The Evolutionary Repertoires of the Eukaryotic-Type ABC Transporters in Terms of the Phylogeny of ATP-binding Domains in Eukaryotes and Prok
http://www.100md.com
分子生物学进展 2004年第11期
Bioinformatics Center, Institute for Chemical Research, Kyoto University Uji, Kyoto, Japan
E-mail: kanehisa@kuicr.kyoto-u.ac.jp.
Abstract
ABC (ATP-binding cassette) transporters play an important role in the communication of various substrates across cell membranes. They are ubiquitous in prokaryotes and eukaryotes, and eukaryotic types (EK-types) are distinguished from prokaryotic types (PK-types) in terms of their genes and domain organizations. The EK-types and PK-types mainly consist of exporters and importers, respectively. Prokaryotes have both the EK-types and the PK-types. The EK-types in prokaryotes are usually called "bacterial multidrug ABC transporters," but they are not well characterized in comparison with the multidrug ABC transporters in eukaryotes. Thus, an exhaustive search of the EK-types among diverse organisms and detailed sequence classification and analysis would elucidate the evolutionary history of EK-types. It would also help shed some light on the fundamental repertoires of the wide variety of substrates through which multidrug ABC transporters in eukaryotes communicate. In this work, we have identified the EK-type ABC transporters in 126 prokaryotes using the profiles of the ATP-binding domain (NBD) of the EK-type ABC transporters from 12 eukaryotes. As a result, 11 clusters were identified from 1,046 EK-types ABC transporters. In particular, two large novel clusters emerged, corresponding to the bacterial multidrug ABC transporters related to the ABCB and ABCC families in eukaryotes, respectively. In the genomic context, most of these genes are located alone or adjacent to genes from the same clusters. Additionally, to detect functional divergences in the NBDs, the Kullback-Leibler divergence was measured among these bacterial multidrug transporters. As a result, several putative functional regions were identified, some corresponding to the predicted secondary structures. We also analyzed a phylogeny of the EK-type ABC transporters in both prokaryotes and eukaryotes, which revealed that the EK-type ABC transporters in prokaryotes have certain repertoires corresponding to the conventional ABC protein groups in eukaryotes. On the basis of these findings, we propose an updated evolutionary hypothesis in which the EK-type ABC transporters in both eukaryotes and prokaryotes consisted of several kinds of ABC transporters in putative ancestor cells before the divergence of eukaryotic and prokaryotic cells.
Key Words: ABC transporter ? multidrug ? comparative genomics
Introduction
The ATP-binding cassette (ABC) transporter superfamily is one of the major classes of active transporters, which convey a wide variety of substrates across the cell membrane and are widespread in eubacteria, archaea, and eukaryotes. This superfamily has two kinds of proteins, the prokaryotic type (PK-type) and the eukaryotic type (EK-type). Most EK-type ABC transporters are known as multidrug transporters that export various compounds and drugs from cells. Thus, previous studies on the characterization of the multidrug ABC transporters were mainly focused on eukaryotic cells as opposed to prokaryotic cells because the application of multidrug transporters has great potential in the medical and pharmacological fields. Therefore, the three-dimensional structures of ABC transporters in bacteria are important targets for research, as has been determined recently (Chang and Roth 2001; Locher, Lee, and Rees 2002). However, there has been no comprehensive sequence analysis of EK-type ABC transporters in the context of bacterial multidrug transporters. Therefore, we conducted the identification, classification, and evolutionary analysis of the EK-type ABC transporters in prokaryotes. We also surveyed their substrates from published literature, which can help us understand the relationships between multidrug transporters and their substrates. Moreover, the application of this work to other multidrug transporter families and their substrates may help to increase the understanding of their wide range of specificities. Therefore, the systematic and detailed classification of EK-type ABC transporters in prokaryotes is important to glean more information about the evolutionary relationships of ABC transporters and the relationships between their genes and substrates.
The PK-type is distinguished from the EK-type by gene and domain organizations. The structure of a typical PK-type ABC transporter consists of three types of components: (1) two integral proteins each having six transmembrane segments, (2) two peripheral proteins that bind and hydrolyze ATP, and (3) a periplasmic substrate-binding protein. In many cases, the genes encoding these three types of components, especially the membrane protein and ATP-binding protein components, form an operon as observed in known archaea and eubacteria genomes (Tomii and Kanehisa 1998). On the other hand, in a typical EK-type ABC transporter, the membrane-spanning protein and the ATP-binding protein are fused, and the transporter consists of two polypeptide chains, each having a membrane-spanning domain (MSD) and an ATP-binding domain (NBD). The EK-type transporter has no substrate-binding component, but the MSD carries out substrate selectivity. In PK-type ABC transporters, the NBD component is the most conserved, the MSD component is somewhat less conserved, and the substrate-binding protein is the least conserved in terms of sequence similarity (Tam and Saier 1993; Saurin and Dassa 1994). In a previous study (Tomii and Kanehisa 1998), PK-type ABC transporters were classified into functional (ortholog) groups by a hierarchical cluster analysis using the sequence similarities of NBDs. In EK-type ABC transporters, the NBD is the most-conserved domain, and it is expected that the sequence similarity of NBDs can also be used for functional classification.
In general, eukaryotes only have EK-type ABC transporters, but prokaryotes not only have PK-type ABC transporters but also have certain kinds of EK-type ABC transporters. Most of these are known as "bacterial multidrug transporters." Their sequence repertoires and substrates are not well characterized in comparison with the ABC multidrug transporters in eukaryotes, although the number of completely sequenced prokaryotic genomes is expanding. Thus, the study of EK-type ABC transporters in prokaryotes should shed some light on the fundamental substrate repertoires of eukaryotes.
The conventional ABC transporter families consist of the ABCA, ABCB, ABCC, ABCD, and ABCG families. Most ABC transporter families in human are involved in genetic diseases. The ABCA family is not well characterized in terms of its function and substrates. However, one member of the ABCA family, ced-7 (cel:C48B4.4 in the KEGG database) is known to transport molecules that are involved in programmed cell death. It has been suggested that the lysine residue of the Walker A motif interacts with the phosphate group of the bound ATP (Saraste, Sibbald, and Wittinghofer 1990), and the point mutation experiment on the lysine residue revealed that the N-terminal NBD (N-NBD) of ced-7 is functionally more important than the C-terminal NBD (C-NBD)in avoiding cell death. (Wu and Horvitz 1998).
The ABCB family contains three subfamilies that are known as pgp (p-glycoprotein), atm, and tap (transporter associated with antigen processing). The pgp family has been well studied in mammals such as human and Chinese hamster. Various compounds have been identified as substrates such as peptides, vinca alkaloids, anthracyclines, and epipodophyllotoxins, among others (Sharom 1997). The atm family is found on mitochondria, and tap family is found on the endoplasmic reticulum (ER). One of the atm members, atm1 (sce:YMR301C), is involved in transporting iron-sulfur cytosolic protein. The tap family is known to transport antigen peptides to the ER to form the MHC class I molecules in the immune system.
The ABCC family is also known as mrp (multidrug resistance–associated protein), which has similar domain organizations as the ABCB family. Some of the mrp members (ex. ABCC1, ABCC2) have five additional transmembrane segments in their N-terminals, but the function of these additional MSDs is not clear. The mrp was found in the multidrug-resisting lung cancer cell in which pgp was not overexpressed (Cole et al. 1992). Mrp exports glutathione by drug stimulation (Cole and Deeley 1996) as well as certain substrates in conjugated forms with anions such as glutathione glucuronate and sulfate (Jedlitschky et al. 1996).
The ABCD family is also known as pmp (peroxisomal membrane protein). The proteins of this family, pxa1p and pxa2p found in sce, are implicated in the importation of long-chain acyl-CoA for shortening fatty acid by beta-oxidation (Verleur et al. 1997). The ABCG and its related family consist of white and pdr (pleiotropic drug resistance) families. The white protein (dme:CG2759-PA) is a half-type ABC transporter that imports guanine with brown protein (dme:CG17632-PA) and tryptophan with scarlet protein (dme:CG4314-PA) (Ewart and Howells 1998). The pdr proteins are observed in fungi and plants and, in sce, are known to exhibit distinct but a wide overlapping range of drug resistance profiles for fungicides, flavonoids, and other substances (Kolaczkowski et al. 1998).
The ABCE and ABCF families are not ABC transporters but have the ATP-binding cassette domain in their sequences. They correspond to ribonuclease inhibitors and translation initiation factors. Because of the limited scope of our analysis, these proteins were not included in our work. The ABCH family, which consists of three sequences in dme, was recently defined by Dean, Rzhetsky, and Allikmets (2001), who found that this family is contained in the ABCA family.
Our work presents a global and comprehensive classification of the EK-type ABC transporters in more than a hundred genomes. First, our classification of the identified EK-type ABC transporters in prokaryotes reveals certain relationships among categorized transporter clusters. Second, we described the novel two clusters in prokaryotes, which are related to the ABCB and ABCC eukaryotes and were found the majority of prokaryotes. In addition to this, the measurement of the Kullback-Leibler (KL) divergence of major clusters of bacterial ABC transporters revealed several regions that are suggestive of functional importance, some of which correspond to the predicted secondary structure. Third, based on phylogenetic analysis, we propose an updated evolutionary scenario of the repertoire of the EK-type ABC transporters in putative ancestor cells.
Materials and Methods
Data Source
We collected the amino acid sequences of protein-encoding genes of 12 eukaryotes, 111 eubacteria, and 15 archaea from the KEGG/GENES nucleotide/protein sequence database (July 17, 2003) (Kanehisa et al. 2002). All the organisms and their abbreviations can be referenced from KEGG/GENES at http://www.genome.ad.jp/kegg/kegg2.html, and those of the 12 eukaryotes and 126 prokaryotes used in our study are listed in the Appendix.
Identification of EK-type ABC Transporters
Our workflow is illustrated in figure 1. The identification of EK-type ABC transporters was performed using 12 eukaryote genomes from the KEGG/GENES database. We first identified the putative ABC transporters that have the ABC_tran protein motif in Pfam_ls (global alignment model) of the Pfam database release 9.0 (Bateman et al. 2002) using an expectation value of 0.001. In this work, we defined this ABC_tran protein motif region as a NBD. This region is about 200 residues in length and contain Walker A (P-loop), Walker B, and ABC signature (Walker C) motifs. Next, we removed the sequences that have no membrane-spanning segments using TM-HMM2 (http://www.cbs.dtu.dk/services/TMHMM) (Krogh et al. 2001). TM-HMM2 predicts the number of membrane-spanning segments, given some query sequences. After the putative ABC proteins were obtained, we annotated the sequences that contained the extracted NBDs from the Pfam database and generated a UPGMA (unweighted pair group method with arithmetic mean) (Sokal and Michener 1958) dendrogram based on their NBDs. We manually removed nontransporter proteins with the help of the UPGMA dendrogram and the Pfam domain annotation. After this process, 474 sequences remained as EK-type ABC transporters in eukaryotes. These 474 sequences were classified by the complete linkage hierarchical clustering method using their NBDs.
FIG. 1.— The workflow used in identifying and classifying EK-type ABC transporters.
Classification of EK-type ABC Transporters
Each NBD of every ABC transporter protein was aligned with all other extracted NBDs using the SSEARCH version 3.4t20 program (Pearson 1996), which is an implementation of the Smith-Waterman algorithm (Smith and Waterman 1981) with the BLOSUM 50 matrix and gap penalties of –12 and –2 for opening and extension, respectively. We defined the distance of two sequences based on their NBDs as the inverse of the base 2 logarithm of the Smith-Waterman alignment score (SW score). Then, we performed a complete linkage clustering of these sequences based on their NBDs. The complete linkage method is implemented in the R version 1.7.0 statistics programs (http://www.r-project.org). We obtained the conventional human ABC transporter classification groups using a threshold of 5.8 (= 55.7 SW score) as the similarity criterion.
Identification of EK-type ABC Transporters in Prokaryotes
The classified NBDs in the complete linkage clustering were aligned using Clustal W version 1.82 (Higgins, Bleasby, and Fuchs 1992; Chenna et al. 2003) with default parameters. The individual HMMs were constructed by the hmmbuild and hmmcalibrate programs in HMMER version 2.3.1 (Eddy 1998). These constructed HMMs and the ABC_tran protein motif model were used for detecting the EK-type NBDs in the prokaryotic genomes in the KEGG/GENES database. We selected those NBDs that had an e-value under 1.0e-10 because it was a relaxed enough threshold criteria to detect all of the EK-type ABC transporters in prokaryotes. This rather strict criteria was used because the precollected protein set before the TM-HMM2 step contains many PK-type ABC transporters that do not have MSDs. The detected sequences were filtered by TM-HMM2 to obtain proteins that have at least three predicted membrane-spanning segments, resulting in 2,896 entries for the candidates of the EK-type ABC transporters, with most of the PK-type ABC transporters being removed. Then, we drew a UPGMA dendrogram and manually removed candidates that were related to flagellar, symporter, antiporter, kinase, and other functions using their annotations and HMM scores. Finally, 1,046 ABC transporters from prokaryote genomes were obtained.
