RESEARCH ARTICLE◥

ION CHANNELS

Structures of human Nav1.7 channelin complex with auxiliary subunitsand animal toxinsHuaizong Shen1,2,3*, Dongliang Liu1,2,3*, Kun Wu4, Jianlin Lei2,5, Nieng Yan1,2,3†‡

Voltage-gated sodium channel Nav1.7 represents a promising target for painrelief. Here we report the cryo–electron microscopy structures of the humanNav1.7-b1-b2 complex bound to two combinations of pore blockers and gatingmodifier toxins (GMTs), tetrodotoxin with protoxin-II and saxitoxin withhuwentoxin-IV, both determined at overall resolutions of 3.2 angstroms. Thetwo structures are nearly identical except for minor shifts of voltage-sensingdomain II (VSDII), whose S3-S4 linker accommodates the two GMTs in a similarmanner. One additional protoxin-II sits on top of the S3-S4 linker in VSDIV. Thestructures may represent an inactivated state with all four VSDs “up” and theintracellular gate closed. The structures illuminate the path toward mechanisticunderstanding of the function and disease of Nav1.7 and establish the foundationfor structure-aided development of analgesics.

Among the nine subtypes of human voltage-gated sodium (Nav) channels, Nav1.7, whichis encoded by SCN9A and highly expressedin peripheral sensory neurons, has a directassociation with pain syndromes (1–4).

Mutations in Nav1.7 are found in many painsyndromes, including both extreme pain dis-order and indifference to pain (table S1) (5–10).An accurate structural model of Nav1.7 wouldfacilitate drug discovery for this promising target(11–13).Eukaryotic human Nav channels share high se-

quence similarity (fig. S1) (14, 15). Cryo–electronmicroscopy (cryo-EM) structures of representa-tive Nav channels from insect, electric eel, andhuman reveal identical architecture of the corea subunit (16–18). A single polypeptide chain,the a subunit, folds to four homologous repeats,each containing six transmembrane helices de-signated S1 to S6. The S1 to S4 segments in eachrepeat constitute the voltage-sensing domain (VSD)that attaches to the central ion-conducting pore

domain (PD) enclosed by the S5 and S6 helicesfrom the four repeats. The VSDs and PD seg-ments conform to the canonical domain-swappedassembly that is prevalent in the voltage-gatedion channel superfamily (19, 20). The sequencesbetween S5 and S6 comprise the selectivity filter(SF) that is sandwiched by two half-membrane-penetrating reentrant pore helices P1 and P2.Four distinct residues on the corresponding SFloci in the four repeats, Asp-Glu-Lys-Ala (DEKA),are the signature motif for Na+ selection (21) (fig.S1). Although the a subunit alone is sufficient forvoltage-dependent gating of ion permeation, it issubject to regulation by one or more b auxiliarysubunits (22). All four b subunits, b1 to b4, affectthe channel properties of Nav1.7, although b1 andb2 are commonly coexpressed with the Nav1.7 asubunit for biophysical characterization (22, 23).Structures of the Nav1.4-b1 complex from bothelectric eel and human reveal the interactiondetails between a and b1 (17, 18), but the bindingmode for other b subunits remains to be struc-turally elucidated.Nav channels are targeted by various natural

toxins and therapeutic drugs. These chemicalcompounds and peptides are generally classifiedinto two classes, pore blockers exemplified by thesmall-molecule neurotoxins tetrodotoxin (TTX)and saxitoxin (STX) and peptidic gatingmodifiertoxins (GMTs) that lock the channel in a par-ticular functional state, hence altering the fir-ing and propagation of action potentials (24, 25).Some toxins, particularly those with subtypespecificity, provide leads for potential thera-peutics for treatment of pain (26). The struc-tures of a Nav channel from American cockroach,NavPaS, bound to TTX, STX, and Dc1a, a GMTfrom the venom of desert bush spider, provide

a glimpse of the modulation of Nav channels byanimal toxins (27).Among the GMTs that target Nav channels,

two tarantula toxins, protoxin-II (ProTx-II) andhuwentoxin-IV (HWTX-IV), exhibit potent inhi-bition ofNa+ currentmediated byNav1.7 (28–32).Both toxins inhibit the activation ofNav channelsby binding to the linker between the S3 and S4segments (L3-4 loop) in the second VSD (VSDII),which is known as “site 4” (30, 33, 34). ProTx-IIwas also characterized to interact with VSDI andinhibit channel inactivation by binding to VSDIV

in the activated channel (31, 35). Structural eluci-dation of themolecular details of toxin interactionwith Nav1.7 may offer a path to potential drugdiscovery.In this study, we report the structures of

human Nav1.7 in the presence of both b1 and b2subunits at resolutions of 3.2 Å determined usingsingle-particle cryo-EM. Two combinations oftoxins were supplemented, ProTx-II with TTXand HWTX-IV with STX. For simplicity, we willrefer to these structures asNav1.7-PTandNav1.7-HS,respectively.

