Báo cáo khoa học: The unique pharmacology of the scorpion a-like toxin Lqh3 is associated with its flexible C-tail pdf

Although different a-toxins are similarly toxic to mice when injected subcutaneously and similarly affect Keywords pH-dependent toxin binding; scorpion a-like toxin; structure–function r

is associated with its flexible C-tail

Izhar Karbat1, Roy Kahn1, Lior Cohen1, Nitza Ilan1, Nicolas Gilles2, Gerardo Corzo3, Oren Froy1, Maya Gur1, Gudrun Albrecht4, Stefan H Heinemann4, Dalia Gordon1and Michael Gurevitz1

1 Department of Plant Sciences, George S Wise Faculty of Life Sciences, Tel Aviv University, Ramat Aviv, Tel Aviv, Israel

2 CEA Saclay, De´partement d’Inge´nie´rie des Prote´ines, Gif-sur Yvette, France

3 Instituto de Biotecnologı´a, Universidad Nacional Auto´noma de Me´xico, Cuernavaca Morelos, Me´xico

4 Center for Molecular Biomedicine, Department of Biophysics, Friedrich Schiller University Jena, Germany

Voltage-gated sodium channels (Navs) are responsible

for the depolarization phase of the action potential in

most excitable cells Due to their pivotal role in

excita-bility, Navs are targeted by a large variety of toxins

that modify their gating, such as long-chain scorpion

toxins These toxins are 61–76 residue-long

polypep-tides that share a similar a⁄ b scaffold and are divided

into two classes, a and b, according to their mode of action and different receptor sites [1,2] Scorpion a-toxins prolong the action potential by slowing chan-nel inactivation upon binding at a site that involves extracellular regions of channel domains 1 and 4 [2–4] Although different a-toxins are similarly toxic to mice when injected subcutaneously and similarly affect

Keywords

pH-dependent toxin binding; scorpion a-like

toxin; structure–function relationships; toxin

effect on inactivation; toxin receptor site on

sodium channel

Correspondence

D Gordon and M Gurevitz, Department of

Plant Sciences, George S Wise Faculty of

Life Sciences, Tel Aviv University, Ramat

Aviv, Tel Aviv 69978, Israel

Fax: +972 3 6406100

Tel: +972 3 6409844

E-mail: dgordon@post.tau.ac.il and

mamgur@post.tau.ac.il

(Received 9 January 2007, revised 6

February 2007, accepted 12 February 2007)

doi:10.1111/j.1742-4658.2007.05737.x

The affinity of scorpion a-toxins for various voltage-gated sodium channels (Navs) differs considerably despite similar structures and activities It has been proposed that key bioactive residues of the five-residue-turn (residues 8–12) and the C-tail form the NC domain, whose topology is dictated by a cis or trans peptide-bond conformation between residues 9 and 10, which correlates with the potency on insect or mammalian Navs We examined this hypothesis using Lqh3, an a-like toxin from Leiurus quinquestriatus hebraeus that is highly active in insects and mammalian brain Lqh3 exhibits slower association kinetics to Navs compared with other a-toxins and its binding to insect Navs is pH-dependent Mutagenesis of Lqh3 revealed a bi-partite bioactive surface, composed of the Core and NC domains, as found in other a-toxins Yet, substitutions at the five-residue turn and stabilization of the 9–10 bond in the cis conformation did not affect the activity However, substitution of hydrogen-bond donors⁄ accep-tors at the NC domain reduced the pH-dependency of toxin binding, while retaining its high potency at Drosophila Navs expressed in Xenopus oocytes Based on these results and the conformational flexibility and rear-rangement of intramolecular hydrogen-bonds at the NC domain, evident from the known solution structure, we suggest that acidic pH or specific mutations at the NC domain favor toxin conformations with high affinity for the receptor by stabilizing the bound toxin-receptor complex More-over, the C-tail flexibility may account for the slower association rates and suggests a novel mechanism of dynamic conformer selection during toxin binding, enabling a-like toxins to affect a broad range of Navs

Abbreviations

Aah2, alpha toxin 2 from the scorpion Androctonus australis hector; BmK M1, alpha toxin from the scorpion Buthus martensii Karsch; CHO, Chinese hamster ovary; Lqh2, Lqh3, LqhaIT, alpha toxins from the scorpion Leiurus quinquestriatus hebraeus; Na v , voltage-gated sodium channel.

