Molecular basis for the cation selectivity of Salmonella typhimurium melibiose permease

Abstract

Cation selectivity and coupling are important attributes of cation-coupled symporters. Salmonella typhimurium melibiose permease (MelB_St) catalyzes the co-transport of galactosides with cations (H⁺, Li⁺, or Na⁺). 3-D crystal structures of MelB_St have revealed the molecular recognition for sugar substrates, but the cation binding and coupling mechanisms have not been defined to atomic levels. In its human homolog MFSD2A, a lethal mutation was mapped at its Na⁺-binding pocket; however, none of the structures in this subfamily resolved its cation binding. In this study, molecular dynamics simulations showed the binding interactions of Na⁺ and Li⁺ to MelB_St. Interestingly, Thr121, the lethal mutation position in MFSD2A, forms stable interaction with Na⁺ but is at a distance from Li⁺. Most mutations among 11 single-site Thr121 mutants of MelB_St exhibited little effects on the galactoside binding, but largely altered the cation selectivity with severe inhibitions on Na⁺ binding. Few mutants (Pro and Ala) completely lost the Na⁺ binding and Na⁺-coupled transport, but their Li⁺ or H⁺ modes of activity were largely retained. It can be concluded that Thr121 is necessary for Na⁺ binding, but not required for the binding of H⁺ or Li⁺, so a subset of the Na⁺-binding pocket is enough for Li⁺ binding. In addition, the protein stability for some mutants can be only retained in the presence of Li⁺, but not by Na⁺ due to the lack of affinity. This finding, together with other identified thermostable mutants, supports that the charge balance of the cation-binding site plays an important role in MelB_St protein stability.

Introduction

The melibiose permease of Salmonella typhimurium (MelB_St) is a secondary active transporter catalyzing stochiometric symport of galactopyranoside with a Na⁺, H⁺ or Li⁺ (1,2). MelB is a prototype of the major facilitator superfamily subgroup 2 (MFS_2) (3,4) among the largest MFS membrane transporters (5,6). MelB homologs in human play important roles from lipid uptake such as the Na⁺-dependent phospholipids transporter MFSD2A (7) to cancer development (8,9). The high-resolution structures for MelB_St (3,10) were solved, and the structures for two eukaryotic homolog MFSD2A proteins have been also resolved (11,12). Structurally, MelB is composed of 12 transmembrane α-helices organized into two pseudo-symmetric bundles (Fig. 1a, b). The outward-facing structures bound with α-nitrophenyl galactoside (α-NPG) or dodecyl melibioside (DDMB) showed that the sugar specificity determinant pocket hosts the specific galactosyl moiety (3).

Fig. 1. Sugar-bound outward-facing crystal structure of D59C MelB_St.

(a) Overall fold and helical packing [PDB ID, 7L17]. A cartoon representation of the D59C MelB_St structure bound with α-NPG is viewed parallel to the membrane, shown in cartoon representation and superimposed with a surface representation. The cytoplasmic middle loop is colored in yellow. Transmembrane helix is numbered in Roman numerals. The bound α-NPG is colored in yellow; the position Asp59 on helix II is shown in ball and stick. (b) Viewed from cytoplasmic side. (c) The galactoside binding and cation-binding pocket. The bound α-NPG molecule is colored in yellow, and the major hydrogen-bonding interactions with MelB_St are indicated in broken line. All sugar-binding residues are labeled in pink color and the transmembrane helices are shown in parentheses brackets. A loose cation-binding pocket was proposed for this D59C mutant, and all possible residues for Na⁺ binding are labeled in black.

This sugar specificity determinant pocket is formed by both of the N- and C-terminal helical bundles involving in five helices (I, IV, V, X, and XI), particularly the helices I and IV (Fig. 1b). Two negatively charged Asp residues at positions 19 (helix I) and 124 (helix IV), and two positively charged residues Lys18 (helix I) and Arg149 (helix IV), as well as Trp128 (helix IV), form multiple hydrogen-bonding interactions with all four hydroxyl groups on the galactosyl moiety. All Cys mutants at position 18, 19, 124, or 128 completely abolished all three-modes melibiose active transport coupled to H⁺, Li⁺, or Na⁺(10).

The D59C MelB_St with the mutation at its cation-binding site has been determined as an uniport mutant (Fig. 1). This single-site Cys mutation selectively abolishes the cation binding and all three modes of melibiose active transport, but still catalyzes melibiose transport and support melibiose fermentation (3), as well as the melibiose exchange (10). Interestingly, D59C MelB_St exhibits superior protein stability and sugar-binding affinity (3). Both Asp55 and Asp59 (helix II) have been shown to contribute to Na⁺ binding (13-18). In the structure [pdb id,7L17], the position 59 is much less solvent exposed, and Asp55 is involved in a salt-bridge with Lys377 (helix XI). Single-Cys mutants on Asn58 (helix II) or Thr121(helix IV) have been shown to selectively inhibit Na⁺-coupled melibiose transport with less or no inhibition on the transport coupled to Li⁺ (10); in addition, the MelB ortholog of Klebsiella pneumoniae (MelB_Kp) uses H⁺ or Li⁺ (but not Na⁺) as the coupling cation (19-21), where Ala residue is present on position Asn58. In the structure, Thr121 is in close proximity to both Asp55 and Asp59, and the Asn58 side chain is extended toward to the membrane. Since the D59C mutant, which was used for the structure determination, does not bind a cation, the cation binding was not resolved and the detailed coordination is unknown in MelB_St (Fig. 1c); interestingly, the cation site directly connects to the bound galactosyl moiety. The helix IV seems to host both cation- and sugar substrate-binding sites, which can dictate the coupling between the two substrates and contribute to the observed cooperative binding (18,22-24).

In this study, we performed molecular dynamics simulations to characterize the Na⁺ and Li⁺ binding to MelB_St and focused on the less-characterized Thr121 (helix IV). As showed previously, a single-site T121C mutant selectively inhibited melibiose transport coupled to Na⁺ or H⁺ cations, but with no obvious inhibition on the Li⁺-coupled transport (10). Bio-informatics studies show that this Thr residue is evolutionally strictly conserved even across those functionally unrelated, likely cation-coupled transporters (3), including the MFSD2A. The potential importance of Thr121 in the cation-binding site was also indicated by a lethal mutation on the identical position T159M in MFSD2A (25). While three structures of Na⁺-coupled MFS transporters have been determined, none solved the cation binding problem. Here we presented evidence that Thr121 is part of the cation-binding site of MelB_St and it is essential for Na⁺ binding, but not for H⁺ or Li⁺ binding. In addition, this study also provided important information to support the argument that the charge balance of this cation-binding pocket plays an important role in MelB_St protein stability.

Results

Binding of Na⁺ or Li⁺ with MelB_St by molecular dynamics simulations

To simulate the Na⁺ and Li⁺ binding, the WT MelB_St derived from the D59C structure [PDB, 7L17] was subjected to all-atoms MD simulations embedded in a lipid environment consisting of POPE and POPG with a ratio similar to that in the E. coli membranes (26). After nearly 1 μs of MD simulations of MelB_St in complex with melibiose and either Na⁺ or Li⁺, the overall protein structure remains stable. The Na⁺ is stably coordinated by 5 coordinating oxygen atoms from 4 residues (Asp55, Asn58, Asp59, and Thr121) with an average coordination bond length of 2.3 Å (Fig. 2a). The Asp55 and Asp59 are stabilized by forming salt-bridges with Lys377 (helix XI), and Thr121 is stabilized though hydrogen bonds with Gly117. The Li⁺ is stably coordinated by 4 coordinating oxygen atoms mainly from only 3 residues (Asp55, Asn58 and Asp59) with an average coordination bond length of 2.0 Å (Fig. 2ab). In contrast to Na⁺, the coordination of Li⁺ by the Thr121 is less stable.

