In silico identification and structural characterization of ClAPRR2 and ClPSY1 as independent genetic targets for a white-rinded, orange-fleshed watermelon (Citrullus lanatus) accessible to colour-vision-deficient consumers

The watermelon (Citrullus lanatus) usually possesses green skin pigmentation due to chlorophyll and red flesh pigmentation due to lycopene, which is a combination of colors that is less contrasting for people with red-green color vision deficiency. In this research, comparative bioinformatics analysis was used to find out if two genes – CIAPRR2 (rind color regulator, chromosome 9) and CIPSY1 (flesh carotenoid biosynthesis enzyme, chromosome 1) are good candidate genes, which are also independently suited to creating a new variety of watermelon with white rind and orange flesh with contrasting colors. Coding sequences were obtained from the Cucurbit Genomics Database and further used for generating homology models using SWISS-MODEL. CIAPRR2 gene was modelled using the template of Cucurbita moschata APRR2 protein with 80.4% coverage and 0.57 GMQE. As expected, the structure of the protein had a typical two-domain structure of the type-B two-component response regulator with an N-terminal receiver domain and C-terminal DNA-binding output domain, which is typical for this type of transcription factor activating the expression of chlorophyll-biosynthesis genes. CIPSY1 gene was modelled using the template of bacterial squalene/phytoene synthase CrtM (34.4% identity, 0.71 GMQE). A non-synonymous SNP associated with orange/golden flesh coloration was found near the substrate-binding cavity. CIAPRR2 and CIPSY1 are located on different chromosomes, are differently expressed (in rind and flesh), and function through different mechanisms (transcriptional regulation and enzyme activity), which makes them good candidates for the development of a stacked trait by marker-assisted selection or multiplex genome editing. Being homology models, these results need experimental verification. The CIPSY1 model has excellent stereochemistry (93.3% Ramachandran favoured, 0% outliers), while the CIAPRR2 model is of moderate quality (78.3% favoured).