|Home | About | Journals | Submit | Contact Us | Français|
Neurotransmitter:sodium symporters (NSS), targets for psychostimulants and therapeutic drugs, play a critical role in neurotransmission. Whereas eukaryotic NSS exhibit Cl−-dependent transport, bacterial NSS feature Cl−-independent substrate transport. Recently we showed in LeuT and Tyt1 that mutation of an acidic side chain near one of the Na+-binding sites renders substrate binding and/or transport Cl− dependent. We reasoned that the negative charge - provided either by Cl− or by the transporter itself - is required for substrate translocation. Here we show that Tyt1 reconstituted in proteoliposomes is strictly dependent on the Na+ gradient and is stimulated by an inside negative membrane potential and by an inversely-oriented H+ gradient. Remarkably, Na+/substrate symport elicited H+ efflux, indicative of Na+/substrate symport-coupled H+ antiport. Mutations that render the transport phenotype Cl−-dependent essentially abolish the pH dependence. We propose unifying features of charge balance by all NSS members with similar mechanistic features but with different molecular solutions.
Members of the eukaryotic neurotransmitter:sodium symporter (NSS) family play an essential role in the nervous system by terminating synaptic transmission and recycling various neurotransmitters, including biogenic amines and amino acids1. The transporters for the biogenic amines dopamine (1), norepinephrine (2), and serotonin (3), (DAT, NET, SERT, respectively) are of particular interest because they are the primary targets for widely abused psychostimulant drugs such as cocaine (4) and amphetamine(5)2, and for antidepressants3. These secondary active transporters catalyze the (re)-uptake of their respective substrates from the extracellular milieu against their concentration gradient across the plasma membrane of presynaptic neurons4-6 in a cotransport (symport) mechanism driven by the Na+ electrochemical gradient7, 8. Several members of the prokaryotic NSS family have been identified, namely TnaT9, LeuT10, Tyt111, and MhsT12. These amino acid transport systems, like their eukaryotic counterparts, exhibit a strict Na+ dependence; however, in contrast to the Cl−-dependence of most eukaryotic NSS, all of the prokaryotic NSS studied to date are Cl− independent9-13.
Recently, we showed that introducing a negatively charged amino acid at or near one of the two Na+-binding sites of the GABA (6) transporter GAT1 (S331→E) renders substrate transport Cl− independent, whereas the reciprocal mutations in LeuT and Tyt1 result in Cl−-dependent substrate binding and/or transport13. Interestingly, whereas GAT1-wild-type transport activity is pH independent, GABA transport in the Cl−-independent GAT1 mutant (S331E) is markedly stimulated by acidic internal pH13. We hypothesized that the carboxyl side chain of E introduced in place of S331 can bind and unbind H+ during the transport cycle and that in Cl−-independent NSS, H+ may be counter-transported, i.e. by a Na+/substrate-coupled H+ antiport mechanism, to facilitate the return step of the “empty” transporter13. Indeed, we also observed a pH-dependent substrate transport phenotype for Tyt1-wild-type (WT) in E. coli 11 and when reconstituted into proteoliposomes12. Since both the electrical membrane potential (ΔΨ) and the chemical gradient of ions across the membrane (i.e. ΔμNa+ or ΔμH+) contribute to the distribution of conformational states of a transporter14, we sought to analyze the effect of membrane potential and ionic composition on Na+-coupled Tyr (7) symport by Tyt1 reconstituted in proteoliposomes. In this system, artificial ion gradients ( or ΔμH+) can be imposed across the proteoliposome membrane by varying the internal and external buffer composition, and a membrane potential (ΔΨ; inside negative) can be created by an outward-directed K+ diffusion gradient in the presence of valinomycin (8).
Here we report that the activity of Tyt1 and another Cl−-independent bacterial homolog of the NSS family, MhsT, is dependent on the Na+ gradient and is stimulated by membrane potential (inside negative), and an inversely-oriented H+ gradient. Remarkably, we identify a heretofore unknown H+ flux that is coupled to Na+/substrate symport, consistent with a Na+/substrate symport/H+ counter-transport mechanism.
