Research Article

Structure of a Eukaryotic CLC Transporter Defines an Intermediate State in the Transport Cycle

See allHide authors and affiliations

Science  29 Oct 2010:
Vol. 330, Issue 6004, pp. 635-641
DOI: 10.1126/science.1195230

Controlling Chloride Channels

The CLC proteins are a large family of channels and transporters that transfer chloride ions across cell membranes. While structures of two prokaryotic CLCs have been determined, these do not include the cytoplasmic regulatory domains found in eukaryotic transporters, and the structures do not reveal the mechanism of Cl/H+–coupled transport. L. Feng et al. (p. 635, published online 30 September; see the Perspective by Mindell) describe the structure of a eukaryotic CLC protein and found that the regulatory domains interacted closely with the transmembrane domain so that conformational changes are transmitted to the ion pathway. A gating glutamate in the eukaryote transporter is in a different conformation to prokaryotic structures, explaining the 2:1 stoichiometry of Cl/H+ exchange in eukaryotes.


CLC proteins transport chloride (Cl) ions across cell membranes to control the electrical potential of muscle cells, transfer electrolytes across epithelia, and control the pH and electrolyte composition of intracellular organelles. Some members of this protein family are Cl ion channels, whereas others are secondary active transporters that exchange Cl ions and protons (H+) with a 2:1 stoichiometry. We have determined the structure of a eukaryotic CLC transporter at 3.5 angstrom resolution. Cytoplasmic cystathionine beta-synthase (CBS) domains are strategically positioned to regulate the ion-transport pathway, and many disease-causing mutations in human CLCs reside on the CBS-transmembrane interface. Comparison with prokaryotic CLC shows that a gating glutamate residue changes conformation and suggests a basis for 2:1 Cl/H+ exchange and a simple mechanistic connection between CLC channels and transporters.

CLC proteins form a large family of membrane proteins that transfer chloride ions across cell membranes. Present in all kingdoms of life, existing in both surface and intracellular membranes, CLCs mediate a wide range of physiological processes (1, 2). In muscle, they govern resting membrane potential; in kidneys, they facilitate transepithelial fluid flow; and in intracellular compartments, they control pH through coupled Cl/H+ exchange (25). Mutations of CLC proteins underlie numerous inherited diseases including myotonia congenita, Bartter’s syndrome, Dent’s disease, osteopetrosis, retinal degeneration, and lysosome storage diseases (2, 5).

CLCs are divisible into two subgroups, Cl channels and secondary active transporters (6). The channels catalyze passive diffusion down the Cl electrochemical gradient, whereas the transporters (Cl/H+ exchangers) actively couple Cl movement in one direction to H+ movement in the opposite direction (610). Thus, in the transporters, “downhill” movement of one ion can drive “uphill” movement of the other. Even though channels and transporters catalyze energetically different reactions, conservation of specific amino acids informs us that these functionally distinct CLC subgroups must share the same basic architecture (1113).

CLC proteins are homodimers with a separate ion pathway within each subunit (14, 15). The two pathways act largely independently, but in some channels, they cooperate to turn on and off simultaneously (16). Each subunit of the dimer consists of a transmembrane (TM) component, which forms the ion pathway, and, in eukaryotic CLCs, a cytosolic cystathionine beta-synthase (CBS) domain component, which affects membrane localization and regulates the TM component (1722).

Crystal structures of two closely related prokaryotic CLC-transporter proteins have been determined (15, 23). These have defined the protein architecture within the membrane, including Cl binding sites along the ion-transport pathway, but they have not yet provided a clear mechanism for Cl/H+ exchange, an explanation for the exchange stoichiometry of two Cl ions for one H+, or a plausible hypothesis for understanding how CLCs can encode both channels and secondary active transporters. Moreover, the prokaryotic transporters of known structure do not contain the CBS domains found in all eukaryotic CLCs (2). The many disease-causing mutations discovered within the CBS domains and the effects of site-directed mutations will gain a more mechanistic interpretation in light of a eukaryotic CLC structure (2, 21, 2428).