Fig. 2. Cation-binding sites for Na⁺ or Li⁺ simulated by MD simulations.

All-atom MD simulations of the Asp59C structure after changed to WT were performed in a lipids environment as described in the Methods. Na⁺ (a) and Li⁺ (b) were manually placed in the binding site. (a & b) The structures of the binding site of Li⁺ and Na⁺ were taken from the two 1 μs MD trajectories. The Na⁺ and Li⁺ are represented by the yellow and purple balls, respectively, and the side chains of neighboring atoms are represented by sticks. The backbone of the protein is represented as ribbons. The coordination interactions between the cation and the neighboring residues are indicated by dashed lines with the separation distances labeled. (c, d) Distribution of distance. During the 1-μs simulations, Thr121 side chain flipped between the bound Li⁺ and Asp59. Distribution of the closest distance between the hydroxyl oxygen atom on the side chain of the Thr121 (OG1(T121)) and the Na⁺ or Li⁺ ions. Distribution of closest distance between the hydroxyl hydrogen atom on the side chain of the Thr121 (HG1(T121)) and the carboxyl oxygen atoms of the Asp59 (OD(D59)) (d). The distributions for the Li⁺ and Na⁺ bound MelB_St are colored in red and blue, respectively. The Thr121 stably binds the Na⁺ but not the Li⁺ during the 1 μs MD simulation.

The Li⁺ binding to MelB_St is different from Na⁺. Li⁺ is closer to the Asp55, Asn58 and Asp59 on helix II. The probability distribution of the closest distance between the hydroxyl oxygen on the side chain of Thr121 (OG1(T121)) and the Li⁺ or Na⁺ ions revealed that there is a single peak for Na⁺-OG1 distances, but there are two peaks for Li⁺-OG1 distance (Fig. 2c). The first peaks are located at 2.1 Å and 2.4 Å for the Li⁺ and Na⁺, respectively, which correspond to direct coordination between the OG1 atom of Thr121 and the cation. The Na⁺-OG1 distance at the first peak is slightly larger than the Li⁺-OG1 distance due to the larger radius of the Na⁺. Importantly, in the distance distribution of Li⁺-OG1, there is a second peak at ~ 4.1 Å, which arise from the breakage of the coordination interaction between the Li⁺ and the OG1 atom of Thr121. In other words, during the 1 μs simulation, there is significant amount of time when the Thr121 did not coordinate the Li⁺. In contrast, the second peak is absent in the distance distribution of Na⁺-OG1, which indicates stable coordination of the Na⁺ by the Thr121 throughout the 1 μs simulation. A closer inspection into the dynamics of the Li⁺-bound MelB_St revealed that the hydroxyl group on the side chain of Thr121 frequently donates hydrogen bond to the carboxyl group of the Asp59. To quantify the observation, the distribution of the closest distance between the hydrogen atom on the hydroxyl group of the Thr121 (HG1(T121)) and the carboxyl oxygen atoms (OD(D59)) was analyzed. For the Li⁺-bound MelB_St, there are two peaks for the distances between HG1 of Thr121 and OD of Asp59. The first peak appears at ~1.7 Å, corresponding to the hydrogen bond distance between the HG1 and the OD atoms. This peak is missing in the Na⁺-bound MelB_St (Fig. 2d). It is likely that Li⁺ is favorably attracted by Asp55, Asp59 and Asn58 so that the Thr121 is released to form hydrogen bond with Asp59.

Site-directed mutagenesis of MelB_St Thr121

To further characterize the role of Thr121, a total number of 11 single-site mutants on Thr121 were constructed using the WT MelB_St as the template. Non-polar residues (Ala, Gly, Pro, or Leu including Met, the lethal mutation at the homolog site of human MFSD2A) and polar residues (Ser, Asn, or Cys) were generated to test if the hydroxyl group on the Thr sidechain and its precise positioning are important. Asp and Arg mutants were included to test how a negative charge or bulky positive charge affect the cation binding. The bulkiest Trp residue was for testing the volume effect.

Protein expression

The membrane expression levels of the MelB_St WT and mutants in the E. coli DW2 cells was examined by western blotting. As judged by an anti-His tag antibody, all Thr121 mutants exhibited similar levels of membrane expression, with Cys, Gly, and Ala mutants slightly better than others; however, as shown in the histogram, all mutations inhibited the protein expressions by about 50% when compared to the WT (Fig. 3; Table 1).

Fig. 3. Membrane expression and western blotting.

Membranes containing the WT and mutant MelB_St in a buffer containing 50 mM NaPi, pH 7.5, 100 mM NaCl, 20 mM melibiose and 10% glycerol were prepared as described in Methods. An aliquot of 20 μg membranes of each sample were analyzed by SDS-15%PAGE, and MelB_St proteins were detected by western blotting using anti-His tag antibody as described in Methods.

Table 1.

Summary

		WT	Ser	Cys	Ala	Pro	Gly	Asn	Asp	Leu	Met	Arg	Trp
Expression (%)a		100	51.83 ± 2.3*	59.05 ± 7.3	56.41 ± 4.4	46.73 ± 5.4	57.89 ± 2.7	50.97 ± 9.6	45.89 ± 7.1	43.74 ± 7.0	47.76 ± 9.1	45.26 ± 4.2	44.69 ± 5.4
Fermentationb		Good	Poor	Good	S	S	S	Good	Good	Good	Good	No	No
Transport at 30 min (%)c	H+	100	47.8 ± 10.69	16.5 ± 2.01	34.0 ± 3.41	106.8 ± 4.54	18.5 ± 2.28	8.7 ± 2.83	8.0 ± 2.66	4.4 ± 1.42	49.6 ± 11.56	−1.0 ± 0.78	0.4 ± 0.12
	Na+	100	22.8 ± 0.18	23.0 ± 1.21	3.7 ± 0.25	11.3 ± 0.13	9.1 ± 1.10	2.5 ± 0.28	2.7 ± 0.12	1.0 ± 0.11	30.5 ± 4.02	0.5 ± 0.27	0.2 ± 0.12
	Li+	100	84.7 ± 1.18	75.3 ± 0.55	93.6 ± 5.48	109.2 ± 2.39	41.7 ± 4.15	22.5 ± 0.76	21.9 ± 0.83	15.0 ± 1.86	31.0 ± 5.09	−0.5 ± 0.39	0 ± 0.05
Sugar binding (IC50, mM)d	H+	+	+	+	+	+	+	+	+	+	+	+	−
	Na+	+	+	+	+	+	+	+	+	+	+	+	−
	Li+	0.73 ± 0.04	1.99 ± 0.12	2.58 ± 0.03	2.48 ± 0.04	5.77 ± 0.52	+	+	+	+	+	+	−
Na+ binding (K0.5(Na+), mM)e		0.83 ± 0.25	8.24 ± 0.22	30.34 ± 1.81	/**	/	+	+	−	−	+	−	−
Li+ binding (K0.5(Na+), mM)f		1.88 ± 0.35	3.48 ± 0.43	0.67 ± 0.01	0.58 ± 0.05	0.42 ± 0.09	+	+	+	+	+	−	−
Soluble MelBSt at 52 °C (%)g	Na+	41.6 ± 3.5	10.85 ± 4.0	3.89 ± 0.2	3.99 ± 0.5	3.32 ± 1.6	2.14 ± 2.1	3.50 ± 1.2	7.94 ± 1.7	1.38 ± 0.3	1.84 ± 1.2	2.87 ± 1.6	2.35 ± 0.1
	Li+	44.89 ± 4.0	33.78 ± 1.3	46.62 ± 1.3	51.51 ± 4.2	7.99 ± 2.2	8.68 ± 1.5	21.25 ± 3.0	3.79 ± 1.2	19.62 ± 0.5	31.51 ± 3.2	2.55 ± 0.9	2.73 ± 1.5
Tm (°C)h	Li+	62.54 ± 0.32		60.30 ± 0.18	64.71 ± 0.23
	Li++Mel	63.87 ± 0.04		61.03 ± 0.13
pKa estimationi	Mel	8.75 ± 0.02***	7.75 ± 0.02	8.36 ± 0.03	8.75 ± 0.02	9.18 ± 0.02

