Corrections

ECOLOGYCorrection for “Recovery of a top predator mediates negativeeutrophic effects on seagrass,” by Brent B. Hughes, Ron Eby, EricVan Dyke, M. Tim Tinker, Corina I. Marks, Kenneth S. Johnson,and Kerstin Wasson, which appeared in issue 38, September 17,2013, of Proc Natl Acad Sci USA (110:15313–15318; first publishedAugust 27, 2013; 10.1073/pnas.1302805110).

The authors note that Fig. 2 appeared incorrectly. Theauthors note that they “unintentionally labeled Fig. 2C ‘Grazerbiomass (g DW) *shoot (cm)−1’ instead of ‘Grazer biomass(mg DW) *shoot (cm)−1.’ ” The corrected figure and its leg-end appear below. This error does not affect the conclusionsof the article.

www.pnas.org/cgi/doi/10.1073/pnas.1401578111

Fig. 2. (A) Interaction web of top-down and bottom-up effects in the eelgrass study system. The top predator is the sea otter (E. lutris), the mesopredators arecrabs (Cancer spp. and Pugettia producta), the epiphyte mesograzers are primarily an isopod (I. resecata) and a sea slug (P. taylori), and algal epiphyte competitorsof eelgrass primarily consist of chain-forming diatoms, and the red alga Smithora naiadum. Solid arrows indicate direct effects, dashed arrows indicate indirecteffects, and the plus and minus symbols indicate positive and/or negative effects on trophic guilds and eelgrass condition. C, competitive interaction; T, trophicinteraction. (Original artwork by A. C. Hughes.) (B–E) Survey results testing for the effects of sea otter density on eelgrass bed community properties (Tables S2and S3). Elkhorn Slough (sea otters present and high nutrients) eelgrass beds (n = 4) are coded in red, and the Tomales Bay reference site (no sea otters, lownutrients) beds (n = 4) are coded in blue. (B) Crab biomass and size structure of two species of Cancer crabs; (C) grazer biomass per shoot and large grazer density;(D) algal epiphyte loading; and (E) aboveground and belowground eelgrass biomass. DW, dry weight; FW, fresh weight.

3644–3645 | PNAS | March 4, 2014 | vol. 111 | no. 9 www.pnas.org

MICROBIOLOGYCorrection for “Programmed Allee effect in bacteria causes atradeoff between population spread and survival,” by Robert Smith,Cheemeng Tan, Jaydeep K. Srimani, Anand Pai, Katherine A.Riccione, Hao Song, and Lingchong You, which appeared inissue 5, February 4, 2014, of Proc Natl Acad Sci USA (111:1969–1974; first published January 21, 2014; 10.1073/pnas.1315954111).The authors note that the following statement should be added

to the Acknowledgments: “This work was partially supported bya Society in Science–Branco Weiss Fellowship (to C.T.).”

www.pnas.org/cgi/doi/10.1073/pnas.1401930111

SYSTEMS BIOLOGYCorrection for “Analysis of proteome dynamics in the mouse brain,”by John C. Price, Shenheng Guan, Alma Burlingame, Stanley B.Prusiner, and Sina Ghaemmaghami, which appeared in issue 32,August 10, 2010, of Proc Natl Acad Sci USA (107:14508–14513; firstpublished August 10, 2010; 10.1073/pnas.1006551107).The authors note that the following grant should be added to

the Acknowledgments: “NIH Grant AG002132.”

www.pnas.org/cgi/doi/10.1073/pnas.1401576111

NEUROSCIENCECorrection for “Mapping the receptor site for α-scorpion toxinson a Na+ channel voltage sensor,” by Jinti Wang, Vladimir Yarov-Yarovoy, Roy Kahn, Dalia Gordon, Michael Gurevitz, ToddScheuer, and William A. Catterall, which appeared in issue 37,September 13, 2011, of Proc Natl Acad Sci USA (108:15426–15431;first published August 29, 2011; 10.1073/pnas.1112320108).The authors note that the following statement should be added

as a new Acknowledgments section: “This work was supported byNational Institutes of Health Research Grants R01 NS015751 (toW.A.C.) and U01 NS058039 (to W.A.C. and M.G.), by ResearchGrant IS-4313-10 from the US-Israel Binational AgriculturalResearch and Development Foundation (to M.G. andW.A.C.), andby Israeli Science Foundation Grant 107/08 (to M.G. and D.G.).”

www.pnas.org/cgi/doi/10.1073/pnas.1401985111

PNAS | March 4, 2014 | vol. 111 | no. 9 | 3645

CORR

ECTIONS

Mapping the receptor site for α-scorpion toxinson a Na+ channel voltage sensorJinti Wanga, Vladimir Yarov-Yarovoya,1, Roy Kahnb, Dalia Gordonb, Michael Gurevitzb, Todd Scheuera,and William A. Catteralla,2

aDepartment of Pharmacology, School of Medicine, University of Washington, Seattle, WA 98195-7280; and bDepartment of Plant Sciences, The GeorgeS. Wise Faculty of Life Sciences, Tel-Aviv University, Ramat-Aviv 69978, Tel-Aviv, Israel

Contributed by William A. Catterall, July 29, 2011 (sent for review May 10, 2011)

