Evolution of permease diversity and energy-coupling mechanisms with special reference to the bacterial phosphotransferase system

248

Biochimica et Biophysica Acta, 1018 (1990) 248-251

Elsevier BBAEBC 00027

Evolution of permease diversity and energy-coupling mechanisms with special reference to the bacterial phosphotransferase system Milton H. Saier, Jr. 1, Long-Fei Wu 1, Michael E. Baker and Jonathan Reizer 1

2 Gaye

Sweet 3, Aiala Reizer 1

1 Department of Biology, C.016 and 2 Department of Medicine, M-023, University of California, San Diego, CA (U.S.A.) and 3 Department of Biology, University of Konstanz, Konstanz (F.R.G.)

(Received I May 1990)

Key words: Transmembranesolute transport; Permease; Energy-coupling;Evolution; Transport, active; Group translocation; Phosphotransferasesystem

Different classes of apparently unrelated permeases couple different forms of energy to solute transport. While the energy coupling mechanisms utilized by the different permease classes are clearly distinct, it is proposed, based on structural comparisons, that many of these permeases possess transmembrane, hydrophobic domains which are evolutionarily related. Carriers may have arisen from transmembrane pore-forming proteins, and the protein constituents or domains which are specifically responsible for energy coupling may have had distinct origins. Thus, complex permeases may possess mosaic structures. This suggestion is substantiated by recent findings regarding the evolutionary origins of the bacterial phosphoenolpyruvate-dependentphosphotransferase system (PTS). Mechanistic implications of this proposal are presented.

At least five distinct mechanisms are responsible for the transport of hydrophilic organic molecules across the cytoplasmic membranes of living cells [1-3]: (a) non-saturable, energy-independent diffusion mediated by pore-forming, intergral membrane proteins such as the Na+-channel of nerve cells or the glycerol facilitator of Escherichia coli; (b) facilitated diffusion catalyzed by single-species, stereospecific facilitators such as the glucose cartier or the anion exchanger of the human red blood cell (uniport or antiport); (c) chemiosmoticallycoupled active transport catalyzed by two-species facilitators such as the lactose: H + or the melibiose: Na + carrier of E. coli (symport or secondary active transport); (d) chemically-driven active transport catalyzed, for example, by cation translocating ATPases or by the multicomponent periplasmic binding protein-dependent transport systems such as the maltose permease of E. coli (primary active transport); and (e) group translocation catalyzed by the bacterial p h o s p h o e n o l p y r u v a t e - d e pendent phosphotransferase system. Mutant analyses and sequence comparisons, as well as functional considAbbreviations: PTS, phosphotransferase system; ArsB, oxyanion pump; MIP, major intrinsic protein. Correspondence: M.H. Saier, Jr., Department of Biology, C-016, University of California, San Diego, La Jolla, CA 92093, U.S.A.

erations, suggest that uniporters, antiporters and symporters function by essentially the same cartier-type mechanism with respect to the translocation step, and that they differ from each other only with respect to the number and nature of the species translocated [3-6]. It has been shown in site-specific mutagenesis studies that amino acyl residues in the ninth and tenth of the twelve putative transmembrane a-helices of the l a c t o s e : H + symporter are involved in proton translocation [7]. However, mutation of these as well as other residues can alter the sugar binding site of the permease, suggesting that the sugar and cation binding sites may overlap, and that they are present within the hydrophobic transmembrane segments of the protein [4,8]. The facts (a) that the melibiose cartier as well as other types of permease can utilize Na + in place of H + as the transported cation, (b) that the cation specificity of the melibiose facilitator can be altered by point mutations, and (c) that the cation specificity of the permease is determined in part by the anomeric configuration of the galactoside substrate provide evidence that the same translocation mechanism operates with either cation [4,8,9]. A unified mechanism of solute:cation symport seems likely. The multicomponent binding protein-dependent permeases such as those specific for histidine, maltose and oligopeptides are now known to be driven by ATP

0005-2728/90/$03.50 © 1990 ElsevierSciencePublishers B.V. (BiomedicalDivision)