^aRelative expression derived from Fig. 3.

^bInterpretation of fermentation results presented in Fig. 4; S, Colonies were small but red.

^crelative transport activity at 30-min time point presented in Fig. 4.

^dall data were from Trp→D²G assays. IC₅₀ were carried out with MelB_St in nanodiscs and the presence of Li⁺ (Fig. 7). The bound D²G was displaced by melibiose binding as described in the Methods. For most cases, the FRET binding results in Fig. 5 were explained as positive (+) or negative (−) qualitatively.

^eK_0.5(Na+), Na⁺ stimulation constant on the Trp→D²G FRET carried out with MelB_St in nanodiscs as determined in Fig. 6. For Ala and Pro mutants, no fitting due to lack of feature of the titration curve. For other cases, the Na⁺ binding data in Fig. 5 were explained as positive (+) or negative (−) qualitatively.

^fK_0.5(Li+), Li⁺ stimulation constant on the Trp→D²G FRET were carried out with MelB_St in nanodiscs as determined in Fig. 6. For Ala and Pro mutants, no fitting due to lack of feature of the titration curve. For other cases, the Li⁺ binding data in Fig. 5 were explained as positive (+) or negative (−) qualitatively.

^gData were from Fig. 8, and the relative amount of soluble MelB_St proteins compared to the total MelB_St after 52 °C treatment as described in Methods.

^hthe melting temperature was determined by CD spectroscopy in Fig. 9. All purified protein samples contain 0.01% DDM and 100 mM LiCl in the assay buffer.

ⁱthe calculated pKa values by PROPKA program as described in Methods.

^*SE, number of tests = 2-3

^**No fitting.

^***SE, number of tests = 1000.

[³H]Melibiose transport with intact cells

The E. coli DW2 (melA⁺melB⁻lacZ⁻Y⁻) intact cells expressing the WT or individual single-site mutants were used for [³H]melibiose transport assay at 0.4 mM (specific activity of 10 mCi/mmol) in the absence or presence of 20 mM NaCl or LiCl. While the chromosomal melA⁺ encoding α-galactosidase is present, it is not expressed without an induction by melibiose (27,28). Thus, this trace-flux assay determines melibiose accumulation against concentration gradient (29). As showed previously (2), the WT catalyzed melibiose accumulation couples to H⁺ in the absence of Na⁺ and Li⁺ (Fig. 4); in the presence of Na⁺ or Li⁺, the initial rates and steady-state level of melibiose accumulation are largely increased. The Na⁺-coupled transport is slightly greater than the Li⁺-coupled melibiose transport (Fig. 4). The effects of Thr121 mutations on all three modes of transport are varied (Table 1), and most mutations exhibited stronger inhibitions on melibiose active transport couped to Na⁺ or H⁺ than the coupling with Li⁺.

Fig. 4. Melibiose transport time course and fermentation.

MelB_St expression in E. coli DW2 (melA⁺B⁻lacY⁻) cells and active transport assay were described in Methods. The transport was carried out at 0.4 mM [³H]melibiose in the absence or presence of 20 mM NaCl or 20 mM LiCl at 23 °C. Accumulated melibiose was plotted against incubation time. E. coli DW2 cells that do not carry the expressing plasmid were included as the negative control and presented in the same panel for the positive control, the WT MelB_St. The results of melibiose fermentation on MacConkey agar plate containing 30 mM melibiose as the sole carbohydrate source were presented in each transport panel. Red colonies, normal melibiose fermentation; yellow colonies, no fermentation; pink colonies, reduced fermentation rates.

Several mutants (Pro, Ser, Cys, Ala, Gly, Asn, Asp, and Leu mutations) altered the cation selective preference (Fig. 4); thus, their symport activities with Li⁺ were better than that with Na⁺. Remarkably, the Cys, Ser, Pro and Ala mutants remained excellent melibiose transport coupled to Li⁺. With regard to the H⁺-coupled transport, only the T121P mutant retained the high steady-state level of melibiose accumulation. With regard to the Na⁺-coupled transport, most mutations showed severe inhibitions, and T121C and T121S mutants retained the best activity, but still less than 50% of the WT. T121P, A, G mutants completely lost the Na⁺ coupling. With the Asn, Asp and Leu mutants, all the three modes of transport activities were largely inhibited.

The Met mutation largely inhibited all three modes of transport, but retained the Na⁺-coupled transport better than Li⁺- or H⁺-coupled activities. Arg and Trp mutants with a bulky sidechain (the T121R mutant also with a positive charge) then completely abolished all three modes of transport (Fig. 4), while both proteins were expressed at a level comparable to all other mutants (Fig. 3a).

Melibiose fermentation

Melibiose fermentation activity was assessed by growing the cells on the MacConkey agar plates containing 30 mM melibiose as the sole carbohydrate source and sufficient Na⁺. WT MelB_St fermented melibiose well and the colonies and background turned to magenta color (Fig. 4, inset; Table 1), an indicator for a fast melibiose transport and degrees of acidification (30). The negative control DW2 cells with no MelB grew as yellow colonies, a sign for no melibiose fermentation, which could be yielded from no or very slow melibiose transport, or no α-galactosidase. Among all mutants, the T121R and T121W exhibited yellow colonies, indicating no fermentation, which is consistent with the results from the active transport. Most other mutants (including Cys, Asn, Asp, Leu and Met mutants) fermented melibiose similar to the WT, although the Asn, Asp, and Leu mutants showed no Na⁺- or H⁺-coupled activities. The T121S formed pink colonies, indicating a low-level melibiose fermentation.

The Pro, Ala, and Gly mutants, interestingly, had a growth problem and formed colonies at small sizes on this selective media; however, they grew normally on the regular LB agar plates or LB broth media. While the size is small, the colonies are red indicating fermentation of melibiose.

