BetP is an Na+‐coupled betaine‐specific transporter of the betaine–choline–carnitine (BCC) transporter family involved in the response to hyperosmotic stress. The crystal structure of BetP revealed an overall fold of two inverted structurally related repeats (LeuT‐fold) that BetP shares with other sequence‐unrelated Na+‐coupled symporters. Numerous structures of LeuT‐fold transporters in distinct conformational states have contributed substantially to our understanding of the alternating access mechanism of transport. Nevertheless, coupling of substrate and co‐transported ion fluxes has not been structurally corroborated to the same extent. We converted BetP by a single‐point mutation—glycine to aspartate—into an H+‐coupled choline‐specific transporter and solved the crystal structure of this mutant in complex with choline. The structure of BetP‐G153D demonstrates a new inward‐facing open conformation for BetP. Choline binding to a location close to the second, low‐affinity sodium‐binding site (Na2) of LeuT‐fold transporters is facilitated by the introduced aspartate. Our data confirm the importance of a cation‐binding site in BetP, playing a key role in a proposed molecular mechanism of Na+ and H+ coupling in BCC transporters.
The betaine–choline–carnitine (BCC) transporter family comprises secondary carriers that facilitate the transport of a wide range of substrates that bear trimethylammonium groups ([‐N+(CH3)3]), such as betaine, choline, l‐carnitine and γ‐butyrobetaine (Ziegler et al, 2010). Although these substrates resemble each other structurally, they are transported by distinct coupling mechanisms. The zwitterionic betaine is co‐transported with sodium (Farwick et al, 1995), the positively charged choline is co‐transported with protons (Lamark et al, 1991) and l‐carnitine is exchanged against γ‐butyrobetaine by a sodium‐independent antiport mechanism (Jung et al, 2002). Na+‐ and H+‐coupled BCC symporters like BetP from Corynebacterium glutamicum and BetT from Pseudomonas syringae (Chen and Beattie, 2008) or Escherichia coli (Tøndervik and Strøm, 2007) are involved in the response to osmotic stress, regulating transport of osmolytes according to the external osmolality. BCC antiporters like the l‐carnitine/γ‐butyrobetaine antiporter CaiT from E. coli do not have a role in stress response (Jung et al, 2002). BetP is the best‐characterized member of the BCCT family and the recently solved crystal structure (PDB entry 2WIT) (Ressl et al, 2009) has contributed to the understanding of osmolyte transport and stress‐induced transporter activation (Krämer and Ziegler, 2009; Ziegler et al, 2010). BetP shares the so‐called LeuT‐fold of a 10 transmembrane (TM)‐helix core formed by two inverted structurally related five TM‐helix repeats with other sequence‐unrelated transporters (Ziegler et al, 2010). Almost each individual structure of LeuT‐fold transporters revealed a different conformational state of the alternating access cycle (Abramson and Wright, 2009). Two different models, the gating model and the rocker‐switch model, describe conformational changes during transport based on the wealth of structural data (Forrest et al, 2010). Most recently, the conformations of both open states (outward facing and inward facing) were available for one single transporter, the Na+‐coupled hydantoin symporter Mhp1, revealing features of both, rocker‐switch and gating model (Shimamura et al, 2010). However, the molecular mechanisms of alternating access in LeuT‐fold transporters is still very much under debate especially the role of the first helix of the first repeat in the gating of substrate and co‐substrate/counter‐substrate‐binding site (Boudker et al, 2007). Another intriguing question is the impact of the coupling ion on the substrate coordination, and thereby on the substrate specificity. BetP is highly specific for its only known substrate to date, betaine, which is transported with an apparent Km of 3.5 μM (Rübenhagen et al, 2000). The crystal structure of BetP revealed aromatic residues from TM4 and TM8, which are conserved in all BCC transporters, coordinating the trimethylammonium group of betaine by cation–π interactions (Ressl et al, 2009). Subsequent structures reported for the BetP‐homolog l‐carnitine/γ‐butyrobetaine antiporter CaiT (Schulze et al, 2010; Tang et al, 2010) confirmed the same aromatic binding motif for the trimethylammonium groups of l‐carnitine and γ‐butyrobetaine. Similar architectures of the aromatic binding sites in BetP and CaiT suggested that the exclusive specificity towards betaine in BetP is not related to the conserved residues in TM4 and TM8 (Ziegler et al, 2010). Residues in TM3 mediate binding of substrate carboxyl groups; however, the coordination differs significantly for betaine (Ressl et al, 2009), l‐carnitine (Tang et al, 2010) and γ‐butyrobetaine (Schulze et al, 2010) reflected by the fact that residues in TM3 are only conserved for BCC transporters sharing the same substrate specificity (Ziegler et al, 2010). TM3, which corresponds to the first helix of the first repeat in BetP, carries a conserved motif of three glycines in its midsection (Ziegler et al, 2010). Here, we have investigated the role of the glycine stretch in substrate specificity and ion coupling. The flexibility of the stretch was changed systematically by point mutations. A single mutation of Gly153 to aspartate resulted in intriguing changes in substrate and co‐substrate specificity. The mutant BetP‐G153D showed choline transport not only driven by electrochemical sodium potential but also by proton motive force (pmf). Based on a 3.35‐Å crystal structure of BetP‐G153D in complex with choline, we describe a possible mechanism for the sodium and proton coupling in the BCCT family.
