Reports: DNI655975-DNI6: Toward Improved Non-Aqueous Biocatalysis: Mapping Relationships between Enzyme Structure, Interfacial Solvent Dynamics, and Enzyme Dynamics

Katie R. Mitchell-Koch, PhD, Wichita State University

This work involves the characterization of the dynamics of organic solvents within the solvation shell of the protein Candida antarctica lipase B (CALB) and other solvent-compatible enzymes. Because there was no established methodology for simulating enzymes in organic solvent, we first developed a protocol for determining how many and which crystallographic waters to keep when simulating enzymes under non-aqueous conditions.  This was evaluated by determining under which simulation conditions the protein reached an equilibrated structure fastest.  It was found that keeping buried or slow-diffusing waters leads to fastest protein equilibration times in organic solvents.  Another outcome of the study was the observation that the location of crystallographic waters kept affects conformational dynamics of the enzyme (opening/closing of the substrate binding cleft), but the nature of the effect is solvent-dependent. The protocol for carrying out protein simulations in organic solvent has been used for subsequent work, in which interfacial solvent dynamics of acetonitrile, n-butanol, and tert-butanol in the solvation layer of CALB have been studied.  Translational and rotational dynamics have been characterized, in addition to protein-solvent hydrogen bond lifetimes and residence times within the solvation layer.  Findings support two of the hypotheses laid out in the proposal, namely that the dynamics of organic solvents in the solvation layer are heterogeneous (as has been found for biomolecular hydration dynamics), and that the regions of fast and slow solvent dynamics vary by the nature of the solvent. 

Work is underway to determine to what extent various features of the protein surface (hydrophobicity, topology, etc.) influence solvent dynamics.  It appears that the curvature of the protein surface has the highest correlation with the dynamics of the solvation shell within a given region, with some influence from hydrophobicity also observed.  The hydration dynamics around CALB have been examined, to provide a comparison with native (aqueous) protein solvation and structure-dynamics-function.  For CALB in water, the radial distribution function of the solvation shell has been calculated by region.  We have seen that water dynamics correlate with density of solvent at the interface.  In this case, higher density regions display faster dynamics, as seen in bulk water.  This finding leads to another area of inquiry for characterizing organic solvents: are there relationships between solvent structure in the solvation shell and the dynamics of the solvent? The density-diffusion relationship in bulk water is considered anomalous, so extending the work to organic solvents will be an interesting research direction.

The connection between interfacial solvent dynamics and enzyme dynamics is another key question in this work.  Previous experimental work has shown that the primary conformational dynamics of CALB involve the opening and closing of helix a5 for substrate binding and release.  In our simulations, protein dynamics have been characterized using root mean squared fluctuation (RMSF).  Work is underway to characterize protein dynamics in a statistically significant way.  It has been observed that, when comparing results across four solvents (acetonitrile, n-butanol, tert-butanol, water), protein flexibility by region correlates with the mobility of the solvent in the surrounding solvation shell. This is specifically noted around the a5 helix, whose conformational dynamics influence enzymatic reaction rates.  We will tie simulations results to published kinetics data in different solvents.

Two graduate students have been supported by the PRF-DNI award during grant year 1, and both are anticipated to graduate with PhDs in December.  The funded graduate students have developed skills in coding and statistical analysis.  In the case of one student, the project provided exposure to molecular dynamics simulations, adding a new skill in computational chemistry.  Her prior work had exclusively used electronic structure methods.  The breadth of her education seems to be paying off: she is in the process of interviewing for postdoctoral research and industrial jobs that would be excellent opportunities and use of her expertise, and there has been significant interest from several potential employers. The primary graduate student funded by the work has developed into a skilled scientist who works with considerable independence, and this project has pushed her into new areas of analysis and theory development.  After defending in December, she will be joining me as a postdoctoral researcher in the spring to continue on with this work.  The project will benefit immensely from the focus of this postdoctoral researcher, given her extensive skills, intellect, and familiarity with the work. 

The first 9 months of work funded by the PRF-DNI award served as a starting point for a proposal submitted to the National Science Foundation in September 2016 with collaborator Prof. Vinh Nguyen, a THz spectroscopist at Virginia Tech.  The grant was funded by the Chemical Structure, Dynamics, and Mechanisms-B program of the Chemistry Division in August 2017.  Having the PRF doctoral new investigator award allowed for preliminary data that was included in the NSF proposal, and the combined funding secured for research efforts in non-aqueous biocatalysis and solvent dynamics has helped lay the foundations of my ongoing research career and tenure package (to be submitted August 2018).  Having an experimental collaborator also strengthens the research, providing spectroscopic data to interpret alongside modeling results.  It will also enable my group to improve the modeling protocol for enzymes in organic solvent, which is a fairly understudied field in spite of its relevance to industrial processes.  Spectroscopic results will be used to refine force field choices and parameters, enabling simulation with higher fidelity.  In the meantime, it looks like we are laying the groundwork for advancing theories to understand the dynamics of solvents at the biomolecular interface, with a prospect for practical applications in solvent selection and rational bioengineering of proteins for improved non-aqueous biocatalysis.  Non-aqueous biocatalysis is particularly attractive for petroleum-based reagents, as the processes take place at mild temperatures, while use of organic solvents (in contrast to the typical aqueous buffers used in enzymatic reactions) solubilizes hydrocarbon substrates. In the end, the outcomes of the funded work are poised to enable more efficient manufacture of commodity and fine chemicals from petroleum stocks.