The Importance of Being Aromatic: Pi interactions in Sodium Symporters

Abstract

In the LeuT family of sodium solute symporters 13-17% of the residues in transmembrane domains are aromatic. The unique properties of aromatic amino acids enable them to play specialized roles in proteins, but their function in membrane transporters is underappreciated. Here we analyze the π bonding pattern in the LeuT (5TMIR) family, and then describe the role of a triad of aromatic residues in sodium-dependent sugar cotransporters (SGLTs). In SLC5 symporters 3 aromatic residues in TM6 (SGLT1 W289, Y290 and W291) are conserved in only those transporting sugars and inositols. We used biophysical analysis of mutants to discover their functional roles, which we have interpreted in terms of CH-π, π-π, and cation-π bonding. We discovered that: 1) glucose binding involves CH-π stacking with Y290; 2) π T-stacking interactions between Y290 and W291 and H-bonding between Y290 and N78 (TM1) are essential to form the sodium and sugar binding sites; 3) Na⁺:sugar stoichiometry is determined by these residues; and 4) W289 may be important in stabilizing the structure through H-bonding to TM3. We also find that the WYW triad plays a role in Na⁺ coordination at the Na1 site, possibly through cation-π interactions. Surprisingly, this Na⁺ is not necessarily coupled to glucose translocation. Our analysis of π-interactions in other LeuT proteins suggests that they also contribute to the structure and function in this whole family of transporters.

An advance in understanding symporters and antiporters was the crystallization of 7 transporters from 7 unrelated gene families. These proteins have a common architecture, the LeuT fold, consisting of an inverted repeat of 5 trans-membrane helices (5TMIR) and a central occluded substrate binding site (1-4). This, along with biophysical and biochemical evidence, indicates that the LeuT (5TMIR) structural family share a common alternating access transport mechanism.

The structures are rich in aromatic residues at the ends of helices and in the membrane interior, e.g., the aromatic residues are 15-17% of the total in the transmembrane (TM) core domain of vSGLT and LeuT (TM1-TM10) (1, 2, 5) (Figures 1). Aromatics can form two types of bonds: CH - π bonds between pairs of aromatic side chains in a T-stacking or an off-centered parallel orientation (6, 7), and cation – π bonds that occur between basic and aromatic side chains (8, 9). In general, CH-π bonds in proteins occur in networks and membrane proteins have a larger number of such networks than soluble proteins suggesting a role in membrane protein stability (10); There is computational evidence that there may be cooperatively between adjacent π-π and cation – π bonding sites (11). Aromatic bonds in LeuT transporters form networks between TM helices of membrane proteins: in vSGLT the network engages TM6 with TM1, 2, 3, and 10, and TM2 with TM10 (Figure 1A); and in LeuT the network ties TM1 with TM2 & TM6, TM3 with TM4, TM8 & TM10, TM4 with TM9, and TM5 with TM8 (Figure 1B).

Figure 1.

(A) Aromatic network in vSGLT. Potential aromatic interactions between TM in the core structure of vSGLT (TM1-10) was extracted from two conformations of vSGLT, inward facing occluded (3DH4) and inward facing open (2XQ2) (2, 5). The linkages shown are those conserved amongst vSGLT and hSGLTs 1-6. The cation-π interactions were identified using the computer program (CaPTURE) and the H-bond and CH – π interactions were identified by visual inspection of the structures. For cation-π and H-bonds, the solid lines are interactions present in both 3DH3 4 and 2XQ2, whereas the broken lines are interactions not present in both conformations. CH - π interactions are present in one or both conformations.

(B) π Interaction network in LeuT. The potential aromatic interactions between TM in the core structure of LeuT (TM1-10) was extracted from three conformations of LeuT, outward facing open (3TT1), outward facing occluded (2A65), and inward facing open (3TT3) (1, 21). The cation-π interactions were identified using the computer program (CaPTURE) and the CH – π interactions were identified by visual inspection of the structures. Solid lines are cation-π interactions present in all three conformations, broken lines are interactions not present in all three conformations. CH - π interactions are present in one or more conformations.

Aromatic residues make unique associations with other residues, e.g., Y263 in vSGLT is part of a triad of aromatics on TM6 (Y262, Y263, W264) conserved in glucose and inositol transporters but not the choline, iodide, multivitamin or short chain fatty acid transporters (Figure 2) (12, 13). Y263 forms a T-stacking π–π bond with W264, and Y262 is H-bonded to W134 on TM3. Whereas Y263 is the inner gate (2, 5), the outer gate is formed in part by a π–π bond between Y87 (TM2) and F424 (TM10). The triad is part of a larger network of inter-helical aromatic bonds in vSGLT (Figure 1A).

