View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Elsevier - Publisher Connector FEBS Letters 580 (2006) 1017–1022 Minireview Membrane topology of human ABC proteins Ga´bor E. Tusna´dya, Bala´zs Sarkadib, Istva´n Simona, Andra´sVa´radia,* a Institute of Enzymology, Hungarian Academy of Sciences, Karolina u´t 29, 1113 Budapest, Hungary b National Medical Center, and the Membrane Research Group of the Hungarian Academy of Sciences, Daro´czi ut 24, 1113 Budapest, Hungary Received 11 October 2005; revised 11 November 2005; accepted 11 November 2005 Available online 1 December 2005 Edited by Gerrit van Meer Several computer-assisted empirical prediction methods are Abstract In this review, we summarize the currently available information on the membrane topology of some key members available to generate the hydrophobicity profiles and other se- of the human ABC protein subfamilies, and present the predicted quence-based profiles for a putative transmembrane protein. domain arrangements. In the lack of high-resolution structures However, such analyses should be regarded only as ‘‘educated for eukaryotic ABC transporters this topology is based only on guesses’’, serving as good starting points for an experimental prediction algorithms and biochemical data for the location of elucidation of the membrane topology. various segments of the polypeptide chain, relative to the mem- Another experimental strategy is to predict the three-dimen- brane. We suggest that topology models generated by the avail- sional structure of a eukaryotic ABC transporter by homology able prediction methods should only be used as guidelines to modeling, based on the available crystal structure of a pro- provide a basis of experimental strategies for the elucidation of karyotic ABC transporter. In principle this is a valid approach, the membrane topology. as by now four high-resolution prokaryotic ABC-transport Ó 2005 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. structures are available [1–4], which are discussed in detail in another review of this issue. Indeed, such homology models Keywords: Prediction constrains; Experiment-backed have been constructed in the case of human Pgp1/MDR1/ prediction; Domain arrangement; Transmembrane proteins ABCB1 [5] and for the membrane-embedded core (6 + 6 trans- membrane helices) domain of human MRP1/ABCC1 [6]. These are again presented in other reviews of this special issue. In our group we are currently establishing a database col- lecting published information on the experimental data rele- 1. Introduction vant to the membrane topology of various families of membrane proteins (Tusna´dy and Simon, unpublished). Such At present there are no high-resolution structural data avail- data are sometimes generated with the aim of systematically able of any eukaryotic ABC transporter. Therefore, laborious establishing the membrane topology of a given membrane pro- biochemical experiments are necessary to elucidate their mem- tein, but the database also contains experimental findings brane topology, i.e., the position and orientation of membrane which per se hold information for the location of a given seg- spanning segments within the polypeptide chain. These meth- ment of the polypeptide chain relative to the membrane. These ods include epitope insertion, localization of glycosylation include, e.g., cysteins involved in disulfide-bridge formation in sites, limited proteolysis and immunochemical techniques. It native state, which are usually not localized in the cytosol, or is worth to note that expression of truncated or non-functional peptide motifs interacting with intracellular proteins, indicat- ABC proteins, or proteins expressed in heterologous expres- ing that this motif should be intracellular as well. We have sion system (e.g., mammalian ABC proteins in bacteria), as developed a membrane topology prediction approach which well as fusion proteins can yield protein with aberrant mem- is able to utilize the experimental data deposited in the above brane topology. database as constrains. A recent paper also argues for improv- It is generally accepted that the minimum structural require- ing topology predictions by using experimental data [7]. ment for a functional ABC transporter is the presence of two In the present review, we apply our topology prediction transmembrane domains (TMD) and two ATP-binding cas- strategy (based on incorporation of experimental data into sette (ABC or NBD) units. In humans these may be present the prediction algorithm) to provide information to the within one polypeptide chain (‘‘full transporters’’), or within most plausible membrane arrangements for the human ABC a membrane-bound homo- or heterodimer of ‘‘half-transport- transporters. ers’’. The membrane topology models define the domain arrangements for these variable ABC transporters. 