Relocation of active site carboxylates in major facilitator superfamily multidrug transporter LmrP reveals plasticity in proton interactions

The expression of polyspecific membrane transporters is one important mechanism by which cells can obtain resistance to structurally different antibiotics and cytotoxic agents. These transporters reduce intracellular drug concentrations to subtoxic levels by mediating drug efflux across the cell envelope. The major facilitator superfamily multidrug transporter LmrP from Lactococcus lactis catalyses drug efflux in a membrane potential and chemical proton gradient-dependent fashion. To enable the interaction with protons and cationic substrates, LmrP contains catalytic carboxyl residues on the surface of a large interior chamber that is formed by transmembrane helices. These residues co-localise together with polar and aromatic residues, and are predicted to be present in two clusters. To investigate the functional role of the catalytic carboxylates, we generated mutant proteins catalysing membrane potential-independent dye efflux by removing one of the carboxyl residues in Cluster 1. We then relocated this carboxyl residue to six positions on the surface of the interior chamber, and tested for restoration of wildtype energetics. The reinsertion at positions towards Cluster 2 reinstated the membrane potential dependence of dye efflux. Our data uncover a remarkable plasticity in proton interactions in LmrP, which is a consequence of the flexibility in the location of key residues that are responsible for proton/multidrug antiport.

. In addition to these carboxyl residues (in red or orange), the clusters also contain one or more aromatic residues (in blue) and polar residues (not shown). To investigate the functional role of these catalytic carboxylates and clusters, we changed the composition of Cluster 1 by removal of Glu-327 and relocation of this carboxyl residue to 6 positions (in yellow) in the interior chamber. Positions for reinsertion of Glu-327 are indicated in smaller, yellow-filled, black rectangles. One of these reinsertions, T324E, is partially hidden behind Y323 and E327Q in the figure; the N-terminus is also hidden.

Insertion of carboxyl residues in E327Q LmrP. In transport experiments in the drug-hypersensitive
Lactococcus lactis NZ9000 ∆lmrA ∆lmrCD expressing LmrP proteins, the ionophores nigericin and valinomycin were employed to selectively manipulate the composition and magnitude of the ∆p. Nigericin mediates the antiport of H + and K + down their concentration gradients, thereby selectively dissipating ∆pH in an electroneutral (membrane potential-independent) manner. Furthermore, valinomycin mediates electrogenic uniport of K + , allowing the electrophoretic uptake of K + in cells with dissipation of the ∆ψ 14 . ATP-depleted cells were first pre-equilibrated with 2 μ M propidium or ethidium, after which the efflux reaction was initiated with the addition of 25 mM glucose in the absence and presence of valinomycin and/or nigericin. The control experiments in Fig. 2A and B show that WT LmrP mediates ∆ψ and ∆pH-dependent substrate/proton antiport for both propidium and ethidium. As the insertion of extra carboxyl residues with additional proton coupling would yield LmrP mutants whose transport reaction is dependent on ∆ψ and ∆pH as seen for WT LmrP, such mutants might not be readily detectable. Therefore, we performed our insertion studies in the His 6 -tagged LmrP mutant in which the catalytic carboxyl residue Glu-327 is replaced by the corresponding neutral amide (E327Q). This mutant was found to mediate electroneutral proton-substrate antiport for propidium, but electrogenic transport for ethidium (Fig. 2C,D). These data have been obtained before 10 and are in line with the current results.
