The 3-4 loop of an archaeal glutamate transporter homolog experiences ligand-induced structural changes and is essential for transport

Abstract

Glutamatergic synaptic transmission is terminated by members of the excitatory amino acid transporter (EAAT) family of proteins that remove glutamate from the synaptic cleft by transporting it into surrounding glial cells. Recent structures of a bacterial homolog suggest that major motions within the transmembrane domain translocate the substrate across the membrane. However, the events leading to this large structural rearrangement are much less clear. Two reentrant loops have been proposed to act as extracellular and intracellular gates, but whether other regions of these proteins play a role in the transport process is unknown. We hypothesized that transport-related conformational changes could change the solvent accessibilities of affected residues, as reflected in protease sensitivity or small-molecule reactivity. In the model system Glt_Ph, an archaeal EAAT homologue from Pyrococcus horikoshii, limited trypsin proteolysis experiments initially identified a site in the long extracellular loop that stretches between helices 3 and 4 that becomes protected from proteolysis in the presence of a substrate, L-aspartate, or an inhibitor, DL-TBOA in the presence of Na⁺, the cotransported ion. Using a combination of site-directed cysteine-scanning mutagenesis and fluorescein-5-maleimide labeling we found that positions throughout the loop experience these ligand-induced conformational changes. By selectively cleaving the 3-4 loop (via introduced Factor Xa sites) we demonstrate that it plays a vital role in the transport process; though structurally intact, the cleaved proteins are unable to transport aspartate. These results inculcate the 3-4 loop as an important player in the transport process, a finding not predicted by any of the available crystal structures of Glt_Ph.

The excitatory amino acid transporter (EAAT) family of glutamate transporters terminate glutamatergic synaptic transmission by transporting ambient glutamate from the synaptic region into glial cells and neurons (1), thus facilitating robust transmission of the glutamate signal across the synaptic cleft. EAATs are well characterized functionally, driving glutamate uptake by the cotransport of 1H⁺ and 3Na⁺ ions and the countertransport of 1K⁺ (2). EAATs also possess an uncoupled chloride conductance that is enhanced by Na⁺ or glutamate. Less is known, however, about the structural rearrangements that accomplish transport. The X-ray crystal structures of an archaeal member of this family from Pyroccocus horikoshii, Glt_Ph, have advanced our understanding of this process considerably (3–5), yet much is still a mystery. Glt_Ph shares ∼35% sequence identity with the EAATs (3) and is functionally similar (4, 6); coupling the transport of aspartate to the cotransport of 3Na⁺ ions (7) and possessing a ligand-activated, uncoupled chloride conductance (8). Thus, it serves as an excellent model with which to probe conformational changes that drive transport in the EAATs.

Crystal structures presumed to represent the Glt_Ph extracellular facing state reveal a substrate binding pocket forms toward the extracellular face of the protein between the tips of reentrant loops HP1 and HP2 (Fig. 1) (3, 4). Structures of the apo-state and with the broadly specific EAAT competitive inhibitor, DL-threo-b-benzyloxyaspartate (TBOA) bound (9) together with functional data implicate HP2 as an extracellular gate (10–16). The structure of a cross-linked form of Glt_Ph suggests the protein undergoes a major conformational change to move the substrate across the membrane, moving the substrate binding pocket approximately 18 Å toward the cytoplasmic side of the protein. Though biochemical data accompanying the structure suggest that the crystallized state is close to a native structure of the protein and it is strongly supported by an independent computational study (17), previous functional and modeling data imply a smaller movement (18–20). Therefore, independent assessments of conformational changes are necessary to validate this structure. Furthermore, the crosslinked structure leaves open many questions regarding the dynamics of transport.

Fig. 1.

The 3-4 loop of a Glt_Ph monomer. (A) Structure of a single monomer of TBOA-bound Glt_Ph(4) (Protein Data Bank accession 2NWW) is shown, with the 3-4 loop highlighted in green and the trypsin cleavage site K125 in orange. HP1 (yellow) and HP2 (red) are indicated, as is a bound TBOA molecule (blue). (B) Amino acid sequence of the 3-4 loop of Glt_Ph.