The α-scorpions toxins bind to the resting state of Na+ channelsand inhibit fast inactivation by interaction with a receptor siteformed by domains I and IV. Mutants T1560A, F1610A, andE1613A in domain IV had lower affinities for Leiurus quinquestria-tus hebraeus toxin II (LqhII), and mutant E1613R had w73-foldlower affinity. Toxin dissociation was accelerated by depolariza-tion and increased by these mutations, whereas association ratesat negative membrane potentials were not changed. These resultsindicate that Thr1560 in the S1-S2 loop, Phe1610 in the S3 seg-ment, and Glu1613 in the S3-S4 loop in domain IV participate intoxin binding. T393A in the SS2-S6 loop in domain I also had loweraffinity for LqhII, indicating that this extracellular loop may forma secondary component of the receptor site. Analysis with theRosetta-Membrane algorithm resulted in a model of LqhII bindingto the voltage sensor in a resting state, in which amino acid resi-dues in an extracellular cleft formed by the S1-S2 and S3-S4 loopsin domain IV interact with two faces of the wedge-shaped LqhIImolecule. The conserved gating charges in the S4 segment are inan inward position and form ion pairs with negatively chargedamino acid residues in the S2 and S3 segments of the voltagesensor. This model defines the structure of the resting state ofa voltage sensor of Na+ channels and reveals its mode of interac-tion with a gating modifier toxin.

mutagenesis | voltage-gated sodium channel

Voltage-gated Na+ channels initiate the action potential innerve and muscle (1, 2). They activate rapidly upon de-

polarization and then inactivate within milliseconds (1, 2). Sev-eral families of neurotoxins bind to six distinct receptor sites onNa+ channels and alter their function (3). α-Scorpion toxins andβ-scorpion toxins bind to neurotoxin receptor sites 3 or 4 andinhibit fast inactivation or enhance activation, respectively (4).Their combined effects activate Na+ channels at negative mem-brane potentials and prolong their opening, leading to repetitivefiring, depolarization, and conduction block. As a consequence,these toxins kill organisms by inducing paralysis and causingcardiac arrhythmia. Because of their high affinity and specificity,the gating-modifier toxins are powerful tools to study structureand function of Na+ channels (4, 5). Identification of the mo-lecular determinants for binding of toxins that modify activationor inactivation will provide important information about struc-tural mechanisms of channel gating and give insight into designof therapeutic agents that could prevent toxin action.Voltage-gated Na+ channels in eukaryotes are heteromeric in-

tegral membrane proteins composed of a pore-forming α-subunitof 220 to 260 kDa and auxiliary β-subunits of 30 to 40 kDa (2).The α-subunit contains the ion pore, voltage sensors, and neuro-toxin binding sites 1 to 6 (2). It consists of four homologousdomains (I to IV) that each contains six transmembrane seg-ments (S1–S6) and a short membrane-penetrating segment(SS1–SS2) between segments S5 and S6. The S5 and S6 segmentsform the central pore and the SS1 and SS2 segments form theion selectivity filter at its extracellular end. The S1 to S4 seg-ments form the voltage sensor. The S4 segment in each domain

contains repeated motifs of a positively charged amino acid res-idue followed by two hydrophobic residues. These positivecharges in the S4 segments serve as gating charges and moveacross the membrane electric field in response to changes inmembrane potential, initiating conformational changes that openand close the pore (6, 7). The cytoplasmic linker connectingdomains III and IV forms the fast inactivation gate, which folds inand blocks the pore during fast inactivation (2, 8–10).α-Scorpion toxins are single polypeptides cross-linked by four

disulfide bridges, which form a highly conserved, tightly foldedcore consisting of an α-helix and a three-stranded β-sheet (11).Neurotoxin receptor site 3 is formed by amino acid residues inthe S5-S6 extracellular linker in domain I and the extracellularlinker connecting segments S3 and S4 in domain IV, which alsocontributes to the receptor sites for structurally unrelated toxinsfrom sea anemone and spider venoms (12–16). α-Scorpion toxinsbind in a state-dependent manner, with high affinity for theresting state (17, 18). The voltage-sensor trapping model of toxinaction posits that binding of α-scorpion toxins prevents thenormal outward movement of the IVS4 segment in response todepolarization, thereby trapping it in an inward position anduncoupling activation from fast inactivation (4). Prolonged de-polarization overcomes the effect of toxin binding and drivesdissociation as the IVS4 segment moves outward (4, 17).Previous work showed that the conserved acidic amino acid

residueE1613 in IVS3-S4 is involved in binding ofα-scorpion toxins(14), and subsequent studies showed that analogous amino acidresidues are important for binding of other toxins to voltage-gatedK+ and Ca2+ channels (19–22). In this study, we identified aminoacid residues in the ISS2-S6 loop, IVS1-S2 loop, and IVS3 segmentas components of neurotoxin receptor site 3, and this informationwas used to generate a molecular model of the toxin-receptorcomplex in the resting state using the Rosetta Membrane program.

ResultsInhibition of Na+ Channel Inactivation by Recombinant α-ScorpionToxins. To determine the effect of binding of α-scorpion toxins,WT NaV1.2a channels were transiently expressed in the embry-onic kidney cell line tsA-201 and analyzed by whole-cell voltageclamp. To determine the Kd, cells were voltage clamped to −100mV, depolarized to 0 mV to measure the Na+ current, and testedfor modification of fast inactivation by LqhII (Leiurus quinques-triatus hebraeus, toxin II) (23). LqhII was synthesized by recom-binant expression in Escherichia coli, purified, and renatured

Author contributions: J.W., V.Y.-Y., T.S., and W.A.C. designed research; J.W., V.Y.-Y., andT.S. performed research; R.K., D.G., and M.G. contributed new reagents/analytic tools;J.W., V.Y.-Y., T.S., and W.A.C. analyzed data; and J.W., V.Y.-Y., R.K., D.G., M.G., T.S., andW.A.C. wrote the paper.

The authors declare no conflict of interest.1Present address: Department of Physiology and Membrane Biology, University of Cali-fornia, Davis, CA 95616.