Figure 2. Sequence alignment of helix 6 of 12 human members of the SLC5 gene family.

The amino acid sequence of helix 6 was from the crystal structure of Vibrio parahaemolyticus vSGLT, and the human sequences were aligned using ClustalW (http://www.ebi.ac.uk) with manual adjustments. Strictly conserved and highly conserved residues are highlighted in red and blue, respectively (single letter code or red dashes). Gaps are indicated by black dashes. The location of the aromatic motif is highlighted in yellow box. SMIT, sodium myo-inositol; CHT, choline; SMVT: sodium multivitamin; SMCT: sodium monocarboxylic acid; NIS, sodium iodide symporter (13).

There is high amino acid identity and similarity between vSGLT and hSGLT1 (32% identity, 69% similarity) and key ligand binding and gate residues are conserved, including the aromatic triad, W289, Y290 and W291 (Figure 2 and (14)). Unlike vSGLT, glucose transport by hSGLT1 is coupled to 2 Na⁺ ions. Of the two Na⁺ binding sites in hSGLT1, the putative Na2 site has been identified by homology with vSGLT and LeuT, but the identity of Na1 is unknown (14). In LeuT the Na1 site is located close to the leucine binding site (1). In single cells, we are able to measure transport kinetics of wild-type and mutant human SGLT1 as a function of voltage (K_0.5 values and Hill coefficients for sugar and Na⁺), the number of transporters in the plasma membrane, and transport coupling ratio between Na⁺ and sugar. Such assays are not yet readily available for bacterial symporters. In this study we have mutated in turn the conserved triad residues, W289, Y290, and W291 in hSGLT1 (Figure 2) and measured kinetics of Na⁺/sugar symport.

Our results show that the triad is a major determinant in Na⁺ and sugar binding and transport by SGLTs, and that aromatic bonding (cation–π, CH–π) plays a central role in the function of hSGLT1 and other members of the LeuT structural family.

MATERIALS and METHODS

Oligonucleotide-directed mutagenesis

Mutations in hSGLT1 were introduced in the plasmid hSGLT1-pBluescript using QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The mutagenic primers were designed taking into consideration the codon usage for human to optimize the protein expression. All constructs were verified by full gene sequencing.

Expression of hSGLT1 and mutants in Oocytes

cRNA was synthesized and injected into Xenopus laevis oocytes as described in supplementary information (14). Plasmid hSGLT1-pBluescript and its mutants were linearized with XbaI. Capped cRNAs were synthesized in vitro (T3 mMESSAGE mMACHINE kit, Applied Biosystems, Austin, TX). Stage V and VI Xenopus laevis oocytes were injected with 50 ng of cRNA and incubated at 18° C for 4-7 days in Barth’s solution (88 mM NaCl, 10 mM HEPES-Tris, 2.4 mM NaHCO₃, 1.0 mM KCl, 0.8 mM MgSO₄, 0.4 mM CaCl₂, and 0.3 mM Ca(NO₃)₂, pH 7.5) supplemented with 5 μg/ml gentamicin, 100 μg/ml streptomycin, 100 units/ml penicillin, and 5.75 μg/ml ciprofloxacin. This protocol follows guidelines approved by the UCLA Chancellor’s Committee on Animal Research.

Electrophysiological experiments

Steady state kinetics was performed using a two-electrode voltage clamp (14). The oocytes were bathed in NaCl buffer (100 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, and with 10 mM HEPES-Tris, pH 7.5) with or without α-methyl-D-glucopyranoside (αMDG, a nonmetabolized sugar). A standard pulse protocol was used to determine the maximal presteady state charge movement Q_max, which is an index of the number of transporters expressed in the oocyte plasma membrane, and to measure the effect of membrane potential on transport. The membrane potential of oocytes was held at −50 mV (V_h) and stepped from −50 mV to −150 mV in 20 mV decrements for 100 ms before returning to V_h. All electrophysiological experiments were performed at 22°C. Data were recorded using pClamp software (Axon Instruments, CA).