Measurement of substrate binding by Trp→D²G FRET with RSO membrane vesicles

Trp→D²G FRET is a well-established assay for the detection of substate binding to MelB including galactosides, Na⁺, or Li⁺ binding (2,31,32). It is based on the overlapping spectra of MelB Trp emission and dansyl excitation. The FRET signal can be displaced by MelB substrates but not by sucrose or other non-MelB substrate sugars, which establish the specific interaction of fluorescent analogue dansyl-2-galactoside (D²G) to MelB. The specific binding can be detected in native membrane samples (2,30,33), purified in detergent (10,18,22), or reconstituted into proteoliposomes (26). To assess the mutational effects in situ, right-side-out (RSO) membrane vesicles were prepared by osmotic lysis from E. coli DW2 cells carrying the WT or individual mutants to determine the binding of substrates (galactoside, Na⁺ or Li⁺). With the WT MelB_St (Fig. 5), D²G at a final concentration of 10 μM (similar to the K_d value) was added into the RSO vesicle sample, and fluorescent intensity was recorded at an exciting wavelength at 290 nm and an emission wavelength at 495 nm that is the maximal wavelength for Trp → dansyl galactoside FRET spectra in MelB_St (2). Consecutive addition of melibiose decreased the fluorescent intensity; on a separate trace, addition of water only exhibited a small dilution effect. The difference in the fluorescent intensity between the two traces reflects the Trp→ bound dansyl galactoside FRET, i.e., the evidence for the binding of galactosides.

Fig. 5. Substrate binding determined by Trp→D²G FRET assay.

Right-side-out (RSO) membrane vesicles in a Na⁺-free buffer prepared from E. coli DW2 cells expressing MelB_St at 1 mg/ml were mixed with D²G at 10 μM (black arrows) and fluorescence intensity at 490 nm was recorded using an excitation wavelength 290 nm. At an interval of 1 min, melibiose at a final concentration of 130 mM (pink arrows), NaCl at 50 mM (green arrow), or LiCl at 50 mM (blue arrow), was supplemented as indicated. Each panel contains two traces, one with water (gray arrows) instead of melibiose but at equal volume for the correction of the dilution effect.

Addition of NaCl or LiCl prior to the melibiose displacement increased the fluorescent intensity. This increase is mainly derived by an increase in the number of D²G binding as a result of positive binding cooperativity between galactoside and cation (18,34). In addition, Trp environmental changes induced by the cation binding could be another factor. With the WT MelB_St, the total intensity increase induced by Na⁺ is significantly smaller than that by Li⁺, which may indicate the two ternary complexes might adopt two different conformations (2).

The T121 mutants exhibited varied phenotypes on the Trp→D²G FRET measurement (Fig. 5; Table 1). Except for the T121W mutant, all other mutants displayed galactosides binding. Further, T121N, T121S, and T121R mutants in the absence of Na⁺ or Li⁺ exhibited the D²G-binding signal similar to that of the WT; however, none of them catalyzed H⁺-coupled melibiose transport as the WT did.

Pro mutant retained the binding for D²G, melibiose, and Li⁺, with reduced FRET intensity changes, but lost the Na⁺ binding. In consistent to the transport data, the T121P mutant selectively eliminated Na⁺ binding and Na⁺-coupled melibiose symport but retained the other two modes, i.e., H⁺/melibiose and Li⁺/melibiose symport.

Ala and Gly mutants also exhibit galactoside binding and Li⁺ stimulation, but the Na⁺-binding signal was largely reduced. T121C, T121N, T121M, and T121S mutants showed good binding signals for both Na⁺ or Li⁺; T121D and T121L lost Na⁺ binding.

Determination of cation binding by measuring the cation-stimulation constants (K_0.5) with lipid nanodiscs.

The increase in FRET intensity (_diffFRET) induced by Na⁺ or Li⁺ only reflects the binding qualitatively. The FRET intensity changes derived from some mutants are too small to determine quantitatively the cation binding affinity. The attempts to generate purified proteins with the routine MelB_St purification protocol failed due to poor stability of these mutants in detergent. Thus, to maintain the stability as shown in Fig. 8, reconstitution into lipid nanodiscs immediately following the elution step during the affinity purification was used.

Fig. 8. Thermostability tests detected by western blotting.

(a) Thermostability in the presence of Na⁺. The membrane samples at protein concentration of 10 mg/ml were mixed with detergent 1.5% DDM for 90 min on ice, and the samples were then incubated in increased temperatures of 52 °C and 54 °C, respectively, for another 90 min before ultracentrifugation at 355,590 g for 45 min. Equal volume of the supernatants was analyzed by SDS-15%PAGE using the membranes containing same amount of membrane proteins for the DDM extract as the control for total MelB_St. (b) Thermostability in the presence of Li⁺. To remove Na⁺, membrane samples were extensively washed with LiCl-containing buffer (50 mM Tris-HCl, pH 7.5, 100 mM LiCl, 20 mM melibiose and 10% glycerol) and subjected to the test as described in panel a. Histograms, the soluble fractions were derived from panel a (in the presence of Na⁺) and panel b (in the presence of Li⁺). Pink bar, the total membrane proteins; green or gray bars, soluble MelB_St after treated at 52 °C or 54 °C, respectively, and ultracentrifuged to remove the aggregated proteins. The number of tests is 2, and each variation bar indicated the SE values.

With the Trp →D²G FRET assay (Fig. 6a), instead of adding 50 mM NaCl or LiCl, stepwise titrations of Na⁺ or Li⁺ were carried out till no further FRET change occurred. The cation simulation constant (K_0.5), which reflects the cation-binding affinity, can be determined from a hyperbolic function. With the lipids nanodiscs containing the WT MelB_St, the K_0.5(Na+) and K_0.5(Li+) values were 0.83 mM and 1.88 mM, respectively, which are in a range of the results from RSO membrane vesicles (Table 1) (2,27,30). The conserved T121S mutant exhibited 10-fold reduced Na⁺-binding affinity with K_0.5(Na+) of 8.24 ± 0.22 mM, whereas the affinity to Li⁺ was only decreased by less than 2-fold. T121C mutation largely inhibited the Na⁺ binding with a K_0.5(Na+) value of 30.34 ± 1.81 mM, which is greater than 30-fold poor than that for the WT, but its affinity to Li⁺ is 3-fold increased with a K_0.5(Li+) of 0.67 ± 0.01 mM.

Fig. 6. Cation-binding affinity.

Trp→D²G FRET assay was used to determine cation binding affinity by measuring the stimulation constant on the FRET. MelB_St proteins were reconstituted into lipid nanodiscs as described in the Methods, and the Trp→D²G FRET was recorded at 490 nm as described in the legend of figure 5. NaCl or LiCl, at a 1 min interval, was added stepwise until no change in FRET signal was approached. Addition of an identical volume of water at each step was used for the control. The K_0.5 determination was carried out as described in the Methods.

With T121P and T121A mutants, there was no Na⁺ binding detected even with increased NaCl concentration to greater than 500 mM. Both mutants, however, exhibited an increased binding affinity to Li⁺ by 3~4 -fold and the K_0.5(Li+) values for T121P and T121A mutants were 0.42 ± 0.09 mM and 0.58 ± 0.05 mM, respectively.

Determination melibiose-binding affinity by determining the melibiose concentration to displace half-maximal Trp →D²G FRET (IC₅₀) with lipid nanodiscs.

With the Trp →D²G FRET assay, instead of adding the saturating concentration of melibiose, titrations of melibiose step by step were carried out till no further FRET change occurred (Fig. 7). The melibiose IC₅₀, which reflects the melibiose affinity, can be determined from a hyperbolic function. With the lipids nanodiscs containing the WT MelB_St, the IC₅₀ for melibiose to displace 50% bound D²G values were 2.73 ± 0.04 mM in the presence of Li⁺ (Table 1). As indicated from the time-trace records, melibiose binding affinities to the mutants including T121S, T121C and T121A are in the similar range of the WT, and T121P mutant exhibited slightly greater IC₅₀ (5.77 ± 0.52 mM).