Limited proteolysis has had success in identifying regions of transporter proteins that undergo changes in conformation (21, 22). Here, we apply this method to Glt_Ph and, in combination with fluorescein maleimide accessibility measurements, we find that a long loop stretching between helices 3 and 4 (Fig. 1 A and B) experiences substrate and inhibitor induced conformational changes. By introducing specific proteolytic cleavage sites into this loop, we demonstrate that while the cleaved protein is intact and properly folded, it is no longer competent for transport. Thus the conformational change we observe is an essential component of the transport mechanism.

Results

Trypsin Cleaves Glt_Ph in a Substrate and Inhibitor Dependent Manner.

We used limited proteolysis to locate regions of Glt_Ph that move as substrate binds to the protein. If changes in protein conformation alter the accessibility of a site to protease cleavage, we expect to observe different cleavage fragments under conditions favoring alternate conformations. We gently trypsinized purified Glt_Ph (solublized in n-dodecyl-β-D-maltopyranoside, DDM) in varying combinations of substrates and inhibitors. Digestion of Glt_Ph (45 kD at full-length) in the absence of substrate or inhibitor yields two distinct fragments, running at approximately 25 and 30 kDa on a polyacrylamide gel, as well as several diffuse lower molecular weight bands (Fig. 2A). However, adding either aspartate or the inhibitor TBOA results in the disappearance or drastic reduction in intensity of the smaller of these fragments (arrow) and an increase in the intensity of the full-length protein band (FL), reflecting protection of the proteolytic site from Trypsin. (Note that the intensity of the larger proteolytic fragment is unchanged.) Such protection implies that binding of substrate or inhibitor induces a conformational change in the protein that alters the accessibility of the cleavage site. Both Na⁺ and either aspartate or TBOA are required for protection; it does not occur when Na⁺ is replaced with K⁺ (Fig. 2B) or when aspartate is replaced with the minimally transported glutamate or the nontransported serine (Fig. 2C). The protection from cleavage by aspartate (Fig. 2D) and Na⁺ (Fig. 2E) depends on the doses of both, with half-maximal effects between 100 nM and 1 μM aspartate and around 2 mM Na⁺. These values are near the published K_Ms of transport for aspartate (120 nM) and Na⁺ (3.9 mM) (6), suggesting that the observed conformational change is related to the transport process. Similar dose-dependence is observed with Na⁺ in the presence of TBOA (Fig. S1).

Fig. 2.

Limited trypsin proteolysis of Glt_Ph. Uncleaved Glt_Ph (first lane) and Glt_Ph proteolysis fragments separated by SDS/PAGE and stained with Coomassie Blue (A) in the presence (+) and absence (-) of 1 mM L- aspartate or 5 mM TBOA; (B) in the presence of 10 mM Na⁺ or K⁺ and 1 mM aspartate; or (C) in the presence of aspartate (Asp), glutamate (Glu) or serine (Ser) and 10 mM Na⁺. Also illustrated are trypsin cleavage fragments produced in the presence of varying concentrations of asparate (D) or Na⁺ (E). Positions of bands representing full-length Glt_Ph (FL) and the protected 25 kDa cleavage fragment (arrow) are noted.

Using N-terminal sequencing of the protected band, we identified the protected trypsin site as lysine 125 (K125), located in the extracellular loop between helices 3 and 4. We confirmed the location of the trypsin cleavage site by removing it, mutating the target lysine (125) to cysteine (note that this mutation was introduced in the C321S background, which lacks the single native cysteine) (6). The K125C mutant is fully functional (Fig. S2A) yet it is far more resistant to trypsin proteolysis than wild-type Glt_Ph (Fig. S2B) and does not show ligand-dependent trypsin sensitivity (Fig. S2C). The multiple bands observed in the cleaved K125C mutant presumably represent cuts at alternative trypsin sites.

Accessibility Changes Throughout the 3-4 Loop.

We investigated whether the conformational change revealed by trypsin proteolysis involves other portions of the 3-4 loop using a combination of cysteine-scanning mutagenesis and site-directed fluorescence labeling. We reasoned that alternate conformations induced by substrate or inhibitor might change the accessibility (and thereby reactivity) of a single introduced cysteine to fluorescein-5-maleimide (FM), a cysteine-specific, bulky, hydrophilic fluorescent probe. By running the protein on a gel (which removes the unreacted dye) and imaging it on a UV transilluminator, we semiquantitatively assess the extent of labeling, correcting for variations in the amount of protein loaded by subsequent Coomassie Blue staining of the same gel.

