If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Program in Chemical Biology, University of Michigan, Ann Arbor, Michigan, USAProgram in Biophysics, University of Michigan, Ann Arbor, Michigan, USADepartment of Molecular, Cellular, and Developmental Biology, University of Michigan, Ann Arbor, Michigan, USA
The small multidrug resistance (SMR) family is composed of widespread microbial membrane proteins that fulfill different transport functions. Four functional SMR subtypes have been identified, which variously transport the small, charged metabolite guanidinium, bulky hydrophobic drugs and antiseptics, polyamines, and glycolipids across the membrane bilayer. The transporters possess a minimalist architecture, with ∼100-residue subunits that require assembly into homodimers or heterodimers for transport. In part because of their simple construction, the SMRs are a tractable system for biochemical and biophysical analysis. Studies of SMR transporters over the last 25 years have yielded deep insights for diverse fields, including membrane protein topology and evolution, mechanisms of membrane transport, and bacterial multidrug resistance. Here, we review recent advances in understanding the structures and functions of SMR transporters. New molecular structures of SMRs representing two of the four functional subtypes reveal the conserved structural features that have permitted the emergence of disparate substrate transport functions in the SMR family and illuminate structural similarities with a distantly related membrane transporter family, SLC35/DMT.
From atomistic descriptions of membrane transport mechanism to global spread of multidrug resistance over the last century, small multidrug resistance (SMR) proteins have provided broad insights along multiple research fronts since the family’s discovery in the mid 1990s (
). With just four transmembrane helices and ∼100 residues, SMR proteins are among nature’s smallest membrane transport proteins, making them ideal systems for biochemical and biophysical investigation. These same properties impeded high resolution structural characterization for many years, however, since the proteins are almost entirely embedded in the membrane, too small for cryo-EM, and with little polar surface area to form crystal contacts. Recently, new high resolution crystal structures have been determined for two functionally distinct SMR subtypes (
). These complement the body of mechanistic data that have been assembled over the years and provide an opportunity to consider the molecular underpinnings of functional diversity among SMR transporters.
In general, the SMRs transport positively charged solutes across the membrane coupled to the antiport of protons (Fig. 1A). The resting membrane potential and pH gradient of most bacteria implies that they typically function in the active efflux of substrates. Four major functional subtypes have been described within the SMR family, and according to our bioinformatic analysis described later in this review, at least 97% of bacterial SMR genes correspond to one of these four subtypes. The first, and likely primal, SMR subtype transports guanidinium ion, a small cationic byproduct of nitrogen metabolism, and is referred to as Gdx (guanidinium export) (
). These are also known by the name of the gene encoding them, sugE. The second subtype, which we refer to as Qac (quaternary ammonium cation), are promiscuous exporters of hydrophobic cationic compounds, including quaternary ammonium antiseptics like benzalkonium and cetyltrimethylammonium (also known as cetrimonium), and polyaromatic cationic biocides like methyl viologen (also known as paraquat), acriflavine, and ethidium (
), and this subtype includes the well-studied multidrug exporter from Escherichia coli, EmrE. Associated gene names for the Qac transporters include emrE, ebrA/ebrB, qacE, qacG, qacH, and others. The third subtype (gene name mdtI/mdtJ) has been implicated in the transport of small polyamine metabolites like spermidine and putrescine (
). In this review, we will first describe unique topological considerations shared by all four SMR subtypes, then analyze the occurrence and distribution of the different SMR subtypes among bacterial genomes. We will review recent advances in our understanding of each SMR subtype, with particular emphasis on recent high resolution structures (Table 1), and finally, analyze structural homology between the SMRs and a distantly related family of transporters, SLC35 (also known as Drug/Metabolite Transport (DMT)).
