1. Structural Biology and Molecular Biophysics
  2. Neuroscience
Download icon

Coupled ion binding and structural transitions along the transport cycle of glutamate transporters

  1. Grégory Verdon  Is a corresponding author
  2. SeCheol Oh
  3. Ryan N Serio
  4. Olga Boudker  Is a corresponding author
  1. Weill Cornell Medical College, United States
Research Article
  • Cited 67
  • Views 3,621
  • Annotations
Cite this article as: eLife 2014;3:e02283 doi: 10.7554/eLife.02283

Abstract

Membrane transporters that clear the neurotransmitter glutamate from synapses are driven by symport of sodium ions and counter-transport of a potassium ion. Previous crystal structures of a homologous archaeal sodium and aspartate symporter showed that a dedicated transport domain carries the substrate and ions across the membrane. Here, we report new crystal structures of this homologue in ligand-free and ions-only bound outward- and inward-facing conformations. We show that after ligand release, the apo transport domain adopts a compact and occluded conformation that can traverse the membrane, completing the transport cycle. Sodium binding primes the transport domain to accept its substrate and triggers extracellular gate opening, which prevents inward domain translocation until substrate binding takes place. Furthermore, we describe a new cation-binding site ideally suited to bind a counter-transported ion. We suggest that potassium binding at this site stabilizes the translocation-competent conformation of the unloaded transport domain in mammalian homologues.

https://doi.org/10.7554/eLife.02283.001

eLife digest

Molecules of glutamate can carry messages between cells in the brain, and these signals are essential for thought and memory. Glutamate molecules can also act as signals to build new connections between brain cells and to prune away unnecessary ones. However, too much glutamate outside of the cells kills the brain tissue and can lead to devastating brain diseases.

In a healthy brain, special pumps called glutamate transporters move these molecules back into the brain cells, where they can be stored safely. However, when brain cells are damaged—by, for example, a stroke or an injury,—the glutamate stored inside spills out, killing the surrounding cells. This leads to a cascade of dying cells and leaking glutamate, which causes even more damage and slows the recovery.

Glutamate transporters ensure that there are more glutamate molecules inside cells than outside. However, it requires energy to maintain this gradient in the concentration of glutamate molecules. The transporters get this energy by moving three sodium ions into the cell with each glutamate molecule, and moving one potassium ion out of the cell. However, it is not clear how these transporters ensure that they move the glutamate molecules and the sodium ions at the same time.

Now, Verdon, Oh et al. have uncovered the 3D structure of a glutamate transporter homologue at each step of the transport process. These structures reveal that, on the outside of the cell membrane, sodium ions attach to the so-called ‘transporter domain’ and make it better able to bind glutamate. The transporter domain then carries the sodium ions and glutamate through the cell membrane and releases them into the cell. Verdon, Oh et al. suggest that a potassium ion then binds to the empty transport domain, stabilizing it into a more compact shape that easily makes the return trip to the outside of the cell.

Most experiments on glutamate transporters, including the work of Verdon, Oh et al., are carried out on model proteins taken from bacteria. An important challenge for the future will be to obtain structural information on human glutamate transporters, as these could be therapeutic targets for the treatment of various neurological conditions.

https://doi.org/10.7554/eLife.02283.002

Introduction

Glutamate transporters, or excitatory amino acid transporters (EAATs), reside in the plasma membranes of glial cells and neurons, where they catalyze the re-uptake of the neurotransmitters glutamate and aspartate (L-asp) (Danbolt, 2001). EAATs terminate neurotransmission events supporting memory formation and cognition, and also prevent excitotoxicity caused by overstimulation of glutamate receptors. Dysfunction of EAATs is linked to neurological disorders, poor recovery from stroke and traumatic brain injuries (Yi and Hazell, 2006; Sheldon and Robinson, 2007; Kim et al., 2013). To maintain steep trans-membrane glutamate gradients, EAATs transport one substrate molecule together with three sodium ions (Na+) and one proton. After their release into the cytoplasm, counter-transport of one potassium ion (K+) resets the transporter for the next cycle (Zerangue and Kavanaugh, 1996; Levy et al., 1998; Owe et al., 2006).

Key mechanistic and structural insights into this family of transporters come from studies on an archaeal homologue from Pyrococcus horikoshii, GltPh (Figure 1—figure supplement 1), which symports L-asp together with three Na+ ions (Groeneveld and Slotboom, 2010); however, it shows no dependence on counter-transport of K+ under the conditions tested (Ryan et al., 2009). GltPh, like EAATs, is a homo-trimer (Gendreau et al., 2004; Yernool et al., 2004). Each protomer consists of a central scaffolding trimerization domain and a peripheral transport domain containing the substrate and ion binding sites (Boudker et al., 2007; Reyes et al., 2009). When bound to Na+ and L-asp (‘fully bound’ from here on), each transport domain moves by ∼15 Å across the membrane from an outward- to an inward-facing position, in which the substrate binding site is near the extracellular solution and the cytoplasm, respectively (Reyes et al., 2009). Structurally symmetric helical hairpins, HP1 and HP2, occlude the bound substrate from the solvent and are thought to serve as gates (Boudker et al., 2007; Huang and Tajkhorshid, 2008; Shrivastava et al., 2008; Reyes et al., 2009; DeChancie et al., 2010; Focke et al., 2011; Zomot and Bahar, 2013). Two Na+-binding sites (Na1 and Na2), neither of which directly coordinates the substrate, were identified crystallographically using thallium (Tl+) (Boudker et al., 2007). The location of the third Na+-binding site is being debated (Holley and Kavanaugh, 2009; Huang and Tajkhorshid, 2010; Larsson et al., 2010; Tao et al., 2010; Bastug et al., 2012; Teichman et al., 2012). A highly conserved non-helical Asn310-Met311-Asp312 (NMD) motif interrupts trans-membrane segment (TM) 7 (see below). It lines the back of the substrate- and ion-binding sites and is involved in binding of the ligands (Rosental et al., 2006; Tao et al., 2006; Rosental and Kanner, 2010). The main chain carbonyl oxygen of Asn310 contributes to Na1 site, while the side chain of Met311 protrudes between the substrate, Na1 and Na2 binding sites (Boudker et al., 2007).

Symport requires that neither the substrate nor the ions alone are efficiently transported (Crane, 1977). Therefore to traverse the membrane, the transport domains of GltPh and EAATs must be loaded with both Na+ ions and substrate. To complete the transport cycle, the transport domain of GltPh must also translocate readily when it is free of both solutes (apo), while in EAATs it requires binding of a K+ ion. To establish the structural underpinnings of these processes, we determined crystal structures of the outward- and inward-facing states of GltPh in apo and ions-only bound forms (Tables 1, 2 and 3). We find that the apo transport domain shows identical structures when facing outward or inward. While ligand-binding sites are distorted, the domain remains compact, suggesting that it relocates across the membrane as a rigid body, similarly to when it is fully bound (Reyes et al., 2009). Ion binding to Na1 site, located deep in the core of the transport domain, triggers structural changes that are propagated to the extracellular gate HP2, at least in part, by the side chain of Met311 in the NMD motif. Consequently HP2, which in the apo form is collapsed into the substrate binding and Na2 sites, frees the sites, assuming conformations more similar to the conformation observed in the fully bound transporter. We suggest that these Na+-dependent structural changes underlie the high cooperativity of Na+ and substrate binding, which is thought to be one of the key coupling mechanisms (Reyes et al., 2013). Furthermore, in the structure of Na+-bound outward-facing GltPh we observe opening of HP2 tip, which may facilitate L-asp access to its binding site and prevent the inward movement of the Na+-only bound transport domain, as previously suggested (Focke et al., 2011). Remarkably, soaks of apo GltPh crystals in Tl+ reveal new cation-binding sites within the apo-like protein architecture. One such site overlaps with the substrate-binding site. Because binding of a cation to this site would compete with binding of Na+ and the transported substrate, it is well suited to serve as a binding site for a counter-transported ion. We propose that the closed translocation-competent conformation of the transport domain free of Na+ and substrate is intrinsically stable in GltPh but not in EAATs, in which K+ binding at the newly identified site is required, coupling transport cycle completion to K+ counter-transport.

