Positively Charged Amino Acid Residues Located Similarly in Sea ...

3 downloads 0 Views 3MB Size Report
Jun 17, 2015 - Francis Miranda and Dr. Claude Granier for fruitful discussions. .... 190-195, BenjamidCummings Inc., Redwood City, CA. Natl. Acad. Sci.
Vol. 269, No. 24, Issue of June 17, PP. 16785-16788,1994 Printed in U.S.A.

THEJOURNAL OF B I O ~ I CCHEMISTRY AL 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Positively Charged Amino AcidResidues Located Similarly in Sea Anemone and Scorpion Toxins* (Received for publication, February 22, 1994)

Erwann P. LoretSO, Roberto Menendez Soto del Vallen, Pascal Mansuelle$,FranGois SampieriS, and Hem6 RochatS From the Waboratoire de Biochimie, CNRS URA 1455,Faculte de Medecine, Secteur Nord, 13916 Marseille Cedex 20, France a n d the YCentro Nacional de Inuestigaciones Cientificas, 6880-6990 La Habana, Cuba

toxins from Bunodosumu grunuliferu called BgI1 and Bg 111.' Specific groups of sea anemone and scorpion toxins compete on the same pharmacologicalsite, on the volt- Although their toxicity is markedly different, their sequences age-gated sodium channel of mammal excitable memshow that they are almost identical, with an asparagine inbranes. However, these scorpion and sea anemone tox- stead of an aspartic in position 16. Bg I1 is highly toxic on ins are two distinct protein families. Here we purified mammals and was a good candidate to be compared withAah and sequenced a new sea anemone toxin, Bg11, highly 11, the mostactivescorpiontoxin.Molecularmodelingwas toxic to mammals and also a less toxic mutant, Bg 111. made by energy minimization and dynamics using sequence T w o Bg I1 models were determined from sequence hohomologies with two sea anemones having their three-dimenmologies with two sea anemone toxin two-dimensional sional structure determined by two-dimensional nuclear magN M R structures. Only one model conformed to circular netic resonance. Each structure provides a Bg I1 model, but dichroism data obtained from BgI1 and was compared only one conformed to circular dichroism data. Comparison of with an x-ray structure aofscorpion toxin. The comparison of the two structures shows that5 amino acid resi- Bg I1 and Aah I1 shows that the two proteins havea surface of dues are located similarly in the sea anemone toxin andsimilar size. When the two proteins are overlaidin function of the scorpion toxin. From these 5 residues, 4 are basic their center of mass and their common surface adjusted, 5 residues occurred to be similarly located in the two toxins: 4 residues, constituting two distinct positively charged basic residues and 1 tyrosine. poles on the surface of these toxins. In the sea anemone of the mutant isolated, a negative charge beside one MATERIALS ANDMETHODS positive poles decreases the toxicity. These results show Purification-Sea anemone were collectedin theGulf of Mexico area. that positively charged amino acid residues could be essential for the activityof these toxinsand outline the Gel filtration was made with Sephadex G15 (Pharmacia) in10%acetic roleofelectrostaticbonds in theinteraction of sea acid with a 100 x 4-cm column. Reversephase HPLC was made with a C8 Hibar-Merck column (30 x 1 cm) using water containing 0.1% trianemone and scorpion toxins with their receptor.

fluoroacetic acid and acetonitrile containing 0.1% trifluoroacetic acid. IonexchangeHPLCwasmade with an analytical sulfopropyl TSK Sea anemone toxinsare short proteins(4000-5000 Da) cross- column using 10 and 500 mM ammonium acetate (pH 4)buffers. Activity Test-Toxicity was assessed by a lethality test after24 h and linked by three disulfidebridges ( 1 4 ) . Their binding is potencarried out using 20 2-gmale C57/BL 6 mice. The toxic solution, tial-dependent and induces openingof the channel followed by supplemented with 1%serum albumin, was injected intracerebrovena slow closing phase (5-8). They are divided i n three classes, tricularly. Synaptosomalpreparations were obtained from homogenates two made u p of molecules containing 46-49 amino acid resiof rat brain (see Ref. 12 fordetails of the toxicity test and binding assay). dues and one of shorter polypeptides containing27-31 residues ''I iodination of AaH I1 was made with lactoperoxidase,and iodinated (3). Sequence homologies were used to classify the 46-49-resi- derivatives were purified with C8-HPLC. CD Measurements-Proteins were in 10 mM phosphate buffer (pH due-long toxinsas type 1 and type 2. These twotypes have the 7.5) for W-CD measurements. CD spectra were measured in 50-pm same electrophysiological activity(9). Scorpion toxins acting on path length cells from 260 to 180 nm with a Jobin-Yvon UV CD specmammalian sodium channel are proteins of 60-65 residues trophotometer. Measurements were performedat 20 "C.CD spectra are long cross-linked by four disulfide bridges. They are divided reported as A€ per amide. The content in secondary structures was (Y and p; a-type toxins bind, in a voltage-depend- determined by the analysis of the CD data according to the method of into two types, ent manner, on site111, inducing a prolongation of the channel Manavalan and Johnson (13). Purified proteins were analyzed for opening, whereas p-type toxins bind on site IV, inducing repeti- amino acid content and the concentration determined on a Beckman 126AA System Gold HPLC Amino Acid Analyzer. tive closing and opening of the channel (10). The binding of Molecular Modelsof Bg 1I-The Bg 1Ia and Bg IIb models were build p-toxins is not voltage-dependent.Sea anemone toxins compete with the Insight 11, Homology,Discover, and Delphi programs from with a-type scorpion toxins and can prevent their binding on BIOSYM Technologies,Inc., running on a Silicon Graphics XS 24 R3000 rat brain synaptosomes, although the reverse is not true (11). Indigo Workstation. The structures were optimized with the consistent Unexpectedly, there is no homology among the sea anemone valence force field in terms of the internal energies, using the Van der Waals energy to monitor each step of the modeling. The pH was set up and scorpion toxin sequences. at 7. Minimization was performed withsteepest descent and ctnjugate Here, we used a new protocol to purify two sea anemones gradient algorithms with a maximum derivative of 0.1 kcal/A in the final steps. Dynamic steps were performed at 300 K for 10 ps.

