Crystal Structure of the Maltose Transporter in a Pretranslocation Intermediate State

See allHide authors and affiliations

Science  03 Jun 2011:
Vol. 332, Issue 6034, pp. 1202-1205
DOI: 10.1126/science.1200767


Adenosine triphosphate (ATP)–binding cassette (ABC) transporters convert chemical energy from ATP hydrolysis to mechanical work for substrate translocation. They function by alternating between two states, exposing the substrate-binding site to either side of the membrane. A key question that remains to be addressed is how substrates initiate the transport cycle. Using x-ray crystallography, we have captured the maltose transporter in an intermediate step between the inward- and outward-facing states. We show that interactions with substrate-loaded maltose-binding protein in the periplasm induce a partial closure of the MalK dimer in the cytoplasm. ATP binding to this conformation then promotes progression to the outward-facing state. These results, interpreted in light of biochemical and functional studies, provide a structural basis to understand allosteric communication in ABC transporters.

Active transport systems transfer substrates across a membrane against a concentration gradient by coupling to a thermodynamically “downhill” reaction. In the prevailing alternating access model, the transporter cycles between two conformations, alternately exposing a substrate-binding site to either side of the membrane (1). Crystal structures of homologous transporters in multiple conformations provide insights into the changes that mediate alternating access (28). It is becoming clear that the two-state model may be an oversimplification of active transporters; intermediate steps are likely to be involved to channel energy from internal and external sources to enable the transition between the inward- and outward-facing states.

Adenosine triphosphate (ATP)–binding cassette (ABC) transporters offer an opportunity to elucidate a detailed mechanism of active transport. Crystal structures of many isolated nucleotide-binding domains (NBDs) [reviewed in (9)], as well as full-length transporters, have been reported (2, 1016). Together, these structures show that all ABC transporters contain two cytosolic NBDs attached to two transmembrane domains (TMDs). While the various conformations adopted by these different structures indicate how rigid-body rotations of the TMDs can coincide with closing and opening of the NBD interface [reviewed in (17)], our understanding of how the presence of transport substrate is communicated to the NBDs to initiate the transport cycle remains incomplete (18).

We used the maltose transporter (MalFGK2) from Escherichia coli as a model system to analyze the molecular events that couple ATP hydrolysis to substrate translocation. The maltose transporter is an importer composed of two TM subunits, MalF and MalG, and two subunits of a cytoplasmic adenosine triphosphatase (ATPase), MalK. Like many uptake systems in Gram-negative bacteria, a periplasmic maltose-binding protein (MBP), is required to stimulate the ATPase activity of the transporter (19). In the absence of maltose, MBP exists in equilibrium between an open and closed conformation (2024), and binding of maltose stabilizes the closed conformation. Two structures of MalFGK2 were determined by x-ray crystallography (2, 3). In the absence of MBP, MalFGK2 forms an inward-facing conformation with the TM maltose-binding site exposed to the cytoplasm (3). An outward-facing conformation, crystallized in complex with open MBP and ATP, shows that closure of the NBDs of MalK is concomitant with the transfer of maltose from MBP to the TM subunits (2). These structures capture two states in the transport cycle: The inward-facing conformation represents the resting state where the transporter has a very low ATPase activity (19), and the outward-facing conformation represents a catalytic intermediate where ATP is poised for hydrolysis. Because MBP stimulates ATP hydrolysis and initiates the transport process (19), it must interact with the resting state conformation to form a “pretranslocation” (pre-T) complex that is metastable in order to advance to the outward-facing conformation in the presence of ATP (25). Here, we present the crystal structure of the initial complex formed between closed MBP and MalFGK2 (Fig. 1A). As an essential intermediate between the inward- and the outward-facing conformations, this structure suggests a mechanism by which substrate bound on the periplasmic surface influences the conformation of the NBDs at the intracellular surface.

Fig. 1

Structure of the maltose transporter in a pretranslocation conformation. (A) Ribbon representation (left) and a surface slab view (right) of the pretranslocation conformation. Maltose molecules are shown in stick model (red, O atom; gray, C atom). (B) Surface representation of MalF and MalG viewed from MBP, showing the fingerprint of MBP at the membrane surface in both the pretranslocation and outward-facing states. Atoms of MalFG within 5 Å of MBP are colored in magenta. Periplasmic loops are labeled; the MalF P2 loop is removed for clarity.

