Computational methods are increasingly gaining importance as an aid in identifying active sites. Mostly these methods tend to have structural information that supplement sequence conservation based analyses. Development of tools that compute electrostatic potentials has further improved our ability to better characterize the active site residues in proteins. We have described a computational methodology for detecting active sites based on structural and electrostatic conformity - C ata L ytic A ctive S ite P rediction (CLASP). In our pipelined model, physical 3D signature of any particular enzymatic function as defined by its active sites is used to obtain spatially congruent matches. While previous work has revealed that catalytic residues have large pKa deviations from standard values, we show that for a given enzymatic activity, electrostatic potential difference (PD) between analogous residue pairs in an active site taken from different proteins of the same family are similar. False positives in spatially congruent matches are further pruned by PD analysis where cognate pairs with large deviations are rejected. We first present the results of active site prediction by CLASP for two enzymatic activities - β-lactamases and serine proteases, two of the most extensively investigated enzymes. The results of CLASP analysis on motifs extracted from Catalytic Site Atlas (CSA) are also presented in order to demonstrate its ability to accurately classify any protein, putative or otherwise, with known structure. The source code and database is made available at www.sanchak.com/clasp/. Subsequently, we probed alkaline phosphatases (AP), one of the well known promiscuous enzymes, for additional activities. Such a search has led us to predict a hitherto unknown function of shrimp alkaline phosphatase (SAP), where the protein acts as a protease. Finally, we present experimental evidence of the prediction by CLASP by showing that SAP indeed has protease activity in vitro.

Error-prone PCR, DNA shuffling, and saturation mutagenesis are techniques used by protein engineers to mimic the natural "evolutionary walk" that conjures new enzymes. Rational design is often critical in efforts to accelerate this "random walk" into a "resolute sprint." Previous work by our group established a computational method for detecting active sites (CLASP) based on spatial and electrostatic properties of catalytic residues, and a method to quantify promiscuous activities in a wide range of proteins (PROMISE). Here, we describe a rational design flow (DECAAF) based on the PROMISE methodology to choose a protein which, when subjected to minimal mutations, is most likely to mirror the scaffold of a desired enzymatic function. Modeling the diversity in catalytic sites and providing precise user control to guide the search is a key goal of our implementation. The flow details have been worked out in a real-life example to select a plant protein to substitute for human neutrophil elastase in a chimeric antimicrobial enzyme designed to bolster the innate immune defense system in plants.

Edited by Miguel De la RosaWe demonstrate the inhibition of the native phosphatase activity of a cold active alkaline phosphatase from Vibrio (VAP) (IC 50 of 44 ± 4 (n = 4) lM at pH 7.0 after a 30 min preincubation) by a specific b-lactam compound (only by imipenem, and not by ertapenem, meropenem, ampicillin or penicillin G). The homologous scaffold was detected by an in silico analysis that established the spatial and electrostatic congruence of the active site of a Class B2 CphA metallo-b-lactamase from Aeromonas hydrophila to the active site of VAP. The tested b-lactam compounds did not inhibit Escherichia coli or shrimp alkaline phosphatase, which could be ascribed to the lower congruence indicated by CLASP. There was no discernible b-lactamase activity in the tested alkaline phosphatases. This is the first time a scaffold recognizing imipenem in an alkaline phosphatase (VAP) has been demonstrated.