Elevator-type mechanisms of membrane transport

Structural studies of membrane transporters from diverse protein families have revealed that alternating access may by achieved in many ways (reviewed recently [1]). The so-called “moving barrier” mechanism is a frequently used solution (Figure 1). Proteins operating by this mechanism bind the transported substrate in a deep cavity, which is accessible to the aqueous environment from one side of the membrane only. A conformational change then closes off the access path to the binding site (gate closure), and opens up a new path to the other side of the membrane (gate opening). Moving barrier transporters thus work with two separate gates. Synchronization of opening and closing of the two gates is crucial: intermediate occluded states with both gates closed may be visited, but states with both gates open are prohibited. During the conformational transitions in the protein, the substrate remains bound at roughly the same position relative to the bilayer plane, until the conformational switching has been completed and a route to the aqueous solution on the opposite side of the membrane has opened. In many cases, the substrate-binding site is located halfway through the bilayer between two proteins domains that move around the substrate when switching between inward- and outward-facing states. The transport protein thus serves as a “moving barrier’. Prominent examples of proteins using a moving barrier mechanism include members of the major facilitator superfamily, in which two homologous protein domains swivel around the substrate as a rocker switch [2,3] (Figure 1a); the LeuT-fold proteins in which one protein domain moves as a rocking bundle relative to a fixed second (non-homologous) domain [4] (Figure 1b); and mitochondrial carriers, where three homologous domains pivot around the substrate in a concerted way as a diaphragm [5] (Figure 1c).

The elevator-type transport mechanism offers an alternative solution to achieve alternating access [1]. Proteins using this mechanism consist of a moving and fixed domain (often termed “transport” and “scaffold” domain, respectively). Switching between outward- and inward-facing states involves the sliding of the entire transport domain through the bilayer as a rigid body. In contrast with proteins using a moving-barrier mechanism, the substrate-binding site translocates some distance across the bilayer during transport along with the transport domain (Figure 2). Because of the displacement of the substrate the elevator mechanism has been described as “moving carrier’. Alternatively, the name “fixed barrier mechanism” has been proposed [1], but as we will discuss below, some elevator proteins may not have a fixed barrier. Therefore, we prefer the names “elevator-type” or “moving carrier” mechanism. It is noteworthy that the classification of a transporter mechanism as “moving barrier” or “moving carrier” is based solely on the structural changes that take place in the proteins during transport, and that it does not have predictive value for the transporter's substrate specificity, coupling ion specificity (in secondary active transporters), or for the kinetic mechanism.

The first elevator-type mechanism was described in 2009 for the aspartate transporter Glt_Ph [6], a member of the glutamate transporter or SLC1 (Solute Carrier 1) family, but the name “elevator” was not used until 2011 [7]. In recent years, elevator-type mechanisms have been proposed for numerous other proteins (Table 1). Many of the proteins shown in Table 1 are sodium-coupled secondary active transporters, but a sub-set of ATP-binding cassette (ABC) transporters, phosphotransferase system (PTS) transporters and unclassified transport proteins also appear to use elevator-type mechanisms. The abundant representation of secondary active transporters in Table 1 may simply be a reflection of the large number of families of secondary transporters that have evolved [8]. In this review, we focus on the global structural changes that take place in elevator-type membrane transporters. We do not discuss the kinetics of switching between outward- and inward-facing states, which may depend on the occupancy of the solute-binding site, or binding of compounds to allosteric sites, such as co-transported ion(s) in secondary active transporters, or nucleotides in ATP-binding cassette (ABC) transporters. For details of the intricate mechanisms of coupling of transport to co-ion translocation or ATP hydrolysis we refer to recent reviews [9–12].

In proteins using the elevator mechanism, the substrate moves some distance across the membrane during the conformational switching. In Table 1, the extent of the movement is indicated as the “vertical distance’, the displacement of the substrate in z-direction if the membrane plane is defined as the xy plane. In many cases, the domain movement is more complex than a simple translation, and the total distance over which the substrate is displaced is larger than the vertical distance (Table 1). Structurally, elevator-type membrane transporters show large diversity, indicating that the vertical movement can be realised in multiple ways, but many of the proteins have some characteristics in common. First, the transported substrates bind exclusively, or predominantly, to the transport domain, which is a prerequisite for joined movement of the transport domain and substrate, relative to the rigid scaffold domain. Second, in many cases the transport domain contains structural elements named helical hairpins (HPs) that form the gates, which must be open to allow access of the substrate to the bindings site, and closed to make the elevator movement possible. An open gate prevents sliding of the transport domain relative to the scaffold domain because of steric incompatibility. Third, almost all proteins using elevator transport mechanisms have a membrane topology with inverted repeats [13], resulting in internal pseudosymmetry, which has been used to model the outward-facing conformation based on an inward-facing structure or vice versa [14–17]. Finally, elevator-type transport proteins are often homodimers or homotrimers. Subunit contacts in the oligomers are made exclusively by the scaffold domains, while the transport domains are located peripherally (Figure 3). It is not entirely clear what is the functional significance of the oligomeric state. For homotrimeric members of the glutamate transporter family, it has been shown that the three protomers function independently [18–25], but it is possible that cooperativity may occur in other protein families.