Transporter expression

The number of wild-type and mutant SGLT1 proteins expressed in the oocyte plasma membrane was estimated by measuring Q_max (maximal pre-steady state charge movement) in the presence of 100 mM Na buffer (15). Q_max provides an index of the number of transporters: 1 nC of Q_max is equivalent to 1 × 10⁹ SGLT1 proteins expressed in the plasma membrane (15, 16). Q_max was estimated by fitting the charge (Q) versus voltage data to a single Boltzmann function: (Q – Q_hyp)/Q_max = 1/[1 − exp (zδ(V_m – V_0.5) F/RT], where Q_max = Q_dep – Q_hyp, Q_hyp and Q_dep are the charge at the hyperpolarizing and depolarizing limits, V_m is membrane potential, V_0.5 is midpoint voltage, and zδ is apparent valence of the voltage sensor.

The data in Table 1 shows that the expression level all the 290 and 291 mutants was comparable to the wild-type hSGLT1, > 10 × 10⁹ transporters in the plasma membrane (Q_max > 10 nC). Expression for W289C was approximately 25% of that for wild type, 5 × 10⁹ transporters in the membrane but for W289F and W289Y we were unable detect Q values above the control oocytes, probably due to defects in trafficking of these proteins from the cytoplasm into the plasma membrane.

Table 1. Summary of wild-type and mutant human SGLT1 kinetics.

	K0.5αMDG (mM)	K0.5Na (mM)	nNa	Imax (nA)	SGLT density (×109)	“Turnover #” (s−1)
SGLT1	0.6 ± 0.2	3 ± 0.1	1.5 ± 0.1	930 ± 232	14 ± 4	69 ± 4
W289C	n.m.	n.m.	n.m.	n.m.	5 ± 2	n.m.
W289Y	~100	n.m.	n.m.	40 ± 14	n.m.	n.m.
Y290C	>>100	32 ± 4	1.6 ± 0.1	1315 ± 102	11 ± 1	>125 ± 5
Y290F	35 ± 10	20 ± 1	1.8 ±0.2	1280 ± 361	15 ± 1	87 ± 19
W291C	>100	72 ± 29	1.9 ± 0.3	200 ± 48	9 ± 2	>19
W291F	4 ± 1	3 ± 1	1.5 ± 0.1	1269 ± 361	21 ± 7	61 ± 3

Kinetics parameters are the means ± S.E. of 3-5 oocytes from at least 2 donor frogs. K_0.5^αMDG and K_0.5^Na are the half-saturation concentrations for αMDG and Na⁺; n^Na is Hill coefficient; and I_max is maximal sugar-induced current. The number of transporters expressed in the plasma membrane is estimated from the maximal charge Q_max: 1 nC of Q_max is equivalent to 1 × 10⁹ SGLT1 proteins expressed in the oocyte plasma membrane. The “turnover #” is obtained from the ratio I_max^−150mV/Q_max. All parameters were estimated at −150mV, except n^Na, which was at −50 mV. Note that K_0.5^αMDG is independent of membrane voltage between −50 and −150 mV while K_0.5^Na is voltage insensitive for voltages more negative than −110 mV (13).

Measurement of steady state kinetics

Cotransport rates were obtained by subtracting the basal Na⁺ current from the total current in the presence of sugar. The maximal sugar-induced current current (I_max) and apparent sugar affinity (K_0.5^αMDG) were then estimated by fitting the I_αMDG versus [αMDG]_o data to:

$${\mathrm{I}}_{\alpha \mathrm{MDG}}={\mathrm{I}}_{\max }{\left[\alpha \mathrm{MDG}\right]}_{\mathrm{o}}∕({{\mathrm{K}}_{0.5}}^{\alpha \mathrm{MDG}}+{\left[\alpha \mathrm{MDG}\right]}_{\mathrm{o}})$$ Equation 1

The apparent affinity for Na⁺ (K_0.5^Na) was measured in different concentrations of Na⁺ with saturating concentration of αMDG. For the mutants with high K_0.5^αMDG, i.e. > 100 mM, we estimated the kinetic constants in the presence of 100 mM αMDG. The external Na⁺ concentration was varied by replacing NaCl with choline chloride. The sugar-induced current (I_αMDG) was plotted as a function of external Na⁺ concentration ([Na]_o) and fitted to equation 2, where n is the Hill coefficient:

$${\mathrm{I}}_{\alpha \mathrm{MDG}}={\mathrm{I}}_{\max }{\left({\left[\mathrm{Na}\right]}_{\mathrm{o}}\right)}^{\mathrm{n}}∕[{\left({{\mathrm{K}}_{0.5}}^{\mathrm{Na}}\right)}^{\mathrm{n}}+{\left({\left[\mathrm{Na}\right]}_{\mathrm{o}}\right)}^{\mathrm{n}}]$$ Equation 2