rat skeletal muscle Navs [5–7], they exhibit profound

differences in potency when injected into mice brain,

and in their affinity for insect and rat-brain neuronal

preparations [7,8] Accordingly, scorpion a-toxins were

divided into three pharmacological groups (Fig 1): (a)

Classical anti-mammalian toxins that bind with high

affinity to rat brain synaptosomes and are practically

nontoxic to insects [1]; (b) a-toxins highly active on

insects that bind with high affinity to insect Navs and

are weakly toxic in mammalian brain; and (c) a-like

toxins that are active in both mammalian brain and

insects (Fig 1 [8,9])

To correlate the selectivity of a-toxins with their

structure, the bioactive surface of the anti-insect

LqhaIT (from L quinquestriatus hebraeus) and its

putative equivalent in the anti-mammalian Aah2 were

investigated and shown to consist of two domains [10]

Four residues located on short loops that connect the

conserved secondary structure elements of the molecule

core form the Core domain, while the five-residue-turn

(residues 8–12) and the C-terminal segment (residues

56–64) form the NC domain The division of the

bio-active surface into two domains is supported by

muta-genesis of the a-like toxin BmK M1 (from Buthus martensii Karch) [11–13] As the amino acid composi-tion and spatial arrangement of the NC domain varies among a-toxins, it was suggested to confer toxin pref-erential binding to various Navs The high insecticidal potency of LqhaIT was correlated with a protruding conformation of the NC domain, a feature typifying all scorpion a-toxins active on insects This protrusion, mediated by a nonproline cis peptide bond between residues 9 and 10 of the five-residue turn, differs mark-edly from the flat conformation dictated by a trans peptide bond conformation between residues 9 and 10, which characterizes the NC domain in toxins highly active in the rat brain [10,14] In this respect, the high potency of a-like toxins for both insect and various mammalian Navs [7] cannot be readily explained and was addressed here using Lqh3, the most pharmaco-logically characterized toxin with known structure of the a-like group [15] Lqh3 is highly toxic to insects and competes with LqhaIT on binding to insect Navs Lqh3 differs from classical anti-mammalian a-toxins

as it inhibits Nav inactivation in cell bodies of hippo-campus brain neurons, on which the anti-mammalian

Fig 1 Sequence alignment of scorpion a-toxins representing three pharmacological groups Positions are numbered according to Aah2 Dashes indicate gaps for best alignment Residues of the five-residue turn and C-tail are shaded Residues of the conserved Core domain are in bold Lqh2, Lqh3, LqhaIT [9], Lqh6, and Lqh7 [40] are from the scorpion L quinquestriatus hebraeus; Aah2 is from Androctonus australis hector; Lqq3 and Lqq5 are from L quinquestriatus quinquestriatus; Bmk-M1, Bmk-M2, Bmk-M4, and Bmk-M8 are from Buthus martensii Karch; Bom3 and Bom4 are from Buthus occitanus mardochei [9].

Lqh2 is inactive, and is unable to affect Nav1.2 in the

rat brain, on which Lqh2 is highly active [16]

More-over, it has been shown that the pharmacological

properties of Lqh3 are unique in that its binding

affin-ity for insect channels drops >30-fold at pH 8.5 versus

pH 6.5, and its rate of association with receptor site-3

on both insect and mammalian Navs is 4–15-fold

slower compared with LqhaIT and Lqh2 [6,17,18]

To clarify the molecular basis of the unique

pharma-cological features of Lqh3, we analyzed its bioactive

surface seeking for residues involved with its slow

association kinetics and sensitivity to pH changes upon

interaction with insect Navs Our data reveal that

resi-dues at the NC domain, which may serve as hydrogen

bond acceptors or donors, are specifically associated

with these features Re-examination of the solution

structures of Lqh3 disclosed a high conformational

flexibility of its C-tail, which may interconvert between

two distinct conformers that differ in their

intramole-cular hydrogen-bonding pattern Based on these

obser-vations we suggest that the unique pharmacological

features of scorpion a-like toxins are associated with

the flexibility of the C-tail

Results

The bioactive surface of Lqh3

Twenty-four residues were substituted and the toxin

mutants were produced in Escherichia coli as a fusion

peptide (His-Apamin-Lqh3), folded in vitro, and

puri-fied by RP-HPLC (see Experimental procedures)

