The function of a protein is implicit in its sequence: if you construct a protein with the proper sequence and provide it an amenable biological environment, it will fold up and start performing its nanoscale task. Many steps of this process, however, hold great mysteries, and predicting this connection between sequence and function is currently a challenge. PSI laboratories in the Enzyme Function Initiative have developed an effective pipeline for the prediction of function in bacterial enzymes. An earlier installment of the PSI Featured System described one of the early successes: the prediction of function of Tm0936, which turned out to be a deaminase (PDB entry 2plm). Work has continued on this class of enzymes, and the group recently reported the successful prediction of function in a wide variety of related enzymes.
The approach bootstraps from information that is known, beginning with the enormous body of sequence information of bacterial proteins. Then, these sequences are grouped by similarity, identifying clusters of sequences that are predicted to have similar functions. In some of these clusters, structures may be known, giving a foundation for prediction of function of the other proteins. A variety of structure-based predictive methods are then brought to bear, including homology modeling of the proteins and docking simulation to evaluate the potential binding of many different types of substrates.
A New Deaminase
The structure shown here is a deaminase from Chromobacterium violaceum (PDB entry 4f0s) discovered through this approach and building from the previous structure of Tm0936. It is also a deaminase, but docking analysis and enzymatic testing has shown that it has a different taste for substrates. Tm0936 prefers adenine nucleotides with homocysteine bound to the sugar, where the new enzyme Cv1032 prefers a more streamlined methylthio group on the substrate.
Network of Enzymes
The PSI study explored a diverse selection of enzymes in the deaminase network, shown with letters in the diagram, uncovering many interesting new functions. Both of the enzymes described above are from the large cluster shown in blue on the right side of the diagram. Some of the outliers in the clusters at the bottom include Moth1224, which was discovered to deaminate guanine, and Avi5431, which was shown to deaminate adenine bases that have been damaged by oxidation.
Active Sites Up Close
The structure of Cv1032 reveals how it recognizes a different substrate than
TM0936. Both enzymes have a similar machinery to perform the deamination
reaction, and a similar set of amino acids to bind to the base and sugar. TM0936,
however, has two arginines that recognize the carboxyl group of the homocysteine,
which are mutated in Cv1032 to remove the possibility of forming this favorable
Click on the JSmol tab to view a Jmol of this interaction.
Deaminases Tm0936 and Cv1032 (PDB entries 2plm and 4f0s)
Two deaminase structures are overlapped in this Jmol. Two catalytic amino acids are shown in green and the zinc ion and associated amino acids are shown in turquoise. Four amino acids, shown in ball-and-stick with atomic colors, are involved in substrate recognition in Tm0936. Two of these are conserved in Cv1032 and form similar hydrogen bonds with the substrate, but two are mutated, giving Cv1032 its different specificity for substrates. Use the checkbox to switch between the two structures.
Gerlt, J. A. et al. The Enzyme Function Initiative. Biochem. 50, 9950-9962 (2011).
References to Structures
Hitchcock, D. S. et al. Structure-guided discovery of new deaminase enzymes. J. Am. Chem. Soc. 135, 13927-13933 (2013).
Hermann, J. C. et al. Structure-based activity prediction for an enzyme of unknown function. Nature 448, 775-779 (2007).