2. Methodology We used the Human ABC Proteins Database (http://nutri- gene.4t.com/humanabc.htm) for the subfamily classification. *Corresponding author. Fax: +361 466 5465. We included only those members of a given subfamily into E-mail address: [email protected] (A. Va´radi). the analysis whose full sequence were know. We choose the 0014-5793/$32.00 Ó 2005 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2005.11.040 1018 G.E. Tusna´dy et al. / FEBS Letters 580 (2006) 1017–1022 sequence of the longest or the most common splicing variant if family (Fig. 1). If there was experimental data defining the alternatively spliced cDNAs were described. We omitted the topology of a segment, or the whole polypeptide of one mem- four ABC-proteins belonging to the ABCE and ABCF sub- ber of the subfamily, it was also used as constraint(s) in the families, as these proteins contain no transmembrane domains. prediction for each sequence in the subfamily. Information ex- Protein sequences with each subfamilies were aligned by tracted from the literature related to the position of a given the ClustalW server (http://www.ebi.ac.uk/clustalw) [8]. The segment is color coded as described in the legend of Fig. 1. ABCB half-transporters and full transporters were aligned separately. For computer-assisted prediction the HMMTOP (hidden 3. Results Markov model for topology prediction) transmembrane topol- ogy prediction server has been utilized [9,10] (http://www. 3.1. The ABCA-subfamily enzim.hu/hmmtop). This method is based on the principle that The Human ATP-Binding Cassette Transporters Database topology of transmembrane proteins is determined by the max- lists 12 members of the ABCA-subfamily, these are full trans- imal divergence of amino acid composition of local sequence porters with plasma membrane localization, and it is notewor- segments. A special feature of this method that it can handle thy that these proteins are the largest by the linear sequence experimentally observed topology information and provides within the entire human ABC-protein family. the most probable topology given these topology data as con- The membrane topologies of two members of the subfamily strains. As an ‘‘initial condition’’, the intracellular locations of were studied in detail. These are ABCA1, in which mutations the ABC-domains (defined by the Walker A and Walker B are responsible for the Tangier disease, and ABCA4, the ret- motifs in the linear sequence) were preset. ina-specific ABC transporter whose mutations cause several The results of the above analyses are presented as linear retinopathies. The two proteins share 50% identical amino block diagrams of the aligned sequences within a given sub- acids. ECD1 TMD1 NBD1 ECD2 TMD2 NBD2 ABCA A1 A2 A3 A4 A5 A6 A7 A8 A9 A11 A13 ∆(100-2600) TMD1 NBD1 TMD2 NBD2 TMD NBD ABCB B1 ABCB B2 (full) B4 (half) B3 B5 B6 B11 B7 B8 TMD0 TMD1 NBD1 TMD2 NBD2 B9 ABCC C1 B10 (long) C2 C3 TMD NBD C6 ABCD D1 C8 D2 C9 D3 C10 D4 TMD1 NBD1 TMD2 NBD2 NBD TMD ABCC C4 ABCG G1 (short) C5 G2 C7 G4 C11 G5 C12 G8 Fig. 1. Predicted topologies of human ABC proteins. Human ABC proteins were classified by using the class definition of Human ABC Protein Database. Protein sequences within each subfamily were aligned by ClustalW. Topology predictions for each sequence were made by HMMTOP using the constraints that ABC domain(s) is/are in the cytoplasmic site of the membrane (2500 amino acids from the first extracellular loop of ABCA13 has been deleted for better visualization). If experimental results about the topology of a segment or the entire polypeptide of one member of the subfamily were available, these were used as constraints in the prediction for each member in the subfamily. The list of publications with the original data is available on the TOPDB web site (http://topdb.enzim.hu/). Experimentally established sequence regions are drawn by darker thick line, experimentally established residues are marked by triangle above the sequence. Sequence regions predicted to be cytoplasmic or extracytosolic are colored by light red and light blue, respectively. Predicted transmembrane segments are colored by yellow. ABC domains are colored in middle tone of red. Domains of ABC proteins are shaded and indicated above the topology chart of each subfamily: TMD is transmembrane domain, ECD is extracellular domain, NBD is nucleotide binding domain, i.e., ABC domain. G.E. Tusna´dy et al. / FEBS Letters 580 (2006) 1017–1022 1019 First Illing et al. [11] suggested a model for ABCA4 with two ABCB10, as well as ABCB9, with a putative lysosomal locali- large extracellular domains (ECDs) between the first and sec- zation. The full transporters of the subfamily are localized in ond transmembrane helices of each predicted TMDs. An ele- the plasma membrane (in the apical membrane compartment), gant experimental proof of the Illing model has been while the half-transporters are found in the membranes of var- published [12]. Eight N-glycosylation sites were mapped by ious organelles. It is well established that ABCB2/TAP1 and mutagenesis within the bovine ABCA4 sequence, four within ABCB3/TAP2 form a heterodimer, which actively translocates the N-terminal half and four within the C-terminal half.