Table 1Structural data and models available for SMR transporters Gdx-Clo and EmrE
In general, bacterial membrane proteins are inserted into the membrane according to the ‘positive inside rule’, in which the cytoplasmic face of the protein has an excess of positively charged residues arginine and lysine relative to the periplasmic face (
). Dual topology proteins lack the typical biased charge distribution and are thus inserted into the membrane in both inward- and outward-facing orientations (Fig. 1B, left), where they can oligomerize with antiparallel subunits (
In addition to dual topology homodimers, there are also numerous examples of SMR gene duplications that have given rise to co-expressed genes within a single operon that assemble as obligate heterodimers (
), the assembly may be an evolutionary antecedent to an architecture that is extremely common among membrane transport proteins, the inverted repeat (Fig. 1B, right), in which a single protein possesses structurally homologous domains arranged antiparallel with respect to each other (
). Alternatively, the fusion of 4-transmembrane (TM) dual topology proteins might simply be an evolutionarily rare event, since this process requires addition of a transmembrane linker helix to connect the N and C termini of the two monomers (
To gauge the distribution of SMR genes across diverse microbes, we evaluated bacterial genomes from the Joint Genome Institute’s curated set of ∼1000 Genomic Encyclopedia of Bacteria and Archaea (GEBA) genomes (
). This set of genomes was selected to reduce sampling bias and maximize phylogenetic diversity in microbial sequences used for evolutionary studies. Available sequencing data tend to be biased toward pathogenic isolates (
) using a profile Hidden Markov Model (profile HMM) constructed for the SMR family (pfam 00893). Profile HMMs for each subtype (Gdx, Qac, polyamine transport, and lipid transport) were constructed from functionally annotated clusters in a sequence similarity network of reference SMR proteins (
), and SMR sequences were assigned to the subtype that corresponded to the lowest e-value calculated by HMMR. SMR sequences were annotated “other” if the e-value was >10−20. Consensus sequences for each SMR subtype are shown in Figure 2, A and B, and sequence information and annotations for individual SMR proteins from this set of genomes is available for download from the Deep Blue Data repository hosted by the University of Michigan with unique identifier doi.org/10.7302/0ynd-b343.
Approximately 2/3 of the bacterial genomes from the GEBA set have at least one gene encoding an SMR protein, and ∼1/3 of the GEBA genomes encode two or more SMR genes (Fig. 3A). This count of genomes with multiple SMR genes reflects both paired SMR genes that encode heterodimers, as well as genomes with more than one SMR functional subtype. The majority of SMR genes have no other SMR gene within 100 base pairs, suggesting they are expressed independently. As expected for dual topology proteins (
), the Arg/Lys bias distribution for these genetic singletons is centered at 0, and only 3% encode protomers with an Arg/Lys bias greater than ±2 (Fig. 3B). SMR genes are also found as adjacent gene pairs. In our dataset, >95% of adjacent gene pairs encode subunits with opposite Arg/Lys biases. The Arg/Lys distributions are centered around +4 and −4 for inward- and outward-facing protomers, respectively, which assemble to form heterodimeric transporters (Fig. 3B). Approximately 10% of Qac transporters and ∼20% of Gdx transporters are encoded by such paired genes, as are all polyamine transporters (Fig. 3C). In the GEBA genome set, all the SMR lipid transporters are encoded by singleton genes, although functional pairs have been identified in some bacteria (
Most bacterial phyla possess genes encoding SMRs. The most prevalent SMRs are Gdx, which are found in about ∼50% of all bacterial genomes, including 80% of Actinobacteria, half of Proteobacteria and Bacteroides, and ∼30% of Firmicutes (Fig. 3, D and E). Qac genes are also frequent, found in ∼25% of bacterial genomes overall, including ∼50% of Proteobacteria, ∼30% of Actinobacteria, and ∼25% of Firmicutes. Many species possess both Gdx and Qac transporters. The lipid and polyamine transporters are less common and found mainly in Proteobacteria, where they are found in <10% of species (Fig. 3F). Thus, the SMR transporters are widespread among bacteria and dual topology Gdx and Qac transporters are the predominant SMR variants.