In a recent study13 we have shown the importance of the charge of the amino acid side chains at position 7.38 or 7.42 (N327 and S331 in GAT1, respectively; for indexing system see15, 16) for the distinction between Cl−-dependent and Cl−-independent transport (Supplementary Table 1). Whereas the Cl−-dependent members of the NSS family possess neutral amino acids at those positions, Cl−-independent NSS members have a negatively charged amino acid at either position. Figure 1a shows the effect of replacing NaCl with Na-gluconate (9) on the time course of Na+-dependent Tyr transport by intact E. coli expressing Tyt1 variants. Whereas Tyt1-WT exhibited no significant difference between the uptake of 1 μM 3H-Tyr in the presence of 10 mM NaCl (3935 ± 176 pmol × mg−1 × min−1) or Na-gluconate (3741 ± 238 pmol × mg−1 × min−1), the Cl−-dependent mutant Tyt1-D2597.38N/A2637.42N mediated transport only when Cl− was present (with an initial rate of 42.4 ± 6.2 pmol × mg−1 × min−1; Fig. 1a, Inset). In the presence of Na-gluconate the activity of this mutant was indistinguishable from that observed in control cells lacking recombinant Tyt1. Despite the significant activity of Tyt1-D259N/A263N in the presence of 100 mM NaCl, the initial rate of transport and the steady state of Tyr accumulation was only ~ 1 % and ~ 10 %, respectively, of that determined for Tyt1-WT at the given test conditions.
Tyt1-WT exhibited initial rates of 1 μM 3H-Tyr uptake that were greatly dependent on the external pH with a peak at pH 8.5, regardless of whether the assay was performed in the presence of NaCl or Na-gluconate (Fig. 1b). Initial rates for Tyt1-WT-mediated Tyr uptake varied between ~ 150 pmol Tyr × min−1 × mg of total cell protein−1 at pH 6.5 and ~ 2500 pmol Tyr × min−1 × mg of total cell protein−1 at pH 8.5. In striking contrast, for Tyt1-D259N/A263N we observed significantly reduced initial rates of 1 μM 3H-Tyr transport in the presence of 100 mM NaCl (about 30 pmol Tyr × min−1 × mg of total cell protein−1), and these rates were independent of pH over the entire range tested. A change in the external pH also had no effect on the transport activity by Tyt1-D259N/A263N in the presence of Na-gluconate; it was only about 20 % of that observed in the presence of NaCl.
Fig. 1c shows the effect of increasing NaCl concentrations on the initial rate of Tyr transport by Tyt1-WT (left panel) and the Cl−-dependent Tyt1 mutant (right panel). In an attempt to improve transport we introduced a third mutation in the Cl− site, T2607.39N to make the sequence more similar to that of the Cl−-dependent NSS members. Combining the D259N/A263N mutations in Tyt1 with T260N resulted in a transporter construct with an enhanced expression level in E. coli or L. lactis, but with other properties nearly identical to the double mutant D259N/A263N (data not shown); consequently, this mutant was used for all subsequent studies. Testing the effect of increasing NaCl concentrations (≤ 50 mM) on the uptake activity of Tyt1-WT revealed an apparent of 1 ± 0.1 mM with a Hill coefficient of 1.5 ± 0.2, consistent with previous results11. NaCl concentrations > 50 mM resulted in inhibition of Tyr transport in intact E. coli, similar to observations in other Na+-dependent symport proteins17. Tyr transport by Tyt1-D259N/T260N/A263N exhibited a ~100-fold increase in the apparent (92 ± 9.4 mM; Hill coefficient of 1.5 ± 0.2 mM [data fit from 0 – 400 mM]; Fig. 1c, right panel). Note, however, that the calculated substantially underestimates the true of this mutant due to the superimposed effect of high Na+ concentrations inhibiting Tyr uptake, as evident in Tyt1-WT.