Table 1

X-ray crystallographic data and refinement statistics for GltPh-R397A and GltPh-K55C-A364CHg (GltPhin) structures deposited at the PDB

https://doi.org/10.7554/eLife.02283.005
GltPhin
apoTl+-bound (apo conf.)alkali-freeTl+-bound (bound conf.)
Data collection
 Space groupC2221C2221C2221C2221
 Cell dimensions
a, b, c (Å)109.93, 201.81, 207.14106.98, 196.56, 206.50106.95, 196.84, 207.48110.83, 200.43, 206.40
 α, β, γ (°)90.00, 90.00, 90.0090.00, 90.00, 90.0090.00, 90.00, 90.090.00, 90.00, 90.00
 Resolution (Å)100.0–3.25 (3.31–3.25)100.0–3.75 (3.81–3.75)100.0–3.50 (3.56–3.50)100.0–4.0 (4.14–4.0)
Rsym or Rmerge10.9 (88.6)14.0 (94.4)8.0 (88.1)16.3 (75.2)
II12.3 (1.2)8.95 (1.1)13.5 (1.2)7.9 (1.3)
 Completeness (%)98.7 (88.1)99.7 (99.8)94.4 (92.7)65.2 (6.5)
 Redundancy5.6 (2.8)3.8 (3.7)3.3 (3.2)3.4 (3.5)
Refinement
 Resolution (Å)15.0–3.2515.0–3.7515.0–3.515.0–4.0
 No. reflections34534215652544611105
Rwork/Rfree22.2/25.823.0/25.726.3/27.825.8/29.6
 No. atoms
 Protein9121911490888985
 Ligand/ion3939
B-factors
 Protein108.5141.8144.2137.2
 Ligand/ion135.3170.8214.1102.3
 R.m.s. deviations
 Bond lengths (Å)0.0100.0130.0050.012
 Bond angles (°)1.6801.8611.1161.407
PDB code4P194P1A4P3J4P6H
GltPh-R397A
ApoNa+-boundNa+/aspartate-bound
Data collection
 Space groupP21P31P31
 Cell dimensions
a, b, c (Å)112.37, 424.42, 113.99110.58, 110.58, 306.92116.96, 116.96, 313.52
 α, β, γ (°)90.00, 119.40, 90.0090.00, 90.00, 120.0090.00, 90.00, 120.00
 Resolution (Å)100.0–4.00 (4.14–4.00)50.0–3.39 (3.51–3.39)100.0–3.50 (3.63–3.50)
Rsym or Rmerge7.8 (62.2)14.0 (>100)8.4 (>100)
II9.3 (1.3)13.8 (1.4)10.6 (0.4)
 Completeness (%)67.9 (13.0)87.3 (12.0)98.1 (96.6)
 Redundancy1.8 (2.0)11.8 (8.6)4.5 (4.2)
Refinement
 Resolution (Å)20.0–4.012.0–3.4115.0–3.50
 No. reflections520684836655613
Rwork/Rfree24.9/26.628.4/29.324.3/26.8
 No. atoms
 Protein352771758018192
 Ligand/ionN/A654/12
 WaterN/A66
B-factors
 Protein139.5152.097.1
 Ligand/ionN/A145.184.7/86.9
 WaterN/A102.6144.6
 R.m.s. deviations
 Bond lengths (Å)0.0100.0100.015
 Bond angles (°)1.3931.4681.735
PDB code4OYE4OYF4OYG
Table 2

Completeness of datasets corrected for anisotropy

https://doi.org/10.7554/eLife.02283.006
Tl+-bound GltPhin (bound conformation)Na+-bound GltPh-R397A
Resolution range (Å)Completeness (%)Resolution range (Å)Completeness (%)
100.0–8.6299.350.00–7.3099.6
8.62–6.8499.97.30–5.79100.0
6.84–5.97100.05.79–5.06100.0
5.97–5.4399.95.06–4.60100.0
5.43–5.0499.94.60–4.27100.0
5.04–4.7469.64.27–4.02100.0
4.74–4.5039.24.02–3.82100.0
4.50–4.3123.63.82–3.6598.6
4.31–4.1414.43.65–3.5163.0
4.14–4.006.53.51–3.3912.0
Apo GltPh-R397A
Resolution range (Å)Completeness (%)
100.0–8.6285.0
8.62–6.8475.6
6.84–5.9775.5
5.97–5.4375.3
5.43–5.0475.2
5.04–4.7475.8
4.74–4.5075.3
4.50–4.3175.4
4.31–4.1451.7
4.14–4.0013.0
Table 3

X-ray crystallographic data and refinement statistics for GltPh-R397A and GltPh-K55C-A364CHg structures not deposited at the PDB

https://doi.org/10.7554/eLife.02283.007
GltPh-R397AGltPhin
Tl+-bound (apo conf.)Tl+/Na+ (apo conf.)Tl+/k+ (apo conf.)
Data collection
 Space groupP21C2221C2221
 Cell dimensions
a, b, c (Å)115.18, 428.53, 116.61108.11, 198.86, 206.34106.59, 198.48, 205.82
 α, β, γ (°)90.00, 119.49, 90.0090.00, 90.00, 90.0090.00, 90.00, 90.00
 Resolution (Å)30.0–5.0 (5.18–5.00)100.0–4.0 (4.07–4.00)100.0–4.15 (4.22–4.15)
Rsym or Rmerge10.9 (>100)15.0 (92.2)13.9 (94.1)
II13.8 (1.9)8.9 (1.5)9.2 (1.5)
 Completeness (%)86.4 (75.1)99.9 (100)94.5 (90.2)
 Redundancy5.5 (5.8)3.9 (3.9)4.0 (3.9)
Refinement
 Resolution (Å)20.0–5.015.0–4.015.0–4.15
 No. reflections347471818415419
Rwork/Rfree22.0/26.528.2/31.728.3/31.2
 No. atoms
 Protein3510791359135
 Ligand/ionN/AN/AN/A
 WaterN/AN/AN/A
B-factors
 Protein223.00183.6194.4
 Ligand/ionN/AN/AN/A
 WaterN/AN/AN/A
 R.m.s. deviations
 Bond lengths (Å)0.0080.0060.008
 Bond angles (°)1.1861.2661.440

Results

Remodeling of the apo transport domain

To determine the structure of apo GltPh, we used R397A mutant that shows a drastically decreased affinity for substrate (Figure 1A). When fully bound, GltPh-R397A crystallizes in the outward-facing state, like wild type GltPh, except that L-asp coordination is slightly altered because the mutant is missing the key coordinating side chain of Arg397 (Figure 1B, Figure 1—figure supplement 2; Bendahan et al., 2000; Boudker et al., 2007). These results suggest that R397A is suitable to capture the apo and ions-only bound outward-facing states for their structural characterization. However, removal of Arg397 may affect local electrostatics, potentially altering ion binding; thus these studies should be interpreted with caution. Apo GltPh-R397A also crystallized in an outward-facing conformation that is similar to the structure reported for a close GltPh homologue (Jensen et al., 2013). To obtain an apo inward-facing state, we used GltPh-K55C-A364C mutant trapped in the inward-facing state upon cross-linking with mercury (Reyes et al., 2009) (GltPhin, Figure 1—figure supplement 1). The positions and orientations of the transport domains relative to the trimerization domains remain essentially unchanged in the apo and fully bound forms of GltPh-R397A and GltPhin (Figure 2). In contrast, the conformations of the transport domains themselves differ significantly. Most remarkably, the apo conformations of the transport domain are nearly identical in the outward- and inward-facing states (Figure 3A, Figure 3—figure supplement 1, Figure 3—figure supplement 2A) and are therefore independent of the transport domain orientations and crystal packing environments.

Figure 1 with 2 supplements see all
Substrate binding to GltPh-R397A.

(A) Raw binding heat rates measured by isothermal titration calorimetry (top) and binding isotherms (bottom) obtained for GltPh-R397A (left) and wild type GltPh (right) at 25°C in the presence of 100 mM NaCl. The solid lines through the data are fits to the independent binding sites model with the following parameters for GltPh-R397A and wild type GltPh, respectively: enthalpy change (ΔH) of −3.2 and −14.3 kcal/mol; the apparent number of binding sites (n) of 0.8 and 0.7 per monomer; dissociation constant (Kd) of 6.6 µM and 27 nM. Note that L-asp binding to the wild type transporter is too tight at 100 mM NaCl to be accurately measured in this experiment. The binding Kd has been estimated to be ∼1 nM (Boudker et al., 2007). (B) L-asp binding site in GltPh-R397A (left) and wild type GltPh (right). L-asp and residues coordinating the side chain carboxylate are shown as sticks with carbon atoms colored light brown and blue, respectively. Potential hydrogen bonds (distances less than 3.5 Å) between the L-asp side chain carboxylate and transporter residues are shown as dashed lines. Note that Y317, which forms cation-π interactions with guanidium group of R397 in wild type GltPh, interacts directly with L-asp in GltPh-R397A.

https://doi.org/10.7554/eLife.02283.008
Apo protomer structures.

(A) GltPh protomers in the outward-facing state (left) and a GltPhin protomer (right) viewed from within the plane of the membrane. Shown are superimpositions between apo (colors) and fully bound protomers (grey).

https://doi.org/10.7554/eLife.02283.011
Figure 3 with 2 supplements see all
Structures of the apo transport domain.