Despite these similarities, global elevator movements and local gating motions vary widely between different protein families (Table 1). Using currently available structural data, elevator mechanisms can be classified into three types with pronounced differences in the way gating is achieved. The classification is based on proteins for which structures are available of multiple conformational states. For many of the proteins in Table 1, only a single structure has been solved, and therefore it is not yet possible to unambiguously classify them.

The glutamate transporter (SLC1) family of solute transporters is structurally well-characterized with 39 available structures of four different family members: the prokaryotic sodium-dependent aspartate transporters Glt_Ph and Glt_Tk, the human sodium- and potassium-dependent glutamate transporter EAAT1 (Excitatory Amino Acid Transporter 1), and the human neutral amino acid exchanger ASCT2 (Alanine Serine Cysteine Transporter 2) (Table 1 and reviewed in [26]) . While Glt_Ph is the prototypical elevator transporter, ASCT2 is the first SLC1 member, for which four key conformations have been resolved structurally: outward-open, outward–occluded [27], inward-open [28] and inward–occluded [29]. We will use these structures to describe the one-gate, fixed barrier elevator movement (Figure 2a).

Like all members of the SLC1 family, neutral amino acid transporter ASCT2 is a homotrimer. Each monomer consists of 8 transmembrane segments (TMs) that form a scaffold domain (TM1–2, TM4–5) and a transport domain (TM3, TM6–8). The transport domain additionally contains two helical hairpins (HP1 and HP2). In the outward-facing states the substrate-binding site is close to the extracellular side of the membrane, and the only difference between open and closed conformations is the position of HP2, which works as a gate to provide access to the binding site from the extracellular aqueous environment [27] (Figure 2a). When the gate is closed, the transported substrate is occluded within the transport domain, which makes the elevator movement possible. The binding site relocates by a distance of ∼19 Å perpendicular to the membrane plane between the outward- to the inward-facing orientation. Strikingly, HP2 was also found to be the gate on the intracellular side, hence the name one-gate elevator mechanism [28]. HP1 plays a role in substrate coordination in the binding site, but in contrast with HP2, it does not change its conformation during the transport cycle. The scaffold domain has two highly tilted helices (TM2 and TM5) along which the transport domain slides. These helices determine the minimal distance that the substrate-binding site must travel, and have been named the fixed barrier [1].

The fixed barrier elevator mechanism with one gate is likely conserved among the SLC1 family, as evidenced by recent single particle cryo-EM structures of Glt_Tk [30], and molecular dynamics simulations of Glt_Ph [7]. Fixed barrier elevators with one gate may also occur in other families of transporters, for which the number of structurally resolved states is not as large as for the SLC1 family. Transporters of the Phosphotransferase System (PTS), which are responsible for the uptake and phosphorylation of carbohydrates and other compounds such as ascorbate (reviewed in [31]) have characteristic elevator elements, such as transport and scaffold domains, HP gates, and homo-oligomer architecture. Structures of MalT [32,33] and ChbC [34] indicate that they use a fixed barrier and most likely a single gate.

ATP-binding Cassette (ABC) transporters do not use elevator-type mechanisms of transport, with the exception of the non-canonical subfamily of ECF (energy-coupling factor) transporters. ECF transporters are involved in uptake of vitamins or other micronutrients (reviewed in [11]). Two sub-types exist (Group I and II) which may differ in the mechanistic details, but the ensemble of available structural information is consistent with elevator-type behaviour in all ECF transporters. ECF transporters make use of an integral membrane subunit named the S-component that binds the transported substrate on the extracellular side of the membrane (Figure 2d). In many cases, access to the binding site is controlled by two loops, which act as gate (loop 1 and loop 3). In the bound state, with closed gate, the substrate is occluded and the S-component can “topple over” in the membrane, which brings the substrate-binding site to the cytoplasm. In the toppled state the same loops 1 and 3 can move to expose the binding site to the cytoplasm (similar to a one-gate elevator). The S-component may be considered as the equivalent of the transport domain, whereas the counterpart of the scaffold domain is a second integral membrane subunit, named EcfT or T-component (Figure 2d). The use of separate subunits instead of linked domains provides extra functionality, as dissociation and association are part of the transport cycle in some ECF transporters [35]. The EcfT subunit is additionally associated with ATPase subunits for allosteric coupling of the conformational changes to ATP binding and hydrolysis, which are the hallmark of ABC transporters.