Changes in activity were monitored in toxicity assays

on blowfly larvae and binding assays using cockroach

neuronal membrane preparations CD spectroscopy

was used as a measure of secondary structure signature

to discern effects that were due to structural

perturba-tions from those associated directly with toxin activity

From a total of 49 mutants, the CD spectrum of only I59R altered (Fig 2B) Of the 24 modified residues, substitution of 15 had a weak (DDG¼ 1.1 kcalÆmol)1)

to moderate (DDG¼ 1.5 kcal ⁄ mol) effect on activity (Table 1)

Substitutions in the five-residue turn (residues 8–12) had no significant effect on the activity to insects, even when charges were neutralized or inverted (Table 1) These results imply that the five-residue turn in Lqh3

is most likely not involved in direct interaction with the channel receptor, and that it tolerates considerable changes with no perturbation of toxin folding, in con-trast to LqhaIT [19] and BmK M1 [14] Substitutions

in the loop preceding the a-helix had large effects on activity, as shown by the replacement of His15 by bulky aliphatic or charged residues (H15F⁄ L ⁄ R), Phe17 by Ala, and Pro18 by Arg or Gly (Table 1) Substitutions in the loop connecting the second and the third b-strands highlighted the importance for activity of Phe39 and Leu45, as was shown for their equivalents in LqhaIT and Bmk-M1 [10,12] His15, Phe17, Pro18, Phe39 and Leu45 constitute a distinct amino acid cluster on the molecule surface intercon-nected by hydrophobic–aromatic interactions resem-bling the Core domain reported for LqhaIT [10] Substitutions I59A⁄ R had a marked effect on the binding affinity (Table 1) Ile59 is mostly buried in the molecule and forms hydrophobic contacts with Gly4, Tyr5, Ile6, and Ala7 of the N-terminal region [15] While I59R altered the CD signature of the molecule, the CD spectrum of I59A was similar to that of the unmodified toxin, which suggested that Ile59 might form contact with the receptor site, as was suggested for the equivalent residue in other a-toxins [10,11,20] Neutralization or inversion of the charge of Lys64 (K64A⁄ D) and His66 (H66A ⁄ E) significantly affec-ted the activity, while a conserved substitution had

a minor effect (Table 1), which suggested that a

Fig 2 The bioactive surface of Lqh3 (A) The toxin backbone is shown in ribbon Resi-dues, whose substitution affected the func-tion (see Table 1) are space-filled and colored according to their chemical nature (aliphatic, green; aromatic, magenta; posit-ive, blue) (B) CD spectra of the recombinant HA-Lqh3 and representative mutants.

positively charged C-tail was important for activity.

Substitutions at the negatively charged patch

com-posed of Glu10, Glu61 and Glu63, which is unique to

Lqh3 compared to other a-toxins, had no effect on the

activity (Table 1)

We further examined if the bioactive surface of

Lqh3 towards insect Navs coincided with that

pre-sented toward rat skeletal muscle Navs by analyzing

the effects of various substitutions on Nav1.4 and

Drosophila melanogaster DmNav1 Navs expressed in

Chinese hamster ovary (CHO) cells and in Xenopus

oocytes, respectively (Table 2) Most substitutions

that markedly reduced the binding affinity for

cock-roach neuronal membranes reduced the toxin potency

towards rNav1.4 and DmNav1 to a similar extent

However, H15A, which had only a slight effect on

the toxicity and binding affinity for insects and on

potency at DmNav1, profoundly affected the potency

at rNav1.4 This analysis highlighted also substitution

H66E, which had a larger effect on the potency at

DmNav1 than at rNav1.4 (Table 2) Thus, the

bio-active surface of Lqh3 towards insect and rat

skel-etalmuscle Navs is similar, but not identical, where

Table 1 Effects of mutations in Lqh3 on binding to cockroach neuronal membranes The change in apparent binding affinity is presented as the ratio of Kmuti over Kwti Kwti and Kmuti were obtained in competition against125I-LqhaIT binding at pH 7.2, as previously described [18] The K wt

i value is 1.0 ± 0.1 n M , n ¼ 6 K i determination is described in the Experimental procedures The change in binding energy was calcu-lated as DDG ¼ -RT ln(Kwti ⁄ Kmuti ).