Expression of recombinant Tyt1 in L. lactis cells and its subsequent purification12 yielded highly pure full-length transporter as judged by protein-analytical techniques (Supplementary Fig. 1 online). Proteoliposomes made from preformed liposomes that were destabilized with Triton X-100 at the onset of solubilization and reconstituted with purified Tyt1 at a protein-to-lipid ratio of 1:200 (w/w) showed the highest transport activity among the conditions tested (Supplementary Fig. 2 online) and were used for all subsequent studies. The Na+-dependence of Tyt1-mediated Tyr transport into proteoliposomes was confirmed by measuring the time course of 1 μM 3H-Tyr in the presence or absence of 25 mM NaCl (Fig. 2a). Diluting proteoliposomes with an internal buffer composition of 100 mM KPi, pH 6.5 into 50 mM Tris/Mes, pH 8.5 in the presence of 25 mM NaCl (generation of and negative ΔμH+ [pHin < pHout]) resulted in an initial rate of Tyr transport (180 ± 12 nmol × min−1 × mg of Tyt1−1) that was ~570 times higher than that in the absence of Na+ (equimolar replacement with Tris/Mes). Consistent with our previous observation of severely reduced Tyr uptake by the Cl−-dependent Tyt1-D259N/T260N/A263N in intact E. coli, reconstitution of this mutant into proteoliposomes produced only about 1 % of the initial rate of 3H-Tyr transport catalyzed by Tyt1-WT in the presence of 25 mM NaCl (Fig. 2b). Note that in contrast to the whole-cell measurements described above in E. coli, these data reflect the protein-specific activity, i.e. secondary effects of other intrinsic membrane proteins in E. coli that may account for the effects reported above can be ruled out. The activity of Tyt1-WT was indistinguishable when measured in NaCl or Na-gluconate. In contrast, transport by reconstituted Tyt1-D259N/T260N/A263N was chloride dependent, as we demonstrated previously in intact E. coli13 - in Na-gluconate 3H-Tyr accumulation was similar to that observed in control liposomes.
Whereas Tyt1-WT in proteoliposomes exhibited a similar apparent as observed in E. coli12, Na+-stimulated 3H-Tyr uptake in Tyt1-D259N/T260N/A263N-containing proteoliposomes did not reach saturation at the highest NaCl concentration tested (400 mM; Fig. 2c), consistent with our results in intact E. coli cells (Fig. 1c, right panel).
Next, we tested whether ΔμH+ is sufficient as the sole driving force of tyrosine transport (Fig. 2d). Tyt1-WT-containing proteoliposomes were created in the presence of 25 mM NaCl at pH 6.5 or pH 7.5. Dilution of these proteoliposomes in 25 mM external NaCl () resulted in Tyr uptake activity that was indistinguishable from the activity observed in control liposomes under the same experimental conditions, regardless of whether the direction of ΔμH+ was inward or outward. Likewise, in the absence of ΔμNa+ the generation of a valinomycin-mediated ΔΨ (in the presence or absence of ΔμH+ in either direction) resulted in Tyr uptake by Tyt1-WT identical to that observed in control liposomes (data not shown).
Figure 3a shows the effect of an inwardly-directed membrane potential (ΔΨ) on Na+-coupled Tyr transport in the absence of a pH difference across the liposome membrane (pHin = pHout = 6.5; ΔμH+ = 0). The initial rate of Na+-driven transport (determined after 10 seconds) was 9.4 ± 2.4 nmol Tyr × min−1 × mg Tyt1−1. This rate was increased almost 2-fold (18.1 ± 3.9 Tyr × min−1 × mg Tyt1−1) when a K+ diffusion membrane potential was generated by the addition of valinomycin. In the presence of gramicidin, which dissipated ΔμNa+, the accumulation of 3H-Tyr in Tyt1-containing proteoliposomes was indistinguishable from that observed in control liposomes (not shown). In contrast to the result obtained in Fig. 3a, a valinomycin-induced outwardly-directed membrane potential (−ΔΨ) blunted Na+-coupled Tyr transport (Fig. 3b) even under optimized ΔμH+ conditions (pHin = 6.5, pHout = 8.5; see Fig. 3c).
We performed a systematic comparison of the effect of ΔμH+ on the -driven transport reaction across the liposome membrane (Fig. 3c). Transport increased with increasing pHout and decreasing pHin, yielding the highest value (250.9 ± 15.9 nmol Tyr × mg Tyt1−1 × min−1) at pHin of 6.5 and pHout of 8.5.
According to our hypothesis13, the negatively charged side chain of aspartate/glutamate 7.38 or 7.42 has to be protonated for the reorientation step of the transporter. If this proton was released to the extracellular milieu upon reorientation of the transporter, H+ antiport would be associated with Na+ symport and the transport cycle would be accelerated by an outward H+ gradient. Indeed, although ΔμH+ did not drive Tyr transport in the absence of ΔμNa+ (Fig. 2c), the initial rate of Na+-dependent Tyr transport by Tyt1-WT in proteoliposomes with an internal pH of 6.5 was increased by about 500 % when the external pH was raised from 6.5 to 8.5 (Fig. 3c). The same qualitative increase was observed at an internal pH of 7.5 and 8.5, but to a lesser extent (see Fig. 3c in which the transport rates are shown). In contrast, under the same experimental conditions Tyt1-D259N/T260N/A263N-mediated Tyr transport was maximally stimulated only to about 80 %. This result clearly demonstrates the combined effects of ΔΨ, ΔμNa+, and negative ΔμH+ on Tyt1-WT-mediated substrate transport, whereas the ΔμH+ effect is absent or blunted in the Cl−-dependent mutant.