ResultsStructural determination of Nav1.7-PTand Nav1.7-HS

To simplify expression and purification of Nav1.7,we screened natural variants, each carrying adisease mutation (table S1). Out of 50 testedvariants, Nav1.7 (Glu

406→Lys, or E406K), a var-iant found in a patient with primary erythermal-gia (36), expressed at least threefold higher thanthewild-type channel.Whole-cell electrophysiolog-ical characterization of Nav1.7 in the presenceof b1 and b2 subunits shows that Nav1.7 (E406K)exhibits a hyperpolarizing shift in activation(−10.6 mV) and prolonged fast-inactivation du-ration compared with the wild type (Fig. 1A, fig.S2, and table S2). These observations are con-sistent with the alteration of channel proper-ties found for other mutants associated withprimary erythermalgia (7, 37–39).Out of 16 liters of human embryonic kidney–

293F (HEK-293F) cells cotransfected with plas-mids for Strep and FLAG twin-tagged Nav1.7and nontagged b1 and b2, about 100 to 200 mgof complex was obtained after tandem affinitypurification and size exclusion chromatography(SEC) (Fig. 1B). The proteins purified in thebuffer containing 0.04% (weight/volume) glyco-diosgenin (GDN, Anatrace) were further concen-trated to about 2 mg/ml. A combination ofProTx-II (50 mM) and TTX (50 mM), or HWTX-IV (50 mM) and STX (11 mM), was separatelyadded to the concentrated sample 30 min beforemaking cryo grids.Cryo-EM images were collected and processed

following standard protocols (Fig. 1, B and C, andfig. S3A). From 263,205 and 275,630 selectedparticles, respectively, the structures of Nav1.7-PTand Nav1.7-HS were both determined at 3.2-Åresolutions (Fig. 1D; figs. S3B and S4; and tableS3). For both complexes, the Nav1.7 core a sub-unit is well resolved. The extracellular immuno-globulin (Ig) domain and the single transmembrane

RESEARCH

Shen et al., Science 363, 1303–1308 (2019) 22 March 2019 1 of 6

1State Key Laboratory of Membrane Biology, TsinghuaUniversity, Beijing 100084, China. 2Beijing AdvancedInnovation Center for Structural Biology, TsinghuaUniversity, Beijing 100084, China. 3Tsinghua-Peking JointCenter for Life Sciences, School of Life Sciences andSchool of Medicine, Tsinghua University, Beijing 100084,China. 4Medical Research Center, Beijing Key Laboratoryof Cardiopulmonary Cerebral Resuscitation, Beijing Chao-Yang Hospital, Capital Medical University, Beijing 100020,China. 5Technology Center for Protein Sciences, Ministryof Education Key Laboratory of Protein Sciences, Schoolof Life Sciences, Tsinghua University, Beijing 100084,China.*These authors contributed equally to this work. †Present address:Department of Molecular Biology, Princeton University, Princeton,NJ 08544, USA.‡Corresponding author. Email: nyan@princeton.edu

on May 27, 2020

http://science.sciencem

ag.org/D

ownloaded from

helix (TM) of b1 are fully resolved, whereasonly the Ig domain of b2 is discernible to alower resolution of 4 to 5 Å (Fig. 1D). Twoblobs of densities at moderate resolutions arefound, respectively, above VSDII and VSDIV

in Nav1.7-PT, and one adheres to the extra-cellular edge of VSDII in Nav1.7-HS. The dif-ference between these two reconstructions andbetween them and our previous Nav1.4 map sug-gests that these additional densities belong toProTx-II andHWTX-IV. However, the densitiesare at peripheral regions and were only resolvedto moderate resolutions of ~5 Å, which does

not allow accurate docking of the toxin struc-tures (Fig. 2A and fig. S4A).For both structures, side chains could be as-

signed for residues 114 to 1768 except for theintracellular linkers I-II (residues 418 to 725)and II-III (residues 973 to 1174). Residues 20 to192 were built for the b1 subunit, and the crystalstructure of b2 (Protein Data Bank 5FEB) wasdocked into the density corresponding to b2-Ig with minor adjustment (Fig. 2B). Elevenglycosylation sites, six on Nav1.7, four on b1,and one on b2, were assigned (table S3). Asmost of the two structures containing Nav1.7-

b1-b2 is identical, we will not distinguish the twocomplexes during structural illustration unlessfor specific discussions. Several minor struc-tural shifts between Nav1.7-b1-b2 and Nav1.4-b1 (18) are described in the supplementarytext and fig. S5.

