ARAMEMNON, a novel database for Arabidopsis integral membrane proteins

Abstract

A specialized database (DB) for Arabidopsis membrane proteins, ARAMEMNON, was designed that facilitates the interpretation of gene and protein sequence data by integrating features that are presently only available from individual sources. Using several publicly available prediction programs, putative integral membrane proteins were identified among the approximately 25,500 proteins in the Arabidopsis genome DBs. By averaging the predictions from seven programs, approximately 6,500 proteins were classified as transmembrane (TM) candidate proteins. Some 1,800 of these contain at least four TM spans and are possibly linked to transport functions. The ARAMEMNON DB enables direct comparison of the predictions of seven different TM span computation programs and the predictions of subcellular localization by eight signal peptide recognition programs. A special function displays the proteins related to the query and dynamically generates a protein family structure. As a first set of proteins from other organisms, all of the approximately 700 putative membrane proteins were extracted from the genome of the cyanobacterium Synechocystis sp. and incorporated in the ARAMEMNON DB. The ARAMEMNON DB is accessible at the URL http://aramemnon.botanik.uni-koeln.de.

Figure 1

Uniformity of TM span predictions. For each class of proteins with a median TM span number between one and 14, the proportion of proteins indicated for that same number of TM domains is predicted by all seven programs used or by six, five, four, or less than four programs, respectively.

Figure 2

Graphical representation of TM span predictions. ARAMEMNON displays plots of the TM predictions by seven programs. A, TM span predictions for an aquaporin MIP-like protein (At3g54820). All seven programs uniformly predict six TM spans at almost the same positions, and TmHMM 2.0, HmmTop 2.0, TmPred and TMap predict the same orientation within the membrane. The shading intensity of the membrane span candidate segments indicates the mean hydrophobicity range according to a normalized hydrophobicity scale (Eisenberg et al., 1984): white, 0 to 0.24; light gray, 0.25 to 0.49; dark gray, 0.50 to 0.74; and black, 0.75 to 0.99. B, TM span predictions for the PPT2 protein (At3g01550). The predictions differ in the number of TM spans, location of TM spans, and orientation of the protein within the membrane. C, For each prediction, the details are shown. #, Predicted TM spans starting from the N terminus; Pos, location of the TM span in the protein; HyPhob, mean hydrophobicity within the membrane span candidate segment (Eisenberg et al., 1984). The shading intensity correlates to that in A; AmPhil, relative maximal amphiphilicity within the membrane candidate segment.

Figure 3

TPT-related PTs identified by ARAMEMNON DB. A, List-form display of TPT-related proteins. The columns show (from left to right): amino acid similarity excluding gaps and Smith-Waterman score; organism (At, Arabidopsis) and chromosome number; gene annotation (with links to relevant publications); gene name (with links to TIGR and Munich Information Center for Protein Sequences DBs); protein accession number in GenBank (with link to National Center for Biotechnology Information [NCBI]); button to call protein, cDNA, genomic DNA, 5′- and 3′-untranslated region sequence display; button to call the TM and signal sequence prediction display; button to call the family structure (cluster) display, as shown in B. B, Columns from left to right: average subcluster amino acid similarity levels; members of the subclusters; chromosomal location of genes. C, NJ tree based on a multiple alignment of PT sequences performed by ClustalX. The numbers beside branches indicate the frequency (%) with which the branch was found in 1,000 bootstrap replicas. The shaded branches correspond to the subclusters generated by ARAMEMNON DB.

Figure 4

Amino acid permease (AAP) protein family structure generated by ARAMEMNON DB in comparison to the NJ tree. A, ARAMEMNON family structure (cluster) display of the AAP proteins. B, Schematic graphic to illustrate the relationships between the AAP protein subclusters listed in A. C, NJ tree of the AAP family, based on a multiple alignment of amino acid permease sequences performed by ClustalX.

Figure 5

Family structure and TM domain topology of the putative K/VAG transporters. A, Graphical illustration the of At1g21870-related membrane protein family structure generated by the ARAMEMNON DB. B, NJ tree of the At1g21870-related proteins. C, The membrane topologies of the K/VAG transporters were predicted by using the TM protein prediction program Eiconda 0.9. Shading of the boxes that symbolize the TM spans is as in Figure 2. The graphical outputs are drawn to scale and were manually aligned. The length of the proteins in amino acids is indicated to the right.