Stoichoimetry Determinations

The Na⁺ to sugar transport ratio was determined by measuring radiolabeled sugar or Na⁺ uptakes and sugar-induced inward charge simultaneously (14, 17, 18). In brief, oocytes expressing the mutant or wild-type transporter were mounted in a micro-flow chamber in the two-electrode voltage clamp and continuously superfused with Na⁺ buffer (in mM: 100 NaCl, 2 KCl, 1 MgCl₂, 1 CaCl₂, 10 HEPES) buffered with Tris to pH 7.5. The membrane potential of the oocyte was clamped at −50 or −90 mV. The experiment was initiated after a stable baseline current was reached by adding 1mM αMDG containing ¹⁴C-αMDG tracer (Perkin Elmer). After the desired uptake interval, αMDG was removed and the oocyte washed until the baseline was re-established. αMDG uptake into each oocyte was determined by scintillation counting. Charge uptake was determined by integrating the current record over the time of the experiment, and converting the inward charge from coulombs to p-equivalents of Na⁺ using Faraday’s constant. To confirm that inward charge was indeed equivalent to Na⁺ uptake, parallel experiments were performed to simultaneously determine sugar-dependent inward charge and ²²Na uptakes using the same protocol as for charge and sugar uptakes. To optimize the specific activity of isotope these ²²Na measurements were carried out at 10 mM NaCl (80 mM Choline chloride replacing 80 mM NaCl) and αMDG concentrations varying between 10 and 50 mM to generate sufficient sugar –induced currents.

Statistics

Fits of data to equations were performed using either Sigmaplot 10 (SPSS, Inc., Chicago IL) or Clampfit 10.1 (Axon Instruments, Union City, CA). For data obtained on a single oocyte, the statistics are given by the estimates and the error of the fit. When data is from a population, the statics are given by the means and the standard error of the means.

Structural analysis

Aromatic interactions within the core domain (TM1-TM10) of transporter structures were extracted from the Protein Data Base files, e.g. for vSGLT (2QX2 and 3DH4) and LeuT (2A65, 3TT1 and 3TT3). Energetically significant cation– π interactions were identified using the computer program (CaPTURE) available at http://capture.caltech.edu (19). CH-π interactions and H-bonds were identified by visual inspection of the crystal structures in PyMol (DeLano Scientific LLC).

RESULTS

Functional Analysis of the WYW Motif

Steady-state Kinetics

Mutants were expressed in oocytes and kinetics of Na⁺ and sugar transport was obtained using Na⁺-dependent sugar activated SGLT currents. The sugar stimulated current is an inward Na⁺ current that is directly proportional to sugar transport (13, 14).

Sugar

An example of the electrophysiological assay in an oocyte expressing Y290F is shown in Figure 3. In NaCl buffer addition of 100 mM αMDG rapidly produced an inward current due to the coupled inward transport of Na⁺ and sugar. After washout of the substrates with Na⁺ -free buffer (choline Cl), the current was restored to the starting level in Na⁺ buffer. The sugar-induced current (I_αMDG) was obtained by subtracting the baseline current in Na⁺ buffer from the current in the presence of αMDG. All Y290 and W291 mutants except Y290S exhibited sugar-induced currents. No sugar-induced currents were observed for Y289C even though there was expression of the mutant in the plasma membrane (Table 1). This indicates that the sugar K_0.5 for Y289C is far in excess of 100 mM. On the other hand, 100 mM αMDG generated a current of 40 ± 14 (n = 4) and < 10 nA for W289Y and W289F, but in these cases protein expression in the membrane was below the resolution of our method.

Figure 3. Na⁺ current in a single oocyte injected with Y290F cRNA.

The membrane potential of the oocyte was held at −50 mV. The horizontal dashed line indicates the baseline current in Na⁺ medium in the absence of substrate. Addition of 100 mM αMDG to the extracellular Na⁺ solution induced an inward current, which was inhibited by 100 μM Pz. The oocyte was washed in Na⁺-free buffer (filled box) before returning to Na⁺ buffer (blank box).