Phylogenetic Analysis of EK-type ABC Transporters
Our final set of EK-type ABC transporters contained 474 sequences in eukaryotes and 1,046 sequences in prokaryotes. The Ward clustering dendrogram was obtained using R and the inverse value of the Smith-Waterman alignment scores as described above. The KL divergence was calculated using the Web site (http://www.cbs.dtu.dk/gorodkin/appl/plogo.html) (Gorodkin et al. 1997) with the amino acid distribution of our data set. The neighbor-joining (NJ) tree (Saitou and Nei 1987) was generated from ClustalW, NJ plot and in-house scripts. The branches of the NJ tree were merged when they belonged to the same cluster in the Ward cluster analysis. We used the average length of two branches to create a new merged branch. The gene list of the Ward cluster dendrogram, the constructed HMMs, and the phylogenetic tree are available at http://web.kuicr.kyoto-u.ac.jp/igarashi/abc/.
Result
EK-type ABC Transporters in Eukaryotes
Initially, to increase sensitivity for searching the EK-types in prokaryotes, we defined a classification for the NBDs of eukaryotes in our data set according to the hierarchical classification of EK-type ABC transporters in eukaryotes, as shown in figure S1 in the Supplementary Material online. In our classification, we found eight clusters that corresponded to the conventional ABC categorizations (see Introduction) according to the complete linkage clustering using the threshold of 5.8 for the logarithm of the SW score. Note that domain organizations are known to be either full-type (MSD-NBD-MSD-NBD or NBD-MSD-NBD-MSD) or half-type (MSD-NBD or NBD-MSD), and we were able to distinguish between these types in our clusters. For each full-type, a half-type similar to either the N-NBD or the C-NBD of the full-type exists in the same cluster.
We noticed that one cluster, however, which corresponded to a subgroup of the ABCA family, did not cluster together with the rest of its family. The N-NBDs of this group are similar to the NBDs of the ABCB family. A closer look revealed that it actually belongs to the ABCA6 subfamily, which contains nine sequences found only in multicellular eukaryotes (hsa:10351, hsa:23461, hsa:23461_1, hsa:10349, mmu:217265, mmu:76184, mmu:217258, mmu27404, and mmu:217262). However, this cluster may have clustered separately because this classification method focuses only on one NBD in each cluster in each integrating step.
To confirm the numbers of the identified EK-types with the previous studies, we list the numbers of EK-types in eukaryotes in table 1. These numbers are slightly different from others previously reported (Dean, Rzhetsky, and Allikmets 2001; Sanchez-Fernandez et al. 2001), because the genomes of hsa, dme, and other organisms are continually being updated, and we incorporated alternative splicing genes into our data set. For the majority, the numbers of ABC transporters in sce coincided with the report by (Decottignies and Goffeau 1997).
Table 1 The Number of EK-type ABC Transporters in Eukaryotes
Hierarchical Clustering of EK-type ABC Transporters
We detected all of the repertoires of EK-type ABC transporters in prokaryotes and eukaryotes and explored the sequences, domains, functional features, and evolutionary relationships among EK-type ABC transporters using the Ward and NJ phylogenetic method. We adopted the Ward hierarchical cluster analysis for classification and the NJ method for exploring the evolutionary relationships between eukaryotes and prokaryotes.
Based on the Ward cluster dendrogram (fig. 2), we obtained 11 clusters that mainly consisted of (1) prokaryotic macrolide-specific transporters, (2) ABC-2 prokaryotes, (3) ABCDs, fused PK-type ABC transporters, yojI, and miscs, (4) ABCGs, (5) ABCG (N-NBD) eukaryotes, (6) ABCA eukaryotes, (7) ABCB eukaryotes, (8) ABCB eukaryotes and similar prokaryotes, (9) ABCC (C-NBD) eukaryotes, (10) ABCC (N-NBD) eukaryotes, and (11) ABCC eukaryotes and similar prokaryotes. We also surveyed the distribution of the NBDs in individual taxa (table 2) and examined each cluster's sequence, domain feature, and function. The number of NBDs in individual organisms is shown table S1 in the Supplementary Material online.
FIG. 2.— The Ward cluster dendrogram of the EK-type ABC transporters in eukaryotes and prokaryotes. Black lines on the left side show relationships of identical sequences that have multiple NBDs. Orange lines indicate the relations of adjacent genes in prokaryotic genomes. The leaves of this clustering correspond to the sequences and are colored red (NBDs used in the Ward clustering analysis), blue (one of multiple NBDs in full-types), purple (FtsX, FHA, peptidase_C38, peptidase_C39, or cNMP-binding motifs), green (ABC_membrane motif defined in Pfam), yellow (nonassigned region), and black (other domains). Blue boxes delineate the individual clusters.
Table 2 The Observation of NBDs of the EK-type ABC Transporters
EK-type ABC Transporters in Prokaryotes
Cluster 1
Cluster 1 consists of ybjZ and its orthologs, so they can be said to represent the family of macrolide-specific exporters because it has been demonstrated that ybjZ (eco:b0879) specifically exports macrolide (Kobayashi, Nishino, and Yamaguchi 2001). We have found that the ybjZ genes have transmembrane segments in their C-terminal regions as annotated (FtsX) by Pfam.
Cluster 2
Cluster 2 consists of ABC-2–type transporters that export oligosaccharides (Reizer, Reizer, and Saier 1992). They have two NBDs in their N-terminal regions. Interestingly, the taxa distribution in the macrolide-specific (cluster 1) and ABC-2–type (cluster 2) clusters are similar, despite the fact that they do not have any substrate correlations or NBDs similarities; they mostly consist of proteobacteria, plantomyces, and green sulfur bacteria.
Cluster 3
Cluster 3 consists of several kinds of proteins, such as members of the ABCD eukaryote family, uncharacterized type (yojI), fused PK-type, and others. In particular, yojI orthologs were found only in proteobacteria (eco, ecj ecc, sty, stt, stm sfl, and sfx), and they have a similar domain organization as the half-types of ABCB and ABCC, but their functions are unknown. However, we note that they have especially close similarity (SW score > 515) to particular cluster 1 members such as pvdE (pae:PA2397) and syrD (rba:RB9612), which are involved in pyoverdine synthesis (McMorran et al. 1996) and transporting syringomycin (Quigley, Mo, and Gross 1993), respectively. Pyoverdine is involved in iron translocations, and syringomycin is known to transport cations. In addition, their distribution among taxonomy does not overlap. The orthologs of pvdE are observed among the proteobacteria gamma subdivision and cyanobacteria (pst:PSPTO2153, pae:PA2397, ppu:PP4216, and tel:tlr1649). The orthologs of syrD are observed among the proteobacteria beta and delta/epsilon subdivisions and plantomyces (rso:RS03729, cje:Cj1587c, and rba:RB9612). Thus, on the basis of homolog substrates and their ortholog distribution, it is suggested that yojI and its orthologs may transport cations with chelating molecules. As for the fused PK-type in this cluster, we observed cysA (rba:RB3287), which transports sulfate (Wooff et al. 2002), and ftsE (dra:DR1550) which is involved in cell division and salt transport (de Leeuw et al. 1999). These fused PK-type ABC transporters are formed from PK-types that are frequently observed in prokaryotic genomes. In that respect, yojI, pvdE, and syrD are different from the usual fused PK-types because they do not have PK-type homologs.
Clusters 4 and 5
Cluster 4 consists of genes whose NBDs are similar to the ABCG family. These half-type ABC transporters contain white (dme:CG2759-PA) and scarlet (dme:CG4314-PA) genes, which transport amino acids (Ewart and Howells 1998). The members of cluster 5 are found only in plants, but the ones from cluster 4 are present in actinobacteria, cyanobacteria, and eukaryotes (ana:all7196, mbo:Mb1776, mtc:MT1789, mtu:Rv1747, sco:SCO1806, sco:SCO5818, sma:SAV1159, sma:SAV2446, sma:SAV6472, syn:sll0778, tel:tll1329, tel:tll1340, and tel:tlr1253).
Cluster 6 to 10
Clusters 6, 7, 9, and 10 consist of eukaryotic genes such as ABCA, ABCB, N-NBD, and C-NBD in ABCC, which were described in the previous section. We note that clusters 2 (ABC-2) and 6 (ABCA) are close to each other in the clustering, although ABCA is only identified in eukaryotes and ABC-2 is only identified in prokaryotes.
Clusters 8 and 11: Two Novel Clusters of Bacterial Multidrug ABC Transporters
Until now, bacterial multidrug transporters have not been classified distinctly, but we were able to obtain two novel clusters based on their NBD similarities, clusters 8 and 11 in our dendrogram. Cluster 8 consists of 497 NBDs in prokaryotes and 25 NBDs in eukaryotes. Cluster 11 consists of 329 NBDs in prokaryotes and 12 NBDs in eukaryotes, the latter of which are organelle transporters such as mitochondria and ER. These clusters are related to ABCB and ABCC of eukaryotes, respectively. Particularly, most of the bacteria in this analysis have genes from both clusters 8 and 11 in their genomes. However, several taxa do not have cluster 8 or 11 NBDs. All neisseria, spirochete, and bacteroid and part of mycoplasma, mycobacterium, and archaea lack cluster 11 NBDs (table 2). On the other hand, all of the delta/epsilon subdivision of proteobacteria, chlamydia, crenarchaeota, and part of mycoplasma, actinobacteria, spirochete, hyperthermophilic bacteria, and euryarchaeota lack cluster 8 NBDs. It is of little probability that only the delta/epsilon subdivision of proteobacteria never contained cluster 8 NBDs, because all the other proteobacteria have cluster 8 NBDs. Thus, we propose that they had lost these cluster 8 NBDs after divergence into the subdivisions. All chlamydia organisms have only one NBD of cluster 11, and these NBDs have high similarity with each other (SW score > 936). Overall, in specific taxa (i.e., proteobacteria gamma, lactobacillales, mollicutes, actinobacteria, and euryarchaeota) some cluster 11 genes are located adjacent to cluster 8 genes, but most genes of cluster 8 and 11 are located alone or adjacent to genes of the same cluster.
To detect the residue variation and the degree of conservation between clusters 8 and 11, we measured their KL divergence (Kullback and Leibler 1951), also known as relative entropy, in their multiple alignment (fig. 3). This multiple alignment is 319 residues in length because of many gaps observed in the SDR (structurally diverse region) and other regions. The SDR was considered as a control region of the NBD to target its cognate MSD (Schmitt et al. 2003). Figure 3 shows that the sequences of cluster 8 conserved better than those of cluster 11, both overall and among individual motif regions. In particular, the degree of conservation tends to increase in the vicinity of the ABC signature and Walker B motif. Moreover, several well-conserved regions were found that do not correspond to the motifs. In cluster 8, we observed several conserved residues "FYD" and "GxDI" between Walker A and Q-loop. Because these conserved residues do not correspond to the predicted secondary structure, we assume that they play a functional role as opposed to a structural role. The distinctive regions from the 156th to the 159th (ATxEE) amino acid residues in SDR may be involved in the recognition of each MSD. The well-conserved residues in the region from the 185th to 198th are hydrophobic amino acids, corresponding to the predicted helix region. The threonine at the 303rd residue is also conserved well and corresponds to the predicted helix region, but its contribution is as of yet unclear. ABC transporters work by ATP hydrolysis energy that is generated at two NBDs. However, both do not always take same function as shown in ced-7, which was noted in the Introduction. Therefore, most clusters 8 and 11 also have such difference in their functions, which may be concerned with the activity of ATP hydrolysis.
FIG. 3.— The subtraction of the Kullback-Leibler (KL) distance of cluster 11 from cluster 8 is shown. The residues that have large difference in KL distances between cluster 8 and 11 are shown with their loci and frequently observed amino acids in parenthesis. Several motifs and predicted secondary structures are also shown. The secondary structure prediction was performed by Jpred using the NBD of bha:BH0941 in cluster 8. The dotted line of the predicted secondary structure indicates the gaps of the multiple alignment.
We surveyed the known substrates of the sequences in clusters 8 and 11 from the literature in an attempt to find some pattern among them (table 3). As a result, we found that both clusters included genes that transport peptides, low-molecular-weight sugars, iron compounds, and hydrophobic drugs. These compounds are also transported by ABC transporters in eukaryotes. Thus, the spectrum of substrates in eukaryotic ABC transporters is similar to the EK-type ABC transporters in prokaryotes, but it seems that there is no significant difference between the spectrum of substrates of clusters 8 and 11 at this time, and further research is required. A systematic compound search against each transporter, and more three-dimensional crystal structure data would aid further work in elucidating the difference in function of these clusters.
Table 3 Functional Description of Representative Genes in Clusters 8 and 11
Phylogenetic Analysis of EK-type ABC Transporters
An unrooted NJ phylogenetic tree of our data set is presented in figure 4, and in this figure, clusters 8 and 11, both of whose NBDs are of bacterial multidrug transporters, fell on different branches. This supports the possibility that clusters 8 and 11 have evolved separately. In addition to this, cluster 7 (ABCB eukaryotes) and the clusters 9 and 10 (ABCC eukaryotes) are close to cluster 11 and cluster 8, respectively. This indicates that the eukaryote multidrug transporters ABCB and ABCC may have originated from bacterial multidrug transporters. Our tree also indicates that the macrolide-specifc exporters of cluster 1 seem to have a different evolutionary origin than bacterial multidrug transporters because of its far distance. We also note that clusters 2 (ABCA) and 6 (ABC-2) are close in our tree, thus, implying that they are near in evolution in terms of their NBDs.