HA-Lqh3 mutant Kmuti ⁄ Kwti DDG (kcalÆmol)1) Lqh3 mutant Kmuti ⁄ Kwti DDG(kcalÆmol)1)

Table 2 Comparison of the effects of selected Lqh3 mutants on rat skeletal muscle Navs (rNav1.4) expressed in CHO cells and on the Drosophila DmNav1 channel expressed in Xenopus oocytes The apparent effective concentration 50% (EC 50 ) of each mutant

on rNav1.4 and DmNav1 were determined in at least three inde-pendent experiments (see Experimental procedures) and normal-ized to the potency of unmodified Lqh3 (EC mut

50 ⁄ EC wt ) The effect

on DmNa v 1 was assayed at pH 7.1 (EC 50 ¼ 10.5 ± 1.6 n M ).

Lqh3 mutant

EC 50 (rNa v 1.4) (n M )

EC mut

50 ⁄ EC wt rNa v 1.4 (EC mut

50 ⁄ EC wt DmNav1)

His15 and His66 seem to contribute to the differential

interaction of Lqh3 with channel receptors of various

origin

In total, the bioactive surface of Lqh3 is composed

of two distinct domains, the Core and NC domains,

formed by residues of the loop preceding the a-helix,

the loop connecting the second and the third b-strands,

and the C-tail (Fig 2A)

Effect of substitutions in Lqh3 on its

pH-dependent binding

The binding affinity of Lqh3 for cockroach neuronal

membranes decreased 32-fold when assayed at pH 8.5

compared to pH 7.2 (Table 3) We have further tested

the effect of pH transitions on Lqh3 interaction with

DmNav1 channels expressed in Xenopus oocytes

Under control conditions, pH alterations of the bath

solution in the range 7.0–8.5 had no effect on the

sodium current amplitude, and a slight reduction of the

peak current was observed at pH 6.5 (not shown) The

effect of Lqh3 on DmNav1 increased markedly upon

transition from basic to more acidic pH with an

estima-ted half saturation at pH 7.2 (Fig 3A–C) This increase

was slow and typically saturated after 10–15 min

(Fig 3D) The slow kinetics was also evident when the

toxin was pre-equilibrated at the tested pH prior to

application onto the oocyte, suggesting that Lqh3

sensitivity to pH is associated with some later stage in

the binding process to the channel This is

corrobor-ated by previous binding studies, which demonstrcorrobor-ated

that Lqh3 association rate did not change between

pH 7.5 and 6.5, and the increased affinity was due to

decrease in the dissociation rate constant at lower

pH [18]

To clarify the molecular basis of the pH-dependent binding, we examined two mechanisms previously suggested to affect toxin binding It was suggested that a cis–trans isomerization of the nonproline cis-peptide bond between residues 9 and 10 of scorpion a-like toxins might function as a molecular switch that determines their preference for various Navs [14]

In Lqh3, the peptide-bond between Pro9 and Glu10 appears in solution as a mixed population of cis and trans conformations, and a slow pH- and tem-perature-dependent interconversion between these two isomeric forms was reported [15,21] Thus, a pH-dependent isomerization of the P9-E10 bond in Lqh3 could underlie its pH-dependent binding We tested this hypothesis by constructing a toxin double mutant, in which Cys substituted both residues Modeling of the double mutant (P9C-E10C) predicted that the position of these two Cys residues on the tight five-residue-turn would force their side chains

to adopt a solvent exposed conformation and create

a vicinal disulfide bond in a cis conformation (Fig 4A,B) The toxin mutant was successfully expressed and folded in vitro, and exhibited identical toxicity (EC50¼ 75 ± 5 ng ⁄ 100 mg blowfly larvae) and binding affinity for cockroach neuronal mem-branes (Ki¼ 1.06 ± 0.07 nm, n ¼ 3) as those of the unmodified toxin (Table 1) The molecular mass of the P9C-E10C toxin mutant was determined to be

7040 ± 0.1 Da, which corresponded exactly to the theoretical mass calculated, assuming that the newly introduced Cys residues were both oxidized This

Table 3 Effect of mutations on the pH dependence of Lqh3 binding to cockroach neuronal membranes All binding experiments were per-formed using 125 I-LqhaIT, a pH-independent marker of receptor site-3 [18], and the data represent mean ± SE of 2–4 independent experi-ments; ND, not determined Ki(pH 8.5) ⁄ K i (pH 7.2) represents the change in ratio when the analysis was performed at pH 8.5 versus 7.2.

