Inducing conformational preference of the membrane protein transporter EmrE through conservative mutations

  1. Maureen Leninger
  2. Ampon Sae Her
  3. Nathaniel J Traaseth  Is a corresponding author
  1. New York University, United States

Abstract

Transporters from bacteria to humans contain inverted repeat domains thought to arise evolutionarily from the fusion of smaller membrane protein genes. Association between these domains forms the functional unit that enables transporters to adopt distinct conformations necessary for function. The small multidrug resistance (SMR) family provides an ideal system to explore the role of mutations in altering conformational preference since transporters from this family consist of antiparallel dimers that resemble the inverted repeats present in larger transporters. Here, we show using NMR spectroscopy how a single conservative mutation introduced into an SMR dimer is sufficient to change the resting conformation and function in bacteria. These results underscore the dynamic energy landscape for transporters and demonstrate how conservative mutations can influence structure and function.

https://doi.org/10.7554/eLife.48909.001

eLife digest

Cells are bound by a thin membrane layer that protects the cell’s interior from the outside environment. Within this layer are various transporter proteins that control which substances are allowed in and out of the cell. These transporters actively move substances across the membrane by loading cargo on one side of the layer, then changing their structure to release it on the other side.

Membrane transporters are typically made up of multiple repeating units. In more complex transporters, the genetic sequence for each of these structural units is fused together into a single gene that codes for the protein. It is thought that the repeated pattern evolved from smaller membrane protein genes that had duplicated and fused together. But, what are the evolutionary advantages of having more complex transporters being produced from a single, fused gene? To investigate this, Leninger, Sae Her, and Traaseth examined a simple transporter found in Escherichia coli bacteria, called EmrE, which contains two identical protein subunits that associate together to transport toxic molecules across the membrane.

Experiments revealed that changing a single amino acid (the building blocks that make up proteins) in one of the two subunits to make them minimally different from each other, dramatically modified the transporter’s structure and function. The subtle amino acid change disrupted the balance of inward- and outward-facing proteins. This altered the transporter’s ability to remove toxic chemicals from E. coli and reduced the bacteria’s resistance to drugs.

The effects of a minor change to one of the identical halves of the EmrE transporter demonstrates how sensitive membrane transporters are to mutations. Furthermore, this observation could help explain why evolution favored more complex transporters comprised of fused genes in which single amino acid changes can greatly alter how the transporter operates.

https://doi.org/10.7554/eLife.48909.002

Introduction

Membrane transport proteins catalyze the movement of ions and molecules across the membrane by binding substrates on one side of the bilayer and undergoing conformational changes (Jardetzky, 1966; Zhang et al., 2016). Structural and functional experiments have shown support for the alternating access model in which a transporter samples at least two different conformations that expose the substrate to the inside and outside environment of the cell membrane. The presence of inverted structural repeats within a single polypeptide chain is widespread in membrane protein transporters (Shimizu et al., 2004; Forrest and biology, 2013) and has been proposed as a mechanism to enable alternating access exchange (Forrest et al., 2008). Proteins with structural repeats are thought to arise evolutionarily from the fusion of smaller membrane protein genes, such as three or four transmembrane (TM) domain transporters that associate into oligomers to achieve the functional state (Xu et al., 2014; Bay and Turner, 2009). One class of these proteins is the small multidrug resistance (SMR) family (Schuldiner, 2014; Schuldiner, 2009; Paulsen et al., 1996) found in bacteria and archaea that contain four TM domains and assemble into antiparallel homodimers or heterodimers which resemble the inverted repeat structure in larger transporters. Homodimer transporters such as the multidrug efflux pump EmrE are able to insert into two opposing directions in the membrane (i.e. dual topology) whereas heterodimers are comprised of two genes that each insert into the membrane in a single orientation (i.e. single topology). The latter form the paired SMR (pSMR) subfamily and are thought to arise from gene duplication and evolution of dual topology SMR genes (Bay and Turner, 2009; Rapp et al., 2007; Rapp et al., 2006; Kolbusz et al., 2010). The similarity of the quaternary structure displayed by the SMR family with the pseudo two-fold symmetry seen in larger transporters suggests a role for gene duplication and fusion of smaller genes to give rise to transporters commonly found in higher level organisms (Yan, 2013; Lolkema et al., 2008). The widespread prevalence of inverted repeats and divergent evolution from homo-oligomeric proteins suggests that asymmetry between functional subunits may have a fitness advantage, such as for evolving substrate specificity and direction of transport.

This work aims to establish a structure-function correlation for how single mutations introduced into one subunit of the EmrE dimer – a minimal heterodimer – induce a shift in the conformational equilibrium between inward-open and outward-open states. EmrE is a native E. coli transporter that couples drug efflux of cationic/aromatic compounds with the proton gradient (Schuldiner, 2009). It has recently been shown that the coupling stoichiometry can proceed in a 1:1 or 2:1 proton:drug ratio, where the protons are bound by the glutamic acid residues at position 14 in the dimer (Robinson et al., 2017). Since EmrE consists of an asymmetric and antiparallel homodimer, the relative energies between inward-open and outward-open states are identical for the wild-type protein. NMR spectroscopy has been a sensitive technique to probe this asymmetry by revealing a separate set of signals in the spectrum for monomers A and B within the dimer (Cho et al., 2014; Gayen et al., 2013; Morrison et al., 2012). Determining the effect of a single mutation in the dimer would provide insight into the energy landscape by mimicking a primitive evolutionary event such as those that may have led to the pSMR subfamily. Previously, we discovered that conservative mutations located within or close to loop 2 of EmrE modulated the overall rate of conformational exchange between inward-open and outward-open states (Gayen et al., 2016). Loop 2 is comprised of a short stretch of residues (approximately Tyr53, Ile54, and Pro55) and adjoins TM2 and TM3 of EmrE (Figure 1A). Loop 2 of monomer A is located on the open side of the transporter and does not make intermolecular contacts with monomer B in the tetraphenylphosphonium bound form, while loop 2 from monomer B is proximal to loop 3 of monomer A and near the bottom of the substrate cavity formed by TM1-3 (Figure 1A) (Chen et al., 2007; Vermaas et al., 2018; Ovchinnikov et al., 2018). The differential contacts in the structure and distinct chemical shifts for each monomer suggest a minimal heterodimer might influence the conformational equilibrium. Indeed, our preliminary experiments showed that a heterodimer formed between wild-type and a mutant from loop 2 (I54L) or a mutant from the N-terminal region of TM3 (I62L) could induce a change in the conformational equilibrium between inward-open and outward-open states when bound to the high affinity compound tetraphenylphosphonium (Gayen et al., 2016). While these measurements support the presence of differential contacts of loop 2 within monomers A and B, no biological significance was offered for these findings or whether the conformational equilibrium was perturbed for other essential forms of the transporter needed to accomplish transport (i.e., proton-bound states, apo states).

Figure 1 with 2 supplements see all
Conservative mutations within EmrE.

(A) The top panel shows the primary sequence of EmrE and indicates the TM helices and loops. Residues colored in blue, purple, and red correspond to Leu51, Ile62, and tyrosine residues, respectively. The middle and bottom panels display a cartoon representation of the X-ray structure of EmrE (3B5D) (Chen et al., 2007) with the two monomers differentially colored: monomer A is in gray; monomer B is in light blue. Residues corresponding to residues 51–62 are highlighted in yellow and the blue and purple spheres correspond to Leu51 and Ile62, respectively. The middle panel shows a view of the open side of EmrE with residues 51–62 colored for monomer A. The bottom panel shows a view from the closed side of EmrE with the same residues colored for monomer B. (B) PISEMA spectra of wild-type EmrE, L51I, and I62L. All spectra are displayed at 5-times the standard deviation of the noise. Tyr53 located within loop 2 showed a notable spectral perturbation in L51I and I62L PISEMA spectra relative to the wild-type spectrum; a spectral expansion is displayed in Figure 1—figure supplement 1. Due to these perturbations, the assignment was confirmed using a single-site tyrosine mutation with a subsequent PISEMA dataset collected (see Figure 1—figure supplement 2).

https://doi.org/10.7554/eLife.48909.003

Here, we report a conservative mutation (L51I) located in the C-terminal end of TM2 that has the greatest extent of altering the conformational equilibrium when paired with a wild-type monomer. The L51I/wild-type heterodimer was systematically investigated in different states of the transport cycle, including conditions where the essential anionic residue Glu14 was protonated or deprotonated and bound to drugs. These measurements revealed that both protonated and deprotonated drug-free forms of the L51I/wild-type heterodimer displayed a greater change in the conformational equilibrium relative to the drug-bound state. More importantly, growth inhibition experiments and efflux assays established that the conformational bias observed in vitro correlated with the functional output in E. coli. These findings suggest a non-negligible role for conservative mutations in the duplication-divergence evolutionary theory (Ohno, 1967; Taylor and Raes, 2004) thought to govern the creation of larger membrane proteins from smaller genes.