To assess the substrate binding properties of the individual Tyt1 variants, we performed equilibrium binding studies by means of scintillation proximity12. In striking contrast to 3H-Tyr flux measurements, binding of 1 μM 3H-Tyr to purified Tyt1-WT and to Tyt1-D259N/T260N/A263N was pH-independent over a range from pH 6.5 to 9.0 (Fig. 4a). This result suggests that the H+ dependence of uptake occurs at a step in the cycle after substrate binding. 3H-Tyr bound to purified Tyt1-WT and -D259N/T260N/A263N with a similar of 5.8 ± 0.8 μM and 8.4 ± 1.7 μM, respectively, when measured in the presence of 200 mM NaCl at pH 7.5. Higher NaCl concentrations did not exert an inhibitory effect on the equilibrium binding activity of Tyt1-WT (data not shown), and the apparent was ~ 90 mM as we showed previously12. In contrast, binding of 1 μM 3H-Tyr to Tyt1-D259N/T260N/A263N did not saturate at NaCl concentrations as high as 400 mM (data not shown).
The apparent affinity of Tyr uptake () in proteoliposomes with an internal pH of 6.5 remained the same (~ 0.5 μM) when the external pH was varied between 6.5 and 8.5 (Fig. 4b). Consistent with the notion of protonation of the transporter being required for maximum substrate flux, increasing the external pH in those experiments from 6.5 to 8.5 resulted in a ~ 6-fold increased maximum velocity of Tyr transport ( of 48 ± 5.1, 253 ± 11.8, and 292 ± 8.1 nmol × mg Tyt1−1 × min−1 at pHout 6.5, 7.5, and 8.5, respectively), and, in turn, in the turnover number of transport (ktrsp = 0.04 ± 0.004 × s−1, 0.2 ± 0.01 × s−1, and 0.25 ± 0.007 × s−1). Similar to the reduced initial rates of Tyr transport in whole-cell measurements (Fig. 1), at 200 mM NaCl the Cl−-dependent/pH-independent Tyt1-variant D259N/T260N/A263N exhibited a ~ 62-fold reduced velocity of transport (; ktrsp = 0.004 ± 0.0002 × s−1)) and a ~9-fold increased (4.9 ± 0.8 μM), compared to Tyt1-WT (Fig. 4c).
Given the stimulating effect of a negative ΔμH+ on substrate transport by Tyt1 reconstituted in proteoliposomes, we speculated that Na+-dependent substrate symport by Tyt1 is coupled to H+ antiport. To test this hypothesis we measured substrate-induced intra-compartmental pH changes using fluorescent probes. Significant ligand-induced changes in the 5(6)-carboxyfluorescein succinimidyl ester (cFSE) fluorescence intensity ratio, indicative of alkalization of the cytoplasm, were detected in L. lactis NZ9000 expressing recombinant Tyt1 but not in control cells (Fig. 5a). Small, albeit consistent fluorescence changes were observed when only the co-transported cation Na+ was added to the cell solution, but the simultaneous addition of Na+ and Tyr led to the largest signal. Similar results were observed when another bacterial NSS homolog, MhsT, a Na+-dependent hydrophobic amino acid transporter of the alkaliphilic B. halodurans12 was expressed in L. lactis (Fig. 5b). The cytoplasmic alkalization reached saturation within ~2 min upon the addition of Na+ and Tyr, consistent with the time course of 3H-Tyr transport mediated by Tyt1 (Fig. (Fig.1a1a and and2a2a).