(A) Superimposition of the nearly identical apo transport domains in the outward- and inward-facing states. HP1, HP2, and NMD motif are colored yellow, red, and green, respectively. The remainder of the domain is blue. (B) Superimposition of the fully bound (light colors, PDB accession number 2NWX) and apo GltPhin (dark colors) transport domains. (C) The NMD motif and adjacent TM3. Met311 is shown as sticks, and the light blue spheres indicate the Cα positions for T92 and S93. (D) The HP2-TM8a structural modules in the fully bound (pink) and apo (red) transport domains superimposed on TM8a and HP2b to emphasize the re-orientation of the HP2a. (E) The Na+ and L-asp binding sites in the fully bound (left) and apo forms (right).

https://doi.org/10.7554/eLife.02283.012

The conformational differences between fully bound and apo forms of the transport domain include a concerted movement of HP2 and TM8a, which form the extracellular surface of the domain, and local rearrangements at the ligand binding sites, involving HP2, the NMD motif and TM3 (Figure 3B–E, Figure 3—figure supplement 2B, Figure 4). In HP2, the last helical turn of HP2a unwinds, and HP2a together with the loop region at HP2 tip collapse into the substrate and Na2 binding sites. Within the NMD motif, the side chain of Asn310 rotates away from TM3 and partially fills the empty Na1 site, while the side chain of Met311 undergoes an opposite movement, flipping away from the binding sites (Figure 4). Finally, TM3 bends away from the NMD motif, particularly around Thr92 and Ser93 (Figure 3B,C). Notably, these residues together with the side chain of Asn310 form one of the proposed third Na+-binding sites (Huang and Tajkhorshid, 2010; Bastug et al., 2012). Thus, all known ligand-binding sites are distorted in the apo forms (Figure 4).

Figure 4 with 1 supplement see all
Remodeling of L-asp and Na+ binding sites in the apo conformations.

Close-up views of the fully bound (left) and apo (right) transport domains at L-asp binding site (top), Na1 and Na2 sites (middle), and one of the proposed locations for the third Na+ binding site (Huang and Tajkhorshid, 2010; Bastug et al., 2012) (dashed circle).

https://doi.org/10.7554/eLife.02283.003

The overall structures of the apo transport domain remain as closed and compacted as in the fully bound forms (Figure 4—figure supplement 1). Therefore, we propose that the unloaded transport domains traverse the membrane as rigid bodies as deduced previously for the fully loaded transport domains (Reyes et al., 2009).

Insight into the coupling mechanism

In GltPh, cooperative binding of Na+ ions and L-asp is central to tightly coupled transport of the solutes (Reyes et al., 2013). Our structures of the apo and fully bound GltPh suggest that binding of L-asp and Na+ at the Na2 site is coupled because the same structural element, the tip of HP2, contributes to both sites and is restructured upon binding. Thus, structural changes in HP2 upon binding of either L-asp or Na+ ion should greatly favor binding of the other.

Met311 in the NMD motif is the only residue that is shared between the Na1 site and the substrate and Na2 sites and also undergoes a conformational change upon ligand binding. To examine whether the structural changes in HP2 upon binding of L-asp and Na+ at the Na2 site could occur independently from those in the NMD motif upon Na+ binding at the Na1 site, we modeled transport domains with HP2 in the bound conformation and the NMD motif in the apo conformation, or vice versa (Figure 5A). In both models, the side chain of Met311 clashes with residues in HP2, suggesting that the conformational changes in HP2 and the NMD motif must be concerted.

Met311 is key to the allosteric coupling.

(A) Structural models combining HP2 bound to L-asp and Na+ at Na2 site with apo conformation of the NMD motif (left), and apo conformation of HP2 with the NMD motif bound to Na+ at Na1 site (right). Met311 and clashing residues in HP2 are shown as sticks and transparent spheres. (B) The dependence of L-asp dissociation constant, Kd, on Na+ activity plotted on a log–log scale for mutants within the context of GltPhin (left) and unconstrained GltPh (right). The data were fitted to straight lines with slopes shown on the graph or to arbitrary lines for clarity. Dashed lines and corresponding slopes correspond to published dependences for GltPhin and GltPh (Reyes et al., 2013).

https://doi.org/10.7554/eLife.02283.015

We then mutated bulky Met311 to either another bulky residue, leucine, or to a smaller residue, alanine, which is not expected to experience similar clashes. For these mutants, generated in the context of unconstrained wild type GltPh and inward cross-linked GltPhin, we measured the dependence of L-asp dissociation constant on Na+ concentration (Figure 5B). While this dependence is very steep for the wild type GltPh constructs (Reyes et al., 2013) and nearly as steep for the M311L mutants, it is substantially shallower for the M311A mutants. The most parsimonious interpretation of these results is that M311A mutation reduces binding cooperativity between the substrate and Na+ ions. However, it is also possible, though we think unlikely, that the mutation abrogates ion binding at one or more Na+-binding sites in the tested concentration range (1–100 mM). Mutating the equivalent methionine to smaller residues in EAAT3 also resulted in less steep dependence of the ionic currents on Na+ concentration (Rosental and Kanner, 2010). Based on these results, we hypothesize that Met311 is key to the allosteric coupling between the Na1, L-asp and Na2 sites. Consistently, bulky methionine or leucine residues are found at this position in ∼85% of glutamate transporter homologues. However, it should be noted that methionine is conserved in the Na+-coupled GltPh and EAATs, while a characterized proton-coupled homologue has leucine at this position (Gaillard et al., 1996). Hence, it is possible that the methionine thioether, which is proximal to both Na1 and Na2 sites, plays a direct role in Na+ binding.

Our hypothesis further predicts that binding of an ion at Na1 site should prime the transporter to accept its substrate. Therefore, we crystallized GltPh-R397A in the presence of 400 mM Na+, but in the absence of L-asp. We also soaked crystals of apo GltPhin in Tl+, an ion with strong anomalous signal that seems to mimic some aspects of Na+ in GltPh and EAATs (Boudker et al., 2007; Tao et al., 2008). The obtained outward- and inward-facing structures pictured the transport domains in conformations overall similar to those observed in the fully bound transporter: straightened TM3, Met311 pointing toward the binding sites, extended helix in HP2a and HP2 tip raised out of the substrate binding site (Figure 6A–D). Indeed, the structure of Tl+-bound GltPhin is indistinguishable from the fully bound GltPhin and both Na1 and Na2 sites are occupied by Tl+ ions (Figure 6A). The structure of Na+-bound GltPh-R397A differs significantly from the fully bound GltPh-R397A only at the tip of HP2 (Figure 3—figure supplement 2, also see below). The coordinating residues at the Na1 site are correctly positioned and the site is likely occupied by a Na+ ion. The Na2 site still shows a distorted geometry: the last helical turn of HP2a points away from the site due to the altered conformation of the tip of HP2 (Figure 6C). Collectively, our results demonstrate that binding of the coupled ions, notably at the Na1 site, is sufficient to trigger isomerization of the transport domain from the apo conformation to the bound-like conformation. The energetic penalty associated with this isomerization likely explains why Na+ ions alone bind weakly to the transporter (Reyes et al., 2013). This experimental observation contrasts with highly favorable calculated binding energies (approximately −10 kcal/mol for Na1) that were obtained using fully bound protein conformation and where the reference ion-free state is the same as the bound state (Larsson et al., 2010; Bastug et al., 2012; Heinzelmann et al., 2013).

Figure 6 with 3 supplements see all
Structures of ions-only bound transport domain.

(A) Superimposition of the fully bound transport domains (grey) and Tl+-bound GltPhin transport domain in the bound-like conformation (colors), with the averaged anomalous difference Fourier map contoured at 8σ (cyan mesh). (B) Superimposition of the fully bound (grey) and Na+-only bound GltPh-R397A (colors) transport domains. (C) Na+ and L-asp binding sites with fully-bound structure shown in white and Na+-bound structure in colors. Hinge glycine residues are shown as spheres. The modeled Na+ ion in Na1 site is pink. (D) Superimposition of the HP2-TM8 in the fully bound transport domain (grey) and in GltPh-R397A bound to Na+ only (colors), showing similar conformations of HP2a. (E) WebLogo representation of the consensus sequence and relative abundance of residues in HP2 tip. (F) Surface representation of the transport domain of GltPh-R397A bound to Na+ only showing access to the substrate-binding site. L-asp was placed into the binding site for reference.

https://doi.org/10.7554/eLife.02283.016

Na+-mediated gating in the outward-facing state

The structure of the Na+-only bound GltPh-R397A shows HP2 in a conformation overall similar to that observed in the fully bound transporter, but with an opened tip (Figure 6B–D,F, Figure 6—figure supplement 1). This opening is smaller than the opening observed previously in the structure of GltPh in complex with the blocker L-threo-β-benzyloxyaspartate (Figure 6—figure supplement 2; Boudker et al., 2007), and it is hinged at two well-conserved glycine residues at positions 351 and 357 (Figure 6C). Interestingly, among the nine amino acids forming the tip in GltPh (residues 351 to 359), five are glycines in the consensus sequence generated for the glutamate transporter family, although not all are present in each homologue (Figure 6E, Figure 6—figure supplement 3). We suggest that the glycines support the structural flexibility of the HP2 tip in all members of the family, but that the structural specifics of the tip opening may vary among homologues.