Results and discussion

A single conservative mutant in the EmrE dimer alters the conformational equilibrium

Prior to investigating whether mutations in one subunit of an EmrE dimer could induce a change in the conformational equilibrium in vitro, NMR experiments were carried out on homodimer samples of wild-type EmrE and mutants (L51I, I62L). EmrE samples were prepared in magnetically aligned bicelles consisting of a 3.5/1 molar ratio of long chain (O-14:0-PC) and short chain (6:0-PC) phospholipids at a pH value of 5.8 that corresponds to the Glu14 proton-bound form of EmrE. Solid-state NMR experiments were carried out using the PISEMA experiment (Wu et al., 1994), since this technique gives the largest frequency separation between peaks corresponding to the two monomers within the asymmetric dimer (Gayen et al., 2013). PISEMA spectra of homodimer samples of wild-type, L51I, and I62L selectively labeled with 15N-tyrosine are shown in Figure 1B. Each spectrum displayed 10 peaks, which confirms the presence of twice the number of peaks as tyrosine residues in the primary sequence of EmrE (Tyr4, Tyr6, Tyr40, Tyr53, Tyr60), and is consistent with previous observations (Gayen et al., 2013; Morrison et al., 2012; Gayen et al., 2016; Dutta et al., 2014b).

To analyze whether heterodimers might lead to a preferred conformation, we prepared mixtures of wild-type EmrE with the L51I or I62L mutant. In each experiment, only the wild-type or mutant was isotopically enriched with 15N-tyrosine and mixed with its partner protein at natural abundance. Using this approach, only one protein was NMR active, while the other was NMR silent. PISEMA spectra corresponding to wild-type/L51I or wild-type/I62L where wild-type was isotopically enriched showed a primary set of intense peaks that corresponded to monomer B (Figure 2A). These signals were superimposable onto those peaks in the wild-type EmrE PISEMA spectrum (Figure 2—figure supplement 1A). Next, we carried out the reverse experiments where the mutants were isotopically enriched and wild-type was NMR silent. In these spectra, the isotopically enriched mutant (L51I or I62L) in the mixed dimer also showed an intense set of peaks (Figure 2B), yet these signals did not overlap with those of isotopically enriched wild-type in the mixed dimer (Figure 2—figure supplement 1B). The peak positions did however match onto the corresponding peaks in the L51I and I62L homodimer PISEMA spectra (Figure 2—figure supplement 1A). These data indicate that wild-type EmrE has a preference for monomer B in the heterodimer while L51I or I62L prefers monomer A for the proton-bound form of EmrE (Figure 2C). It is important to note that the mixture of wild-type EmrE and the L51I or I62L mutant produces a statistical fraction of heterodimers and homodimers in the samples. While our analysis focused on the most intense set of signals, we were able to resolve a weaker set of signals that likely corresponded to homodimers at a lower contour level (Figure 2—figure supplement 2). Nevertheless, these observations demonstrate that a minimal heterodimer comprised of a single conservative mutation can strongly influence the conformational equilibrium and perturb the energy landscape of EmrE.

Figure 2 with 2 supplements see all
Heterodimer experiments showing conformational bias using oriented sample solid-state NMR.

(A) PISEMA spectra of isotopically enriched wild-type EmrE mixed with L51I (top) or I62L (bottom) at natural abundance. (B) PISEMA spectra of isotopically enriched L51I (top) or I62L (bottom) mixed with wild-type EmrE at natural abundance. The underlined protein in each panel was isotopically enriched with 15N tyrosine, while the partner protein was unlabeled and NMR silent. Each PISEMA spectrum is shown at 5-fold above the standard deviation of the noise within the dataset. (C) Cartoon representation depicting how heterodimers lead to a change in the conformational equilibrium, where the mutant adopts monomer A conformation in the heterodimer and the wild-type monomer B population.

https://doi.org/10.7554/eLife.48909.006

Assessing the conformational equilibrium for different states within the transport cycle

All experiments in Figure 2 were performed on the proton-bound and drug-free form of EmrE. However, substrate transport requires EmrE to sample additional conformations within its catalytic cycle. To investigate how other states might impact the conformational equilibrium, we turned to solution NMR spectroscopy since it is a more sensitive technique to probe structure and conformational dynamics. To confirm the observations from PISEMA experiments in Figure 2, we collected solution NMR experiments of mixed dimers under proton-bound sample conditions in 2H-14:0-PC/2H-6:0-PC (1/2 mol/mol) isotropic lipid bicelles. The mixed samples of wild-type EmrE and L51I were prepared in a similar manner as for solid-state NMR experiments with two notable differences: (1) the NMR active protein was isotopically enriched with 13C at the Cδ position of isoleucine residues and (2) the ratio of isotopically labeled wild-type EmrE to L51I at natural abundance was 1/3. The latter was possible due to the increased sensitivity of methyl detection in solution NMR spectroscopy. Figure 3A shows the 1H/13C HMQC spectrum of isotopically enriched wild-type EmrE in a mixed dimer sample with L51I. From this spectrum, we observed an intense set of signals corresponding to monomer B, which confirms the results of our solid-state NMR experiments in aligned lipid bicelles. Thus, under sample conditions used to collect solution and solid-state NMR spectra (i.e. isotropic and aligned bicelles, respectively), the skewed conformational equilibrium is preserved.

Figure 3 with 1 supplement see all
Determination of conformational equilibria among different states within the transport cycle probed with solution NMR.

HMQC spectra of wild-type (middle row, green spectra) or isotopically enriched wild-type mixed with 3-fold excess L51I at natural abundance (bottom row, black spectra). The sample conditions were: (A) proton-bound (pH = 5.6), (B) deprotonated, apo (pH = 9.1), (C) ethidium-bound (pH = 9.1), and (D) tetraphenylphosphonium-bound (pH = 9.1). In each case, the isotopically enriched protein was 13C labeled at the Cδ position of isoleucine. Wild-type spectra serve as a reference since the population between monomers A and B is 50/50. Peaks labeled with red correspond to monomer B peaks, while those in black correspond to monomer A peaks.

https://doi.org/10.7554/eLife.48909.009

Next, we sought to investigate whether other states of the transport cycle, such as deprotonation of Glu14 and drug bound forms, may alter the conformational equilibrium. Previously, we reported pH induced chemical shift perturbations within EmrE’s NMR methyl spectrum that were centered at a pH value of 7.0 and corresponded to a biologically relevant pKa of Glu14 (Gayen et al., 2016). Thus, we acquired a 1H/13C HMQC spectrum with a mixed sample of 13C enriched EmrE and natural abundance L51I at a pH value of 9.1. Similar to the proton-bound dataset, this spectrum showed intense peaks that corresponded to the signals stemming from monomer B (Figure 3B). This is in stark contrast to the spectrum of wild-type EmrE at high pH that showed approximately equal intensities of monomer A and B peaks (Figure 3B). This result supports the conclusion that the skewed conformational equilibrium is preserved under conditions in which Glu14 is deprotonated. Note that we also observed monomer A signals at a lower contour level, which arises due to the statistical nature of mixing as discussed above (Figure 3—figure supplement 1A). Using the relative peak intensities of A and B signals in the spectrum, we estimated the equilibrium population of wild-type EmrE to assume monomer B in the heterodimer at ~96% (range of 90% to 99% based on standard deviation). This value corresponds to a free energy of ~1.8 kcal/mol induced by the L51I mutation in the heterodimer.

To determine the effect of drug binding on the conformational equilibrium, we acquired NMR spectra for the mixed dimers in the presence of ethidium and tetraphenylphosphonium. These drug substrates are commonly used in the EmrE literature for measuring resistance, transport, and binding (Yerushalmi et al., 1995; Curnow et al., 2004; Robinson et al., 2018). Addition of ethidium induced a significant amount of spectral broadening and ablation of peak intensities in both the wild-type sample and that of the mixed dimer where wild-type was isotopically enriched (Figure 3C). These data suggest intermediate chemical exchange, which likely stems from motion corresponding to exchange between inward-open and outward-open states (Cho et al., 2014). From these data, it is likely that the equilibrium is less skewed in the mixed sample upon ethidium binding. Namely, if the populations were maintained to the same extent as in the mixed dimer sample at pH 9.1 (Figure 3B), the effect of intermediate chemical exchange would not induce peak broadening beyond detection. Hence, the effect of ethidium binding suggests that the conformational bias is reduced when the heterodimer is bound to a drug substrate. Furthermore, the line-broadening supports the presence of drug-induced dynamics in the heterodimer, which would enable conformational exchange needed for drug transport.