To validate our findings in a cell-free system, we measured changes in pyranine fluorescence in proteoliposomes reconstituted with purified Tyt1 (Fig. 5c). Significant alkalization of the proteoliposomes’ lumen was observed when Na+ and Tyr were added together, and, to a smaller extent when Tyr was added in the nominal absence of Na+ (Fig. 5c, left panel). Rapid alkalinization was also observed upon addition of Na+ and Tyr in the presence of the K+ ionophore valinomycin (Supplementary Fig. 3 online), excluding the possibility that H+ extrusion results from the accumulation of positive charge in the lumen of the proteoliposomes due to the electrogenic nature of Tyt1 function. In contrast, no significant changes were detected upon addition of the substrates to control liposomes devoid of Tyt1 or to proteoliposomes containing Tyt1-D259N/T260N/A263N (Fig. 5c, right panel). While this is consistent with the lack of H+ antiport in this mutant, we cannot rule out the possibility that the low transport activity of this mutant makes it impossible to observe changes in the fluorescence.
To obtain an estimate of the coupling stoichiometry of Na+/Tyr symport-coupled H+-translocation we calculated (maximum velocity of H+ translocation; see SUPPLEMENTARY METHODS) of three independent experiments (51.3 ± 2.3 nmol H+ × (mg Tyt1 × min)−1) and compared it with the obtained in Tyt1-containing proteoliposomes at pHin = pHout = 7.5 (38.4 ± 0.6 nmol Tyr × (mg Tyt1 × min)−1). Under those experimental conditions, the H+-to-Tyr coupling ratio was calculated to be 1.34 ± 0.06. Note, however, that this estimate is based on a number of approximations of liposome size and geometry (see SUPPLEMENTARY METHODS). [Au: Appropriate to call out Supplementary Methods in both instances in this paragraph, as we do not allow callouts to Supplementary Information?]
The structural rearrangement associated with opening the “cytoplasmic gate” of the protein to enable release of Na+ and substrate to the intracellular milieu changes the local environment around E290 and greatly increases the penetration of water into the permeation pathway18. To understand the consequences of these changes in the local environment produced by the modeled conformational rearrangement, we carried out comparative pKa calculations using the MM_SCP approach19 and molecular models of LeuT and Tyt1 embedded in lipid bilayer membranes. The molecules were either in the crystal structure-like conformation, or in the modeled inward-facing conformations derived from our intracellular-pulling Steered Molecular Dynamics (SMD) study18. Comparing the pKa obtained from the calculated acid/base equilibrium of E290 in the crystal structure-like conformation of LeuT (pKa = 3.6) to that in the inward-facing model (pKa = 6.9) shows that the conformational change associated with transport produces a very significant increase in the propensity for the protonated state of the acidic side chain. Similar results were obtained for the D259 of Tyt1 (pKa values of 2.7 and 6.4, respectively). These results support our interpretation of the observed pH effect on transport.
Analysis of the mechanism producing the calculated changes in the pKa values demonstrates how the pH dependence is integrated in the functional mechanism of the transporter. The pKa values of 3.6 for E2907.38 in LeuT and 2.7 for D2597.42 in Tyt1 in their crystal structure-like conformations indicate that they are primarily ionized (deprotonated) at physiological pH, in agreement with a previous study20. In both cases, the results pertain to the occluded state of the transporter in which E290 and D259 are deeply buried in the protein core and in complex with Na1. Analysis of the microenvironment around E290 shows that the titratable moiety is surrounded mostly by polar residues in addition to the Na+ ion, which yields a hydrophilic microenvironment that is fairly typical for D or E21. This is reflected in the values, calculated as described in METHODS, of the interaction energy (wint) and the transfer energy (Δwtr) of −3 and 2 Kcal/mol, respectively; these contributions largely cancel each other out, and in the present case lead to only a modest lowering of the pKa value from its reference value of 4.4 in water. Thus, the fact that the transfer energy is > 0 shows that, in spite of the effect of Na+, the microenvironment is still less hydrophilic than if it consisted entirely of water and hence the modification in pKa is small. Interestingly, the pKa value of E290 in the crystal structure-like LeuT model without Na+ and substrate18 is 1.7 units above the reference value, as compared to 0.8 pKa units below the reference value when Na+ is in the immediate vicinity of E290.