CmCLC is a Cl/H+ exchange transporter. To overcome the instability of eukaryotic CLCs, we took an approach similar to that used in the study of prokaryotic voltage-dependent K+ channels (29) and identified CmCLC from a thermophilic red alga Cyanidioschyzon merolae (fig. S1) (30, 31), which migrated as a monodisperse peak on a size-exclusion column. On the basis of limited proteolysis, a final construct excluding 86 N-terminal and 7 C-terminal amino acids was expressed in Trichoplusia ni insect cells and purified for functional and crystallographic studies.

CmCLC was reconstituted into lipid vesicles and studied initially with a fluorescence assay (Fig. 1A) (32). Vesicles loaded with 450 mM KCl (pH 7.4) were diluted 20-fold into an assay buffer containing 450 mM K+ gluconate (pH 7.5). Under these conditions, if CmCLC is a Cl/ H+ exchanger, Cl efflux will drive H+ influx against its concentration gradient and cause quenching of the fluorophore 9-amino-6-chloro-2-methoxyacridine (ACMA). This reaction is initiated by the addition of the K+ selective ionophore valinomycin, which collapses the electrical gradient built up by an electrogenic transporter. On the other hand, if CmCLC is a Cl channel, H+ transport and, thus, a fluorescence change should not be observed. We first validated this assay with a known Cl/H+ exchanger, EcCLC (7). The robust fluorescence decrease observed upon addition of valinomycin reflects H+ influx into the vesicles driven by Cl efflux (Fig. 1B). Control experiments with empty vesicles and vesicles containing EcCLCs with a mutation of the “gating glutamate” [Glu148 → Ala148 (E148A) (33)], which are known to transport Cl without H+ exchange activity, showed little fluorescence change (Fig. 1B). However, in assays with vesicles containing CmCLC, a clear fluorescence reduction was observed (Fig. 1C). When the gating glutamate of CmCLC was mutated (E210A), the fluorescence signal was minimized (Fig. 1C). Thus, CmCLC is a secondary active transporter, and Cl/H+ exchange depends on the presence of the gating glutamate.

Fig. 1

Analysis of CmCLC transport activity. (A) Depiction of the fluorescent-based flux assay. Vesicles (yellow) were loaded with a high concentration of KCl and then diluted into the flux buffer with a low concentration of Cl and ACMA. Flux was initiated by addition of valinomycin. The X indicates that protonated ACMA does not exit. (B) Fluorescence changes for WT (red) and E148A mutant (blue) EcCLCs compared with empty (black) vesicles. CCCP, carbonyl cyanide m-chlorophenyl hydrazone. (C) Fluorescence changes for WT and E210A mutant CmCLCs and empty vesicles. (D) Recordings of pH (red) and [Cl] (black) over time in response to the addition of valinomycin to a solution of Cl-loaded vesicles. Electrode response was calibrated by the addition of 5 μM NaOH and 20 μM NaCl before each experiment. The initial rates of Cl efflux and H+ influx provide an estimate of the exchange ratio (inset).

The exchange stoichiometry in EcCLC and other CLC transporters was shown to be two Cl ions to one H+ (7, 9, 10, 34, 35). To estimate the exchange stoichiometry in CmCLC, we directly measured the Cl concentration change and pH change in a flux assay developed by Miller et al. for EcCLC (6). We determined the initial rate of Cl increase and H+ decrease upon valinomycin addition in parallel under identical conditions. From the initial rates, the average Cl/ H+ flux ratio for CmCLC is 2.25 ± 0.22 (Fig. 1D).

Structure determination. CmCLC was crystallized in space group C2 with two copies of the CmCLC dimer in each asymmetric unit. The crystals were grown in the presence of 150 mM Cl and diffracted x-rays to 3.5 Å resolution (fig. S2). The experimental phases were derived from single- and multiwavelength diffraction experiments on selenomethionine-containing crystals (table S1). The initial experimental map after solvent flattening allowed tracing of most of the main chain. After several rounds of model rebuilding and refinement, most of the TM component and part of the CBS component could be built with the aid of 16 selenomethionine markers. The remaining segments were built initially as polyalanine chains. To improve the model, we replaced 30 amino acids throughout the protein with methionine in 20 CmCLC mutants and crystallized them with selenomethionine labels. Anomalous difference analysis with the selenium edge identified 24 out of 30 mutation sites to aid in model building (Fig. 2A). Native plus mutant methionines provided one or more selenomethionine markers on almost every helix and strand and on some loops (Fig. 2A). By establishing the register and allowing the building of most of the side chains, this extensive labeling improved the accuracy and completeness of the model, which is refined to working and free residuals (Rwork/Rfree) = 0.259/0.284 (table S1).