Finally, we investigated the effect of tetraphenylphosphonium binding on the conformational equilibrium. Unlike ethidium, tetraphenylphosphonium binds with greater affinity to EmrE and significantly reduces the rate of conformation exchange (i.e. slow chemical exchange regime) (Cho et al., 2014; Morrison and Henzler-Wildman, 2014). We carried out the same NMR experiment by adding tetraphenylphosphonium to a mixed dimer sample of 13Cδ-Ile labeled EmrE mixed with natural abundance L51I (Figure 3D). Unlike the effect of ethidium binding, the spectra were nicely resolved in the presence of tetraphenylphosphonium. We observed that monomer B peaks were more intense relative to those of monomer A; however, the signal intensities stemming from monomer A were significantly greater compared to any of the drug-free samples (see I58A, I62A, and I88A peaks in Figure 3D). Specifically, the ratio of monomer B to A peak intensities were reduced upon addition of tetraphenylphosphonium relative to the deprotonated sample. Using the relative peak intensities of A and B, we estimated the equilibrium population of wild-type to assume monomer B in the heterodimer at ~86% (standard deviation range of 79% to 92%), which corresponds to ~1.1 kcal/mol induced by the single mutant in the heterodimer. This result is in agreement with the ethidium binding experiment. These data also provide evidence that drug binding would not trap the transporter into a single conformation in the transport cycle and is consistent with previous work proposing the substrate controls the rate of conformational exchange (Morrison and Henzler-Wildman, 2014).

Functional assays in E. coli reveal biological significance for equilibrium changes

E. coli growth inhibition assays against ethidium bromide were carried out to determine the biological significance of our NMR observations. Initially, we tested whether the mutations L51I and I62L were able to confer resistance when expressed individually. Figure 4A shows a growth inhibition assay using serial 10-fold dilutions on an LB agar plate with wild-type, L51I, I62L, and a control vector. From these data, L51I and I62L displayed a strong phenotype toward ethidium and were indistinguishable relative to wild-type EmrE. Thus, each mutation appears to be fully functional and able to couple to the electrochemical potential to accomplish active drug efflux.

Figure 4 with 1 supplement see all
E. coli growth inhibition assays containing plasmids of wild-type EmrE, conservative mutants, and single topology variants of EmrE.

(A) LB agar plates in the absence or presence of ethidium bromide spotted with E. coli containing plasmids corresponding to vector (control), wild-type EmrE, L51I, and I62L. Serial 10-fold dilutions are displayed in each panel from left to right. (B) LB agar plates in the absence or presence of ethidium bromide spotted with E. coli containing plasmids corresponding to vector (control), EmrEin, EmrEout, EmrEin and EmrEout co-expressed, and wild-type EmrE. Serial 10-fold dilutions are displayed in each panel from left to right. A schematic of the constructs and sequences of EmrEin and EmrEout are given in Figure 4—figure supplement 1.

https://doi.org/10.7554/eLife.48909.011

To probe the effect of heterodimer induced changes to the conformational equilibrium, we designed transporter complementation experiments by expressing two genes: (1) gene 1 consisted of wild-type EmrE, L51I, or I62L and (2) gene 2 consisted of single topology variants of EmrE. The latter utilized similar constructs as previous work showing insertion topology can be controlled by mutating arginine and lysine residues within the primary sequence (Rapp et al., 2007) (i.e. the positive inside rule; Heijne, 1986). Using this technology, gene 2 consisted of N- and C-termini facing the cytoplasmic direction (EmrEin) or periplasmic direction (EmrEout), respectively. Resistance assays verified literature results (Rapp et al., 2007) that EmrEin and EmrEout expressed individually were unable to confer a phenotype to ethidium, whereas co-expression displayed a strong phenotype toward this compound (Figure 4B).

Next, we carried out resistance assays using mixtures of wild-type, L51I, or I62L with EmrEin or EmrEout. The experiments where wild-type EmrE was co-expressed with EmrEin or EmrEout revealed identical levels of conferred resistance relative to wild-type EmrE alone (Figure 5A). To ensure that wild-type was forming a heterodimer with EmrEin or EmrEout, we designed a control experiment in which an additional mutation (E14Q) was engineered into the EmrEin or EmrEout constructs (i.e. E14Qin or E14Qout). E14Q was selected since EmrE requires two glutamic acid residues to confer drug resistance (Rapp et al., 2007). Thus, if a dimer of wild-type and E14Qin or E14Qout formed, we would observe a reduced growth phenotype. Indeed, the results of these experiments displayed growth inhibition when wild-type was co-expressed with E14Qin or E14Qout (Figure 5A). Taken together with the robust phenotype of wild-type co-expressed with EmrEin or EmrEout, these results confirmed that the transporter complementation approach leads to heterodimer formation in the cell membrane.

Figure 5 with 2 supplements see all
Assays to determine the biological significance of conformational bias observed in NMR experiments.

(A) LB agar plates in the absence or presence of ethidium bromide spotted with E. coli containing plasmids corresponding to vector (control), wild-type EmrE, and combinations of wild-type EmrE co-expressed with EmrEin, EmrEout, E14Qin, or E14Qin. Serial 10-fold dilutions are displayed in each panel from left to right. (B) LB agar plates in the absence or presence of ethidium bromide spotted with E. coli containing plasmids corresponding to vector (control), L51I, I62L, and combinations of L51I or I62L co-expressed with EmrEin or EmrEout. Serial 10-fold dilutions are displayed in each panel from left to right. (C) Resistance assays in liquid media in the presence of ethidium bromide (240 μg/mL). E. coli cultures containing the indicated plasmid were grown to an OD600 of ~1.0 and treated with ethidium bromide. The point of ethidium addition represents time = 0. The error bars reflect the standard deviation from three trials. Error bars not observed are smaller than the points. (D) Ethidium efflux assays with E. coli transformed with plasmids corresponding to vector (control) and L51I co-expressed with EmrEin or EmrEout. The normalized ethidium fluorescence is plotted as a function of time following the addition of glucose at time zero. (E) The normalized fluorescence value from panel D at 3600 s. (F) Schematic depiction skewed conformational equilibria for the different heterodimers as indicated within the panel. (G) EmrE transport model proposed by Robinson et al. (2017). ‘E’ refers to the EmrE dimer, ‘H’ is proton, and ‘S’ is the drug substrate. The states on the left are cytoplasmic-facing, while those on the right are periplasmic-facing. The double asterisks indicate equilibria most influenced by the heterodimers displaying conformational bias (drug-free states), while a single asterisk indicates equilibria influenced to a minor extent.

https://doi.org/10.7554/eLife.48909.013

We next carried out experiments with L51I and I62L co-expressed with EmrEin or EmrEout. Remarkably, when EmrEin was co-expressed with L51I or I62L, we observed a significant reduction in ethidium resistance on LB agar plates (Figure 5B). To the contrary, EmrEout co-expressed with L51I or I62L retained the same phenotype as the single-site mutant alone. To provide additional support, we carried out resistance assays in liquid media. These experiments were performed by growing E. coli cultures containing plasmids to an optical density at 600 nm of 1.0, then adding ethidium and recording the culture density as a function of time. Similar to the LB agar plate results, we found that the L51I or I62L mutant expressed with EmrEin displayed a faster drop in the optical density, which signified a reduced ability to confer resistance relative to the same mutant expressed with EmrEout (Figure 5C). It is important to note that if heterodimers of L51I or I62L and EmrEin did not form, we would have observed the same resistance phenotype of L51I or I62L alone. However, the reduced phenotype and the control experiments discussed above (Figure 5A; Figure 5—figure supplement 1) indicate a specific association in the transporter complementation experiments that are influenced by the conformational equilibrium within the assembled heterodimer in the cell membrane.

To provide direct evidence that the overall rate of ethidium transport was different between the mutant co-expressed with EmrEin and EmrEout, we performed an ethidium efflux assay by measuring the intrinsic fluorescence of ethidium. E. coli were treated with ethidium bromide and the ionophore carbonyl cyanide m-chlorophenylhydrazone (CCCP), which causes the cytoplasmic ethidium concentration and fluorescence to increase. Upon addition of glucose and removal of CCCP, the membrane potential is reestablished, and the fluorescence decreases as ethidium is transported out of the cytoplasm. This assay was performed with L51I co-expressed with EmrEin or EmrEout. Immediately following the addition of glucose, the fluorescence dropped ~3 fold faster for L51I co-expressed with EmrEout than L51I co-expressed with EmrEin (Figure 5D). Furthermore, the fluorescence value at steady state (3600 s) was significantly lower for L51I co-expressed with EmrEout compared to L51I co-expressed with EmrEin (Figure 5E). This indicates a reduced cytoplasmic ethidium concentration for the L51I/EmrEout sample. Taken together, these data are consistent with the resistance assay results and support the conclusion that a change in the conformational equilibrium can influence the overall rate of drug efflux. Note that a previous study reported an ethidium efflux rate difference of 1.6-fold between wild-type E. coli and an AcrAB-TolC efflux pump knockout strain that led to a 30-fold difference in the minimum inhibitory concentration (Paixão et al., 2009). This shows that somewhat modest transport rate differences can have deleterious effects on bacterial growth, which stems from increased ethidium accumulation that leads to irreversible binding to DNA (Jernaes and Steen, 1994; Lambert and Le Pecq, 1984).