An additional, function-related alteration in pKa is caused by significant changes in the microenvironment of E290 that result from the adoption of the inward-facing conformation. Although the titratable groups of E290 (and D259) are still mostly buried in protein (~ 95 %), which shields them from the solvent, the structural rearrangement and the interactions of polar residues with a few isolated water molecules produce a significant change in the microenvironment of the acidic side chains. Thus, in the inward-facing conformation, the neighborhood of the buried acidic residues is composed of more hydrophobic residues than before the rearrangement to the inward-facing conformation (even the surrounding polar residues contribute mostly their hydrophobic portions to this environment). The net result is that the local environment around E290 is much more hydrophobic than in the LeuT crystal structure10, which increases the energy required for transferring the residue from water into the protein to 4.5 Kcal/mol. The interaction energy is −1.1 Kcal/mol, yielding a net gain of 3.4 Kcal/mol, which shifts the value of the pKa upwards (relative to the reference value) by 2.5 pH units, to 6.9.
Here we have demonstrated the existence of Na+/substrate symport-mediated H+ antiport by Cl−-independent members of the NSS family, a novel mechanistic finding for this class of membrane proteins. The ‘classical’ form of metabolic energy gradient across bacterial membranes, according to the chemiosmotic hypothesis of Mitchell22 is the protonmotive force (pmf). This driving force consists of an electrical membrane potential (ΔΨ) and a chemical proton gradient (ΔμH+) across the bacterial plasma membrane. For many Na+ dependent solute symport systems, such as members of the Solute:Sodium Symporter (SSS) family (e.g. PutP23 and hSGLT124), the Dicarboxylate/Amino Acid:Cation Symporter (DAACS) family (e.g. SdcS25), and the Bile Acid:Sodium Symporter (BASS) family (e.g. Abst26) ΔμH+ contributes to the electrochemical force of –driven substrate symport due to its effect on ΔΨ. In striking contrast to this prototypical scheme for Na+-coupled solute transport, however, we show here that ΔμH+ directly acts against substrate transport catalyzed by the Cl−-independent NSS members Tyt1 and MhsT. Thus, detailed kinetic studies of Tyt1 in proteoliposomes, (Fig. (Fig.22&3), revealed that a negative ΔμH+ is necessary to maximally stimulate Na+-coupled substrate symport. In addition, our results in intact cells and Tyt1-containing proteoliposomes indicate that Na+/substrate symport creates significant changes in the internal pH, consistent with internal alkalization stemming from H+ export (Fig. 5).
In contrast to the Cl−-independent Tyt1-WT and MhsT-WT, the constructs in which the aspartate at position 7.38 (D259 and D263 in Tyt1 and MhsT, respectively) was mutated to asparagine, are Cl−-dependent and exhibit diminished dependence on the negative ΔμH+. Thus, the Cl−-dependent members of the NSS family have neutral amino acids at position 7.38 or 7.42, whereas all Cl−-independent NSS members have a negatively charged amino acid at one of these positions (Supplementary Table 1). Together with our recent findings13, the results of the current work allow us to propose that the negatively charged amino acids located at position 7.38/7.42 are directly involved in H+ translocation (Fig. 6a). The involvement of acidic amino acids in H+ translocation has been suggested for H+ translocating proteins, i.e. LacY27 and in the H+-translocating F1Fo ATPase of E. coli (for review see28). Consequently, we interpret our results indicating that the conformational transitions associated with the transport cycle modulate the microenvironment of the charged residue and, in turn, alter its pKa, to suggest a facilitated protonation/deprotonation during the transport cycle.
Analysis of the significant pKa increase by > 3 units for E290 in LeuT and D259 in Tyt1 when comparing the inward-facing to the crystal structure-like conformation shows that E290 in LeuT and D259 in Tyt1 become protonated in the inward-facing conformations in which water molecules penetrate and modulate their local microenvironments. This is likely required in the inward facing state for the Na+ and substrate-depleted transporter to facilitate movement to the outward facing configuration for the next cycle (Fig. 6 a). Such a proposed mechanism is supported by the finding that Tyr transport by Tyt1-WT is increased by 500% when the external pH is raised from pH 6.5 ([H+] = 0.316 μM) to pH 8.5 ([H+] = 0.003 μM) while the internal pH is maintained at pH 6.5 (Fig. 3c), thus generating a negative ΔμH+. In contrast to the Cl−-dependent members of the NSS family, e.g., in GAT1 (Fig. 6 b) where reorientation of the transporter occurs with a neutral side chain at 7.38/7.42, in the Cl−-independent NSS members, as well as the Cl−-independent GAT1-S331E mutant13, the negative side chain of D or E at position 7.38 or 7.42 accepts an H+ (or H3O+) due to the changes in the pKa. Reorientation appears to be the rate limiting step in the transport cycle, as judged by the dramatic effect of ΔμH+ in stimulating turnover. However, substrate binding to the mutated Tyt1 variant, which lacks the vectorial component of Tyt1-mediated substrate translocation across the membrane, is essentially unaffected by pH (Fig. 4a).