To test whether the trans-membrane movement of the transport domain is possible when the tip of HP2 is opened, we modeled the open tip conformation in the context of the previously reported early transition intermediate structure (Figure 7; Verdon and Boudker, 2012). In this structure, the transport domain tilts towards the trimerization domain but does not yet undergo a significant translation toward the cytoplasm. We find that such intermediate state with the opened tip of HP2 can be achieved without major steric clashes, while further progression of the transport domain to the inward-facing position could be impeded because the tip is likely to clash with TM5 in the trimerization domain (Figure 7B). Also in the inward-facing state HP2 is packed against the trimerization domain and cannot open in a manner observed in the outward-facing state. Consistently, HP2 is closed in GltPhin bound to Tl+ (Figure 6A).

Modeled Na+-bound early transition intermediate between the outward- and inward-facing states.

(A) Surface representations of the protomer in the fully bound intermediate state (PDB code 3V8G) (left), and the modeled Na+-bound intermediate with an open HP2 tip (right) viewed from the extracellular space (top). The model reveals no clashes, suggesting that the observed opening of HP2 is structurally compatible with the intermediate orientation of the transport domain. The arrows indicate the point of access to the domain interface with potentially increased solvent accessibility. (B) Side views of thin cross-sections of the closed fully bound (left) and open Na+-only bound (right) intermediate state. The protomers are sliced normal to the membrane plane, as indicated by the dashed lines in A.

https://doi.org/10.7554/eLife.02283.020

Therefore, opening of the HP2 tip upon Na+ binding in the outward-facing state may serve as a structural mechanism preventing uncoupled uptake of Na+ ions. We suggest that the structural changes in the NMD motif and HP2 that are triggered upon Na+ binding at the Na1 site may lead to the loss of direct interactions between the tip of HP2 and the rest of the transport domain, resulting in tip opening. Subsequent binding of L-asp and Na+ at the Na2 site is then required to provide compensatory interactions, allowing HP2 tip to close. Similar conformational behavior has been observed for transporters with the LeuT fold: when bound to Na+ ions only, substrate binding sites are open to the extracellular solution, and substrate binding is required for occlusion (Weyand et al., 2008; Krishnamurthy and Gouaux, 2012).

We do not see a transition into an open conformation in the inward-facing GltPhin bound to Tl+ ions (Figure 6A). This may be because Tl+ ions do not faithfully mimic Na+ ions and fail to induce an open state or it may be because Na+ bound inward-facing state is, indeed, closed. This latter possibility does not contradict the requirements of symport because the measured dissociation constant for Na+ ions in the inward-facing state (250 mM) (Reyes et al., 2013), is far above Na+ concentration in the cytoplasm (10 mM) and therefore, Na+-bound inward-facing state is not expected to be significantly populated.

We and others have proposed that transition intermediates mediate fluxes of polar solutes, including anions, because potentially hydrated cavities form in such intermediates at the interface between the trimerization and transport domains (Stolzenberg et al., 2012; Verdon and Boudker, 2012; Li et al., 2013). Interestingly, because the tip of HP2 forms part of this interface in the fully bound intermediate state of the transporter, opening of the tip in the Na+-only bound form may increase solvent accessibility to the interface (Figure 7A).

New cation-binding sites

While soaking apo GltPhin crystals in Tl+ solutions, we observed that only in approximately one third of crystals Tl+ ions bound to the Na1 and Na2 sites, inducing transition from apo- to bound-like conformation as described above. In the majority of the crystals, we observed no conformational changes of the transport domain and Tl+ ions incorporated at two previously uncharacterized sites (Figure 8), within the small cavities that remain under the collapsed HP2 (Figure 8—figure supplement 1). One site, termed Na2', involves residues of HP2 and TM7a that form the Na2 site, but in a different ion coordinating geometry due to the conformational difference in HP2 (Figure 8B). The second site, termed Ct, overlaps with the L-asp binding site and is formed by the side chains of highly conserved Asp394 and Thr398 in TM8 and main chain carbonyl oxygen atoms of HP1 and HP2 (Figure 8B,C). Tl+ soaks of the outward-facing apo GltPh-R397A also showed no conformational changes of the transport domain, with Tl+ binding at the Ct site, but not at the Na2' site (Figure 8A). The ion selectivity of the Ct site remains ambiguous, because neither 300 mM K+ nor 10 mM Na+ efficiently inhibited incorporation of Tl+ (150 mM) at this site in GltPhin (Figure 8—figure supplement 2). Crystals deteriorated at higher Na+ concentrations. In contrast, the Na2' site seems to show a preference for Na+, which even at low concentration (10 mM) interfered significantly with Tl+ binding.

Figure 8 with 2 supplements see all
New cation binding sites.

(A) Superimpositions of GltPhin (left) and outward-facing GltPh-R397A (right) transport domains in the apo form (grey) with Tl+-bound apo-like conformations (colors). Averaged anomalous difference Fourier maps are contoured at 8σ (cyan mesh). (B) Modeled Tl+ ions bound to the Ct and Na2’ sites (left), and L-asp aspartate and Na+ ions bound to the Na1 and Na2 sites in the fully bound transport domain (right). (C) Close-up view of Tl+ in the Ct site of apo-like GltPhin and L-asp in the fully bound transporter.

https://doi.org/10.7554/eLife.02283.021

The functional relevance of these sites is speculative at present. However, it is remarkable that the Ct site is positioned exactly at the same place as the amino group of the bound L-asp and share several coordinating moieties. Therefore, binding of a cation at the Ct site and binding of the substrate are mutually exclusive. Because the Ct site is observed only in the apo-like conformation, cation binding at this site would also inhibit the transition into the bound-like conformation upon Na+ binding at the Na1 site. Finally, the Ct site is observed in both the inward- and outward-facing states, suggesting that the apo-like transport domain could carry the ion across the membrane. These are the exact properties expected for the K+-binding site in EAATs. Moreover, it has been previously proposed that K+ binds to EAATs at a similar position (Holley and Kavanaugh, 2009). Most remarkably, in an insect K+-independent dicarboxylate transporter, an asparagine to aspartate mutation at the position equivalent to Asp394 in GltPh changes the transporter substrate specificity to amino acid glutamate, and also leads to dependence on K+ counter-transport (Wang et al., 2013). Therefore this aspartate plays a key role in both binding the amino group of substrate and coupling to K+ counter-transport. Consistently, Asp394 in GltPh coordinates both the amino group of the bound substrate and Tl+ in the Ct site. Notably, while Tl+ mimics, to some extent, Na+ ions in GltPh and EAATs, it is a better mimic of K+ ions in EAATs (Boudker et al., 2007; Tao et al., 2008).

Movement of HP1 in the inward-facing state

To examine whether a complete removal of Na+ and K+ ions had an effect on the structure of GltPhin, we soaked apo GltPhin crystals (typically grown in the presence of K+) in alkali-free buffer. Interestingly, we observed a small, but reproducible structural change in several crystals examined: HP1 and TM7a that form the transport domain cytoplasmic surface moved slightly towards TM8, with the tip of HP1 detaching from that of HP2 (Figure 9, Figure 9—figure supplement 1). This movement is observed clearly in one protomer (chain B in 4P3J), in which these helices are not involved in crystal packing contacts. It is reminiscent of the isomerization of the structurally symmetric HP2 and TM8a on the extracellular side of the domain observed upon the transition from bound to apo forms (Figure 9B). It was suggested previously that HP1 participates in intracellular gating in GltPh (Reyes et al., 2009; DeChancie et al., 2010). Indeed, the observed movement of HP1 generates a small opening, leading to the substrate and Ct sites (Figure 9—figure supplement 2), and it is reminiscent of the movement observed in molecular dynamics simulations (DeChancie et al., 2010; Zomot and Bahar, 2013). However, this conformational difference is too small to be interpreted unambiguously.

Figure 9 with 2 supplements see all
Movements of the HP1-TM7a structural module.

(A) Superimposition of GltPhin transport domains when bound to Tl+ in the apo-like conformation (grey) and when prepared in an alkali-free solution (colors). (B) The transport domains of the fully bound GltPh (light colors) and alkali-free GltPhin (dark colors) superimposed on TM6. Arrows indicate movements of the structurally symmetric HP1-TM7a and HP2-TM8a modules. (C) Per residue main chain R. M. S. D. values calculated for the structures of the inward-facing transport domains bound to Tl+ in apo-like conformation and alkali-free shown in A and superimposed on HP2. The bars above the plot represent secondary structure elements colored as in A.

https://doi.org/10.7554/eLife.02283.024

Discussion

Our apo and ions-only bound structures reveal a remarkable structural plasticity of GltPh transport domain that is likely a conserved feature in the glutamate transporter family. In addition to the large trans-membrane rigid-body movements of the transport domain between outward- and inward-facing orientations, local conformational changes within the domain accompany binding and release of the transported substrate and ions (Figure 10). These local changes provide a structural explanation of how Na+ gradients are harnessed to drive concentrative substrate uptake, supporting two previously proposed coupling mechanisms (Focke et al., 2011; Reyes et al., 2013): allosterically coupled binding of the substrate and symported Na+ ions, and opening of HP2 upon Na+ binding, which impedes the inward trans-membrane movement of the Na+-only bound transport domain.

Proposed transport cycle for GltPh and EAATs.