Fig. 7. Melibiose binding affinity.

Trp→D²G FRET assay was used to determine the melibiose binding affinity expressed as IC₅₀ for melibiose displacement of D²G binding Stepwise displacements of D²G by melibiose at increasing concentrations were measured on a time trace and IC₅₀ values were determined as described in the Methods.

Thermostability detected by western blotting

To test the thermostability of the mutant proteins (Fig. 8a), membranes (10 mg/mL) in a Na⁺- and melibiose-containing buffer (50 mM NaPi, pH 7.5, 200 mM NaCl, and 20 mM melibiose) were solubilized with a detergent dodecyl-β-D-maltopyranoside (DDM) (1.5%) on ice, then incubated at 52 °C or 54 °C for 90 min. After an ultracentrifugation to remove aggregates, equal volume of supernatant was analyzed by SDS-15%PAGE and western blotting using untreated membranes diluted with water instead of DDM as the control. The WT MelB_St can be detected even after 90-min heat incubation at 54 °C; however, for all mutants, hardly any MelB_St is visualized even after 52 °C treatment.

To test the Li⁺ effect (Fig. 8b), all membranes were extensively washed by the same buffer but containing LiCl at 200 mM instead of NaCl. Interestingly, the presence of Li⁺ largely increased the protein stability for several mutants (Fig. 8, Histogram); such as the thermostability of T121A and T121C mutants are similar to the WT. With a treatment at 54 °C, the MelB_St protein band can be detected clearly with T121S mutant and visible for T121D and T121M mutants. Further, between the two salts, the stability of the WT MelB_St is greater in the presence of Li⁺ than Na⁺.

Melting temperature (T_m) detected by CD spectroscopy.

For a quantitative analysis, the WT MelB_St and T121A and T121C mutants were purified in DDM in the presence of NaCl and subjected to CD spectra analysis and thermal denaturation testes for determining the melting temperature (T_m) (Fig. 9), as described previously (22,26). In this study, we focused on the mutational effect. Overall, the CD spectra in the absence or presence of substrates were undistinguishable (Fig. 9a), featured with strong negative ellipticity sub-maxima at 209 nm and 222 nm. The thermal denaturation tests were performed at temperatures between 25 and 90 °C and the ellipticity changes at 210 nm were recorded to monitor the unfolded fractions (Fig. 9b). The T_m value, at which 50% protein unfolded, for the WT bound with Li⁺ is 62.63 ± 0.53 °C, and ternary state with both Li⁺ and melibiose is 63.89 ± 0.04 °C (Table 1). Mutant T121C exhibited slighted reduced T_m values, and mutant T121A showed slightly increased T_m values of 65.36 ± 0.88 °C. The T_m results are consistent with the data from the thermostability tests detected by western blotting (Fig. 8).

Fig. 9. Determination of melting temperature by thermo-denaturation monitored by CD spectroscopy.

(a) CD spectra. MelB_St at a protein concentration of 4 μM in the CD assay buffer (20 mM NaPi, pH 7.5, 100 mM LiCl, 10% glycerol, and 0.01% DDM) in the absence or presence of melibiose at 50 mM were prepared for the measurements. The CD spectra were recorded between 200 and 260 nm at 25 °C after subtracted from each corresponding buffer control. (b) Thermal denaturation. The CD ellipticity changes were recorded at 210 nm at a 1 °C interval with temperature ramping at 1 °C per minute, and expressed as a percentage of the unfolded fraction. Black, WT MelB_St; orange, T121C mutant; red, T121A mutant. Dashed line, no sugar; solid line, with 50 mM melibiose.

Estimated pKa of Asp59 in the WT MelB_St and mutants.

Previous competitive assay of Na⁺ binding as a function of pH with isotherm titration calorimetry (ITC) revealed the absolute pKa value of MelB_St cation-binding site is about 6.25 or 6.59 in the absence or presence of melibiose{Hariharan, 2017 #7289}. In our current model, the Asp59 is the sole sidechain responsible for the proton binding in MelB_St. To further examine if the local hydrophobicity near the Thr121 affects the pKa of Asp59, four Thr121 mutants (including Ser, Cys, Pro and Ala mutations) were created computationally, and the pKa values of Asp59 in the WT and each mutant were estimated by molecular dynamics simulations combined with the PROPKA program (Table 1) (35,36). It is well-known that it is challenging to computationally predict the absolute pKa values of ionizable residues deeply buried in proteins, and our simple PROPKA calculations indeed predicts higher pKa values than the experimental measurements. However, the relative changes in the calculated pKa between the WT and mutants under the same simulation conditions are often more reliable and could provide useful information. Both the T121S and T121C mutants decrease the Asp59 pKa relative to the WT due to the presence of more hydrophilic residues (Table 1). On the other hands, the calculated pKa values for the T121A and T121P mutants are either unaffected or elevated, respectively, due to the presence of side chains with similar or more hydrophobicity compared to that in the WT.

Discussion

A transporter that uses Na⁺ as the coupling cation can often use Li⁺ and/or H⁺. For most Na⁺-coupled bacterial secondary transporters, their cation site has not been extensively studied. In MelB_St and MelB_Ec, the specificity, selectivity and stoichiometry have been well characterized (1,2,18,37-39). All studies support that the three cations H⁺, Na⁺, and Li⁺ share the same binding pocket (2,18,40-42). In MelB_St, previous ITC measurements determined the absolute K_D values for Na⁺ and H⁺, which are 0.64 mM and 0.56 μM (pK_a of 6.25), respectively (18). The binding affinities increase in the presence of melibiose with a cooperativity number of 8 and 2, respectively; thus, K_D(Na+) and K_D(H+) are decreased to 90 μM and 0.26 μM, respectively (18). Melibiose affinity was also increased in the presence of Na⁺, so a positive cooperativity of melibiose and Na⁺ binding has been firmly established, which has been proposed to be the core principle of symport mechanisms (22).

While MelB_St cation binding was extensively characterized biochemically, our knowledge at molecular levels is still limited. The high-resolution 3D structures for three Na⁺-coupled MFS transporters were solved (3,11,12); unfortunately, none of them resolved the Na⁺ binding. Our all-atom MD simulations in this study showed that Asp55, Asn58, Asp59, and Thr121 of MelB_St contributes to the cation binding (Fig. 2), which is consistent with previous functional analysis and extend the knowledge of cation binding in MelB_St. The Na⁺ is stably coordinated by five coordinating atoms from all four residues with an average coordination bond length of 2.3 Å. This Na⁺ coordination pattern in MelB_St is similar to that observed in the MFSD2A(12) or the multidrug and toxic compound extrusion (MATE) transporters (43) The Li⁺ is stably coordinated by four coordinating atoms from three residues with an average coordination bond length of 2.0 Å, and the coordination by the Thr121 is less stable. Interestingly, the helix II contributes three critical residues for cation coordination (Asp55, Asn58, and Asp59), and the helix IV hosting the major sugar-binding residues (Asp124, Trp128, Tyr120) only contributes the Thr121 residue for cation coordination. The Na⁺ and Li⁺ are only slightly different in their ionic radii, and the difference in the two simulated binding modes are small, but the two cations affect MelB’s conformational change and transport activity very differently. For example, melibiose change rate in the presence of Na⁺ is fast, but Li⁺ inhibits melibiose exchange rate (2). The novel insights into the difference of Na⁺ and Li⁺ binding interactions with MelB_St presented here may provide important clues for exploring the mechanistic effects on the activities and coupling mechanisms of MelB in the future.