Note however, that whereas Keynan & Kanner7, Hilgeman & Lu29, and Zomot et al.13 propose the coupled translocation of 1 GABA, 2 Na+ and 1 Cl− by GAT1, there is also evidence in GAT1 and DAT suggesting that under some conditions there is no net transport of Cl− (“Cl−-cycling”)30,31. In addition, Na+/Cl−/serotonin symport-coupled K+ and/or H+ antiport by SERT has been proposed32-34..
In spite of these observed differences, a unifying feature of the proposed mechanism for all NSS members is that the charge of the cotransported Na+ ions must be compensated. Our findings point to the general conclusion that this is accomplished either by acidic groups in the protein (H+ antiporting NSS members), or by the movable charge of the co-transported Cl− (Cl−-dependent NSS members).
From a structural perspective (Fig. 7), [Au: ok to call out Figure 7 as a whole here, to avoid rearranging text about Fig. 7c and 7b below?] E290 in LeuT, a titratable residue, is closely associated with the Na+ binding pocket and is in direct interactions with two Na1 site residues, T254 and N286 (Fig. 7a); conversely, as described in RESULTS, Na1 contributes significantly to the hydrophilic environment of E290, which, in combination with the Glu−-Na+ electrostatic interaction renders E290 deprotonated at physiological pH. In turn, the rearrangement of the microenvironment in the absence of Na1 makes it less hydrophilic, destabilizing the deprotonated form of E290. Thus the Na+ binding/dissociation and the E290 deprotonation/protonation events are likely to be closely associated in the transition between inward- and outward-facing conformations. In our well-equilibrated Tyt1 model, similar orientation of D259 relative to the Na1 site is also observed, though D259 is one turn away from the aligned position of E290 in LeuT (Fig. 7c). Such immediate proximity of Na+ and H+ binding sites along their transport pathway is reminiscent of NhaA, an Na+/H+ antiporter35. Based on extensive simulation, Arkin et al. proposed that in NhaA, D164 is a Na+ binding site and D163 controls the alternating access mechanism, which includes protonation/deprotonation events of both D163 and D164 and results in Na+/H+ antiport36 (Fig. 7b).
In terms of charge movements, Na+/H+ antiport is equivalent to the symport of Na+/Cl−. Interestingly, in the Cl−-dependent mammalian NSS the Cl− site has been found as well to be close to the Na1 site13. We find that in our DAT model with embedded Na+ and Cl−37, S321 and N353 consistently interact with both Na1 and Cl− along the 16 ns equilibration trajectory (Fig. 7d). This evidence for direct coupling of Na+ and H+ (or Cl−) binding suggests that the structural features surrounding E290 and the Na1 site of LeuT may indeed bring about the Na+/H+ antiport mechanism of Cl−-independent NSS revealed by this study.
The literature shows that the use of different solutions to achieving the same transport phenotype in closely related proteins (or in proteins in which an acidic amino acid (D or E) was replaced with a polar one (T or S)), is not to be unique to the NSS family. For example, replacement of D85 with T in the light-driven proton pump bacteriorhodopsin transformed the mutated bacteriorhodopsin into a chloride ion pump, which, like halorhodopsin, actively transports chloride ions in the direction opposite from the proton pump38.
Despite the opposite charge of the coupling ions, the NSS family of proteins shares a similar molecular mechanism if it is understood that a key step in the reaction cycle (Fig. 6) of NSS is the maintenance of an ionic counterbalance that compensates for the translocated Na+.
Tyt1 and MhsT were expressed in E. coli MQ614 [SVS1144 mtr aroP tnaB271::Tn5 tyrP1 pheP::cat]11 or Lactococcus lactis NZ9000 using pQ2 and pQmhsT or pNZ2 and pNZmhsT, respectively12. Mutations were introduced in the tyt1 or mhsT genes according to the procedures described in the QuikChange® Site-Directed Mutagenesis Kit (Stratagene) using mutagenic oligonucleotides (Invitrogen) with pQ2 or pQmhsT as template. The following amino acid replacements were introduced in Tyt1 ([D259N260A263]→[N259S260S263]) and MhsT (D268→S268). All modifications were confirmed by DNA sequencing (Agencourt Bioscience Corp).