The pre-T complex was obtained by co-crystallizing MalFGK2 in the presence of maltose with either wild-type (WT) MBP or the MBP(G69C/S337C) mutant [see supporting online material (26)]. The MBP(G69C/S337C) mutant contains two cysteines that form an interdomain disulfide bond, stabilizing the closed, substrate-bound conformation (27). The two pre-T complex structures are identical, with an overall root mean square deviation (RMSD) less than 1.0 Å (fig. S1). Crystals obtained using the MBP(G69C/S337C) mutant diffracted anisotropically to a slightly higher resolution of 3.1 Å (tables S1 and S2), and thus are used here to describe the pre-T state.

The crystal structure of the pre-T complex contains five subunits: MBP in a maltose-bound, closed conformation; the two transmembrane subunits MalF and MalG encapsulating an occluded maltose-binding pocket; and a semi-open MalK dimer with no nucleotide bound (Fig. 1A). The configuration of the pre-T state lies between the inward- and outward- facing states, thus representing an intermediate in the alternating access model (fig. S2 and movies S1 and S2).

The structure of MBP is very similar to that of isolated MBP in the closed form (23) (RMSD of 0.48 Å for 370 Cα), indicating that neither interactions with MalFGK2 nor the introduced disulfide bond alter the conformation of the binding protein (fig. S3). At the membrane surface, the two lobes of MBP interact with MalF and MalG, and the circular interface is smaller than that observed in the open MBP, outward-facing structure (Fig. 1B). The total buried surface at the interface is 2438 Å2 in the pre-T complex and 2872 Å2 in the outward-facing complex (2), which also contains 15 more hydrogen bonds or salt bridges, consistent with the higher affinity shown in biochemical studies (28).

The TM region of the transporter can be divided into two units, one consisting of MalF TM1-3 and MalG TM2-6, and the other consisting of MalG TM1 and MalF TM4-8 (Fig. 1A and fig. S4). During the transition between different conformational states, these two TM units move relative to each other as rigid bodies. In addition to the maltose molecule present in the closed MBP, a second maltose molecule was identified in the TM site of the pre-T structure (Fig. 2). This TM-bound maltose is likely nonphysiological in the pre-T state, caused by the high concentration of maltose used in crystallization, but it shows that residues contacting maltose can adopt the same conformation as observed in the outward-facing structure. Access to the binding site in the pre-T state is restricted from both sides of the membrane. The periplasmic gate, similar to that in the inward-facing structure, is formed by residues at the intersection of kinked TM helices, including V230, P231 of MalG and F441, V442 of MalF (Fig. 2B and fig. S5A). However, the configuration of the cytoplasmic gate is different than in either the inward- or outward-facing structures. In contrast to the outward-facing structure, in which MalG TM 4, 5 and MalF TM 6, 7 form a tightly packed helix bundle near the membrane inner leaflet, in the pre-T state these helices are separated, except for residues T176, L221 of MalG and Y383, L429 of MalF, which create van der Waals interactions to shield the TM maltose-binding site from the cytoplasm (Fig. 2B and fig. S5B).

Fig. 2

The pretranslocation state is an intermediate conformation between the inward- and outward-facing states. (A) Cartoon representation of three different states of the transport cycle. Maltose is shown as a black ball. The open and filled circles in MalK represent empty and occupied ATP binding sites, respectively. (B) Structure of transmembrane maltose-binding site observed in the three states. For clarity, only two helices from each TM subunit are shown. Residues forming the periplasmic and cytoplasmic gates are shown in space-filling model. Maltose is shown in ball-and-stick model. (C) Surface representations of the nucleotide-binding domains (NBDs) in the three states. The RecA-like and the helical subdomains are differentiated by color shade. The D loop and the Walker A (WA) motifs are labeled. The rotation of the helical subdomain with respect to the RecA-like domain during transition from the pretranslocation state to the outward-facing conformation is indicated.