2To whom correspondence should be addressed. E-mail: wcatt@u.washington.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1112320108/-/DCSupplemental.

15426e15431 | PNAS | September 13, 2011 | vol. 108 | no. 37 www.pnas.org/cgi/doi/10.1073/pnas.1112320108

(Materials and Methods and ref. 24). In the absence of toxin, Na+

currents activated and inactivated completely within a few milli-seconds (Fig. 1A). Addition of 1 nM LqhII slowed the inactivationof a fraction of WT channels, and 100 nM slowed inactivation ofmost WT channels (Fig. 1A). The fraction of conductanceremaining at 4 ms after the peak, which is proportional to thenumber of channels modified by LqhII (Fig. 1A), was plottedversus toxin concentration (Fig. 1B). These data were fit witha Hill equation (Eq. S1) using a fixed Hill coefficient of 1.0. Fromthis analysis, the Kd for LqhII binding was estimated to be 0.47 ±0.05 nM forWT, consistent with other related scorpion toxins (14,17, 25, 26). The voltage-dependence of channel activation andinactivation was very similar in the absence and the presence ofLqhII (Tables S1 and S2).

Molecular Mapping of Amino Acid Residues in the α-Scorpion ToxinReceptor Site in Domain IV. Previous work with site-directed anti-bodies showed that α-scorpion toxins bind to the extracellularloops connecting S5 and S6 in domains I and IV (12, 13), and site-directed mutagenesis studies showed that the IVS3-S4 loop con-tains amino acid residues that participate in toxin binding, withGlu1613 as a primary binding determinant (14). We reevaluatedthe effects of mutations of Glu1613 in the IVS3-S4 loop usingrecombinant LqhII. The voltage-dependence of activation ofE1613A and E1613R was not significantly changed (Table S1),whereas the voltage-dependence of inactivation was negativelyshifted for E1613A but not E1613R (Table S2). The Na+ currentsconducted by WT, E1613A, or E1613R appeared identical in the

absence of toxin (Fig. 1). However, 1 nMLqhII was less effective inslowing inactivation of E1613A than WT, and even 10 nM LqhIIwas less effective in impairing inactivation of E1613R than 1 nMLqhII was on WT (Fig. 1 A, C, and D). Maximal modification wasreached at 100 nM LqhII for both E1613A and E1613R mutants.The E1613A mutation reduced the affinity for LqhII w3.9-fold,and E1613R reduced the affinityw73-fold (Fig. 1G). These resultsare consistent with the previous conclusion that Glu1613 is animportant component of the receptor site for α-scorpion toxins.To develop an accurate molecular model of the receptor site,

it is necessary to identify all of the amino acid residues in dif-ferent extracellular loops that are required for toxin binding. Tocomplete mapping of the receptor site, we studied mutations ofthe remaining amino acid residues in the IVS1-S2 and IVS3-S4loops. Hydrophobic amino acids were converted to alanine toremove side chain interactions with minimum perturbation ofsecondary structure. Charged amino acids were replaced withamides of similar size: glutamine for glutamate, lysine, and ar-ginine, and asparagine for aspartate. Additional residues in theS1, S2, S3, and S4 segments in domain IV were also screened byalanine-scanning mutagenesis. We present the functional analy-sis of the most informative mutants first, followed by a 2D mapand 3D model of the toxin receptor site.Mutant F1610A in the IVS3 segment had reduced affinity for

LqhII (Fig. 1E), with w10.7-fold increase in Kd compared withWT (Fig. 1H). The voltage-dependence of activation was similarfor WT and F1610A (Table S1), whereas the voltage dependenceof fast inactivation of F1610A was negatively shifted by w13 mV(Table S2). Phe1610 is located 3-aa residues on the N-terminalside of Glu1613 in the IVS3-S4 loop. The decrease in affinitywith alanine mutations at both sites suggests that these sidechains form an essential part of the toxin-binding site.The results for mutant T1560A in IVS1-S2 also revealed a sig-

nificant reduction in affinity for LqhII (Fig. 1F), with no significantchange in the voltage-dependence of channel activation or fastinactivation (Tables S1 and S2). The concentration-responsecurve was shifted to higher concentration, with a Kd w5.9-foldhigher than WT (Fig. 1H). This decrease in affinity suggests thatthe side chain of this residue may interact with the toxin directly.

Kinetics of Dissociation of LqhII. Binding of α-scorpion toxins isreversed by strong depolarizations that activate the voltagesensor (14, 17). Cells expressing each channel type were voltage-clamped to −100 mV to allow maximum binding, and thendepolarized to 100 mV for 1 to 1,024 ms to induce dissociation,hyperpolarized to −100 mV for 20 ms to reverse inactivation,and depolarized to 0 mV for measurement of Na+ current (Fig.2A). For WT channels in the presence of 100 nM LqhII, pro-gressively longer depolarizations caused faster and more com-plete Na+ channel inactivation, indicating LqhII dissociation(Fig. 2A). Dissociation of toxin and loss of toxin action werevirtually complete after a 256-ms depolarization to 100 mV (Fig.2B). For WT in the presence of 100 nM LqhII, the dissociationtime constant was 95 ± 3 ms at 100 mV, 135 ± 17 ms at 80 mV,and 288 ± 53 ms at 60 mV (Fig. 2B). These results show that therate of toxin dissociation is voltage-dependent, with more rapiddissociation during stronger depolarization (Fig. 2C).The same protocol was used to measure the dissociation rate

for the three mutant channels (Fig. 2C). The WT channel hadthe slowest dissociation at all membrane potentials. E1613A andF1610A mutations decreased the time constant by w4.3- andw6.7-fold, respectively, at 100 mV. The effect was smaller forT1560A, with w1.9-fold decrease in time constant at 100 mV.The dissociation time constants for all three mutant channelswere voltage-dependent, with lower values at more depolarizedpotentials (Fig. 2C). The time course of toxin dissociation wasalso determined in the presence of 30 nM toxin. The dissociationtime constants showed strong voltage-dependence but little