Apparent sugar affinities (K_0.5^αMDG) and maximal sugar-induced currents (I_max) were estimated in 100 mM NaCl buffer at −150 mV by fitting I_αMDG versus [αMDG]_o to Eqn 1. The K_0.5^αMDG for wild-type SGLT1 was 0.4 mM, while substitution of the side chains at Y290 and W291 with cysteine (Y290C and W291C) resulted in a dramatic increase in K_0.5^αMDG to more than 100 mM (Table 1). Figure 4 shows single experiments with each mutant. For wild type and W291F and Y290F mutants currents saturated or were close to saturation at 100 mM sugar (K_0.5 of 0.4, 3.7 and 23 mM). At 100 mM sugar, I_αMDG for Y290C and W291C was less than 50% of saturation, indicating that αMDG K_0.5 for these mutants was greater than 100 mM. The data is summarized in Table 1. A phenylalanine side chain at 290 partially restored sugar transport (K_0.5 35 vs. >100 mM) showing the importance of both the aromatic side chain and the tyrosine –OH group in Na⁺/glucose cotransport by hSGLT1. At position 291 the tryptophan and phenylalanine side chains are almost equivalent in supporting sugar transport (K_0.5^αMDG 4 vs. 0.6 mM).

Figure 4. Steady-state kinetics of αMDG transport by hSGLT1 and the mutants in 100 mM NaCl at −150 mV.

The sugar-induced currents are normalized to the predicted I_max of each oocyte, and the normalized αMDG currents (I_αMDG/I_max) were plotted as a function of [αMDG]_o and fitted to equation (Eqn 1). Similar results were obtained in at least three different batches of oocytes. The kinetic constants for each mutant are summarized in Table 1 from at 3 batches of oocytes.

Turnover

The turnover rate, an estimate of the number of transport cycles each protein undergoes per second, was estimated from the ratio I_max^−150mV/Q_max (15) (16). For the phenylalanine mutants, the turnover number was similar to WT (~70 s⁻¹) (Table 1). However, only minimal rates can be estimated for W291C and Y290C, as the K_0.5^αMDG for these mutants is > 100 mM at −150mV. For Y290C the turnover number was >> 100 s⁻¹ indicating either an increase in the actual turnover or uncoupling of Na⁺ and glucose transport.

Sodium

Apparent Na⁺ affinity (K_0.5^Na) was measured in saturating [sugar]_o. Sugar-induced currents (I_αMDG) at −50mV for wild-type and mutants were plotted as a function of external Na⁺ concentration ([Na]_o) and fitted to the equation Eqn 2 (Figure 5). All 290 and 291 mutants exhibited sigmoidal Na⁺ activation kinetics with Hill coefficients between 1.5 and 2.2 (Figure 5 and Table 1). This shows that two Na⁺ ions are required for activation of Na⁺/glucose cotransport. Table 1 shows the K_0.5^αMDG for most mutants at −150 mV was 5 to 10 times higher than wild-type SGLT1 (15, 26, and 32 mM for Y290F, Y290C and W291C vs.3 mM for WT). The exception was W291F where K_0.5^Na was identical to WT (3 mM).

Figure 5. Na⁺ activation of hSGLT1 αMDG currents at −50 mV.

The sugar-induced current was measured in single cells at different Na⁺ concentrations at saturating concentration of sugar for each mutant. The sugar-induced current was plotted as a function of external Na⁺ concentration and the data were fitted to the Hill equation (Eqn 2) where n is the Hill coefficient, and K_0.5^Na is apparent Na⁺ affinity. For comparison, the sugar-induced current of each oocyte was normalized with I_max. Similar results were obtained in at least three different batches of oocytes (Table 1).

Transport stoichoimetry

The coupling between Na⁺ and sugar transport for hSGLT1 is 2:1 (14, 18, 20) but mutations of outer gate residues increased coupling to as high as 6:1, i.e. uncoupled Na⁺/glucose transport(14). Here, we measured the coupling ratio of inner gate mutants W290F and W291F: it was impractical to measure the stoichiometry W290C and W291C due to K_0.5^αMDG values >100 mM). Figure 6 shows the plots of Na⁺ (charge) vs. αMDG uptakes for individual oocytes expressing WT hSGLT1, W291F and Y290F mutants. For WT the slope was 2.1 ± 0.1, whereas for both mutants the slopes were close to 1:1. We confirmed that the sugar induced inward charge uptake was Na⁺ uptakes by measuring ²²Na uptakes: the ratios of charge: ²²Na were 1.1 ± 0.1 (3), 0.7 ± 0.1 (3) and 1.2 ± 0.3 (3) for WT, W290F and W291F.

Figure 6. Determination of Na⁺ : glucose coupling ratio of SGLT1 and mutants.