Fig. 2

Structure of CmCLC. (A) Selenium sites identified by single-wavelength anomalous dispersion. Wire representation of a CmCLC monomer is shown in blue. The selenium sites of the selenomethionine-labeled WT protein are shown as green spheres, and those corresponding to “marker mutants” are red. (B and C) Ribbon representation of a CmCLC dimer from the side of the membrane with the extracellular solution above. The two subunits are colored in blue and red, respectively.

Molecular architecture. Each monomer of the CmCLC dimer contains 24 α helices and 6 β strands (Fig. 2, B and C, Fig. 3A, and fig. S1). The TM monomer of CmCLC has an antiparallel architecture similar to that of EcCLC (fig. S3A) (15). The cytoplasmic CBS domains share the fold characteristics of other CBS-containing proteins (fig. S3B) (36). Two CBS subdomains on each subunit (CBS1 and CBS2) are related by a pseudo–twofold-symmetry axis and are tightly packed against each other through β sheets (Fig. 3, A and B). A long and ordered linker following the last TM helix, αR, makes sharp turns, crosses over CBS2, and reaches CBS1 (Fig. 3B). This arrangement brings CBS2 in close proximity to the TM and places CBS1 farther away from the membrane. As a result, the C-terminal end of the protein is positioned closest to the membrane.

Fig. 3

Structure of CmCLC subunit. (A) Stereo view of a CmCLC subunit from the side of the membrane with the extracellular solution above. The black capital letters indicate elements of secondary structure; the blue C and N denote the C and N termini, respectively. d-e, f-g, and r-s correspond to secondary structure elements not present in EcCLC. (B) Surface representation of a CmCLC subunit. The TM domain, gray; CBS1 subdomain, pale blue; and CBS2, darker blue. The linker between the TM domain and CBS1 is shown as a yellow cord.

The two subunits in the CmCLC dimer are related by a dyad perpendicular to the membrane and bury an extensive interface: 3200 Å2 in the TM components and 1830 Å2 in the CBS-domain components (Fig. 2B). The CBS interface is similar to that observed in the isolated domain structures from CLC-5 and CLC-Ka and is consistent with extensive mutagenesis studies on other members of the CLC family (21, 22, 36, 37).

Although CmCLC and EcCLC have modest sequence identity (<25%), the structure of the CmCLC TM superimposes well onto EcCLC with a main-chain root mean square deviation (RMSD) of 1.7 Å over a range of 366 aligned residues (fig. S3A and fig. S4A). Helices forming the dimer interface superimpose best (1.2 Å main-chain RMSD for αH, αI, αP, and αQ), whereas helices on the perimeter farthest from the interface, particularly αJ and αK, superimpose worst (3.2 Å main-chain RMSD for αK and the C-terminal half of αJ and αM) (fig. S4, B and C). The larger deviations on the perimeter are mainly due to differences in the tilt angle of the helices. At present, we do not know whether these deviations reflect a true structural difference of equivalent conformational states or different conformations.

Interface between the TM and the CBS. In some membrane proteins, for example, eukaryotic voltage-dependent channels of the Kv1 class, regulatory, or structural domains in the cytoplasm are loosely attached to the TM component through extended polypeptide chains (38). In contrast, in CmCLC we observe an extensive protein-protein interface burying 3600 Å2 between the CBS and TM components. The shape complementarity index is 0.69, similar to an antibody-antigen interface (39). This area and complementarity index suggest a highly specific interaction.

Three structural features of the interface seem relevant to the role of the CBS domains in regulating TM function. First, the polypeptide linker between the TM and CBS not only connects the two together, but also makes multiple contacts with TM helix αR (Fig. 4A). Helix αR extends into the ion pathway and holds Y515 in place to form part of the central Cl binding site (23, 40, 41). Second, the CBS domain contacts helix αD, which also extends into the ion pathway and contains another important amino acid, S165, that is involved in Cl ion coordination and selectivity (Fig. 4A) (23, 34, 35, 42, 43). This explains how conformational changes within the CBS domains could be transmitted directly to the ion pathway of the TM domain (fig. S6). Third, the loop connecting αH to αI, which forms part of the TM dimer interface, comes in direct contact with the CBS domains (fig. S5, A and B), raising the possibility that the CBS domains could influence cooperative interactions between the two TM components (44).