The Thr121 residue forms stable coordination interaction with the Na⁺ but not the Li⁺. Our systematic mutagenesis in combination of melibiose transport, substrate binding, and stability assays consistently support the observation from the MD simulations that Thr121 residue is part of the cation binding pocket and essential for the binding to Na⁺, but not necessary for the binding of Li⁺. Any mutants that can bind Na⁺ are also capable in binding of Li⁺. All mutants, that can significantly retain the Na⁺ binding and Na⁺-coupled active transport, contain a polar group (either O or S) on the γ (Ser or Cys) or the δ (Met) position. Interestingly, T121N mutation also carries a carbonyl O at δ position, but the amide group may interrupt the binding of Na⁺ and Li⁺. The T121C mutant retains the highest transport activity coupled to Na⁺; it has an unaltered initial transport rate, but the steady-state accumulation level was only 30% of the WT (Fig. 4; Table 1).

The γ carbon in the methyl group on the Thr121 sidechain is in close proximity to Asp59 (Fig. 1c), and it might partially contribute to the local hydrophobicity around Asp59 and an elevated pKa of this residue for the binding of H⁺. Neither Cys nor Ser contains this hydrophobic methyl group, and accordingly the H⁺-coupled active transport of melibiose by the two mutants are also poor. With regard to the steady-state level of the H⁺-coupled melibiose accumulation, the T121P mutant is the best, followed by Met, Ala, and Ser mutants, albeit with a large difference in activity levels. All other mutants nearly completely lost the H⁺/melibiose symport activity. The computationally estimated pKa values of Asp59 in the T121S, T121C, T121A and T121P mutants are consistent with our interpretations that the local hydrophobic environment modulates the proton affinity of Asp59, thereby affecting the H⁺-coupled melibiose transport.

Placing Pro or Ala at position Thr121 increased the affinity for Li⁺, with little effects on the Li⁺/melibiose symport activity (Table 1; Fig. 4). In contrast, these residues with no or a hydrophobic carbon sidechain selectively eliminated Na⁺ binding (Figs. 5-6), which narrowed the cation specificity of MelB_St. Placing the small polar residues Cys or Ser on Thr121 largely inhibited the Na⁺ binding and the Na⁺-coupled melibiose symport, but caused only mild inhibitions of Li⁺ binding, with an unnoticeable change in the Li⁺-coupled transport. These mutations largely altered cation selectivity. When introducing a residue with a large, bulky, or hydrophobic sidechain, the binding to either Na⁺ or Li⁺ were severely inhibited. The different effects on Na⁺ or Li⁺ binding from Cys and Ser compared with Pro and Ala may suggest that the precise positioning of a polar group is required for coordinating Na⁺. All of our experimental data support that the binding of Li⁺ only involves a subset of the Na⁺-binding site, which is consistent with the MD simulations results showing that Li⁺ coordination number in MelB_St is less than that of Na⁺. Over all, the results indicate that Thr121 is essential for Na⁺ binding, but it is not necessary for the Li⁺ binding, nor the H⁺ binding.

Protein stability is critical for structural and functional characterizations experimentally. The stability of MelB_St is strongly influenced by the charge balance of the cation-binding pocket. This binding site contains two negatively charged residues (Asp55 and Asp59), two polar residues (Asn58 and Thr121), and a positively charged Lys377 residue. Ly377 is positioned between the Asp55 and Asp59, which yields an unbalanced charge distribution (2 negative and 1 positive). In the D59C mutant crystal structure, Lys377 forms a salt-bridge interaction with Asp55(3). This single D59C mutation of MelB_St, by removing one negative charge and neutralizing charge distribution in the cation-binding site, largely increased thermostability of MelB_St by ~6 °C (3). Further, the D55C mutation also increased thermostability by 2 °C (22). The WT MelB_St with an empty cation-binding site is less stable, and the binding of the cation Na⁺ increases the thermostability by 1.5 °C and the binding of Na⁺ and melibiose increases the thermostability by 2.6 °C (22).

In this study, the protein stability of Thr121 mutants exhibited a clear cation dependence. Due to a weak or lack of Na⁺ binding, all mutants showed largely reduced stabilities in the presence of Na⁺ (Fig. 6). Remarkably, neutralization of the cation-binding site for mutants T121A, T121C and T121S by a bound Li⁺ regained the protein thermostability. Especially in the presence of Li⁺, the T121A mutant exhibited higher Li⁺ binding affinity (Table 1) and also slightly greater T_m than that of the WT (Table 1; Figs. 8 and 9). Taken all together, a charge-unbalanced cation-binding site causes protein less stable or unstable, and MelB_St can be stabilized by the binding of Na⁺ or Li⁺, or mutated to remove one negative charge. This finding supports the view that the charge balance of the cation-binding pocket modulates MelB_St stability. The results provide useful knowledge and important insights for generating stable protein samples for structural and functional studies.

Materials and Methods

Materials and regents

[1-³H]Melibiose (5.32 Ci/mmol) was custom synthesized by PerkinElmer and unlabeled melibiose was purchased from Acros Organics (Fisher Scientific). MacConkey agar media (lactose free) was purchased from Difco. Detergent n-dodecyl-β-D-maltopyranoside (DDM) were purchased from Anatrace. 2′-(N-dansyl)aminoalkyl-1-thio-β-d-galactopyranoside (D²G) was kindly provided by Drs. Gerard Leblanc and the late H. Ronald Kaback. E. coli Extract Polar was purchased from Avanti. All other materials were reagent grade and obtained from commercial sources. Oligodeoxynucleotides were synthesized by Integrated DNA Technologies. All mutants were constructed by QuickChange™ Site-Directed Mutagenesis Kit (Invitrogen), and confirmed by DNA sequencing analysis.

Bacterial strains and plasmids.

E. coli DW2 strain (melA⁺, ΔmelB, ΔlacZY) was used for expression and functional analysis. The expression plasmid pK95 ΔAH/MelB_St/CHis₁₀ encoding the MelB_St with a His₁₀-tag at the C-terminus (2,38) was used as the template for constructing the site-directed mutagenesis.

Cell growth for transport assay and protein expression.

E. coli DW2 cells (melA⁺, melB⁻, lacZ⁻Y⁻) transformed with the plasmid carrying MelB were grown in Luria-Bertani (LB) broth with 100 mg/L ampicillin in a 37 °C shaker overnight. The overnight cell growth was diluted by 5% to fresh LB broth with 0.5% glycerol and 100 mg/L of ampicillin. Constitutive overexpression was obtained by shaking at 30 °C for 5 h for the functional tests and for examining the protein expression level.

[1-³H]Melibiose transport assay.

E. coli DW2 cells expressing MelB_St were washed with 50 ml of 100 mM KP_i (pH 7.5) three times, followed by washing with the assay buffer (50 ml 100 mM KP_i, pH 7.5, 10 mM MgSO₄). The cell pellets were resuspended with the assay buffer and adjusted to A₄₂₀ = 10 (~0.7 mg proteins/ml). The transport was initiated by mixing 50 μL of cells with 2 μL of 10 mM [³H]melibiose at a specific activity of 10 mCi/mmol in the absence or presence of 500 mM NaCl or 500 mM LiCl, respectively, which yielded a final concentration of 0.4 mM melibiose in the absence or presence of 20 mM Na⁺ or 20 mM Li⁺, respectively. The transport time courses were carried out by separating the cells from solutions at zero, 5 s, 10 s, 30 s, 1 m, 2 m, 5 m, 10 m, and 30 m by a dilution and fast filtration. The filters were subjected to the radioactivity measurement using a Liquid Scintillation Counter. The radioactivity was converted to nmol melibiose/mg proteins. The E. coli DW2 cells without a plasmid were used as the negative control.