Ion binding to the Na1 site of the outward-facing apo transport domain triggers isomerization into bound-like conformation, formation of the L-asp and Na2 binding sites and HP2 opening, impeding translocation of the domain. Closure of HP2, coupled to L-asp and Na2 binding, allows translocation. After the release of the ligands into the cytoplasm by as yet an unknown gating mechanism, the domain is in a compact apo state, and returns to the extracellular side. Notably, binding of cations to the inward-facing state does not lead to a crystallographically observed gate opening that would impede translocation. However, Na+ affinity in this state is only ∼250 mM (Reyes et al., 2013), and it will remain largely unbound when facing the cytoplasm. Hence, uncoupled Na+ transport should be limited. In EAATs, an open conformation of the gates might be more favored in the apo state, and K+ binding at the Ct site might be required to stabilize translocation-competent conformation of the apo transport domain.

https://doi.org/10.7554/eLife.02283.027

The apo transport domains in the outward- and inward-facing states are essentially identical and as compact as when they are fully bound, consistent with previous spectroscopic experiments (Focke et al., 2011). Therefore, the apo transport domain is likely able to transition readily between the cytoplasmic and extracellular orientations. Consistently, previous spectroscopic studies showed that the transport domains continuously sample the outward- and inward-facing positions with nearly equal probabilities either when bound to Na+ and L-asp or when free of the solutes (Akyuz et al., 2013; Erkens et al., 2013; Georgieva et al., 2013; Hanelt et al., 2013). Moreover, these transitions are more frequent in the apo transporter, consistent with a lack of large energetic barriers (Akyuz et al., 2013). In GltPh, the compact translocation-competent apo conformation of the transport domain is stabilized by interactions between the collapsed HP2, and HP1, TM7, and TM8. In EAATs, by contrast, we speculate that these interactions are insufficient and that K+ binding to the Ct site is required to stabilize the translocation-competent closed conformation that can return to the outside, ensuring coupling between substrate uptake and counter-transport of K+ ion. Local structural differences in EAATs in the vicinity of the Ct site may underlie the higher affinity and specificity of this site for K+ ion.

In conclusion, we have shown structurally that ion binding and unbinding events in GltPh and, by analogy, in EAATs control the conformational state of the transporter, determining its competence to bind substrate and undergo transitions between the outward- and inward-facing states. Studies establishing the location of the Na3 binding site; the potential role of the Ct site in binding K+; and the gating mechanism in the inward-facing state will be necessary to verify and refine our proposed mechanisms.

Materials and methods

DNA constructs, mutagenesis, protein expression, and purification

Request a detailed protocol

R397A mutation was introduced by PCR into GltPh containing seven point mutations to histidine (Yernool et al., 2004), referred as wild type GltPh for brevity. Proteins were produced in Escherichia coli DH10b strain (Invitrogen, Inc., Grand Island, NY) as fusions with a thrombin cleavage site, and a metal-affinity octa-histidine at their carboxyl-terminus. Proteins were purified by nickel-affinity chromatography, digested with thrombin to remove the affinity tag, and purified by size exclusion chromatography (SEC) in appropriate buffers as described below. Protein concentrations were determined by measuring the absorbance at 280 nm using an extinction coefficient of 26,820 M−1.cm−1.

Crystallization, and soaking experiments

Outward-facing state

Request a detailed protocol

GltPh-R397A was purified by SEC in 10 mM HEPES/Tris, pH 7.4, 200 mM choline chloride and 7 mM n-decyl-α-D-maltopyranoside (Anatrace, inc., Maumee, OH). Crystallization experiments were setup using the hanging-drop vapor diffusion method, by mixing protein (∼7 mg/ml) and well solutions (1:1 vol:vol), and incubated at 4°C. Na+-bound GltPh-R397A crystals were grown in 18–20% PEG 400, 0.1 M citric acid/Tris pH 4.5 and 0.4 M NaCl. Fully bound GltPh-R397A crystals were obtained in the same crystallization conditions supplemented with 5 mM L-asp. The crystals were cryo-protected by soaking in the well solution supplemented with 10% glycerol and 7 mM n-decyl-α-D-maltopyranoside and frozen in liquid nitrogen. Apo crystals were grown in 18–20% PEG 400, 0.1 M citric acid/Tris pH 4.5, and 0.4 M choline chloride. Tl+-bound GltPh-R397A crystals were obtained by soaking GltPh-R397A apo crystals in several changes of solutions containing 18–20% PEG 400, 0.1 M citric acid/Tris pH 4.5, 0.1 M TlNO3, 7 mM n-decyl-α-D-maltopyranoside and 10% glycerol.

Inward-facing state

Request a detailed protocol

GltPh-K55C-A364C was purified by SEC in 10 mM HEPES/KOH, pH 7.4, 150 mM KCl, and 7 mM n-decyl-α-D-maltopyranoside. Cross-linking was carried out by mixing protein at ∼3 mg/ml with a 10-fold molar excess of HgCl2 for 20 min on ice. The samples were diluted ∼10-fold with the SEC buffer and concentrated to remove partially the excess of HgCl2. For crystallization, the protein samples at 2.75 mg/ml were supplemented with 20–50 mM MgCl2 and 0.2 to 0.5 mM of a mixture of E. coli polar lipid extract and egg PC (3:1 wt:wt) (Avanti Polar Lipids, Inc., Alabaster, AL) and incubated on ice for 45 min. Crystallization was carried out using hanging drop diffusion methods in 96-well plates at 4°C. Initial crystallization conditions were identified using a replica of the crystallization screen MemGold (Molecular Dimensions., Altamonte Springs, FL), in which Na+-containing compounds were replaced with K+-containing compounds. The screen was prepared using the liquid handler Formulator (Formulatrix, Inc., Waltham, MA). GltPhin crystals were grown in 14–20% PEG 400, and 0.1 M potassium citrate, pH 5.0 to 6.0. For Tl+ soaking, crystals were washed in 20% PEG 400, 20 mM MES/Tris, pH 6.5, 20 mM KNO3, 20 mM MgNO3, and 10 mM n-decyl-α-D-maltopyranoside, and then incubated in 20% PEG 400, 20 mM MES/Tris, pH 6.5, 5 mM MgNO3, 10 mM n-decyl-α-D-maltopyranoside, and 150 mM TlNO3. In ion competition experiments, the soaking solution was further supplemented with either NaNO3 or KNO3. To obtain an alkali-free structure, crystals were soaked in solution containing 20% PEG 400, 20 mM MES/Tris, pH 6.5, 5 mM MgCl2, and 10 mM n-decyl-α-D-maltopyranoside. Crystals were directly frozen in liquid nitrogen.

Data collection and structure determination

Request a detailed protocol

Diffraction data were collected at the National Synchrotron Light Source beamlines X25 and X29 (Brookhaven National Laboratory). Data from crystals soaked in Tl+ were collected at a wavelength of 0.97 Å. Data were processed using HKL2000 (Otwinowski and Minor, 1997), and further analyzed using the CCP4 program suite (Collaborative Computational Project, 1994). Anisotropy correction was performed as described previously (Strong et al., 2006). Briefly, resolution limits along the a, b, and c axes were determined using the UCLA–MBI Diffraction Anisotropy server (http://services.mbi.ucla.edu/anisoscale/) and applied as cutoffs to truncate the dataset obtained after processing of diffraction images. After scaling in HKL2000, structure factors were anisotropically scaled using PHASER (McCoy et al., 2007), and a negative B factor correction was applied to these structure factors using CAD. Initial phases were determined by molecular replacement with PHASER (McCoy et al., 2007), using the structure of GltPh either in the outward-facing state (PDB code 2NWX) or the inward-facing state (PDB code 3KBC) as the search model. Refinement was carried out by rounds of manual model building in COOT (Emsley and Cowtan, 2004) and refinement in REFMAC5 with TLS (Winn et al., 2001). With the exception of the analysis of the data from alkali-free GltPhin crystals, where protomers in the trimer were clearly not identical, the electron density maps and the anomalous difference Fourier maps were three or sixfold averaged in real space. Strict non-crystallographic symmetry constrains were also applied during structural refinement in REFMAC5 when necessary. Structures of the transport domain were superimposed and R.M.S.D.s calculated using VMD software (Humphrey et al., 1996). All structural figures were prepared using Pymol (DeLano Scientific, LLC) (DeLano, 2008).

Isothermal titration calorimetry (ITC)

Request a detailed protocol

ITC experiments were performed as described previously (Reyes et al., 2013). Briefly, GltPh mutant proteins were purified by SEC in 10 mM HEPES/Tris, pH 7.4, 200 mM choline chloride, 0.5 mM n-dodecyl-β-D-maltopyranoside and concentrated to 4 mg/ml. The protein was diluted to 40 μM in buffer containing 20 mM HEPES/Tris, pH 7.4, 200 mM choline chloride, 1 mM n-dodecyl-β-D-maltopyranoside and various NaCl concentrations. ITC experiments were performed using a small cell NANO ITC (TA instruments, Inc., New Castle, DE) at 25°C. Protein samples were placed into the instrument cell and titrated with L-asp solution prepared in the same buffer. The isotherms were analyzed using the NanoAnalyze software (TA instruments, Inc., New Castle, DE), and fitted to independent binding sites model.