An additional carboxyl residue was then introduced in various locations on the surface of the internal binding cavity in E327Q LmrP. Locations Thr-33, Ala-53, Asn-123, Thr-324, Thr-331 and Ala-355 were selected using the inward-facing three-dimensional LmrP homology model that was previously constructed based on the crystal structure of the glycerol-3P/Pi antiporter GlpT 11 (Fig. 1A,B). Positions were chosen with respect to their proximity to the catalytic carboxylates Asp-235 (located in Cluster 1) and Asp-142 (located in Cluster 2). In our structure model, (i) position 324 is located in closer proximity to Cluster I [Glu-327 (~4.  . Ionophores nigericin or valinomycin or both were added 3 min prior to the addition of glucose to measure transport in the presence of ∆ψ or ∆pH only, or without ∆p (Control), respectively. The normalised fluorescence start level before glucose addition is set at 100% to facilitate direct comparison of data for LmrP mutants with different dye preloading activities. Histograms display the mean ± s.e.m. of dye fluorescence end levels in 3 independent experiments with independently prepared batches of cells compared to control for which the ∆p was absent (one-way analysis of variance; ****P < 0.0001), and demonstrates the reproducibility of data obtained in these assays. Inset in (E) shows the same data as the main figure at a longer time scale. the background of E327Q LmrP. Western blot analysis of total membrane proteins from inside-out membrane vesicles showed that except for A53E/E327Q LmrP, which was expressed at 43% of the expression level of WT LmrP, the other mutants showed equal amounts of accumulated protein in the plasma membrane compared to WT LmrP (Fig. 3A).
To examine the energetics of LmrP E327Q mutants with an inserted carboxyl residue, propidium efflux in intact cells was measured in the absence or presence of ionophores as described for Fig. 2. The introduction of E at position 331 substantially restored electrogenic propidium transport pointing to the ability of the T331E/E327Q double mutant to utilise both the ∆ pH and the ∆ ψ in the efflux reaction (Fig. 3B). In contrast, the double mutants A355E/E327Q and N123D/E327Q showed propidium efflux activity in the presence of either the ∆ p or the ∆ pH, but negligible transport activity in the presence of the ∆ ψ only (Fig. 3C,D). These mutants therefore display the same phenotype as the single mutant E327Q background in which the mutations were generated. The double mutants T324E/E327Q, T33E/E327Q and A53E/E327Q (Fig. 3E-G) did not mediate detectable propidium efflux in the presence of ∆ p or its components. As all mutants were expressed in the plasma membrane (Fig. 3A), these data suggest that an acidic residue in position 33, 53 and 324 somehow interferes with the propidium transport reaction. Insertion of carboxyl residues in E327A LmrP. To examine changes in ethidium transport for the carboxylate insertion mutants, Glu-327 was replaced by A in His 6 -tagged LmrP. In control experiments, the single mutant E327A showed active propidium and ethidium efflux in the presence of the ∆p (interior negative and alkaline) or its ∆pH component, while transport activity was reduced to a negligible level with the ∆ψ only (Fig. 2E,F). Thus, E327A LmrP mediates electroneutral transport of these cationic dyes, and provides an ideal background to test for carboxylate insertions that restore the ∆ψ dependence in ethidium transport.
The carboxylate substitutions at positions 324, 355 and 331 were selected for further experiments as they represent the three categories of mutants that were encountered in the propidium transport experiments in E327Q LmrP (Fig. 3): transport-inactive, tolerated but silent, and catalytically-active relocations of Glu-327, respectively. Double mutants T331E/E327A and A324E/E327A were equally well expressed as WT LmrP in the plasma membrane of L. lactis, whereas A355E/E327A was expressed at 68% of the WT level (Fig. 4A). To test whether the introduction of the carboxyl residues on the surface of the internal cavity can restore the ability of the E327A mutant to mediate ∆ ψ -dependent ethidium transport, similar substrate efflux assays were carried out as described for the E327Q proteins (Fig. 3). In experiments with propidium, comparable results were obtained in the E327A background as for E327Q ( Fig. 4B-D). For ethidium transport, the introduction of E at position 331 and 355 in E327A LmrP, each supported ∆ ψ -dependent transport (Fig. 4E,F). Although A355E/E327A showed a weaker activity in ∆ ψ -dependent ethidium transport compared to T331E/E327A, the activity of A355E/E327A was significant compared to E327A LmrP ( Fig. 2F) (P < 0.01). Mutant T324E/E327A transported ethidium (Fig. 4G) in a comparable fashion as E327A LmrP but at lower rates ( Fig. 4B) suggesting that T324E is not catalytically active in ethidium transport in the E327A background but that the replacement is tolerated. Taken together, T331E and A355E are the only carboxyl residues in our test panel that impose electrogenicity on transport, for T331E, of propidium and ethidium transport, and for A355E, of ethidium transport only. These carboxylates are therefore most likely functional in proton coupling with these substrates.
