histology analysis of EDi27 MYCN tumors derived from each protocol showing (top) H&E staining and (bottom) IHC for PHOX2B. Scale bars = 100 um.

Fig. 5

Protocol #3 and 4 generate adrenergic neuroblastoma tumors. (A) RNA was extracted from three EDi27 tumors from each protocol and analyzed by RNA-seq. Transcriptome profiles of human EDi27 tumors were compared against neuroblastoma, other human neural crest derived tumors (melanoma, pheochromocytoma and paraganglioma), and other small round blue cell tumors (medulloblastoma, ewing sarcoma, rhabdomyosarcoma). Patient data was obtained from the Treehouse Childhood Cancer Initiative (B) Transcriptomes of each EDi27 tumor was compared against neuroblastoma tumors based on the adrenergic score vs mesenchymal score. Neuroblastoma patient data was obtained from the TARGET database.

Fig. 6

Stromal and immune cells contribute to molecular differences between human neuroblastoma patient samples and Protocol #4 tumors. (A) Expression of stromal and immune markers were analyzed between TARGET human MYCN-amplified neuroblastoma samples (> = 90% tumor purity, n = 6) and tumors from Protocol #4 (n = 3). Bars represent the log₂ fold-change expression levels of the Protocol #4 tumors compared to the TARGET samples. (B) Gene ontology analysis of genes differentially expressed between Protocol #4 tumors and TARGET tumors using the REACTOME database.