Alanine scanning of the glycine stretch in TM3
Multiple sequence alignment (Figure 1A) indicates that the glycine motif (Gly149‐x‐Gly151‐x‐Gly153) located in the unwound region halfway across the membrane in TM3 (Figure 1B) defines Na+‐coupled betaine symporters in the BCCT family. We investigated this motif by an alanine scanning with respect to betaine transport activation upon an osmotic shock (Figure 1C and D). Replacement of Gly151 by alanine resulted in the strongest decrease in uptake rates of [14C]‐betaine in E. coli MKH13 cells (Figure 1C) and a lack of activation at higher osmolalities. BetP‐G149A although being significantly less active compared with the wild type (WT) still shows osmo‐dependent activation. BetP‐G153A retains WT activity at low osmolality; however, the activation profile is altered. Alanine exchange of the glycine motif significantly altered betaine transport, although it did not abolish it completely. The most drastic changes occur when the middle of the stretch is affected (Gly151; Figure 1B and C). Flexion of TM3 around its midsection might be part of a mechanism that renders the substrate‐binding site accessible during conformational changes in the catalytic cycle. A similar conformational flexibility in the midsection of the first helix of the first repeat has been described previously for LeuT (Yamashita et al, 2005; Shi et al, 2008). Thus, we conclude that one important role of the glycine motif is to provide a functional conformational flexibility to TM3, which is the first helix of the first repeat in BetP in agreement with a mechanism described by the gating model (Zhao et al, 2010).
Sodium‐coupled betaine and choline transport in BetP‐G153D
In the choline‐specific BCC transporter BetT, an aspartate residue lies in a position that corresponds to Gly153 in BetP (Figure 1A), while Gly149 and Gly151 are conserved. Choline and betaine are structurally very similar, both harbouring a positively charged trimethylammonium group (Figure 2A). However, betaine is zwitterionic comprising a negatively charged carboxyl group, while choline is positively charged due to its neutral hydroxyl group. We hypothesize that the presence of a negatively charged residue in TM3 is critical for choline to bind. Therefore, we introduce the point mutation G153D into BetP. In the presence of an electrochemical Na+ potential, BetP‐G153D transports betaine (Figure 2B), albeit with reduced affinity and Vmax compared with the WT protein when measured in proteoliposomes (Table I). By assuming that betaine transport is similarly catalysed in BetP and in BetP‐G153D, the decreased affinity of the mutant form towards betaine might be caused by the decreased flexibility of the unfolded stretch in TM3. Further, BetP‐G153D also transports choline (Figure 2B), while choline transport was not detected in the WT. The transport rate for choline in BetP‐G153D was comparable to that of betaine in the WT, although the affinity of BetP‐G153D for choline in the presence of sodium was significantly lower than for betaine (Table I). One of the tryptophans contributing to the substrate‐binding site was exchanged against a tyrosine that is the corresponding residue found in BetT at the same location (W189Y in TM4 of Supplementary Figure S1A). However, this additional mutation in the aromatic binding box did not cause any further significant change in the kinetics of Na+‐coupled betaine or choline transport (Table I). We suggest that to the inserted aspartate has the key role in the coordination of choline during Na+‐coupled transport. A similar coordination of choline by hydrogen bonds with an aspartate residue was reported earlier in the choline‐binding protein ChoX of the ABC transporter ChoXWU from E. coli (Oswald et al, 2008) (Supplementary Figure S1B).
Proton‐coupled transport of BetP‐G153D
BetT, in contrast to BetP, is a proton‐coupled symporter (Lamark et al, 1991). Proton coupling and translocation is assumed to involve charged residues, which might be protonated during transport. With regard to the E. coli H+‐coupled lactose symporter LacY, approximately five charged residues have been proposed to coordinate proton translocation (Smirnova et al, 2009). BetT does not harbour any additional charged residues along the putative substrate translocation pathway when compared with BetP. Asp97 in TM3 of BetT is the only charged residue in this context and therefore the most likely to be critically involved in proton translocation by BetT.