The structure of the MalK dimer resembles a pair of tweezers with a fulcrum point formed by the two regulatory domains and pincers formed by the two NBDs that open and close depending on the ATP hydrolysis state (fig. S2) (29). The outward-facing TMDs coincide with a closed MalK dimer, in which two ATP molecules are bound at the dimer interface (Fig. 2C). In the inward-facing, resting state, the MalK dimer relaxes to an open configuration, in which the NBDs are well separated from each other. In the pre-T state, the MalK dimer is semi-open, as the distance between the two NBDs is between those of the open and the closed forms (Fig. 2C and fig S2). The structure of a NBD monomer can be divided into two subdomains: a RecA-like subdomain containing the Walker A/B motifs that are found in many ATPases and a helical subdomain that is specific to ABC transporters. The relative orientations between the RecA-like and the helical subdomains are similar in the resting state and pre-T state. In contrast, transition from the pre-T state to the outward-facing state involves a 15° rotation of the helical subdomain in addition to the movement of the entire NBD (Fig. 2C). Consistently, EPR studies demonstrated that in detergent solutions and in lipid bilayers, the helical subdomain rotation occurs only in the presence of both adenylyl-imidodiphosphate (AMP-PNP) and maltose-MBP (30).

One key feature of the pre-T state is that the D loops are located at the MalK dimer interface (Figs. 2C and 3A). D165 of the D loop is highly conserved among members of the ABC transporter family, and mutations at this position substantially reduce ATPase activity (31). In the pre-T complex, D165 and A166 make H-bonds to the Walker A S38 and H192 of the opposite subunit, respectively (Fig. 3A). In the closed-dimer form of the outward-facing state, both S38 and H192 interact with the γ-phosphate of ATP (Fig. 3A). ATP binding would perturb the pre-T structure by causing local conformational changes of S38, H192 and the D loop, thereby initiating progression toward the outward-facing conformation. In contrast, in the inward-facing resting state, where MBP is absent, the Walker A/B motifs are distant from the dimer interface (Fig. 2C); thus, ATP binding would not induce such a global structural rearrangement necessary for hydrolysis.

Fig. 3

ATP binding to the pretranslocation state promotes progression to the outward-facing state. (A) Interactions between the D loop and Walker A motifs in the pretranslocation and outward-facing states. Hydrogen bonds between either D165 or A166 to the Walker A S38 and switch H192, respectively, of the opposite subunit are indicated by dashed lines. (B) Binding of AMP-PNP to the pretranslocation crystals containing cross-linked MBP or WT MBP. Shown is the ATP binding site, together with nonweighted Fo-Fc electron density maps (blue) contoured at 2.5σ with the nucleotide omitted. AMP-PNP is shown in stick model. The two MalK subunits are distinguished in red and green.

To further investigate the effect of ATP binding, we took advantage of the fact that pre-T crystals were obtained using both maltose-loaded WT MBP and the locked closed MBP mutant. After soaking with the nonhydrolysable ATP analog, AMP-PNP, both the WT and the mutant crystals diffracted anisotropically to about 3 Å resolution (tables S1 and S2); however, the resulting structures were different. AMP-PNP soaking of the WT MBP-MalFGK2 resulted in a complete conversion to the outward-facing conformation, with clear density for the entire AMP-PNP molecule (Fig. 3B and fig. S6). In contrast, crystals with the MBP(G69C/S337C) mutant remained in the pre-T conformation, with AMP-PNP bound to the Walker A motif of each MalK subunit (Fig. 3B and fig. S6). The electron-density map showed density corresponding to the adenoside, the α- and β-phosphates, and the Mg2+ ion, but not the γ-phosphate of the nucleotide (Fig. 3B), suggesting that the γ-phosphate is disordered in the semi-open MalK configuration. These data suggest that the formation of the closed NBD dimer in the outward-facing state is necessary to orient the γ-phosphate for hydrolysis. Consistently, in high-resolution structures of monomeric NBDs, although densities were observed for the γ-phosphate, its position varies among different structures (3235).