Fluorescence-based binding assays

Request a detailed protocol

Fluorescence-based binding assays were performed as described previously (Reyes et al., 2013). In brief, 100 μg/ml of protein in 20 mM HEPES/Tris, pH 7.4, 200 mM choline chloride, 0.4 mM n-dodecyl-β-D-maltopyranoside, 200 nM styryl fluorescent dye RH421 (Invitrogen, Inc., Grand Island, NY) were titrated with L-asp in the presence of various concentrations of NaCl at 25°C. Fluorescence experiments were carried out using a QuantaMaster (Photon International Technology, Inc., Edison, NJ). The RH421 dye was excited at 532 nm, and the fluorescence was collected at 628 nm. Fluorescence emissions were measured after at least 1000 s equilibration. The data were analyzed using SigmaPlot12 (Systat software, Inc., San Jose, CA). Fractional fluorescence changes were corrected and normalized with respect to the dilution factors and maximal fluorescence changes, respectively. Corrected fluorescence changes were plotted as a function of ligand concentration and fitted to the Hill equation. Sodium activity was calculated as γ × [Na+], where γ is the activity coefficient. The activity coefficient is calculated with the Debye-Hückel equation as described (Reyes et al., 2013). All the experiments were performed at least in triplicate.

Sequence analysis

Request a detailed protocol

Sodium:dicarboxylate symporter family sequences were harvested from PFAM database (PF00375) (Finn et al., 2008), parsed to remove incomplete sequences and sequences with over 70% identity and aligned in ClustalW (Larkin et al., 2007). The alignment was manually adjusted and the final dataset containing 463 aligned sequences was used to generate a consensus sequence using WebLogo (Crooks et al., 2004).

Molecular modeling

Request a detailed protocol

To model the structures of the transport domains with HP2 in the bound conformation and the NMD motif in the apo conformation, and vice versa, we superimposed the structures of the fully bound and apo forms of the transport domains using TM6, HP1, and TM7. We then generated new coordinates files combining the coordinates of TM7, including the NMD motif, from the bound form and the coordinates for HP2 from the apo form or vice versa. In both of these models, we observed steric clashes between Met311 and residues in HP2. To construct a model of the intermediate state with an open tip of HP2, we superimposed the structure of Na+-only bound GltPh-R397A and the intermediate state (PDB accession code 3V8G) using TM6, HP1, and TM7. We then replaced HP2 in the structure of the intermediate with HP2 from Na+-only bound GltPh-R397A. We moved slightly the side chain of Lys55 that was involved in a minor steric clash with the HP2 tip. We observed no major steric clashes in the resulting model.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
    The gradient hypothesis and other models of carrier-mediated active transport
    1. RK Crane
    (1977)
    Reviews of Physiology, Biochemistry and Pharmacology 78:99–159.
  7. 7
  8. 8
  9. 9
  10. 10
    The PyMOL molecular graphics System
    1. WL DeLano
    (2008)
    Palo Alto, CA, USA: DeLano Scientific LLC.
  11. 11
    Coot: model-building tools for molecular graphics
    1. P Emsley
    2. K Cowtan
    (2004)
    Acta Crystallographica Section D, Biological Crystallography D60:2126–2132.
    https://doi.org/10.1107/S0907444904019158
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
    Conformational heterogeneity of the aspartate transporter Glt(Ph)
    1. I Hanelt
    2. D Wunnicke
    3. E Bordignon
    4. HJ Steinhoff
    5. DJ Slotboom
    (2013)
    Nature Structural & Molecular Biology 20:210–214.
  20. 20
  21. 21
    Interactions of alkali cations with glutamate transporters
    1. DC Holley
    2. MP Kavanaugh
    (2009)
    Philosophical Transactions of the Royal Society of London Series B, Biological sciences 364:155–161.
    https://doi.org/10.1098/rstb.2008.0246
  22. 22
  23. 23
  24. 24
    VMD: visual molecular dynamics
    1. W Humphrey
    2. A Dalke
    3. K Schulten
    (1996)
    33–38, Journal of Molecular Graphics, 14, 27–28, 10.1016/0263-7855(96)00018-5.
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
    Stoichiometry of the glial glutamate transporter GLT-1 expressed inducibly in a Chinese hamster ovary cell line selected for low endogenous Na+-dependent glutamate uptake
    1. LM Levy
    2. O Warr
    3. D Attwell
    (1998)
    The Journal of Neuroscience 18:9620–9628.
  31. 31
    Transient formation of water-conducting states in membrane transporters
    1. J Li
    2. SA Shaikh
    3. G Enkavi
    4. PC Wen
    5. Z Huang
    6. E Tajkhorshid
    (2013)
    Proceedings of the National Academy of Sciences of the United States of America 110:7696–7701.
    https://doi.org/10.1073/pnas.1218986110
  32. 32
  33. 33
    Processing of X-ray diffraction data collected in oscillation mode
    1. Z Otwinowski
    2. W Minor
    (1997)
    Methods in Enzymology 276:308–326.
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55

Decision letter

  1. Benoit Roux
    Reviewing Editor; University of Chicago, United States

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

Thank you for sending your work entitled “Coupled ion binding and structural transitions along the transport cycle of glutamate transporters” for consideration at eLife. Your article has been very favorably evaluated by a Senior editor, John Kuriyan, a Reviewing editor, and 3 reviewers, one of whom, Lucy Forrest, has agreed to reveal her identity.

The Reviewing editor and the reviewers discussed their comments before we reached this decision, and the Senior editor has assembled the following comments to help you prepare a revised submission.

Boudker and colleagues present a manuscript describing structures and binding data for the prokaryotic asparate transporter GltPh, an important reference system for the neuronal glutamate transporters of the EAAT family. The manuscript contains an array of important data, including substrate-free states, and sodium-only states, that together provide crucial insights into ion-substrate coupling and transporter conformational dynamics. The several crystal structures for ion-bound apo state allow for the first time a dissection of the exact mechanism of coupling in this system. The findings are in line with previous work and the proposed mechanisms of translocation of the apo states and the Na coupling makes perfect sense. The discussion on a tentative location of K+ binding site, while offering a plausible explanation (and a novel one) remains a little speculative. However, the authors are careful in the interpretation of the K sites and these sites might be the ones used in EAATs. Nevertheless, this is novel and rigorous study of a high caliber! One of the reviewers describes the paper as “a great manuscript”.

The following comments are collated from the individual reviews. The authors should pay attention to them when submitting a revised manuscript, and provide an itemized response. The revised manuscript will be dealt with by the editors, and will not be sent out for further external review, so you should expect a quick final decision.

Specific comments:

Reviewer 1:

1) Introduction section; not clear what “reflect weaker coupling” refers to. The slope only tells something about how many Na could be involved in the conformational changes, not anything about coupling. Maybe it would be good to clearly state what the measurement really mean. A Hill coefficient of 1 doesn't mean that only 1 ion binds. Is this the same case here?

2) The off rate of aspartate is dependent on all conformational changes in the cycle and the coupling could be very indirect by Na affecting another step in the cycle. This should be stated.

3) Not clear what is meant by “trigger isomerization from apo... to bound ... conformation”; Isomerization of what?

4) It is also not clear why the isomerization would make the Na affinity low. Previous molecular modelling has proposed that Na binds with low affinity to the GltPh sites (without any isomerization needed to explain this low affinity).

5) Are the two Tl+ sites occupied at the same time or in different crystals?

6) Figure 10. It is not clear from the model figure how the substrate and ions come of the transporter in the inward state?

7) In the Introduction, it is stated that Na1 binding form the Na2 and asp binding sites. However, this is confusing, since also Na1 binding opens the Hp2 gate which should prevent Na2 binding. Please state carefully what Na1 does.

8) Figure 6e is not very legible.

9) Previous experiments and modeling have suggested that the apo state returns by adopting a similar conformation as the fully loaded transporter. The present crystals fully support this.

Reviewer 2:

1) General remark. R/A mutation used in this paper may affect overall electrostatic of the pocket and thus bias packing of ions. Some cautious remarks should be put in the manuscript.

2) While describing replacement of M311 with Leu the authors states: “similarly bulky” for L311 replacement. I am unsure if it is exactly correct statement. Furthermore, M311 was assigned a variety of roles in the past including tentative participation in stabilization of Tl+ in the binding pocket due to high polarizability. Any comments/discussions to that end would be beneficial. Tl+ marking of Na1/Na2 site remains a little bit of a mystery. Thus, it is not only volume but also chemistry of a side-chain that may be important.

3) The hypothesis about ion “priming” the site for substrate acceptance is interesting. Zhao et al. (PLoS Comput Biol 9(10): e1003296) described recently a fairly similar mechanism for MhP1 where transporter appears to get “primed” for substrate by binding of an ion stabilizing one of the states. Some discussion on this seemingly general mechanism, which may exist in other systems may benefit this manuscript. It is entirely up to the authors, but again it may represent a general mechanism of allosteric coupling during secondary transport.