A B

DC

E

G H

F

Fig. 1. Electrophysiological effect of LqhII on WT and mutant NaV1.2channels. (A, and C–F). Normalized voltage-clamp current traces from tsA-201 cells expressing WT and mutant channels in the absence (Control, black)and in the presence of 1 nM (red), 10 nM (green), or 100 nM LqhII (blue).Cells expressing NaV1.2a channel were held at −100 mV and Na currentswere elicited with a 30-ms step to 0 mV. (A) WT, (C) E1613A, (D) E1613R, (E)F1610A. (F) T1560A. (B and G–H). Concentration-response relations for LqhIIblock of fast inactivation from cells expressing WT and mutant channels. (B)WT (Kd = 0.47 ± 0.05 nM, n = 3–8). (G) E1613A (Kd = 1.8 ± 0.4 nM, n = 3–5)and E1613R (Kd = 34.3 ± 5.7 nM, n = 3–4). (H) F1610A (Kd = 5.0 ± 1.8 nM,n = 3–6) and T1560A (Kd = 2.8 ± 0.8 nM, n = 4–8).

Wang et al. PNAS | September 13, 2011 | vol. 108 | no. 37 | 15427

NEU

ROSC

IENCE

concentration-dependence (Fig. 2D), indicating that the disso-ciation step is a state-dependent, unimolecular reaction.

Kinetics of Reassociation of LqhII. The rates of toxin associationwere assessed for WT and mutants using a 200-ms prepulse to100 mV to cause toxin dissociation, followed by progressivelylonger hyperpolarizing pulses to allow toxin rebinding, and a fi-nal test depolarization to 0 mV to assess LqhII action (Fig. 3A).For WT channels, toxin dissociated during the prepulse, as in-dicated by the rapid inactivation when only 50 ms were allowedfor rebinding (Fig. 3B). Toxin gradually rebound after longerrepolarizations with complete reassociation at 12.8 s (Fig. 3B).The time course of toxin reassociation was exponential and in-dependent of membrane potential (Fig. 3 B and C).The same protocol was used to measure the association rate for

the three mutants at 100 nM toxin (Fig. 3C). The time constantsfor toxin binding to mutant channels were similar toWT and werevoltage-independent. Similar experiments at 30 nM LqhII showedw3.3-, w4.5-, and w5.7-fold decreases compared with 100 nMtoxin, but little change fromWT (Fig. 3D). Thus, the time constantfor binding showed strong concentration-dependence. However,there was no effect of the mutations and little or no voltage-de-pendence, indicating that the association step is a diffusion-con-trolled bimolecular reaction with the resting state.

Amino Acid Residues Required for Toxin Binding in Domain I. Anti-bodies directed against the extracellular SS2-S6 loop in domain Iwere effective in immunoprecipitating peptides covalently la-beled by a photoreactive α-scorpion toxin and in reducingbinding of the α-scorpion toxin LqTxV (Leiurus quinquestriatustoxin V) (12, 13). Moreover, recent structural studies of KV1.2channels show that the equivalent of the SS2-S6 loop interactswith the voltage sensor of the neighboring subunit, implying in-teraction of the ISS2-S6 loop of Na+ channels with the voltagesensor of domain IV (27). To identify possible sites of interactionof LqhII in the ISS2-S6 loop, we tested the amino acid residuesfrom F385 to Y401 by alanine-scanning mutagenesis. An in-

crease ofw3.4-fold in Kd was observed for mutant T393A (Fig. 4A and B), resulting in a rightward shift in concentration-responsecurve, whereas this mutation had no effect on the voltage-dependence of activation or inactivation (Tables S1 and S2).Mutation of other amino acid residues in the ISS2-S6 loop hadlittle or no effect on toxin binding, so this loop may form a sec-ondary component of the toxin-binding site.The rates of dissociation of LqhII from T393A channels were

measured in the presence of 30 or 100 nM toxin using the same

A

C

B

D

Fig. 2. Voltage-dependent dissociation rates of LqhII. (A) Traces demon-strating time-dependent dissociation of LqhII at 100 mV. Cells expressing WTNaV1.2 channels were incubated in 100 nM LqhII for 6 min at a holding po-tential of −100 mV to allow binding. The rate of toxin dissociation was de-termined with the illustrated pulse paradigm by stepping to a depolarizingpulse of 100, 80, or 60mV, for 1 to 1,024ms, returning to −100mV for 20ms toallow recovery from fast inactivation and then assessing the effect of thedepolarizing pulse with a 10-ms test pulse to 0 mV (test). (B) Time course ofdissociationof 100nMLqhII from cells expressingWT channels at 100mV (τ = 95± 3ms, n = 3), 80mV (τ = 135 ± 17ms, n = 3), and 60mV (τ = 289± 53ms, n = 3).Time constants were determined by fits of monoexponential functions to thedata. (C andD) Time constants of dissociation as a function of potential forWT,E1613A, F1610A, and T1560A channels. (C) 100 nM LqhII. (D) 30 nM LqhII.