Purification, resonctitution, and functional assays were performed as described12 (see also SUPPLEMENTARY METHODS online). Briefly, purified protein was reconstituted in pre-formed liposomes made of E. coli polar lipid extract and egg L-α-phosphatiylcholine (both Avanti) at a 2:1 (w/w) ratio. Protein activity was measured by means of transport studies in proteoliposomes and binding studies using the scintillation proximity assay (SPA) using Cu2+ chelate YSi scintillation SPA beads (GE Healthcare) in conjunction with the poly-His-tagged recombinant proteins. 0.5 μg of purified and desalted protein were used per assay in assay buffer composed of Tris/Mes and Na-salt (equimolar replacement of Tris/Mes with Na-salts to obtain a total of 450 mM) and pH as indicated/20% glycerol/1 mM TCEP/0.1% n-dodecyl-β-D-maltopyranoside and the indicated concentration of L-3H-Tyr. For the determination of the dependence of the ion species/concentration on radioligand binding to the purified protein, the eluted protein fraction was desalted with Zeba® Desalt Spin Columns (Pierce) previously pre-equilibrated with 200 mM Tris/Mes, pH 7.5/20 % glycerol/1 mM TCEP/0.1% n-dodecyl-β-D-maltopyranoside.
L. lactis NZ9000 were cultivated and harvested as described above, washed and resuspend in 50 mM HEPES-K, pH 8.0. Incorporation of 1 μM 5-(and -6)-carboxyfluorescein diacetate succinimidyl ester (cFDASE; Invitrogen), which is subsequently hydrolyzed by esterases to 5 (and 6-)-carboxyfluorescein succinimidyl ester (cFSE) in the cytoplasm, was achieved by incubation for 15 min at 30 °C39. The suspension was washed and resuspend in 50 mM potassium phosphate buffer, pH 7.0, and 10 mM lactose was added for 30 min to eliminate non-conjugated cFSE. Free cFDASE was removed by washing the cells three times in 50 mM potassium phosphate buffer. For the determination of the internal pH changes within L. lactis, the cells were diluted to an optical density at 600 nM (OD600) of 0.1 in 50 mM Tris/Mes, pH 7.5. The fluorescence intensity was measured at an excitation wavelength of 490 nm and excitation wavelength of 530 nM (at 5 nm slit widths). The effect of ligand-induced changes was tested by the addition of 20 mM NaCl and/or 10 μM Tyr after the recordings reached a stable baseline (routinely 200 s).
pH changes in proteoliposomes were measured by changes in pyranine fluorescence (excitation wavelength of 450 nm, excitation wavelength of 508 nM; 5 nm slit widths)40, 41. Pyranine-containing proteoliposomes were diluted into 5 mM potassium phosphate, pH 7.4 to a concentration of 4 μg protein/mL, and 20 mM NaCl and/or 10 μM Tyr were added after the fluorescence intensity was stable.
All experiments were repeated at least in duplicate. Figures represent typical experiment, and, unless otherwise noted, data points represent the mean of a triplicate determination (n = 3) +/− S.D. Data fits of kinetic analyses were performed using non-linear regression algorithms in Sigmaplot (SPSS Inc., Chicago, IL) or Prism (GraphPad, San Diego, CA) and errors represent the S.E.M. of the fit.
pKa calculations for particular amino acid residue side chains were performed according to the MM-SCP (Microenvironment Modulated-Screened Coulomb Potential) computational approach for calculating acid/base equilibria in proteins19. This method is based on expressing the protonation state of a titratable residue in a protein in terms of a thermodynamic cycle42, 43 and is described in detail in the SUPPLEMENTAL METHODS.
We thank Dr. Baruch Kanner for helpful discussion, Dr. Mary Ann Gawinovicz of the Columbia University Medical Center Protein Core Facility for the MALDI-TOF analysis of Tyt1, and Lynn Chung for technical assistance. Computations were performed on the Ranger at the Texas Advanced Computing Center (TG-MCB090022) and the computational infrastructure of the Institute for Computational Biomedicine at Weill Cornell Medical College. This work was supported by National Institutes of Health Grants DA022413 and DA17293 (JAJ), P01 DA012923 (HW), DA015170 (ELM), and K99 DA023694 (LS).
The authors declare no competing financial interests.