Guanidinium exporters (Gdx)
Although they are the most common SMRs encoded in bacterial genomes, the Gdx transporters were also the last to be functionally annotated. The proteins were originally reported to play a role in activity of the chaperone GroEL and called SUG (Suppressor of GroEL mutations) (
). The subtype appeared to contribute to low levels of resistance to a narrow subset of drugs, but the activity was not robust, and the proteins from this subtype remained poorly characterized until their physiological role in export of guanidinium (Gdm+) was established (
). However, electrophysiological transport experiments show that the Gdx are not exquisitely selective for Gdm+ either—although Gdx proteins strictly exclude guanidinyl metabolites with polar substituents, like arginine, they transport guanidinyl compounds with single hydrophobic substitutions at WT-like levels (
). Many of the structural features observed for this homolog had been proposed for Qac protein EmrE based on prior biophysical and biochemical experiments, establishing common structural attributes of the SMR family. Gdx-Clo possesses the expected antiparallel topology, and the two subunits assemble as an asymmetric homodimer with an aqueous cavity opened to one side of the membrane (Fig. 4A). Dimerization is mediated primarily by TM helix 4. The extramembrane loops also form extensive hydrogen bonded cross-subunit interactions to seal the closed side of the transporter. Each subunit of the dimer is composed of two discrete lobes delineated by a conserved Gly-Ile-Gly motif that acts as a fulcrum in TM helix 3 (TM3). The subunits differ according to a ∼35° rotation between N- and C-terminal lobes that stems from a difference in the angle of the Gly-Ile-Gly kink (Fig. 4B). The outward facing to inward facing conformational transition involves a structural swap between the two subunits, each changing the degree of rotation between the N- and C-terminal lobes. As a result, the inward- and outward-open conformations are 2-fold symmetric to each other, and the structural swap opens an identical, symmetry-related aqueous cavity on the opposite side of the membrane (Fig. 4A). This elegant mechanism for the alternating access of the substrate-binding site—a prototype for the “rocker-switch” mechanism used by many other transporters—was first predicted and demonstrated for EmrE (
), and the same TM3 fulcrum motif (Gly-hydrophobic-Gly or GXG) is conserved in the Gdx, Qac, and polyamine subtypes (Fig. 2).
The substrate-binding site is located at the bottom of the aqueous cavity, with the positively charged Gdm+ situated between a pair of negatively charged glutamates, E13A and E13B, one contributed by each subunit. These “central glutamates” are conserved and essential in the Gdxs, the Qacs, and the polyamine transporters (
). In addition to binding positively charged substrate, the central glutamates are protonatable at physiological pH and carry protons across the membrane during the opposing leg of the antiport cycle (
). This common binding site for the small molecule substrate and the two antiported protons favors alternating binding site occupancy by the substrates and sets the 2:1 H+:Gdm+ stoichiometry measured for Gdx (
). Y59A, in contrast, points away from the substrate-binding pocket and into the aqueous vestibule. The divergent poses of Y59 presented a mechanistic proposal for conformational change by the SMR transporters (
) whereby the rotameric switch of Y59A from the aqueous vestibule toward E13B displaces the substrate Gdm+ from its interaction with E13B. The Gdm+, in turn, engages with E13A, displacing Y59B, which undergoes the converse rotameric switch, away from the central glutamates. This “tyrosine switch” has been proposed to trigger the global conformational swap that opens an aqueous cavity to the other side of the membrane, where Y59B ultimately rests (Fig. 4D) (
), suggesting that the tyrosine switch is fundamental to transport by the SMRs.
In addition to the central glutamates and the tyrosine switch, the structures of Gdx-Clo revealed a third structural feature that is likely to be conserved among other SMR subtypes, the membrane portal (
). This portal is defined by TM2A and TM2B, which form one side of the binding pocket and splay apart on the open side of the transporter. The gap between these helices is lined by hydrophobic sidechains and could, in principle, permit substrate access between the aqueous substrate-binding site and the membrane interior (Fig. 4E). In EmrE, spectroscopic experiments lead to the suggestion that the hydrophobic residues lining this portal act as a gate that permits access for the lipophilic substrates to the binding site (
), permitting hydrophobic or amphipathic substrates to diffuse between, or have simultaneous access to, the membrane and the substrate-binding pocket.