Closed intracellular gate

In the structures of Nav1.4 from both electric eeland human, the intracellular gate is penetratedby an elongated detergent molecule, digitonin orGDN (17, 18). No such density is observed in theEM maps of the Nav1.7 complexes, although a

Shen et al., Science 363, 1303–1308 (2019) 22 March 2019 2 of 6

Fig. 1. Structural determination of human Nav1.7 (E406K) incomplex with auxiliary subunits and toxins. (A) Electrophysiologicalcharacterization of wild-type Nav1.7 and the variant Nav1.7 (E406K)associated with primary erythermalgia in the presence of the b1and b2 auxiliary subunits. See materials and methods, fig. S2, andtable S2 for details. Nav1.7 (E406K) was used for structuraldetermination. For simplicity, we will refer to this variant as Nav1.7with regard to structural description. V, voltage; G, conductance; I,electric current; max, maximum value. (B) SEC purification ofthe human Nav1.7-b1-b2 complex. Mass spectrometric analysis(top) of the upper bands on the coomassie blue–stained SDS–polyacrylamide gel electrophoresis (bottom), indicated by thered arrow, confirmed the presence of all three subunits. The

concentrated protein complex was supplemented with the toxincombinations of either ProTx-II with TTX (PT) or HWTX-IVwith STX (HS) for cryo-sample preparation. UV, ultraviolet; mAU,milli–absorbance units. (C) Representative electron micrograph(top) and two-dimensional (2D) class averages (bottom). Thegreen circles indicate representative particles in distinct orientations.The white scale bar in the bottom panel indicates 10 nm.(D) EM reconstructions of the human Nav1.7-b1-b2 in complex withPT (Nav1.7-PT) or HS (Nav1.7-HS), both at 3.2-Å resolution. Thetop panel shows the gold-standard Fourier shell correlation (FSC)curves for the 3D reconstruction of the two complexes. In the bottompanel, the local resolution map was calculated with Relion 2.1 andpresented in Chimera.

RESEARCH | RESEARCH ARTICLEon M

ay 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

Shen et al., Science 363, 1303–1308 (2019) 22 March 2019 3 of 6

Fig. 2. Overall structures ofNav1.7-PT and Nav1.7-HS. (A) Oneblob of density is above VSDII inNav1.7-HS, and two are above VSDII

and VSDIV in Nav1.7-PT. These den-sities belong to HWTX-IV and ProTx-II,respectively. The structures and thedensities, which are low-pass filteredto 4.5 Å, are prepared in Chimera.(B) Overall structure of Nav1.7 incomplex with b1 and b2. Nav1.7 iscolored by repeats. The III-IV linker,which carries the fast-inactivationmotif Ile-Phe-Met (IFM), is highlightedin orange. The sugar moieties areshown as black sticks, and the IFMresidues are shown as spheres. Allstructure figures were prepared inPyMol (51) or Chimera.

Fig. 3. Closed intracellular gate.(A) The intracellular gate of Nav1.7 isslightly tightened compared with thatof Nav1.4. Shown here and in (C) areintracellular views of the superimposedPDs of Nav1.7-HS (domain colored) andNav1.4 (pale pink).The orange arrowsindicate the slight rotation of the S6segments in repeats II to IVof Nav1.7relative to those in Nav1.4. (B) Twoconformers of Tyr1755 in Nav1.7-HSresult in the shift of the constrictionsite along the permeation path ofNav1.7. Shown in the left two panels arethe respective permeation paths,calculated by HOLE (52), of Nav1.7-HSwith Tyr1755 up (left, green) or down(right, purple).The corresponding poreradii of human Nav1.7 (green andpurple) and Nav1.4 (gray) are comparedin the right panel.The residuesconstituting the respective constrictionsites in the two conformations of Nav1.7-HSare shown as sticks. Ea, extracellulara helix. (C) The intracellular gateof Nav1.7 is constituted by four hydrophobicresidues on the corresponding locus ofeach S6, Leu398-Leu964-Ile1453-Tyr1755.TheGDNmolecule observed in Nav1.4, shown as pink thin sticks, cannot beaccommodated by the closed gate in Nav1.7. Phe

1755 of Nav1.7 in the type I (down)conformation is shownasball and sticks. (D) Rotation of twoaromatic rings closesthe fenestration on the interface between repeats I and IV. Shown here is a sideviewof thesuperimposedPDsofNav1.7 andNav1.4. The type I and II conformersofMet1754 and Tyr1755 are shown as cyan and dark gray sticks, respectively, and theconformational shifts are indicated by orange arrows.The type II conformers ofthese two residues exhibit similar conformations to the corresponding ones in

Nav1.4 (shownas thinpinksticks).TheadjacentPhe387 andPhe391 onS6I (silver forNav1.7) adopt distinct conformers from the corresponding ones in Nav1.4 (pink),leading to closure of the fenestration on this side wall of the pore domain.Conformational changes of the corresponding Phe residues from Nav1.4 toNav1.7 are indicated by dark gray arrows. See fig. S6 for a detailed analysis.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.