The effect of the T331E replacement on the affinity of E327A LmrP for propidium and ethidium was evaluated by measuring the kinetics of facilitated substrate influx in ATP-depleted cells. Following the addition of propidium or ethidium at concentrations ranging from 1 μ M to 30 μ M, the rates of substrate uptake over the initial 30 s were determined. These rates were corrected for dye influx in control cells without LmrP expression, and were subsequently plotted against the substrate concentration ( Fig. 4H,I). The propidium binding affinity of T331E/E327A LmrP (expressed as the substrate concentration K t giving a half-maximal rate of transport V max ) was similar in comparison to WT LmrP Similarly, no significant difference was observed for ethidium in this comparison (K t = 8.1 ± 5.3 μ M versus 7.7 ± 5.1 μ M). Furthermore, the V max values for dye uptake by T331E/E327A LmrP were similar to those obtained for WT LmrP (propidium, As the rates of facilitated dye uptake by E327A LmrP were too low for kinetic analysis, the data suggest that insertion of E at position 331 in the E327A background restores the transport capabilities of the single mutant to those of WT LmrP, and hence, that T331E compensates for the loss of E at position 327. T331E substitution in WT LmrP. As a carboxyl residue at position 331 is functionally active in drug/proton antiport in E327Q and E327A mutants, the effect of this replacement was also tested in the WT background (Fig. 5). The T331E mutant was generated in His 10 -tagged LmrP, and expressed to a similar level as WT (Fig. 5A). It was found that T331E LmrP mediates propidium efflux to a lower steady-state level compared to WT LmrP ( Fig. 5B), which is consistent with the increased propidium resistance of T331E LmrP-expressing cells compared to WT LmrP-expressing cells (Fig. 5D). For ethidium, T331E LmrP showed a similar transport as WT LmrP (Fig. 5C). To delineate whether the enhanced propidium transport by T331E LmrP was due to enhanced proton coupling and/or enhanced interaction with propidium, the binding affinity of T331E LmrP for propidium was first measured using the facilitated influx assay in lactococcal cells as described for T331E/E327A LmrP (Fig. 4H). The K t of T331E LmrP and WT LmrP for propidium was found not to be significantly different, 1.2 ± 0.4 μ M and 1.3 ± 0.3 μ M, respectively. Furthermore, propidium binding to purified T331E LmrP and WT LmrP was measured by fluorescence anisotropy (Fig. 5F). In this assay, propidium at a fixed concentration of 1 μ M or 0.2 μ M was titrated with increasing concentrations of purified protein in detergent solution until a steady level of fluorescence anisotropy was achieved. These fixed ligand concentrations are in the same range or 5 to 10-fold below the apparent K t of Wt LmrP and T331E LmrP for propidium, and hence, enable propidium binding in a protein concentration-dependent, hyperbolic fashion (Fig. 5F). As the measurements yielded binding data for Wt LmrP and T331E LmrP that were overlapping, these data indicate that there was no significant difference in the propidium binding affinity of both proteins.
