BackgroundNon-productive binding of enzymes to lignin is thought to impede the saccharification efficiency of pretreated lignocellulosic biomass to fermentable sugars. Due to a lack of suitable analytical techniques that track binding of individual enzymes within complex protein mixtures and the difficulty in distinguishing the contribution of productive (binding to specific glycans) versus non-productive (binding to lignin) binding of cellulases to lignocellulose, there is currently a poor understanding of individual enzyme adsorption to lignin during the time course of pretreated biomass saccharification.ResultsIn this study, we have utilized an FPLC (fast protein liquid chromatography)-based methodology to quantify free Trichoderma reesei cellulases (namely CBH I, CBH II, and EG I) concentration within a complex hydrolyzate mixture during the varying time course of biomass saccharification. Three pretreated corn stover (CS) samples were included in this study: Ammonia Fiber Expansiona (AFEX™-CS), dilute acid (DA-CS), and ionic liquid (IL-CS) pretreatments. The relative fraction of bound individual cellulases varied depending not only on the pretreated biomass type (and lignin abundance) but also on the type of cellulase. Acid pretreated biomass had the highest levels of non-recoverable cellulases, while ionic liquid pretreated biomass had the highest overall cellulase recovery. CBH II has the lowest thermal stability among the three T. reesei cellulases tested. By preparing recombinant family 1 carbohydrate binding module (CBM) fusion proteins, we have shown that family 1 CBMs are highly implicated in the non-productive binding of full-length T. reesei cellulases to lignin.ConclusionsOur findings aid in further understanding the complex mechanisms of non-productive binding of cellulases to pretreated lignocellulosic biomass. Developing optimized pretreatment processes with reduced or modified lignin content to minimize non-productive enzyme binding or engineering pretreatment-specific, low-lignin binding cellulases will improve enzyme specific activity, facilitate enzyme recycling, and thereby permit production of cheaper biofuels.

Biological-mediated conversion of pretreated lignocellulosic biomass to biofuels and biochemicals is a promising avenue toward energy sustainability. However, a critical impediment to the commercialization of cellulosic biofuel production is the high cost of cellulase enzymes needed to deconstruct biomass into fermentable sugars. One major factor driving cost is cellulase adsorption and inactivation in the presence of lignin, yet we currently have a poor understanding of the protein structure-function relationships driving this adsorption. In this work, we have systematically investigated the role of protein surface potential on lignin adsorption using a model monomeric fluorescent protein. We have designed and experimentally characterized 16 model protein variants spanning the physiological range of net charge (-24 to +16 total charges) and total charge density (0.28-0.40 charges per sequence length) typical for natural proteins. Protein designs were expressed, purified, and subjected to in silico and in vitro biophysical measurements to evaluate the relationship between protein surface potential and lignin adsorption properties. The designs were comparable to model fluorescent protein in terms of thermostability and heterologous expression yield, although the majority of the designs unexpectedly formed homodimers. Protein adsorption to lignin was studied at two different temperatures using Quartz Crystal Microbalance with Dissipation Monitoring and a subtractive mass balance assay. We found a weak correlation between protein net charge and protein-binding capacity to lignin. No other single characteristic, including apparent melting temperature and 2nd virial coefficient, showed correlation with lignin binding. Analysis of an unrelated cellulase dataset with mutations localized to a family I carbohydrate-binding module showed a similar correlation between net charge and lignin binding capacity. Overall, our study provides strategies to identify highly active, low lignin-binding cellulases by either rational design or by computational screening genomic databases. Biotechnol. Bioeng. 2017;114: 740-750. © 2016 Wiley Periodicals, Inc.