Rationalization for why a change in the conformational equilibrium influences phenotype

How does the NMR observed conformational change of the heterodimer correlate with the functional results? The transporter complementation experiments create topologically defined heterodimers due to the single topology of EmrEin and EmrEout and the antiparallel association of the dimer quaternary structure. Using this information and knowledge from NMR experiments that the mutant had a preference for monomer A in the heterodimer, we can make inferences about how a shift in the conformational equilibrium influences function. Specifically, the mutant paired with EmrEout in a heterodimer would favor the inward-open/cytoplasmic-facing conformation (Figure 5F). Likewise, the mutant paired with EmrEin will bias toward the outward-open/periplasmic-facing conformation under the same conditions (Figure 5F). Hence, the resistance assay results suggest that the conformational equilibrium favoring the outward-open/periplasmic state has a deleterious impact on the overall transport cycle. In contrast, a heterodimer with a conformation favoring the inward-open/cytoplasmic facing conformation gives no measurable reduction in phenotype. This finding is in harmony with our prior observation that the role of the pH gradient is to favor the inward-open conformation that is poised for drug binding (Gayen et al., 2016). Thus, reducing the effect of the pH gradient by biasing the equilibrium in the opposite direction apparently is sufficient to alter the net transport as observed in growth inhibition experiments and efflux assays.

It is important to underscore that the equilibrium constant between inward-open and outward-open states of the heterodimer ultimately stems from differential kinetic rate constants pertaining to the inward-open to outward-open and outward-open to inward-open transitions (see kinetic model [Robinson et al., 2017] in Figure 5G). Thus, we hypothesize that the primary kinetic steps in the transport cycle that influence the observed phenotype are the inward-open to outward-open and outward-open to inward-open rates for the drug-free heterodimer (Figure 5G, see double asterisks). We make this conclusion based on two observations. First, the populations were more skewed for the drug-free states than those for the drug bound states. Second, NMR experiments with ethidium indicate that the conformational exchange was not halted in the heterodimer. The latter means that the conformational bias does not result in a locked-state that would ultimately lead to loss-of-function (i.e. no substrate turnover).

To provide support for this conclusion, we performed a numerical simulation of pH-driven transport into liposomes using the kinetic model and parameters introduced by Robinson et al. (2017). The differences were to change the inward-open to outward-open and outward-open to inward-open conformational exchange rates based on the populations observed in our heterodimer experiments and to account for the fact that drug binding affinities to wild-type(A)/mutant(B) and mutant(A)/wild-type(B) are not necessarily the same (see Figure 5—figure supplement 2). Using these simulations, we found that the rate of drug transport was ~5.6 fold faster for a heterodimer of mutant and EmrEout versus a heterodimer of mutant and EmrEin (Figure 5—figure supplement 2). In addition, the former achieved a small but significant increase in the final concentration of drug within the vesicles, which further suggests that differences in kinetic rates in EmrE’s transport cycle give rise to a change in the cytoplasmic drug concentration between heterodimers with opposite insertion topologies in E. coli. These calculations correlate with the enhanced phenotype and increased rate of ethidium efflux and provide evidence that shifting conformational equilibria via specific rate constants in a transport cycle has a direct impact on functional output.

Conclusion

Despite the prominence of inverted repeat domains within transporters, there are only a few examples of mutational studies to influence conformational equilibria. These experiments involve the creation of loss-of-function, non-conservative mutations (e.g. Trp to Gly) at key structural contacts with the goal of crystallographically trapping particular conformations in the transport cycle (Smirnova et al., 2013; Latorraca et al., 2017). Our observations demonstrate that a minimal heterodimer comprised of a single conservative mutation is sufficient to disrupt the preferred resting conformation and underscore the presence of a dynamic energy landscape where relatively small energy differences exist among conformations in the transport cycle. From an evolutionary perspective, we hypothesize there might be fitness advantages for having separate genes that form a hetero-oligomer or a single polypeptide chain where the structural repeats are contained within one protein. In these cases, a single mutation can be introduced within the functional assembly that is not possible for a homo-oligomeric protein. Functional differences between closely related proteins stemming from one mutation in a hetero-oligomer versus multiple mutations in a homo-oligomer could potentially include influencing the propensity of a transporter to carry out antiport or symport. In fact, in addition to EmrE’s known antiport activity, it is also able to carry out symport under certain conditions (Robinson et al., 2017), including import of polyamines upon mutation (Brill et al., 2012). Therefore, the observation that conservative mutations can display skewed conformational equilibria and influence function opens the possibility of somewhat modest evolutionary paths for achieving symport or antiport.

Materials and methods

Protein expression and purification

Request a detailed protocol

Protein expression and purification was carried out as previously reported (Gayen et al., 2013). EmrE is expressed as a fusion construct with maltose-binding protein in the pMAL vector (New England Biolabs Inc) in E. coli BL21(DE3) cells. For oriented sample NMR experiments, 15N-tyrosine labeled EmrE was expressed with IPTG for 4 hr at 25°C in an amino acid mixture consisting of 19 amino acids at natural abundance (300 mg/L) and 15N-tyrosine (120 mg/L). Unlabeled EmrE used in the heterodimer experiments was expressed in LB medium at natural abundance. For solution NMR experiments, wild-type and L51I were expressed in a fully perdeuterated background as previously described (Gayen et al., 2016). Wild-type protein was isotopically enriched with 13C at the Cδ position of isoleucine methyl groups with the addition of 50 mg/L 2-ketobutyric acid-4-13C, 3,3-2H2 sodium salt hydrate 1 hr before induction. The bacterial cells following expression were lysed and the fusion protein was purified with amylose affinity chromatography. The fusion protein was cleaved with tobacco etch virus protease (TEV) and EmrE was further purified by size exclusion chromatography using a Superdex 200 10/300 column (GE Healthcare) in 0.06% w/v n-dodecyl-β-D-maltopyranoside (DDM).

NMR sample preparation

Request a detailed protocol

For oriented sample solid-state NMR studies, purified EmrE in DDM detergent was reconstituted into 1,2-di-O-tetradecyl-sn-glycero-3-phosphocholine/dihexanoyl-sn-glycero-3-phosphocholine (O-14:0-PC/6:0-PC) bicelles at a molar ratio of 3.5/1 (i.e. q = 3.5). The bicelles were made with a protein concentration ~2 mM with a combined lipid concentration of 25% (w/v) in 80 mM HEPES and 20 mM NaCl at pH = 5.4. The heterodimer samples were prepared by mixing the two proteins (wild-type/L51I or wild-type/I62L) with a molar ratio 1/1.2, where the protein in excess was at natural abundance (i.e. NMR silent). The two proteins used to prepare the heterodimer sample were incubated together at 37°C in the presence of 50 mM DTT for 1 hr immediately prior to reconstitution into lipids. The reconstitution into the bicelles was carried out in the same manner as those for homodimer samples (Leninger and Traaseth, 2018).

For solution NMR studies, purified EmrE in DDM detergent was reconstituted into dimyristoyl-sn-glycero-3-phosphocholine (14:0-PC) and dihexanoyl-sn-glycero-3-phosphocholine (6:0-PC) bicelles with a molar ratio of 1/2. The acyl chains of 14:0-PC and 6:0-PC were perdeuterated (Avanti Polar Lipids) to reduce the lipid signals in 1H/13C heteronuclear correlation experiments. The heterodimer samples were prepared by mixing isotopically enriched wild-type with natural abundance L51I in a 1/3 molar ratio. The two proteins used to prepare the heterodimer sample were incubated together at 37°C for 1 hr immediately prior to reconstitution into lipids. The heterodimers contained 0.533 mM total protein (0.133 mM wild-type, 0.4 mM L51I). The long-chain lipid (14:0-PC) to total protein ratio of the solution NMR samples was ~150/1 (mol/mol). Control homodimer samples with isotopically enriched wild-type EmrE only were prepared in a similar fashion. The sample buffer was 150 mM Na2HPO4 and 20 mM NaCl. The experiments with ethidium bromide and tetraphenylphosphonium were carried out at concentrations of 2.1 mM and 1.05 mM, respectively.