FIG. 4.— The unrooted NJ phylogenetic tree of NBDs of EK-type ABC transporters in eukaryotes and prokaryotes. Red branches indicate cluster 11 (ABCC prokaryote). Orange branches indicate cluster 7 (prokaryotes ABCB). The green branches indicate cluster 4 (ABCG). Yellow branches indicate cluster 3 (ABCD, fused PK-type, and miscs). Black branches indicate other clusters (10,9: ABCC [C-NBD, N-NBD], 1: macrolide-specific transporters, 2: ABC-2, 5: ABCG (N-NBD), and 6: ABCA). The number of merged branches are given in angle brackets. The thickness of the lines corresponding to the number of merged branches. The thinnest lines indicate fewer than 10 merged branches, the medium lines correspond to 10 to 29 merged branches, and the thickest lines correspond to 30 or more merged branches. We emphasize mainly those branches that have over 30 branches merged.
Discussion
In this work, we used two hierarchical clustering methods. The Ward method is a classic classification method that attempts to minimize the loss associated with each grouping. At each step in the analysis, the union of every possible cluster pair is considered and the two clusters whose fusion results in the minimum increase in information loss are combined. Ward defines information loss in terms of an error sum-of-squares criterion. Our second clustering method, the complete linkage hierarchical clustering method, provides results using absolute as opposed to relative values. This allows us to distinguish clearly similar clusters based on such metrics as the SW-score.
Before the identification of EK-types in prokaryotes, we identified and classified the EK-type ABC transporters in eukaryotes. The classification results agreed with the conventional human ABC transporter classification. The similarities between N-NBD and C-NBD in the same family are higher than the similarities between N-NBD and C-NBD of other families. It is suggested that there is an interaction between the N-terminal side and C-terminal side in an ABC transporter, which does not allow drastic mutations in them.
In the analysis of EK-type ABC transporters in eukaryotes and prokaryotes, we could obtain clear clusters that were not obtained by other classification methods such as complete-linkage clustering, single-linkage clustering, and UPGMA. We demonstrated that several certain repertoires of sequence and domain organization were observed in the EK-type ABC transporters in prokaryotes, including the bacterial multidrug ABC transporters. The identified EK-type NBDs in prokaryotes corresponds to the NBDs of the ABCB, ABCC, and ABCG families of eukaryotic ABC transporters. Furthermore, in the bacterial multidrug ABC transporters, there are two novel and large groups, clusters 8 and 11. These two clusters are difficult to detect by simple sequence similarity search. Additionally, we note that cluster 11 sequences are related to ABCC eukaryotes. However, they are not accompanied by the additional MSDs that are well conserved in the mammalian ABCC family known as mrp. Thus, we suggest that cluster 11 members have a decreased possibility of transporting the substrates of mrp.
In terms of the genomic context, we found that there are adjacent half-type transporters, which result in transporters that function as full-types. We claim that these adjacent half-type transporters are the pre-evolution versions of the full-types. In addition, in clusters 8 and 11, some half-type transporters were found that have additional domains in their N-terminals. These additional domains were identified as peptidase_C39, peptidase_S8, and cNMP-binding domains in Pfam. These domains imply that these ABC transporters are dual-function proteins. The peptidase_C39 domains were observed in prokaryotes in this work and function to cleave off signal peptides during maturation (Havarstein, Diep, and Nes 1995). The peptidase_S8 domains were observed only in ddi and function as serine proteases. The cNMP-binding domain may play a role in signal transductions. Therefore, it is likely that such domains are fused into the N-terminals of the half-type transporters so that they were able to obtain another function. We propose that these phenomena are the results of the efficiency of expression and function through gene evolution.
In this section, we assess the evolutionary relationships of each family for constructing an updating evolutionary hypothesis of ABC transporters. An evolutionary hypothesis of the exporter and importers of ABC transporters was proposed by (Saurin, Hofnung, and Dassa 1999). They postulated that the importer originated from prokaryotes and that the exporter originated from both eukaryotes and prokaryotes. In addition to this, they argued that it would be improbable to derive the eukaryotic exporters from prokaryotic exporters in the context of their phylogenetic analysis. They also proposed that putative ancestor cells that had originated from eukaryotic and prokaryotic cells already contained the specific divergent exporters. On the basis of this hypothesis and the result of our classification and phylogenetic analysis, we constructed an evolutionary hypothesis of the EK-type ABC transporters, as illustrated in figure 5.
FIG. 5.— Our proposed evolutionary scenario of the NBDs of the EK-type ABC transporter. We consider only NBDs in this scheme. The dotted line indicates a horizontal transfer.
We assume that several NBD repertoires of the ABC transporters that currently exist would have also been present in putative ancestor cells. Our classification and phylogenetic analysis (clusters 8 and 11) supports the fact that the two major bacterial multidrug transporters are evolutionarily related to the ABCB and ABCC families of eukaryotes. Furthermore, because the ABCA and ABC-2 families are also close to each other in our phylogenetic tree, we claim that they may have shared same NBDs in their origins and changed their domain organizations after putative ancestor cells diverged into eukaryotes and prokaryotes. However, further work is necessary to ascertain the origins and evolutionary relationships between them.
A part of the ABCG proteins of prokaryotes have the FHA (forked head–associated) domain, which is mainly observed in eukaryotic genomes and is involved in binding phosphorylated peptides. In our analysis, the ABCG family was also mainly observed in eukaryotes. Therefore, it is probable that the ABCG family originated in eukaryotes and propagated through horizontal gene transfer to cyanobacteria and actinobacteria, among others. Moreover, the macrolide-specific transporters and fused PK-type transporters may have appeared in prokaryotes after the divergence of eukaryotes and prokaryotes because they are found only in prokaryotes. As for the ABCD family, we found that it is localized at an ancient cellular component peroxisome that is observed in virtually all eukaryotes. Thus, it is likely to have existed since the very early stages of eukaryotic cells. Therefore, we propose that the set of NBDs of ABC transporters in putative ancestor cells are similar to ABCB, ABCC, ABCA, and ABC-2.
In addition, we observed several fused PK-type ABC transporters, but prokaryotes typically have nonfused PK-type ABC transporters as importers. The nonfused PK-types consist of operons, and their domain (component) organizations are not conserved as well as in the EK-types. In general, it is expected that genes that interact with each other tend to consist of operons, thus increasing the probability that they are fused through evolution to function more efficiently (Marcotte et al. 1999). Therefore, it is speculated that the PK-type genes have not yet undergone as much evolutionary time as have the EK-type genes.
Conclusion
In this work, we identified EK-type ABC transporters in approximately a hundred organisms and analyzed their repertoires, distributions, and relationships. In particular, we identified two novel large clusters in bacterial multidrug transporters and elucidated their differential regions, which are proposed to be important regions for their functions. We also developed a methodology using an HMM for each family that proved very useful for identifying families in EK-type ABC transporters in newly sequenced genomes. This methodology gave better results than the general motif model database Pfam. The proposed hypothetical evolutionary scenario is not a conclusive account, but it is of much use for further functional analysis of EK-type ABC transporters.
Appendix
The abbreviations of organisms are as follows:
From eukaryotes, Homo sapiens (hsa), Mus musculus (mmu), Rattus norvegicus (rno), Drosophila melanogaster (dme), Caenorhabditis elegans (cel), Arabidopsis thaliana (ath), Oryza sativa (osa), Plasmodium falciparum (pfa), Dictyostelium discoideum (ddi), Saccharomyces cerevisiae (sce), Schizosaccharomyces pombe (spo), and Encephalitozoon cuniculi (ecu)
From bacteria, Escherichia coli K-12 MG1655 (eco), Escherichia coli K-12 W3110 (ecj), Escherichia coli O157 EDL933 (ece), Escherichia coli O157 Sakai (ecs), Escherichia coli CFT073 (ecc), Salmonella typhi CT18 (sty), Salmonella typhi Ty2 (stt), Salmonella typhimurium (stm), Yersinia pestis CO92 (ype), Yersinia pestis KIM (ypk), Shigella flexneri 301 (sfl), Shigella flexneri 2457T (sfx), Buchnera sp. APS (buc), Buchnera aphidicola Sg (bas), Buchnera aphidicola Bp (bab), Wigglesworthia brevipalpis (wbr), Haemophilus influenzae (hin), Pasteurella multocida (pmu), Xylella fastidiosa 9a5c (xfa), Xylella fastidiosa Temecula1 (xft), Xanthomonas campestris (xcc), Xanthomonas axonopodis (xac), Vibrio cholerae (vch), Vibrio vulnificus CMCP6 (vvu), Vibrio parahaemolyticus (vpa), Pseudomonas aeruginosa (pae), Pseudomonas putida (ppu), Pseudomonas syringae pv. tomato (pst), Shewanella oneidensis (son), Coxiella burnetii (cbu), Neisseria meningitidis MC58 (nme), Neisseria meningitidis Z2491 (nma), Ralstonia solanacearum (rso), Nitrosomonas europaea (neu), Helicobacter pylori 26695 (hpy), Helicobacter pylori J99 (hpj), Helicobacter hepaticus (hhe), Campylobacter jejuni (cje), Rickettsia prowazekii (rpr), Rickettsia conorii (rco), Mesorhizobium loti (mlo), Sinorhizobium meliloti (sme), Agrobacterium tumefaciens C58 UWash (atu), Agrobacterium tumefaciens C58 Cereon (atc), Brucella melitensis (bme), Brucella suis (bms), Bradyrhizobium japonicum (bja), Caulobacter crescentus (ccr), Bacillus subtilis (bsu), Bacillus halodurans (bha), Bacillus anthracis (ban), Bacillus cereus (bce), Oceanobacillus iheyensis (oih), Staphylococcus aureus N315 (sau), Staphylococcus aureus Mu50 (sav), Staphylococcus aureus MW2 (sam), Staphylococcus epidermidis (sep), Listeria monocytogenes (lmo), Listeria innocua (lin), Lactococcus lactis (lla), Streptococcus pyogenes SF370 (spy), Streptococcus pyogenes MGAS8232 (spm), Streptococcus pyogenes MGAS315 (spg), Streptococcus pyogenes SSI-1 (sps), Streptococcus pneumoniae TIGR4 (spn), Streptococcus pneumoniae R6 (spr), Streptococcus agalactiae 2603 (sag), Streptococcus agalactiae NEM316 (san), Streptococcus mutans (smu), Lactobacillus plantarum (lpl), Enterococcus faecalis (efa), Clostridium acetobutylicum (cac), Clostridium perfringens (cpe), Clostridium tetani (ctc), Thermoanaerobacter tengcongensis (tte), Mycoplasma genitalium (mge), Mycoplasma pneumoniae (mpn), Mycoplasma pulmonis (mpu), Mycoplasma penetrans (mpe), Mycoplasma gallisepticum (mga), Ureaplasma urealyticum (uur), Mycobacterium tuberculosis H37Rv (mtu), Mycobacterium tuberculosis CDC1551 (mtc), Mycobacterium bovis (mbo), Mycobacterium leprae (mle), Corynebacterium glutamicum (cgl), Corynebacterium efficiens (cef), Streptomyces coelicolor (sco), Streptomyces avermitilis (sma), Bifidobacterium longum (blo), Tropheryma whipplei Twist (twh), Tropheryma whipplei TW08/27 (tws), Fusobacterium nucleatum (fnu), Rhodopirellula baltica (rba), Chlamydia trachomatis (ctr), Chlamydia muridarum (cmu), Chlamydophila pneumoniae CWL029 (cpn), Chlamydophila pneumoniae AR39 (cpa), Chlamydophila pneumoniae J138 (cpj), Chlamydophila caviae (cca), Borrelia burgdorferi (bbu), Treponema pallidum (tpa), Leptospira interrogans (lil), Bacteroides thetaiotaomicron (bth), Synechocystis sp. PCC6803 (syn), Thermosynechococcus elongatus (tel), Anabaena sp. PCC7120 (ana), Chlorobium tepidum (cte), Deinococcus radiodurans (dra), and Aquifex aeolicus (aae), Thermotoga maritima (tma).
From archaea, Methanococcus jannaschii (mja), Methanosarcina acetivorans (mac), Methanobacterium thermoautotrophicum (mth), Methanopyrus kandleri (mka), Archaeoglobus fulgidus (afu), Halobacterium sp. (hal), Thermoplasma acidophilum (tac), Thermoplasma volcanium (tvo), Pyrococcus horikoshii (pho), Pyrococcus abyssi (pab), Pyrococcus furiosus (pfu), Aeropyrum pernix (ape), Sulfolobus solfataricus (sso), Sulfolobus tokodaii (sto), and Pyrobaculum aerophilum (pai).
