Structural and functional basis of amino acid specificity in the invertebrate cotransporter KAAT1

Abstract

The substrate specificity of KAAT1, a Na⁺- and K⁺-dependent neutral amino acid cotransporter cloned from the larva of the invertebrate Manduca sexta and belonging to the SLC6A gene family has been investigated using electrophysiological and radiotracer methods. The specificity of KAAT1 was compared to that of CAATCH1, a strictly related transporter with different amino acid selectivity. Competition experiments between different substrates indicate that both transporters bind leucine more strongly than threonine and proline, the difference between KAAT1 and CAATCH1 residing in the incapacity of the latter to complete the transport cycle in presence of leucine. The behaviour of CAATCH1 is mimicked by the S308T mutant form of KAAT1, constructed on the basis of the atomic structure of a leucine-transporting bacterial member of the family, which indicates the participation of this residue in the leucine-binding site. The reverse mutation T308S in CAATCH1 conferred to this transporter the ability to transport leucine in presence of K⁺. These results may be interpreted by a kinetic scheme in which, in presence of Na⁺, the leucine-bound state of the transporter is relatively stable, while in presence of K⁺ and at negative potentials the progression of the leucine-bound form along the cycle is favoured. In this context serine 308 appears to be important in allowing the change to the inward-facing conformation of the transporter following substrate binding, rather than in determining the binding specificity.

The two neutral amino acid transporters KAAT1 and CAATCH1, cloned from the midgut of the larva of Manduca sexta (Castagna et al. 1998; Feldman et al. 2000), represent an interesting tool for investigating structure–function relationships concerning ion and substrate selectivity in the SLC6A transporter family. In fact both sequence similarities and functional analogies indicate that they may be assigned to this family, to which many important vertebrate neurotransmitter transporters belong, as part of a nutrient amino acid transporters (NAT) subset (Boudko et al. 2005). KAAT1 and CAATCH1, composed of 634 and 633 amino acids, respectively, are very similar to each other, differing in only 63 residues, and are 35–45% identical to the neurotransmitter transporters of the Na⁺/Cl⁻-dependent transporter superfamily (Castagna et al. 1998; Feldman et al. 2000). Accordingly, their hydropathicity profiles and the atomic structure of LeuT_Aa, the Aquifex aeolicus leucine transporter (Yamashita et al. 2005), suggest 12 transmembrane domains with cytosolic carboxy- and amino-termini.

At variance with the strict Na⁺ dependence of known vertebrate transporters, KAAT1 and CAATCH1 are able to exploit in addition a K⁺ electrochemical gradient to power up the uptake of neutral amino acids from the intestinal lumen (Hanozet et al. 1992; Castagna et al. 1997). This peculiar characteristic is believed to be related to the high-potassium, low-sodium condition of the larva intestine (Harvey et al. 1975; Hennigan et al. 1993) caused by its dietary habits. In both ionic conditions the amino acid uptake is electrogenic, giving rise to a transmembrane transport-associated current. The two other types of current observed in the majority of ion-coupled cotransporters, that is the presteady-state (transient) currents, and the uncoupled (leak current), are also present in KAAT1 and CAATCH1 (Bossi et al. 1999; Feldman et al. 2000; Quick & Stevens, 2001). Moreover, CAATCH1 exhibits a peculiar behaviour of an amino acid-gated cation channel (Quick & Stevens, 2001).

Again in contrast with the high substrate specificity of the vertebrate neurotransmitter transporters, KAAT1 and CAATCH1 are capable of transporting a rather wide spectrum of neutral amino acids (Castagna et al. 1998; Feldman et al. 2000). Quite interestingly, the potency order, in terms of amplitude of transport-associated currents, is different in the two transporters, and furthermore it depends on the driver cation (Feldman et al. 2000; Soragna et al. 2004).

These clear differences in ionic requirements and amino acid selectivity, together with the limited number of divergent residues, are obviously quite useful when trying to identify the structural determinants of the ionic dependence and amino acid interactions. Indeed these have been already exploited in our laboratory to build chimeras that led to restriction of the regions of interaction with the organic substrates to the central part (transmembrane segments 4–8) of the protein sequence (Soragna et al. 2004).

A great impetus to the understanding of the structure–function relation in the Na⁺/Cl⁻-dependent transporter family has been given by the resolution of the atomic structure of a bacterial member of the family, the Aquifex aeolicus leucine transporter LeuT (Yamashita et al. 2005). Among the many suggestions deriving from this structure, the leucine-binding site of this transporter was pinpointed with great accuracy, showing that the amino acid to be transported forms specific interactions with several main-chain and side-chain atoms that may be traced in the sequence of other members of the family, opening the possibility to investigate structure–specificity relationships (Beuming et al. 2006).

The precise definition of the leucine-binding site in LeuT, together with the different ability to transport leucine of KAAT1 and CAATCH1, prompted us to investigate the structural determinants involved in substrate interaction in these transporters.

Methods

Site-directed mutagenesis

The mutant cDNAs KAAT1_S308T, KAAT1_A313P and CAATCH1_T308S were constructed by site-directed mutagenesis (Quickchange Site-Directed Mutagenesis Kit, Stratagene Inc.), as explained in detail elsewhere (Forlani et al. 2001) in the presence of 250 nm overlapping primers containing in their sequence the mutated codons:

KAAT1_S308T: ACTCAAGTGTTCTTCTCTCTGACAGTGTGCACTGGAGCT

CAATCH1_T308S: CAAGTGTTCTTCTCTCTGTCAGTGTGCACCGGACCG

KAAT1_A313P: CTGTCAGTGTGCACTGGACCTATTATCATGTTCTCC

cRNA preparation and Xenopus laevis oocyte expression

The experimental procedure has been described in detail elsewhere (Bossi et al. 2007). Briefly, the cDNAs encoding the original and mutant cotransporters constructs were linearized with Not I. cRNAs were in vitro synthesized in the presence of Cap Analogue and 200 units of T7 RNA polymerase. All enzymes were supplied by Promega Italia, Milan, Italy.

