Mathematical Model Based Understanding and Engineering of the Phenylpropanoid Pathway in Arabidopsis

2-Phenylethanol (2-PE) is a naturally occurring aromatic with advantageous properties compared to ethanol making it a potential biobased oxygenate in petroleum-derived gasoline. In plants, biosynthesis of 2-PE competes for the common precursor phenylalanine with phenylpropanoid pathway which directs about 30% of fixed carbon towards lignin biosynthesis. The efficient extraction of cellulosic biomass is hindered by the presence of lignin. Therefore genetic manipulation around the phenylalanine branch point provides a promising strategy to redirect carbon flux away from lignin to enhance biomass quality, and realize an economically valuable by-product 2-PE. Transgenic Arabidopsis thaliana were generated that overexpress the phenylacetaldehyde synthase from Arabidopsis (PAAS) in tandem with the phenylacetaldehyde reductase (PAR) from tomato. We selected primary stem as the experimental system for analyzing competition for phenylalanine between phenylpropanoid metabolism and engineered 2-PE pathway. Excised 5-week-old stems were fed with different concentrations of ¹³C₆-ring labeled phenylalanine, with both the amount and isotopic enrichment of downstream intermediates quantified with LC-MS/MS at multiple time points after feeding. A kinetic model of the phenylpropanoid network was constructed, and the parameters were identified through non-linear optimization with training datasets, and validated with data from an independent experiment. In silico analysis of the results from our model predicted that the cytosolic phenylalanine concentration limits 2-PE production in these transgenic plants. This prediction was tested by combining overexpression of PAR and PAAS with overexpressing a feedback-insensitive 3-deoxy-D-arabino-heptulosonate 7-phosphate (DAHP) synthase, the latter of which has previously been shown to have hyper-induced phenylalanine biosynthesis. More than an order of magnitude increase in 2-PE accumulation was successfully achieved using this strategy, compared to overexpression of the 2-PE pathway alone. We are currently investigating the effect on lignin accumulation and composition. Thus kinetic modeling combined with in vivo time-course metabolite profiling shows to be a promising approach to define a strategy for feedstock improvements.