The concentrative nucleoside transporter CNT (a member of the SLC28 family) is a homotrimer [36], with each monomer subdivided into a transport domain (TM1–2, TM4–5, TM7–8 and HP1, HP2) and a scaffold domain (TM3 and TM6). In this case, the binding site for the nucleoside is located at the interface between scaffold and transport domains, but most of the interactions with the substrate come from the residues in the transport domain. CNT uses different gates on the extra- and intracellular sides [36] (Figure 2b). Comparison of structures of CNT in outward-open and outward-closed states revealed different conformations of TM4b, suggesting that this half-TM is an extracellular gate. On the intracellular side, HP1b is the movable element, which gates access to the binding site. The transitions between the outward- and inward-facing states involve a ∼8 Å translocation of the substrate-binding site (perpendicular to the membrane plane), in which it passes a fixed barrier formed by TM3 and TM6 of the scaffold domain. CNT is the only elevator transporter, for which multiple intermediate conformations, where the position of transport domain is distributed between the inward and outward states, have been resolved structurally.

It is possible that the location of the binding site between two domains in CNT necessitates the use of two gates, whereas an occluded binding site within the transport domain, as found in SLC1 transporters, may allow the use of a single gate. Most of the transporters with proposed elevator-like transport mechanisms have substrate-binding sites positioned at the interface of two domains (Table 1). Transporters of AbgT family [37] and the structurally related Na⁺/succinate transporter VcINDY [14] (DASS family), the Na⁺/citrate transporter SeCitS [38] (2HCT family), anion exchanger 1 (AE1), a member of SLC4 family [39] and the structurally related uracil:proton symporter UraA [40,41] from SLC23 family (seven transmembrane segment inverted repeat [42]), and bicarbonate transporter BicA [43] of the SLC26 family are organized in two domains (transport and scaffold) and bind the substrate at the domain interface. All of these proteins may use an elevator mechanism with fixed barrier and two gates [37], but additional structural characterization is needed to classify the gating mechanism of these transporters.

The bile acid transporter ASBT, and structurally related sodium-proton antiporters have 10 and 13 transmembrane helices respectively, with a transport domain (also called core domain) consisting of TM3–5, TM8–10 in ASBT (TM3–5, TM10–12 in sodium-proton antiporters), and a scaffold domain (TM1–2, TM6–7 in ASBT or TM1–2, TM7–9 in sodium-proton antiporters). Despite the movement of the substrate-binding site across the membrane during sliding of the transport domain relative to the scaffold (the hallmark of the elevator mechanism), ASBT does not have a fixed barrier (Figure 2c). Thus, this transporter combines an elevator movement with a moving barrier, which is a typical feature of non-elevator-type mechanisms (Figure 1) [44]. Unlike most other elevator transporters, ASBT and the related sodium-proton antiporters NapA and NhaA do not have helical hairpins. Possibly HPs are suitable for gating when a fixed barrier is used, but are not required for moving barrier elevators (Figure 2c).

ASBT is exceptional among elevator-type transporters because it is a monomeric protein. Another monomeric transporter, for which an elevator mechanism has been postulated, is CcdA [17]. CcdA is the smallest elevator-type protein and is involved in the transport of reducing equivalents from the cytoplasm to the extracellular environment, by using a pair of cysteine residues that can be oxidized to form a disulfide bridge. The protein consists of six transmembrane helices, which are organized in two inverted structural repeats [17]. Comparison of the outward-facing conformation, solved using NMR spectroscopy, and inward-facing conformation, which was computationally modelled using information from the inverted topology, showed that protein forms a unique “O-shaped scaffold” in the centre of which TM1 and TM4 may move as an elevator between inward- and outward-facing states with the active-site cysteines bridging a distance of 12 Å [17]. Structural information on CcdA is still very limited, and further work is required to confirm the elevator mechanism.