In the background of the cysless construct (6), we mutated residues, from 111 to 129 in the 3-4 loop individually to cysteine (23). We also introduced a cysteine at a control position, 34, in the short extracellular loop between helices 1 and 2, not expected to be involved in these conformational changes. Each mutant was purified and reconstituted into lipid vesicles under conditions suitable to drive transport (24).

We then labeled the reconstituted cysteine mutants with FM in the presence or absence of aspartate or TBOA (at all times Na⁺ was present at 100 mM), removing and quenching aliquots with excess L-cysteine at increasing time points. To visualize the results of the labeling reaction, we used the technique described by Geertsma et al. (25), which eliminates the need for prior processing to remove the lipids. Quantification of the fluorescent and Coomassie stained bands allows the extent of labeling to be expressed as a normalized “fluorescence to protein” ratio. Detailed results of the labeling reaction in an example loop mutant (Q121C) and in a negative control (G34C) are shown in Fig. 3. Examining the time course of labeling reveals that Q121C is labeled at a much higher rate by FM in the presence of aspartate than in its absence, whereas the rates are identical for the control [Fig. 3A (gels) and B (graphs)]. Because it is impractical to examine the full time course of each reaction at each position using our gel-based method, we chose a time point (6 min.) where the different rates produced substantially different extents of labeling and used the extent of labeling as a surrogate for the irreversible pseudo-first order rate in further experiments (Fig. 3C). We observed a similar, but even more dramatic, effect on labeling when we compared the rates in the presence and absence of TBOA (Fig. 3C).

Fig. 3.

Substrate and inhibitor induced effects on Fluorescein-5-maleimide labeling of Glt_Ph single cysteine mutants G34G and Q121C. Time course of fluorescein-5-maleimide (FM) labeling of Glt_Ph mutant Q121C reconstituted into proteoliposomes in the presence and absence of 5 mM L-aspartate. The reaction was quenched with L-cysteine at the indicated times in minutes. (A) Top: the band of interest after separation from free label by SDS/PAGE, imaged using a UV transilluminator (FM). Bottom: the same gel after staining with Coomassie Blue, imaged using visible light (CB). Note that contamination caused stray fluorescence overlaying the 1 min time point with no ligand—that point was excluded from further analysis and is not plotted in the graph below. (B) Quantification of the Q121C labeling time course shown in A Left, or that of G34C, Right; solid symbols without aspartate, open symbols with aspartate. Labeling is expressed as the intensity UV fluorescent band F normalized to the intensity of the Coomassie stained protein P. (C) Extent of FM labeling of mutants Q121C and G34C in the absence of ligand (-) or the presence of 5 mM L-aspartate, or 5 mM DL-TBOA. The reaction was quenched after 6 mins. Na⁺ was present at 100 mM in all conditions.

We ruled out substantial functional effects of the 3-4 loop mutations by measuring Na⁺-driven ³H-L-aspartate transport into vesicles (Fig. 4A). Like the cysless background protein, all the mutants support substrate uptake. Though there are variations in the transport activity, none of the mutants is severely impaired. Scanning the entire 3-4 loop in the absence of aspartate or TBOA reveals variations in the extent of FM labeling along the length of the loop, with a greater than fivefold increase between the least and most accessible residues (Fig. 4B). Performing the labeling reaction in the presence of substrate or inhibitor increased the extent of FM labeling at multiple positions (Fig. 4C). Interestingly, there are no positions at which either ligand decreases the rate of labeling. TBOA always acts at the same residues as aspartate, and its effect is always greater than that of aspartate. We compared the effects of substrate and inhibitor by normalizing fluorescence to protein ratios at each position to those measured in the absence of either aspartate or TBOA at that position. The magnitude of the changes induced by substrate or inhibitor varies along the loop (Fig. 4C). In general the greatest increases in labeling occur with a peak every 5 or 6 residues (positions 111, 116, 121, and 127).

Fig. 4.