Mutant

Ki, pH 6.5 (n M )

Ki, pH 7.2 (n M )

Ki, pH 8.5 (n M )

Ki(pH 8.5) ⁄

K i (pH 7.2)

7 0 6 0 5 0 4 0 3 0 2 0

0 8 0 6 0 4 0 2 0

s) ( e m i T

5 7

0 7

5 6

l o r t n C 0 8

s m 5

B

0 1

8 0

6 0

4 0

2 0

0 8 5 7 0 7 5 6

H p

A

0 1

8 0

6 0

4 0

2 0

0 0

04

03

02

01

00

M ) n ( ] n i x o T [

1 7 H p 5 7 H p

Fig 3 pH-dependent effect of Lqh3 on DmNav1 channels expressed in Xenopus oocytes (A) Concentration-response relationship of Lqh3 at

pH 7.85 (s) and pH 7.1 (d) Data were fit using the Hill equation (Eqn 1, Experimental procedures) and the EC50values obtained were 86.6 ± 15.1 n M (n ¼ 3; pH 7.85) and 10.5 ± 1.6 n M (n ¼ 3; pH 7.1) (B) Effect of Lqh3 at various pH values Oocytes were incubated with

50 n M Lqh3 dissolved in buffer at pH 8.0, and the toxin effect was continuously monitored by step depolarizations to )10 mV from a holding potential of )80 mV The toxin effect was allowed to saturate for 10 min and the external solution was then replaced by 50 n M Lqh3 in

pH 7.5 buffer This procedure was repeated stepwise down to pH 6.5 Current traces from a representative oocyte are shown (C) Toxin effect (Iss⁄ I peak ) at each pH in the range 6.5–8.0 was normalized to the maximal effect obtained at pH 6.5 and plotted as a function of the

pH Each point represents mean ± SEM from three oocytes (D) Kinetics of the effect developed upon transition from pH 7.5 to pH 7.0 for the cell presented in B The steady-state to peak current ratio was determined at intervals of 15 s from the transition to pH 7.0 and is plot-ted against the incubation time The kinetics was fit by a single exponential with s ¼ 496 s.

6 H 0 E

3 E

9

3 Å 4 Å.0

8 Q

9 P

0 E

1 N

2 C

8 Q

9 C

0 C

1 N

2 C

Fig 4 Conformations of the five-residue turn and the C-terminal segment of Lqh3 (A,B) Fixation of the peptide bond between residues 9 and 10 in Lqh3 in a cis conformation by an engineered vicinal disulfide bond The five-residue turn of Lqh3 (A) is compared with its modeled equivalent in the P9C-E10C mutant (B) The modeling was based on the structure of Lqh3, and energy minimized in vacuo using the GROMOS96 implementation of Swiss-pdbViewer [39] The arrows point to the cis peptide bond between residues 9 and 10 (C) Hydrogen bond network that involves the sidechains of Glu10, Glu63 and His66.

finding suggested that the two Cys residues were

indeed linked by a vicinal disulfide bond (Fig 4B)

Still, the binding affinity of the double mutant

remained highly dependent on pH, similar to the

unmodified Lqh3 (Table 3) Therefore, we concluded

that the cis–trans isomerization of the P9-E10

pep-tide-bond was most likely unrelated to the pH

dependence of Lqh3

To test the possibility that the pH-dependent

bind-ing of Lqh3 is associated with protonation of surface

histidines [18], we examined the effects of toxin

mutants H15A⁄ L, H36A, H43A ⁄ R, and H66A ⁄ R ⁄ E

on the binding affinity for cockroach synaptosomes at

various pH values (Table 3) Whereas substitutions of

His15, His36 and His43 did not reduce Lqh3

sensitiv-ity to pH, substitutions of His66 had a clear impact

with the utmost decrease obtained with H66R

(Table 3) Unexpectedly, substitutions of neutral or

negatively charged residues of the five-residue turn

(Q8A and E10Y) and C-tail (E63R), which were not

assigned to the bioactive surface, reduced markedly

the dependence of binding affinity on pH (Table 3)

Combined with the slow build up of Lqh3 effect on

DmNav1 upon pH transitions, these results indicate

that the dependence of Lqh3 binding on pH is not

dic-tated by the protonation of His residues per se These

findings prompted us to examine structural features of the NC domain that could explain its relatedness with the pH dependency

Lqh3 pH-dependent binding is associated with the conformational flexibility of the C-tail Inspection of Lqh3 solution structure reveals that the C-terminal segment is by far more flexible than its equivalent in LqhaIT (Fig 5A,B) The conformational heterogeneity focuses on a short loop spanning resi-dues 60–64 (Fig 5B), and is mediated by alternations

in a hydrogen bond network among the negatively charged carboxyl groups of Glu10 and Glu63, and the guanidinium moiety of His66 (Fig 4C), substitution of which clearly affected the pH-dependent binding of Lqh3 (Table 3)