A B

DC

Fig. 3. Rates of LqhII association after repolarization. (A) Current tracesdemonstrating time-dependent reassociation of LqhII at −100 mV. Cellsexpressing WT channels were incubated in 100 nM LqhII. Toxin was disso-ciated from the channel using a 200-ms depolarization to 100 mV. The rateof toxin association was measured by hyperpolarizing to −120, −100, or −80mV for increasing times. The effect of repolarization on toxin action wasassessed with a 10-ms test pulse to 0 mV (test). (B) The time course of as-sociation of 100 nM LqhII from cells expressing WT channel was determinedat −120 mV (τ = 1.6 ± 0.01 s, n = 3), −100 mV (τ = 1.2 ± 0.2 s, n = 3), and −80mV (τ = 1.2 ± 0.3 s, n = 3). (C and D) Time constants of association asa function of voltage for WT, E1613A, F1610A, and T1560A channels. (C) 100nM LqhII. (D) 30 nM LqhII.

A

C

B

D

Fig. 4. Characterization of mutation T393A in domain ISSII-S6. (A) Nor-malized current traces during steps to 0 mV from cells expressing T393Achannels in the absence (control) and in the presence of 1 nM and 100 nMLqhII. (B) Concentration-response relations for LqhII removal of fast in-activation in cells expressing T393A channels (Kd = 1.6 ± 0.5 nM, n = 2–5). (C)Time constants for LqhII dissociation for T393A and WT channels as a func-tion of potential in the presence of 30 and 100 nM LqhII (n = 3). (D) Timeconstants of association for T393A and WT channels in the presence of 30 or100 nM LqhII (n = 3).

15428 | www.pnas.org/cgi/doi/10.1073/pnas.1112320108 Wang et al.

protocol illustrated in Fig. 2A. The rates of toxin dissociationwere voltage-dependent and generally similar to WT for T393Achannels (Fig. 4C). A slight increase in time constant for disso-ciation of LqhII from T393A was observed at 60 mV for bothtoxin concentrations (Fig. 4C).The rates of toxin association were assessed for T393A

channels using the same protocol as in Fig. 3A. The associationtime constant was voltage-independent and showed w1.2-,w1.7-, and w2.2-fold increase compared with WT at −120,−100, and −80 mV using 100 nM toxin (Fig. 4D). The timecourse of toxin association also showed a similar increase com-pared with WT when measured at 30 nM toxin. Thus, this mu-tation causes primarily a decrease in the rate of toxin associationat negative membrane potentials.

Molecular Map of Neurotoxin Receptor Site 3. The Kd values for allNa+ channel mutants are summarized in Fig. 5, together withdata from our previous work (14). Most mutants in domain IVwere identical to WT, with the exception of mutations of threeamino acid residues: T1560 in the S1-S2 loop, F1610 near theextracellular end of the S3 segment, and E1613 in S3-S4 loop.These results suggest that the side chains of these amino acids indomain IV form part of the binding pocket for α-scorpion toxins.For the SS2-S6 loop in domain I, only T393A caused a sub-

stantial increase in Kd. The structure of KV1.2 channels revealsthat the voltage sensor from one subunit interacts closely withthe pore of the adjacent subunit in a clockwise direction (27),implying that the ISS2-S6 loop is located adjacent to the IVS3-S4loop in Na+ channels. Therefore, it is likely that the amino acidresidues in IVS1-S2 and IVS3-S4 form the primary receptor sitefor α-scorpion toxins, whereas Thr393 in the ISS2-S6 loop pro-vides a secondary site of interaction.

Structural Model of the α-Scorpion Toxin/NaV1.2 Channel Complex.We have developed a structural model of the LqhII-NaV1.2 com-plex using the Rosetta docking method (28, 29) and resting statemodels of the KV1.2 channel (30–32) as a template for the NaV1.2domain IV voltage sensor (Fig. 6; see Fig. S1 for the sequencealignment between the voltage sensor in domain IV of NaV1.2 and

the voltage sensor in the KV1.2 channel). Ourmodel shows that theLqhII toxin has an extensive interface of interaction with the ex-tracellular water-accessible cavity of the domain IV voltage sensor(Fig. 6). The overall orientation of the α-scorpion toxin LqhIIrelative to the voltage sensor (Fig. 6) is similar to the orientation ofthe structurally homologousCentruroides suffusus suffusus toxin IVrelative to the voltage sensor in domain II (33). The wedge-shapedLqhII molecule sits in the cleft between the IVS1-S2 and IVS3-S4loops and makes intimate contact on two sides of the wedgethrough interactions with amino acid side chains (Fig. 6). LqhIIresidues Phe15, Trp38, and Asn44 in the core domain of LqhII areburied deeply in the cleft between IVS1-S2 and IVS3-S4 (Fig. 6 Band D). Evidently, interactions of the core domain of LqhII withamino acid residues in this cleft in the voltage sensor contribute themajority of the binding energy that drives formation of the toxin-receptor complex. The key amino acid residues in the Na+ channelare also positioned to interact with the bound toxin (Fig. 6 A andC). Thr1560 in IVS1-S2 interacts with one face of the wedge-sha-ped toxin and Glu1613 interacts with the other face. In the moststable channel conformation shown in Fig. 6, the side chain ofPhe1610 points away from the bound LqhII. However, a slightlydifferent conformation in which the IVS3 helix is unwound be-ginning at Phe1610 allows its interaction with Phe15 and Trp38 inthe core domain of LqhII and retains potential interactions ofThr1560 and E1613 with bound LqhII. It is conceivable that toxinbinding induces unwinding of the outermost turn of the IVS3 helixand that voltage-dependent outward movement of the S4 segmentinduces rewinding of the outer end of the S3 segment into helicalconformation and consequent dissociation of the bound toxin.Alternatively, it is possible that this unwound conformation of theouter end of S3 is found in the resting state of the channel, eventhough it appears to be less stable in Rosetta modeling.