The effect of the insertion of T331E in WT LmrP on the energetics of transport was tested in proteoliposomes in which purified LmrP or T331E LmrP were functionally reconstituted at equal levels in the phospholipid bilayer (Fig. 6A). In this system, dye accumulation was monitored by fluorimetry through inclusion of 1 mg/ml sheared calf thymus DNA in the lumen of the liposomes 6 . For the imposition of the ∆ p, a transmembrane ∆ pH (interior acidic) and ∆ ψ (interior positive) were imposed artificially by the preparation of the (proteo) liposomes in NH 4 SCN containing buffer (pH 6.8), and subsequent 100-fold dilution in NH 4 SCN-free buffer (pH 7.5), thus allowing the passive diffusion of NH 3 and SCN − to the external buffer 6 . With the addition of 2 μ M propidium to the external buffer, a significantly higher ∆ p-dependent accumulation of propidium was obtained in proteoliposomes containing T331E LmrP compared to WT LmrP, whereas the no gradient controls showed a similar low background-level fluorescence (Fig. 6B). In contrast, when 2 μ M ethidium was added in the assay a similar level of ethidium accumulation was observed for both T331E and WT LmrP compared to the no gradient controls (Fig. 6C). The basic thermodynamic relationship between the driving force and the propidium concentration gradient across the proteoliposomal membrane at equilibrium is given by:  in which Z = 2.3RT/F = approx. 59 mV at 25 °C, and n is the number of H + that participates in the antiport reaction. The elevated propidium accumulation level in proteoliposomes for ∆ p-dependent transport by T331E LmrP compared to WT LmrP (Fig. 6D) therefore indicates a higher value of n for T331E LmrP in the propidium transport reaction. This conclusion is also in agreement with propidium efflux measurements in cells showing efflux to lower end levels for T331E LmrP compared to WT LmrP (Fig. 5B). Taken together, these data suggest that the enhanced propidium transport activity of T331E LmrP compared to WT LmrP is not due to enhanced interaction of the mutant with propidium but that the extra carboxyl residue introduced at position 331 functions as an additional proton-binding site in the propidium transport reaction.

Discussion
Although previous work identified three catalytic carboxylates in LmrP (Asp-142, Asp-235 and Glu-327), it was suggested that these carboxylates only form two proton binding groups in the transport reaction for monovalent ethidium 10 . In the presence of ethidium, Glu-327 and Asp-235 in wild-type LmrP are proposed to form a carboxyl-carboxylate pair containing Asp-235 as a single proton acceptor moiety that is stabilized through hydrogen bonding with the adjacent undissociated carboxyl moiety in Glu-327 or amide moiety in E327Q. This stabilization is likely to elevate the pKa of Asp-235 to a value closer to the physiological intracellular pH where Asp-235 becomes a functional proton binding group. Therefore, the Δ p dependence of ethidium transport is similar for wildtype (Glu-327) and the E327Q mutant in the experiments in Fig. 2B and D, and reflects apparent electrogenic 2H + /ethidium + antiport. However, when E327Q is replaced by E327A the lower pKa of Asp-235 would disable this carboxyl group in proton binding under physiological conditions, yielding apparent electroneutral 1H + /ethidium + antiport as observed in our data (Fig. 2F). When divalent cationic propidium is bound on the surface of the interior chamber of LmrP, Glu-327 and Asp-235 might not directly interact with each other due to changes in the distance and geometry of their side chains compared to the ethidium-bound state, and both carboxyl residues now function as independent proton acceptor moieties 10 . Such changes in side-chain interactions in response to the binding of different substrates have been observed in co-crystal structures of the multidrug binding transcriptional repressor QacR 15,16 . Together with Asp-142, the independent activities of Asp-235 and Glu-327 in LmrP now result in apparent electrogenic 3H + /propidium 2+ antiport. Consistent with the experimental data for E327Q (Fig. 2C) and E327A (Fig. 2E), this stoichiometry is reduced to apparent electroneutral 2H + / propidium 2+ antiport upon the loss of Glu-327.