Xenopus laevis frogs were anaesthetized in MS222 (tricaine methansulphonate) 0.10% (w/v) solution in tap water; portions of ovary were removed through an incision on the abdomen. The frogs were humanely killed after the collection.

The oocytes were treated with collagenase (Sigma Type IA) 1 mg ml⁻¹ in ND96 Ca²⁺-free, for at least 1 h at 18°C. Healthy-looking V- and VI-stage oocytes were selected and injected with 12.5 ng of the appropriate cRNA in 50 nl of water, using a manual microinjection system (Drummond). Oocytes were injected with 12.5 μg of each cRNAs. The oocytes were incubated at 18°C for 3–4 days in NDE solution (ND 96 solution: 96 mm NaCl, 2 mm KCl, 1.8 mm CaCl₂, 1 mm MgCl₂, 5 mm Hepes supplemented with 50 μg ml⁻¹ gentamicin and 2.5 mm Na⁺ pyruvate at pH 7.6), before electrophysiological studies.

Electrophysiology and data analysis

A two-microelectrode voltage-clamp was used to perform electrophysiological experiments (Geneclamp 500B, Axon Instruments, Union City, CA, USA). The holding potential was kept at −60 mV, and the typical protocol consisted of 200 ms voltage pulses spanning the range from −160 to +20 mV in 20 mV steps. Four pulses were averaged at each potential; signals were filtered at 1 kHz and sampled at 2 KHz. Experimental protocols, data acquisition and analysis were done using the pCLAMP 8 software (Axon Instruments).

Uptake experiments

For uptake measurements, groups of 10–14 oocytes were incubated for 60 min in 120 μl of the uptake solution (mm: 100 NaCl, 2 KCl, 1 CaCl₂, 1 MgCl₂, 20 Hepes/Tris, pH 8) with 1100 kBq ml⁻¹ of [³H]leucine, [³H]proline or [³H]threonine (GE Healthcare Europe, GmbH), rinsed in ice-cold wash solution (mm: 100 choline chloride, 2 KCl, 1 CaCl₂, 1 MgCl₂, 10 Hepes/Tris, pH 8), and dissolved in 250 μl of 10% SDS for liquid scintillation counting. In competition experiments, 0.5 mm cold leucine was added to the uptake solution. The uptake rates were always corrected for the background uptake measured in non-injected oocytes.

Solutions

In electrophysiological experiments the external control solution had the following composition (mm): NaCl, 98, MgCl₂, 1, CaCl₂, 1.8, Hepes free acid, 5; in the other solutions NaCl was replaced by KCl or TMACl. The pH was adjusted to 7·6 by adding the corresponding hydroxide for each alkali ion and TMAOH for TMA⁺ solution. Amino acids (leucine, threonine, proline) at the desired concentrations (up to 3 mm in the dose–response experiments) were added to the same solutions to induce transport-associated currents. The oocyte was perfused continuously in a rapid solution exchange chamber (Warner Instruments model RC-1Z, http://www.warneronline.com/), and the washout time was estimated to be about 1 s.

Mathematical modelling

From the kinetic scheme of Fig. 9 the following system of differential equations may be written:

in which T₁ to T₅ represent the probability of occupancy of the different states and k_ij are the unidirectional rate constants whose expressions and values are listed in Table 1.

Figure 9.

Five-state kinetic scheme for the interpretation of KAAT1 amino acid competition T₁ represents the empty, outward-facing state of the transporters, to which cations (and possibly Cl⁻) may bind leading to state T₂ which enables interaction with amino acids. The scheme is intended to simulate the competition of leucine (left branch of the scheme T₃ → T₅) with other amino acids (aa) (right branch T₄ → T₅) and with central leak pathway T₂ → T₅. T₅ represents the inward-facing conformation that releases the substrates to the cytosol before rearranging to the initial state to complete the cycle.

Table 1.

Parameters used in the simulations

	k012 (m−1 s−1)	k021 (s−1)	k25 (s−1)	k52 (m−1 s−1)	k15 (s−1)	k51 (s−1)	k23 (m−1 s−1)	k32 (s−1)	k035 (s−1)	k53 (m−2 s−1)	k24 (m−1 s−1)	k42 (s−1)	k045 (s−1)	k54 (m−2 s−1)
KAAT1 Na+	126	359	2	70.2	20	2000	1 × 106	0.2	4	7.02 × 105	3 × 105	50	500	1.1 × 105
KAAT1 K+	126	359	2	70.2	20	2000	1 × 106	0.2	3000	3.51 × 108	300	300	200	7020

Rates denoted with the ‘0’ superscript represent the zero voltage value. The constants k₅₂, k₅₃ and k₅₄ were calculated to ensure micro reversibility (Läuger, 1991). The values of α, β and δ were, respectively, set at 0.5, 0.3, 0.2 for Na⁺ and 0.3, 0.6, 0.1 for K⁺. The external and internal concentrations of the organic substrates were set at 500 μm and 10 μm, respectively. For Na⁺ they were 100 mm and 7 mm, while for K⁺ they were both 100 mm.

The system was solved numerically to obtain steady-state values and time-dependent solutions, using MATLAB under variable voltage and substrate concentration conditions.

