The human serine-arginine rich splicing factor 6 (SRSF6) is part of the SR-protein family consisting of 12 members and is involved in (alternative-) splicing regulation. It is composed of an N-terminal RRM domain, followed by a pseudo-RRM and a C-terminal serine/arginine-rich disordered domain. Despite extensive studies on the functions of SRSF6, its determinants for specific RNA-recognition have remained unresolved due to a lack of high-resolution structures based on unambiguous boundaries of individual domains and the involved disordered regions. We here provide the first near-complete NMR backbone and sidechain assignments of human SRFS6 RRM1 and backbone assignments of the pseudo-RRM2 (ΨRRM2), as well as the tandem RRM. The derivable carbon secondary chemical shifts were used to define secondary structure elements as well as exact domain boundaries, showcasing a canonical β 1 α 1 β 2 β 3 α 2 β 4 -fold for either RRM. Consequently, we defined the inter-domain linker with a length of 37 residues. Furthermore, 15 N-relaxation data measured for the single and tandem RRMs indicate that the two RRMs tumble as one entity. However, the minor chemical shift differences between the respective constructs suggest a merely transient interaction with no defined interface between the two RRMs. Lastly, we show that AlphaFold3 models are majorly supported by our solution NMR data, but there are subtle inconsistencies with regard to the formation of the β 4 -strand. In sum, our initial work underlines the role of NMR for visualizing crucial functional details of biology and sets the critical basis for future high-resolution structures of and residue-resolved determinants of specific RNA-recognition by the splicing regulator SRSF6. • Heterologous expression and purification of SRSF6 constructs for structural studies. • First deposited backbone assignments of SRSF6 RRM1, ΨRRM2 and tandem RRM. • Experimentally proven secondary structure and domain boundaries. • SRSF6 RRMs interact transiently with each other. • Experimental data matches closely with AlphaFold3 models.
Ehr et al. (Fri,) studied this question.