4) Furthermore, the idea of shuttling of a “compact” state of this transporter is interesting and well supported by two side-by-side publications in Nat Stuct Biology, but still some additional discussion on possible energetics of this process would be advantageous! The large protein has to move a great length across the membrane plain.

Reviewer 3:

1) My only substantive query is why several of the structures have not been deposited in the PDB. These are all structures at >4 Angstrom resolution, but I'm not sure that is sufficient reason; this may come down to journal policy.

2) I didn't follow the arguments regarding the evidence for priming of the transporter by Na1. For example, the Na2 site is distorted in GltPh-R397A+Na, so its structure is not really similar to the bound state. Maybe it would help to separate out the figures of the outward-open R397A+Na from those of the inward-facing Tl-bound. Or perhaps the phrasing just needs a little reworking.

3) In the Introduction section I was confused by these statements: “Met311 in the NMD motif is the only residue that is shared between the Na1 site and the substrate and Na2 sites and also undergoes a conformational change upon ligand binding.” Which are the reference states in each case? Are the authors only comparing the trimerization domains?

4) I would welcome quantitation of the similarity between domains in the different states, using RMSDs of the backbone and of all atoms. It would put the differences in context relative to the crystallographic resolution.

5) In Figure 4–figure supplement 1, it is not yet explained what is meant by the Ct site, so the labeling is confusing.

6) It would be good to label/show S92 and T93 in TM3 in Figure 3 or Figure 3–supplement, as they are referenced in the text.

7) The phrase “in the context of unconstrained GltPh and GltPh-in” is not immediately clear.

8) The legend of Figure 6c should clarify which structure is in colors. In that figure, I found it difficult to keep track of which structure was which.

9) Which figure is referred to for the Tl data? Could the authors comment on why Na2 is bound in GltPh-in in the absence of substrate? How does this relate to the point about coupling requiring?

10) I would say that since Figure 7b does not illustrate the progression of the transport domain beyond the intermediate structure, the authors should rephrase to “because the tip would presumably clash with TM5”

11) Figure 9 legend: I think that the word dynamic should be removed from the legend title; the figure clearly shows changes between two structures under two conditions, but dynamics aren't clearly demonstrated by the figure.

12) PDB identifiers will need to be added to the tables.

https://doi.org/10.7554/eLife.02283.028

Author response

Reviewer 1:

1) Introduction section; not clear what “reflect weaker coupling” refers to. The slope only tells something about how many Na could be involved in the conformational changes, not anything about coupling. Maybe it would be good to clearly state what the measurement really mean. A Hill coefficient of 1 doesn't mean that only 1 ion binds. Is this the same case here?

We have rephrased this sentence. It now reads: “reduced binding cooperativity between the substrate and Na+ ions”. Indeed, we do not imply that fewer Na+ ions bind to the transporter, only that the extent of cooperativity between binding of the ions and the substrate is diminished.

2) The off rate of aspartate is dependent on all conformational changes in the cycle and the coupling could be very indirect by Na affecting another step in the cycle. This should be stated.

This could be indeed the case in the wild-type transporter. However, we note that the loss of cooperativity is also observed in the transporter that is cross-linked in the inward facing state, suggesting that binding/dissociation in this isolated state is affected. We hope that using “binding cooperativity” rather than “coupling” in the revised manuscript eliminates the ambiguity.

3) Not clear what is meant by “trigger isomerization from apo... to bound ... conformation”: Isomerization of what?

Here, we refer to the isomerization of the transport domain. We have made the corresponding correction in the text.

4) It is also not clear why the isomerization would make the Na affinity low. Previous molecular modelling has proposed that Na binds with low affinity to the GltPh sites (without any isomerization needed to explain this low affinity).

Computational studies appear to predict binding free energies for Na1 ranging from – 7 kcal/mol (Heinzelmann et al., 2013) to -12.5 kcal/mol (Larsson et al., 2010). We have restated this more clearly in the revised text and included the references.

5) Are the two Tl+ sites occupied at the same time or in different crystals?

To clarify, we soaked crystals of the apo crystals of the inward facing GltPhin in Tl containing buffers. Although the same soaking conditions were used, we obtained two types of crystals. In one, Tl+ ions incorporated at the canonical Na1 and Na2 sites, and the protein was in a bound -like conformation. In the other, Tl+ ions incorporated at the two new Ct and Na2’ sites in the apo conformation. Both Ct and Na2’ sites were occupied in all three protomers. To determine whether the two sites were occupied at the same time in each molecule in the crystal, we would have to refine the occupancies of the Tl sites. Our occupancy estimates are 40-80%. However, the resolution of the data is not sufficient to estimate reliably these occupancies. Hence, we think that the sites are very likely occupied at the same time, but cannot state this with certainty. Therefore, we did not include such discussion in the text.

6) Figure 10. It is not clear from the model figure how the substrate and ions come of the transporter in the inward state?

At present, the release/gating mechanism is still not known. We have added a comment in the legend of Figure 10.

7) In the Introduction, it is stated that Na1 binding form the Na2 and asp binding sites. However, this is confusing, since also Na1 binding opens the Hp2 gate which should prevent Na2 binding. Please state carefully what Na1 does.

We have restated this more clearly. In brief, Na+ binding to the Na1 site is coupled through movements of Met311 to the structural transitions in HP2. While HP2 is collapsed in the apo form of the transporter and occupies the substrate binding and Na2 sites, in the Na+-bound form it frees these sites and assumes conformations that are more similar to the fully bound form. We have clarified this point in the text.

8) Figure 6e is not very legible.

We have increased the size of the panel and the font size for the residue numbers.

9) Previous experiments and modeling have suggested that the apo state returns by adopting a similar conformation as the fully loaded transporter. The present crystals fully support this.

We have included in the discussion a reference to Focke et al. (J. Neuroscience 2011), who have suggested that the apo form would be as compact and closed as the fully bound form.

Reviewer 2:

1) General remark. R/A mutation used in this paper may affect overall electrostatic of the pocket and thus bias packing of ions. Some cautious remarks should be put in the manuscript.

We have added the following statement in the Introduction section: “These results suggest that R397A is suitable to capture the apo and ions-only bound outward-facing states for their structural characterization. However, removal of Arg397 may affect local electrostatics, potentially altering ion binding; thus these studies should be interpreted with caution.”

2) While describing replacement of M311 with Leu the authors states: “similarly bulky” for L311 replacement. I am unsure if it is exactly correct statement. Furthermore, M311 was assigned a variety of roles in the past including tentative participation in stabilization of Tl+ in the binding pocket due to high polarizability. Any comments/discussions to that end would be beneficial. Tl+ marking of Na1/Na2 site remains a little bit of a mystery. Thus, it is not only volume but also chemistry of a side-chain that may be important.

We have replaced “similarly bulky” by “another bulky”. We agree with the reviewer that the role of the methionine residue side chain in Tl+ and/or Na+ binding remains poorly understood. We also note that from sequence analysis, it appears that Na+-coupled members of the family have Met at this position. In contrast, the members that are more likely to be proton-coupled (those, for example, that lack Asp405, one amino acid whose side chain is clearly involved in Na+-binding) have Leu at this position. While our data do not allow us to explain the role that the other group of Met311 may play in Na+ binding, we made a relevant comment in the revised manuscript.

3) The hypothesis about ion “priming” the site for substrate acceptance is interesting. Zhao et al. (PLoS Comput Biol 9(10): e1003296) described recently a fairly similar mechanism for MhP1 where transporter appears to get “primed” for substrate by binding of an ion stabilizing one of the states. Some discussion on this seemingly general mechanism, which may exist in other systems may benefit this manuscript. It is entirely up to the authors, but again it may represent a general mechanism of allosteric coupling during secondary transport.

We feel that the “priming” role for the Na+ ions in Mhp transporters proposed by Zhao and Noskov is distinct from their role in GltPh. In Mhp, the proposed key role of Na+ ions is to stabilize outward facing state compared to the inward facing state. In GltPh, we propose that the role of Na+ ions is local: structurally organizing the substrate-binding site within either the outward-facing or the inward-facing state. Hence, we feel that inclusion of the reference to Zhao and Noskov’s work may bring confusion. We have however added references to the crystal structures of Na+-only bound Mhp and LeuT transporters that, like GltPh, show open structures.

4) Furthermore, the idea of shuttling of a “compact” state of this transporter is interesting and well supported by two side-by-side publications in Nat Stuct Biology, but still some additional discussion on possible energetics of this process would be advantageous! The large protein has to move a great length across the membrane plain.

We agree with the reviewer that it is fascinating that such large conformational changes can occur readily in membranes. We note also that large and rapid transitions somewhat reminiscent of GltPh have recently been proposed for the Na+/H+ exchanger. While the dynamics is an important question for us, the current manuscript does not add significantly to the topic, other than to suggest that the apo transport domain may move in a manner overall similar to that of the fully loaded protein. We feel that extending the discussion on energetics here would not be appropriate.