Acknowledgements
We thank Jean-Phillipe Vert, Ruy Jauregui, and Susumu Goto for critical reading of the manuscript and helpful comments. We also thank Hiroyuki Toh for helpful discussions and comments for the KL-divergence analysis. This work was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology of Japan, the Japan Society for the Promotion of Science, and the Japan Science and Technology Corporation.
References
Ahn, J. H., J. G. Pan, and J. S. Rhee. 1999. Identification of the tliDEF ABC transporter specific for lipase in Pseudomonas fluorescens SIK W1. J Bacteriol. 181:1847–1852.
Bateman, A., E. Birney, L. Cerruti, R. Durbin, L. Etwiller, S. R. Eddy, S. Griffiths-Jones, K. L. Howe, M. Marshall, and E. L. Sonnhammer. 2002. The Pfam protein families database. Nucleic Acids Res. 30:276–280.
Becker, A., H. Kuster, K. Niehaus, and A. Puhler. 1995. Extension of the Rhizobium meliloti succinoglycan biosynthesis gene cluster: identification of the exsA gene encoding an ABC transporter protein, and the exsB gene which probably codes for a regulator of succinoglycan biosynthesis. Mol. Gen. Genet. 249:487–497.
Benabdelhak, H., S. Kiontke, C. Horn, R. Ernst, M. A. Blight, I. B. Holland, and L. Schmitt. 2003. A specific interaction between the NBD of the ABC-transporter HlyB and a C-terminal fragment of its transport substrate haemolysin A. J. Mol. Biol. 327:1169–1179.
Chang, G., and C. B. Roth. 2001. Structure of MsbA from E. coli: a homolog of the multidrug resistance ATP binding cassette (ABC) transporters. Science 293:1793–1800.
Chenna, R., H. Sugawara, T. Koike, R. Lopez, T. J. Gibson, D. G. Higgins, and J. D. Thompson. 2003. Multiple sequence alignment with the Clustal series of programs. Nucleic Acids Res. 31:3497–3500.
Christensen, P. U., J. Davey, and O. Nielsen. 1997. The Schizosaccharomyces pombe mam1 gene encodes an ABC transporter mediating secretion of M-factor. Mol. Gen. Genet. 255:226–236.
Cole, S. P., G. Bhardwaj, J. H. Gerlach, J. E. Mackie, C. E. Grant, K. C. Almquist, A. J. Stewart, E. U. Kurz, A. M. Duncan, and R. G. Deeley. 1992. Overexpression of a transporter gene in a multidrug-resistant human lung cancer cell line. Science 258:1650–1654.
Cole, S. P., and R. G. Deeley. 1996. Multidrug resistance associated with overexpression of MRP. Cancer Treat Res. 87:39–62.
de Leeuw, E., B. Graham, G. J. Phillips, C. M. ten Hagen-Jongman, B. Oudega, and J. Luirink. 1999. Molecular characterization of Escherichia coli FtsE and FtsX. Mol. Microbiol. 31:983–993.
Dean, M., A. Rzhetsky, and R. Allikmets. 2001. The human ATP-binding cassette (ABC) transporter superfamily. Genome Res. 11:1156–1166.
Decottignies, A., and A. Goffeau. 1997. Complete inventory of the yeast ABC proteins. Nat. Genet. 15:137–145.
Eddy, S. R. 1998. Profile hidden Markov models. Bioinformatics 14:755–763.
Ewart, G. D., and A. J. Howells. 1998. ABC transporters involved in transport of eye pigment precursors in Drosophila melanogaster. Methods Enzymol. 292:213–224.
Fetherston, J. D., V. J. Bertolino, and R. D. Perry. 1999. YbtP and YbtQ: two ABC transporters required for iron uptake in Yersinia pestis. Mol. Microbiol. 32:289–299.
Gorodkin, J., L. J. Heyer, S. Brunak, and G. D. Stormo. 1997. Displaying the information contents of structural RNA alignments: the structure logos. Comput. Appl. Biosci. 13:583–586.
Havarstein, L. S., D. B. Diep, and I. F. Nes. 1995. A family of bacteriocin ABC transporters carry out proteolytic processing of their substrates concomitant with export. Mol. Microbiol. 16:229–240.
Higgins, D. G., A. J. Bleasby, and R. Fuchs. 1992. CLUSTAL V: improved software for multiple sequence alignment. Comput. Appl. Biosci. 8:189–191.
Hwang, J., M. Manuvakhova, and P. C. Tai. 1997. Characterization of in-frame proteins encoded by cvaA, an essential gene in the colicin V secretion system: CvaA* stabilizes CvaA to enhance secretion. J. Bacteriol. 179:689–696.
Idei, A., E. Kawai, H. Akatsuka, and K. Omori. 1999. Cloning and characterization of the Pseudomonas fluorescens ATP-binding cassette exporter, HasDEF, for the heme acquisition protein HasA. J. Bacteriol. 181:7545–7551.
Jedlitschky, G., I. Leier, U. Buchholz, K. Barnouin, G. Kurz, and D. Keppler. 1996. Transport of glutathione, glucuronate, and sulfate conjugates by the MRP gene-encoded conjugate export pump. Cancer Res. 56:988–994.
Kanehisa, M., S. Goto, S. Kawashima, and A. Nakaya. 2002. The KEGG databases at GenomeNet. Nucleic Acids Res. 30:42–46.
Ketchum, C. J., W. K. Schmidt, G. V. Rajendrakumar, S. Michaelis, and P. C. Maloney. 2001. The yeast a-factor transporter Ste6p, a member of the ABC superfamily, couples ATP hydrolysis to pheromone export. J. Biol. Chem. 276:29007–29011.
Khudyakov, I., and C. P. Wolk. 1997. hetC, a gene coding for a protein similar to bacterial ABC protein exporters, is involved in early regulation of heterocyst differentiation in Anabaena sp. strain PCC 7120. J. Bacteriol. 179:6971–6978.
Kispal, G., P. Csere, C. Prohl, and R. Lill. 1999. The mitochondrial proteins Atm1p and Nfs1p are essential for biogenesis of cytosolic Fe/S proteins. EMBO J. 18:3981–3989.
Kobayashi, N., K. Nishino, and A. Yamaguchi. 2001. Novel macrolide-specific ABC-type efflux transporter in Escherichia coli. J. Bacteriol. 183:5639–5644.
Kolaczkowski, M., A. Kolaczowska, J. Luczynski, S. Witek, and A. Goffeau. 1998. In vivo characterization of the drug resistance profile of the major ABC transporters and other components of the yeast pleiotropic drug resistance network. Microbiol. Drug Resist. 4:143–158.
Krogh, A., B. Larsson, G. von Heijne, and E. L. Sonnhammer. 2001. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J. Mol. Biol. 305:567–580.
Kullback, S., and R. A. Leibler. 1951. On information and sufficiency. Ann. Math. Stat. 22:76–86.
Lin, W., K. J. Fullner, R. Clayton, J. A. Sexton, M. B. Rogers, K. E. Calia, S. B. Calderwood, C. Fraser, and J. J. Mekalanos. 1999. Identification of a vibrio cholerae RTX toxin gene cluster that is tightly linked to the cholera toxin prophage. Proc. Natl. Acad. Sci. USA 96:1071–1076.
Locher, K. P., A. T. Lee, and D. C. Rees. 2002. The E. coli BtuCD structure: a framework for ABC transporter architecture and mechanism. Science 296:1091–1098.
Marcotte, E. M., M. Pellegrini, H. L. Ng, D. W. Rice, T. O. Yeates, and D. Eisenberg. 1999. Detecting protein function and protein-protein interactions from genome sequences. Science 285:751–753.
Mauel, C., M. Young, P. Margot, and D. Karamata. 1989. The essential nature of teichoic acids in Bacillus subtilis as revealed by insertional mutagenesis. Mol. Gen. Genet. 215:388–394.
McMorran, B. J., M. E. Merriman, I. T. Rombel, and I. L. Lamont. 1996. Characterisation of the pvdE gene which is required for pyoverdine synthesis in Pseudomonas aeruginosa. Gene 176:55–59.
Ortiz, D. F., T. Ruscitti, K. F. McCue, and D. W. Ow. 1995. Transport of metal-binding peptides by HMT1, a fission yeast ABC-type vacuolar membrane protein. J. Biol. Chem. 270:4721–4728.
Pearson, W. R. 1996. Effective protein sequence comparison. Methods Enzymol. 266:227–258.
Pestova, E. V., L. S. Havarstein, and D. A. Morrison. 1996. Regulation of competence for genetic transformation in Streptococcus pneumoniae by an auto-induced peptide pheromone and a two-component regulatory system. Mol. Microbiol. 21:853–862.
Pittman, M. S., H. Corker, G. Wu, M. B. Binet, A. J. Moir, and R. K. Poole. 2002. Cysteine is exported from the Escherichia coli cytoplasm by CydDC, an ATP-binding cassette-type transporter required for cytochrome assembly. J. Biol. Chem. 277:49841–49849.
Quigley, N. B., Y. Y. Mo, and D. C. Gross. 1993. SyrD is required for syringomycin production by Pseudomonas syringae pathovar syringae and is related to a family of ATP-binding secretion proteins. Mol. Microbiol. 9:787–801.
Reizer, J., A. Reizer, and M. H. Saier Jr. 1992. A new subfamily of bacterial ABC-type transport systems catalyzing export of drugs and carbohydrates. Protein Sci. 1:1326–1332.
Reuter, G., T. Janvilisri, H. Venter, S. Shahi, L. Balakrishnan, and H. W. van Veen. 2003. The ATP binding cassette multidrug transporter LmrA and lipid transporter MsbA have overlapping substrate specificities. J. Biol. Chem. 278:35193–35198.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425.
Sanchez-Fernandez, R., T. G. Davies, J. O. Coleman, and P. A. Rea. 2001. The Arabidopsis thaliana ABC protein superfamily, a complete inventory. J Biol Chem 276:30231–30244.
Saraste, M., P. R. Sibbald, and A. Wittinghofer. 1990. The P-loop—a common motif in ATP- and GTP-binding proteins. Trends Biochem. Sci. 15:430–434.
Saurin, W., and E. Dassa. 1994. Sequence relationships between integral inner membrane proteins of binding protein-dependent transport systems: evolution by recurrent gene duplications. Protein Sci. 3:325–344.
Saurin, W., M. Hofnung, and E. Dassa. 1999. Getting in or out: early segregation between importers and exporters in the evolution of ATP-binding cassette (ABC) transporters. J. Mol. Evol. 48:22–41.
Schmitt, L., H. Benabdelhak, M. A. Blight, I. B. Holland, and M. T. Stubbs. 2003. Crystal structure of the nucleotide-binding domain of the ABC-transporter haemolysin B: identification of a variable region within ABC helical domains. J. Mol. Biol. 330:333–342.
Sharom, F. J. 1997. The P-glycoprotein efflux pump: how does it transport drugs? J. Membr. Biol. 160:161–175.
Smith, T. F., and M. S. Waterman. 1981. Identification of common molecular subsequences. J. Mol. Biol. 147:195–197.
Sokal, R. R., and C. D. Michener. 1958. A statistical method for evaluating systematic relationships. University of Kansas Science Bulletin, 38:1409–1438.
Stanfield, S. W., L. Ielpi, D. O'Brochta, D. R. Helinski, and G. S. Ditta. 1988. The ndvA gene product of Rhizobium meliloti is required for beta-(1-2)glucan production and has homology to the ATP-binding export protein HlyB. J. Bacteriol. 170:3523–3530.
Tam, R., and M. H. Saier Jr. 1993. Structural, functional, and evolutionary relationships among extracellular solute-binding receptors of bacteria. Microbiol. Rev. 57:320–346.
Tauch, A., S. Krieft, A. Puhler, and J. Kalinowski. 1999. The tetAB genes of the Corynebacterium striatum R-plasmid pTP10 encode an ABC transporter and confer tetracycline, oxytetracycline and oxacillin resistance in Corynebacterium glutamicum. FEMS Microbiol. Lett. 173:203–209.
Tomii, K., and M. Kanehisa. 1998. A comparative analysis of ABC transporters in complete microbial genomes. Genome Res. 8:1048–1059.
Verleur, N., E. H. Hettema, C. W. van Roermund, H. F. Tabak, and R. J. Wanders. 1997. Transport of activated fatty acids by the peroxisomal ATP-binding-cassette transporter Pxa2 in a semi-intact yeast cell system. Eur. J. Biochem. 249:657–661.
Wooff, E., S. L. Michell, S. V. Gordon, M. A. Chambers, S. Bardarov, W. R. Jacobs Jr., R. G. Hewinson, and P. R. Wheeler. 2002. Functional genomics reveals the sole sulphate transporter of the Mycobacterium tuberculosis complex and its relevance to the acquisition of sulphur in vivo. Mol. Microbiol. 43:653–663.
Wu, Y. C., and H. R. Horvitz. 1998. The C. elegans cell corpse engulfment gene ced-7 encodes a protein similar to ABC transporters. Cell 93:951–960.
