In silico characterization of protein chimeras: relating sequence and function within the same fold.

The exploration of novel proteins via recombination of fragments derived from structurally homologous proteins has enormous potential for medicine and biotechnology. This modular exchange of sequence material puts novel activities, substrate specificities, and stability within reach of a semi-random search. This article takes stock of the growing resource of experimentally characterized chimeric proteins within a homologous protein family to build sequence-function models that can effectively guide the construction of new libraries. A novel framework for predicting structural viability of chimeric proteins, only assuming knowledge of their sequence and their parental structure, is presented. Removing a major barrier in previous work, the model processes any sequence that derives from parents with similar folds. The method naturally mixes test and training data from site-directed recombination, DNA shuffling, or random mutagenesis experiments. We train a model from a site-directed recombination library with state-of-the-art prediction accuracy on hold-out test data from the same experimental source and convincing performance on chimeras with a different origin. Specifically, the model is used to assess the structural viability of P450 chimeras deriving from proteins with only 18% sequence similarity to those used for model tuning.

Allelic variants of drug metabolizing enzymes as risk factors in psoriasis.

The onset or exacerbation of psoriasis, a T-cell-dependent skin disease with autoimmune features, can be triggered by drugs such as antimalarials and beta-blockers. Xenobiotics may also play a role in idiopathic psoriasis. It has been hypothesized that different metabolic efficiencies caused by variant alleles of xenobiotic metabolizing enzymes could lead to the accumulation of xenobiotics or their reactive metabolites in target organs. Subsequently, neoantigens or cryptic peptides could be presented and initiate an aggressive T cell response. In this context, we analyzed a broad array of xenobiotic metabolizing enzymes in up to 327 Caucasian psoriasis patients and compared them to 235 control persons. Alleles tested include four phase I and three phase II enzymes. Significantly more carriers of the variant alleles of CYP1A1 (alleles *2A and *2C) were found in healthy controls than in patients, suggesting a protective role for these alleles. No significant difference between patients and controls could be found, however, for the other phase I alleles 1B1*1 and 1B1 *3, 2C19*1A and 2C19*2A, and 2E1*1A and 2E1*5B. Of the variant alleles coding for phase II enzymes only GSTM1, but not GSTT1 or NQOR, correlated with a risk to contract psoriasis. Some combinations of phase I and phase II enzymes suggested protective or risk-associated effects. Interestingly, heterozygosity for CYP2C19 alleles *1A and *2A was associated with increased risk for "late onset" psoriasis, whereas this genotype was protective for psoriatic arthritis. This is the first large-scale study on these enzymes and the results obtained support the concept that different activities of metabolizing enzymes can contribute to disease etiology and progression.