RESEARCH | RESEARCH ARTICLEon M

ay 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

similar concentration of GDN was used forpurification and cryo-sample preparation ofthe Nav1.4 and Nav1.7 complexes. The S6 seg-ments in repeats II to IV of Nav1.7 undergoslight counterclockwise twisting relative tothose in Nav1.4 when viewed from the intra-cellular side (Fig. 3A). These seemingly minorchanges result in the upward shift of the intra-cellular gate by ~5 Å, enclosed by four hydropho-bic gating residues on the corresponding locusof each S6, Leu398-Leu964-Ile1453-Tyr1755, thatare highly conserved among the Nav channels(Fig. 3, B and C, and fig. S1). Single point muta-tions of the corresponding residues of Leu398 andLeu964 are found in Nav1.5 in patients with longQT syndrome and Brugada syndrome, respective-ly (40, 41), whereas mutations of the Tyr1755-corresponding residues in Nav1.1 and Nav 1.5are associated with seizure and long QT syn-drome, respectively (42, 43).The local EMmap reveals two distinct rotamers

of Tyr1755 (Fig. 3D and fig. S6A). Although the

main chain remains unchanged, the rotation ofthe aromatic ring results in two distinct con-formers, which we refer to as type I and type II,for intracellular gating (Fig. 3, B and C). The typeI Tyr1755 conformer is similar to the correspond-ing Tyr1593 in Nav1.4 (Fig. 3D, black sticks). Theconstriction with a diameter of ~4 Å is definedby Leu398 and Ile1453 (Fig. 3B, left). The narrowedgate can no longer accommodate GDN. In thetype II conformer, the side chain of Try1755 rotatesaround Cb by nearly 180°. Consequently, thearomatic ring directly cuts into the permea-tion path, completely sealing the intracellulargate (Fig. 3, B and C). The side chain of thepreceding residue Met1754 also has two con-formations. One is the same as that in Nav1.4,and the other, which we also refer to as thetype II conformer, swings toward S6I (Fig. 3Dand fig. S6A). The adjacent Phe387 and Phe391

on S6I both display distinct rotamers fromthose in Nav1.4 (Fig. 3D and fig. S6B). Suchconformational shifts effectively avoid clashing

with the type II conformer ofMet1754, meanwhileleading to closure of the fenestration on the sidewall constituted by repeats I and IV (Fig. 3D andfig. S6C).

Pore blockade by TTX and STX

TTX and STX have been used as potential thera-peutics for treatment of pain (44–46), and STXexhibits a lower affinity with Nav1.7 than withother subtypes (47). We recently reported theinsect NavPaS in complex with TTX and STX(27). Given that not all the coordinating residuesare invariant between NavPaS and Nav1.7, accu-rate structural models for Nav1.7 bound to theseprototypal neurotoxins are necessary to facilitatefuture drug discovery. The resolutions of the cen-tral region are higher than the averaged 3.2 Å forboth Nav1.7-PT and Nav1.7-HS, enabling reliablemodel building for TTX and STX and surround-ing residues (fig. S7).The binding sites and most interactions for

TTX and STX are conserved between Nav1.7 and

Shen et al., Science 363, 1303–1308 (2019) 22 March 2019 4 of 6

Fig. 4. Pore blockade by TTX and STX. (A) Specific interactionsbetween TTX and Nav1.7. Extracellular views of Nav1.7 (left) and thesuperimposed SF region of Nav1.7 and NavPaS (right) are shown.NavPaS is colored light purple. The coordination of TTX by Nav1.7is nearly identical to that by NavPaS except for residue variations onP2III. The SF residues Asp361-Glu930-Lys1406-Ala1698 (DEKA) areshown as thick sticks and color labeled. The red circles indicate thedifference in the coordination of the toxin by Nav1.7 and NavPaS.(B) Specific interactions between STX and Nav1.7. The conserved Metand Asp on helix P2 in the other eight human Nav subtypes, amongwhich Asp is expected to participate in STX binding, are replaced

by Thr1409 and Ile1410, respectively, in Nav1.7. The polar interactionsare represented by red dashed lines. The red circles highlightThr1409-Ile1410 and the corresponding residues in NavPaS.(C) Sequence alignment of the SF elements and P2 helices in eachrepeat of human Nav channels and NavPaS. The panel is adaptedfrom our recently published sequence alignment, with minoradjustment (27). The residues whose side chains participate in TTXand STX coordination are indicated by gray and brown squares,respectively, below the alignment. The varied toxin-coordinatingresidues between Nav1.7 and other channels are in white text on abrown background.