DiscussionAmino Acid Residues in S1-S2, S3, and S3-S4 Segments in Domain IVForm the Primary Receptor Site for α-Scorpion Toxins. We foundpreviously that the acidic amino acid residue Glu1613 in the S3-S4 extracellular loop in domain IV contributes significantly toα-scorpion toxin binding (14). In this work, we found two addi-

Fig. 5. Summary of effects of mutations in domains I and IV on LqhII affinity. Fold-changes in Kd measured either by displacement of bound 125I-LqTxV withunlabeled LqTxV adapted from our previous work (open bars) (14) or fold-changes in Kd, as determined using electrophysiology experiments described inResults (filled bars, n = 2–13). Different substitutions for E1613 are shown as stacked bars of different colors. Ala and Gly residues (A396, A397, G398, G1608,and A1612) were not studied. R395Q and R395H gave no Na+ current.

Wang et al. PNAS | September 13, 2011 | vol. 108 | no. 37 | 15429

NEU

ROSC

IENCE

tional nearby mutations that caused substantial reduction inbinding affinity for LqhII: T1560A in IVS1-S2 and F1610A nearthe extracellular end of IVS3. It is likely that these three aminoacid residues form the primary component of the receptor sitefor α-scorpion toxins. Amino acid residues aligned with Phe1610and Glu1613 contribute to the ω-agatoxin receptor site in CaV2.1channels (19, 20) and the hanatoxin and grammotoxin receptorsites in KV channels (21, 22). Thus, these studies reveal a com-mon five-residue motif of two hydrophobic residues in positions1 and 2 followed by Glu in position 5 (ΦΦXXE) that may con-tribute to the actions of all of these gating modifier toxins (4).

Amino Acid Residues in Domain I Form a Secondary Site of Interactionfor α-Scorpion Toxins. Antipeptide antibody experiments sug-gested that neurotoxin receptor site 3 is formed by amino acidresidues in the segment between amino acids 382 and 400 in theextracellular S5-S6 loop in domain I, as well as by the S3-S4 loopin domain IV (12, 13). Our experiments identify Thr393 as aprimary point of interaction in this segment of the channel. Thestructure of KV1.2 channels revealed that the voltage-sensing

module from one subunit interacts closely with the pore-formingmodule of the adjacent subunit in a clockwise direction when theKV1.2 channel structure is viewed from the extracellular side ofthe membrane (27). This organization would place the IS5-S6segment adjacent to the IVS1-S2 and IVS3-S4 segments in NaVchannels. β-Scorpion toxins, which are similar to α-scorpiontoxins in structure but enhance activation by binding to the S3-S4loop on domain II, have been found to interact with amino acidresidues in the SS2-S6 loop in the adjacent pore-forming modulein domain III, as well as with amino acid residues in IIS3-S4 (34).Consistent with these previous studies, we found one amino acidresidue (Thr393) in the ISS2-S6 loop that contributes signifi-cantly to α-scorpion toxin binding, although its contribution wassubstantially less than IVS1-S2 and IVS3-S4. These results in-dicate that α-scorpion toxins have a secondary site of interactionwith this segment of domain I. The conformation of LqhII boundto domain IV (Fig. 6) places it in position for the essential aminoacid residues in its N- and C-terminal domains to interact withamino acid residues in the S6-SS2 segment of domain I.

Voltage-Dependence of α-Scorpion Toxin Binding. α-Scorpion toxinsbind with high affinity to the resting state, and their binding isreversed by depolarizations that activate Na+ channels (17, 18).Recombinant LqhII dissociates more rapidly at more depolar-ized potentials, as previously observed for other α-scorpiontoxins (14, 17, 26). These results indicate that the conformationof the toxin-receptor complex changes during depolarizationfrom a high-affinity conformation in the resting state to a lowaffinity conformation in the activated state. Evidently, high-affinity toxin binding to the resting state of the voltage sensoropposes the outward movement of the S4 segment in domain IVin response to depolarization, thereby uncoupling channelopening from fast inactivation. However, prolonged depolarizationcan drive voltage-dependent dissociation of bound α-scorpiontoxin (17, 18). Mutations T1560A, F1610A, and E1613A signif-icantly decreased the dissociation time, demonstrating that ala-nine mutations at these sites reduced the stability of the toxin-channel interaction.By comparing the time constants for LqhII binding to WT

channels at different potentials, we found that toxin binding tothe resting state of Na+ channels is voltage-independent. Theseresults indicate that the charged amino acid residues of the toxindo not enter the transmembrane electric field during binding.Furthermore, the mutations T1560A, F1610A, and E1613A didnot change the association time constant significantly at thesehyperpolarized potentials. Therefore, the changes in toxinbinding affinity to the resting state caused by these mutations indomain IV must result from changes in the rate of dissociation.However, the T393A mutation in domain I increased the asso-ciation time constant at hyperpolarized potentials, indicatingthat this mutation may affect toxin binding by a differentmechanism. The decrease in the rate of toxin association causedby mutation T393A raises the possibility that the –CH2OH groupof Thr393 may catalyze the movement of LqhII into its receptorsite in addition to its contribution to binding affinity.