It has been noticed that the TMs of the scaffold domains of many elevator-type transporters are shorter than those in transport domains, and often highly tilted [1]. As a consequence, the distance between the external and internal aqueous solutions is substantially smaller than the thickness of the bulk bilayer. Such thinning not only reduces the extent of elevator movement required to transfer the substrate between the aqueous solutions on either side of the membrane, but may also induce membrane distortion, which in turn could facilitate the sliding movement of the transport domain. Molecular dynamic simulations of ECF transporters in a lipid bilayer predict possible membrane distortion near the EcfT scaffold, which might facilitate toppling of the S-component when it is near the scaffold [11,45]. Recent MD simulations of a lipid bilayer around Glt_Ph show different extents of membrane deformation depending on the position of the transport domain [46] (Figure 4a). Protomers of Glt_Ph in the outward-facing state induce very little local membrane curvature [46], but the lipid bilayer strongly bends around protomers in the inward-facing state. The energetic penalty of such deformation may be balanced by specific protein–lipid interactions.

Most structures of elevator-type transporters have been determined in the absence of a lipid bilayer, using detergent-solubilized proteins, which precludes accurate analysis of the protein–lipid interface. Nonetheless, these structures can provide indications of specific lipid-binding sites (Figure 4). For example, many non-protein densities were found in structures of ASCT2 determined by single particle cryo-electron microscopy (Figure 4b). These densities likely correspond to phospholipid molecules or cholesterol, although unambiguous identification was not possible at the attained resolution. The observed densities were located around the entire perimeter of the scaffold domain, also in the space between transport and scaffold domains, and close to the substrate binding site [28,29]. Lipids binding at these positions could be important for protein stability and might allosterically affect protein activity. A crystal structure of EAAT1 in the presence of the allosteric inhibitor UCPH101 demonstrated that the inhibitor's binding site is located between transport and scaffold domains [47], exactly where a putative cholesterol molecule was observed in ASCT2 [27–29] (Figure 4c). Also in other families of elevator-type transporters, lipids were found to intercalate between the scaffold and transport domains [38,48]. These observations indicate that specific lipid–protein interactions might affect elevator-like movements of the transporter, and that lipid-binding sites may be targeted for drug design.

In only very few cases have the effects of the lipid environment been studied experimentally. In Glt_Ph the relation between lipid composition and transport activity was studied in proteoliposomes. The activity of Glt_Ph was higher in liposomes containing the non-bilayer lipid Phosphatidylethanolamine (PE), than in liposomes composed of Phosphatidylcholine (PC) [49]. This effect may be caused by specific interactions between the protein and lipid headgroups, or by colligative properties of the bilayer such as lipid disorder, both of which could affect the elevator-type movements. For ASCT2, glutamine uptake activity in proteoliposomes was enhanced by the presence of cholesterol [29], but again it has not been established whether this effect is due to binding of cholesterol at specific sites, or to colligative effects such as thickness or fluidity. Lipid interactions are also essential for dimer stability of NhaA, which falls apart to monomers in the presence of high detergent concentrations, but is assembled back if cardiolipin is added [50]. In vivo, allosteric modulation by lipid molecules has been observed in Xenopus oocytes expressing EAAT4 that displayed increased glutamate-induced currents when arachidonic acid was added [51]. The presence of cholesterol was found to be crucial for functioning and localization of EAAT2 [52].

The above examples show that lipids may affect protein function directly via interactions with amino acid residues, which could accelerate or slow down transport domain movements or stabilize the scaffold domain in the membrane. In addition, colligative bilayer properties are likely to affect the functioning of elevator-type transporters, because the lipid–protein interface must rearrange substantially during transport. Finally, also the domain structure of the proteins may affect the bilayer morphology, and consequently elevator dynamics.

1. Importance of the field. Since the first description of an elevator-type transport mechanism for Glt_Ph over a decade ago [6], a variety of protein folds have emerged that support elevator movements, not only in secondary active transporters but also in different transporter classes (Table 1). Many of these transporters are potential targets in pharmacological studies and understanding of their transport and gating mechanisms might help with the development of new drugs.

2. A summary of the current thinking. In elevator-type transport mechanisms, one protein domain brings the substrate-binding site from one side of the membrane to the other by sliding through the lipid bilayer. The extent of the elevator movement, ranging from 21 Å in Glt_Tk to 7.5 Å in ASBT, and number of gating elements (one or two) vary between different proteins (Table 1).