Results

Amino acid selectivity and competition in KAAT1 and CAATCH1

Among leucine, threonine and proline applied at the same 500 μm concentration, leucine is the amino acid eliciting the largest currents in KAAT1 when K⁺ is the driver ion, while in presence of Na⁺, threonine and proline are transported more efficiently than leucine (Soragna et al. 2004; Feldman et al. 2000).

Conversely, in the strictly related transporter CAATCH1, leucine has the peculiar effect of blocking the constitutive leak (uncoupled) current in presence of either Na⁺ or K⁺ (Feldman et al. 2000; Soragna et al. 2004). This blocking effect is also exerted in CAATCH1 by other amino acids such as alanine, histidine and methionine, and the latter was reported to be able to inhibit the transport current induced by proline as well (Feldman et al. 2000; Soragna et al. 2004). These observations encouraged us to devise a series of competition experiments to investigate more deeply the interactions of the different amino acids with KAAT1 and CAATCH1. Figure 1A shows the result of applying, in presence of Na⁺, 500 μm leucine alone or in addition to 500 μm proline or threonine, to a CAATCH1-expressing oocyte kept at the holding potential of −60 mV. Besides the known leak-blocking effect visible in Fig. 1Aa, Fig. 1Ab clearly shows that leucine completely blocks the proline-induced currents, as well as those generated by threonine. It is also interesting to note that the recovery of the currents after the leucine exposure is very slow, compared to the onset of the currents, suggesting a slow dissociation rate of leucine from the transporter.

Figure 1.

In CAATCH1 in Na⁺ solution leucine act as a blocker of both leak and transport-associated currents A, in an oocyte kept at −60 mV holding potential, a decrease in inward current is produced by application of 500 μm leucine (a). In the same oocyte the large inward current elicited by application of 500 μm proline was promptly blocked by the simultaneous addition of 500 μm leucine (b). Leucine also blocked the threonine transport current in another oocyte (c). Note in all cases the very slow recovery after leucine exposure. The dotted lines mark the holding current in the absence of amino acids. B, uptake of leucine, threonine and proline in Na⁺ solution (the radioactive amino acid is denoted by an asterisk) alone or in combination, showing the block exerted by leucine on the uptake of the other two amino acids (both significantly different at P < 0.001). Data are expressed as a percentage of the proline uptake and are means ± s.e.m. from 10 to 14 oocytes in at least three independent experiments.

The electrophysiological results are confirmed by the uptake data shown in Fig. 1B, where it can be seen that in Na⁺ solution the uptake of radioactive threonine or proline is completely inhibited by the concurrent presence of leucine. In CAATCH1 in Na⁺ solution, leucine appears then not only to block the leak current, but to inhibit the transport-associated current and the uptake of the two other amino acids threonine and proline.

Figure 2 illustrates parallel experiments performed on KAAT1, showing similar results. Addition of leucine in Na⁺ solution has an inhibitory action on the larger threonine and proline currents, reducing the amplitude at the level corresponding to the application of 500 μm leucine alone. On the other hand, the competition between proline and threonine, appears to be dominated by threonine, since addition of the latter to proline brings the current to the level of threonine alone, while addition of proline to threonine, does not result in any current increase (Fig. 2B).

Figure 2.

Competition between substrates in KAAT1 A, in Na⁺ solution the transport currents elicited by 500 μm threonine or proline are reduced by the addition of 500 μm leucine; the current level during leucine inhibition is similar to the level in response to application of leucine alone (dotted line). B, in another oocyte addition of threonine 500 μm to 500 μm proline in the presence of Na⁺ increases the amplitude of the transport current to a level comparable to that of threonine alone; conversely, addition of proline to threonine does not increase the current level. C, in K⁺ solution the effect of 500 μm leucine adds up to that of threonine, reaching a level comparable to that of leucine alone (dotted line); 500 μm proline has only a transient effect at this potential (−80 mV) that does not prevent the action of leucine. Similarly, adding proline does not alter the threonine response. Holding potential −60 mV in A and B, −80 mV in C. D, uptake of leucine, threonine or proline (the radioactive amino acid is denoted by an asterisk) alone or in combination, in Na⁺ solution. The reductions in the uptake of threonine and proline uptake when in combination with leucine are both significantly different (P < 0.05) from the respective values alone. Data are expressed as a percentage of the leucine alone uptake and are means ± s.e.m. from 10 to 14 oocytes in at least three independent experiments.

In the presence of K⁺, the interaction between substrates gives different results in KAAT1, although leucine appears to retain a dominating role (Fig. 2C). Combinations of leucine and threonine or leucine and proline (all at 500 μm) elicit currents of amplitudes similar to leucine alone, even though now this is larger. Differently from CAATCH1, in KAAT1 the velocity of the recovery phases after leucine exposure is comparable to that of the onsets of the currents, with time constants of a few seconds, in both ionic conditions.

The electrophysiological results are again confirmed by radioactive tracer measurements, as illustrated in Fig. 2D: the uptake of threonine or proline alone is comparable to that of leucine alone in Na⁺ solution, while the uptake of both amino acids is strongly reduced when non-radioactive leucine is present at the same time.

While a 500 μm concentration is saturating for leucine in sodium solution (Mari et al. 2004; Vincenti et al. 2000), the apparent affinity for the other two amino acids, and for leucine in potassium, needs to be determined. The full dose–response curves for the three substrates shown in Fig. 3 indicate that indeed higher concentrations of proline and threonine are required to reach the maximal transport capacity. The K_0.5 values for the three amino acids at V_m = −60 mV are 12 ± 2, 82 ± 16 and 464 ± 39 μm, for leucine, threonine and proline, respectively, in Na⁺ solution, while in K⁺ solution at −80 mV K_0.5,leu is 289 ± 25 μm and K_0.5,thr is in the lower millimolar range. At the same time the maximal proline- and threonine-induced currents largely exceed that elicited by leucine in Na⁺ (Fig. 3A), while in potassium, saturating leucine and threonine give rise to comparable currents (Fig. 3B), in agreement with previous results (Feldman et al. 2000; Soragna et al. 2004).