The significance of the portal for the Gdx subtype is less readily apparent, however, since the physiological substrate Gdm+ is small and hydrophilic and would be expected to access the binding pocket directly from aqueous solution. Nonetheless, structures of Gdx-Clo with phenyl-bound and octylGdm+-bound showed that these non-natural substrates utilize the membrane portal to accommodate their hydrophobic substituents, while their guanidinyl headgroups bind between the central glutamates in the binding pocket, similar to Gdm+ (
). Should a compound such as arginine or agmatine enter the binding site in the same orientation, its polar tail would likewise be positioned to extend from the binding pocket through the membrane portal. But the hydrophobic membrane interior would not favorably interact with the polar substituents, and thus, the membrane itself could contribute to selectivity against natural guanidinylated metabolites (
) that could be mitigated by a Gdx exporter. Examples of such common biocides include the agricultural antifungal dodine (decylGdm+) and pharmaceuticals like metformin, which is excreted into wastewater, where it is slow to degrade and accumulates to levels of environmental concern (
), this SMR subtype confers resistance to the quaternary ammonium compounds used as common hospital and household antiseptics. These antimicrobial agents were introduced in the 1930s, and evolutionary analysis suggests that it was around this time that the immediate ancestor of the clinically important vector for multidrug resistance, the class I integron, emerged (
). This ancestral class I integron likely consisted of an integron/integrase sequence to capture drug resistance genes, a transposable element to facilitate its spread among microbial populations, and a single resistance gene: an SMR transporter of the Qac subtype (
). Sequence analysis suggests that Qac SMRs have been dynamically associated with these and other drug resistance gene arrays over the last hundred years, gained and lost multiple times as these elements have spread among both pathogenic and environmental bacteria (
). Today, Qac transporters remain adaptive to subinhibitory concentrations of quaternary ammonium antiseptics found in wastewater and surface runoff and remain among the most common genes isolated from human-adjacent environments (
). By conferring this selective advantage against ubiquitous environmental biocides, the Qac SMRs coselect for other resistance genes in the cassettes that provide resistance against more potent clinical antibiotics (
), contributing to the continued spread of multidrug resistance.
Meanwhile, the E. coli variant, EmrE, has become one of the best studied bacterial multidrug exporters over the last 25 years. EmrE was originally shown to transport a variety of polyaromatic, cationic antimicrobial compounds (
) (Fig. 5A). Early, low resolution electron microscopy (EM) of 2D crystals demonstrated the unusual antiparallel architecture and established an elementary understanding of the helical connectivity and protein fold (
). Although high resolution structural information lagged, biochemical and biophysical studies provided a detailed molecular picture of the protein and its transport cycle. As a reference, we provide a summary of the scanning mutagenesis studies that have been performed for EmrE in Table 2. Although too extensive to discuss individually here (see reference (
These EmrE crystal structures were determined with structurally diverse substrates bound, including planar polyaromatics and tetrahedral quaternary phosphoniums and ammoniums (Table 1). The electron density determined using X-ray crystallography shows high correspondence with previous low resolution EM data obtained for EmrE in lipid bilayers (
), implying that the crystal structures represent a native, low energy conformation. The molecular details of the crystal structures also agree with specific predictions from spectroscopic and mutagenic studies (
) and will only be briefly summarized here. In the NMR models, the subunits of the dimer are arranged more parallel with respect to each other, and the loops are unpacked, permitting aqueous access to the binding site from both sides of the membrane. The hydrogen bond network in the binding site is also rearranged with respect to the recent crystal structures. It is possible that the NMR models represent functional intermediate states, such as those suggested by prior EPR experiments (
), and that the differences between the crystallography or EM density (which largely agree with each other) and NMR models and are due to differences in the experimental conditions. However, it is also important to note that the NMR models are based on relatively few experimental measurements of distances between backbone atoms and bound substrate and that the models (including sidechain placement) are generated computationally, based on these distance restraints and backbone chemical shift measurements. Since the recent crystal structures are in the best agreement with the EM maps and are the only structural models of EmrE with experimental electron density that supports sidechain placement without further computational modeling, we will focus our analysis on these crystal structures (
) (Protein Data Bank codes 3B5D, 3B62, 3B61) are low quality and incomplete (only Cα atoms are modeled) and experimental maps are unavailable in the Protein Data Bank. These prior structures are broadly considered inadequate for molecular inference (
Both the recent structures and the long history of functional data show that EmrE has many major mechanistic features in common with Gdx-Clo, including the central glutamates in the binding pocket that contribute to alternate binding of drug and protons (
). Based on information from the crystal structures of EmrE with benzyltrimethylammonium (the headgroup of the common household antiseptic benzalkonium) and Gdx-Clo with octylGdm+, a model for benzalkonium binding was constructed (
), illustrating how the membrane portals of both the Gdx and Qac subtypes can be exploited to bind substrates with extended alkyl substituents (Fig. 5B).
