Enzyme optimization: moving from blind evolution to statistical exploration of sequence-function space

Implications of available design space for identification of non-immunogenic protein therapeutics

Immunogenicity/antibody responses are major issues for parenteral proteins and nanotherapeutics (nanovectors, diagnostics, theranostics, etc.), and robust antibody responses require T-helper epitopes. T-helper epitopes consist of specific amino acids at specific positions (anchor positions) in immunogens which contact the major histocompatibility complex (MHC), provide most of the energy for MHC binding and constitute the binding motif for the corresponding MHC alleles. We developed an algorithm that considers motifs to design vaccines lacking unwanted T-cell epitopes, and found numbers of such vaccines can be astronomical (Lee et al. 2009). The algorithm can be used to design reduced immunogenicity proteins, and numbers of predicted proteins are also immense. Reducing T-helper epitope content reduces protein immunogenicity, but the depth of mutagensis needed to eliminate immunogenicity is commonly assumed to be too great for retention of protein bioactivity. However, very deep, but successful substitution, insertion and deletion mutagenesis have been reported. These reports and design space the algorithm reveals suggest development of non-immunogenic therapeutics might be more feasible than commonly assumed.

Enzyme engineering for enantioselectivity: from trial-and-error to rational design?

The availability of tailored enzymes is crucial for the implementation of biocatalysis in organic chemistry. Enantioselectivity is one key parameter defining the usefulness of an enzyme and, therefore, the competitiveness of the corresponding industrial process. Hence, identification of enzymes with high enantioselectivity in the desired transformation is important. Currently, this is achieved by screening collections and libraries comprising natural or man-made diversity for the desired trait. Recently, a variety of improved methods have been developed to generate and screen this diversity more efficiently. Here, we present and discuss the most important advances in both library generation and screening. We also evaluate future trends, such as moving from random evolution to more rational.

Machine learning integration for predicting the effect of single amino acid substitutions on protein stability

BACKGROUND

Computational prediction of protein stability change due to single-site amino acid substitutions is of interest in protein design and analysis. We consider the following four ways to improve the performance of the currently available predictors: (1) We include additional sequence- and structure-based features, namely, the amino acid substitution likelihoods, the equilibrium fluctuations of the alpha- and beta-carbon atoms, and the packing density. (2) By implementing different machine learning integration approaches, we combine information from different features or representations. (3) We compare classification vs. regression methods to predict the sign vs. the output of stability change. (4) We allow a reject option for doubtful cases where the risk of misclassification is high.

RESULTS

We investigate three different approaches: early, intermediate and late integration, which respectively combine features, kernels over feature subsets, and decisions. We perform simulations on two data sets: (1) S1615 is used in previous studies, (2) S2783 is the updated version (as of July 2, 2009) extracted also from ProTherm. For S1615 data set, our highest accuracy using both sequence and structure information is 0.842 on cross-validation and 0.904 on testing using early integration. Newly added features, namely, local compositional packing and the mobility extent of the mutated residues, improve accuracy significantly with intermediate integration. For S2783 data set, we also train regression methods to estimate not only the sign but also the amount of stability change and apply risk-based classification to reject when the learner has low confidence and the loss of misclassification is high. The highest accuracy is 0.835 on cross-validation and 0.832 on testing using only sequence information. The percentage of false positives can be decreased to less than 0.005 by rejecting 10 per cent using late integration.

CONCLUSION

We find that in both early and late integration, combining inputs or decisions is useful in increasing accuracy. Intermediate integration allows assessing the contributions of individual features by looking at the assigned weights. Overall accuracy of regression is not better than that of classification but it has less false positives, especially when combined with the reject option. The server for stability prediction for three integration approaches and the data sets are available at http://www.prc.boun.edu.tr/appserv/prc/mlsta.

Chemical synthesis using synthetic biology

An immense array of naturally occurring biological systems have evolved that convert simple substrates into the products that cells need for growth and persistence. Through the careful application of metabolic engineering and synthetic biology, this biotransformation potential can be harnessed to produce chemicals that address unmet clinical and industrial needs. Developing the capacity to utilize biology to perform chemistry is a matter of increasing control over both the function of synthetic biological systems and the engineering of those systems. Recent efforts have improved general techniques and yielded successes in the use of synthetic biology for the production of drugs, bulk chemicals, and fuels in microbial platform hosts. Synthetic promoter systems and novel RNA-based, or riboregulator, mechanisms give more control over gene expression. Improved methods for isolating, engineering, and evolving enzymes give more control over substrate and product specificity and better catalysis inside the cell. New computational tools and methods for high-throughput system assembly and analysis may lead to more rapid forward engineering. We highlight research that reduces reliance upon natural biological components and point to future work that may enable more rational design and assembly of synthetic biological systems for synthetic chemistry.