In order to assess whether a 1 hr incubation time at 37°C was sufficient to achieve complete mixing of the heterodimer, a control experiment was carried out by comparing NMR spectra after 1 hr and 19 hr of incubation at 37°C. The NMR spectra collected of these experiments showed that shorter and longer incubation times gave statistically the same ratios of monomer A and B peak intensities, which demonstrates that complete mixing was achieved with 1 hr (Figure 3—figure supplement 1B). Note that the 1 hr incubation time in DDM is consistent with the reduced dimer stability of EmrE in DDM detergent micelles compared to that in lipid bicelles or lipid bilayers (Dutta et al., 2014a).

Solid-State NMR spectroscopy

Request a detailed protocol

Oriented sample solid-state NMR experiments were acquired using an Agilent DD2 spectrometer at a 1H frequency of 600 MHz equipped with a 1H/15N double resonance probe manufactured by Revolution NMR (design of Peter Gor’kov; Gor'kov et al., 2007). Experiments were carried out using magnetically aligned lipid bicelle samples that were flipped with the addition of 3 mM YbCl3 to orient the bicelle normal parallel to the magnetic field. PISEMA (Wu et al., 1994) spectra were acquired with SPINAL-64 1H decoupling during acquisition and phase modulated Lee-Goldberg (Vinogradov et al., 1999) (PMLG) 1H-1H decoupling in the indirect dimension. The radiofrequency field for SPINAL decoupling was 50 kHz, while the effective field for PMLG was 41.7 kHz. Spectra were acquired with ~1500 scans and 14 increments in the indirect dimension with a recycle delay of 3 s. The indirect dimension axis was corrected with the scaling factor of 0.82. The 15N direct dimension was referenced to 15NH4Cl (solid) at 41.5 ppm.

Solution NMR spectroscopy

Request a detailed protocol

Solution NMR experiments were acquired using a Bruker Avance III spectrometer at a 1H frequency of 600 MHz equipped with a TCI cryogenic probe. 1H/13C HMQC experiments were acquired at 25°C using 1H and 13C spectral widths of 10,000 Hz and 4,000 Hz, respectively. The total acquisition (1H) and evolution times (13C) corresponding to t2 and t1 were 59.9 msec and 18.5 msec, respectively. Spectra were acquired with 12 or 24 scans with a recycle delay of 1 s.

Quantification of population from heterodimer samples

Request a detailed protocol

The equilibrium shown in Equation 1 corresponds to a heterodimer composed of wild-type EmrE and mutant:

(1) WTAmutantB mutantAWTB

Subscripts 'A' and 'B' correspond to monomers A and B in the asymmetric dimer. A simple calculation of the peak intensity ratio for A and B resonances in the heterodimer samples does not correspond to the equilibrium constant since the peak intensities contain a statistical fraction of homodimers (fhomo) and heterodimers (fhet) based on the ratio of isotopically labeled and natural abundance proteins present for mixing. Note that fhomo and fhet consider only homodimers or heterodimers that contain the isotopically enriched protein. The observed peak intensities for monomer A (IA,obs) and B (IB,obs) signals in a heterodimer sample are given by Eqns. 2 and 3:

(2) IA,obs=IA fhomo+fhet pA
(3) IB,obs=IB fhomo+fhet pB

Equation 2 divided by Equation 3 gives:

(4) IA,obsIB,obs=IA IB fhomo+fhet pA fhomo+fhet pB

IA/IB is the ratio of ‘intrinsic’ intensities of monomer A and B peaks obtained from the homodimer spectrum. This is needed since A and B peaks in the homodimer spectrum are not exactly the same. fhomo and fhet were set to 1/7 and 6/7, respectively, by assuming statistical mixing of the molar ratio of isotopically labeled wild-type to natural abundance L51I mutant in solution NMR experiments (i.e., 1 part labeled to 3 parts unlabeled). pA and pB are populations of wild-type in monomer A or B conformations when in a heterodimer with a mutant (see Equation 1). The addition of these populations is given by Equation 5 and their ratio is the equilibrium constant (K) in Equation 6:

(5) pA+pB=1
(6) K=pBpA

Substitution of pB from Equation 5 into Equation 4 gives the following:

(7) IA,obsIB,obs=IA IB fhomo+fhet pA fhomo+fhet (1-pA)

pA was solved from Equation 7 for drug-free (pH = 9.1) and tetraphenylphosphonium bound forms of wild-type/L51I heterodimers using solution NMR HMQC spectra of wild-type EmrE and isotopically labeled wild-type mixed with natural abundance L51I. The average and error range (derived from the standard deviation) of the populations were determined from the following residues that displayed well-resolved signals in each spectrum: Ile11, Ile54, Ile58, Ile62, Ile68, Ile88, and Ile101.

Resistance assays on solid media

Request a detailed protocol

Growth inhibition assays were performed in pET Duet-1 vectors with constructs designed similar to that of Rapp et al. (2007). All of the constructs are shown in Figure 4—figure supplement 1. EmrEin consists of three mutations relative to wild-type EmrE (R29G, R82G, S107K) and induces the N- and C-termini to face the cytoplasm. EmrEout consists of three mutations from wild-type (T28R, L85R, R106A) and induces the N- and C-termini to face the periplasm. EmrEin and EmrEout were placed in the second cloning site. An additional mutation to EmrEin and EmrEout constructs were made by mutating E14 to Q14. These constructs are also single topology and are referred to as E14Qin and E14Qout. The L51I and I62L constructs are single site mutants of wild-type EmrE (UniProt P23895).

Luria-Bertani (LB) media and Luria agar powder for resistance assay were purchased from Research Products International. Each construct in the pET Duet-1 vector was transformed into BL21(DE3) and grown at 37°C up to an OD600 of ~1.5. The cultures were diluted to an OD600 of 1.0 with fresh LB supplemented with carbenicillin (100 μg/mL) and were then serially diluted by 10-fold to achieve final dilutions of 100 to 106. All of the dilutions used fresh LB media containing 100 μg/mL carbenicillin. 3 μl of each dilution were pipetted onto plates containing 20 μM IPTG, 100 μg/mL carbenicillin, and 94 μg/ml ethidium bromide. Control experiments were carried out by plating cells onto LB agar plates containing 20 μM IPTG and 100 μg/mL carbenicillin. All resistance assays on solid medium were repeated at least two times.

Resistance assays in liquid media

View detailed protocol

Liquid assays were performed using the same constructs as for the solid media resistance assay. E. coli BL21 (DE3) were transformed with the pET Duet-1 vectors and grown at 37°C until the OD600 was ~1.5. The cultures were then diluted to an OD600 of 1.0 with fresh LB supplemented with carbenicillin (100 μg/mL) and ethidium bromide (240 μg/mL). Cultures were incubated for 11.5 hr and the OD600 was measured every hr up to 6 hr and one additional time 11.5 hr after the initial exposure to ethidium bromide. A full set of resistance assays in liquid medium were repeated three times. The error bars reflect the standard deviation for replicate trials.

Ethidium bromide efflux assay

Request a detailed protocol

Mutant constructs used in resistance assay were transformed into BL21(DE3) and grown to mid log phase at 37°C (OD600 ~1.0). The cells were spun down and diluted to the OD600 of 0.1 in minimal media A (40 mM K2HPO4, 22 mM KH2PO4, 2 mM sodium citrate, 0.8 mM MgSO4, and 7.6 mM (NH4)2SO4, pH 7.0). The resuspended cells were treated with 80 μM carbonyl cyanide m-chlorophenyl hydrazine (CCCP) for 5 min. Ethidium bromide (10 μg/mL) was added to the cells and incubated for 30 min at 37°C while shaking. Cells were spun down for 10 min and the pellet was kept on ice until the florescence experiment. For the efflux assay, cells were resuspended in minimal media A with 10 μg/mL ethidium bromide. To initiate the assay, 0.2% (w/v) glucose was added. Control experiments were carried out by treating the cells in the same manner except without adding glucose. The fluorescence decays were measured with a Molecular Devices FlexStation 3 instrument using an excitation wavelength of 530 nm and an emission wavelength of 600 nm. The fluorescence was recorded over a time period of 3600 s. Each experiment was acquired in duplicate and repeated at least two times. The error bars reflect the duplicate experiments carried out on the same day.

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files.

References

  1. Book
    1. Ohno S
    (1967)
    Sex Chromosomes and Sex-Linked Genes
    Berlin, New York: Springer-Verlag.
  2. Book
    1. Schuldiner S
    (2014)
    What Can a Living Fossil Tell Us About Evolution and Mechanism ofIon-Coupled Transporters: The Story of Small Multudrug Transporters
    In: Reinghard K, Christine Z, editors. MembraneTransport Mechanism. Springer Berlin Heidelberg. pp. 233–248.