York, G. M., and G. C. Walker. 1998. The Rhizobium meliloti ExoK and ExsH glycanases specifically depolymerize nascent succinoglycan chains. Proc. Natl. Acad. Sci. USA 95:4912–4917.(Yoshinobu Igarashi, Kiyok)
E-mail: kanehisa@kuicr.kyoto-u.ac.jp.
Abstract
ABC (ATP-binding cassette) transporters play an important role in the communication of various substrates across cell membranes. They are ubiquitous in prokaryotes and eukaryotes, and eukaryotic types (EK-types) are distinguished from prokaryotic types (PK-types) in terms of their genes and domain organizations. The EK-types and PK-types mainly consist of exporters and importers, respectively. Prokaryotes have both the EK-types and the PK-types. The EK-types in prokaryotes are usually called "bacterial multidrug ABC transporters," but they are not well characterized in comparison with the multidrug ABC transporters in eukaryotes. Thus, an exhaustive search of the EK-types among diverse organisms and detailed sequence classification and analysis would elucidate the evolutionary history of EK-types. It would also help shed some light on the fundamental repertoires of the wide variety of substrates through which multidrug ABC transporters in eukaryotes communicate. In this work, we have identified the EK-type ABC transporters in 126 prokaryotes using the profiles of the ATP-binding domain (NBD) of the EK-type ABC transporters from 12 eukaryotes. As a result, 11 clusters were identified from 1,046 EK-types ABC transporters. In particular, two large novel clusters emerged, corresponding to the bacterial multidrug ABC transporters related to the ABCB and ABCC families in eukaryotes, respectively. In the genomic context, most of these genes are located alone or adjacent to genes from the same clusters. Additionally, to detect functional divergences in the NBDs, the Kullback-Leibler divergence was measured among these bacterial multidrug transporters. As a result, several putative functional regions were identified, some corresponding to the predicted secondary structures. We also analyzed a phylogeny of the EK-type ABC transporters in both prokaryotes and eukaryotes, which revealed that the EK-type ABC transporters in prokaryotes have certain repertoires corresponding to the conventional ABC protein groups in eukaryotes. On the basis of these findings, we propose an updated evolutionary hypothesis in which the EK-type ABC transporters in both eukaryotes and prokaryotes consisted of several kinds of ABC transporters in putative ancestor cells before the divergence of eukaryotic and prokaryotic cells.
Key Words: ABC transporter ? multidrug ? comparative genomics
Introduction
The ATP-binding cassette (ABC) transporter superfamily is one of the major classes of active transporters, which convey a wide variety of substrates across the cell membrane and are widespread in eubacteria, archaea, and eukaryotes. This superfamily has two kinds of proteins, the prokaryotic type (PK-type) and the eukaryotic type (EK-type). Most EK-type ABC transporters are known as multidrug transporters that export various compounds and drugs from cells. Thus, previous studies on the characterization of the multidrug ABC transporters were mainly focused on eukaryotic cells as opposed to prokaryotic cells because the application of multidrug transporters has great potential in the medical and pharmacological fields. Therefore, the three-dimensional structures of ABC transporters in bacteria are important targets for research, as has been determined recently (Chang and Roth 2001; Locher, Lee, and Rees 2002). However, there has been no comprehensive sequence analysis of EK-type ABC transporters in the context of bacterial multidrug transporters. Therefore, we conducted the identification, classification, and evolutionary analysis of the EK-type ABC transporters in prokaryotes. We also surveyed their substrates from published literature, which can help us understand the relationships between multidrug transporters and their substrates. Moreover, the application of this work to other multidrug transporter families and their substrates may help to increase the understanding of their wide range of specificities. Therefore, the systematic and detailed classification of EK-type ABC transporters in prokaryotes is important to glean more information about the evolutionary relationships of ABC transporters and the relationships between their genes and substrates.
The PK-type is distinguished from the EK-type by gene and domain organizations. The structure of a typical PK-type ABC transporter consists of three types of components: (1) two integral proteins each having six transmembrane segments, (2) two peripheral proteins that bind and hydrolyze ATP, and (3) a periplasmic substrate-binding protein. In many cases, the genes encoding these three types of components, especially the membrane protein and ATP-binding protein components, form an operon as observed in known archaea and eubacteria genomes (Tomii and Kanehisa 1998). On the other hand, in a typical EK-type ABC transporter, the membrane-spanning protein and the ATP-binding protein are fused, and the transporter consists of two polypeptide chains, each having a membrane-spanning domain (MSD) and an ATP-binding domain (NBD). The EK-type transporter has no substrate-binding component, but the MSD carries out substrate selectivity. In PK-type ABC transporters, the NBD component is the most conserved, the MSD component is somewhat less conserved, and the substrate-binding protein is the least conserved in terms of sequence similarity (Tam and Saier 1993; Saurin and Dassa 1994). In a previous study (Tomii and Kanehisa 1998), PK-type ABC transporters were classified into functional (ortholog) groups by a hierarchical cluster analysis using the sequence similarities of NBDs. In EK-type ABC transporters, the NBD is the most-conserved domain, and it is expected that the sequence similarity of NBDs can also be used for functional classification.
In general, eukaryotes only have EK-type ABC transporters, but prokaryotes not only have PK-type ABC transporters but also have certain kinds of EK-type ABC transporters. Most of these are known as "bacterial multidrug transporters." Their sequence repertoires and substrates are not well characterized in comparison with the ABC multidrug transporters in eukaryotes, although the number of completely sequenced prokaryotic genomes is expanding. Thus, the study of EK-type ABC transporters in prokaryotes should shed some light on the fundamental substrate repertoires of eukaryotes.
The conventional ABC transporter families consist of the ABCA, ABCB, ABCC, ABCD, and ABCG families. Most ABC transporter families in human are involved in genetic diseases. The ABCA family is not well characterized in terms of its function and substrates. However, one member of the ABCA family, ced-7 (cel:C48B4.4 in the KEGG database) is known to transport molecules that are involved in programmed cell death. It has been suggested that the lysine residue of the Walker A motif interacts with the phosphate group of the bound ATP (Saraste, Sibbald, and Wittinghofer 1990), and the point mutation experiment on the lysine residue revealed that the N-terminal NBD (N-NBD) of ced-7 is functionally more important than the C-terminal NBD (C-NBD)in avoiding cell death. (Wu and Horvitz 1998).
The ABCB family contains three subfamilies that are known as pgp (p-glycoprotein), atm, and tap (transporter associated with antigen processing). The pgp family has been well studied in mammals such as human and Chinese hamster. Various compounds have been identified as substrates such as peptides, vinca alkaloids, anthracyclines, and epipodophyllotoxins, among others (Sharom 1997). The atm family is found on mitochondria, and tap family is found on the endoplasmic reticulum (ER). One of the atm members, atm1 (sce:YMR301C), is involved in transporting iron-sulfur cytosolic protein. The tap family is known to transport antigen peptides to the ER to form the MHC class I molecules in the immune system.
The ABCC family is also known as mrp (multidrug resistance–associated protein), which has similar domain organizations as the ABCB family. Some of the mrp members (ex. ABCC1, ABCC2) have five additional transmembrane segments in their N-terminals, but the function of these additional MSDs is not clear. The mrp was found in the multidrug-resisting lung cancer cell in which pgp was not overexpressed (Cole et al. 1992). Mrp exports glutathione by drug stimulation (Cole and Deeley 1996) as well as certain substrates in conjugated forms with anions such as glutathione glucuronate and sulfate (Jedlitschky et al. 1996).
The ABCD family is also known as pmp (peroxisomal membrane protein). The proteins of this family, pxa1p and pxa2p found in sce, are implicated in the importation of long-chain acyl-CoA for shortening fatty acid by beta-oxidation (Verleur et al. 1997). The ABCG and its related family consist of white and pdr (pleiotropic drug resistance) families. The white protein (dme:CG2759-PA) is a half-type ABC transporter that imports guanine with brown protein (dme:CG17632-PA) and tryptophan with scarlet protein (dme:CG4314-PA) (Ewart and Howells 1998). The pdr proteins are observed in fungi and plants and, in sce, are known to exhibit distinct but a wide overlapping range of drug resistance profiles for fungicides, flavonoids, and other substances (Kolaczkowski et al. 1998).
The ABCE and ABCF families are not ABC transporters but have the ATP-binding cassette domain in their sequences. They correspond to ribonuclease inhibitors and translation initiation factors. Because of the limited scope of our analysis, these proteins were not included in our work. The ABCH family, which consists of three sequences in dme, was recently defined by Dean, Rzhetsky, and Allikmets (2001), who found that this family is contained in the ABCA family.
Our work presents a global and comprehensive classification of the EK-type ABC transporters in more than a hundred genomes. First, our classification of the identified EK-type ABC transporters in prokaryotes reveals certain relationships among categorized transporter clusters. Second, we described the novel two clusters in prokaryotes, which are related to the ABCB and ABCC eukaryotes and were found the majority of prokaryotes. In addition to this, the measurement of the Kullback-Leibler (KL) divergence of major clusters of bacterial ABC transporters revealed several regions that are suggestive of functional importance, some of which correspond to the predicted secondary structure. Third, based on phylogenetic analysis, we propose an updated evolutionary scenario of the repertoire of the EK-type ABC transporters in putative ancestor cells.
Materials and Methods
Data Source
We collected the amino acid sequences of protein-encoding genes of 12 eukaryotes, 111 eubacteria, and 15 archaea from the KEGG/GENES nucleotide/protein sequence database (July 17, 2003) (Kanehisa et al. 2002). All the organisms and their abbreviations can be referenced from KEGG/GENES at http://www.genome.ad.jp/kegg/kegg2.html, and those of the 12 eukaryotes and 126 prokaryotes used in our study are listed in the Appendix.
Identification of EK-type ABC Transporters
Our workflow is illustrated in figure 1. The identification of EK-type ABC transporters was performed using 12 eukaryote genomes from the KEGG/GENES database. We first identified the putative ABC transporters that have the ABC_tran protein motif in Pfam_ls (global alignment model) of the Pfam database release 9.0 (Bateman et al. 2002) using an expectation value of 0.001. In this work, we defined this ABC_tran protein motif region as a NBD. This region is about 200 residues in length and contain Walker A (P-loop), Walker B, and ABC signature (Walker C) motifs. Next, we removed the sequences that have no membrane-spanning segments using TM-HMM2 (http://www.cbs.dtu.dk/services/TMHMM) (Krogh et al. 2001). TM-HMM2 predicts the number of membrane-spanning segments, given some query sequences. After the putative ABC proteins were obtained, we annotated the sequences that contained the extracted NBDs from the Pfam database and generated a UPGMA (unweighted pair group method with arithmetic mean) (Sokal and Michener 1958) dendrogram based on their NBDs. We manually removed nontransporter proteins with the help of the UPGMA dendrogram and the Pfam domain annotation. After this process, 474 sequences remained as EK-type ABC transporters in eukaryotes. These 474 sequences were classified by the complete linkage hierarchical clustering method using their NBDs.
FIG. 1.— The workflow used in identifying and classifying EK-type ABC transporters.
Classification of EK-type ABC Transporters
Each NBD of every ABC transporter protein was aligned with all other extracted NBDs using the SSEARCH version 3.4t20 program (Pearson 1996), which is an implementation of the Smith-Waterman algorithm (Smith and Waterman 1981) with the BLOSUM 50 matrix and gap penalties of –12 and –2 for opening and extension, respectively. We defined the distance of two sequences based on their NBDs as the inverse of the base 2 logarithm of the Smith-Waterman alignment score (SW score). Then, we performed a complete linkage clustering of these sequences based on their NBDs. The complete linkage method is implemented in the R version 1.7.0 statistics programs (http://www.r-project.org). We obtained the conventional human ABC transporter classification groups using a threshold of 5.8 (= 55.7 SW score) as the similarity criterion.
Identification of EK-type ABC Transporters in Prokaryotes
The classified NBDs in the complete linkage clustering were aligned using Clustal W version 1.82 (Higgins, Bleasby, and Fuchs 1992; Chenna et al. 2003) with default parameters. The individual HMMs were constructed by the hmmbuild and hmmcalibrate programs in HMMER version 2.3.1 (Eddy 1998). These constructed HMMs and the ABC_tran protein motif model were used for detecting the EK-type NBDs in the prokaryotic genomes in the KEGG/GENES database. We selected those NBDs that had an e-value under 1.0e-10 because it was a relaxed enough threshold criteria to detect all of the EK-type ABC transporters in prokaryotes. This rather strict criteria was used because the precollected protein set before the TM-HMM2 step contains many PK-type ABC transporters that do not have MSDs. The detected sequences were filtered by TM-HMM2 to obtain proteins that have at least three predicted membrane-spanning segments, resulting in 2,896 entries for the candidates of the EK-type ABC transporters, with most of the PK-type ABC transporters being removed. Then, we drew a UPGMA dendrogram and manually removed candidates that were related to flagellar, symporter, antiporter, kinase, and other functions using their annotations and HMM scores. Finally, 1,046 ABC transporters from prokaryote genomes were obtained.