Fig. 4

The TM-CBS interface. (A) A CmCLC subunit highlighting the TM and CBS interface. The CBS is shown in light blue and the TM as gray ribbons with helices αR and αD highlighted in red. S165 and Y515 are shown as sticks. Yellow, carbon; red, oxygen. Top panels show the stereo view of the zoomed-in boxed area in the bottom panel. (B) Mutations on the TM-CBS interface. Mutations of hCLC-1, CLC-0, and CLC-7 are mapped on the structure through sequence alignment (fig. S1) and shown as spheres. There are four groups: (i) hCLC-1 and CLC-0 mutations in the CBS domain that affect common gating (H835R in hCLC-1, H736A and E763K in CLC-0; green), (ii) dominant Thomsen’s disease mutations in the intracellular loops of the hCLC-1 TM (G200R, F297S, R300Q; red), (iii) dominant osteopetrosis mutations in the hCLC-7 CBS domains (F758L, R762L, G765B, R767W, A788D; yellow), and (iv) dominant osteopetrosis mutations in the loops of the hCLC-7 TM (G215R, L213F, F318L; pink).

Several disease-causing mutations in human CLCs localize to the TM-CBS interface (Fig. 4B). Studies have shown that H835R mutations in hCLC-1 and H736A and E763K in the ortholog CLC-0 alter the “common gating” process that affects both subunits in unison (25). These amino acids localize to the surface of the CBS that faces the TM (Fig. 4B, green spheres). Several hCLC-1 mutations that underlie Thomsen’s disease exist on the TM-dimer interface (44, 45). We find that several other mutations causing this disease exist on the TM-CBS interface (Fig. 4B). More than 30 osteopetrosis-related mutations have been identified in a Cl/ H+ exchanger hCLC-7 (2). Although mutations involved in recessive osteopetrosis tend to be spread throughout the structure (46), most dominant mutations are located on the TM-CBS interface (Fig. 4B, yellow and pink spheres) (4649). It appears that altering communication between CBS and TM domains may be a common mechanism underlying many CLC-related diseases.

Truncations of the hCLC-1 C terminus affect gating of the channel (50, 51), and several disease-causing mutations are known to occur in this region. Because CLCs vary in the length of the unstructured C terminus following CBS2, the importance of this region to channel function was puzzling. The crystal structure offers a possible explanation; the C terminus is in close proximity to the TM and participates in the TM-CBS interface (Figs. 3A and 4A). Hence, it is possible that the C-terminal peptide affects transport function through its interaction with helix αR or other regions of the TM component.

Ion-transport pathway. The Cl-transport pathway of CmCLC is shaped like an hourglass with aqueous vestibules on the extracellular and intracellular surfaces leading up to a narrow segment (Fig. 5A, blue mesh). In between the vestibules, the transport pathway narrows into a selectivity filter or transport region where ions become dehydrated as they cross the membrane. The α helices αN, αF, and αD are oriented with their N termini pointed toward the transport region, making amide nitrogen atoms available for anion coordination. Overall, the transport region of CmCLC is very similar to that of EcCLC with respect to the positioning of the oriented α helices and the locations of residues Y515 and S165, which are known to play an important role in Cl transport (23, 34, 35, 40, 42).

Fig. 5

Ion-transport pathway. (A) Stereo view of the ion-transport pathway viewed from the dimer interface. The protein is shown as a blue ribbon, with selected residues S165, E210, and Y515 shown as sticks. Aqueous cavities connecting to the Cl binding sites from the extracellular (top) and the intracellular (bottom) solutions are shown as pale blue mesh; Cl ions are shown as pink spheres. (B) Stereo view of the electron-density map around the ion-transport pathway. Weighted 2fo-fc (where fo is the observed structure factor amplitude, and fc is the model structure factor amplitude) electron density at 3.5 Å, contoured at 1.35σ, is shown in pale blue. The refined model is shown as sticks. A bromine anomalous-difference Fourier map at 4.2 Å from crystals grown in Br is contoured at 9σ (red). (C) Stereo view of the ion-transport pathway. Possible hydrogen bonds between Cl ions (pink spheres) and protein are denoted by dashed lines, as are the possible hydrogen bonds between E210 and Y515.