Decision letter

  1. Olga Boudker
    Senior and Reviewing Editor; Weill Cornell Medicine, United States
  2. Timothy A Cross
    Reviewer; Florida State University, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Inducing conformational preference of the membrane protein transporter EmrE through conservative mutations" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Olga Boudker as the Reviewing Editor and Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Timothy A Cross (Reviewer #1).

In the manuscript, the authors examine the conformational equilibrium of homodimeric EmrE transporter. Individual EmrE promoters insert and assemble in anti-parallel manner and each assumes conformation A or confirmation B. Their structural transition from A to B and from B to A conformations leads to the transition between states open to the periplasm or cytoplasm. Because protomers are identical, there is no conformational bias toward either outward or inward facing states. The authors introduce single subtle mutations into a protomer and show that when such protomers assemble with WT protomers they demonstrate conformational preference for conformation A, while WT protomers assume preferentially conformation B. Now, when mutants are assembled with WT protomers whose insertion topology is fixed by additional mutations relying on "positive inside" principle, the authors presumably generate in cells assemblies that have preferential outward- or inward-facing configurations. They show that conformational preference for the inward-facing states results in active assemblies, while conformational preference for the outward-facing state leads to reduced activity. This is an elegant study that probes the role of the so-called "energy landscape" in defining the rate of transport. The reviews have found both the solid-state and the solution NMR spectroscopies to be well done and cell assays appropriate and informative. There were several aspects of the work that, however, raised concerns.

Essential revisions:

1) Key experiments in this manuscript cover ground that was already covered in the authors' 2016 Nature Chemical Biology paper (Gayen et al., 2016). It is a really interesting observation that a single mutation in only one subunit changes the energy landscape. This observation, originally reported in Gayen et al., is replicated using a different NMR strategy here. It would be critical for the authors to articulate what is different and new in the current study. It was also puzzling why different mutants were used in this paper compared to the 2016 paper. On the related note, the Introduction section was perceived as insufficient to introduce non-specialists to the EmrE class of transporters and to explain what the goals of the current experiments are.

2) The reviewers were surprised by how facile the preparation of the heterodimers was. Mixing for 1 hour at 37 °C is often insufficient to achieve subunit exchange in membrane proteins. Slower kinetics is typical for membrane proteins in general (Jefferson…Bowie, JACS, 2013), and has been reported for EmrE specifically (Rotem…Schuldiner JBC 2001). The authors should present evidence that complete mixing has been achieved. Further, fast kinetics also raises question whether monomers are present in the mix at significant levels during data collection. The authors should address this concern. Finally, there exists the possibility that the subunits actually exchange during the experiment. The experiments with labeled Tyr have very close to equal amounts of labeled and unlabeled protein, so the effect of subunit exchange would be expected to be pretty substantial. How would this change the results?

3) The reviewers further felt that the paper would be significantly strengthened if the authors took more quantitative approach to the analysis of populations. "Conformational preference" is an imprecise term. The authors should consider analyzing populations quantitatively to estimate free energies associated with preferred conformations. This is particularly pertinent because mutations are, indeed, very subtle and putting numbers on their effects would be informative.

4) The reviewers were surprised further by how detrimental to the bacterial survival the mutations were. How significant should be the population shift to lead to a complete loss of resistance to ethidium bromide? Even in the authors' simulations, the difference in uptake rate is only three-fold and it is difficult to reconcile such relatively subtle kinetic effects with complete loss of resistance. It would be important for the authors to comment on this more explicitly.

5) The Figure 5D is confusing. Why is there a conformational preference for the inward-facing state in EmrEin/EmrEout dimer? Should not it, like WT, have no conformational preference? What is exactly depicted in the middle panel? Here, the preference should be for the outward-facing state shouldn't it? Perhaps, the figure should be remade to make the point the authors are making clearer. It might be better to show the entire transport cycle and point out which rates are likely affected by the asymmetry generated by mutations within the assemblies with EmrEin or EmrEout variants. Such depiction may also allow to clarify which conclusions are fully supported by the data and which are more speculative in nature.

Reviewer #1:

I found this work to be very clever and the results to be exciting. Both the solid-state and solution NMR spectroscopies were well done as were the assay studies. However, as a membrane protein spectroscopist and as someone who has not kept up with the EmrE literature, I found the start of the Results and Discussion text very difficult to comprehend. The Introduction did not include an introduction to EmrE, but only to the class of transporters – as a result I found the Results and Discussion section difficult to comprehend – a paragraph in the Introduction about EmrE in/out, EmrE A and B etc. would be helpful for a broader audience.

The study involves the functional characterization of two conservative mutants of an antiparallel dimeric transporter, EmrE. These mutants L51I and I62L result in preferential roles when paired with a WT monomer. Furthermore, these WT/mutant hybrids result in modified functionality. Clever mutation of EmrE by changing the inward facing charge on the structure results in an inward facing structure EmrEin and an outward facing structure EmrEout, these mutants in combination displayed resistance to ethidium. Furthermore, the EmrE dimers of L51I and dimers of I62L display resistance to ethidium, but L51I with EmrEin and I62L with EmrEin both displayed reduced resistance to ethidium. However, the mutants paired with EmrEout displayed WT resistance to ethidium. Based on the structural studies with NMR spectroscopy the authors were able to note that the conservative EmrE mutants have a preference for monomer A in the heterodimers with WT. For EmrEout with a mutant the heterodimer preference is inward-open/cytoplasmic-facing conformation and for EmrEin the preference was for the outward-open/periplasmic-facing conformation under the same drug free conditions. This led the authors to justifiably make two conclusions: 1) that the "outward-open/periplasmic (facing) state has a deleterious impact on the overall transport cycle" and 2) that the "inward-open/cytoplasmic facing conformation" gives no such deleterious effect. This is really an elegant detective story involving these subtle mutations at the ends of two TM helices that anchor an extramembrane loop.

Reviewer #2:

This manuscript from the Traaseth lab investigates the effect of a pair of conservative mutations, L52I and I62L, on protein conformation and drug export function in EmrE.

The best experiments in this manuscript cover ground that was already covered in the authors' 2016 Nature Chemical Biology paper (Gayen et al., 2016). It is a really interesting observation that a single mutation in only one subunit changes the energy landscape. This observation, originally reported in Gayen et al., is replicated using a different NMR strategy here. However, my overall impression of the manuscript is that the follow-up experiments are too observational and qualitative. There are a lot of aspects of this study that I find surprising (beyond the fact that a one methyl group alteration does anything at all!), and presenting these surprising observations without deeper digging is somewhat unsatisfying.

I'm surprised that the subunit mixing is so extensive. The 1 hr, 37 °C incubation that the authors use to achieve heterodimers suggests faster subunit exchange kinetics than is typical for membrane proteins in general (Jefferson…Bowie, JACS, 2013), or what has been reported for EmrE specifically (Rotem…Schuldiner JBC 2001). Are there monomers in equilibrium too? I would feel more comfortable if the authors had some way to assess monomer vs. heterodimer vs. homodimer populations in their samples. Moreover, I do not know what temperature/acquisition time was used to collect the NMR data shown here (or how the kinetics differ in detergent vs. bicelles vs. solid state bilayer), but if subunit mixing reaches equilibrium in an hour, there certainly exists the possibility that the subunits actually exchange during the experiment. The experiments with labeled Tyr have very close to equal amounts of labeled and unlabeled protein, so the effect of subunit exchange would be expected to be pretty substantial. How would this change the results?

The term "conformational preference" is used throughout. Is there some reason the authors don't frame this as conformational equilibrium? I assume that this is some kind of equilibrium process - the heterodimers also go through conformational exchange, except that now the equilibrium more strongly favors the mutant subunit in one state, and the wildtype subunit in the other. This is one area where a more rigorous analysis of equilibrium constants would be very useful. Can the differences in peak volume be measured/reported? A numerical ratio comparing two residues in the "A" and "B" positions would be a better metric than trying to evaluate how peak intensity and width change from the bird's eye view. For example in Figure 3A, I'm just not able to evaluate the author's assertion that the heterodimer has shifted the population towards "B." There aren't, to my eye, enough assigned peaks and direct A:B comparisons.

I'm also very surprised that the EmrE-in/mutant heterodimers are so sensitive to ethidium (as sensitive as abolishing one of the glutamates!). Survival changes by four orders of magnitude! While this paper is short on quantitation, in the previous Nat. Chem Biol paper, the authors estimated that the heterodimer favors one conformation over the other by about 2:1 (compared to 1:1 for WT). In terms of Delta G, this is a small change. The authors agree that their data does not indicate that the mutant locks the transporter in one configuration, and have shown that conformational exchange still occurs for the homodimeric mutant. So I'm puzzled by the extent to which this mutant disables drug export.