Transport of betaine and choline in BetP WT and BetP‐G153D energized by pmf was measured in E. coli lipid proteoliposomes. In BetP‐G153D choline transport could indeed be coupled to the electrochemical proton gradient (Figure 3A) and abolished in the absence of an inwardly directed proton gradient (Figure 3B). Transport rates decreased to ∼25% in the presence of the protonophore CCCP (Figure 3B). The apparent Km for choline in BetP‐G153D decreased to 35 μM in the presence of protons (Table I). Neither BetP‐G153D nor the WT catalysed the co‐transport of betaine with protons, and the WT failed to co‐transport choline with protons, too. Due to the differences in ΔG values, when transport is driven by pmf and smf of different extent, proton‐coupled choline transport activity cannot quantitatively be compared with Na+‐coupled choline or betaine transport activity; however, they seem to be in at least a similar range. We conclude that a single‐point mutation is sufficient to convert BetP from an Na+‐coupled betaine‐specific transporter to an H+‐coupled choline‐specific transporter. In the presence of sodium, this mutant transports betaine and choline. Therefore, a single residue in TM3 determines both substrate and co‐substrate specificity in a BCC transporter, pointing towards a common molecular mechanism of Na+ and H+ coupling in this transporter family.
Structure of BetP‐G153D
The mutation G153D was introduced in the N‐terminally truncated BetPΔN29 and 3D crystals were grown in excess of choline. The structure of BetPΔN29G153D (PDB entry 3PO3) was solved to 3.35 Å (Table II), without imposition of a three‐fold non‐crystallographic symmetry to account for the conformational asymmetry of individual protomers within the trimer (Tsai et al, 2011). A choline molecule was observed in one of the three protomers within the trimer (light blue protomer in Figure 4A) located in a central binding site that is fully accessible from the cytoplasm (Figure 4B). This open inward‐facing state constitutes a new conformation of BetP in comparison to the betaine‐bound structure of BetP (PDB entry 2WIT) (Figure 4B) reported previously as occluded state (Ressl et al, 2009). In the light of the open inward‐facing state observed for BetP‐G153D, hereafter, we will refer to the betaine‐bound structure (PDB entry 2WIT) as occluded inward‐facing conformation. Superimposition of both structures (Figure 4C) supports a rigid‐body movement of the bundle domain (first two helices of each repeat) relative to the scaffold of adjacent helices (helices 3 and 4 of each repeat) that is comparable to the conformational changes described very recently for vSGLT (Supplementary Table SI) (Watanabe et al, 2010). The intracellular halves of TM3 and TM8 are displaced by 6° and 5°, respectively (Figure 4C). Side chain displacements of residues Ala144, Met144, Ile302, Gln303, Phe380, Phe384, Ile388 and Ser471 facilitate the accessibility of the binding site from the cytoplasm (Figure 4C, inset). The substrate choline binds by a hydrogen bond formed by its hydroxyl group with the carboxyl group of Asp153 (Figure 5A and B) located at the end of the unwound region in TM3. The trimethylammonium group is coordinated by the carbonyl groups of Ala148 and Met150 in TM3, by Van der Waals interactions with Trp377 and Phe380 in TM8, and by the hydroxyl group of Ser468 in TM10. Trp377 is a crucial residue in substrate transport in BetP and replacement against leucine abolished betaine transport (Ressl et al, 2009). Asp153 forms an additional hydrogen bond to Ser253 in TM5. In the inward‐facing open conformation, the positively charged trimethylammonium group of choline is located close to the position of one of the potential sodium‐binding sites (Supplementary Figure S2A). This site corresponds to the Na2 site in the outward‐facing state of LeuT (Yamashita et al, 2005) (Supplementary Figure S2B). In comparison to the betaine location in the occluded inward‐facing betaine‐bound state of BetP, choline is not located in the aromatic box formed by residues from TM4 and TM8. This shift in substrate‐binding site is reflected by subtle changes in the positions of Trp189, Trp377 and Phe380 as observed in the superimposition of inward‐facing occluded and inward‐facing open states (Figure 6).