αMDG uptake and inward charge were measured simultaneously over 5-10 min in each oocyte expressing wild type SGLT1 or Y290F or W291F. Inward charge and sugar uptake was calculated as described in Materials and Methods. The slope is the Na⁺:glucose coupling ratio. For each mutant, the experiments were repeated at least 3 times with different durations for sugar uptake. Note: for visual clarity only one line is shown for the mutants, but the slopes for each mutant are reported separately.

DISCUSSION

Our understanding of the mechanism of sodium substrate cotransport increased with the advent of crystal structures of the LeuT structural family in several different conformations (2, 4, 5, 21) (22-28). These structures confirm the alternate access transport model where substrates bind to the protein on one side of the membrane, go through an intermediate state where the substrate is occluded from the aqueous media on both sides of the membrane, before being released to the other side of the membrane. The substrate-binding sites, as well as the external and internal gates, have been identified. Furthermore, molecular dynamic simulations of vSGLT have provided insights into the structural basis of the underlying conformational changes, e.g. a plausible mechanism for Na⁺ and sugar release from vSGLT into the cytoplasm is that transition from outward to inward facing occluded conformations weakens Na⁺ binding to the Na2 site, resulting in disruption of the H-bond between inner gate residue Y263 and N64. Thus a prediction is that Y263 adopts a new rotamer position to permit sugar exit (5).

To understand the mechanism of symport we have been using the human sodium/glucose transporter, hSGLT1, as a model system, and have generated homology models of hSGLT1 based on vSGLT and other LeuT fold structures in different conformations. To infer the role of key residues, we related functional data from experiments on mutants. We have confirmed the importance of conserved residues forming the sugar binding site and the gates, and provided structural insights into Na⁺ - induced external glucose binding and translocation to the internal aqueous release channel (14). Here we focus on an aromatic motif, W289, Y290 and W291, conserved in the glucose and inositol transporters. Y290 corresponds to the inner gate residue in vSGLT (Y263) that is involved in both sugar binding and gating of sugar release to the cytoplasm, (5).). Replacing the aromatic side chains of Y290 and W291 with cysteine dramatically reduced K_0.5^αMDG and K_0.5^Na for Na⁺/glucose cotransport, and appeared to increase the turnover number (14). Here, we have expanded the studies to include mutants of each of the three aromatic resides in the motif. The results show that W289, Y290, and W291 are part of an interactive network that forms the sugar binding site, the Na1 binding site, and in coupling of Na⁺ and sugar cotransport.

Na⁺ and sugar-binding functions of the triad

The kinetics of hSGLT1 Y290C and Y291C (14) showed that the K_0.5’s for both sugar and Na⁺ were dramatically decreased relative to WT (Table 1). The low affinity for sugar (K_0.5^αMDG >100 mM) is compatible with involvement of Y290 in sugar binding, but the loss of Na⁺ affinity (K_0.5^Na ~10-fold higher compared to WT) was surprising, and suggested that residues 290 and 291 may be directly or indirectly involved in Na⁺ binding at the Na1 site.

Replacement of Y290 with phenylalanine (Y290F) decreased K_0.5^αMDG compared to Y290C, from ~170 to 35 mM, presumably due to restoring the CH -stacking interactions. Nevertheless, K_0.5^αMDG for Y290F is 50-fold higher than that of WT (0.6 mM). The tyrosine –OH group is within H-bonding distance of N78 (TM1) (Figure 7) and K_0.5^αMDG for N78A/C/S mutant was 10-fold higher than in WT [(14) and unpublished data]. Thus, H-bonding between N78 and Y290 appears crucial for the interaction of sugar and Na⁺ with the transporter. We think that the H-bond between N78 and Y290 has two effects on K_0.5^αMDG; positioning the aromatic ring for optimal interaction with sugar, and that the H-bond through the phenolic hydroxyl increases the electrostatic potential of the aromatic system, thereby increasing the strength of interaction with the sugar (29).

Figure 7. Homology model of the inward facing occluded structure of hSGLT1 based on the crystal structure of vSGLT.

The reduced model only shows TM helices 1, 3, and 6 (LeuT numbering (3), with the atoms of the side chains of residues N78, S149, Y153, W289, Y290, and W291 shown as spheres (C black, O red, and N blue). Glucose is shown as a stick model and Na⁺ in the Na2 site as a yellow sphere.