Upon closing the NBD dimer, the LSGGQ motif from the opposing subunit is located at the ATP binding site (Fig. 3B). The LSGGQ motif, also known as the signature motif, is highly conserved in ABC transporters and is essential for ATPase function (36, 37). Its role in orienting the γ-phosphate is reminiscent of the “arginine finger” in many RecA-like ATPases (38). For example, in the structure of F1-ATPase, the arginine finger (R373) from the α subunit inserts into the catalytic β subunit in a position equivalent to that of the LSGGQ motif (39) (fig. S7). In F1-ATPase, ATP binding and hydrolysis modulate the interface between the α and β subunits (40). Analogously, in ABC transporters, ATP hydrolysis is coupled to opening and closure of the NBD dimer interface, which in turn are linked to the orientations of the substrate-binding site relative to the membrane.

Electron paramagnetic resonance (EPR) studies demonstrated that in detergent solutions in the presence of MBP, AMP-PNP induces formation of the outward-facing conformation (41). Here, the same conformational changes occurred inside the crystal lattice. The structural rearrangements are substantial, including opening of the two lobes of MBP, rotations of the TM domains, and closure of the MalK dimer. Usually such large-scale conformational changes would destroy a crystal; however, in this case, both conformational states crystallize in the same triclinic space group with similar packing arrangements (fig. S8), fortuitously allowing the conversion within the crystal. Prevention of MBP opening by an introduced disulfide between its two lobes (G69C/S337C) effectively locks the crystals in the pre-T state, suggesting that the conformational changes involved in the transition are tightly coupled.

The crystal structure of the pre-T conformation extends our knowledge of how coordinated motions in the maltose transporter are communicated more than 80 Å across the membrane to allow active transport (Fig. 4). In the absence of MBP, the resting state MalFGK2 has little ATPase activity (19), indicating that there is an energetic barrier between the inward- and outward-facing conformations that cannot be overcome by ATP binding alone. EPR studies have shown that binding of both MBP and ATP are required to form the outward-facing conformation (41). The crystal structure of the pre-T state shows that MBP induces rotations of the transmembrane subunits and a partial closure of the MalK dimer, bringing two catalytic residues, the Walker A S38 and switch H192, to the dimer interface. In contrast to the resting state, ATP binding to the pre-T state drives progression to the outward-facing state, indicating that MBP binding lowers the energy barrier for closure of the MalK dimer necessary for ATP hydrolysis. MBP binds in the closed form to induce the pre-T state, thus explaining the well-known enhancement of ATP hydrolysis by maltose (19). This property of the maltose transporter and MBP emphasizes the important role of the substrate in regulating the ATPase activity by stabilizing a specific conformation of the binding protein that productively interacts with the transporter. Genetic data have shown that the requirement of MBP to activate ATP hydrolysis can be bypassed by mutations in the transmembrane subunits (42). Mapping these mutations into the structure suggests that they destabilize the resting state, equivalent to lowering the energy barrier between the inward- and outward-facing states (3).

Fig. 4

An extended alternating access model for ABC importers. When maltose diffuses into the periplasm through the outer-membrane channel maltoporin, it binds to MBP and stabilizes a closed conformation that interacts with the resting state MalFGK2. Binding of MBP to MalFGK2 brings the NBDs closer such that ATP would promote a concerted motion of MalK closure, reorientation of the TM subunits, and opening of MBP. Formation of the outward-facing conformation transfers maltose from MBP to the TM binding site and, at the same time, positions ATP at the catalytic site for hydrolysis. Although a crystal structure of the posthydrolysis state of the entire transporter is still lacking, EPR studies and analysis of the isolated NBD structures suggest that ADP is not sufficient to stabilize the closed dimer conformation (41, 43, 44). Thus, once ATP is hydrolyzed, the TMDs will likely reorient toward the cytoplasm and the substrate will be released into the cell though diffusion.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S8

Tables S1 and S2


Movies S1 and S2

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Acknowledgments: We thank A. Davidson for providing the MBP(G69C/S337C) construct and comments on this manuscript and staff at the Advanced Photon Source beamline 23-ID for assistance with data collection. We are also especially grateful for thoughtful and constructive comments from the anonymous reviewers. This work was supported by an NIH grant (GM070515 to J.C.) and a postdoctoral fellowship from the American Heart Association (to M.L.O). J.C. is an investigator of the Howard Hughes Medical Institute. Coordinates and structure factors have been deposited in the Protein Data Bank under accession nos. 3PV0, 3PUY, and 3PUZ.

Stay Connected to Science

Navigate This Article