Unique and redundant function of SOX2 and SOX17 in regulating the germ cell tumour fate

Embryonal carcinomas (ECs) and seminomas are testicular germ cell tumours. ECs display expression of SOX2, while seminomas display expression of SOX17. In somatic differentiation, SOX17 drives endodermal cell fate. However, seminomas lack expression of endoderm markers, but show features of pluripotency. Here, we use ChIP-sequencing to report and compare the binding pattern of SOX17 in seminoma-like TCam-2 cells to SOX2 in EC-like 2102EP cells and SOX17 in somatic cells. In seminoma-like cells, SOX17 was detected at canonical (SOX2/OCT4), compressed (SOX17/OCT4) and other SOX family member motifs. SOX17 directly regulates TFAP2C, PRDM1 and PRDM14, thereby maintaining latent pluripotency and suppressing somatic differentiation. In contrast, in somatic cells canonical motifs are not bound by SOX17. In sum only 10% of SOX17 binding sites overlap in seminoma-like and somatic cells. This illustrates that binding site choice is highly dynamic and cell type specific. Deletion of SOX17 in seminoma-like cells resulted in loss of pluripotency, marked by a reduction of OCT4 protein level and loss of alkaline phosphatase activity. Further, we found that in EC-like cells SOX2 regulates pluripotency-associated genes, predominantly by partnering with OCT4. In conclusion, SOX17 (in seminomas) functionally replaces SOX2 (in ECs) to maintain expression of the pluripotency cluster. Overall design: Chromatin immunoprecipitation (ChIP) was carried out using 200 µg chromatin lysate and 5 µg antibody (anti-SOX2: #ab59776 (abcam); anti-SOX17: #AF1924 (R&D)). SOX17 ChIP was performed on chromatin lysate of TCam-2 seminoma-like cells for three biological replicates. SOX2 ChIP was performed on chromatin lysate of 2102EP EC-like cells for three biological replicates. The protocol was performed according to the Simple ChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology). 2% Input (= 2 µg chromatin) served as control. ChIP-seq libraries were generated using the MicroPlex Library Preparation Kit v2 (Diagenode). Quality of the ChIP-seq libraries was assessed on a 2400 Tapestation (Agilent, Santa Clara, USA) and library pools were quantified using Qubit (Thermo Fisher Scientific). Samples were then sequenced on the Illumina High Seq 2500 v4 (Illumina, San Diego, USA) in rapid mode using 25 M single-end reads (1 x 75 bp) per sample. Reads were mapped to the human genome (hg38) using CLC Genomics Workbench 12 (Qiagen, Hilden, Germany) and data was analysed using HOMER (Hypergeometric Optimization of Motif EnRichment) Software (http://homer.ucsd.edu/homer/).