RESEARCH | RESEARCH ARTICLEon M

ay 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

NavPaS (Fig. 4, A and B). The major distinctionoccurs in the coordination by repeat III. Twoinvariant residues in the other eight human Navsubtypes on the P2 helix, Met and Asp, are re-placed by Thr1409 and Ile1410, respectively, in Nav1.7(Fig. 4C). The hydroxyl group and main-chainamide of Thr1409 are both hydrogen bonded tothe C10-OH of TTX, and the hydroxyl group ofThr1409 forms two hydrogen bonds with the car-bamoyl group of STX (Fig. 5, A and B). In othersubtypes, it is likely that the Asp correspondingto Ile1410 interacts directly with TTX and STX.The structure supports a previous analysis thatvariation fromMet1409 and Asp1410 in other Navchannels to Thr and Ile, respectively, in Nav1.7may account for the lowered affinity betweenNav1.7 and STX (47). The molecular details forthe recognition of TTX and STX by Nav1.7 may

facilitate future optimization of ligands for painrelief.

Conformational differences of VSDII inthe presence of ProTx-II and HWTX-IV

The structures of Nav1.7-PT and Nav1.7-HSare nearly identical except for overall shiftsof the b2 subunit and local differences inVSDII (Fig. 5A). Cys

895 on the extracellularloop of repeat II forms a disulfide bond withb2-Cys55 (fig. S1). Pivoting around this di-sulfide bond, the Ig domains in the two re-constructions deviate from each other by about15° (Fig. 5A). The shift may reflect the over-all flexibility of the immunoglobulin G (IgG)domain relative to Nav1.7 owing to the lackof specific interactions other than the di-sulfide bond.

The structural shifts of VSDII between Nav1.7-PT andNav1.7-HS aremainly found in the S3 andS4 segments (Fig. 5B). The entire S4 segmentin VSDII exists as a 310 helix in both reconstruc-tions, but a detailed comparison of Nav1.7-PT andNav1.7-HS shows slight motions between the twoS4 segments. The last two gating charge (GC)residues, Arg844 (R5) and Lys847 (K6) can be wellsuperimposed, but the segment deviates at Arg841

(R4) and above. The helical turns in Nav1.7-PT areslightly more elongated, shifting the Ca atoms ofR2 to R4 toward the extracellular side relative tothe corresponding ones in Nav1.7-HS. The move-ments of the side chains are more pronounced,with the guanidinium groups of R2 and R3 bothabove the conserved coordinating residue Asn780

(An1) in Nav1.7-PT, whereas only that of R2 isabove An1 in Nav1.7-HS (Fig. 5B, right).

Shen et al., Science 363, 1303–1308 (2019) 22 March 2019 5 of 6

Fig. 5. Binding sites andpotential working mecha-nism for HWTX-IV andProTx-II. (A) Structural dif-ferences between Nav1.7-HS(color coded) and Nav1.7-PT(white). An extracellular viewof the two superimposedstructures is shown. b2-Igdomains are shown in semi-transparent surface repre-sentation.The lack of specificinteraction between b2-Igand Nav1.7, other than thedisulfide bond, may accountfor the poor resolutions andthe flexible positioning of b2-Igin the two reconstructions.VSDII, which provides thedocking site for both GMTs, ishighlighted by pink shading.(B) Slight conformationalchanges of VSDII betweenNav1.7-HS and Nav1.7-PT. TheS4 segment exists as a 310

helix in both structures.Note that the Ca atoms andthe side groups of GC resi-dues R2 to R4 in Nav1.7-PTmove slightly toward theextracellular side relativeto those in Nav1.7-HS.(C) HWTX-IV and ProTx-IIbind to the similar site 4 onVSDII. In the superimposedoverall structures ofNav1.7-HS and Nav1.7-PT,ProTx-II (orange) andHWTX-IV (light purple) are largely overlapped. The densities for thetoxins are low-pass filtered to 4.5 Å in Chimera. (D) Distinct bindingmodes for ProTx-II by VSDII and VSDIV. When ProTx-II–loaded VSDII

and VSDIV are superimposed, the difference in the position of thetoxin is evident. See fig. S8 for a more detailed analysis on thebinding of ProTx-II and HWTX-IV by Nav1.7. (E) Potential workingmechanisms for GMTs. The cartoon is derived from the structures of

NavPaS in complex with Dc1a (left) and Nav1.7 bound to GMTs(right). An entity other than a VSD, such as the PD and thesurrounding lipid bilayer, provides the support to anchor the GMT,which then can lock the target VSD in a functional state throughspecific interactions. ECL, extracellular loop. The changingcolors for GMT in the right panel indicate the amphiphilic natureof GMTs like HWTX-IV and ProTx-II.