Figure 3.

KAAT1 apparent affinity and competition at saturating concentrations A, in Na⁺ solution at −60 mV, leucine has significantly higher apparent affinity, but much lower maximal, current compared to proline and threonine. B, in K⁺ solution at −80 mV, leucine retained a greater affinity than threonine, although both K_0.5 values were much higher than in Na⁺, and the maximal currents were comparable; parameters were not determined for proline because of the very small current at this potential in K⁺. C, competition between leucine and the other two substrates at saturating concentrations confirms the leucine dominance in Na⁺ solutions. D, in K⁺, saturating 3 mm leucine causes additional currents that add up to the threonine and proline ones approaching the leucine-alone level.

To assess whether the leucine dominance shown in Fig. 2 could be overcome by higher concentrations of threonine and proline, we repeated the competition experiments using higher (in most cases saturating) values for each amino acids (500 μm for leucine and 3 mm for proline and threonine in Na⁺, and 3 mm for all three in K⁺). The results are illustrated in Fig. 3C and D, showing that in Na⁺, leucine still reduces the responses induced by the other two substrates, although now, consistently with the lower affinity for proline with respect to threonine, the reduction is smaller with the latter amino acid. In the presence of potassium, the higher current level elicited by 3 mm leucine is maintained when leucine is applied in addition to 3 mm threonine, while when the combination 3 mm proline plus 3 mm leucine is applied the leucine-alone level is slightly reduced (Fig. 3D).

Identification of putative selectivity determinants

In addition to the functional studies, and taking advantage of the high-resolution atomic structure of LeuT recently published (Yamashita et al. 2005), it is now possible to tackle the question of substrate selectivity also on structural grounds. The initial indication that in KAAT1 and CAATCH1, substrates interact with the central part of the sequence (Soragna et al. 2004) finds support in the results from LeuT, where the leucine-binding site appears to be formed with the contribution of transmembrane segments 1, 3, 6 and 8 (Yamashita et al. 2005).

We have therefore focused our attention on the differences in residues between KAAT1 and CAATCH1 in TM segments 1, 3, 6 and 8, considering also the corresponding amino acids in other leucine-transporting and non-transporting members of the family. The sequence alignment of LeuT with KAAT1 and CAATCH1 and some selected neurotransmitter transporters of the SCL6A family, not capable of transporting leucine, is shown in Fig. 4, where the alignments recently proposed (Beuming et al. 2006) have been used, and another non-neuronal leucine transporter, SCL6A19 (B⁰AT1), is also included (Böhmer et al. 2005).

Figure 4.

Sequence alignment of a number of SCL6 family transporters in the regions involved in substrate binding according to LeuT_Aa Strictly conserved residues are in bold. Residues differing in KAAT1 and CAATCH1 are in bold and underlined. ▪ and □ indicate residues involved in coordinating Na⁺ ions (1 and 2, respectively); • marks leucine-binding residues. The leucine-transporting members are denoted by *. Arrows indicate residues 6.56 and 6.61 generic numbering. (256 and 261 in LeuT, 308 and 313 in KAAT1 and CAATCH1).

As shown in Fig. 4, the KAAT1 and CAATCH1 sequences are identical in TM1, while in TM3 they differ in two residues at positions 138 (3.41, according to general numbering (Beuming et al. 2006), and 156 (3.59), the corresponding residues in LeuT, tryptophan 99 and glycine 117 not being involved in leucine binding (Yamashita et al. 2005). Similarly, in TM8 the three differing residues between KAAT1 and CAATCH1 in positions 408, 411 and 412 (8.68, 8.71, 8.72) do not correspond to leucine-binding residues in LeuT. In TM6 KAAT1 and CAATCH1 have four differing residues, and one of them, serine 308 (6.56) in KAAT1 (but threonine 308 in CAATCH1) corresponds to serine 256 of LeuT, a residue taking part in leucine binding through its side-chain hydroxyl. Another differing residue, A313 (6.61) in KAAT1 (P313 in CAATCH1), corresponds to A261 in LeuT and it is important in stabilizing the structure (Yamashita et al. 2005). More generally, comparing the sequences of the three leucine-transporting proteins shown in Fig. 4 (LeuT, B⁰AT1 and KAAT1), to those of the other listed proteins that do not transport leucine (GAT1, GlyT1b and CAATCH1), we see that serine 256 in LeuT, is confirmed only in the equivalent position S280 in B⁰AT1, but it is replaced by different residues in the other neurotransmitter transporters. A more complete analysis recently published (Beuming et al. 2006) shows that most prokaryotic (80%) and eukaryotic neutral amino acid transporters have a serine in the position equivalent to LeuT 256, while more strictly selective neurotransmitter transporters have a glycine or an alanine.

This analysis suggests that the capacity to transport leucine may be related to the presence of a serine residue in position 6.56 in the sixth transmembrane segment. We have therefore proceeded to make point mutants replacing S308 with threonine in KAAT1, and T308 with serine in CAATCH1, trying to alter the leucine selectivity of the two transporters in opposite ways. Furthermore, mutation A313P in KAAT1 was used as control.