To examine whether changes in this hydrogen bond network alter Lqh3 sensitivity to pH, we constructed a double mutant, in which Tyr and Arg substituted Glu10 and Glu63 to eliminate the intramolecular polar interactions of His66 with these two Glu residues The binding affinity of the E10Y-E63R mutant to cockroach neuronal membranes (Table 3), as well as its potency at DmNav1 channels at neutral pH, was similar to that of the unmodified toxin (Fig 6A)

h

A

B

0

.

3

5

.

2

0

.

2

5

.

1

0

.

1

5

.

0

6 4 2 0 8 6 4 2 0 8 6 4 2 0 8 6 4 2 0 8 6 4 2 0 8 6 4 2 0 8

6

4

2

e d i s e R

3 q

h

L I T

Fig 5 Conformational heterogeneity in LqhaIT and Lqh3 (A) Solution structures of LqhaIT (PDB ID: 1LQI) and Lqh3 (PDB ID: 1FH3) in a ‘sausage’ representation The Ca carbon trace is depicted as a tube with a radius proportional to the mean rmsd observed within the various conformers in the NMR ensemble a-Helices are highligh-ted in red; b-strands are colored in cyan The arrows point to the C-terminal segment

of the molecule (B) The rmsd of the Ca atoms in the solution structures of LqhaIT and Lqh3 For each model in the NMR ensemble (LqhaIT )29 structures [41]; Lqh3–30 structures [15]) the rmsd of each

Ca atom was calculated using the mean structure as reference The rmsd of the indi-vidual models were averaged and presented for each toxin residue.

Surprisingly, the pH-dependence of E10Y-E63R

mutant binding to cockroach sodium channels

decreased markedly (Table 3), and it was highly potent

at DmNav1 channels even at basic pH (Fig 6B)

Discussion

Insight into the molecular basis of preferential

interac-tions of scorpion a-toxins with insect or mammalian

Navs was thus far obtained mainly from mutagenesis

and comparison of bioactive surfaces and overall

struc-tures of pharmacologically distinct toxins These

analy-ses were based on available crystal structures of

a-toxins and their mutants and highlighted the NC

domain as a rigid structural entity, whose precise

topology dictates toxin specificity for various Nav

sub-types [10,14,20] Here we focused on the a-like toxin

Lqh3 because of its unique pharmacological features,

which suggested that structural flexibility rather than rigidity had an important role on its function

Comparison of the bioactive surface of Lqh3

to those of other a-toxins Molecular dissection of Lqh3 highlighted a bi-partite functional surface composed of a Core domain and an

NC domain (Fig 2), as was previously shown for the anti-insect toxin LqhaIT [10] The chemical nature of the Core domain is highly conserved among various scorpion a-toxins, and is predominated by positively charged and aromatic⁄ hydrophobic residues In Lqh3, substitution of Core-domain residues (His15, Phe17, Pro18, Phe39 and Leu45) had a profound effect on the binding energy (Table 1) Residue 15 (His or Glu in a-like toxins) is especially peculiar: It was not assigned

to the bioactive surface of LqhaIT or BmK M1 [10–13,19], but in Lqh3 it seems to be within atomic proximity of the channel receptor, because substitu-tions which increased its side chain volume (H15F⁄ L ⁄ R) reduced profoundly the binding affinity (Table 1) In addition, residue 15 is involved in toxin selectivity, as implied from the different effects of mutant H15A on insect versus rat skeletal muscle Navs (Table 2) Thus, the Core domain of Lqh3 plays an important role in both, interaction with the receptor site and toxin selectivity

The NC domain, composed of the five-residue turn and the C-terminal segment, varies in amino acid com-position and conformation among a-toxins (Fig 1), and was therefore suggested to play a role in toxin selectivity [10,13,14,19,22,23] In scorpion a-toxins (e.g LqhaIT, Aah2, BmK M1 and Lqh3), residue 58 (59 in Lqh3) is involved in an intricate network of intramo-lecular contacts, which contribute to C-tail stabiliza-tion relative to the molecule core Therefore, chemical modifications or substitutions at this region resulted in

a number of instances in marked alterations in struc-ture and function [11,22–25] Although the residue in position 58 is conserved in most scorpion a-toxins (Arg or Lys), its equivalent in a number of a-like tox-ins is hydrophobic⁄ aliphatic (e.g Ile59 in Lqh3; Fig 1), and is highly important for activity, as shown

in Lqh3 (Table 1) The mutagenic dissection high-lighted the importance of the C-tail residues Ile59, Lys64 and His66 for activity and selectivity, but not of residues at the five-residue turn (Tables 2 and 3) Whereas substitutions at the five-residue turn of LqhaIT and BmK M1 were shown to greatly affect the activity [10,11,19], mutagenesis at this region in Lqh3 had no effect (Table 1), suggesting that this structural motif was not involved in direct contact with