Using rational design, we modified the relative distance between the membrane-embedded carboxyl residues in the interior chamber of LmrP. Out of the 6 tested positions, T331E recovered WT-like electrogenic transport for propidium and ethidium in E327A LmrP, whereas A355E recovered electrogenic transport for ethidium but not for propidium in this background. The T331E substitution also recovered WT-like propidium transport capabilities in the E327Q background. These findings point to a functional redundancy between T331E and Glu-327. In these experiments, the deleted Glu-327 carboxylate migrates from being close to Asp-235 (~4.7 Å) to positions where it has a large distance from Asp-235, ~11.3 Å in the case of T331E, and ~17.4 Å in the case of A355E, and where this carboxylate is functionally active in proton coupling. This result demonstrates the plasticity in the location of proton-binding moieties in LmrP. As the T331E and A355E mutations involve the relocation of Glu-327 from Cluster 1 towards Cluster 2, our data point to the existence of an area in the interior chamber, with Cluster 1 and 2 at its boundaries, in which carboxylates are active in proton-coupling.
The transport measurements indicate that the insertion of T331E in WT LmrP allows the mutant transporter to sustain a larger inwardly directed propidium concentration gradient in cells, and outwardly directed propidium concentration gradient in proteoliposomes in which the transporter is inserted in an inside-out fashion. Together with the observations on ∆ p-dependent propidium transport in proteoliposomes, these results strongly suggest that T331E acts as proton binding group in addition to the existing Asp-142, Asp-235 and Glu-327. The established stoichiometry of 3H + /propidium 2+ for WT LmrP 10 will therefore most likely increase to 4H + /propidium 2+ . This conclusion is consistent with recent observations on the MFS transporter MdfA from E. coli showing that an increase in proton/drug stoichiometry requires an additional acidic residue in the multidrug recognition pocket of this transporter 17 . As the insertion of T331E in Wt LmrP did not enhance the ethidium concentration gradients in cells and proteoliposomes, T331E might share a functional similarity with Glu-327 in that neither of these carboxylates appears to play an active role in proton coupling during the ethidium transport reactions by T331E LmrP and Wt LmrP (see discussion for Glu-327 in Wt LmrP above). Our detailed understanding of these phenomena will require the high-resolution structural elucidation of the ethidium-LmrP and propidium-LmrP complexes in which the side-chain interactions of Asp-142, Asp-235, Glu-327 and T331E in inward-facing and outward-facing states of LmrP are resolved.
Similar to the native catalytic carboxylates Glu-327 and Asp-235, the newly introduced T331E and A355E are located in the C-terminal half of LmrP. As none of the other substitutions at the positions 33, 53 and 123 in the N-terminal half was catalytically active, the data suggest that although, like MdfA, LmrP is promiscuous regarding the positions where carboxyl residues can be functionally inserted, a bias might exist towards positions in the C-terminal half. This finding is reminiscent of previous observations on the MFS oxalate transporter OxlT, a GlpT homolog from Oxalobacter formigenes 18 . The ligand-binding residues in OxlT all lie in the C-terminal domain (on TM8 and TM11), whereas those in the glycerol-phosphate/phosphate antiporter GlpT lie at symmetrical positions (on TM1 and TM7) at the interface between the N-and C-terminal halves 19 .
Secondary-active multidrug transporters can confer drug resistance on the cell by facilitating active drug export across a plasma membrane that provides a barrier for passive influx of drugs 20 . Drugs can be expelled against their concentration gradient or lipid-water partition coefficient by coupling drug extrusion to the movement of protons and/or sodium ions down their electrochemical gradient. Our understanding of the mechanistic principles behind this biochemical coupling process requires knowledge of the ion translocation pathway(s), as well as the link between ion movement and conformational changes within the transporter that result in alternating access of drug binding surfaces. The mechanism of LacY has been intensely studied [21][22][23] , and was found to rely on proton binding to critical carboxylates, the dissociation of which is based on competitive binding of the cationic side chain of a neighbouring basic residue. The location of these catalytic carboxylates is therefore very precise and crucial for transport activity. A similar type of mechanism is also relevant for proton coupling in the trimeric RND multidrug transporter AcrB, where Lys-940, Arg-971, Asp-407 and Asp-408 are part of a proton relay network within the membrane domain of each AcrB protomer. This network is functionally coupled to drug binding at, and drug translocation through, the periplasmic domains in the AcrB trimer 24 . Previous studies on basic residues in LmrP highlighted the important role of Arg-260 and Lys-337 in the functional properties of this transporter 25 . Although these residues are predicted in our structure model to be oriented towards the Scientific RepoRts | 6:38052 | DOI: 10.1038/srep38052 head group region of the outer leaflet (Arg-260) and inner leaflet (Lys-357) of the cytoplasmic membrane, helical rotations during the transport cycle or a small 1 to 2 residue shift of the polypeptide in our structure model could place these basic residues in the central cavity where they might directly interact with the catalytic carboxylates. However, the protonation state of the catalytic carboxylates in LmrP could also be regulated by mechanisms that do not require the proximity of basic residues, e.g. by conformation-dependent changes in the physico-chemical properties of the local environment near the catalytic clusters. If relevant, the lack of AcrB and LacY-type stereospecific proton dissociation mechanisms in LmrP could be one of the reasons why this transporter tolerates relocations of carboxylates between cluster regions in which these carboxylates are catalytically active in proton/ drug antiport.