249 hydrolysis [10-12]. Similarly, the extrusion of toxic anions from E. coli is energized by ATP hydrolysis [13]. These processes do not involve substrate modification. By contrast, permeases of the group translocating phosphotransferase system (PTS) utilize phosphoenolpyruvate to energize transport, and the sugar substrates are phosphorylated during transport. Two permease-specific phosphorylation sites energize transmembrane sugar translocation, and these sites are believed to be topologically distinct from the integral membrane parts of the permeases. Thus, the fl-glucoside or mannitol permease, each of which consists of a single polypeptide chain (possibly present in the membrane as a homodimer), appears to possess both of its phosphorylation sites within hydrophilic domains of the protein, while the high-affinity sugar binding site resides in a structurally independent, hydrophobic domain of the protein [14,15]. These observations suggest that like the cation symport and binding protein-dependent permeases, the PTS permeases contain sugar binding sites localized within the hydrophobic, transmembrane, a-helical regions of the proteins. Further, like the ATP-driven transport systems (the F1F0-ATPases, the binding protein-dependent systems, and the oxyanion-translocating ATPase [9,12,13]) as well as the cation-translocating substrate-decarboxylating transport systems [9], the energy coupling domains of the PTS permeases are topologically distinct, being localized to cytoplasmically exposed domains or polypeptides within the permeases [14-161. While the information summarized above clarifies our picture of permease topography and reveals that energy coupling can be structurally and functionally dissected from substrate binding and translocation, it does not provide new insight into the actual translocation mechanisms. Relevant to these mechanisms are recent permease sequence and structural analyses suggesting that different classes of transport systems may be evolutionarily related (Fig. 1). Thus, (a) the sugar: cation symporters such as the l a c t o s e : H + and melibiose:Na + permeases appear to each consist of twelve transmembrane helical segments (possibly derived from an ancestral protein half as large with only six transmembrane helical segments [5,17]) with both the N- and C-termini localized to the cytoplasmic side of the membrane [4,6-8]. (b) The integral membrane constituents of the binding protein-dependent systems probably span the membrane six times, and they exist in the membrane as heterodimers where the two subunits of the dimeric structure exhibit sequence similarity with each other [12]. (c) A total of six or seven transmembrane segments has been predicted for the two integral membrane constituents of the Na+-translocat ing oxaloacetate decarboxylase [9]. (d) The PTS permeases, which appear to be present in the membrane as homodimers, possess subunits which undoubtedly span

ca~'ier-Lype faci]it~om

binding protein perz~ases

poee-Ly

eases

pore-type ancestral peP|ease

Fig. 1. Schematic diagram of a proposal suggesting that the integral membrane domains of several permease classes exhibit common structural features and have a common evolutionary origin. The ancestral permease, common to the different permease systems depicted, is suggested to be a pore-type transmembrane protein similar in structure to the glycerol permease of E. coli (a pore-type facilitator). Carrier-type facilitators such as the lactose and melibiose permeases of E. coli may have evolved their symport (carrier-like) properties by evolving intramembrane binding sites for the solute and the coupling cation as well as the potential for two metastable conformational states, of about equal free energies, which permit exposure of the central binding site alternately to the two sides of the membrane. The binding protein-dependent permeases may utifize relatively static solute binding sites within the transmembrane channel. They may have recruited high-affinity solute recognition proteins (present in the periplasm) as well as energy-coupling-regulatory proteins (localized to the cytoplasmic face of the membrane). The PTS permeases similarly possess their energy coupling domains localized to the cytoplasmic face of the membrane, and a 'cartier-type' translocating mechanism like that observed for the facilitators is suggested. The common ancestral permease is proposed to have given rise only to the hydrophobic domains of the permeases and not to the auxiliary proteins or domains that function in solute recognition and energy coupling. The proposal illustrated in this figure may apply to other classes of permeases such as the oxyanion-translocating ATPases and the substrate-decarboxylating, cation-translocating transport systems.

the membrane at least six times [14-16]. Indeed, sequence comparisons have revealed that within their transmembrane regions, a segment of the fructosespecific PTS permease of Rhodobacter capsulatus exhibits about 25% identity with membrane imbedded segments of two homologous insulin-responsive glucose facilitators of animals [18], that the mannitol permease of E. coli may be homologous to the N a + / H + antiporter of E. coli, and that the PTS permeases of E. coli exhibit significant sequence similarity with mitochondrial transport proteins which also show sequence identity with the glycerol facilitator (Pao, G.M., Sweet, G., Baker, M.E. and Saier, M.H., Jr., unpublished observations). Finally, the integral membrane component of the oxyanion pump (ArsB) has a molecular weight similar to those of the symporters; ArsB may span the membrane 10 or 12 times [13]. These structural similarities suggest a common origin, but what might the evolutionary precursor have been? Recent sequence comparisons have shown that the