Melibiose fermentation on MacConkey agar plates.

E. coli DW2 cells (ΔmelBΔlacYZ) were transformed with a plasmid carrying the WT or mutants and plated on MacConkey agar plates containing 30 mM melibiose, 100 mg/L of ampicillin, and incubated at 37 °C. After 18 h, the plates were viewed and photographed immediately. Formation of magenta color colonies after 18 h is indicative of a normal melibiose fermentation; pink colony, reduced fermentation; yellow colony indicates on melibiose fermentation due to poor transport rates,

Membrane expression and western blotting.

Cells were grown as described for the transport assay. After 7 h shaking at 30 °C, cells were washed with 50 mM NaP_i, pH 7.5, resuspended with the same buffer, and broken by sonication. The unbroken cells or debris were removed by centrifugation, and the supernatant were subjected to ultracentrifugation at 62,000 rpm for 30 min in a Beckman rotor Ti 70. The membrane pellets were resuspended with 50 mM NaP_i, pH 7.5, 200 mM NaCl, 10 % glycerol, and 20 mM melibiose. After protein assay, 20 μg of total membranes was loaded on to SDS-15% PAGE. The gel was transferred onto the PVDF membrane by the Trans-Blot Turbo transfer system (Bio-Rad) at 1.3 A, 25 V for 20 min. The PVDF membrane was blocked by 3% BSA in TBST buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.05% tween-20) overnight, and then washed with TBST buffer for 15 min twice. The washed membrane was used to react with the HisProbe™-HRP conjugate (Thermo Scientific) in TBST with 1.0% BSA for 1 h, and then washed three times with TBST for 15 min. MelB_St protein was detected using the SuperSignal West Pico chemiluminescent substrate (Thermo Scientific) by the ImageQuant LAS 4000 Biomolecular Imager (GE Health Care Life Science).

Protein thermostability assay in the presence of Na⁺ or Li⁺.

Membranes in buffer containing Na⁺ and melibiose (50 mM NaP_i, pH 7.5, 200 mM NaCl, 10 % glycerol, and 20 mM melibiose) at a final protein concentration of 10 mg/ml were mixed with 1.5% DDM for 90 min on ice. The membrane extracts were then incubated at two increased temperatures of 52 °C and 54 °C, respectively, for another 90 min before ultracentrifugation at 355,590 g in a Beckman OptimaTM MAX ultracentrifuge using a TLA-100 rotor for 45 min at 4 °C. Equal volume of the supernatants were analyzed by SDS-15%PAGE and western blotting. The membranes containing same amount of total protein that were used for the DDM extract was loaded as the control for total MelB_St.

For test Li⁺ effects, the membranes in the Na⁺-containing buffer were diluted by 50-fold with Li⁺-containing buffer (50 mM Tris-HCl, pH 7.5, 200 mM LiCl, 10 % glycerol, and 20 mM melibiose) with other components unchanged, and ultracentrifuged at 62,000 rpm using Ti 70 rotor for 30 min. The pellets were washed with the same buffer by a 50-fold dilution with sonication. The pellets after ultracentrifugation were resuspended in the same Li⁺-containing and subjected to the thermostability testes as described above.

Protein concentration assay.

The Micro BCA Protein Assay (Pierce Biotechnology, Inc.) was used to determine the protein concentration assay.

Right-side-out (RSO) membrane vesicles preparation.

The MelB_St was expressed in the E. coli DW2 as described above, and RSO membrane vesicles were prepared by osmotic lysis and resuspended in 100 mM KPi as described (2,44,45).

Affinity purification of the WT and mutant MelB_St.

Cell growth for a large-scale production of the WT MelB_St and single-site Thr121 mutants from E. coli DW2 cells (melA⁺melB, lacZY) as described (10,46). Briefly, the cells were grown in Luria–Bertani broth supplemented with 50 mM KPi, pH 7.0, 45 mM (NH₄)SO₄, 0.5% glycerol, and 100 mg/liter ampicillin(46). The initial protocols for membrane preparation and MelB_St purification by cobalt-affinity chromatography after extraction by 2% DDM were adapted as described previously (3). MelB_St protein was eluted with 250 mM imidazole in a buffer containing 50 mM NaPi, pH 7.5, 100 mM NaCl, 0.01% DDM, and 10% glycerol.

Overexpression and purification of membrane scaffold protein (MSP).

Overexpression of the membrane scaffold protein MSP1D1E3 with N-terminal 7-His tag followed by spacer sequence and tobacco etch virus (TEV) protease cleavage site (mass 32.6 kDa) from a plasmid pMSP1E3D1 (Addgene; #20066) was carried out in the E. coli BL21 (DE3) strain. MSP1D1E3 yields nanodiscs with a diameter of ~12.1 nm (47). The cells were grown in LB media containing 0.5% glucose and 30 mg/L kanamycin at 37 °C; protein expression was induced by adding 1 mM IPTG at an A₆₀₀ of ~0.6 for 2-3 h. The MSP from the cell lysates were purified with metal-affinity purification using INDIGO Ni-Agarose (Cube Biotech). The eluted MSP at 300 mM imidazole were dialyzed at 4 °C against 20 mM Tris-HCl, pH 7.5, and 100 mM NaCl and concentrated to ~8 mg/ml. To remove the His tag, a 5 mL of 21.5 mg MSP1D1E3 protein was mixed with 0.5 mL of His-tagged TEV protease (1.67 mg/ml in 25 mM KPi, 200 mM NaCl, 10 % glycerol, 2 mM EDTA, 10 mM DTT, pH 8.0), at a ratio of 1:20 (TEV/ MSP1D1E3, mol/mol), and incubated for 48 h at 4 °C. Processed MSP was separated from His-tagged TEV protease and unprocessed His-tagged MSP by Ni-agarose chromatography as a flowthrough, concentrated to ~6-8 mg/ml, frozen in liquid nitrogen, and stored at −80 °C.

Reconstitution of MelB_St into phospholipid bilayer nanodiscs.

A stepwise reconstitution was adapted from the reported protocols (22,48). Briefly, in a 2-ml reaction, 2 mg of the purified MelB_St in DDM at a concentration of 1 mg/ml was mixed with 12.6 mg of E. coli polar lipids extract from a stock of 40 mg lipids/ml (50 mM) in 7.5% DDM, yielding a protein:lipid ratio of 1:420 (mol/mol) or 1:6.3 (mg/mg). The protein/lipid mixture was incubated for 10 min on ice; MSP1D1E3 protein was added at a 6:1 molar ratio of MSP1D1E3:MelB_St, followed by incubation at 23 °C with mild stirring for 30 min. The detergents were removed using 1 g Bio-Beads SM-2 (500 mg beads per 1 mg MelB_St) with mild stirring at 4 °C for 2 h, followed by overnight incubation after adding another portion of Bio-Beads SM-2 (300 mg). Bio-Beads SM-2 was removed by centrifugation at 20,000 g for a few minutes at 4 °C, and the reconstituted phospholipid bilayer nanodiscs were collected from the supernatant.

