CyBase: a database of cyclic protein sequence and structure.

CyBase is a curated database and information source for backbone-cyclized proteins. The database incorporates naturally occurring cyclic proteins as well as synthetic derivatives, grafted analogues and acyclic permutants. The database provides a centralized repository of information on all aspects of cyclic protein biology and addresses issues pertaining to the management and searching of topologically circular sequences. The database is freely available at http://research.imb.uq.edu.au/cybase.

INTRODUCTION AND MOTIVATION

In recent years a number of proteins have been discovered that contain a macrocyclic backbone consisting of a continuous cycle of peptide bonds. Such macrocyclic proteins were unknown a decade ago but have now been discovered in bacteria, plants and animals (1). Unlike bacterial polyketides and small cyclic peptides such as cyclosporin that are constructed by peptide synthetases (2), these new cyclic proteins are ribosomally produced gene products, with backbone cyclization occurring as a post-translational modification. This new class of protein has excited interest because circular proteins have a range of advantages over conventional proteins (3,4). At least one class of cyclic proteins, the cyclotides, has been shown to be resistant to proteolysis and to a wide variety of adverse thermal and chemical conditions (5,6). Furthermore, since the termini of conventional proteins are often flexible, and as the degree of flexibility can be reduced by cyclization, entropic factors can lead to improved receptor binding affinities of circular proteins over corresponding acyclic proteins (7,8).

Five classes of naturally occurring proteins contain cyclic examples. These include the cyclic sex-pilin (9) and three bacteriocins (10–12) from bacterial sources, trypsin inhibitors from sunflower seeds (SFTI-1) (7,8) and the squash plant Momordica cochinchinensis (MCoTI-II) (13), the θ-defensins from macaque monkeys (14) and the cyclotides from plants of the Violaceae and Rubiaceae (15–17). The cyclotides are by far the largest family of circular proteins and ∼60 cyclotide sequences have been reported thus far. Screening programs suggest that the number of sequences may soon number in the thousands (18,19). In addition to this natural diversity, a large number of synthetic mutants, grafted analogues and acyclic permutants of cyclic proteins have been reported, and several proteins of biological interest have been artificially cyclized (20–22). This growth in information necessitates a collation of sequence/structure/function data and the development of a uniform nomenclature to prevent duplication of research and multiple naming schemes. Furthermore, sequence searching of cyclic proteins adds an extra layer of complexity, with most sequence searching tools assuming a linear sequence of amino acids. Because backbone cyclization is a ‘seamless’ post-translational modification, the location of the N- and C-termini cannot be determined from the mature sequence alone. Consequently, when searching cyclic sequences, the point at which the sequence begins and ends in the database is often arbitrary and may confound traditional searching techniques. The special considerations needed for dealing with cyclic sequences and the rapidly expanding data on their structure and function has led us to develop a curated database and web-based information source for cyclic proteins called CyBase.

APPLICATION AND DISCUSSION

CyBase incorporates a MySQL database that contains a repository of information on the amino acid and nucleic acid sequence, structure and activity of cyclic proteins. The layout of the database is shown in Figure 1. At the core of the database is the protein table, which contains information on each cyclic protein characterized. Related to each protein table entry are tables containing information on nucleic acid sequences, structure, activity and literature references. A web-based interface provides access to the information and allows for text-based searching on all data fields and filtering of results by class, source, activity or other attributes. To account for the cyclic nature of the sequences any sequence search uses a concatenation of two copies of the linear representation of the sequence to simulate a cyclic protein. Each protein has a dynamically produced sequence card, which provides cross-links to activity, nucleic acid sequence and structural information contained in the database. As the database is intended to supplement existing biological databases, links to UniProt Knowledgebase, Genbank and PDB are available for each entry and linkage with the KNOTTIN website (23) is planned. A number of other tools are provided, including coloured alignments and calculation of amino acid frequencies of selected sequences.

Figure 1

Schematic of the relational database underlying CyBase.

In general, naturally occurring cyclic proteins are small, with the largest possessing 78 amino acids. This small size makes the combination of mass spectrometry, to obtain sequence information, and NMR, to determine 3D structures, ideal for characterizing these proteins. Accordingly, to facilitate the rapid characterization of newly discovered proteins the database can be queried on molar mass, and for cyclotides, the capability exists for searching on the mass of fragments corresponding to particular inter-cysteine loops, facilitating sequence determination when utilizing reduction/alkylation of cysteine residues and tandem mass spectrometry (24,25).

Analysis of NMR-derived data such as chemical shifts and patterns of NOE connectivity can provide an early indication of the structure of a protein. To facilitate rapid structural characterization of newly discovered cyclic proteins chemical shift and restraint data from NMR-derived structures are included in the database, along with dihedral angle information. From these data, distances and regions containing defined secondary structure are calculated and stored in the database. These data can be presented visually for the analysis of short- and long-range NOE patterns, the backbone dihedral angles and chemical shift patterns. Although NMR is the most common technique for analysing these proteins, X-ray structures are also incorporated into the database and sets of inter-atom distances calculated for comparative purposes. As with the protein and nucleic acid entries, each structure possesses an information card, which contains cross-links to protein and nucleic acid entries.

Updating of the database is facilitated by a range of PERL and PHP scripts. These provide for the automated searching of sequence databases, using BLAST, to provide examples of novel cyclic proteins, and ensure quality control by preventing duplication of sequence data and renaming of already characterized sequences, a particularly important consideration for the cyclotide family, which contains potentially many sequences that may occur in a number of different species. These scripts also provide for the standardizing of residue numbering in new cyclotide structures. Despite these automations the addition of new entries is performed manually to ensure maximum quality of database entries.

We plan to extend the database by utilizing the growing number of cyclotide structures to provide predictions of cyclotide secondary structures based on primary sequence and to develop methods to search the structures in the database based on the similarity between selected inter-atomic distances and NOE connectivities. We also plan to improve the information content of the database by including hydrogen bond and other structural information as well as homology models for cyclotide sequences that have not yet been structurally characterized. CyBase is available at and given the growing interest in backbone cyclization, it is hoped that CyBase will prove to be a useful resource in the field of structural biology. Suggestions should be directed to D.C.