Phylogenetic Analysis of EK-type ABC Transporters
Our final set of EK-type ABC transporters contained 474 sequences in eukaryotes and 1,046 sequences in prokaryotes. The Ward clustering dendrogram was obtained using R and the inverse value of the Smith-Waterman alignment scores as described above. The KL divergence was calculated using the Web site (http://www.cbs.dtu.dk/gorodkin/appl/plogo.html) (Gorodkin et al. 1997) with the amino acid distribution of our data set. The neighbor-joining (NJ) tree (Saitou and Nei 1987) was generated from ClustalW, NJ plot and in-house scripts. The branches of the NJ tree were merged when they belonged to the same cluster in the Ward cluster analysis. We used the average length of two branches to create a new merged branch. The gene list of the Ward cluster dendrogram, the constructed HMMs, and the phylogenetic tree are available at http://web.kuicr.kyoto-u.ac.jp/igarashi/abc/.
Result
EK-type ABC Transporters in Eukaryotes
Initially, to increase sensitivity for searching the EK-types in prokaryotes, we defined a classification for the NBDs of eukaryotes in our data set according to the hierarchical classification of EK-type ABC transporters in eukaryotes, as shown in figure S1 in the Supplementary Material online. In our classification, we found eight clusters that corresponded to the conventional ABC categorizations (see Introduction) according to the complete linkage clustering using the threshold of 5.8 for the logarithm of the SW score. Note that domain organizations are known to be either full-type (MSD-NBD-MSD-NBD or NBD-MSD-NBD-MSD) or half-type (MSD-NBD or NBD-MSD), and we were able to distinguish between these types in our clusters. For each full-type, a half-type similar to either the N-NBD or the C-NBD of the full-type exists in the same cluster.
We noticed that one cluster, however, which corresponded to a subgroup of the ABCA family, did not cluster together with the rest of its family. The N-NBDs of this group are similar to the NBDs of the ABCB family. A closer look revealed that it actually belongs to the ABCA6 subfamily, which contains nine sequences found only in multicellular eukaryotes (hsa:10351, hsa:23461, hsa:23461_1, hsa:10349, mmu:217265, mmu:76184, mmu:217258, mmu27404, and mmu:217262). However, this cluster may have clustered separately because this classification method focuses only on one NBD in each cluster in each integrating step.
To confirm the numbers of the identified EK-types with the previous studies, we list the numbers of EK-types in eukaryotes in table 1. These numbers are slightly different from others previously reported (Dean, Rzhetsky, and Allikmets 2001; Sanchez-Fernandez et al. 2001), because the genomes of hsa, dme, and other organisms are continually being updated, and we incorporated alternative splicing genes into our data set. For the majority, the numbers of ABC transporters in sce coincided with the report by (Decottignies and Goffeau 1997).
Table 1 The Number of EK-type ABC Transporters in Eukaryotes
Hierarchical Clustering of EK-type ABC Transporters
We detected all of the repertoires of EK-type ABC transporters in prokaryotes and eukaryotes and explored the sequences, domains, functional features, and evolutionary relationships among EK-type ABC transporters using the Ward and NJ phylogenetic method. We adopted the Ward hierarchical cluster analysis for classification and the NJ method for exploring the evolutionary relationships between eukaryotes and prokaryotes.
Based on the Ward cluster dendrogram (fig. 2), we obtained 11 clusters that mainly consisted of (1) prokaryotic macrolide-specific transporters, (2) ABC-2 prokaryotes, (3) ABCDs, fused PK-type ABC transporters, yojI, and miscs, (4) ABCGs, (5) ABCG (N-NBD) eukaryotes, (6) ABCA eukaryotes, (7) ABCB eukaryotes, (8) ABCB eukaryotes and similar prokaryotes, (9) ABCC (C-NBD) eukaryotes, (10) ABCC (N-NBD) eukaryotes, and (11) ABCC eukaryotes and similar prokaryotes. We also surveyed the distribution of the NBDs in individual taxa (table 2) and examined each cluster's sequence, domain feature, and function. The number of NBDs in individual organisms is shown table S1 in the Supplementary Material online.
FIG. 2.— The Ward cluster dendrogram of the EK-type ABC transporters in eukaryotes and prokaryotes. Black lines on the left side show relationships of identical sequences that have multiple NBDs. Orange lines indicate the relations of adjacent genes in prokaryotic genomes. The leaves of this clustering correspond to the sequences and are colored red (NBDs used in the Ward clustering analysis), blue (one of multiple NBDs in full-types), purple (FtsX, FHA, peptidase_C38, peptidase_C39, or cNMP-binding motifs), green (ABC_membrane motif defined in Pfam), yellow (nonassigned region), and black (other domains). Blue boxes delineate the individual clusters.
Table 2 The Observation of NBDs of the EK-type ABC Transporters
EK-type ABC Transporters in Prokaryotes
Cluster 1
Cluster 1 consists of ybjZ and its orthologs, so they can be said to represent the family of macrolide-specific exporters because it has been demonstrated that ybjZ (eco:b0879) specifically exports macrolide (Kobayashi, Nishino, and Yamaguchi 2001). We have found that the ybjZ genes have transmembrane segments in their C-terminal regions as annotated (FtsX) by Pfam.
Cluster 2
Cluster 2 consists of ABC-2–type transporters that export oligosaccharides (Reizer, Reizer, and Saier 1992). They have two NBDs in their N-terminal regions. Interestingly, the taxa distribution in the macrolide-specific (cluster 1) and ABC-2–type (cluster 2) clusters are similar, despite the fact that they do not have any substrate correlations or NBDs similarities; they mostly consist of proteobacteria, plantomyces, and green sulfur bacteria.
Cluster 3
Cluster 3 consists of several kinds of proteins, such as members of the ABCD eukaryote family, uncharacterized type (yojI), fused PK-type, and others. In particular, yojI orthologs were found only in proteobacteria (eco, ecj ecc, sty, stt, stm sfl, and sfx), and they have a similar domain organization as the half-types of ABCB and ABCC, but their functions are unknown. However, we note that they have especially close similarity (SW score > 515) to particular cluster 1 members such as pvdE (pae:PA2397) and syrD (rba:RB9612), which are involved in pyoverdine synthesis (McMorran et al. 1996) and transporting syringomycin (Quigley, Mo, and Gross 1993), respectively. Pyoverdine is involved in iron translocations, and syringomycin is known to transport cations. In addition, their distribution among taxonomy does not overlap. The orthologs of pvdE are observed among the proteobacteria gamma subdivision and cyanobacteria (pst:PSPTO2153, pae:PA2397, ppu:PP4216, and tel:tlr1649). The orthologs of syrD are observed among the proteobacteria beta and delta/epsilon subdivisions and plantomyces (rso:RS03729, cje:Cj1587c, and rba:RB9612). Thus, on the basis of homolog substrates and their ortholog distribution, it is suggested that yojI and its orthologs may transport cations with chelating molecules. As for the fused PK-type in this cluster, we observed cysA (rba:RB3287), which transports sulfate (Wooff et al. 2002), and ftsE (dra:DR1550) which is involved in cell division and salt transport (de Leeuw et al. 1999). These fused PK-type ABC transporters are formed from PK-types that are frequently observed in prokaryotic genomes. In that respect, yojI, pvdE, and syrD are different from the usual fused PK-types because they do not have PK-type homologs.
Clusters 4 and 5
Cluster 4 consists of genes whose NBDs are similar to the ABCG family. These half-type ABC transporters contain white (dme:CG2759-PA) and scarlet (dme:CG4314-PA) genes, which transport amino acids (Ewart and Howells 1998). The members of cluster 5 are found only in plants, but the ones from cluster 4 are present in actinobacteria, cyanobacteria, and eukaryotes (ana:all7196, mbo:Mb1776, mtc:MT1789, mtu:Rv1747, sco:SCO1806, sco:SCO5818, sma:SAV1159, sma:SAV2446, sma:SAV6472, syn:sll0778, tel:tll1329, tel:tll1340, and tel:tlr1253).
Cluster 6 to 10
Clusters 6, 7, 9, and 10 consist of eukaryotic genes such as ABCA, ABCB, N-NBD, and C-NBD in ABCC, which were described in the previous section. We note that clusters 2 (ABC-2) and 6 (ABCA) are close to each other in the clustering, although ABCA is only identified in eukaryotes and ABC-2 is only identified in prokaryotes.
Clusters 8 and 11: Two Novel Clusters of Bacterial Multidrug ABC Transporters
Until now, bacterial multidrug transporters have not been classified distinctly, but we were able to obtain two novel clusters based on their NBD similarities, clusters 8 and 11 in our dendrogram. Cluster 8 consists of 497 NBDs in prokaryotes and 25 NBDs in eukaryotes. Cluster 11 consists of 329 NBDs in prokaryotes and 12 NBDs in eukaryotes, the latter of which are organelle transporters such as mitochondria and ER. These clusters are related to ABCB and ABCC of eukaryotes, respectively. Particularly, most of the bacteria in this analysis have genes from both clusters 8 and 11 in their genomes. However, several taxa do not have cluster 8 or 11 NBDs. All neisseria, spirochete, and bacteroid and part of mycoplasma, mycobacterium, and archaea lack cluster 11 NBDs (table 2). On the other hand, all of the delta/epsilon subdivision of proteobacteria, chlamydia, crenarchaeota, and part of mycoplasma, actinobacteria, spirochete, hyperthermophilic bacteria, and euryarchaeota lack cluster 8 NBDs. It is of little probability that only the delta/epsilon subdivision of proteobacteria never contained cluster 8 NBDs, because all the other proteobacteria have cluster 8 NBDs. Thus, we propose that they had lost these cluster 8 NBDs after divergence into the subdivisions. All chlamydia organisms have only one NBD of cluster 11, and these NBDs have high similarity with each other (SW score > 936). Overall, in specific taxa (i.e., proteobacteria gamma, lactobacillales, mollicutes, actinobacteria, and euryarchaeota) some cluster 11 genes are located adjacent to cluster 8 genes, but most genes of cluster 8 and 11 are located alone or adjacent to genes of the same cluster.
To detect the residue variation and the degree of conservation between clusters 8 and 11, we measured their KL divergence (Kullback and Leibler 1951), also known as relative entropy, in their multiple alignment (fig. 3). This multiple alignment is 319 residues in length because of many gaps observed in the SDR (structurally diverse region) and other regions. The SDR was considered as a control region of the NBD to target its cognate MSD (Schmitt et al. 2003). Figure 3 shows that the sequences of cluster 8 conserved better than those of cluster 11, both overall and among individual motif regions. In particular, the degree of conservation tends to increase in the vicinity of the ABC signature and Walker B motif. Moreover, several well-conserved regions were found that do not correspond to the motifs. In cluster 8, we observed several conserved residues "FYD" and "GxDI" between Walker A and Q-loop. Because these conserved residues do not correspond to the predicted secondary structure, we assume that they play a functional role as opposed to a structural role. The distinctive regions from the 156th to the 159th (ATxEE) amino acid residues in SDR may be involved in the recognition of each MSD. The well-conserved residues in the region from the 185th to 198th are hydrophobic amino acids, corresponding to the predicted helix region. The threonine at the 303rd residue is also conserved well and corresponds to the predicted helix region, but its contribution is as of yet unclear. ABC transporters work by ATP hydrolysis energy that is generated at two NBDs. However, both do not always take same function as shown in ced-7, which was noted in the Introduction. Therefore, most clusters 8 and 11 also have such difference in their functions, which may be concerned with the activity of ATP hydrolysis.
FIG. 3.— The subtraction of the Kullback-Leibler (KL) distance of cluster 11 from cluster 8 is shown. The residues that have large difference in KL distances between cluster 8 and 11 are shown with their loci and frequently observed amino acids in parenthesis. Several motifs and predicted secondary structures are also shown. The secondary structure prediction was performed by Jpred using the NBD of bha:BH0941 in cluster 8. The dotted line of the predicted secondary structure indicates the gaps of the multiple alignment.
We surveyed the known substrates of the sequences in clusters 8 and 11 from the literature in an attempt to find some pattern among them (table 3). As a result, we found that both clusters included genes that transport peptides, low-molecular-weight sugars, iron compounds, and hydrophobic drugs. These compounds are also transported by ABC transporters in eukaryotes. Thus, the spectrum of substrates in eukaryotic ABC transporters is similar to the EK-type ABC transporters in prokaryotes, but it seems that there is no significant difference between the spectrum of substrates of clusters 8 and 11 at this time, and further research is required. A systematic compound search against each transporter, and more three-dimensional crystal structure data would aid further work in elucidating the difference in function of these clusters.
Table 3 Functional Description of Representative Genes in Clusters 8 and 11
Phylogenetic Analysis of EK-type ABC Transporters
An unrooted NJ phylogenetic tree of our data set is presented in figure 4, and in this figure, clusters 8 and 11, both of whose NBDs are of bacterial multidrug transporters, fell on different branches. This supports the possibility that clusters 8 and 11 have evolved separately. In addition to this, cluster 7 (ABCB eukaryotes) and the clusters 9 and 10 (ABCC eukaryotes) are close to cluster 11 and cluster 8, respectively. This indicates that the eukaryote multidrug transporters ABCB and ABCC may have originated from bacterial multidrug transporters. Our tree also indicates that the macrolide-specifc exporters of cluster 1 seem to have a different evolutionary origin than bacterial multidrug transporters because of its far distance. We also note that clusters 2 (ABCA) and 6 (ABC-2) are close in our tree, thus, implying that they are near in evolution in terms of their NBDs.