Leucine selectivity in KAAT1 mutants

The behaviour of KAAT1_S308T in the presence of Na⁺ is illustrated in Fig. 5. The mutant is still able to transport proline and threonine, while the leucine transport is strongly impaired. This is shown in Fig. 5A and B, where it can be seen that, while in the wild-type the leucine-induced current is on average 25% of that induced by proline, in the mutant it is never significantly different from zero. This result is confirmed by the uptake measurements (Fig. 5E), showing a significant depression of the leucine uptake in KAAT1_S308T. Mutation of alanine 313 in the corresponding CAATCH1 residue proline, appears instead to increase the leucine-induced currents to about 35% of the one elicited by threonine (Fig. 5C).

Figure 5.

Amino acid specificity of mutated KAAT1 in high-Na⁺ solution Leucine transport currents are abolished in the S308T mutant of KAAT1 (B), compared to the wild-type (A), while they are preserved in the A313P mutant (C). Current data are averages ± s.e.m. from 3 to 8 oocytes from at least two batches, normalized to the values at −160 mV in threonine. All three amino acids were applied at 500 μm. D, apparent affinities of wild-type (open symbols) and S308T (filled symbols) for leucine (squares), proline (circles) and threonine (triangles). E, a statistically significant (P < 0.05) reduction in leucine transport by KAAT1_S308T, compared to the wild type, is also seen in uptake experiments (normalized to threonine). The differences between mutated forms and wild types are not significant for the other two amino acids. Data are expressed as a percentage of the threonine uptake and are means ± s.e.m. from 10 to 14 oocytes in at least three independent experiments. All amino acids were added at 500 μm.

The reduction of the leucine-induced current relative to those elicited by proline and threonine is not due to a decreased leucine affinity compared to that of the other two amino acids, but actually to an opposite effect. Figure 5D plots the substrate concentrations giving rise to the half-maximal current at each potential (K_0.5), for the three amino acids in the wild-type and S308T transporters. For both threonine and proline the K_0.5 values are higher (and thus the apparent affinity lower) in the mutant, while for leucine, that has a wild-type K_0.5 lower than the other two substrates, the mutant K_0.5 could not be determined in this way due to the lack of current (but see below the paragraph on the presteady-state currents, showing stronger binding of leucine to the mutant). Figure 5D also shows all K_0.5 values tend to increase at more depolarized potentials, although this voltage dependence appears weaker for leucine compared to the other two substrates

We also produced the reverse mutant CAATCH1_T308S in an attempt to confer to this transporter the capacity to generate transport currents in the presence of leucine. However in Na⁺ this was not the case, as the behaviour of the mutant was indistinguishable from that of the wild-type form (not shown).

The currents induced by the different amino acids in the presence of potassium in KAAT1 wild-type and S308T mutant are shown in Fig. 6A and B. It can be observed that, although some current may still be generated by the presence of leucine, its amplitude is strongly reduced relative to those induced by threonine.

Figure 6.

Amino acid specificity of mutated KAAT1 in terms of the amplitude of the transport-associated current in the presence of 500 μM of the indicated substrates and in high-K⁺ solution Mutation S308T in KAAT1 strongly reduces the leucine-induced current relative to the threonine current (A and B). C, apparent affinities of wild-type (open symbols) and S308T mutant (filled symbols) for leucine (squares) and threonine (triangles). D and E, the reverse mutation T308S in CAATCH1 increases the magnitude of the proline current compared to threonine, and significantly reverts the blocking effect of leucine in an net inward current. Data are averages ± s.e.m. from 6 to 12 oocytes from at least two batches, normalized to the values at −160 mV in threonine.

Again, the mutation A313P does not appear to alter the selectivity sequence of the wild-type KAAT1 (not shown).

The K_0.5 for leucine and threonine in K⁺, are plotted in Fig. 6C for both wild-type and S308T. Similarly to the results in Na⁺, the apparent affinity of the mutant for threonine is lower, while that for leucine is higher at negative potentials. In the presence of K⁺, the reverse T308S mutation in CAATCH1 has some effects on the amplitudes of the proline-induced current relative to threonine, as shown in Fig. 6D and E, and, particularly interesting, on the direction of the leucine-induced current, that becomes inward at negative potentials.

This series of experiments confirms then that the serine residue in position 308 of KAAT1 is involved in leucine interaction, although its insertion in CAATCH1 is not sufficient to restore leucine transport in Na⁺ solution.

Presteady-state currents

Voltage-dependent presteady-state currents, signalling rearrangements of charges in the membrane field, have been observed in both KAAT1- and CAATCH1-expressing oocytes (Bossi et al. 1999; Quick & Stevens, 2001). This process may be useful in understanding the basis of substrate specificity, as the amplitude of the transport currents often correlates with the maximal displaceable charge. Presteady-state currents in KAAT1 are best studied in the presence of Na⁺, since in the presence of K⁺ they are strongly shifted toward negative potentials (Bossi et al. 1999). Figure 7A shows current traces in a representative KAAT1-expressing oocyte in control Na⁺ solution, and after addition of 500 μm leucine. It can be seen that the slow relaxations present in the control traces (arrows) are abolished by the addition of leucine.

Figure 7.

Presteady-state currents in Na⁺ A, current traces elicited by voltage pulses (range −160 mV to 0 mV) from V_h = −60 mV in a wild-type KAAT1-expressing oocyte in the absence (Aa) or presence (Ab) of 500 μm leucine; the arrows in a point to the slow relaxations in Na⁺ control solution that are abolished by saturating leucine concentrations (b); the isolated presteady-state currents, obtained by subtracting the traces in b from those in a, and after further corrections to bring to zero the steady levels, are shown in c. B, the same from an oocyte expressing wild-type CAATCH1. C, the same for an oocyte expressing S308T_KAAT1. D, the charge movement (Q) obtained by integration of the transients in Ac (□) and in Cc (^) is plotted as an average between ‘on’ and ‘off’ areas, and following the convention of setting the zero charge level at the holding potential. E, time constants of decay from Ac (□) and Cc (^). Data from three representative oocytes.