250 glycerol facilitator, which functions as a nonspecific pore in the E. coli cytoplasmic membrane through which straight chain carbon compounds can pass [19,20], shows sequence identity with probable transmembrane poreforming proteins of animals and plants (Ref. 21 and Sweet, G., unpublished results). These pore-forming proteins include the major intrinsic protein (MIP) of mammalian lens cells and nodulin-26 of nodulating soybean root cells and probably contain six potential transmembrane a-helical segments. In one case, that of the MIP of mammalian lens cells, both the N- and C-termini have been shown to be locahzed to the cytoplasmic face of the membrane. Thus, the ancestral permease may have been a simple, nonspecific, transmembrane channel-protein like the present-day glycerol facilitator (Fig. 1). Gene duplication, followed by divergence and association (either covalent or noncovalent) with proteins allowing various modes of energy coupling may have given rise to a diversity of transport systems with their different modes of energy coupling. This possibility must be considered in spite of the fact that the different classes of permeases seldom exhibit significant sequence identity with each other. It is well established that primary structures of proteins diverge during evolution more rapidly than secondary or tertiary structures. For example, several of the periplasmic, solute binding protein components of the ATP-driven, multicomponent binding protein-dependent bacterial permeases show an insignificant degree of sequence identity with each other, but X-ray crystallographic analyses have revealed that they exhibit essentially the same secondary and tertiary structural features [22]. It is much easier to explain this fact by divergent evolution than by convergent evolution since many tertiary protein structures should be capable of fulfilling a particular functional role. We therefore postulate that the integral membrane domains of many permeases, possessing different physiological functions, different modes of action, and different energy coupling mechanisms, may have arisen from a common ancestor (Fig. 1). Mechanistic implications of this unifying concept suggest that at least several classes of permeases (poretype facilitators, carrier-type uniporters, antiporters, and symporters, active transporters driven by ATP hydrolysis or substrate decarboxylation, and group translocatots) may all function essentially by a pore-type mechanism. Stereospecific solute recognition may have been achieved during evolution by the appropriate introduction within the transmembrane pore of specific amino acyl residues which comprised a substrate binding site. Alternate exposure of this binding site to the two sides of the membrane (a prerequisite for a carrier type mechanism) would result from the existence of the two appropriate conformational states of similar free energies equivalent to the 'mobile carrier' or 'mobile

barrier' concept [3,8]. Inclusion of a cation (H30 + or Na +. H20 ) binding site within the region of the pore, near the solute binding site, would give rise to solute: cation symport [3]. As noted above, the process of coupling chemical energy to transport would require the participation of additional proteins or protein domains within the permease. Both within the PTS and the binding protein-type systems, gene fusion and gene splicing events have occurred repeatedly during evolution, altering the numbers of polypeptide chains which comprise the transport system without changing the essential, overall permease structure. Thus, the energy coupling proteins of the PTS in some cases are fused in various combinations with each other a n d / o r with the permeases [14-16,23,24], and some of these proteins exhibit sequence identity with cytoplasmic enzymes that catalyze reactions which have nothing to do with transport. For example, Enzyme I, the first energy coupling protein of the PTS, is homologous to pyruvate : phosphate dikinase, and some evidence suggests that other PTS energy coupling proteins (Enzymes III) also have a common origin with this cytoplasmic enzyme (Ref. 23 and Reizer, A. and Reizer, J., unpublished results). Similarly, the energy-coupling ATP-binding constituents (or domains) of the binding protein-dependent systems exhibit the properties of water-soluble proteins, and homology of these energy coupling proteins (or domains) with a number of ATPbinding proteins which do not function in transport has been demonstrated [12]. The energy-coupling constituents of permeases may therefore have arisen from ancestral proteins which did not function in transport. Finally, the oxyanion ATPase, the Na+-translocating oxaloacetate decarboxylase, and the H ÷- and Na +translocating FoF1-ATPases of bacteria also may have recruited peripheral membrane proteins, non-covalently associated, as their energy coupling constituents [9,13]. Thus, while it seems that the hydrophobic, transmembrane parts of permeases which participate directly in solute translocation probably arose during evolution from one or a few common ancestral pore-type permeases, the subunits (or domains) which catalyze energy coupling probably arose from distinct ancestral proteins. The more complex permease systems may well represent a mosaic of protein domains with different evolutionary origins. This unifying precept can be tested experimentally by determining the sequence, structural, and mechanistic similarities of the different classes of permeases. It predicts the presence of structurally related pores or channels bounded by the six or twelve characteristic transmembrane helical segments of the permeases, some of which have evolved stereospecific substrate binding sites. Evidence for or against this postulate will also be forthcoming when the three-dimensional structures of these proteins become known.