Reconstituted nanodiscs were further isolated by metal-affinity purification using Ni-NTA beads. The elute containing MelB_St in nanodiscs was further dialyzed against a ligand-free buffer without a detergent (20 mM Tris-HCl, pH 7.5, 50 mM Choline Chloride (ChoCl), and 10% glycerol. Protein concentration was measured at A₂₈₀ nm with an extinction coefficient (Ɛ = 135,110) based on 1 MelB_St and 2 MSP1D1E3 molecules. MelB_St lipid nanodiscs were aliquoted, flash-frozen in liquid nitrogen, and stored at −80°C.

Trp→dansyl fluorescence resonance energy transfer (FRET)

Steady-state fluorescence measurements were performed with an AMINCO-Bowman series 2 spectrometer with RSO membrane vesicles containing MelB_St at 1 mg/ml in 100 mM KP_i, pH 7.5. Using an excitation wavelength at 290 nm, the emission intensity was recorded at 490 nm. On a time-trace, D²G, NaCl or LiCl, then melibiose were sequentially added into the RSO vesicles solution at a 1 min-interval. An identical volume of water instead of melibiose was used for the control on a separate trace.

Determination of the cation stimulation constant of Na⁺ (K_0.5(Na+)) and Li⁺ (K_0.5(Li+)) on the Trp→dansyl FRET intensity.

The reconstituted lipids nanodisc samples were used to increase the binding signal. On the same experimental setup, titration of Na⁺ or Li⁺ after adding 10 μM D²G were performed. An identical volume of water was used for the control. The increase in the intensity after each cation addition was corrected by the dilution effect and plotted as a function of the accumulated cation centration. The K_0.5 value was determined by fitting a hyperbolic function to the data (OriginPro).

Determination of IC₅₀ for melibiose displacement of the Trp→dansyl FRET intensity.

Trp→dansyl FRET was used to determine melibiose affinity using the reconstituted lipids nanodiscs. After the addition of 10 μM D²G and LiCl, melibiose was added consecutively till no further decrease in the FRET intensity was reached. Addition of the identical volume of water was used for the correction of dilution effect. The decrease in FRET was plotted as a function of melibiose concentration. IC₅₀ was determined by the hyperbolic fitting to data.

CD spectroscopy and melting temperature (T_m) determination.

The CD measurements were carried out using Jasco J-815 spectrometer equipped with a Peltier MPTC-490S temperature-controlled cell holder unit. MelB_St at 4 μM in 20 mM NaPi, 100 mM LiCl, 10% glycerol, and 0.01% DDM at a pH 7.5 in the absence or presence of 50 mM melibiose. An aliquot of 180-μL MelB_St sample was placed in 1-mm quartz cuvette on the temperature-controlled cell holder. CD spectra for a wavelength range of 200-260 nm were collected at data pitch of 0.1 nm using a band width of 1 nm and scanning speed of 100 nm/min by using Jasco Spectra Measurement software (V.2). Each spectrum was corrected by subtraction with corresponding buffer in the absence of MelB_St.

The T_m determination was carried out by thermal denaturation tests at temperatures between 25-90 °C. Ellipticity at 210 nm was recorded at a 1 °C interval with the temperature ramp rate at 1 °C per minute, and plotted against temperature. The T_m values were defined as the temperature leading to 50% unfolding, which was determined by fitting the data to the Jasco Thermal Denaturation Multi Analysis Module.

All-atom molecular dynamics simulation.

The crystal structure of the D59C mutant of the MelB_St (pdb, 7L17) was used as the starting structure for the simulation of the WT MelB_St. The Cys59 was mutated back to Asp59. The protein was embedded in a lipid bilayer consisting of POPE and POPG. The ratio of POPE vs. POPG was 7:2, similar to the membrane composition of E. coli, and the total number of lipid molecules is 288. The lipid bilayer was capped by water molecules on each side, and either LiCl or NaCl ions were inserted among the water molecules to generate a solution of ~0.16 M concentration. The melibiose was inserted into the sugar binding site based on the crystal structure of the α-NPG. The simulation system contains ~131,000 atoms, and the periodic boundary condition is ~99 Å ×99 Å × 131 Å. The CharmmGUI web interface (49) was employed to setup the system.

The following simulation procedures applied to each of the two system setups, i.e. the MelB_St + melibiose + LiCl and the MelB_St + NaCl + melibiose. The system was first optimized with the heavy atoms of protein and lipids harmonically restrained (1000 kJ/mol/Ų force constant), followed by 130 ps of dynamics at 300 K temperature in the constant NVT ensemble. The restraints were then gradually reduced over 10 ns of dynamics in the NPT ensemble at 300 K temperature and 1 atm. Finally, 30 ns of dynamics in the NPT ensemble with the same temperature and pressure was carried out for equilibration with no restraints. After equilibration, 1 μs of dynamics in the NPT ensemble with the same temperature and pressure was carried out to generate the production trajectory. The temperature was maintained by a Langevin thermostat with a friction coefficient of 1 ps^-1. The trajectory was propagated with a 2 fs timestep, and the lengths of all bonds that involve a hydrogen atom are constrained. Throughout all simulations, the CHARM36 force field was used to simulate the protein, lipid, melibiose and ions (50-53). The particle mesh Ewald method(54) with a tolerance of 5 × 10⁻⁴ was used to calculate the electrostatic interactions. The cutoff for van der Waals interactions was chosen as 12 Å. The water molecules and ions were treated by the TIP3P model(55),The classical MD equilibration was performed using the OpenMM software package (56).

Estimation of pKa values of MelB_St Asp59.

The WT and four computationally created mutants (T121S, T121C, T121A and T121P) were subjected to pKa calculation using the PROPKA program (35,36). For each system, 100 ns MD simulation was performed in the constant NPT ensemble (T=300 K, P=1 atm) with Asp59 in the protonated state and other ionizable residues in their standard protonated states (deprotonated Asp, Glu, His residues and protonated Thr, Ser, Cys, Lys, Arg, Tyr). In the aqueous solution capping the lipid bilayer, 0.16 M KCl was added. Other simulation parameters remain the same as mentioned above. During the MD simulation, none of the K⁺ ions penetrated into the cation-binding site. One thousand snapshots were extracted from each of the five MD trajectories with a 100 ps interval, and each snapshot was subjected to the pKa calculation of Asp59 using the PROPKA program (35,36). The statistical uncertainty of the average pKa value for each system was obtained by bootstrapping over 1000 random samples of all snapshots.

Abbreviations

WT

wild type

MelB_St

Salmonella typhimurium melibiose permease

MelB_Kp

Klebsiella pneumoniae melibiose permease

MelB_Ec

Escherichia coli melibiose permease

E. coli

Escherichia coli

MSFD2A

Na⁺-dependent phospholipids transporter

DDM

dodecyl-β-D-maltopyranoside

DDMB

dodecyl melibioside

α-NPG

α-nitrophenyl galactoside

D²G

dansyl-2-galactoside

RSO

right-side-out

FRET

fluorescence resonance energy transfer

K _0.5(Na+)

the stimulation constant of Na⁺ for the Trp→dansyl FRET intensity

K _0.5(Li+)

the stimulation constant of Li⁺ for the Trp→dansyl FRET intensity

CD

circular dichroism

T _m

melting temperature

MD simulations

molecular dynamics simulations

pKa

acid dissociation constants