Claes Gustafsson

Directed evolution: the 'rational' basis for 'irrational' design.

The development of powerful genetic manipulation formats has revolutionized the creation of functional biological molecules. Recent advances in directed evolution demonstrate that multiple properties of proteins can be optimized simultaneously and rapidly. Improved proteins often contain multiple and dispersed substitutions that act synergistically to improve enzyme properties and function. The benefits of such multiple changes are often not predictable from a priori structural knowledge. Furthermore, alternative solutions to gaining functional change can be obtained.

Engineering proteinase K using machine learning and synthetic genes.

BackgroundAltering a proteins function by changing its sequence allows natural proteins to be converted into useful molecular tools. Current protein engineering methods are limited by a lack of high throughput physical or computational tests that can accurately predict protein activity under conditions relevant to its final application. Here we describe a new synthetic biology approach to protein engineering that avoids these limitations by combining high throughput gene synthesis with machine learning-based design algorithms.ResultsWe selected 24 amino acid substitutions to make in proteinase K from alignments of homologous sequences. We then designed and synthesized 59 specific proteinase K variants containing different combinations of the selected substitutions. The 59 variants were tested for their ability to hydrolyze a tetrapeptide substrate after the enzyme was first heated to 68°C for 5 minutes. Sequence and activity data was analyzed using machine learning algorithms. This analysis was used to design a new set of variants predicted to have increased activity over the training set, that were then synthesized and tested. By performing two cycles of machine learning analysis and variant design we obtained 20-fold improved proteinase K variants while only testing a total of 95 variant enzymes.ConclusionThe number of protein variants that must be tested to obtain significant functional improvements determines the type of tests that can be performed. Protein engineers wishing to modify the property of a protein to shrink tumours or catalyze chemical reactions under industrial conditions have until now been forced to accept high throughput surrogate screens to measure protein properties that they hope will correlate with the functionalities that they intend to modify. By reducing the number of variants that must be tested to fewer than 100, machine learning algorithms make it possible to use more complex and expensive tests so that only protein properties that are directly relevant to the desired application need to be measured. Protein design algorithms that only require the testing of a small number of variants represent a significant step towards a generic, resource-optimized protein engineering process.

Systematic Variation of Amino Acid Substitutions for Stringent Assessment of Pairwise Covariation

During protein evolution, amino acids change due to a combination of functional constraints and genetic drift. Proteins frequently contain pairs of amino acids that appear to change together (covariation). Analysis of covariation from naturally occurring sets of orthologs cannot distinguish between residue pairs retained by functional requirements of the protein and those pairs existing due to changes along a common evolutionary path. Here, we have separated the two types of covariation by independently recombining every naturally occurring amino acid variant within a set of 15 subtilisin orthologs. Our analysis shows that in this family of subtilisin orthologs, almost all possible pairwise combinations of amino acids can coexist. This suggests that amino acid covariation found in the subtilisin orthologs is almost entirely due to common ancestral origin of the changes rather than functional constraints. We conclude that naturally occurring sequence diversity can be used to identify positions that can vary independently without destroying protein function.

Putting engineering back into protein engineering: bioinformatic approaches to catalyst design.

Complex multivariate engineering problems are commonplace and not unique to protein engineering. Mathematical and data-mining tools developed in other fields of engineering have now been applied to analyze sequence-activity relationships of peptides and proteins and to assist in the design of proteins and peptides with specified properties. Decreasing costs of DNA sequencing in conjunction with methods to quickly synthesize statistically representative sets of proteins allow modern heuristic statistics to be applied to protein engineering. This provides an alternative approach to expensive assays or unreliable high-throughput surrogate screens.

Synergistic modular promoter and gene optimization to push cellulase secretion by Pichia pastoris beyond existing benchmarks.

Although successfully used for heterologous gene expression for more than twenty years, general knowledge about all factors influencing protein expression by Pichia pastoris is still lacking. For high titers of protein clones are optimized individually for each target protein. Optimization efforts in this study were focused on the DNA level, evaluating a set of 48 different individual synthetic genes (TrCBH2) coding for the same protein sequence of a Trichoderma reesei cellulase in combination with three different promoter sequences: PGAP (constitutive) and the synthetic AOX1 promoter variants PDeS (derepressed) and PEn (enhanced, inducible). Expression of active secreted enzyme varied from undetectable to ∼300% of the best known gene, as determined by secreted enzyme activity analyses of supernatants from 96 well plate and bioreactor cultivations. Finally, the best optimized gene and new promoters were combined to engineer highly productive P. pastoris CBH2 expression strains. Although no methanol was used for induction a final titer of more than 18g/l of secreted protein was produced under controlled conditions in small scale bioreactor cultivations after 60-70h of growth limiting glycerol feed. This is the highest concentration of secreted enzyme in P. pastoris published so far and single parts of the expression cassette could be independently optimized showing additive effects for improvements in protein production by P. pastoris.

Mapping of Amino Acid Substitutions Conferring Herbicide Resistance in Wheat Glutathione Transferase

We have used design of experiments (DOE) and systematic variance to efficiently explore glutathione transferase substrate specificities caused by amino acid substitutions. Amino acid substitutions selected using phylogenetic analysis were synthetically combined using a DOE design to create an information-rich set of gene variants, termed infologs. We used machine learning to identify and quantify protein sequence-function relationships against 14 different substrates. The resulting models were quantitative and predictive, serving as a guide for engineering of glutathione transferase activity toward a diverse set of herbicides. Predictive quantitative models like those presented here have broad applicability for bioengineering.