Directed evolution of an enantioselective epoxide hydrolase: uncovering the source of enantioselectivity at each evolutionary stage

Directed evolution of enzymes as enantioselective catalysts in organic chemistry is an alternative to traditional asymmetric catalysis using chiral transition-metal complexes or organocatalysts, the different approaches often being complementary. Moreover, directed evolution studies allow us to learn more about how enzymes perform mechanistically. The present study concerns a previously evolved highly enantioselective mutant of the epoxide hydrolase from Aspergillus niger in the hydrolytic kinetic resolution of racemic glycidyl phenyl ether. Kinetic data, molecular dynamics calculations, molecular modeling, inhibition experiments, and X-ray structural work for the wild-type (WT) enzyme and the best mutant reveal the basis of the large increase in enantioselectivity (E = 4.6 versus E = 115). The overall structures of the WT and the mutant are essentially identical, but dramatic differences are observed in the active site as revealed by the X-ray structures. All of the experimental and computational results support a model in which productive positioning of the preferred (S)-glycidyl phenyl ether, but not the (R)-enantiomer, forms the basis of enhanced enantioselectivity. Predictions regarding substrate scope and enantioselectivity of the best mutant are shown to be possible.

Computational tools for designing and engineering biocatalysts

Current computational tools to assist experimentalists for the design and engineering of proteins with desired catalytic properties are reviewed. The applications of these tools for de novo design of protein active sites, optimization of substrate access and product exit pathways, redesign of protein-protein interfaces, identification of neutral/advantageous/deleterious mutations in the libraries from directed evolution and stabilization of protein structures are described. Remarkable progress is seen in de novo design of enzymes catalyzing a chemical reaction for which a natural biocatalyst does not exist. Yet, constructed biocatalysts do not match natural enzymes in their efficiency, suggesting that more research is needed to capture all the important features of natural biocatalysts in theoretical designs.

Improved PCR method for the creation of saturation mutagenesis libraries in directed evolution: application to difficult-to-amplify templates

Saturation mutagenesis constitutes a powerful method in the directed evolution of enzymes. Traditional protocols of whole plasmid amplification such as Stratagene’s QuikChange™ sometimes fail when the templates are difficult to amplify. In order to overcome such restrictions, we have devised a simple two-primer, two-stage polymerase chain reaction (PCR) method which constitutes an improvement over existing protocols. In the first stage of the PCR, both the mutagenic primer and the antiprimer that are not complementary anneal to the template. In the second stage, the amplified sequence is used as a megaprimer. Sites composed of one or more residues can be randomized in a single PCR reaction, irrespective of their location in the gene sequence.The method has been applied to several enzymes successfully, including P450-BM3 from Bacillus megaterium, the lipases from Pseudomonas aeruginosa and Candida antarctica and the epoxide hydrolase from Aspergillus niger. Here, we show that megaprimer size as well as the direction and design of the antiprimer are determining factors in the amplification of the plasmid. Comparison of the results with the performances of previous protocols reveals the efficiency of the improved method.

Predicting positive p53 cancer rescue regions using Most Informative Positive (MIP) active learning

Many protein engineering problems involve finding mutations that produce proteins with a particular function. Computational active learning is an attractive approach to discover desired biological activities. Traditional active learning techniques have been optimized to iteratively improve classifier accuracy, not to quickly discover biologically significant results. We report here a novel active learning technique, Most Informative Positive (MIP), which is tailored to biological problems because it seeks novel and informative positive results. MIP active learning differs from traditional active learning methods in two ways: (1) it preferentially seeks Positive (functionally active) examples; and (2) it may be effectively extended to select gene regions suitable for high throughput combinatorial mutagenesis. We applied MIP to discover mutations in the tumor suppressor protein p53 that reactivate mutated p53 found in human cancers. This is an important biomedical goal because p53 mutants have been implicated in half of all human cancers, and restoring active p53 in tumors leads to tumor regression. MIP found Positive (cancer rescue) p53 mutants in silico using 33% fewer experiments than traditional non-MIP active learning, with only a minor decrease in classifier accuracy. Applying MIP to in vivo experimentation yielded immediate Positive results. Ten different p53 mutations found in human cancers were paired in silico with all possible single amino acid rescue mutations, from which MIP was used to select a Positive Region predicted to be enriched for p53 cancer rescue mutants. In vivo assays showed that the predicted Positive Region: (1) had significantly more (p<0.01) new strong cancer rescue mutants than control regions (Negative, and non-MIP active learning); (2) had slightly more new strong cancer rescue mutants than an Expert region selected for purely biological considerations; and (3) rescued for the first time the previously unrescuable p53 cancer mutant P152L.

Engineering proteins to acquire or enhance a particular useful function is at the core of many biomedical problems. This paper presents Most Informative Positive (MIP) active learning, a novel integrated computational/biological approach designed to help guide biological discovery of novel and informative positive mutants. A classifier, together with modeled structure-based features, helps guide biological experiments and so accelerates protein engineering studies. MIP reduces the number of expensive biological experiments needed to achieve novel and informative positive results. We used the MIP method to discover novel p53 cancer rescue mutants. p53 is a tumor suppressor protein, and destructive p53 mutations have been implicated in half of all human cancers. Second-site cancer rescue mutations restore p53 activity and eventually may facilitate rational design of better cancer drugs. This paper shows that, even in the first round of in vivo experiments, MIP significantly increased the discovery rate of novel and informative positive mutants.