RESEARCH | RESEARCH ARTICLEon M

ay 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

Considering the identical protein purifica-tion procedure, it is reasonable to attributethe structural shifts of VSDII to the associa-tion of ProTx-II and HWTX-IV (Fig. 5, C andD, and fig. S8). The binding sites for ProTx-IIand HWTX-IV on VSDII overlap, although theinteraction details may vary. The L3-4 loopcontaining residues Ala826-Asp827-Val828-Glu829-Gly830 in VSDII is invisible in both EMreconstructions. Nevertheless, the positions ofProTx-II and HWTX-IV support the previousanalysis that the L3-4 loop on VSDII, known as“site 4” (24), is involved in the association ofboth toxins (31–33, 35). Asn774 on S2 may alsocontribute to the interaction with both toxins(fig. S8, A and B). No direct contact is detectedbetween HWTX-IV and S4II, whereas Leu

831 onS4II appears to interact with ProTx-II, whichmay underlie the further “up” conformation ofS4II in the presence of ProTx-II (Fig. 5, B and C,and fig. S8B).Although the L3-4 loop also represents the

major binding site for ProTx-II on VSDIV, theposition of the toxin deviates markedly from thatin VSDII. Whereas ProTx-II is placed abovethe central cavity of VSDII and sandwiched byS2II and S3II, the L3-4IV loop (also known as site3) is solely responsible for binding to ProTx-II,which is oriented away from S2 and makes nocontact with other elements in VSDIV (Fig. 5Dand fig. S8C).

Discussion

In both reconstructions, the intracellular gate isclosed, and all VSDs exhibit depolarized con-formations, although at distinct up states relativeto the charge transfer center (48) (fig. S9). Suchfeatures are consistent with an inactivated state.In this conformation, the fast-inactivation Ile-Phe-Met motif in the III-IV linker still insertsinto the cavity between the S6 helical bundle andthe S4-5 restriction ring in repeats III and IV(Fig. 2B), providing further structural supportfor the “allosteric blocking”mechanism for fastinactivation (17).The well-characterized ProTx-II and HWTX-IV

both effectively inhibit activation of Nav1.7 at~100 nM (28, 31). They were suggested to bindto VSDII in the resting state (33, 34). Our struc-tures show that they can also bind to the de-polarized conformation of VSDII, probablybecause of the high concentration (50 mM) usedfor complex reconstitution. ProTx-II was alsoshown to inhibit inactivation of Nav channels(31); therefore, the binding of depolarizedVSDIV is expected. Nonetheless, we are cautiousin the interpretation of the observed inter-actions between the two GMTs and VSDs (fig.S8). It was reported that membranes facilitatethe action of these toxins (28, 49, 50). The poorresolution of the toxins may indicate intrinsicflexibility in the absence of lipid bilayer. Struc-tures of Nav channels in complex with theseGMTs in a lipid environment, such as nano-disc, will be necessary to provide a more com-prehensive understanding of the mode of actionof these toxins.

The binding mode for HWTX-IV and ProTx-IIby Nav1.7 is distinct from that for Dc1a by NavPaS(27). Whereas HWTX-IV and ProTx-II both bindto the peripheral region that links S3 and S4 in aVSD, Dc1a inserts into the extracellular cavityenclosed by segments in VSDII and the PD (Fig.5E). Despite this distinction, these GMTs mayobserve a common principle for function. Anentity distinct from the VSDs, such as the PD ofNavPaS for Dc1a and the lipid bilayer for HWTX-IV and ProTx-II, provides the support to anchortheGMTs, which then lock the associated VSDsin certain functional states to achieve gatingmodulation.Despite many remaining questions, successful

recombinant expression, purification, and struc-tural determinationofNav1.7 in thepresenceof bothb1 and b2 subunits and different animal toxinsto near-atomic resolution establishes an impor-tant framework for investigating the structure-function relationship. Dozens of point mutationsthat are associated with pain syndromes cannow be structurallymapped (table S1) and probedwith a precise template. The structuresmay guidedevelopment of therapeutic agents for painrelief.

