Deciphering the molecular underpinnings for the superior performance of engineered SOX17 as an inducer and maintainer of stemness

Abstract

Previously, an engineered SOX17 with three amino acid mutations inside the DNA binding high mobility group (HMG) box domain termed SOX17FNV has been identified as a highly potent pluripotency inducer. While the wild-type SOX17 failed to induce pluripotency, SOX17FNV outperformed the Yamanaka factor SOX2 in mouse pluripotency reprogramming with unknown mechanisms. Thus, this study aims to better understand the molecular basis for the high potency of SOX17FNV, in order to gain insight into transcription factor and cell fate engineering. The highly efficient mouse pluripotency reprogramming with SOX17FNV was first confirmed using a drug-controllable lentivirus system. Subsequently, the direct lineage reprogramming towards induced neural stem cell (iNSC) was performed and SOX17FNV was again found to outperform SOX2. Next, SOX17FNV was overexpressed to replace SOX2 in a SOX2-tet-off mouse embryonic stem cell (mESC) line. Surprisingly, despite the significant sequence differences, SOX17FNV was able to functionally substitute for SOX2 to maintain unimpaired pluripotency with overall similar transcriptomes and chromatin organizations. Interestingly, several cleavage stage genes were specifically upregulated by SOX17FNV and the SOX17FNV-expressing mESCs proliferated more slowly, indicating a more naïve pluripotency state reached and a higher developmental potential. Next, the molecular mechanisms for the superior performance of SOX17FNV were decoded. Though SOX17 was previously reported to interact with beta-catenin, the modulation of Wnt/beta-catenin pathway did not interfere with SOX17FNV-mediated cellular reprogramming, suggesting that SOX17FNV provides a more robust tolerance to Wnt fluctuations. Moreover, biochemical analyses failed to support the previously claimed “direct interaction” between SOX17 and beta-catenin. Liquid-liquid phase separation experiments showed that SOX17FNV and SOX17 could incorporate into Mediator 1-transcription condensates in a more liquid-like manner and more efficiently than SOX2 both in vitro and in cells, indicating a higher potential for gene expression activation. In vitro DNA binding assay using full-length protein from mammalian cells revealed that SOX17FNV changed DNA target preference compared to wild-type SOX17. SOX17FNV co-bound OCT4 and BRN2 more tightly than SOX2 on canonical SoxOct motifs that control the expression of many stem cell genes. Thinking outside of the HMG box, a detailed sequence-structure-function analysis was performed on the intrinsically disordered N and C termini of SOX17. The C-terminus of SOX17 had a more potent reprogramming ability than that of SOX2 and artificially designed transactivation domains (TADs). Moreover, the N terminus was found dispensable for SOX17FNV-mediated high-performance reprogramming, while the C terminus encoded several functionally redundant sequences. Only the central domain and the defined C2 regions were vital for highly efficient cell fate conversion. By removing the nonessential regions, a minimal SOX17FNV termed miniSOX was defined, encoding only 70% length of the full-length protein. The miniSOX performed in a similar manner as SOX17FNV in cellular reprogramming, pluripotency maintenance, and self-organization. Overall, the application of miniSOX could reduce the payload of current gene delivery systems for reprogramming in vivo. The enhanced knowledge of synthetic transcription factors will significantly advance stem cell engineering, personalized stem cell models, and therapeutic interventions.