Reviewer 3:

1) My only substantive query is why several of the structures have not been deposited in the PDB. These are all structures at >4 Angstrom resolution, but I'm not sure that is sufficient reason; this may come down to journal policy.

We have also deposited the structure of Tl-bound GltPhin in the bound-like conformation. Regarding the other structures, we think that the structures do not provide additional structural information. They were used only to evaluate monitor the anomalous signal of Tl+ ions at the Ct and Na2’ sites.

2) I didn't follow the arguments regarding the evidence for priming of the transporter by Na1. For example, the Na2 site is distorted in GltPh-R397A+Na, so its structure is not really similar to the bound state. Maybe it would help to separate out the figures of the outward-open R397A+Na from those of the inward-facing Tl-bound. Or perhaps the phrasing just needs a little reworking.

We have improved this paragraph. The key clarification is: “The obtained outward- and inward-facing structures pictured the transport domains in conformations overall similar to those observed in the fully bound transporter: straightened TM3, Met311 pointing towards the binding sites, extended helix in HP2a and HP2 tip raised out of the substrate binding site (Figures 6a-d).” The remainder of the paragraph better separates the discussion on the inward and outward facings states to avoid confusion. Finally, we added a figure (Figure 3–figure supplement 2), which shows RMSDs per residue between the transport domains of the various states presented. From this figure, it is clear that Na+-binding triggers most of the structural changes that are associated with loading of the transport domain.

3) In the Introduction section I was confused by these statements: “Met311 in the NMD motif is the only residue that is shared between the Na1 site and the substrate and Na2 sites and also undergoes a conformational change upon ligand binding.” Which are the reference states in each case? Are the authors only comparing the trimerization domains?

We wanted to point out that the positions and orientations of the transport domains relative to the trimerization domains are the same in the apo and fully bound forms. The conformations of the transport domains themselves are significantly different. We have rephrased the sentence.

4) I would welcome quantitation of the similarity between domains in the different states, using RMSDs of the backbone and of all atoms. It would put the differences in context relative to the crystallographic resolution.

We have generated new figures, including Figure 3–figure supplement 2 and Figure 9c, where we show RMSDs for main chain atoms of the transport domains calculated between equivalent residues in different structural states. These figures clarify the extent of the structural changes and similarities between the various states.

5) In Figure 4–figure supplement 1, it is not yet explained what is meant by the Ct site, so the labeling is confusing.

We have moved the figure and the corresponding discussion to the section on Ct site so that Figure 4–figure supplement 1 has become Figure 8–Figure supplement 1.

6) It would be good to label/show S92 and T93 in TM3 in Figure 3 or Figure 3–supplement, as they are referenced in the text.

We have added the labels in Figure 3(c) and Figure 3–Figure supplement 1

7) The phrase “in the context of unconstrained GltPh and GltPh-in” is not immediately clear.

We have clarified this: “unconstrained wild-type GltPh and inward cross-linked GltPhin

8) The legend of Figure 6c should clarify which structure is in colors. In that figure, I found it difficult to keep track of which structure was which.

We have clarified the legend. We have also included the substrate in the figure.

9) Which figure is referred to for the Tl data? Could the authors comment on why Na2 is bound in GltPh-in in the absence of substrate? How does this relate to the point about coupling requiring?

The appropriate figure is Figure 6a. We have referenced it in the text. We have also added the discussion on coupling. In brief, it is possible that the closed Tl+-bound inward-facing state observed in crystallo reflects the closed Na+-bound inward -facing state of the transporter. This does not violate the principle of coupled symport because the affinity for Na+ in this state is very low (Kd∼250 mM) and in the cytoplasm where the Na+ concentration is expected to be around ∼ 10 mM, so this state is not expected to be significantly populated.

10) I would say that since Figure 7b does not illustrate the progression of the transport domain beyond the intermediate structure, the authors should rephrase to “because the tip would presumably clash with TM5”.

We have rephrased this part of the sentence as follows: “further progression of the transport domain to the inward-facing position could be impeded because the tip is likely to clash with TM5 in the trimerization domain”.

11) Figure 9 legend: I think that the word dynamic should be removed from the legend title; the figure clearly shows changes between two structures under two conditions, but dynamics aren't clearly demonstrated by the figure.

We have changed the title to “Movements of the HP1-TM7a structural module”

12) PDB identifiers will need to be added to the tables.

PDB codes have been added to the tables.

https://doi.org/10.7554/eLife.02283.029

Article and author information

Author details

  1. Grégory Verdon

    Department of Physiology and Biophysics, Weill Cornell Medical College, New York, United States
    Present address
    1. Department of Life Sciences, Imperial College London, South Kensington, United Kingdom
    2. Research Complex at Harwell, Rutherford Appleton Laboratory, Didcot, United Kingdom
    Contribution
    GV, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    SeCheol Oh
    For correspondence
    g.verdon@imperial.ac.uk
    Competing interests
    The authors declare that no competing interests exist.
  2. SeCheol Oh

    Department of Physiology and Biophysics, Weill Cornell Medical College, New York, United States
    Contribution
    SO, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Grégory Verdon
    Competing interests
    The authors declare that no competing interests exist.
  3. Ryan N Serio

    Department of Pharmacology, Weill Cornell Medical College, New York, United States
    Contribution
    RNS, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  4. Olga Boudker

    Department of Physiology and Biophysics, Weill Cornell Medical College, New York, United States
    Contribution
    OB, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    olb2003@med.cornell.edu
    Competing interests
    The authors declare that no competing interests exist.

Funding

National institutes of Health (NS064357)

  • Olga Boudker

The funder had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the US Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886. We thank Howard H Robinson, Annie Héroux for access to beamlines X29 and X25 at the National Synchrotron Light Source, and the staff for their assistance. We thank Deena Oren (Structural Biology Resource Center, The Rockefeller University) for her assistance with the liquid handling robot. We also thank Michael Sawaya (Institute for Genomics and Proteomics, University of California, Los Angeles, USA) for providing the programs and scripts for performing the anisotropy correction. We thank Irene Kiburu for the advice on calculation of R.M.S.D.s in VMD. We thank Alessio Accardi, Nurunisa Akyuz, Crina Nimegean and Nicolas Reyes for helpful comments on the manuscript.

Contributions

GV, SO, and OB designed the experiments and analyzed the data. GV and SO conducted the experiments, and collected X-ray diffraction data. RS performed preliminary ion-binding experiments for GltPhin-M311A. GV, OB, and SO wrote the manuscript.

Reviewing Editor

  1. Benoit Roux, University of Chicago, United States

Publication history

  1. Received: January 10, 2014
  2. Accepted: May 16, 2014
  3. Accepted Manuscript published: May 19, 2014 (version 1)
  4. Version of Record published: June 10, 2014 (version 2)

Copyright

© 2014, Verdon et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,621
    Page views
  • 375
    Downloads
  • 67
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Structural Biology and Molecular Biophysics
    Adishesh K Narahari et al.
    Research Article Updated

    Pannexin 1 (Panx1) is a membrane channel implicated in numerous physiological and pathophysiological processes via its ability to support release of ATP and other cellular metabolites for local intercellular signaling. However, to date, there has been no direct demonstration of large molecule permeation via the Panx1 channel itself, and thus the permselectivity of Panx1 for different molecules remains unknown. To address this, we expressed, purified, and reconstituted Panx1 into proteoliposomes and demonstrated that channel activation by caspase cleavage yields a dye-permeable pore that favors flux of anionic, large-molecule permeants (up to ~1 kDa). Large cationic molecules can also permeate the channel, albeit at a much lower rate. We further show that Panx1 channels provide a molecular pathway for flux of ATP and other anionic (glutamate) and cationic signaling metabolites (spermidine). These results verify large molecule permeation directly through caspase-activated Panx1 channels that can support their many physiological roles.

    1. Cell Biology
    2. Structural Biology and Molecular Biophysics
    Salah A Baker et al.
    Research Article Updated

    Interstitial cells of Cajal (ICC) generate pacemaker activity responsible for phasic contractions in colonic segmentation and peristalsis. ICC along the submucosal border (ICC-SM) contribute to mixing and more complex patterns of colonic motility. We show the complex patterns of Ca2+ signaling in ICC-SM and the relationship between ICC-SM Ca2+ transients and activation of smooth muscle cells (SMCs) using optogenetic tools. ICC-SM displayed rhythmic firing of Ca2+transients ~ 15 cpm and paced adjacent SMCs. The majority of spontaneous activity occurred in regular Ca2+ transients clusters (CTCs) that propagated through the network. CTCs were organized and dependent upon Ca2+ entry through voltage-dependent Ca2+ conductances, L- and T-type Ca2+ channels. Removal of Ca2+ from the external solution abolished CTCs. Ca2+ release mechanisms reduced the duration and amplitude of Ca2+ transients but did not block CTCs. These data reveal how colonic pacemaker ICC-SM exhibit complex Ca2+-firing patterns and drive smooth muscle activity and overall colonic contractions.