Metabolite profiles of green leaves and coffee beans as predictors of coffee sensory quality in Robusta (Coffea canephora) germplasm from the Democratic Republic of the Congo

Abstract

Research on the chemical composition of coffee beans and its correlation with sensory quality is advancing, but the metabolites in coffee leaves have often been overlooked. This study investigated the metabolite profiles of roasted coffee, green beans, and coffee leaves from 39 C. canephora genotypes through LC–HRMS with an untargeted metabolomics approach. Our results showed that metabolite profiles of roasted coffee, green beans, and coffee leaves can be discriminated based on the coffee's sensory quality. The highest predictive power of coffee sensory quality was achieved with the metabolite profiles of the coffee leaves. Genotypes with varying genetic backgrounds could only be discriminated by the metabolite profiles of the coffee leaves. Metabolite marker compounds predictive of sensory quality were a putative quercetin derivate in green beans and likely physagulin E and argophyllin in coffee leaves. Estimated levels of caffeoylquinic acids, dicaffeoylquinic acids, and caffeoylquinic acid lactones were higher in roasted coffee with a lower sensory quality. These differences were not apparent in their respective green beans and coffee leaves. Overall, our study revealed the promising potential of the metabolite profile of coffee leaves as a predictive tool for coffee sensory quality and the genetic background of C. canephora.

Keywords

• Coffea canephora
• Robusta
• Coffee beans
• Coffee leaves
• LC-HRMS
• PLS-DA