Transport activity and FM accessibilities of Glt_Ph 3-4 loop single cysteine mutants. (A) Transport activity of wild-type, wt, cysless, cl, and single cysteine Glt_Ph mutants at the indicated positions expressed as the initial rate of ³H-L-aspartate uptake into proteoliposomes (3 ug protein/mg lipid) in the presence of an inwardly directed Na⁺ gradient (100 mM Na⁺ outside, 100 mM K⁺ inside). All rates are normalized to that of the cysless mutant. (B) FM labeling of the single cysteine residues before addition of L-aspartate or TBOA. Proteoliposomes containing each cysteine mutant (3 ug protein/mg lipid) were reacted with FM for 6 mins in the absence of L-aspartate or TBOA. FM labeling was quantified as in Fig. 3. (C) Change in FM labeling of each single cysteine mutant upon addition of L-aspartate and TBOA. Each mutant was reconstituted into proteoliposomes in the absence, white bars, or presence of 5 mM L-aspartate, black bars, or 5 mM TBOA , gray bars and labeled as in B. To assess the relative effect of the ligands at each position, the extents of labeling in each condition at each position (F/P) are normalized to the condition with no ligand present. An underestimation in the error of the no ligand condition occurs because of the gel-based method we are using to detect the FM labeling (see Materials and Methods), we indicate this by pale gray error bars. Na⁺ was present at 100 mM in all labeling reactions.

Thus, residues throughout the 3-4 loop undergo large substrate and inhibitor-induced increases in accessibility to fluorescein-5-maleimide, reflecting changes in accessibility to the aqueous environment and demonstrating that the conformational changes impacting K125 do extend throughout the loop. These data, however, tell us nothing about the functional relevance of such movements; are they simply reporting structural rearrangements of another part of the protein, or is the 3-4 loop itself an integral part of the transport process?

The 3-4 loop Is Essential for Transport Activity.

We reasoned that if the 3-4 loop has an active role in the transport mechanism, then cleaving or removing it should affect Glt_Ph’s ability to transport substrate. To test this we used the protease Factor Xa, which specifically cleaves after the sequence I(D/E)GR, to cut the backbone at either or both of two positions bracketing the loop. We engineered the two Factor Xa sites into the wild-type background, allowing us to cleave the loop after residues 114 (Xa114) and/or 125 (Xa125) (see Fig. 1A). Note that the latter site corresponds to the trypsin site described earlier. When purified and digested in DDM, the three mutants are highly susceptible to proteolysis by Factor Xa, whereas the WT protein is unaffected (Fig. 5A). Cleavage of the two single mutants results in the expected fragments, confirming that Factor Xa is specifically cleaving at the intended position. Factor Xa treatment of the double mutant results in the smaller, Xa114, N-terminal fragment and both expected C-terminal fragments, revealing that most of the protein is doubly cut, with some cleaved only at position 114.

Fig. 5.

Factor Xa proteolysis of Glt_Ph (A) Factor Xa cleavage of wild-type Glt_Ph, wt, and mutants containing one, Xa114 and Xa125, or two, Xa114/125, Factor Xa recognition sites within the 3-4 loop. Each protein was incubated with Factor Xa (20 μg/mg protein) or control buffer at pH8.0 for 24 hrs (thrombin was added after 8 hrs). The resulting cleavage products were separated by SDS/PAGE and stained with coomassie blue. Positions of the full-length protein and N- and C-terminal fragments on the gel are indicated. (B) Far-UV circular dichroism spectra and (C) size-exclusion chromatography elution profiles of Xa114 (cyan), Xa125 (green), and Xa114/125 (pink) after partial cleavage by Factor Xa (as A except pH7.4) and untreated wild-type Glt_Ph (black). The gel in C represents 1 ml fractions collected during the SEC elution of mutant Xa114/125 (lanes are aligned with appropriate elution volumes) showing that the cleavage fragments and full-length protein elute together in the major peak at ∼11.4 ml and that the smaller peak at ∼15 ml is due to Factor Xa. (D) Relative initial rates of Na⁺-driven ³H-L-aspartate uptake into proteolipsomes containing wild-type Glt_Ph , Xa114, Xa125, or Xa114/125 either with (+) or without (-) cleavage by Factor Xa. * p < 0.05, ** p < 0.01, ns = no statistical significance according to the student’s t-test.