Given the structural similarities and high sequence conservation, why then can EmrE bind and transport a much more diverse range of substrates than Gdx-Clo? The crystal structures in complex with substrates suggest that, despite shared sequences, EmrE and Gdx-Clo also have important structural differences. Although many of the binding pocket residues that serve as hydrogen bond donors or acceptors are conserved in EmrE, they do not form as an extensive an inter-residue H-bond network as is observed in the binding site of Gdx-Clo. Peripheral binding site residues W63 and S42 are both present, but they do not H-bond with other EmrE sidechains (
). Many potential H-bond interactions have either poor geometry or longer interaction distances than ideal (Fig. 5C). As a result, the central glutamates and conserved binding site Trp (W63 in EmrE) are comparatively unconstrained and able to adopt different rotamers in the presence of different substrates. EmrE’s central glutamates move closer together or farther apart to accommodate flat planar substrates or bulky quaternary compounds (Fig. 5D). At the same time, the binding site Trp rotates over ∼80° to stack against aromatic rings of different substrates bound in different poses (
These structural observations are in accord with NMR and computational studies that suggest that EmrE possesses an unusual degree of structural plasticity that might contribute to substrate polyspecificity (
). However, it should be emphasized that the observed structural perturbations are limited to the sidechains. Larger conformational changes involving the backbone are not necessary to explain the binding of diverse substrates to EmrE, and such perturbations are not observed in any of the five drug-bound crystal structures (
). Likewise, the low pH, proton-bound crystal structures of EmrE (as well as Gdx-Clo) do not exhibit major structural differences relative to the substrate-bound structures, with only local changes in the position of the central glutamates (
). Thus, the doubly protonated state might not exist in a single predominant stable conformation and the reported low pH crystal structure may capture only one species in this ensemble.
Although the crystal structures provided essential insight into the molecular basis for substrate binding by EmrE, the transport of these disparate substrates poses additional problems, requiring the choreography of substrate binding and dissociation, conformational exchange, and proton antiport. Spectroscopic techniques, including NMR and EPR, have been integral to fleshing out a dynamic picture of EmrE. NMR studies in lipid membranes and in bicelles have identified various mutations that slow or eliminate conformational change but preserve substrate binding, isolating residues involved in the first process and not the second (
). NMR measurements have shown that the kinetic behavior of EmrE, including the rate of conformational exchange, differs depending on the substrate, demonstrating that different substrates have different affinities for the transition state of the conformational exchange, as they do in the ground state (
However, emerging evidence also suggests that under certain conditions, EmrE violates tenets of classic transport mechanisms. Conformational exchange of the apo (proton- and drug-free) and single proton–bound transporter have been reported (
). The conformational exchange rate of EmrE with different substrates is not tightly correlated with the rate of substrate transport, hinting that different substrates, particularly high affinity substrates, might undergo futile cycles and remain bound as the transporter transits between inward and outward open states (
) suggests scenarios in which the microscopic rate constants measured for each potential binding event and conformational transition in the transport cycle combine to reduce the stoichiometry noticeably from the 2:1 H+:substrate stoichiometry measured for Gdx (
A limitation of such free exchange transport models is that they permit potential pathways for proton leak. If the inward-to-outward facing transition of the unoccupied or singly protonated transporter is not energetically prohibitive in living bacteria, such transport cycles would contribute to the dissipation of the proton motive force. For EmrE, different mechanisms have been proposed to explain the apparent absence of detrimental leak pathways in vivo and in vitro. In one proposal, the central glutamates are electrostatically independent so that the proton cannot “hop” from the glutamate with the lower pKa to that with the higher pKa, preventing proton release from the singly protonated state after the conformational swap (
). Alternatively, it has been suggested that the proton pathway is gated by a C-terminal histidine residue that is highly conserved among the Qac subtype, which occludes the binding pocket in the absence of drug, preventing proton leak (
). It should also be mentioned that the quaternary phosphonium substrates used in these transport experiments are not encountered by bacteria outside the laboratory. Thus, the transport properties for such anthropogenic chemicals have not been optimized by purifying selection in bacterial populations over evolutionary time. While it is possible that proton slippage and deviations from ideal stoichiometry are evolved properties of the transporter to handle diverse substrates (
), it is also possible that these mechanistic features reflect nonoptimized transport of non-native compounds and that transport of native substrates (whether the drug-like molecules produced by microbes in competitive niches or yet-unknown metabolites) is more parsimonious. Native substrates of the Qac transporters have yet to be identified, however, so this remains an open question.
