Proteins power most of the individual workings of the cell and find wide use in therapeutic and industrial settings: from monoclonal antibodies that target and kill cancer cells to enzymes that convert biomass to fermentable substrates. A central question asked by my research group is how we can better design and engineer functional proteins. To answer this question we will develop and maintain particular expertise in the design and engineering of new protein-protein interactions, protein-analyte interactions, and novel enzymes using experimental approaches coupled to computational modeling and design. We focus on two main applications: (1) the microbial-mediated conversion of biomass to fuels and chemicals that more closely approximate petroleum-derived products; and (2) development of antibody and antibody-like molecules for use as protein therapeutics. 


Computational protein interface design

Proteins engage with all sorts of other proteins, nucleic acids, and small molecules; specific molecular recognition at such interfaces is determined by the interplay of various intermolecular interactions.  Using even approximate energy functions to describe this balance of forces, computational methods seeking to design such interfaces de novo have met with some success, including the design of proteins that can inhibit Influenza.  However, the success rates are too low (<<10%) to warrant wide-scale application of protein design for construction of functional proteins.  We are interested in increasing the success rate by improving search strategies as well as the underlying energy function used for the design calculations.  Our main contribution to the development of a more accurate energy function will come through high-resolution experimental mapping of protein fitness landscapes, described in detail below.

Mapping of protein fitness landscapes

Traditional laboratory evolution of bio-molecules involves screening and/or selecting from a very large pool of variants for improvements in an underlying property (e.g. thermostability, catalytic turnover, binding affinity).  Such methods grossly under-sample the potential phenotypic diversity accessible to a starting variant, as only a handful of substitutions that improve the selected property is typically uncovered.  We are interested in developing methods to observe the sequence-function fitness landscape accessible to a protein sequence.  We (among others) have developed such methods for the class of proteins that bind other proteins and are interested in building a mathematical framework for quantifying these landscapes as well as extending the methodology to other classes of proteins (e.g. enzymes).

Development of protein switches

Protein switches that can turn a signal on or off depending on the presence of an analyte would greatly facilitate the development of designer organisms in synthetic biology as well as more efficient enzymes and molecular transporters.  We are working on general, modular strategies for such protein switches, and are particularly interested in the activation of fluorescent signals both at the transcriptional and post-translational level; such signals are amenable to large-scale fluorescent activated cell sorting (FACS), which would greatly accelerate development of functional proteins.  


One promising avenue in renewable energy is the conversion of plant biomass to transportation fuels and commodity chemicals that are currently derived from petroleum-based resources.  Significant inefficiencies occur in feedstock production, deconstruction of biomass to fermentable substrates, and the conversion of the substrates to fuels and chemicals.  We have identified several problems on the deconstruction and conversion side where protein engineering may play a key role. 

On the deconstruction side, an abundant component of plant biomass is cellulose.  Cellulose is very recalcitrant to degradation, and current research is focused on a two-component system wherein a physical or chemical pretreatment disrupts the crystallinity of cellulose, and then cellulases are added to convert this pretreated cellulose to glucose.  We are working on improving the latter degradation step in several ways by: (1) improving the synergistic effects of cellulase cocktails by engineering modular scaffolding proteins; (2) directing improvements in cellulases by comprehensive mapping of the effects of point mutants on function; and (3) directing product formation to the more energy-efficient metabolite cellobiose.

For conversion, we are investigating several different avenues towards improving fermentations.  Most pretreated biomass includes small metabolites (e.g. furan, p-coumaric acid) that inhibit the growth rate and overall cell mass of fermenting microorganisms; we are interested in identifying and improving intracellular resistance pathways.  A broad hunt is on to develop microorganisms with new or existing metabolic pathways to convert biomass to fermentative products nature has never made, including diesel and jet fuel, and monomers for the plastics industry.  To that end, we are constructing metabolic pathways de novo for the production of oxygenated hydrocarbons.  Finally, transporter proteins ferry substrates and products in and out of cells; we maintain a strong interest in engineering substrate specificity and thermostability in such proteins.


Monoclonal antibodies are the dominant class of protein therapeutics in the clinic owing to facile engineering approaches and superior pharmokinetic properties.  However, the high cost of development and treatment limits antibody-based therapies for viruses except for life-threatening conditions.  Thus, we are investigating as antivirals antibody-like proteins that are cheaper to produce.  We are also exploring ways to modulate the various pharmacokinetic and “effector function” properties of antibodies, allowing for superior characteristics and more cost-effective production.