In addition, I have some concerns about data presentation. Even though the authors are forthright about doing so, I don't like the maneuver of cutting the contour levels to obscure non-A or non-B peaks in Figure 2. I think the spectra included in the supplement, which show both populations of peaks, are a more faithful visual representation of the actual data. I also don't like showing part of the Figure 1B spectrum at 4x noise, and the rest of the spectrum at 5x noise. I'm guessing that some unexplainable peaks appear elsewhere in the spectrum at 4x noise, which could indicate sample heterogeneity.

Reviewer #3:

In the manuscript, the authors examine the conformational equilibrium of homodimeric EmrE transporter. Individual EmrE promoters insert and assemble in anti-parallel manner and each assumes conformation A or confirmation B. Their structural transition from A to B and from B to A conformations leads to the transition between states open to the periplasm or cytoplasm. Because protomers are identical, there is no conformational bias toward either outward or inward facing states. The authors introduce single subtle mutations into a protomer and show that when such protomers assemble with WT protomers they demonstrate conformational preference for conformation A, while WT protomers assume preferentially conformation B. Now, when mutants are assembled with WT protomers whose insertion topology is fixed by additional mutations relying on "positive inside" principle, the authors presumably generate in cells assemblies that have preferential outward- or inward-facing configurations. They show that conformational preference for the inward-facing states results in active assemblies, while conformational preference for the outward-facing state leads to reduced activity. This is an elegant study that probes at the role of the so-called "energy landscape" in defining the rate of transport. I found it overall compelling, but confusing in interpretation and presentation.

In particular, I found Figure 5D and corresponding discussion confusing. Why is there a conformational preference for the inward-facing state in EmrEin/EmrEout dimer. Should not it, like WT, have no conformational preference? What is exactly depicted in the middle panel. Here, the preference should be for the outward-facing state, right? I think that this figure should be remade. It might be better to show the entire transport cycle and point out which rates are likely affected by the asymmetry generated by mutations within the assemblies with EmrEin or EmrEout variants. My first take on this is as follows: the conformational exchange of the drug-bound transporter is comparatively fast and is not rate-limiting to the cycle. Thus, it is the rate of the transition of the substrate-free transporter that determines the overall rate. In the transport cycle, where drug extrusion is coupled to influx of protons, the outward to inward translocation of the protonated transporter would determine the overall rate. Thus, if mutation favors the inward-facing conformation of Mut/EmrEout dimer, it might do so through accelerating the rate of outward to inward transition and therefore speeding up the cycle. Or, alternatively, it could slow the reverse transition from outward to inward and increase the amount of time for the drug to be successfully released into the periplasm.

https://doi.org/10.7554/eLife.48909.018

Author response

Essential revisions:

1) Key experiments in this manuscript cover ground that was already covered in the authors' 2016 Nature Chemical Biology paper (Gayen et al., 2016). It is a really interesting observation that a single mutation in only one subunit changes the energy landscape. This observation, originally reported in Gayen et al., is replicated using a different NMR strategy here. It would be critical for the authors to articulate what is different and new in the current study. It was also puzzling why different mutants were used in this paper compared to the 2016 paper. On the related note, the Introduction section was perceived as insufficient to introduce non-specialists to the EmrE class of transporters and to explain what the goals of the current experiments are.

The novelty of the current work is to show how the energy landscape is perturbed by a single mutation for several states in the transport cycle and to provide functional support for these in vitro observations. The previous publication in 2016 reported a change in the conformational equilibrium for only the tetraphenylphosphonium bound state and did not report any functional significance for the observation. The revised submission underscores the structure-function relationship we aim to develop while clearly delineating what was previously studied and what is novel about our current work (see Introduction, second and third paragraphs).

With regard to mutations, our initial report in 2016 utilized the I54L and I62L mutations. Since this publication, we discovered that the L51I mutation has the most prominent effect in influencing the conformational equilibrium when paired with a wild-type monomer. For this reason, the majority of our studies were performed with this mutation. The revised version states this point in the Introduction section (second and third paragraphs).

As referenced above, the Introduction has been revamped to provide additional background information on EmrE and to more clearly articulate the goals of our experiments. We also highlight what was accomplished in the 2016 paper, which will enable the reader to identify the novelty of our current work.

2) The reviewers were surprised by how facile the preparation of the heterodimers was. Mixing for 1 hour at 37 °C is often insufficient to achieve subunit exchange in membrane proteins. Slower kinetics is typical for membrane proteins in general (Jefferson…Bowie, JACS, 2013), and has been reported for EmrE specifically (Rotem…Schuldiner JBC 2001). The authors should present evidence that complete mixing has been achieved. Further, fast kinetics also raises question whether monomers are present in the mix at significant levels during data collection. The authors should address this concern. Finally, there exists the possibility that the subunits actually exchange during the experiment. The experiments with labeled Tyr have very close to equal amounts of labeled and unlabeled protein, so the effect of subunit exchange would be expected to be pretty substantial. How would this change the results?

The EmrE dimer has been shown to be at least 1000-fold more stable in lipids compared to dodecyl maltoside (DDM) detergent micelles (Dutta et al., 2014). This is pertinent to the monomer exchange experiments reported by Rotem et al., since they were carried out by solubilizing E. coli membranes with DDM (i.e., native lipids were present). This is in contrast to the preparation of our heterodimer samples, which were carried out on purified proteins in DDM (i.e., few native lipids present). In our hands, purified EmrE in DDM in the drug-free state cannot be heated above ~40 °C without noticeable precipitation after 1 hr. This likely reflects the presence of monomers during the mixture procedure that would increase kinetics relative to the Rotem et al. study. To provide evidence that the mixing is complete, we prepared two parallel samples, where one was incubated at 37 °C for 1 hr and the other was incubated at 37 °C for 19 hrs. Solution NMR samples were prepared in the same way. The ratio of subunit A and B signals was statistically the same between the samples, which supports that complete mixing has been achieved (see Figure 3—figure supplement 1).

The next question was whether there are monomers in our NMR samples. The EmrE dimer is at least 1000-fold more stable in lipid bilayers and ~100-fold more stable in lipid bicelles compared to DDM detergent micelles (Dutta et al., 2014). Furthermore, NMR spectra show only two peaks per residue, where each set of signals stems from one monomer in the asymmetric dimer. This has been confirmed by NMR spectroscopy on cross-linked dimers that give the same peak position as wild-type samples (Cho et al., 2014) and also agrees with the asymmetry seen in cryoEM (Tate et al., EMBO J, 2001, 20, 77-81) and X-ray datasets (Chen et al., 2007). This fact is further underscored by the heterodimer experiments presented in the current work, where each monomer in the heterodimer gives rise to one set of signals.

The last question is whether exchange of labeled and unlabeled proteins occur during the NMR experiments. While exchange is theoretically possible, to the extent that it occurs, it would not change any of our conclusions for two reasons. First, exchange in bicelles would be a slow process on the NMR timescale since the peak positions do not move relative to those of the respective homodimers. This is consistent with the stability of the EmrE dimer reported in bicelles (Dutta et al., 2014). Second, the system is at equilibrium as established from the 1 hr and 19 hr mixing experiments (Figure 3—figure supplement 1), so any subunit exchange would be offset to maintain identical peak intensities of monomers A and B. The fact that the Tyr labeled PISEMA experiments used a lower ratio of isotopically enriched protein to natural abundance protein only changes the expected statistical fraction of homodimers and heterodimers in the sample.

3) The reviewers further felt that the paper would be significantly strengthened if the authors took more quantitative approach to the analysis of populations. "Conformational preference" is an imprecise term. The authors should consider analyzing populations quantitatively to estimate free energies associated with preferred conformations. This is particularly pertinent because mutations are, indeed, very subtle and putting numbers on their effects would be informative.

In the resubmission, we quantified the relative populations in the heterodimer equilibrium corresponding to wild-type and the L51I mutant:

WTA∙mutantB⇌mutantA∙WTB

The calculation is described in detail in the Materials and methods section “Quantification of Populations from Heterodimer Samples”. The populations are reported in the Results section “Assessing Conformational Preference for Different States within the Transport Cycle” for the drug-free deprotonated and the tetraphenylphosphonium bound heterodimers derived from solution NMR experiments shown in Figure 3. The populations of wild-type assuming monomer B in the heterodimer were estimated to be 96% for the apo form and 86% for the tetraphenylphosphonium bound form. These values correspond to free energies of ~1.8 kcal/mol for the apo form and ~1.1 kcal/mol for the tetraphenylphosphonium-bound form. In the Results section, we also report the error range which reflects the deviation among residues used for the calculation.

4) The reviewers were surprised further by how detrimental to the bacterial survival the mutations were. How significant should be the population shift to lead to a complete loss of resistance to ethidium bromide? Even in the authors' simulations, the difference in uptake rate is only three-fold and it is difficult to reconcile such relatively subtle kinetic effects with complete loss of resistance. It would be important for the authors to comment on this more explicitly.