One major difference exists between the transport regions of CmCLC and EcCLC. In EcCLC, three Cl ion–binding sites were identified: Sext, Scen, and Sint, from the extracellular to the intracellular side (23). In a mutant of EcCLC in which the gating glutamate E148 (corresponding to E210 in CmCLC) was mutated to a glutamine residue, the Cl analog Br occupied all three sites in difference Fourier maps, and the glutamine side chain was located in the extracellular solution, directed away from the transport region (Fig. 6A) (23). In wild-type (WT) EcCLC, the gating glutamate was bound to the outermost site, Sext, and Br was excluded from that position (Fig. 6A) (23). Thus, the gating glutamate appeared to compete with the halogen anion for the extracellular site. However, in CmCLC, a new conformation is observed: The gating glutamate occupies the central site, and Br, as demonstrated in an anomalous-difference Fourier experiment, occupies Sext and Sint (Fig. 5, A and B). This new conformation is achieved mainly through a reorientation of the gating glutamate side chain. The carboxylate group in the central site interacts with the same hydrogen bond donor groups that interact with the halogen anion at that site in the other structures (Fig. 5C).

Fig. 6

A working model for ion transport. (A) Close-up view of the ion-transport pathway of WT CmCLC, WT EcCLC, and the E148Q mutant of EcCLC, respectively. The foreground helices were removed for clarity. Selected residues are shown as sticks and Cl as pink spheres. For CmCLC, a bromine anomalous-difference Fourier map contoured at 6σ is shown as red mesh. (B) A proposed working model for ion transport in CLC transporters. Residues corresponding to S165, E210, and Y515 in CmCLC are shown as schematic drawings. Red spheres represent Cl ions, and purple spheres denote protons. The negative charge on the carboxyl group of the deprotonated E210 is shown as a dashed circle. (a), (b), (c), (d), (e), and (f) are six major steps in this transport cycle, as described in the text. The illustration in the middle of the circle shows the positions that Cl and the carboxyl group of the gating glutamate could occupy. For an animated version of this figure, see movie S1.

Two amino acids are known to affect the movement of H+ through CLC transporters: the externally located gating glutamate and an internally located “proton glutamate,” E203 in EcCLC (7, 52). In EcCLC, mutations at either position can abolish coupled H+ movement and convert the transporter into a Cl channel with a small conductance (7, 5254). In CmCLC, a threonine residue (T269) is found at the proton glutamate position. We note that the side-chain hydroxyl (Oγ) of T269 is only 7.5 Å away from the carboxylate (Oε) of E210. Either this threonine residue is able to perform the H+ transfer between the intracellular solution and the transport region in CmCLC, or, alternatively, a different amino acid performs this role.

Hypothesis for the mechanism of two Cl/one H+ countertransport. Together with the prokaryotic structures, three structures are now available (Fig. 6A) (15, 23). Although these structures were obtained with two different transporters and a mutant, they likely represent conformations that can occur in the transport cycle. From these structures, we put forth a hypothesis for the transport mechanism that accounts for the countertransport of two Cl ions against one H+ (Fig. 6B). States (a) and (b) represent unprotonated and protonated forms, respectively, of the conformation observed in CmCLC (Fig. 6A, left); states (d) and (e) represent protonated and unprotonated forms, respectively, of the conformation observed in the mutant EcCLC-E148Q (Fig. 6A, right) (23); and state (f) represents the conformation observed in WT EcCLC (Fig. 6A, middle) (15). From state (a) to state (b), a H+ is transferred from the intracellular solution to the carboxylate of the gating glutamate. The protonated carboxylate changes its conformation to the extracellular solution [state (c)], and two Cl ions enter the transport region from the external side [state (d)]. Once the H+ dissociates from the carboxylate to the extracellular solution [state (e)], the carboxylate can enter Sext [state (f)] and then Scen [state (a)], in association with the movement of two Cl ions to the intracellular side. In Scen, the glutamate is again in position to receive a H+ from the intracellular solution. Every transition in this cycle is reversible. Net cycling can only result from the dissipation of a Cl or H+ electrochemical gradient. As described here, this cycle will give rise to an exchange stoichiometry of two Cl ions for one H+.