REFERENCES AND NOTES

1. N. Klugbauer, L. Lacinova, V. Flockerzi, F. Hofmann, EMBO J.14, 1084–1090 (1995).

2. S. D. Dib-Hajj, T. R. Cummins, J. A. Black, S. G. Waxman,Trends Neurosci. 30, 555–563 (2007).

3. A. I. Basbaum, D. M. Bautista, G. Scherrer, D. Julius, Cell 139,267–284 (2009).

4. D. L. Bennett, C. G. Woods, Lancet Neurol. 13, 587–599(2014).

5. M. A. Nassar et al., Proc. Natl. Acad. Sci. U.S.A. 101,12706–12711 (2004).

6. Y. Yang et al., J. Med. Genet. 41, 171–174 (2004).7. T. R. Cummins, S. D. Dib-Hajj, S. G. Waxman, J. Neurosci. 24,

8232–8236 (2004).8. J. J. Cox et al., Nature 444, 894–898 (2006).9. C. R. Fertleman et al., Neuron 52, 767–774 (2006).10. Y. P. Goldberg et al., Clin. Genet. 71, 311–319

(2007).11. E. C. Emery, A. P. Luiz, J. N. Wood, Expert Opin. Ther. Targets

20, 975–983 (2016).12. Y. P. Goldberg et al., Pain 153, 80–85 (2012).13. J. H. Lee et al., Cell 157, 1393–1404 (2014).14. W. A. Catterall, A. L. Goldin, S. G. Waxman, Pharmacol. Rev. 57,

397–409 (2005).15. M. Noda et al., Nature 312, 121–127 (1984).16. H. Shen et al., Science 355, eaal4326 (2017).17. Z. Yan et al., Cell 170, 470–482.e11 (2017).18. X. Pan et al., Science 362, eaau2486 (2018).19. S. B. Long, E. B. Campbell, R. Mackinnon, Science 309,

897–903 (2005).20. F. H. Yu, V. Yarov-Yarovoy, G. A. Gutman, W. A. Catterall,

Pharmacol. Rev. 57, 387–395 (2005).21. H. Terlau et al., FEBS Lett. 293, 93–96 (1991).22. H. A. O’Malley, L. L. Isom, Annu. Rev. Physiol. 77, 481–504

(2015).23. C. J. Laedermann, N. Syam, M. Pertin, I. Decosterd, H. Abriel,

Front. Cell. Neurosci. 7, 137 (2013).24. W. A. Catterall et al., Toxicon 49, 124–141 (2007).25. F. Zhang, X. Xu, T. Li, Z. Liu, Mar. Drugs 11, 4698–4723

(2013).26. S. Yang et al., Proc. Natl. Acad. Sci. U.S.A. 110, 17534–17539

(2013).27. H. Shen et al., Science 362, eaau2596 (2018).28. K. Peng, Q. Shu, Z. Liu, S. Liang, J. Biol. Chem. 277,

47564–47571 (2002).29. R. E. Middleton et al., Biochemistry 41, 14734–14747

(2002).30. W. A. Schmalhofer et al., Mol. Pharmacol. 74, 1476–1484

(2008).

31. Y. Xiao, K. Blumenthal, J. O. Jackson 2nd, S. Liang,T. R. Cummins, Mol. Pharmacol. 78, 1124–1134 (2010).

32. Y. Xiao, J. O. Jackson 2nd, S. Liang, T. R. Cummins, J. Biol.Chem. 286, 27301–27310 (2011).

33. Y. Xiao et al., J. Biol. Chem. 283, 27300–27313(2008).

34. S. Sokolov, R. L. Kraus, T. Scheuer, W. A. Catterall, Mol.Pharmacol. 73, 1020–1028 (2008).

35. F. Bosmans, M. F. Martin-Eauclaire, K. J. Swartz, Nature 456,202–208 (2008).

36. W. Huang, M. Liu, S. F. Yan, N. Yan, Protein Cell 8, 401–438(2017).

37. S. D. Dib-Hajj et al., Brain 128, 1847–1854 (2005).38. C. Han et al., Brain 132, 1711–1722 (2009).39. M. Eberhardt et al., J. Biol. Chem. 289, 1971–1980

(2014).40. J. D. Kapplinger et al., Heart Rhythm 6, 1297–1303

(2009).41. J. D. Kapplinger et al., Heart Rhythm 7, 33–46

(2010).42. C. Napolitano et al., JAMA 294, 2975–2980

(2005).43. G. Fukuma et al., Epilepsia 45, 140–148 (2004).44. N. A. Hagen et al., J. Pain Symptom Manage. 35, 420–429

(2008).45. F. R. Nieto et al., Mar. Drugs 10, 281–305 (2012).46. M. Chorny, R. J. Levy, Proc. Natl. Acad. Sci. U.S.A. 106,