1.0

0.8

0.6

0.4

0.2

8.0 7.5

7.0 6.5

pH 0.0

Lqh3 E10Y-E63R

1.0

0.8

0.6

0.4

0.2

0.0

[Toxin] n M

10 0

10 -1

10 -2 10 1 10 2 10 3

E10Y-E63R Lqh3

A

B

Fig 6 Effects of mutant E10Y-E63R on the properties of

interac-tion with DmNa v 1 channels (A) Concentration–response relations

of the unmodified Lqh3 (s) and mutant E10Y-E63R (n) at pH 7.0.

Data were fitted using Hill equation (Eqn 1, Experimental

proce-dures) and the EC 50 values obtained are: Lqh3–10.5 ± 1.6 n M (n ¼

4), E10Y-E63R )6.5 ± 1.2 n M (n ¼ 3) (B) pH-dependent effect of

E10Y-E63R (n) compared with the unmodified toxin (s) Data were

collected and analyzed as in Fig 3.

the channel receptor site Still, the entire NC domain

is important for activity as indicated by the effect of

substitutions at the five-residue turn and C-tail on

toxin potency and its pH-dependent binding to insect

Navs (Table 2)

Dissociation of the toxin-receptor complex and

the slow association kinetics of Lqh3 are linked

to the flexibility of the C-tail

The substantial decrease in the sensitivity of binding to

alterations in pH of Lqh3 mutants modified at the NC

domain in residues other than His (Table 3), as well as

the slow onset of Lqh3 effect upon pH transitions

(Fig 3D), have raised the possibility that the NC

domain undergoes a slow conformational change along

the toxin binding process with the channel Close

inspection of the published Lqh3 solution structure

[15,21] has indicated a high degree of conformational heterogeneity of the NC domain especially around the short loop spanning residues 60–64 Detailed analysis

of the various backbone conformations of this loop have suggested that the majority (26 out of 29) of Lqh3 NMR models are divided between two main populations (Figs 7A,B), in which the overall topology

of the NC domain varies greatly (Fig 7B–F) They dif-fer in the side chains of His66, Glu10 and Glu63, which project to nearly opposite directions (Figs 7C,D), and in the intramolecular contacts among Gln8, Glu10 (of the five-residue turn), Glu63 and His66 (Fig 7C–F), whose substitution had profound effects on Lqh3 pH-dependent binding (Table 3) As a result, the five-residue turn adopts different confor-mations, although in both populations the backbone torsion angles around the Pro9–Glu10 bond are restricted to allow for a cis conformation Exchange of

Fig 7 Lqh3 solution structure exhibits two distinct conformations at its C-terminal seg-ment (A) Lqh3 NMR ensemble (PDB ID: 1FH3) was divided into two separate popula-tions, designated group A (16 structures) and group B (10 structures), and for each group, a geometric average structure was calculated using MOLMOL [42] The averaged rmsd of the backbone atoms of residues 57–67 from the average structure is presen-ted for each group, as well as for the com-plete NMR ensemble Three structures, which exhibited great structural variations and could not be classified into these two groups were omitted for clarity (B) Ca trace for residues 60–64 of NMR structures clas-sified to group A (red) or group B (blue) (C,D) The side chains of Gln8, Glu10, Glu63 and His66, whose substitution affected Lqh3 pH-dependent binding, is presented for two individual NMR structures that rep-resents two extreme conformations typify-ing the group A (C) and group B (D) structure populations (E,F) Comparison of the overall topology and disposition of the

NC domain relative to the molecule core in group A (E) versus group B (F) model NC-domain residues are colored as in (C, D); for all other residues only backbone atoms are displayed (gray).