Structural data on several transporters revealed that a few key amino‐acid residues determine substrate specificity in secondary transporters (Forrest et al, 2010). Corresponding point mutations led to a change in substrate specificity for example in the glycerol‐3‐phosphate: phosphate antiporter GlpT (Law et al, 2009) or in the creatine transporter (Dodd and Christie, 2007). BetP alters substrate specificity even by only a single‐point mutation. The crystal structure of BetP‐G153D provides evidence for how choline transport is enabled by the presence of the strategically positioned Asp153 in TM3. However, the most intriguing consequence of this mutation is the switching to another coupling ion, which allows speculating about a common mechanism of Na+ and H+ coupling in BCC transporters. The presence of a cation‐binding site that might serve for different mechanistic purposes appears to be the key parameter for a possibility of simultaneous sodium and proton coupling in transporters of the BCCT family. The BetP‐G153D structure suggests this site at a position assigned as Na2 sodium‐binding site (Supplementary Figure S2A). Na2 is conserved in five TM‐helix repeat transporters including vSGLT (Faham et al, 2008), Mhp1 (Weyand et al, 2008) and BetP (Ressl et al, 2009) and was first reported in the 1.65‐Å structure of the leucine transporter LeuT (PDB entry 2AS5) (Yamashita et al, 2005) (Supplementary Figure S2B). The Na2 sodium is situated between the 4TM‐helix bundle formed by the two first helices of each repeat (helices 1 and 6 in Figure 7A) and the hash domain comprising the third and fourth helix of each repeat (helices 3 and 8 in Figure 7A). Molecular dynamics free energy calculations (Caplan et al, 2008; Shi et al, 2008) on LeuT designated the Na2 site as a relatively weak binding site. It was shown recently that the electrostatic component provided by a cation in the Na2 site is crucial in opening the external gates in LeuT‐fold transporters (Shaffer et al, 2009). These electrostatic interactions can also be provided by fixed side chain charges. In ApcT, an H+‐coupled amino‐acid transporter also of the LeuT‐fold, a positively charged lysine (Lys158) is located at the Na2 site, and its protonation might mimic a monovalent cation (Shaffer et al, 2009).
Analogous to LeuT and ApcT, we suggest that in BetP, the binding of sodium to the Na2 site in the outward‐facing open conformation (Ce‐state in Figure 7A) triggers flexion of the periplasmic half of TM3 around its glycine‐rich extended stretch. Thereby, the aromatic box turns accessible for the substrate (Ce*‐state) by opening an extracellular gate (Figure 7A, red bar), although this is rather speculative since an outward‐facing structure of BetP is unknown. Subsequently, betaine or choline binds to the central binding site (SocCe‐state). In LeuT, another sodium ion occupies a high‐affinity sodium‐binding site (Na1) that stabilizes TM6 (Yamashita et al, 2005) and provides an electrostatic coordination for the carboxyl group of leucine. Since transport in BetP requires two sodium ions per betaine molecule, it was assumed that the carboxyl group of betaine is similarly coordinated (SocCe‐state for BetP in Figure 7A). The exact position of this putative Na1 site was not yet assigned for BetP. However for BetP‐G153D, Asp153 purveys an electrostatic component to coordinate the hydroxyl group of choline similar to the coordination observed in the periplasmic choline‐binding protein ChoX (Oswald et al, 2008) (Supplementary Figure S1B). In this scenario, choline flux is still coupled to the sodium flux and sodium stoichoimetry might even not change since the Na1 site is not affected by the mutation in TM3. The main difference is that choline might not be coordinated any longer by sodium.
In the same context, we suggest that binding of the bulky cationic ammonium group of choline to the Na2 site, which would function as a monovalent cation‐binding site in that case, opens the extracellular gate in the absence of sodium (transition of Ce‐ to SCe‐state in Figure 7B). The observed lower affinity for choline in the presence of sodium would be caused by competition between the positive charges of choline and sodium for the Na2‐binding site, suggesting that Na+ has an inhibitory effect during H+‐coupled transport. At some stage during the alternating access cycle, the trimethylammonium group of choline might be positioned in the aromatic pocket similar to what we predict for choline binding (SocCi‐state in Figure 7B) based on a homology model of BetT (Supplementary Figure S3A).
The structural and functional data presented here indicate that the key parameter in H+‐coupled choline transport is the protonation of the aspartate residue. Most likely, the pKa of Asp153 has to shift significantly in different states of the transport cycle. Based on pKa predictions (Li et al, 2005) it can be assumed that on coordination to choline, the pKa of the aspartate shifts to 4.4, therefore Asp153 will be deprotonated at pH 5.5 (transition from SCe‐ to SocCe‐state in Figure 7B). Subsequently, it might be possible that release of a proton triggers the conformational change from outward to inward‐facing conformation and the predicted pKa of Asp153 would rise to 6.5 once choline is released, resulting in protonation of the carboxyl group. The interaction between Asp153 and Ser253, which represents a link between the bundle and the hash domain, might be also important during a conformational change back to an outward‐facing state (Supplementary Figure S3B; transition of Ci‐ to Ce‐state in Figure 7B). Obviously, protonation of Asp153 has a different role in the transport cycle of BetP compared with Lys158 in ApcT, where protonation of this residue renders the central binding site accessible for the substrate (Shaffer et al, 2009).