There are two apparent problems with the cycle as described. First, in the (c)-to-(d) transition, two Cl ions must enter from the extracellular solution, or else stoichiometric coupling and active transport will be lost. Second, (d) and (e) are “forbidden” states for a transporter, because Cl can move uncoupled to H+. Both of these problems can be overcome with one assumption: that there exists a relatively large kinetic barrier to the rapid movement of Cl ions between Sint and Scen. This barrier would ensure that Cl ions would more likely fill Scen and Sext from the extracellular solution. In fact, if the barriers between the transport region and the extracellular solution are small, then state (c) will exist only transiently. A sufficiently large barrier between Sint and Scen would also permit the existence of “forbidden” states (d) and (e) without substantial loss of coupled exchange. This conclusion can be understood through a simple kinetic argument. If the transition from (d) to (e) is determined by the lifetime of the protonated carboxylate in the extracellular solution, then we expect a duration of ~10−6 s if the pKa (where Ka is the acid dissociation constant) is 4.0 and we assume a H+ association rate of 1010 M–1 s–1 (55). Therefore, if the barrier were high enough to limit the Cl conduction rate through states (d) and (e) to 104 s–1, then on average only one Cl every hundred cycles would slip through state (d) uncoupled to H+ movement. If the transition from (e) to (f) were then rapid, little uncoupled slippage will occur during the cycle. In the EcCLC-E148Q mutant, representing a transporter caught permanently in states (d) and (e), the Cl conduction rate was determined to be less than 104 s–1 (53). We would suggest that the slow intrinsic Cl conductivity of mutated CLC transporters that function as channels (such as the EcCLC-E148Q mutant) results from a kinetic barrier to Cl that is an important aspect of the Cl/H+ coupled-exchange mechanism.

Movie S1

Kinetic simulations of the cycle described in the paper.

Kinetic simulations corroborate the explanations above (movie S1). Simulations of the cycle—allowing Cl entry and exit from either side of the membrane during the transitions connecting states (c) and (d) and allowing Cl conduction across the membrane through states (d) and (e)—generate the experimentally observed properties of CLC exchangers: Cl gradients produce uphill H+ transport, H+ gradients produce uphill Cl transport, and a coupling ratio of two Cl ions against one H+ is generated over conditions of membrane voltage, electrolyte, and pH under which CLC exchangers have been studied (7, 9, 10, 34, 35). Moreover, an adjustment of the rate at which Cl ions conduct through states (d) and (e) produces a loss of coupling and converts the CLC exchanger into a gated Cl ion channel.

The cycle also explains other more subtle features of CLC proteins. Certain mutations produce partial uncoupling (Cl/H+ exchange ratios greater than two) rather than complete uncoupling (41). This observation is plausible if different mutations cause different rates of Cl conduction through states (d) and (e) or if they alter the interaction between the gating glutamate and the pore (Fig. 6B). Lowering external pH produces uncoupling in CLC-3 (56). This outcome is expected if, due to a high extracellular H+ concentration, the transition rate from state (e) to state (d) becomes much greater than that from state (e) to state (f) (Fig. 6B). Finally, CLC proteins that function as channels exhibit pH-dependent gating (16). The cycle predicts that these channels may still mediate coupled exchange on a microscopic level that is hidden beneath uncoupled Cl conduction (57).

There may exist undiscovered conformational states of CLC transporters. Still, the mechanism of secondary active transport that we are proposing, although distinct from the classical alternating-access model (58), is plausible because it accounts for experimental observations. It invokes a Cl channel with a barrier to slow the throughput and a glutamate gate that can swing into the channel and be protonated from either side of the membrane. This mechanism explains why the CLC protein structure gives rise to both transporters and channels.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

Table S1


Movie S1

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  3. We thank the staff at beamline X29 (National Synchrotron Light Source, Brookhaven National Laboratory) and K. R. Rajashankar and K. Perry at beamline 24ID-C (Advanced Photon Source, Argonne National Laboratory) for assistance at the synchrotron, members of the MacKinnon laboratory for helpful dicussions, and P. Yuan for comments on the manuscript. R.M. is an Investigator in the Howard Hughes Medical Institute. The x-ray crystallographic coordinates and structure factors have been deposited in the Protein Data Bank with accession identification number 3ORG.
View Abstract

Stay Connected to Science

Navigate This Article