In conclusion, our data suggest that LmrP exhibits a plasticity in proton interactions due to the intrinsic flexibility in the location of the key residues that are responsible for proton/drug exchange. This flexibility enables LmrP to rapidly evolve its polyspecificity through the acquisition of mutations while retaining proton/drug antiport activity.

Methods
Bacterial strains, protein expression, and site-directed mutagenesis. The His 6 -tagged lmrP wild type and mutant genes were cloned in the pNZ8084 plasmid downstream of the nisA promoter region 7,26 . Plasmids were maintained in the drug-hypersensitive L. lactis NZ9000 Δ lmrA Δ lmrCD 27 lacking endogenous multidrug efflux pumps, which was grown at 30 °C in M17 broth (Oxoid) supplemented with 25 mM glucose and 5 μ g/ml chloramphenicol. The cells were grown to an OD 660 of 0.5-0.6 after which the protein expression was induced for 2 h by the addition of a 1:1000 dilution of nisin A-containing supernatant of the nisin-producing L. lactis strain NZ9700 26 (approx. final concentration of 10 pg/ml of nisin A). The construction of lmrP mutants was carried out using the QuikChange method (Agilent Technologies) with mutagenic primer pairs as listed in Table 1. DNA was sequenced to ensure that only the intended changes were introduced. Transport measurements in intact cells. Cells were harvested by centrifugation at 6,500 x g for 10 min at 4 °C and washed once with 50 mM KPi (pH 7.0) containing 5 mM MgSO 4 . Harvested cells were then de-energised in the presence of 0.5 mM of the uncoupler dinitrophenol for 30 min at 30 °C to deplete ATP 28 . The de-energised cells were subsequently washed 3 times with ice-cold buffer, resuspended to an OD 660 of 5.0, and kept on ice to maintain their viability. For monitoring active efflux, cells were diluted in prewarmed buffer to OD 660 of 0.5, and equilibrated with 2 μ M of either ethidium or propidium (fluorescence level of ~150 A.U. was obtained for propidium and up to ~250 A.U. for ethidium). Once the saturation level was obtained, 25 mM glucose was added as the source of energy and active efflux of the dye was measured by fluorimetry. The data were normalised by taking the fluorescence start level before the addition of glucose as 100% for direct comparison of data for LmrP mutants with different dye preloading activity. Where indicated, valinomycin was added 3 min before glucose addition at a concentration of 0.5 μ M. Nigericin was added at a concentration of 0.5 μ M for E327Q mutants and 1 μ M for E327A mutants. Fluorescence measurements were performed in a Perkin Elmer LS-55B at λ excitation of 500 nm and λ emission of 580 nm with slit widths of 5 and 10 nm, respectively, for ethidium, and λ excitation at 535 nm and λ emission at 617 nm with slit widths of 10 nm each for propidium. For measurements of the kinetics of facilitated dye influx, cells were diluted 10-fold in 2 ml of buffer. Propidium or ethidium was added to final concentrations between 1 μ M and 10 μ M, and influx was monitored over the first 100 s.