These currents were isolated (Fig. 7Ac) by subtracting from the control traces those recorded in the presence of 500 μm leucine (Fig. 7Ab). The remaining negative steady-state currents were brought to zero using a standard leak-subtraction procedure. This was necessary to allow the integrated charge to reach a steady value during the ‘on’ transient. The correctness of this procedure was validated by the good equality of the ‘on’ and ‘off’ time integrals, which are plotted in Fig. 7D as the amount of displaced charge, while the time constant of relaxation (τ) is shown in Fig. 7E. Similarly to KAAT1, the presteady-state currents of CAATCH1 are abolished by addition of leucine (Fig. 7B), although no transport-associated current is produced, but on the contrary the steady currents are reduced (Fig. 7Bb).

Analysis of the currents elicited by voltage jumps in KAAT1_S308T reveals that this mutant behaves like CAATCH1 in this respect: the Na⁺-dependent presteady-state currents are preserved (Fig. 7Ca) and resemble those of the wild-type, although no transport-associated currents are generated by addition of leucine (Fig. 7Cb). As shown in Fig. 7D and E, the voltage dependence and absolute values of Q and τ, are also comparable to those of the wild-type.

This behaviour confirms then that substitution of serine 308 with threonine has the effect of stabilizing the binding of leucine to KAAT1, mimicking all the effects of this amino acid on CAATCH1.

The abolishment of the presteady-state currents upon addition of 500 μm leucine in KAAT1_S308T demonstrates that the mutant transporter is still able to interact with leucine. The inhibition of the presteady-state currents by leucine is dose dependent, as it is shown in Fig. 8, with an IC₅₀ value in the low micromolar range.

Figure 8.

A, presteady-state currents obtained by subtracting the current traces in the presence of the indicated leucine concentration from the control traces (Na⁺ solution) in a representative KAAT1_S308T-expressing oocyte. B, average dose–inhibition relationship in S308T-expressing oocytes. The leucine concentration inhibiting half the maximal charge movement, was estimated to be 4.2 ± 0.5 μm (n = 4, mean ± s.e.m.).

Discussion

The intestinal lumen of the larva of the hornworm Manduca sexta displays rather peculiar ionic and electrical conditions: due to the particular diet of the animal, and to the presence of a proton pump and of a H⁺/K⁺ exchanger in the epithelial goblet cells, the potassium concentration is very high (around 200 mm), while the pH is strongly alkaline, and the apical transmembrane potential very negative (−150 mV to −200 mV) (Hanozet et al. 1992; Dow & Peacock, 1989). To achieve uptake of nutrient amino acids in these conditions, the ion-coupled transporters present in the apical membrane of the absorptive cells, KAAT1 and CAATCH1, have developed peculiar functional adaptations, such as the ability to use the potassium electrochemical gradient to energize transport (Harvey et al. 1975; Hennigan et al. 1993), and optimal efficiency at alkaline pH (Peres & Bossi, 2000; Vincenti et al. 2000). Although both belonging to the Na⁺/Cl⁻-dependent family, and 90% identical to each other in terms of sequence, KAAT1 and CAATCH1 display a different amino acid potency order, that is furthermore dependent on the driver cation and, to some extent, on the membrane potential. All these characteristics make KAAT1 and CAATCH1 interesting tools to study the structural determinants of substrate binding, and the principles governing substrate selectivity and affinity.

Substrate competition

The effects of leucine in CAATCH1 indicate that this amino acid acts as a blocker of all the three kinds of current that may be observed in this transporter: the leak or uncoupled current, the voltage-dependent transient current, and the transport-associated current. These results suggest that in CAATCH1 leucine binds strongly to the transporter, locking it in an inactive state, unable to proceed in the transport cycle, preventing the binding of other substrates, inhibiting the intramembrane charge movement and blocking the leak pathway.

The substrate competition experiments in KAAT1 reported above indicate that leucine has a dominant role in this transporter as well. The fact that the amplitude of the transport current converges to that characteristic of leucine, even when other amino acids capable of larger currents are present at the same time, suggests, in analogy to CAATCH1, a stronger binding of this amino acid. However, in contrast to CAATCH1, KAAT1 is able to complete the transport cycle in the presence of leucine, though with less efficiency compared to other amino acids.

Kinetic scheme

To interpret the results presented in this paper, we devised the kinetic scheme shown in Fig. 9, designed to include the minimal necessary number of states. State T₁ represents the empty, outward-facing conformation of the transporter, with which cations (Na⁺ or K⁺), and possibly chloride, may interact to move to state T₂, ready to bind the organic substrate. In order to simulate our competition experiments we have represented in the scheme two possibilities: binding of leucine leads to state T₃, while binding of another amino acid (aa) leads to state T₄. From either of these two states the transporter may reach state T₅, a unified inward-facing state that releases substrates in the cytosol and may rearrange to return to state T₁. The uncoupled leak current is represented in the scheme by the T₂ ⇌ T₅ transition, while the presteady-state currents are generated by the voltage-dependent T₁ ⇌ T₂ transition. In other models (Parent et al. 1992; Forster et al. 1998; Sala-Rabanal et al. 2006) rearrangement of the empty transporter (T₅ ⇌ T₁ transition) is also considered to contribute to the intramembrane charge movement; however, this step does not seem to be important in KAAT1, since in this transporter the presteady-state currents are abolished in the absence of Na⁺ and K⁺ (Bossi et al. 1999). In any case, this difference is not relevant in the present context, because these transitions are not involved in the interaction with the organic substrate. In the scheme of Fig. 9, the transitions of interest for the interpretation of the present results are those involved in substrate binding (T₂ ⇌ T₃ and T₂ ⇌ T₄), and the conformational changes of the fully loaded transporter leading to the release of substrates into the cytoplasm (T₃ ⇌ T₅ and T₄ ⇌ T₅).