How protons in BetP‐G153D are transported in a pathway that is initially designed for sodium remains unknown. In LacY, the binding of lactose requires prior protonation of specific residues (Smirnova et al, 2009), and substrate affinity is pH dependent; both events have been observed during H+‐coupled choline binding in BetP‐G153D. Moreover, no additional charges line the lactose pathway, and water molecules are proposed to act as a co‐factor in LacY during proton translocation by forming a transient hydronium ion (Smirnova et al, 2009). Hydronium is assumed to pass via similar pathways as the sugar in LacY. In BetP‐G153D, a similar mechanism might be possible assuming that a hydronium ion occupies similar positions as choline at different time points in the catalytic cycle (Figure 7B).
Collectively, our data suggest a rather similar mechanism for Na+ and H+ coupling in BCC transporters. In the initial step, a cation (sodium or choline) is bound in the Na2 site thereby opening an extracellular gate to render the substrate‐binding site accessible. Subsequently, a coupling ion might contribute to the generation of the substrate‐binding site directly by interacting with the substrate (Na+ binding to the Na1 site when betaine is the substrate) or, alternatively, a charged residue may mimic this event (choline coordination by aspartate). The primary requirement for H+ coupling, however, compared with Na+ coupling, is the presence of a side chain that undergoes pKa shifts after conformation‐induced interactions (Ser253–Asp153) or on interaction with the substrate (Asp153–choline), respectively. By this means, BCC transporters seem to maintain certain promiscuity towards their coupling ion in the case of a cationic substrate.
Materials and methods
E. coli DH5αmcr were used for the heterologous expression of strep‐betp. The QuikChangeTM kit (Stratagene), in combination with Pfu Turbo DNA polymerase, was used for the replacement of nucleotides in the pASK‐IBA5betP and pASK IBA7betPDeltaN29EEE44/45/46AAA plasmids (Schiller et al, 2004). Membranes were solubilized using β‐dodecyl‐maltoside, and BetP was purified by affinity chromatography via Strep‐Tactin macroprep and size exclusion chromatography (Ziegler et al, 2004; Ressl et al, 2009). Uptake of labelled betaine and choline was measured in E. coli MKH13 cells and proteoliposomes made of E. coli polar lipids (Ott et al, 2008) started by adding 15–400 μM of [14C]‐betaine or [14C]‐choline upon an osmotic shock of 600 mOsmol/kg adjusted with proline. When transport was energized by an electrochemical proton gradient, the internal buffer composed 100 mM KPi at pH 7.5 and the external buffer 100 mM KPi at pH 5.5, 50 mM MES at pH 5.5 and 10 nM valinomycin. The kinetic constants were derived by Michaelis–Menten curve fitting of the uptake rates versus substrate concentration with GraphPad Prism version 5.0c for Mac OS X (Motulsky, 1999). BetPDeltaN29/G153D was crystallized in the presence of choline. A data set to 3.35 Å was collected at ESRF‐ID29, and the crystal structure was determined by molecular replacement against the structure of Chain C of BetP (PDB entry 2WIT). The substrate was positioned when a clear positive peak in the Fo–Fc difference electron‐density map was observed in the binding site after several refinement rounds. Extended version of the methods is given as supporting information.
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
Conflict of Interest
The authors declare that they have no conflict of interest.
We thank Özkan Yildiz for support with the crystallography and model building and Lucy Forrest for helpful discussions and suggestions. Vera Ott made valuable suggestions on the BetP transport experiments. We thank the beam line scientists at the SLS PX‐II at the ESRF‐ID29, who helped during initial screening and data collection. This work was supported by the International Max‐Planck Research School, Frankfurt, and by the DFG (German Research Foundation), Collaborative Research Center 807, Transport and Communication across Biological Membranes.
Author contributions: CP performed all mutations, the activity measurements in cells and proteoliposomes, and crystallization and collected and processed the data; CK carried out the homology modelling; SR directed CP in the early X‐ray experiments and performed the initial crystallographic data analysis; SN made initial proteoliposome measurements; CP and CZ analysed the data; CZ directed the research; and CP, RK and CZ wrote the paper.
Author information: Coordinates and structure factors for BetP‐G153D with bound substrate (PDB entry 3PO3) have been deposited into the Protein Data Bank.
- Copyright © 2011 European Molecular Biology Organization