Cleavage and removal of the 3-4 loop allowed us to investigate the role it plays in the transport process; if the loop is an integral piece of the transport mechanism, then these manipulations should affect the protein’s ability to transport substrate. Cutting the protein backbone, however, is a somewhat radical maneuver that could have repercussions on the structural integrity of the entire protein. To rule out such effects, we used two independent methods to demonstrate that the cleaved proteins remain as intact, correctly folded trimers. First, we used circular dichroism (CD) to characterize the secondary structure of the cleaved mutants, a good indicator of the folding status of a protein. The far-UV CD spectra of these proteins are essentially indistinguishable from that of the uncleaved wild-type protein (Fig. 5B), confirming that the cleaved mutants retain the same overall secondary structure as the native protein and therefore are not misfolded.

Second, we used size-exclusion chromatography (SEC) to assess the quaternary structure of the cleaved proteins, confirming that they retain their trimeric stoichiometry after cleavage. If the mutants aggregate or are no longer correctly assembled, they will elute from the column at a different volume compared with the correctly folded trimeric wild-type protein (26). We partially digested each of the three mutants with Factor Xa before separation by SEC; doing so provided a mix of cleaved and uncleaved proteins and facilitated a direct comparison of the elution profiles of the two forms. All three mutant proteins eluted at approximately 11.4 ml as sharp, symmetrical peaks, superimposable upon each other and on that from the intact wild-type protein (Fig. 5C). SDS/PAGE analysis reveals that the full-length proteins and cleaved N- and C-terminal fragments all elute in this peak and are not present anywhere else in the elution profile. We obtained essentially identical results with fully cleaved protein. Along with the CD data, this confirms that after Factor Xa treatment the mutants remain intact, correctly folded, and trimeric. The gel also reveals that the small peak at ∼15 ml results from the Factor Xa protease itself.

Once convinced that Factor Xa cleavage preserves the structural integrity of the protein, we investigated its functional implications. Wild-type Glt_Ph and the three mutants were treated with Factor Xa or control buffer under the same conditions and reconstituted into lipid vesicles. We investigated whether removal or cleavage of the 3-4 loop had any effect on the transport activity of Glt_Ph using Na⁺-driven ³H-aspartate uptake measurements. To verify that roughly equal amounts of protein incorporate into lipid vesicles after Factor Xa treatment, we recovered the protein from the vesicles and resolved it from lipid using SDS/PAGE (Fig. S3). The uncleaved mutants all support aspartate uptake (Fig. 5D), albeit at a reduced rate compared to the wild-type protein (not surprising given that each mutant contains at least four altered residues). Cleavage, however, essentially abolishes the ability of all three mutants to transport aspartate (Fig. 5D). In contrast, the effect of Factor Xa treatment on the wild-type protein is not statistically significant. These data demonstrate that an intact 3-4 loop is necessary for transport to occur and thereby reveal that the 3-4 loop is an essential component of the transport mechanism.

Discussion

The crystal structures of Glt_Ph have provided extraordinary insights into how glutamate transporters move their substrates across the cell membrane, secluded within a transport domain that rocks back and forth in its entirety. However, these structures have yielded little insight as to whether other regions of the protein are involved in the transport mechanism. Using limited trypsin proteolysis we found such a region; trypsin cleavage at K125 in the extracellular loop connecting transmembrane helices 3 and 4 (the 3-4 loop) is specifically sensitive to the presence of Na⁺, substrate, and inhibitor. Cysteine-scanning/fluorescein maleimide labeling methods support and extend this conclusion, demonstrating that these substrate-induced structural rearrangements occur throughout the the 3-4 loop. Further targeted proteolysis experiments reveal that this loop is an essential part of the transport mechanism.

Our results confirm and substantially expand upon those from a study published by the Kanner group (27). They found multiple trypsin cleavage sites in the mammalian EAAT, GLT-1, from crude rat brain vesicles, some of which appeared protected in the presence of substrates. However, because the fragments were mapped using western blots probed with peptide-directed antibodies, the sites could not be located precisely. Despite this, their data are consistent with one of their cleavage sites localizing to the 3-4 loop, in agreement with the one we observe in Glt_Ph. Our data further reveal the extent of the change in the 3-4 loop and its central importance to the transport process.