Structure of the Complex of α-Scorpion Toxins with the Resting Stateof the Voltage Sensor in Domain IV. The structures of voltage-gatedion channels that have been determined to date have had acti-vated voltage sensors (27, 35–37), which is expected because theactivated state is favored at a membrane potential of 0 mV as ina protein crystal. We have developed a molecular model of theresting state of the voltage sensor of KV1.2 channels using theRosetta Membrane modeling system (30, 32) and the crystalstructure of the open state (36, 37), and we have used theRosetta Membrane Docking algorithm (29) to construct a modelof the α-scorpion toxin-receptor complex. Our structural model(Fig. 6) reveals a close fit of the toxin in the cleft between the

Fig. 6. Full atomandmolecular surface representation of LqhII binding to thevoltage-sensing domain IV of NaV1.2. Segments S1 through S4 of the voltage-sensing domain colored individually and labeled. (A and B) Side view of thestructural model with the voltage-sensing domain segments S1 and S4 on thefront. (C and D) Side view of the structural model with the voltage-sensingdomain segments and with S2 and S3 on the front (rotated 180° when viewedfrom the extracellular side of the membrane compared with orientationshown in A and B). Side chains of key residues for LqhII-NaV1.2 interaction areshown in space-filling representation and all other side chains shown in stickrepresentation. A probe radius of 1.4 Åwas used to scan themolecular surfaceof each structural model. This figure was generated using Chimera (43).

15430 | www.pnas.org/cgi/doi/10.1073/pnas.1112320108 Wang et al.

S1-S2 and S3-S4 extracellular loops of the voltage sensor indomain IV. This docking position closely resembles the positionof the β-scorpion toxin CssIV when bound to its receptor site inthe voltage sensor of domain II of Na+ channels (33).It is noteworthy that our structural model does not reveal

interactions of amino acid residues in the N- and C-terminaldomains of LqhII with the voltage sensor, even though theseamino acid residues are important for toxin binding (23, 24, 38,39). We speculate that these amino acid residues may interactwith the SS2-S6 loop in domain I, which may form a secondarysite for toxin interaction. Our current models for the structure ofthe voltage sensor do not define its spatial relationship with theSS2-S6 extracellular loop in the complete structure. Therefore,a model of this part of the toxin-receptor complex must awaitdetermination of the 3D structure of the resting state of a voltagesensor in a complete voltage-gated ion channel or developmentof models for the entire four-domain Na+ channel structure.

Implications of Voltage-Sensor Structure for its Mechanism of Action.Our model provides an initial view of the resting state confor-mation of a Na+ channel voltage sensor. As expected from our

model of the resting state of KV1.2 channels (30, 32), the S4segment is in a transmembrane orientation in the resting state,and its gating charge-carrying arginine residues are in an inwardposition, poised to move outward upon depolarization of themembrane. The R2 gating charge interacts with Asn1567 on theextracellular side of the S2 segment. The R3 side chain is onthe intracellular side of the gating pore in position to interact withthe conserved negatively charged residues Glu1578 in S2 andAsp1598 in S3. The R4 and R5 gating charges are exposed on theintracellular side of the membrane. These molecular interactionsillustrate how the voltage-sensor structure stabilizes the gatingcharges in their inward position in the resting state and how bindingof LqhII traps the voltage sensor in its resting conformation.

Materials and MethodsLqhII was produced as described (24, 40). Structural modeling was carried outas described (30, 31, 41–43). Site-directed mutagenesis, cell culture andtransfection, and electrophysiological recordingwere carried out as described(44) using the recording solutions described in SI Materials and Methods.Additional technical details are presented in SI Materials and Methods.

1. Hodgkin AL, Huxley AF (1952) A quantitative description of membrane current and itsapplication to conduction and excitation in nerve. J Physiol 117:500e544.

2. Catterall WA (2000) From ionic currents to molecular mechanisms: The structure andfunction of voltage-gated sodium channels. Neuron 26:13e25.

3. Cestèle S, Catterall WA (2000) Molecular mechanisms of neurotoxin action on volt-age-gated sodium channels. Biochimie 82:883e892.

4. Catterall WA, et al. (2007) Voltage-gated ion channels and gating modifier toxins.Toxicon 49:124e141.

5. Bosmans F, Swartz KJ (2010) Targeting voltage sensors in sodium channels with spidertoxins. Trends Pharmacol Sci 31:175e182.

6. Armstrong CM (1981) Sodium channels and gating currents. Physiol Rev 61:644e683.7. Catterall WA (2010) Ion channel voltage sensors: Structure, function, and patho-

physiology. Neuron 67:915e928.8. Vassilev PM, Scheuer T, Catterall WA (1988) Identification of an intracellular peptide

segment involved in sodium channel inactivation. Science 241:1658e1661.9. Stühmer W, et al. (1989) Structural parts involved in activation and inactivation of the

sodium channel. Nature 339:597e603.10. West JW, et al. (1992) A cluster of hydrophobic amino acid residues required for fast

Na+-channel inactivation. Proc Natl Acad Sci USA 89:10910e10914.11. Housset D, Habersetzer-Rochat C, Astier JP, Fontecilla-Camps JC (1994) Crystal struc-

ture of toxin II from the scorpion Androctonus australis Hector refined at 1.3 Å res-olution. J Mol Biol 238:88e103.

12. Tejedor FJ, Catterall WA (1988) Site of covalent attachment of alpha-scorpion toxinderivatives in domain I of the sodium channel alpha subunit. Proc Natl Acad Sci USA85:8742e8746.

13. Thomsen WJ, Catterall WA (1989) Localization of the receptor site for alpha-scorpiontoxins by antibody mapping: Implications for sodium channel topology. Proc NatlAcad Sci USA 86:10161e10165.

14. Rogers JC, Qu Y, Tanada TN, Scheuer T, Catterall WA (1996) Molecular determinantsof high affinity binding of alpha-scorpion toxin and sea anemone toxin in the S3-S4extracellular loop in domain IV of the Na+ channel alpha subunit. J Biol Chem 271:15950e15962.