In contrast, W291F largely recovered the poor sugar and Na⁺ affinities of the cysteine mutant (K_0.5^αMDG decreased from ~100 mM for W291C to 4.4 mM for W291F, K_0.5^Na from 70 to 3 mM Table 1). It is likely that the aromatic side chain at position 291, has a π - π interaction with Y290, stabilizing its position and W291 forms an H-bond with the # 4 –OH group of the sugar, or there is a direct CH – π bonding between the sugar and W291 not evident in the x-ray structure. The loss of the phenolic hydroxyl also affected Na⁺ affinity as the K_0.5^Na for Y290F was 6-fold higher than for WT, 20 vs. 3 mM (Table 1). Neither Y290 nor W291 is within bonding distances to Na⁺ at the Na2 site (Figure 7).

In Y290C and W291C K_0.5^Na increased from 3 to >30 mM and for the W289 mutants there was no apparent Na⁺ binding. Since the putative Na2 site in vSGLT and hSGLT1 is > 10Å away on the other face of TM1 (Figure 7), the effects are likely to be on as yet unidentified Na1 binding site. Possibilities include: 1) that Na⁺ in the Na1 site interacts with W289, Y290 and or W291 through cation-π interactions (8); and/or 2) Na⁺ directly interacts with bound glucose (30, 31). In LeuT the Na1 binding site is close to the leucine binding site and the coordinating groups include the leucine –COOH group and backbone carbonyls in the unwound region of TM1 (1). These results suggest that the Na1 site is close to the sugar-binding site in hSGLT1.

Na⁺ and sugar coupling

The gate side chains are not simply barriers, but play a role in the coupling of Na⁺/sugar transport (14). Mutation of each outer gate residue, L87C, F101C, and F453C, increased the Na⁺/sugar stoichiometry above 2:1. In contrast, we found that the coupling ratios of Na⁺ to sugar transport were reduced from 2:1 in WT to 1:1 in Y290F and W291F, i.e. only one sodium ion was transported with each sugar. However, Na⁺ activation for all Y290 and W291 mutants was sigmoid (Hill coefficients in the 1.5-2.2 range (Table 1)) indicating binding of two Na⁺ ions is required for transport. An explanation for this difference between Na⁺ activation and Na⁺/glucose stoichiometry is that only 1 of the 2 Na⁺ ions that bind to hSGLT1, probably the one at the Na2 site, is released into the cytoplasm with each sugar molecule, but Na⁺ at the Na1 site returns to the external compartment.

Structural Model of Glucose Binding to Triad

Figure 7 is a model of hSGLT1 focused on the aromatic triad W289/Y290/W291 and related structural components in the inward facing occluded conformation(14). Glucose is shown to overlie Y290. The fact that αMDG affinity was reduced by over 2-orders of magnitude in the Y290C mutant (Table 1) indicates the close interaction between the pyranose ring and the aromatic ring as in sugar binding proteins (32). It is now recognized that such stacking is mediated by π hydrogen bonds between the axial-CH groups of the sugar and the aromatic side chain (33). The –OH group of Y290 is shown H-bonded to N78 (TM1) because mutation of each residue, Y290F and N78C, lowered the affinity for glucose by over an order of magnitude (Table 1 and (14).

We further postulate that W289 (TM6) forms an H-bond with either S149 or Y153 (TM3): in both Vibrio SGLT structures the homologous Y262 H-bonds with W134 (2). Effect of mutations W289C or W289F also indicates that W289 is important for sugar and/or Na⁺ binding. W289C was well expressed in the oocyte plasma membrane (Table 1), but was insensitive to external Na⁺ and sugar. Even for W289Y there was poor sugar transport, K_0.5 ~100 mM. These results suggest an interaction between W289 and S149 or Y153 is critical for Na⁺ coupled sugar transport possibly by stabilizing the non-helical region of TM6. By analogy, in vSGLT the interaction between S365 (TM8) and E68 (TM1) has been proposed to stabilize the inward facing empty conformation (5).

The interaction between Y263 and W264 in vSGLT (2, 5) shows a typical edge to face geometry (T-stacking) (6, 7, 33). As shown in Table 1 replacement of the homologous W291 side chain in hSGLT1 with cysteine dramatically reduced the affinity for αMDG (K_0.5 >100 mM) illustrating the importance of this aromatic interaction. These results point to the importance of the interactions between the TM6 aromatic triad and residues on TM1 and TM3 on maintaining the functional integrity of the transporter though a combination of classical and CH-π hydrogen bonds.

Broad Role of Aromatic residues in LeuT Symporters

We have examined the role of the aromatic triad in hSGLT1. To assess the generality we conducted an analysis of the aromatics in symporters. Figures 1 shows the network of aromatic interactions in vSGLT and LeuT. Ten intra-TM bonds and a comparable number (thirteen) occur in LeuT. Given a binding energy of 1.5 – 3.3 kcal/mole per bond, these aromatic interactions are comparable to that of the network of classical H-bonds in other membrane proteins thought to be important in protein stability (34).