We realized that the drug binding constants for the heterodimer in the simulation is not the same for wild-type(A)/L51(B) and L51I(A)/wild-type(B). In the resubmission, the binding constants were adjusted by the ratio of the equilibrium constants governing the AB to BA transition for the heterodimer in the absence and presence of the drug. The revised simulations (Figure 5—figure supplement 2) show that the concentration of drug into the liposome mutant/EmrEout is slightly greater than for the mutant/EmrEin sample. However, there is a notable difference in the initial rate of transport. Based on these simulations, we hypothesized that the difference in the transport rate by mutant/EmrEin and mutant/EmrEout would lead to a greater cytoplasmic accumulation of ethidium for mutant/EmrEin. The higher accumulation would display a reduced minimum inhibitory concentration, causing mutant/EmrEin to be more susceptible at the ethidium concentration used in growth inhibition assays. To provide evidence that the overall rate of ethidium transport was different between mutant/EmrEin and mutant/EmrEout in vivo, we carried out an ethidium efflux assay by following the intrinsic fluorescence of ethidium. In this assay, the membrane potential is disrupted with the ionophore carbonyl cyanide mchlorophenylhydrazone (CCCP), causing the cytoplasmic ethidium concentration in the cell to increase and give rise to higher fluorescence. Upon addition of glucose and removal of CCCP, the membrane potential is re-established and the fluorescence decreases as ethidium is effluxed out of the cytoplasm.

The ethidium efflux assay was carried out for L51I/EmrEout, L51I/EmrEin, and a control plasmid not expressing EmrE. These results showed that the rate of ethidium efflux was ~3-fold faster for L51I/EmrEout than for L51I/EmrEin (Figure 5D), which is in agreement with the simulation results. Furthermore, the fluorescence value at which ethidium fluorescence levels off was lower for L51I co-expressed with EmrEout, indicating a reduction in ethidium accumulation relative to L51I co-expressed with EmrEin (Figure 5E). Overall, these new results support the hypothesis that the L51I/EmrEout heterodimer is able to achieve a lower cytoplasmic ethidium concentration, which enables it to grow at higher ethidium concentrations relative to L51I/EmrEin. Based on these data, we conclude that the altered kinetic rate constants between L51I/EmrEin and L51I/EmrEout give rise to the ability of the latter to confer resistance at higher ethidium concentrations. Note that previous work by Paixão et al. (Paixão et al., 2009) found that an ethidium efflux rate difference of 1.6-fold between wild-type E. coli and its efflux pump knockout strain led to a 30-fold difference in the minimum inhibitory concentration. Therefore, there is literature precedence that a somewhat modest efflux rate difference can have a deleterious effect on bacterial growth. The underlying reason stems from the fact that ethidium accumulation in the cell ultimately leads to irreversible DNA binding that impacts bacterial growth (Jernaes et al., 1994; Lambert et al., 1984).

5) The Figure 5D is confusing. Why is there a conformational preference for the inward-facing state in EmrEin/EmrEout dimer? Should not it, like WT, have no conformational preference? What is exactly depicted in the middle panel? Here, the preference should be for the outward-facing state shouldn't it? Perhaps, the figure should be remade to make the point the authors are making clearer. It might be better to show the entire transport cycle and point out which rates are likely affected by the asymmetry generated by mutations within the assemblies with EmrEin or EmrEout variants. Such depiction may also allow to clarify which conclusions are fully supported by the data and which are more speculative in nature.

Figure 5F (previously 5D) was revised to show a simple depiction that reflects the observed bias in vitro. Namely, EmrEin/EmrEout has no skewed equilibrium, the mutant/EmrEin favors the periplasmic-facing conformation, and the mutant/EmrEout favors the cytoplasmic-facing conformation. We also included the EmrE transport cycle in Figure 5G proposed by Robinson et al., 2017, that was used for the simulations. Within this figure, we indicate steps influenced by the change in equilibrium constant measured in the heterodimer samples. Rate constants we predict to be most affected are indicated with double asterisks, while more minor changes are indicated with a single asterisk.

https://doi.org/10.7554/eLife.48909.019

Article and author information

Author details

  1. Maureen Leninger

    Department of Chemistry, New York University, New York, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Writing—original draft, Writing—review and editing
    Contributed equally with
    Ampon Sae Her
    Competing interests
    No competing interests declared
  2. Ampon Sae Her

    Department of Chemistry, New York University, New York, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Writing—review and editing
    Contributed equally with
    Maureen Leninger
    Competing interests
    No competing interests declared
  3. Nathaniel J Traaseth

    Department of Chemistry, New York University, New York, United States
    Contribution
    Conceptualization, Formal analysis, Writing—original draft, Writing—review and editing
    For correspondence
    traaseth@nyu.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1185-6088

Funding

National Institutes of Health (R01AI108889)

  • Nathaniel J Traaseth

National Science Foundation (MCB1506420)

  • Nathaniel J Traaseth

National Institutes of Health (S10OD016343)

  • Nathaniel J Traaseth

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by NIH (R01 AI108889) and NSF awards (MCB 1506420) to NJT. ML was supported from a Dean’s Dissertation Fellowship from New York University. The NMR data collected with a cryoprobe at NYU were supported by an NIH S10 grant (OD016343).

Senior and Reviewing Editor

  1. Olga Boudker, Weill Cornell Medicine, United States

Reviewer

  1. Timothy A Cross, Florida State University, United States

Publication history

  1. Received: May 30, 2019
  2. Accepted: September 13, 2019
  3. Version of Record published: October 22, 2019 (version 1)

Copyright

© 2019, Leninger et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,145
    Page views
  • 245
    Downloads
  • 14
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Maureen Leninger
  2. Ampon Sae Her
  3. Nathaniel J Traaseth
(2019)
Inducing conformational preference of the membrane protein transporter EmrE through conservative mutations
eLife 8:e48909.
https://doi.org/10.7554/eLife.48909

Further reading

    1. Structural Biology and Molecular Biophysics
    Zeyu Shen, Bowen Jia ... Mingjie Zhang
    Research Article

    Formation of membraneless organelles or biological condensates via phase separation and related processes hugely expands the cellular organelle repertoire. Biological condensates are dense and viscoelastic soft matters instead of canonical dilute solutions. To date, numerous different biological condensates have been discovered; but mechanistic understanding of biological condensates remains scarce. In this study, we developed an adaptive single molecule imaging method that allows simultaneous tracking of individual molecules and their motion trajectories in both condensed and dilute phases of various biological condensates. The method enables quantitative measurements of concentrations, phase boundary, motion behavior and speed of molecules in both condensed and dilute phases as well as the scale and speed of molecular exchanges between the two phases. Notably, molecules in the condensed phase do not undergo uniform Brownian motion, but instead constantly switch between a (class of) confined state(s) and a random diffusion-like motion state. Transient confinement is consistent with strong interactions associated with large molecular networks (i.e., percolation) in the condensed phase. In this way, molecules in biological condensates behave distinctly different from those in dilute solutions. The methods and findings described herein should be generally applicable for deciphering the molecular mechanisms underlying the assembly, dynamics and consequently functional implications of biological condensates.

    1. Structural Biology and Molecular Biophysics
    Seoyoon Kim, Daehyo Lee ... Duyoung Min
    Tools and Resources

    Single-molecule tweezers, such as magnetic tweezers, are powerful tools for probing nm-scale structural changes in single membrane proteins under force. However, the weak molecular tethers used for the membrane protein studies have limited the observation of long-time, repetitive molecular transitions due to force-induced bond breakage. The prolonged observation of numerous transitions is critical in reliable characterizations of structural states, kinetics, and energy barrier properties. Here, we present a robust single-molecule tweezer method that uses dibenzocyclooctyne (DBCO) cycloaddition and traptavidin binding, enabling the estimation of the folding 'speed limit' of helical membrane proteins. This method is >100 times more stable than a conventional linkage system regarding the lifetime, allowing for the survival for ~12 h at 50 pN and ~1000 pulling cycle experiments. By using this method, we were able to observe numerous structural transitions of a designer single-chained transmembrane (TM) homodimer for 9 h at 12 pN, and reveal its folding pathway including the hidden dynamics of helix-coil transitions. We characterized the energy barrier heights and folding times for the transitions using a model-independent deconvolution method and the hidden Markov modeling (HMM) analysis, respectively. The Kramers rate framework yields a considerably low speed limit of 21 ms for a helical hairpin formation in lipid bilayers, compared to μs scale for soluble protein folding. This large discrepancy is likely due to the highly viscous nature of lipid membranes, retarding the helix-helix interactions. Our results offer a more valid guideline for relating the kinetics and free energies of membrane protein folding.