The grey areas in Fig. 9 underline the voltage-dependent steps. In addition to the charge-moving transition T₁ ⇌ T₂, which generates the presteady-state currents in the absence of substrate, the T₃ ⇌ T₅ and T₄ ⇌ T₅ transitions are also considered voltage dependent, since states T₃ and T₄ are electrically charged and, according to previous determinations (Bossi et al. 1999), the T₁ ⇌ T₂ transition does not occur over the entire electrical field of the membrane.

The system of differential equations (see Methods) arising from the scheme has been solved numerically, using the kinetic parameters of Table 1. The rates k₁₂ and k₂₁, were derived from the data in Fig. 7 and from (Bossi et al. 1999), and are given by the following expressions:

where the factors α and β are the fractions of electrical field sensed by the transition, F is the Faraday constant, R the gas constant and T the absolute temperature. The leucine dissociation rate k₃₂ was derived from the rate of recovery of the currents after leucine exposure (see Fig. 2). The outward-facing conformation of the empty transporter was assumed to be favoured, so that k₅₁ and k₁₅ were arbitrarily set at 2000 s⁻¹ and 20 s⁻¹, respectively. The inward leak rate k₂₅ was estimated from the currents in the absence of organic substrates, while k₅₂ was calculated to ensure microreversibility (Läuger, 1991); The remaining rates were adjusted by trial and error until a satisfactory simulation of the experimental results was obtained, taking into account the current amplitudes and the apparent affinities derived from the experiments. Rates k₅₃ and k₅₄ were calculated to satisfy microreversibility, respectively, in the presence of leucine alone, and in the presence of leucine plus a second substrate.

The expressions for the two rates leading from states T₃ and T₄ to state T₅ were:

where δ is the fraction of the electrical field sensed by these transitions, taken as 1 − (α+β).

The total transmembrane current at each potential was given by

where Q is a normalization factor adjusted to give I = −1 at V = −160 mV in the presence of the amino acid giving rise to the largest current in either ionic condition. Simulation of the charge movement was performed by numerically solving system (1) in the absence or presence of substrates to obtain steady-state and time-dependent values. The presteady-state currents, given by I_pre = (α+β) × (k₂₁× (T₂) −k₁₂× (T₁)) were satisfactorily fitted with single exponentials to determine the relaxation time constants, while the normalized steady-state charge distribution was given by the occupancy probability of state T₂.

Simulation results

The voltage dependence of the simulated transport currents induced by leucine and a second amino acid present separately or in combination (each at 500 μm concentration) is shown in Fig. 10A for the Na⁺ condition, and in Fig. 10B for the K⁺ condition (continuous lines). To facilitate comparison with the experimental data (symbols), the I–V relationships are the result of the subtraction of the leak current from the transport current in each situation. It is worthwhile to note in this respect, that the simulations show that the transport current is not simply additive to the leak current, but the presence of transport current through the pathways T₂ → T₃ → T₅ and T₂ → T₄ → T₅ has the effect of lowering the current through the pathway T₂ → T₅, suggesting that the commonly used procedure of defining the transport current as the current in the presence of substrate minus the current in its absence may not be entirely correct.

Figure 10.

Comparison between simulations and experimental observations In A and B the leucine dominance over threonine in Na⁺ solution (A), and over proline in K⁺ solution (B), observed experimentally in wild-type KAAT1 (symbols, data from two representative oocytes) is well reproduced by the model (continuous lines). C, voltage dependence of K_0.5,leu in Na⁺ and in K⁺; open symbols are experimental data from Figs 5D and 6C, filled symbols are from the simulated currents. D, simulation of decreased leucine specificity in the S308T mutant: reducing the k₃₅₀ rate 100-fold (from 4 s⁻¹ to 0.04 s⁻¹) makes the calculated leucine current negligible in Na⁺ (•, relative to the current elicited by threonine (^); the same relative k₃₅₀ decrease (from 3000 s⁻¹ to 30 s⁻¹) in K⁺ simulates the reduced leucine current relative to the threonine current (compare to Fig. 6). The model also simulates the characteristics of the presteady-state currents in Na⁺, both for the wild-type and the mutant (using the same reduction from 4 s⁻¹ to 0.04 s⁻¹ for the rate k₃₅₀). E, charge versus voltage curves obtained as the occupancy probability of state T₂, and vertically shifted to set zero charge at the holding potential of −60 mV; no differences are apparent between wild-type (open symbols) and S308T mutant (filled symbols) either in the absence of leucine (squares) or at saturating leucine (circles), indicating that in both isoforms the presence of leucine abolishes the presteady-state currents (compare with Fig. 7C). The voltage dependence of the time constant is also unaffected by the mutation (F, compare with Fig. 7E).

Clearly the dominating effect of leucine is satisfactorily reproduced by the simulations, supporting the indication that in the presence of Na⁺, leucine dissociation from the binding site to the external solution is slow, and furthermore also the progress to the inward-facing conformation of the transporter occurs at a low rate when leucine is bound.

To simulate the behaviour in the presence of K⁺, a different set of parameters must be used (Table 1). To account for the negative shift in the voltage dependence of the presteady-state and transport currents (Bossi et al. 1999), the rates k₁₂, k₂₁, k₂₅, k₃₅ and k₄₅ were moved by 100 mV toward more negative potentials. Since in K⁺ leucine becomes the most efficient substrate, the k₃₅ rate has been substantially increased, while k₃₂ has been kept constant. Figure 10B shows that the simulated curves are able to satisfactorily reproduce the results of the competition experiments also in K⁺ solution, although the effect of leucine on the current amplitude is opposite.