FIG. 4.— The unrooted NJ phylogenetic tree of NBDs of EK-type ABC transporters in eukaryotes and prokaryotes. Red branches indicate cluster 11 (ABCC prokaryote). Orange branches indicate cluster 7 (prokaryotes ABCB). The green branches indicate cluster 4 (ABCG). Yellow branches indicate cluster 3 (ABCD, fused PK-type, and miscs). Black branches indicate other clusters (10,9: ABCC [C-NBD, N-NBD], 1: macrolide-specific transporters, 2: ABC-2, 5: ABCG (N-NBD), and 6: ABCA). The number of merged branches are given in angle brackets. The thickness of the lines corresponding to the number of merged branches. The thinnest lines indicate fewer than 10 merged branches, the medium lines correspond to 10 to 29 merged branches, and the thickest lines correspond to 30 or more merged branches. We emphasize mainly those branches that have over 30 branches merged.
Discussion
In this work, we used two hierarchical clustering methods. The Ward method is a classic classification method that attempts to minimize the loss associated with each grouping. At each step in the analysis, the union of every possible cluster pair is considered and the two clusters whose fusion results in the minimum increase in information loss are combined. Ward defines information loss in terms of an error sum-of-squares criterion. Our second clustering method, the complete linkage hierarchical clustering method, provides results using absolute as opposed to relative values. This allows us to distinguish clearly similar clusters based on such metrics as the SW-score.
Before the identification of EK-types in prokaryotes, we identified and classified the EK-type ABC transporters in eukaryotes. The classification results agreed with the conventional human ABC transporter classification. The similarities between N-NBD and C-NBD in the same family are higher than the similarities between N-NBD and C-NBD of other families. It is suggested that there is an interaction between the N-terminal side and C-terminal side in an ABC transporter, which does not allow drastic mutations in them.
In the analysis of EK-type ABC transporters in eukaryotes and prokaryotes, we could obtain clear clusters that were not obtained by other classification methods such as complete-linkage clustering, single-linkage clustering, and UPGMA. We demonstrated that several certain repertoires of sequence and domain organization were observed in the EK-type ABC transporters in prokaryotes, including the bacterial multidrug ABC transporters. The identified EK-type NBDs in prokaryotes corresponds to the NBDs of the ABCB, ABCC, and ABCG families of eukaryotic ABC transporters. Furthermore, in the bacterial multidrug ABC transporters, there are two novel and large groups, clusters 8 and 11. These two clusters are difficult to detect by simple sequence similarity search. Additionally, we note that cluster 11 sequences are related to ABCC eukaryotes. However, they are not accompanied by the additional MSDs that are well conserved in the mammalian ABCC family known as mrp. Thus, we suggest that cluster 11 members have a decreased possibility of transporting the substrates of mrp.
In terms of the genomic context, we found that there are adjacent half-type transporters, which result in transporters that function as full-types. We claim that these adjacent half-type transporters are the pre-evolution versions of the full-types. In addition, in clusters 8 and 11, some half-type transporters were found that have additional domains in their N-terminals. These additional domains were identified as peptidase_C39, peptidase_S8, and cNMP-binding domains in Pfam. These domains imply that these ABC transporters are dual-function proteins. The peptidase_C39 domains were observed in prokaryotes in this work and function to cleave off signal peptides during maturation (Havarstein, Diep, and Nes 1995). The peptidase_S8 domains were observed only in ddi and function as serine proteases. The cNMP-binding domain may play a role in signal transductions. Therefore, it is likely that such domains are fused into the N-terminals of the half-type transporters so that they were able to obtain another function. We propose that these phenomena are the results of the efficiency of expression and function through gene evolution.
In this section, we assess the evolutionary relationships of each family for constructing an updating evolutionary hypothesis of ABC transporters. An evolutionary hypothesis of the exporter and importers of ABC transporters was proposed by (Saurin, Hofnung, and Dassa 1999). They postulated that the importer originated from prokaryotes and that the exporter originated from both eukaryotes and prokaryotes. In addition to this, they argued that it would be improbable to derive the eukaryotic exporters from prokaryotic exporters in the context of their phylogenetic analysis. They also proposed that putative ancestor cells that had originated from eukaryotic and prokaryotic cells already contained the specific divergent exporters. On the basis of this hypothesis and the result of our classification and phylogenetic analysis, we constructed an evolutionary hypothesis of the EK-type ABC transporters, as illustrated in figure 5.
FIG. 5.— Our proposed evolutionary scenario of the NBDs of the EK-type ABC transporter. We consider only NBDs in this scheme. The dotted line indicates a horizontal transfer.
We assume that several NBD repertoires of the ABC transporters that currently exist would have also been present in putative ancestor cells. Our classification and phylogenetic analysis (clusters 8 and 11) supports the fact that the two major bacterial multidrug transporters are evolutionarily related to the ABCB and ABCC families of eukaryotes. Furthermore, because the ABCA and ABC-2 families are also close to each other in our phylogenetic tree, we claim that they may have shared same NBDs in their origins and changed their domain organizations after putative ancestor cells diverged into eukaryotes and prokaryotes. However, further work is necessary to ascertain the origins and evolutionary relationships between them.
A part of the ABCG proteins of prokaryotes have the FHA (forked head–associated) domain, which is mainly observed in eukaryotic genomes and is involved in binding phosphorylated peptides. In our analysis, the ABCG family was also mainly observed in eukaryotes. Therefore, it is probable that the ABCG family originated in eukaryotes and propagated through horizontal gene transfer to cyanobacteria and actinobacteria, among others. Moreover, the macrolide-specific transporters and fused PK-type transporters may have appeared in prokaryotes after the divergence of eukaryotes and prokaryotes because they are found only in prokaryotes. As for the ABCD family, we found that it is localized at an ancient cellular component peroxisome that is observed in virtually all eukaryotes. Thus, it is likely to have existed since the very early stages of eukaryotic cells. Therefore, we propose that the set of NBDs of ABC transporters in putative ancestor cells are similar to ABCB, ABCC, ABCA, and ABC-2.
In addition, we observed several fused PK-type ABC transporters, but prokaryotes typically have nonfused PK-type ABC transporters as importers. The nonfused PK-types consist of operons, and their domain (component) organizations are not conserved as well as in the EK-types. In general, it is expected that genes that interact with each other tend to consist of operons, thus increasing the probability that they are fused through evolution to function more efficiently (Marcotte et al. 1999). Therefore, it is speculated that the PK-type genes have not yet undergone as much evolutionary time as have the EK-type genes.
Conclusion
In this work, we identified EK-type ABC transporters in approximately a hundred organisms and analyzed their repertoires, distributions, and relationships. In particular, we identified two novel large clusters in bacterial multidrug transporters and elucidated their differential regions, which are proposed to be important regions for their functions. We also developed a methodology using an HMM for each family that proved very useful for identifying families in EK-type ABC transporters in newly sequenced genomes. This methodology gave better results than the general motif model database Pfam. The proposed hypothetical evolutionary scenario is not a conclusive account, but it is of much use for further functional analysis of EK-type ABC transporters.
Appendix
The abbreviations of organisms are as follows:
From eukaryotes, Homo sapiens (hsa), Mus musculus (mmu), Rattus norvegicus (rno), Drosophila melanogaster (dme), Caenorhabditis elegans (cel), Arabidopsis thaliana (ath), Oryza sativa (osa), Plasmodium falciparum (pfa), Dictyostelium discoideum (ddi), Saccharomyces cerevisiae (sce), Schizosaccharomyces pombe (spo), and Encephalitozoon cuniculi (ecu)
From bacteria, Escherichia coli K-12 MG1655 (eco), Escherichia coli K-12 W3110 (ecj), Escherichia coli O157 EDL933 (ece), Escherichia coli O157 Sakai (ecs), Escherichia coli CFT073 (ecc), Salmonella typhi CT18 (sty), Salmonella typhi Ty2 (stt), Salmonella typhimurium (stm), Yersinia pestis CO92 (ype), Yersinia pestis KIM (ypk), Shigella flexneri 301 (sfl), Shigella flexneri 2457T (sfx), Buchnera sp. APS (buc), Buchnera aphidicola Sg (bas), Buchnera aphidicola Bp (bab), Wigglesworthia brevipalpis (wbr), Haemophilus influenzae (hin), Pasteurella multocida (pmu), Xylella fastidiosa 9a5c (xfa), Xylella fastidiosa Temecula1 (xft), Xanthomonas campestris (xcc), Xanthomonas axonopodis (xac), Vibrio cholerae (vch), Vibrio vulnificus CMCP6 (vvu), Vibrio parahaemolyticus (vpa), Pseudomonas aeruginosa (pae), Pseudomonas putida (ppu), Pseudomonas syringae pv. tomato (pst), Shewanella oneidensis (son), Coxiella burnetii (cbu), Neisseria meningitidis MC58 (nme), Neisseria meningitidis Z2491 (nma), Ralstonia solanacearum (rso), Nitrosomonas europaea (neu), Helicobacter pylori 26695 (hpy), Helicobacter pylori J99 (hpj), Helicobacter hepaticus (hhe), Campylobacter jejuni (cje), Rickettsia prowazekii (rpr), Rickettsia conorii (rco), Mesorhizobium loti (mlo), Sinorhizobium meliloti (sme), Agrobacterium tumefaciens C58 UWash (atu), Agrobacterium tumefaciens C58 Cereon (atc), Brucella melitensis (bme), Brucella suis (bms), Bradyrhizobium japonicum (bja), Caulobacter crescentus (ccr), Bacillus subtilis (bsu), Bacillus halodurans (bha), Bacillus anthracis (ban), Bacillus cereus (bce), Oceanobacillus iheyensis (oih), Staphylococcus aureus N315 (sau), Staphylococcus aureus Mu50 (sav), Staphylococcus aureus MW2 (sam), Staphylococcus epidermidis (sep), Listeria monocytogenes (lmo), Listeria innocua (lin), Lactococcus lactis (lla), Streptococcus pyogenes SF370 (spy), Streptococcus pyogenes MGAS8232 (spm), Streptococcus pyogenes MGAS315 (spg), Streptococcus pyogenes SSI-1 (sps), Streptococcus pneumoniae TIGR4 (spn), Streptococcus pneumoniae R6 (spr), Streptococcus agalactiae 2603 (sag), Streptococcus agalactiae NEM316 (san), Streptococcus mutans (smu), Lactobacillus plantarum (lpl), Enterococcus faecalis (efa), Clostridium acetobutylicum (cac), Clostridium perfringens (cpe), Clostridium tetani (ctc), Thermoanaerobacter tengcongensis (tte), Mycoplasma genitalium (mge), Mycoplasma pneumoniae (mpn), Mycoplasma pulmonis (mpu), Mycoplasma penetrans (mpe), Mycoplasma gallisepticum (mga), Ureaplasma urealyticum (uur), Mycobacterium tuberculosis H37Rv (mtu), Mycobacterium tuberculosis CDC1551 (mtc), Mycobacterium bovis (mbo), Mycobacterium leprae (mle), Corynebacterium glutamicum (cgl), Corynebacterium efficiens (cef), Streptomyces coelicolor (sco), Streptomyces avermitilis (sma), Bifidobacterium longum (blo), Tropheryma whipplei Twist (twh), Tropheryma whipplei TW08/27 (tws), Fusobacterium nucleatum (fnu), Rhodopirellula baltica (rba), Chlamydia trachomatis (ctr), Chlamydia muridarum (cmu), Chlamydophila pneumoniae CWL029 (cpn), Chlamydophila pneumoniae AR39 (cpa), Chlamydophila pneumoniae J138 (cpj), Chlamydophila caviae (cca), Borrelia burgdorferi (bbu), Treponema pallidum (tpa), Leptospira interrogans (lil), Bacteroides thetaiotaomicron (bth), Synechocystis sp. PCC6803 (syn), Thermosynechococcus elongatus (tel), Anabaena sp. PCC7120 (ana), Chlorobium tepidum (cte), Deinococcus radiodurans (dra), and Aquifex aeolicus (aae), Thermotoga maritima (tma).
From archaea, Methanococcus jannaschii (mja), Methanosarcina acetivorans (mac), Methanobacterium thermoautotrophicum (mth), Methanopyrus kandleri (mka), Archaeoglobus fulgidus (afu), Halobacterium sp. (hal), Thermoplasma acidophilum (tac), Thermoplasma volcanium (tvo), Pyrococcus horikoshii (pho), Pyrococcus abyssi (pab), Pyrococcus furiosus (pfu), Aeropyrum pernix (ape), Sulfolobus solfataricus (sso), Sulfolobus tokodaii (sto), and Pyrobaculum aerophilum (pai).
Acknowledgements
We thank Jean-Phillipe Vert, Ruy Jauregui, and Susumu Goto for critical reading of the manuscript and helpful comments. We also thank Hiroyuki Toh for helpful discussions and comments for the KL-divergence analysis. This work was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology of Japan, the Japan Society for the Promotion of Science, and the Japan Science and Technology Corporation.