Polyamines, such as spermidine, putrescine, and cadaverine, play myriad roles in diverse bacteria (
). Although biochemical information is relatively limited for the polyamine transporters, mutagenesis coupled with growth assays has demonstrated that key mechanistic residues for Qac and Gdx function, including the central glutamates, the tyrosine switch, and the binding site tryptophans, are critical for function of the MdtIJ complex (
), implying that the polyamine transporters share mechanistic similarities with the more extensively characterized Gdx and Qac subtypes.
Lipid transport proteins
The most distantly related members of the SMR family are reported to act as glycolipid flippases. The SMR genes (annotated arnE and arnF in E. coli and Salmonella enterica) are found in larger biosynthetic operons that contribute to the chemical modification of lipid A in the outer membrane with 4-amino-4-deoxy-l-arabinose (L-Ara4N) (
). There are indications that the S. enterica flippase has somewhat broader specificity, including genetic complementation of flippase deletion mutants in synthetic pathways that require transport of different glycolipids (
The lipid SMRs are the only subtype that transports noncationic substrates, and these proteins often possess an asparagine in place of the central glutamate (Fig. 2). This replacement also suggests that lipid transport may not be proton coupled, since the central glutamates are also responsible for proton binding in proton-coupled SMRs. Because undecaprenyl phosphate aminoarabinose is synthesized in the cytoplasmic leaflet (
), transport of the lipid down its concentration gradient to the periplasmic leaflet via facilitated diffusion might be sufficient. However, this has not been established experimentally. A membrane portal similar to that observed in structures of EmrE and Gdx-Clo would be an obviously useful feature for lipid transport, permitting the substrate’s prenyl tail access to the membrane while the polar headgroup is ensconced within its protein-binding site, a familiar feature of other lipid flippases (
). Indeed, in this SMR subtype, the hydrophobic character of the TM2 residues is retained, and we therefore conjecture that the membrane portal is conserved as well. The tyrosine switch is also conserved in the lipid SMRs, despite their overall low sequence similarity with other SMR subtypes.
Structural relationship between the SMR and the SLC35/DMT folds
Although other dual topology transporter families have representatives with inverted repeat topology (
). However, structural analysis of transporters from the DMT superfamily, which possess the SLC35/DMT fold, suggests that the SMR fold might nonetheless have been preserved by evolution as an inverted repeat. Structures of transporters with the SLC35/DMT fold, including a bacterial aromatic amino acid exporter, a protozoan drug exporter, and eukaryotic organellar sugar/nucleotide transporters (
), possess striking structural homology to the bacterial SMRs (TM helix RMSD 2.6–3.8 Å) despite sharing no sequence similarity (Fig. 6A). DMT and SMR transporters have previously been proposed to be evolutionarily related (
). However, the SLC35/DMT and SMR structures also bear two notable differences. First, each domain of the SLC35 inverted repeat is composed of five TM helices. In the 3D structure, this pair of inserted helices pack against the membrane portal defined by helices 2A and 2B of the SMR transporters, sealing the portal and eliminating access to the substrate-binding site from the membrane (
). The second major difference between the SMR and SLC35/DMT structures is in the helix connectivity. Whereas in the SMRs, each transport domain is composed of a single, independently folded monomer; in the SLC35/DMT proteins, the transport domains do not correspond simply to the N- and C-terminal halves of the protein. The first transport domain is composed of helices 1, 2, 8, 9, and 10 and the second is composed of helices 3, 4, 5, 6, and 7 (Fig. 6A).