conformations between the two populations would

involve the formation and breakdown of hydrogen

bonds and changes in the tilt and twist angles of the

backbone, and should be sensitive to the pH of the

medium As His residues contribute in part the

hydro-gen bonds that differ in the two molecule populations

(His66, His43; Fig 7C,D), such a conformational

change may provide the basis for the dependence of

Lqh3 binding on pH This hypothesis is supported by

the decreased sensitivity to pH of the E10Y-E63R

mutant, in which key Glu residues that

parti-cipate in hydrogen bond formation were eliminated

(Fig 4C)

On the basis of these structural observations and the

unchanged toxin association rate under various pH

values, as well as the slower toxin dissociation rate at

low pH [18], we speculate that upon toxin binding to

the channel, acidic pH favors toxin conformation with

high affinity for the receptor, and reduces the

proba-bility that the bound toxin spontaneously convert to

unfavorable conformations, hence stabilizing the

toxin-receptor complex In the case of the E10Y-E63R

mutations, elimination of critical hydrogen bonds

(Fig 7) allows it to assume conformation favourable

for the receptor at a wider pH range

To explain the mechanism of slow association of

Lqh3 to various Navs, we propose that the

rate-limit-ing step that governs Lqh3 bindrate-limit-ing is a slow transition

between the two conformational populations of the

toxin depicted in Fig 7 In the course of Lqh3 binding

to its Nav receptor, specific toxin conformers are

selec-ted from a dynamic ensemble of structures with

var-ious C-tail conformations (Fig 7) Such a mechanism

may also rationalize the broad-range potency of this

toxin on insect as well as mammalian peripheral and

brain Navs [5,16,18] A similar explanation might hold

for the slow effect on toxicity and a broad range of

activity of the site-3 sea anemone toxin Av2 [26], in

which the bioactive surface involves a highly flexible

Arg14 loop [26–28] The paradigm of dynamic

con-former selection was recently demonstrated for the

interaction between the cleavage factor component

pcf11 and the C-terminal domain of RNA polymerase II

[29] This C-terminal domain was found to exist in

solution as a dynamic disordered ensemble of

con-formers, and upon binding to pcf11 it assumed a

struc-tured conformation via induced fit This adaptation

ability enables RNA polymerase II C-terminal domain

region to bind specifically a broad range of factors

involved in mRNA processing [29] By analogy, the

ability of Lqh3 and possibly other members of the

a-like group to affect a wide range of Nav subtypes

may be attributed to their conformational flexibility

Experimental procedures

Bacterial strains and insects

and the BL21 (DE3, pLys) strain was used for toxin expres-sion using the pET-14b vector as was described previously [30,31] Sarcophaga falculata blowfly larvae were bred in the laboratory

Lqh3 expression

For expression in E coli we used the cDNA encoding Lqh3 isolated from a cDNA library constructed from the RNA

of the scorpion L quinquestriatus hebraeus Because expres-sion of Lqh3 using the pET-11c vector, as was described for the toxin LqhaIT [19], was poor, we used a fusion-partner strategy, whereby the N-terminus of Lqh3 was

HA-Lqh3 using the pET-14b vector as template DNA Primer 1, 5¢- GGCAGCCATATGTGTAATTGTAAGGCA CCAGAAACTGCACTTTGCGC-3¢, was designed to add

a sequence encoding Apamin and a linker cleavable by thrombin and Fx proteases at an NdeI site behind the

the anticipation for improved folding of the Lqh3 sequence behind The 3¢ region of this primer included 11 bases that overlapped the 5¢ region of Lqh3-cDNA Primer 2, 5¢- GGATCCGGCTGCTAACAAAGCCCGAAAGG-3¢, was designed for the opposite strand in reverse orientation at the 3¢ side of the Lqh3 gene, and contained a BamHI restriction site for insertion into pET-14b The PCR

BamHI and cloned into the corresponding restriction sites

in the polylinker of pET-14b The recombinant Lqh3, which accumulated in inclusion bodies, was folded in vitro following denaturation (in 6 m guanidinium-HCl, 0.1 m

glutathi-one) and renaturation (by 100-fold dropwise dilution into a 0.2 m ammonium acetate pH 8.0, 0.2 mm oxidized

col-umn, and HA-Lqh3 eluted as a single peak at 32% aceto-nitrile with a typical yield of 2 mg toxin per liter of E coli culture The high yield of recombinant toxin seems to involve both higher yield of inclusion bodies and improved

HA-Lqh3 exhibited a very similar activity to that of the native Lqh3 (purchased from Latoxan, Valence, France), in