15. Bosmans F, Martin-Eauclaire MF, Swartz KJ (2008) Deconstructing voltage sensorfunction and pharmacology in sodium channels. Nature 456:202e208.

16. Gilles N, et al. (2002) Variations in receptor site-3 on rat brain and insect sodiumchannels highlighted by binding of a funnel-web spider delta-atracotoxin. Eur J Bi-ochem 269:1500e1510.

17. Catterall WA (1977) Membrane potential-dependent binding of scorpion toxin to theaction potential Na+ ionophore. Studies with a toxin derivative prepared by lacto-peroxidase-catalyzed iodination. J Biol Chem 252:8660e8668.

18. Catterall WA (1979) Binding of scorpion toxin to receptor sites associated with sodiumchannels in frog muscle. Correlation of voltage-dependent binding with activation.J Gen Physiol 74:375e391.

19. Bourinet E, et al. (1999) Splicing of alpha 1A subunit gene generates phenotypicvariants of P- and Q-type calcium channels. Nat Neurosci 2:407e415.

20. Winterfield JR, Swartz KJ (2000) A hot spot for the interaction of gating modifiertoxins with voltage-dependent ion channels. J Gen Physiol 116:637e644.

21. Li-Smerin Y, Swartz KJ (1998) Gating modifier toxins reveal a conserved structuralmotif in voltage-gated Ca2+ and K+ channels. Proc Natl Acad Sci USA 95:8585e8589.

22. Li-Smerin Y, Swartz KJ (2000) Localization and molecular determinants of the Ha-natoxin receptors on the voltage-sensing domains of a K+ channel. J Gen Physiol 115:673e684.

23. Gordon D, Gurevitz M (2003) The selectivity of scorpion alpha-toxins for sodium

channel subtypes is determined by subtle variations at the interacting surface. Tox-

icon 41:125e128.24. Karbat I, et al. (2004) Molecular basis of the high insecticidal potency of scorpion

alpha-toxins. J Biol Chem 279:31679e31686.25. Chen H, et al. (2002) Differential sensitivity of sodium channels from the central and

peripheral nervous system to the scorpion toxins Lqh-2 and Lqh-3. Eur J Neurosci 16:

767e770.26. Leipold E, Lu S, Gordon D, Hansel A, Heinemann SH (2004) Combinatorial interaction

of scorpion toxins Lqh-2, Lqh-3, and LqhαIT with sodium channel receptor sites-3. Mol

Pharmacol 65:685e691.27. Long SB, Tao X, Campbell EB, MacKinnon R (2007) Atomic structure of a voltage-

dependent K+ channel in a lipid membrane-like environment. Nature 450:376e382.28. Gray JJ, et al. (2003) Protein-protein docking with simultaneous optimization of rigid-

body displacement and side-chain conformations. J Mol Biol 331:281e299.29. Wang C, Bradley P, Baker D (2007) Protein-protein docking with backbone flexibility.

J Mol Biol 373:503e519.30. Yarov-Yarovoy V, Baker D, Catterall WA (2006) Voltage sensor conformations in the

open and closed states in ROSETTA structural models of K+ channels. Proc Natl Acad

Sci USA 103:7292e7297.31. Yarov-Yarovoy V, Schonbrun J, Baker D (2006) Multipass membrane protein structure

prediction using Rosetta. Proteins 62:1010e1025.32. Pathak MM, et al. (2007) Closing in on the resting state of the Shaker K+ channel.

Neuron 56:124e140.33. Cestèle S, et al. (2006) Structure and function of the voltage sensor of sodium

channels probed by a beta-scorpion toxin. J Biol Chem 281:21332e21344.34. Leipold E, Hansel A, Borges A, Heinemann SH (2006) Subtype specificity of scorpion

beta-toxin Tz1 interaction with voltage-gated sodium channels is determined by the

pore loop of domain 3. Mol Pharmacol 70:340e347.35. Jiang Y, et al. (2002) Crystal structure and mechanism of a calcium-gated potassium

channel. Nature 417:515e522.36. Long SB, Campbell EB, Mackinnon R (2005) Crystal structure of a mammalian voltage-

dependent Shaker family K+ channel. Science 309:897e903.37. Long SB, Campbell EB, Mackinnon R (2005) Voltage sensor of Kv1.2: Structural basis of

electromechanical coupling. Science 309:903e908.38. Karbat I, et al. (2007) The unique pharmacology of the scorpion alpha-like toxin Lqh3

is associated with its flexible C-tail. FEBS J 274:1918e1931.39. Zilberberg N, et al. (1997) Identification of structural elements of a scorpion alpha-

neurotoxin important for receptor site recognition. J Biol Chem 272:14810e14816.40. Kahn R, et al. (2009) Molecular requirements for recognition of brain voltage-gated

sodium channels by scorpion alpha-toxins. J Biol Chem 284:20684e20691.41. Jeanmougin F, Thompson JD, Gouy M, Higgins DG, Gibson TJ (1998) Multiple se-

quence alignment with Clustal X. Trends Biochem Sci 23:403e405.42. Rohl CA, Strauss CE, Misura KM, Baker D (2004) Protein structure prediction using

Rosetta. Methods Enzymol 383:66e93.43. Pettersen EF, et al. (2004) UCSF Chimera—A visualization system for exploratory re-

search and analysis. J Comput Chem 25:1605e1612.44. Beacham D, Ahn M, Catterall WA, Scheuer T (2007) Sites and molecular mechanisms

of modulation of Nav1.2 channels by Fyn tyrosine kinase. J Neurosci 27:11543e11551.

Wang et al. PNAS | September 13, 2011 | vol. 108 | no. 37 | 15431

NEU

ROSC

IENCE