Cation – π bonds

The role of cation-π interactions has already been well recognized in ligand binding to ion-channels, receptors, and transporters (1, 8, 9, 24, 25). Here we identify significant cation-π interactions in symporters using the computer program CaPTURE (19): for example, 1) In the open inward conformation of vSGLT (2QX2) there are interactions between K285 (TM7) and W53 (TM1), and between K450 and W448 (TM10) (2 - 3.6 kcal/mol) (Figure 1A), but these bonds do not exist in the inward occluded conformation (3DH4); 2) in LeuT a cation-π bond K121 – F167 (2.2 – 2.4 kcal/mole) ties the tops of TM3 and TM4 together in both the outward occluded and inward open conformations (2A65 and 3TT3) but not the outward open conformation (3TT1) (Figure 1B); 3) In all 3 conformations of LeuT (outward open (3TT1), outward occluded (2A65), and inward open (3TT3) a bond occurs between R30 (TM1) and F253 (TM6) (2.6 – 4.1 kcal/mole); and 4) Aromatic residues at the end of TM domains may form cation– π bonds with basic residues on inter-helical loops, e.g. in Mhp1 (2JLN) Y324 (TM8) interacts with K232 on the loop between TM6 and TM7. These observations suggest that cation– π bonds are involved in helix –helix packing and transitioning between conformations.

CH – π bonds

Pairs of aromatic residues buried within the membrane core of vSGLT were identified where the phenyl rings are separated < 7.5 Å in T-stacking or off-centered parallel orientations (Figure 1A). Those in a T-stacking orientation include Y87 (TM2) and F424 (TM10), previously identified as outer gate residues; F102 (TM2) and F266 (TM6); F102 (TM2) and F447 ((TM10); and Y263 and W264 (TM6 (Figure 1A). Those pairs in a parallel orientation include; F266 (TM6) - F442 (TM10). In addition, a pair W134 (TM3) and Y262 (TM6) are within H-bonding distance. This indicated that buried aromatic side chains are involved in defining the tertiary structure of vSGLT.

Ligand – aromatic interactions

Aromatic residues are essential elements in binding ligands. Betaine binding to BetP involves cation – π interactions with 3 Trp residues in an aromatic box (Y189, W194 and W197 (2WIT, (25); (24)). Benzylhydantoin binding to Mhp1 has been modeled as π-π stacking of hydantoin with Y117 and the benzyl stacking with W220 (22). Finally, D-galactose in vSGLT stacks on the aromatic face of Y263 (2), presumably by CH – π bonding. In hSGLT1, the Y290C mutation reduces sugar transport affinity by over 2-orders of magnitude. The magnitude of the difference in sugar affinity between Y290 (or Y290F) and Y290C is compatible with the loss in CH - π binding energy, 2 - 5 kcal/mole (33): according to the Gibbs equation (K₁/K₂ = exp [1-Δ(ΔG)/RT], where K₁, K₂ are the binding constants of WT and mutant proteins), the loss of 5 kcal/mole binding energy would decrease the sugar binding constant by up to 300-fold.

In summary, our results highlight the importance of CH-π and cation-π hydrogen bonding in both ligand binding and the tertiary structure of sodium symporters in the LeuT structural family. Specifically, in hSGLT1 π –π interaction between i) outer gate residues F101 and F453 contribute towards holding sugar in the occluded conformation after binding, ii) sugar and Y290 determine the strength in sugar binding, iii) interactions involving the WYW triad are intimately involved in Na binding to Na1, glucose binding, and release of sugar into the cytoplasm. More generally, we suggest that π-π interactions are involved in determining the tertiary structure of LeuT transporters and transitions between conformational states.

ABBREVIATIONS

SSS

solute sodium symporters

LeuT

bacterial Na⁺-coupled leucine transporter

vSGLT

Na⁺/galactose symporter from Vibrio parahaemolyticus

hSGLT1

human sodium/glucose cotransporter

αMDG

α-methyl-D-glucopyranoside

WT

wild type

5TMIR

5 transmembrane inverted repeat

TM

transmembrane helix

K_0.5^Na

half-saturation concentration for Na⁻

K_0.5^αMDG

half-saturation concentration for αMDG

I_max

maximal sugar-induced current

I_αMDG

sugar-induced current