References
Ahn, J. H., J. G. Pan, and J. S. Rhee. 1999. Identification of the tliDEF ABC transporter specific for lipase in Pseudomonas fluorescens SIK W1. J Bacteriol. 181:1847–1852.
Bateman, A., E. Birney, L. Cerruti, R. Durbin, L. Etwiller, S. R. Eddy, S. Griffiths-Jones, K. L. Howe, M. Marshall, and E. L. Sonnhammer. 2002. The Pfam protein families database. Nucleic Acids Res. 30:276–280.
Becker, A., H. Kuster, K. Niehaus, and A. Puhler. 1995. Extension of the Rhizobium meliloti succinoglycan biosynthesis gene cluster: identification of the exsA gene encoding an ABC transporter protein, and the exsB gene which probably codes for a regulator of succinoglycan biosynthesis. Mol. Gen. Genet. 249:487–497.
Benabdelhak, H., S. Kiontke, C. Horn, R. Ernst, M. A. Blight, I. B. Holland, and L. Schmitt. 2003. A specific interaction between the NBD of the ABC-transporter HlyB and a C-terminal fragment of its transport substrate haemolysin A. J. Mol. Biol. 327:1169–1179.
Chang, G., and C. B. Roth. 2001. Structure of MsbA from E. coli: a homolog of the multidrug resistance ATP binding cassette (ABC) transporters. Science 293:1793–1800.
Chenna, R., H. Sugawara, T. Koike, R. Lopez, T. J. Gibson, D. G. Higgins, and J. D. Thompson. 2003. Multiple sequence alignment with the Clustal series of programs. Nucleic Acids Res. 31:3497–3500.
Christensen, P. U., J. Davey, and O. Nielsen. 1997. The Schizosaccharomyces pombe mam1 gene encodes an ABC transporter mediating secretion of M-factor. Mol. Gen. Genet. 255:226–236.
Cole, S. P., G. Bhardwaj, J. H. Gerlach, J. E. Mackie, C. E. Grant, K. C. Almquist, A. J. Stewart, E. U. Kurz, A. M. Duncan, and R. G. Deeley. 1992. Overexpression of a transporter gene in a multidrug-resistant human lung cancer cell line. Science 258:1650–1654.
Cole, S. P., and R. G. Deeley. 1996. Multidrug resistance associated with overexpression of MRP. Cancer Treat Res. 87:39–62.
de Leeuw, E., B. Graham, G. J. Phillips, C. M. ten Hagen-Jongman, B. Oudega, and J. Luirink. 1999. Molecular characterization of Escherichia coli FtsE and FtsX. Mol. Microbiol. 31:983–993.
Dean, M., A. Rzhetsky, and R. Allikmets. 2001. The human ATP-binding cassette (ABC) transporter superfamily. Genome Res. 11:1156–1166.
Decottignies, A., and A. Goffeau. 1997. Complete inventory of the yeast ABC proteins. Nat. Genet. 15:137–145.
Eddy, S. R. 1998. Profile hidden Markov models. Bioinformatics 14:755–763.
Ewart, G. D., and A. J. Howells. 1998. ABC transporters involved in transport of eye pigment precursors in Drosophila melanogaster. Methods Enzymol. 292:213–224.
Fetherston, J. D., V. J. Bertolino, and R. D. Perry. 1999. YbtP and YbtQ: two ABC transporters required for iron uptake in Yersinia pestis. Mol. Microbiol. 32:289–299.
Gorodkin, J., L. J. Heyer, S. Brunak, and G. D. Stormo. 1997. Displaying the information contents of structural RNA alignments: the structure logos. Comput. Appl. Biosci. 13:583–586.
Havarstein, L. S., D. B. Diep, and I. F. Nes. 1995. A family of bacteriocin ABC transporters carry out proteolytic processing of their substrates concomitant with export. Mol. Microbiol. 16:229–240.
Higgins, D. G., A. J. Bleasby, and R. Fuchs. 1992. CLUSTAL V: improved software for multiple sequence alignment. Comput. Appl. Biosci. 8:189–191.
Hwang, J., M. Manuvakhova, and P. C. Tai. 1997. Characterization of in-frame proteins encoded by cvaA, an essential gene in the colicin V secretion system: CvaA* stabilizes CvaA to enhance secretion. J. Bacteriol. 179:689–696.
Idei, A., E. Kawai, H. Akatsuka, and K. Omori. 1999. Cloning and characterization of the Pseudomonas fluorescens ATP-binding cassette exporter, HasDEF, for the heme acquisition protein HasA. J. Bacteriol. 181:7545–7551.
Jedlitschky, G., I. Leier, U. Buchholz, K. Barnouin, G. Kurz, and D. Keppler. 1996. Transport of glutathione, glucuronate, and sulfate conjugates by the MRP gene-encoded conjugate export pump. Cancer Res. 56:988–994.
Kanehisa, M., S. Goto, S. Kawashima, and A. Nakaya. 2002. The KEGG databases at GenomeNet. Nucleic Acids Res. 30:42–46.
Ketchum, C. J., W. K. Schmidt, G. V. Rajendrakumar, S. Michaelis, and P. C. Maloney. 2001. The yeast a-factor transporter Ste6p, a member of the ABC superfamily, couples ATP hydrolysis to pheromone export. J. Biol. Chem. 276:29007–29011.
Khudyakov, I., and C. P. Wolk. 1997. hetC, a gene coding for a protein similar to bacterial ABC protein exporters, is involved in early regulation of heterocyst differentiation in Anabaena sp. strain PCC 7120. J. Bacteriol. 179:6971–6978.
Kispal, G., P. Csere, C. Prohl, and R. Lill. 1999. The mitochondrial proteins Atm1p and Nfs1p are essential for biogenesis of cytosolic Fe/S proteins. EMBO J. 18:3981–3989.
Kobayashi, N., K. Nishino, and A. Yamaguchi. 2001. Novel macrolide-specific ABC-type efflux transporter in Escherichia coli. J. Bacteriol. 183:5639–5644.
Kolaczkowski, M., A. Kolaczowska, J. Luczynski, S. Witek, and A. Goffeau. 1998. In vivo characterization of the drug resistance profile of the major ABC transporters and other components of the yeast pleiotropic drug resistance network. Microbiol. Drug Resist. 4:143–158.
Krogh, A., B. Larsson, G. von Heijne, and E. L. Sonnhammer. 2001. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J. Mol. Biol. 305:567–580.
Kullback, S., and R. A. Leibler. 1951. On information and sufficiency. Ann. Math. Stat. 22:76–86.
Lin, W., K. J. Fullner, R. Clayton, J. A. Sexton, M. B. Rogers, K. E. Calia, S. B. Calderwood, C. Fraser, and J. J. Mekalanos. 1999. Identification of a vibrio cholerae RTX toxin gene cluster that is tightly linked to the cholera toxin prophage. Proc. Natl. Acad. Sci. USA 96:1071–1076.
Locher, K. P., A. T. Lee, and D. C. Rees. 2002. The E. coli BtuCD structure: a framework for ABC transporter architecture and mechanism. Science 296:1091–1098.
Marcotte, E. M., M. Pellegrini, H. L. Ng, D. W. Rice, T. O. Yeates, and D. Eisenberg. 1999. Detecting protein function and protein-protein interactions from genome sequences. Science 285:751–753.
Mauel, C., M. Young, P. Margot, and D. Karamata. 1989. The essential nature of teichoic acids in Bacillus subtilis as revealed by insertional mutagenesis. Mol. Gen. Genet. 215:388–394.
McMorran, B. J., M. E. Merriman, I. T. Rombel, and I. L. Lamont. 1996. Characterisation of the pvdE gene which is required for pyoverdine synthesis in Pseudomonas aeruginosa. Gene 176:55–59.
Ortiz, D. F., T. Ruscitti, K. F. McCue, and D. W. Ow. 1995. Transport of metal-binding peptides by HMT1, a fission yeast ABC-type vacuolar membrane protein. J. Biol. Chem. 270:4721–4728.
Pearson, W. R. 1996. Effective protein sequence comparison. Methods Enzymol. 266:227–258.
Pestova, E. V., L. S. Havarstein, and D. A. Morrison. 1996. Regulation of competence for genetic transformation in Streptococcus pneumoniae by an auto-induced peptide pheromone and a two-component regulatory system. Mol. Microbiol. 21:853–862.
Pittman, M. S., H. Corker, G. Wu, M. B. Binet, A. J. Moir, and R. K. Poole. 2002. Cysteine is exported from the Escherichia coli cytoplasm by CydDC, an ATP-binding cassette-type transporter required for cytochrome assembly. J. Biol. Chem. 277:49841–49849.
Quigley, N. B., Y. Y. Mo, and D. C. Gross. 1993. SyrD is required for syringomycin production by Pseudomonas syringae pathovar syringae and is related to a family of ATP-binding secretion proteins. Mol. Microbiol. 9:787–801.
Reizer, J., A. Reizer, and M. H. Saier Jr. 1992. A new subfamily of bacterial ABC-type transport systems catalyzing export of drugs and carbohydrates. Protein Sci. 1:1326–1332.
Reuter, G., T. Janvilisri, H. Venter, S. Shahi, L. Balakrishnan, and H. W. van Veen. 2003. The ATP binding cassette multidrug transporter LmrA and lipid transporter MsbA have overlapping substrate specificities. J. Biol. Chem. 278:35193–35198.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425.
Sanchez-Fernandez, R., T. G. Davies, J. O. Coleman, and P. A. Rea. 2001. The Arabidopsis thaliana ABC protein superfamily, a complete inventory. J Biol Chem 276:30231–30244.
Saraste, M., P. R. Sibbald, and A. Wittinghofer. 1990. The P-loop—a common motif in ATP- and GTP-binding proteins. Trends Biochem. Sci. 15:430–434.
Saurin, W., and E. Dassa. 1994. Sequence relationships between integral inner membrane proteins of binding protein-dependent transport systems: evolution by recurrent gene duplications. Protein Sci. 3:325–344.
Saurin, W., M. Hofnung, and E. Dassa. 1999. Getting in or out: early segregation between importers and exporters in the evolution of ATP-binding cassette (ABC) transporters. J. Mol. Evol. 48:22–41.
Schmitt, L., H. Benabdelhak, M. A. Blight, I. B. Holland, and M. T. Stubbs. 2003. Crystal structure of the nucleotide-binding domain of the ABC-transporter haemolysin B: identification of a variable region within ABC helical domains. J. Mol. Biol. 330:333–342.
Sharom, F. J. 1997. The P-glycoprotein efflux pump: how does it transport drugs? J. Membr. Biol. 160:161–175.
Smith, T. F., and M. S. Waterman. 1981. Identification of common molecular subsequences. J. Mol. Biol. 147:195–197.
Sokal, R. R., and C. D. Michener. 1958. A statistical method for evaluating systematic relationships. University of Kansas Science Bulletin, 38:1409–1438.
Stanfield, S. W., L. Ielpi, D. O'Brochta, D. R. Helinski, and G. S. Ditta. 1988. The ndvA gene product of Rhizobium meliloti is required for beta-(1-2)glucan production and has homology to the ATP-binding export protein HlyB. J. Bacteriol. 170:3523–3530.
Tam, R., and M. H. Saier Jr. 1993. Structural, functional, and evolutionary relationships among extracellular solute-binding receptors of bacteria. Microbiol. Rev. 57:320–346.
Tauch, A., S. Krieft, A. Puhler, and J. Kalinowski. 1999. The tetAB genes of the Corynebacterium striatum R-plasmid pTP10 encode an ABC transporter and confer tetracycline, oxytetracycline and oxacillin resistance in Corynebacterium glutamicum. FEMS Microbiol. Lett. 173:203–209.
Tomii, K., and M. Kanehisa. 1998. A comparative analysis of ABC transporters in complete microbial genomes. Genome Res. 8:1048–1059.
Verleur, N., E. H. Hettema, C. W. van Roermund, H. F. Tabak, and R. J. Wanders. 1997. Transport of activated fatty acids by the peroxisomal ATP-binding-cassette transporter Pxa2 in a semi-intact yeast cell system. Eur. J. Biochem. 249:657–661.
Wooff, E., S. L. Michell, S. V. Gordon, M. A. Chambers, S. Bardarov, W. R. Jacobs Jr., R. G. Hewinson, and P. R. Wheeler. 2002. Functional genomics reveals the sole sulphate transporter of the Mycobacterium tuberculosis complex and its relevance to the acquisition of sulphur in vivo. Mol. Microbiol. 43:653–663.
Wu, Y. C., and H. R. Horvitz. 1998. The C. elegans cell corpse engulfment gene ced-7 encodes a protein similar to ABC transporters. Cell 93:951–960.
York, G. M., and G. C. Walker. 1998. The Rhizobium meliloti ExoK and ExsH glycanases specifically depolymerize nascent succinoglycan chains. Proc. Natl. Acad. Sci. USA 95:4912–4917.(Yoshinobu Igarashi, Kiyok)