3. Future directions. Local deformations of the lipid bilayer near elevator-type transporters, which were observed in MD simulations [46], can be studied experimentally by single particle cryo-electron microscopy, using transporters reconstituted in lipid environment [30], similar to what has been done for the lipid scramblase TMEM16 [53]. Also systematic analysis of the relationship between lipid composition, transport activity and dynamics (for instance by single molecule FRET methods [18,54]) will shed further light on the interplay between bilayer and protein. The gating behaviour might affect the order of binding and release of coupled ions and a substrate, and steady state and pre-steady state kinetic measurements may allow insight in the consequences of using one or two gates [55–61].

The authors declare that there are no competing interests associated with the manuscript.

Open access for this article was enabled by the participation of University of Groningen in an all-inclusive Read & Publish pilot with Portland Press and the Biochemical Society.

A.A.G. and D.J.S. wrote the manuscript and prepared the figures.

This work was supported by the Netherlands Organisation for Scientific Research (NWO).

2HCT

2-hydroxycarboxylate transporters

ABC

ATP-binding cassette

AbgT

p-aminobenzoyl-glutamate transporter

AE1

Anion Exchanger 1

ASBT_NM

Neisseria meningitidis apical sodium-dependent bile acid transporter

ASBT_Yf

Yersinia frederiksenii apical sodium-dependent bile acid transporter

ASCT

Alanine Serine Cysteine Transporter

bcChbC

Bacillus cereus chitobiose transporter

bcMalT

Bacillus cereus maltose transporter

BicA

bicarbonate transporter

Bor1

boron exporter 1

CNT_NW

Neisseria wadworthii concentrative nucleoside transporter

Cryo-EM

cryo-electron microscopy

DASS

divalent anion/Na⁺ symporter

EAAT

Excitatory Amino Acid Transporter

ECF ABC

ECF-type (type III) ABC importers

ECF

Energy Coupling Factor

ECF-FolT

Energy Coupling Factor folate transporter

EcNhaA

Escherichia coli Na⁺/H⁺ antiporter

ecUlaA

Escherichia coli ascorbate transporter (‘utilization of l-ascorbate’)

FRET

Förster Resonance Energy Transfer

Glt_Ph

Pyrococcus horikoshii glutamate transporter homologue

Glt_Tk

Thermococcus kadakarensis glutamate transporter homologue

GLUT5

fructose transporter

HP

helical hairpin

KpCitS

Klebsiella pneumonia sodium-ion dependent citrate transporter

LeuT

leucine transporter

LysE

L-lysine exporter

MD

molecular dynamics

MjNhaP1

Methanococcus jannaschii Na⁺/H⁺ antiporter

MtrF

antibiotic exporter (Multiple Transferable Resistance)

NMR

Nuclear Magnetic Resonance

PaNhaP

Pyrococcus abyssi Na⁺/H⁺ antiporter

PC

phosphatidylcholine

PDB

Protein Data Bank

PE

phosphatidylethanolamine

pmUlaA

Pasteurella multocida ascorbate transporter (‘utilization of l-ascorbate’)

PTS

phosphotransferase system

SeCitS

Salmonella enterica sodium-ion dependent citrate transporter

SLC26Dg

Deinococcus geothermalis fumarate symporter

TM

transmembrane

TMEM16

lipid scramblase (TransMEMbrane protein)

TtCcdA

Thermus thermophilus membrane electron transporter

TtNapA

Thermus thermophiles Na⁺/H⁺ antiporter

UapA

purine/H⁺ symporter

UCPH101

2-amino-4-(4-methoxyphenyl)-7-(naphthalen-1-yl)-5-oxo-5,6,7,8-tetrahydro-4H-chromene-3-carbonitrile

UraA

uracil:proton symporter

vcCNT

Vibrio cholera concentrative nucleoside transporter

VcINDY

Vibrio cholera Na⁺/succinate transporter (‘I'm not dead yet’)

X-ray

X-ray crystallography

YdaH

antibiotic exporter

SLC

solute carrier

SLC26Dg

Deinococcus geothermalis fumarate symporter

TM

transmembrane

TMEM16

lipid scramblase (TransMEMbrane protein)

TtCcdA

Thermus thermophilus membrane electron transporter

TtNapA

Thermus thermophiles Na⁺/H⁺ antiporter

UapA

purine/H⁺ symporter

UCPH101

2-amino-4-(4-methoxyphenyl)-7-(naphthalen-1-yl)-5-oxo-5,6,7,8-tetrahydro-4H-chromene-3-carbonitrile

UraA

uracil:proton symporter

vcCNT

Vibrio cholera concentrative nucleoside transporter

VcINDY

Vibrio cholera Na⁺/succinate transporter (‘I'm not dead yet’)

X-ray

X-ray crystallography

YdaH

antibiotic exporter