Figure 10C shows in addition the values of K_0.5 for leucine obtained from the simulations both in Na⁺ and in K⁺ conditions. Clearly the agreement with the experimentally derived values is very good, confirming that the set of parameters used for the simulations is satisfactory.

The effects of the mutation S308T on the leucine-induced currents may be also reproduced by the model, both in Na⁺ and in K⁺, by simply reducing the values of the rate k₃₅. This is shown in Fig. 10D where the I–V curves obtained using a 100-fold reduced k₃₅₀ (from 4 s⁻¹ to 0.04 s⁻¹ in Na⁺, and from 3000 s⁻¹ to 30 s⁻¹ in K⁺, and keeping all the other parameters constant) are compared to the simulations of the threonine-induced currents. According to the experimental observations, while the current in potassium becomes significantly smaller, in sodium the leucine-induced current is negligible.

As expected from the fact that the rate constants for the T₂ ⇌ T₂ transition are derived from experimental data (Bossi et al. 1999), the model satisfactorily simulates the characteristics of the presteady-state currents in the wild-type, in terms of voltage dependence of both the charge movement and of the relaxation time constant (Fig. 10E and F). Furthermore, addition of saturating amounts of substrates abolishes the presteady-state currents both in the wild-type and in the S308T simulation (i.e. using k₃₅₀ = 4 s⁻¹ or 0.04 s⁻¹).

The role of S308

From the structural point of view, we have seen that serine 6.56, a residue contributing to the binding of leucine in LeuT_Aa (Yamashita et al. 2005), is present in KAAT1 and other leucine cotransporters, while it is absent in CAATCH1 and in other eukaryotic members of the family unable to transport this amino acid. As shown in Figs 5 and 6, replacement of serine 308 with a threonine present in CAATCH1, abolishes the transport-associated current in the presence of Na⁺, and strongly reduces it in the presence of K⁺. The analysis of the presteady-state currents in this mutant (Figs 7 and 8) clearly indicates that the transporter is still able to interact with leucine, since in presence of the amino acid these currents are abolished with an IC₅₀ = 4.2 μm, a value lower than the apparent affinity derived from the transport activity of the wild-type (K_0.5≈ 20–50 μm, Fig. 5D). The loss of function of S308T appears then to be due to the inability to complete the transport cycle rather than to impaired interaction with leucine. Thus, KAAT1_S308T exhibits the same characteristics as CAATCH1 concerning the role of this amino acid.

Although the lack of a detailed knowledge of the atomic structure of KAAT1 and CAATCH1 does not allow one to exclude indirect allosteric effects of serine 308, our results confirm that this residue has a major role in the pore structure, as described in LeuT (Yamashita et al. 2005), where the equivalent serine 256, together with glycine 260, appears to adopt an extended, non-helical, conformation that may affect hydrogen bonding and ion coordination, possibly favouring the transition to the inward-facing conformation.

Leucine specificity

From the above considerations the amino acid specificity in KAAT1 appears to be determined by a combination of affinity proper, that is the ratio of dissociation to association rates of the substrate to the transporter, and the role of the rates at which the subsequent steps in the transport cycle occur. These two factors act somehow in contrast to each other to determine the overall transport efficiency, since high affinity implies a lower free-energy level of the bound state, from which it is more difficult to proceed in the transport cycle (that then will occur less freequently). Indeed, in Na⁺ conditions, the dominance of leucine is consistent with low exit rates from state T₃ towards both T₂ and T₅, since its infrequent replacement by other amino acids is accompanied by a low transport rate. However in K⁺ conditions, the leucine dominance goes along with a high turnover rate, at least at the physiological negative potentials. This may be accounted for in the scheme of Fig. 9 by a low dissociation rate T₃ → T₂, together with a high rate of the T₃ → T5 conformational change.

Physiological and biophysical implications

The electrophysiological data show that at the physiologically relevant membrane potential (very negative) and ionic conditions (high K⁺) of the Manduca sexta intestine, leucine is at the same time the preferred and the most efficiently transported amino acid. This occurs at the expense of other potentially transportable amino acids. While understanding the physiological relevance of this observation requires better knowledge of the dietary requirements of the animal in terms of relative quantities of the various amino acids, and of the amino acidic composition of its diet, some speculations may be of interest regarding the biophysical basis of this selectivity mechanism.

In fact the considerations developed here for the competition of different amino acids in these cotransporters may find a remarkable analogy in the behaviour of some ionic channels in which an apparently anomalous competition among ions occurs. As an example, most Ca²⁺-selective channels are found to be permeated much better by Na⁺ when Ca²⁺ is absent, while at physiological Ca²⁺ no Na⁺ can flow even if the Ca²⁺ concentration is much lower than that of Na⁺ (Almers & McCleskey, 1984). This phenomenon has been explained by the existence of high-affinity binding sites for Ca²⁺ in the channel pore that may then select Ca²⁺ over Na⁺, but at the price of a much smaller current flux (Hess & Tsien, 1984).

In KAAT1 in high Na⁺, leucine appears to perform the role played by Ca²⁺ in the channels, while threonine and proline perform the role played by Na⁺, that is leucine is preferred but at the price of a slower transport rate. However the analogy fails in the presence of K⁺, since in this case the preference for leucine goes together with a high transport rate; clearly this diversity reveals the difference between the Ca²⁺ ionic channel in which the model of permeation is based on electrostatic repulsive forces between two adjacent Ca²⁺ ions (Hess & Tsien, 1984), and the conformational change required for the transporter operation.