We observe substrate/inhibitor induced increases in fluorescein maleimide labeling in Glt_Ph that undulate along the length of the loop with a periodicity of 5-6 residues, peaking at positions A111C, A116C, Q121C, and A127C and with some residues between showing no change (Fig. 3C). This periodicity does not correspond to any common secondary structure element, but the variability of the changes, together with a distinct accessibility pattern in the absence of either aspartate or TBOA, suggests the loop in the apo state has a well-defined structure, packed tightly against the bulk of the protein. Our experiments reveal movements upon substrate (or inhibitor) binding that increase the aqueous exposure of residues throughout the loop. Perhaps the increased accessibility to FM we observe throughout the loop reflects a freeing-up of the loop’s packing upon closure of HP2.

The trypsin cleavage site in Glt_Ph, K125, appears to behave differently in the proteolysis and FM labeling experiments; it is protected from proteolysis by aspartate or TBOA (a decrease in aqueous exposure), but both ligands make it more accessible to FM (an increase in aqueous exposure). FM is very small in comparison to trypsin and to label a cysteine residue it needs access only to an individual side chain; trypsin is a bulky molecule and its binding is a complex interaction involving several residues (28). Thus, a decrease in trypsin cleavage may not simply reflect burial of the side chain from the aqueous environment as a decrease in FM labeling does; it could be due to steric hindrance by other parts of the protein, changes in accessibility at any of the interacting regions, or even restriction in the mobility of the backbone. Therefore trypsin and FM accessibility may not be expected to exactly mimic each other. The critical observation is that a change is detected by both techniques.

In all cases TBOA appears as or more potent than aspartate at changing a residue’s accessibility. This may represent bona fide differences between the aspartate- and TBOA-bound states; however, it may also be explained by considering the effect of these ligands on the transport cycle. TBOA is a competitive inhibitor of Glt_Ph (29), because it binds to the protein in a manner similar to aspartate, yet it causes the transport cycle to stall, essentially trapping an outwardly facing state. In contrast, aspartate binding triggers the protein to move among its full range of conformations. Thus, if the state stabilized by TBOA is one with high aqueous exposure of the 3-4 loop, then the inhibitor will highly populate that state and will appear to facilitate FM labeling more potently than the substrate. We initially expected labeling in the presence of TBOA to mimic that with no ligand bound. However, the assumption that the apo-protein structure is similar to the TBOA-bound structure is based on only the conformation of HP2 in the TBOA-bound and apo-states (4). The full apo-protein structure itself has not been published.

Are our proposed movements consistent with the published crystal structures of Glt_Ph? A recent structure of a cross-linked form of Glt_Ph, which likely represents the inward-facing state, reveals a major conformational change in the transmembrane domain of the protein compared with the previously published outward-facing form (5). Our observations of extensive accessibility changes support the idea of a large-scale protein movement underlying transport; however, despite the extensive movement implied by the structures, they provide no obvious explanation for the accessibility changes we observe in the 3-4 loop. The original structures are thought to represent the outward-facing apo/open (TBOA-bound) and closed (aspartate-bound) states (4); while these structures suggest there are potentially small conformational movements at the C-terminal end of the 3-4 loop, they also cannot account for the large changes in accessibility we see throughout the loop. Close inspection of the crystal structures, however, reveals that the 3-4 loop is extensively involved in packing interactions within the crystal lattice: All crystal contacts on the extracellular face of the protein in both the inward- and outward- facing structures are mediated by this loop. These interactions are likely to significantly constrain the flexibility of the loop and restrict its possible conformations. Correspondingly, when substrate-bound crystals in the outward-facing state were treated to obtain an apo-structure, the proposed extracellular gate remained largely closed in the two monomers involved in the lattice interactions (4), hinting that movements of the 3-4 loop may be necessary for the extracellular gate to open. Hence, it is not surprising that the accessibility changes we observe within the 3-4 loop are not apparent from the crystal structures.

Our most striking finding is that the 3-4 loop is an essential part of the transport mechanism in Glt_Ph; severing the backbone within the loop abolishes transport while having no effect on the structural integrity of the protein. This is the first direct evidence that the transport mechanism involves regions of the protein other than the “transport domain,” which comprises helices 3, 6, 7, 8 and the two reentrant loops. One possibility is that the loop acts as a taught string between the two domains of the protein, communicating conformational changes across the membrane. We find that mutations within the 3-4 loop affect the protein’s ability to transport aspartate: All three of the Factor Xa mutants (four consecutive amino acid mutations) and many of the single cysteine mutants have significantly altered transport activities. These effects on the transport process suggest that interactions between the 3-4 loop and the rest of the protein play a role in the transport process. The 3-4 loop stretches across the top of the mobile transport domain and links the moving helix 3 to the immobile helix 4, making it an attractive region to be involved in coordinating the transport process.