Preparation of inside-out membrane vesicles and protein purification. Inside-out membrane vesicles were prepared using as Basic Z 0.75-kW Benchtop Cell Disruptor as described 10 . The expression levels of the his-tagged LmrP proteins in the lactococcal plasma membrane vesicles were compared on immunoblot that was probed with primary mouse anti-polyhistidine tag antibody (Sigma-Aldrich, Cat. No.: H1029) and secondary goat anti-mouse antibody (Sigma-Aldrich, Cat. No.: A4416). LmrP signals were quantified by densitometric analysis of three independent blots using ImageJ software, version 1.43 (National Institutes of Health).  Transport measurements in proteoliposomes. Purified LmrP WT or LmrP T331E protein was functionally reconstituted into liposomes by described methods 6 in preparation buffer (pH 6.8) containing 10 mM K-Hepes, 100 mM K 2 SO 4 , 100 mM NH 4 SCN, and 1 mg/ml of sonicated calf thymus DNA (Trevigen). Purified protein was added at a lipid:protein ratio of 50:1 (w/w) to Triton X-100-destabilised liposomes, and the mixture was incubated for 30 min at room temperature. Detergent was removed by incubation with SM2 biobeads as described 6 . Subsequently, the proteoliposomes were incubated for 15-20 min with DNase (10 μ l/ml) in the presence of 10 mM MgSO 4 to remove any DNA contamination from the lipid bilayer, collected by centrifugation at 165,145 × g for 30 min, resuspended in 150-200 μ l of preparation buffer, and used immediately in transport assays. For measurements of ethidium and propidium transport in (proteo)liposomes in the presence of an imposed Δ p (inside acidic and negative), the proteoliposomes were diluted 100-fold in dilution buffer (pH 7.5) containing 10 mM K-Hepes and 100 mM K 2 SO 4 in a total reaction volume of 2 ml. Ethidium or propidium was added to a final concentration of 2 μ M, and transport was monitored in a Perkin Elmer LS-55B fluorimeter.
Fluorescence anisotropy. Protein purification was done as described under "Preparation of inside-out membrane vesicles and protein purification" and the purified protein was diluted to a concentration of 0.4 mg/ml in 2 ml of 50 mM KPi buffer (pH 7.0) supplemented with 100 mM NaCl, 0.05% DDM, and 1 μ M or 0.2 μ M propidium. For propidium binding measurements, the purified protein was added in 3 steps of 3.45 μ g each, followed by 1 step of 7.0 μ g, 3 steps of 17.3 μ g, 1 step of 3 μ g and 1 step of 17 μ g. Equivalent amounts of elution buffer (Buffer C under "Preparation of inside-out membrane vesicles and protein purification") were added in a protein-free control sample.
Drug resistance. L. lactis was grown in M17 medium supplemented with 25 mM glucose and 5 μ g/ml chloramphenicol. After 16 h of growth at 30 °C, cell cultures were used to inoculate fresh medium at a 1:50 (v/v) dilution. At OD 660 of 0.5, protein production was initiated through the addition of 10 pg/ml nisin. Protein expression was induced for 2 h, after which the culture was diluted 1:10 (v/v) in fresh medium supplemented containing 10 pg/ml nisin in the wells of a 96-well plate. Propidium was then added at concentrations as indicated in Fig. 5D. The wells were covered with 25 μ l of sterile mineral oil to reduce evaporation of water. The growth of cultures in the wells at 30 °C was recorded by measuring the OD 660 in a Versamax microplate reader (Molecular Devices) over a time period of 16 h. Cultures were grown in triplicates for each propidium concentration. The slopes of the exponential phase in growth curves were determined from the linear increase in a log10(OD 660 ) versus time plot, and were plotted against the corresponding propidium concentration in Fig. 5D.
Statistical analyses. The data points plotted for immuno blots and fluorescence studies are obtained in at least