An integrated view of haplotype, gene expression and allelic variation explains the chemotype and cannabinoid yield of Cannabis sativa

The chemotype of Cannabis sativa is defined by the relative accumulation of Δ⁹-tetrahydrocannabinol (THC) and cannabidiol (CBD), which depend on the activities of the cannabinoid synthases THCAS and CBDAS. However, the link between chemotype gene expression and actual THC/CBD levels is unclear. Here, we show an integrated model predicting chemotype and cannabinoid yields. We quantified THC and CBD contents and gene expression in three proprietary commercial varieties. We integrated these data with metabolomic, transcriptomic and genomic information from 27 public accessions. We used a total of 157 RNA-seq datasets and eight sequenced genomes. Microsynteny and phylogenetic analyses revealed a single copy of THCAS and/or CBDAS , whereas CBCAS occurred in variable copy numbers. Expression of THCAS and CBDAS correlated positively with measured cannabinoid levels. Incorporation of GOT1 , encoding the CBGA precursor enzyme, improved model performance (R² = 0.678 for THC; R² = 0.768 for CBD). Eight alleles of THCAS and five alleles of CBDAS discriminated chemotypes and, for CBD, and including alleles in the model outperformed expression data in predictive power (R² = 0.9398). These results provide an updated model of chemotype inheritance that integrates cannabinoid quantification, gene expression, and allelic diversity, clarifying the molecular basis of THC and CBD production in C. sativa . Furthermore, our results indicate that the THCASa3 and THCASa4 alleles have a stronger association with THC production than the most common THCAS allele, found in UniProt. On the other hand, the CBDAS allele from UniProt is the most common and strongly associated with CBD production in our statistic model. Our work allows the direct search for high production of THC and CBD germplasm using combined allelic variants and expression analysis.