6891–6892 (2009).47. J. R. Walker et al., Proc. Natl. Acad. Sci. U.S.A. 109,

18102–18107 (2012).48. X. Tao, A. Lee, W. Limapichat, D. A. Dougherty, R. MacKinnon,

Science 328, 67–73 (2010).49. S. T. Henriques et al., J. Biol. Chem. 291, 17049–17065

(2016).50. A. J. Agwa et al., Biochimica et biophysica Acta Biomembranes

1859, 835–844 (2017).51. W. L. DeLano, The PyMOL Molecular Graphics System

(Schrödinger, 2002); www.pymol.org.52. O. S. Smart, J. G. Neduvelil, X. Wang, B. A. Wallace,

M. S. Sansom, J. Mol. Graph. 14, 354–360, 376 (1996).

ACKNOWLEDGMENTS

We thank X. Li (Tsinghua University) for technical supportduring EM image acquisition. Funding: This work was fundedby the National Key Basic Research (973) Program(2015CB910101 to N.Y.) and the National Key R&D Program(2016YFA0500402 to N.Y. and 2016YFA0501100 to J.L.)from the Ministry of Science and Technology of China,and the National Natural Science Foundation of China (projects31621092, 31630017, and 81861138009 to N.Y.). We thankthe Tsinghua University Branch of China National Centerfor Protein Sciences (Beijing) for providing the cryo-EMfacility support. We thank the computational facility supporton the cluster of Bio-Computing Platform (Tsinghua UniversityBranch of China National Center for Protein SciencesBeijing) and the “Explorer 100” cluster system of TsinghuaNational Laboratory for Information Science and Technology.N.Y. is supported by the Shirley M. Tilghman endowedprofessorship from Princeton University. Authorcontributions: N.Y. conceived the project. H.S., D.L., K.W.,and J.L. performed the experiments. All authors contributedto data analysis. N.Y. wrote the manuscript. Competinginterests: The authors declare no competing interests.Data and materials availability: The EM maps for Nav1.7-HSand Nav1.7-PT have been deposited in EMDB (www.ebi.ac.uk/pdbe/emdb/) with the codes EMD-9781 and EMD-9782,respectively, and the atomic coordinates in the Protein DataBank (www.rcsb.org) with the accession codes 6J8G and6J8H for Nav1.7-HS and 6J8I and 6J8J for Nav1.7-PT.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/363/6433/1303/suppl/DC1Materials and MethodsSupplementary TextFigs. S1 to S9Tables S1 to S3References (53–68)

2 December 2018; accepted 29 January 2019Published online 14 February 201910.1126/science.aaw2493

Shen et al., Science 363, 1303–1308 (2019) 22 March 2019 6 of 6

RESEARCH | RESEARCH ARTICLEon M

ay 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

1.7 channel in complex with auxiliary subunits and animal toxinsvStructures of human NaHuaizong Shen, Dongliang Liu, Kun Wu, Jianlin Lei and Nieng Yan

originally published online February 14, 2019DOI: 10.1126/science.aaw2493 (6433), 1303-1308.363Science 

, this issue p. 1303, p. 1309; see also p. 1278Science structures provide a framework for targeted drug development.vThese and other recently determined Na

1.2.v2 and a toxic peptide, the µ-conotoxin KIIIA. The structure shows why KIIIA is specific for Naβ1.2 bound to vNa present the structure ofet al.2 subunits and with animal toxins. Pan β1 and β1.7 in complex with both vstructures of Na

present theet al.act as pore blockers or gating modifiers (see the Perspective by Chowdhury and Chanda). Shen subunits and by natural toxins that can eitherβfor voltage sensing and ion conductance, but function is modulated by

subunit is sufficientαmany subtypes of these channels, making it challenging to develop specific therapeutics. A core ) channels have been implicated in cardiac and neurological disorders. There arevVoltage-gated sodium (Na

Targeting sodium channels

ARTICLE TOOLS http://science.sciencemag.org/content/363/6433/1303

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2019/02/13/science.aaw2493.DC1

CONTENTRELATED

http://stm.sciencemag.org/content/scitransmed/8/335/335ra56.fullhttp://science.sciencemag.org/content/sci/363/6433/eaav8573.fullhttp://science.sciencemag.org/content/sci/363/6433/1278.fullhttp://science.sciencemag.org/content/sci/363/6433/1309.full

REFERENCES

http://science.sciencemag.org/content/363/6433/1303#BIBLThis article cites 67 articles, 24 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2019 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on May 27, 2020

http://science.sciencem

ag.org/D

ownloaded from