251

Acknowledgements The valuable assistance of Mary Beth Hiller in the preparation of this manuscript is gratefully acknowledged. Work in the authors' laboratories was supported by U.S. Public Health Service Grants 5 RO1 AI 21702 and 2 RO1 AI 14176 from the National Institute of Allergy and Infectious Diseases and the Deutsche Forschungs Gemeinschaft (SFB 156).

References 1 Saier, M.H., Jr. (1985) Mechanisms and Regulation of Carbohydrate Transport in Bacteria, Academic Press, New York. 2 Saier, M.H., Jr. and Boyden, D.A. (1984) Mol. Cell. Biochem. 59, 11-32. 3 Mitchell, P. (1990) Res. Microbiol. 141, 286-289. 4 Botfield, M.C., Wilson, D.M. and Wilson, T.H. (1990) Res. Microbiol. 141, 328-331. 5 Henderson, P.J.F. (1990) Res. Microbiol. 141, 316-328. 6 Maloney, P.C. (1990) Res. Microbiol. 141, 374-383. 7 Roepe, P.D., Consler, T.G., Menezes, M.E. and Kaback, H.R. (1990) Res. Microbiol. 141, 290-308. 8 Brooker, R.J. (1990) Res. Microbiol. 141, 309-315.

9 Dimroth, P. (1990) Res. Microbiol. 141, 332-336. 10 Ames, G.F.-L. (1990) Res. Microbiol. 141, 341-348. 11 Dean, D.A., Davidson, A.L. and Nikaido, H. (1990) Res. Microbiol. 141, 348-352. 12 Higgins, C.F. (1990) Res. Microbiol. 141, 353-360. 13 Rosen, B.P. (1990) Res. Microbioi. 141, 336-341. 14 Sutrina, S.L., Schnetz, K., Rak, B. and Saier, M.H., Jr. (1990) Res. Microbiol. 141, 368-374. 15 Jacobson, G.R. (1990) Res. Microbiol. 141, 365-368. 16 Emi, B. (1990) Res. Microbiol. 141,360-364. 17 Szkutnicka, K., Tschopp, J.F., Andrews, L. and Cirillo, V.P. (1989) J. Bacteriol. 171, 4486-4493. 18 Wu, L.-F. and Saier, M.H., Jr. (1990) Mol. Microbiol., in press. 19 Heller, K.B., Lin, E.C.C. and Wilson, T.H. (1980) J. Bacteriol. 144, 274-278. 20 Sweet, G., Gandor, C., Voegele, R., Wittekindt, N., Beuerle, J., Truniger, V., Lin, E.C.C. and Boos, W. (1990) J. Bacteriol. 172, 424-430. 21 Baker, M.E. and Saier, M.H., Jr. (1990) Cell 60, 185-186. 22 Adams, M.D. and Oxender, D.L. (1989) J. Biol. Chem. 264, 15739-15742. 23 Wu, L.-F., Tomich, J.M. and Saier, M.H., Jr. (1990) J. Mol. Biol., in press. 24 Saier, M.H., Jr., Yamada, M., Erni, B., Suda, K., Lengeler, J., Ebner, R., Argos, P., Rak, B., Schnetz, K., Lee, C.A., Stewart, G.C., Breidt, F., Jr., Waygood, E.B., Peri, K.G. and Doolittle, R.F. (1988) FASEB J. 2, 199-208.