Our work thus demonstrates the importance of an extracellular loop in the transport process. Because such loops are often poorly represented by crystallography, other methods, like those presented here, will be essential tools to analyze their contributions to transporter mechanisms.

Materials and Methods

Protein Purification and Reconstitution.

All single cysteine mutants were made using standard methods for site-directed mutagenesis in a background where the single native cysteine was mutated to serine (C321S, this mutant is fully functional). Glt_Ph protein was purified as described elsewhere (4). Protein was reconstituted into liposomes formed from E.coli polar lipids and 1-Palmitoyl-2-Oleoyl-sn-Glycero-3-Phosphocholine (Avanti Polar Lipids) at a ratio of 3∶1 as described previously (24).

Trypsin Proteolysis.

Limited proteolysis experiments were performed in a buffer containing 10 or 100 mM NaCl, 20 mM Tris/Hepes pH7.4, 1 mM CaCl₂, 700 μM NaEDTA, 7 mM n-dodecyl-β-D-maltopyranoside for 30 min at 37 °C at a ratio of 0.25 BAEE units bovine pancreatic trypsin (Sigma Chemical cat#T8658) per 30 μg Glt_Ph. More rigorous conditions (10 BAEE units trypsin per μg Glt_Ph) were used to assess the resistance of the K125C mutant. Note that we have found that the exact trypsin used is critical to observe the substrate protection effects noted here. The reaction was stopped with 1 mM or 10 mM AEBSF and the resulting cleavage fragments separated by SDS/PAGE.

Transport Assay.

Proteoliposomes with an internal solution of 100 mM KCl, 20 mM Tris/Hepes pH7.4 and 350 μM DTT were diluted into 100 mM NaCl, 20 mM Tris/HEPES pH 7.4, 350 μM DTT, 1 μM valinomycin, 100 nM ³H-L-aspartate (GE Healthcare) at 30 °C. Aliquots were removed and quenched by 10-fold dilution into ice cold 100 mM LiCl, 20 mM Tris/Hepes and filtered over nitrocellulose filters (0.22 μm pore size, Millipore). The filters were washed and assayed for radioactivity using a Trilux beta counter (Perkin Elmer). All data represent the mean ± s.e.m. of at least 3 experiments.

Fluorescein Maleimide Labeling of Single Cysteine Mutants.

Each mutant protein was reconstituted into proteoliposomes either “ligand-free” (neither aspartate nor TBOA), or with either 5 mM aspartate or 5 mM TBOA, in all cases the reaction buffer contained 100 mM NaCl, 20 mM Tris/Hepes pH7.4 and 5 μM TCEP. 200 μM fluorescein-5-maleimide (Molecular Probes) was added to the proteoliposomes and the reaction quenched with 2 mM cysteine in an SDS gel loading buffer and loaded directly onto a gel containing 0.2% SDS (high concentrations of SDS in the loading buffer and gel eliminate the need to remove the lipid from the protein) (25). The FM signal was imaged on a UV transilluminator; subsequent Coomassie staining allowed for normalization to the amount of protein in the band. Fluorescence to protein ratios were normalized within each gel to the labeling of the ligand-free protein, allowing accurate assessment of the relative change in FM labeling induced by either ligand.

Factor Xa Proteolysis.

Glt_Ph Factor Xa mutants were cleaved with Factor Xa at a ratio of 20 ug Factor Xa to 1mg protein for 24 hrs at 37 °C. The reaction was stopped by addition of 1 mM AEBSF and 10 mM EDTA. Uncleaved controls were treated in exactly the same manner with omission of the protease.

Size-Exclusion Chromatography and Circular Dichroism.

Partially cleaved Factor Xa mutants were eluted from a Superdex 200 10/300GL column (GE Healthcare) equilibrated in a buffer containing 20 mM Tris/Hepes pH7.4, 100 mM NaCl, 5 mM glutamate and 7 mM n-decyl-β-D-maltopyranoside. Protein was collected from the major peak and concentrated for measurement of CD spectra using a JASCO J-815 CD spectrometer in a 0.5 mm cell at 25 °C.