The imperative to limit climate change and dependence on fossil resources has led to an increased interest in the use of lignocellulosic biomass (LB) for the production of biofuel, bio-based chemicals and materials. Nevertheless, LB is recalcitrant to enzymatic deconstruction and requires expensive pretreatment steps. Although many studies have focused on the causes of this recalcitrance, the evaluated markers are mostly limited to chemical properties. In this study, we combine 4D (space + time) confocal data acquisition, image processing, and computational modelling to better understand and predict the LB deconstruction occurring at cellular scale as well as its underlying parameters. Using lignin autofluorescence of plant cell walls, we first collected 4D images of poplar wood samples during enzymatic hydrolysis by confocal microscopy. These images were processed using an existing pipeline [1] including cell wall segmentation and tracking to follow the dynamics of voxels’ intensities over time and estimate cellular structural parameters such as cell wall volume, accessible surface area or compactness. A preliminary data analysis was then performed which indicated a linear relation between the variation of the cell wall intensity and the kinetic of cellulose conversion. Considering this observation, we developed an image-based model predicting the variation of voxels’ intensity through time as a proxy for LB deconstruction. The model consists of a set of ordinary differential equations (ODE) that have been calibrated for cell wall different regions separately (inner cell wall, cell’s edges and corners). To do this, the model’s parameters were estimated by minimizing the mean square error (MSE) of the model in each region previously segmented. After calibration, the model showed a very good ability to predict the intensity variation over time for all regions of the sample. Notably, the estimated parameters values in each region confirmed the strong heterogeneity of LB deconstruction at cellular scale. Finally, these values were compared to the quantified structural parameters revealing correlations between (i) the accessible surface area and the cell wall compactness and (ii) the early stage LB deconstruction dynamic predicted by the model.

Computational modelling of lignocellulosic biomass deconstruction during enzymatic hydrolysis

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.