Although the possibility that the similarities between the SMR and SLC35/DMT folds arose via convergent evolution cannot be ruled out; structural correspondence between proteins with similar functions is usually considered evidence for evolutionary relatedness (
). Moreover, we propose that the topological differences between the SMRs and SLC35/DMT folds can be plausibly explained by a divergent evolutionary pathway (Fig. 6B). This model posits an ancestral, dual topology transporter with the SMR fold and a TM helix insertion between TM helices 1 and 2. The introduction of this new TM helix would enforce a reorientation of TM helices 2, 3, and 4 relative to TM helix 1, disrupting the packing between TM helix 1 and the other TM helices in that monomer (Fig. 6B, panel 2). The 3D SMR fold could be preserved, however, by a domain swap during dimer assembly (Fig. 6B, panels 2 and 3) such that the now-inverted TM1 from the first monomer trades positions with TM1∗ in the opposite subunit and vice versa (Fig. 6B, panel 3). TM helix domain swaps have been observed in other membrane proteins (
). Finally, a subsequent duplication/fusion of the 5-helix, domain-swapped ancestral dual topology transporter would give rise to the extant SLC35/DMT fold (Fig. 6B, panel 4). Duplication/fusion of 4-TM dual topology transporters are not unprecedented (
), but this process is more common for dual topology transporters with an odd number of transmembrane helices, since the N and C termini are on the same side of the membrane and connecting them does not require introduction of a transmembrane linker (
). Although no sequence homology can be detected between the SLC35/DMT transporters and SMRs or even between symmetry-related helices of the SLC35/DMTs, lack of sequence homology is not uncommon in other families of evolutionary divergent transporters with shared folds (
The SMRs have provided a tremendously productive system for studying membrane protein evolution, transport mechanism, and microbial multidrug resistance. After 25 years of such studies, the SMR transporters finally joined the high resolution structural era in 2021. These recent crystal structures representing two of the four known functional subtypes have provided a platform for analyzing prior functional studies and understanding the structural features that contribute to substrate binding and transport for each SMR subtype: Gdm+, drugs, polyamines, and glycolipids. Moreover, structural homology with SLC35/DMT transporters suggests that, contrary to the proposal that the SMRs are unusual in having not evolved fused, inverted repeat architecture (
), it is probable that the SMR fold has indeed been preserved through this evolutionary mechanism, albeit with a helical insertion and domain swap along the way. These recent advances in understanding the molecular architecture bolster ongoing efforts to develop antimicrobials that target SMR proteins, either by inhibiting transporter assembly in order to sensitize bacteria to transported compounds (
In addition, the molecular framework described here opens the door for future integrative functional, structural, and computational studies to understand how the SMR scaffold has been tailored to transport diverse substrates as the family has evolved. Such lines of inquiry are urgently important as bacteria continue to evolve around us. The Qac and Gdx subtypes, in particular, have found new roles in human-impacted environments, conferring bacterial resistance to household antiseptics (
). Genes encoding SMR transporters are currently spreading among bacterial populations, encountering new physiological contexts and substrate transport demands, driving co-selection of co-localized antimicrobial resistance genes in environmental reservoirs (
), as the SMR transporters contribute to the ongoing story of natural selection at the human–microbe interface.
Sequences and annotation for SMR proteins identified in the bacterial GEBA genome set are available for download from the Deep Blue Data repository hosted by the University of Michigan with unique identifier doi.org/10.7302/0ynd-b343.
Conflict of interest
The authors declare that they have no competing interests with the contents of this article.
C. B. M. data curation; O. E. B., T. J. Y., and R. B. S. writing–original draft; O. E. B., T. J. Y., C. B. M., and R. B. S. writing–review & editing; O. E. B., T. J. Y., and R. B. S. visualization; R. B. S. project administration; R. B. S. funding acquisition.
Funding and additional information
This work was supported by National Institutes of Health grant R35 GM128768 to R. B. S. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
EmrE, an Escherichia coli 12-kDa multidrug transporter, exchanges toxic cations and H+ and is soluble in organic solvents.