* The costs of publication of this article were defrayed in partby the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C.Section1734solelyto indicate this fact. 5 To whom correspondence should be addressed. Tel.: 33-91-69-88 53; Fax: 33-91-65-75-95.

The abbreviations used are: Bg I1 and Bg 111, toxins purified from the seaanemone B. granulifera; ATX Ia and BDS I, toxins purified from the sea anemone Anemonia sulcata; Bg IIa, model made from ATX Ia; Bg IIb, model made from BDS I, Aah 11, toxin purified from the scorpion venom Androctonus australis hector; HPLC, high performance liquid chromatography.

16785

16786

Similarities in 20

BDS ATX Bg 11

&

&2pG % G &;

Sea Anemone and Scorpion Toxin Structures

:::: 46 aa

FIG.1. BDS I, B g 11, and ATX Ia sea anemone sequences. The boxes indicate sequence homologies between these three toxins. Bg I11 is a mutant where Asn-16 is replaced by an aspartic acid. RESULTS AND DISCUSSION

Sea anemones (B. granulifera) were collected and immediately frozen in liquid nitrogen and then they were homogenized in water at 4 "C and centrifuged at 20,000 x g for 20 min. Toxicity (20 mgkg on mice) was in the supernatant, which was freeze-dried (fraction 1). Fraction 1 was submitted to a 10% Wavelength t n m ) acetic acid extraction, where only the soluble materials were toxic (fraction 2). Fraction 2 was submitted toa gel filtration in FIG.2. Circular dichroic spectra of Bg II (solid line) and Bg 111 10% acetic acid giving one toxic fraction (fraction 3). Fraction 3 (dashed line). Proteins are in 20 mM phosphate buffer (pH7). A€ is on was submitted to reverse phase HPLC giving two toxic frac- a per amide basis expressed as M - ~cm". tions. The most toxic fraction (fraction 4) was submitted toion exchange HPLC givingtwo proteins called Bg I1 and Bg 111.Bg BDS-I has a sequence and a three-dimensional structure difI1 and Bg I11 correspond to 0.2 and 0.1%of the protein material ferent from ATX Ia, the Bg IIb model showed a three-dimenand 44% of the total toxicity from fraction 1. sional structure similar to ATX Ia (Fig. 3).As could be expected, Intracerebroventricular) injection on mice C57/BL6 showed the Bg IIa model is very similar t o ATX Ia; however Bg IIb has that Bg I1 has a higher toxicity (0.4 pgkg) compared with Bg I11 the lowest energy compared with Bg IIa (Table I). Both Bg IIa (21 pgkg). Competition assays in binding to rat brain synap- and Bg IIb had no a-helix structure in theirfinal step of minitosomes showed that Bg I1 and Bg I11 compete with the lZ5I- mization; however, Bg IIb had thesequence 5-12 in helicoidal labeled a-type scorpion toxin Aah 11, whereas Aah 11 cannot structure, and this sequence was easily transformed ina n a-hebind on synaptosomes after preincubation with Bg I1 (results lix by putting residues8 and 11closer, t o form a hydrogen bond. not shown). In competition experiments,the half-effect ob- Attempts to locate a two-turn a-helix in another partof Bg IIb served was 9 nM for Bg I1 and 72 nM for Bg 111. or in Bg IIa was not successful and led to a dramatic increase The first 34 amino-terminal residues were assigned by se- in the totalenergy of the molecule. The Bg IIb model was preferred toBg IIa, because it had the quencing of 1nmol of either S-alkylatedBg I1 and Bg I11 (using a n Applied Biosystems 473A sequencer). Digestion of both S- lowest energy and agreed with CD data results. The x-ray alkylated Bg I1 and Bg I11 with chymotrypsin gave a peptide structure of the a scorpion toxin Aah I1 (18)was superimposed that extended the sequence to residue39. Digestion with Arg-C with Bg IIb. Bg IIb hassurface adjusting witha surface present protease gave a peptide that extended the sequence to residue in the Aah 11, which is 64 residues long, although there is no 48 for Bg I1 as well as Bg 111. Carboxypeptidase A digestion possible superposition of the two backbones (Fig. 4u). If we provided the same C terminus sequence Cys-Lys-Gln-COOH assume that thesurface of a toxin is important for binding on for both Bg I1 and Bg 111. Bg I11 is a mutant where Am-16is its target, this observation provides a first explanation for the replaced by an asparticacid. Bg I1 and Bg I11 show similarities similar pharmacological properties of these two toxins. A secwith type1sea anemonetoxins represented by ATX Ia (Fig. 1). ond observation is the similar spatial location of 4 basic resitwo toxins: Arg-5, Arg-14, Arg-27, and Arg-36 for Bg Asn-16 is conserved among type 1 sea anemonetoxins and dues in the should be important for their toxicity, since Bg I1 is 50 times 11, with respectively, Arg-62, Lys-58, Lys-50, and Arg-18 for Aah 11. In thesetwo toxins the 4 basic residues are grouped in two more active than Bg 111. pairs located at two opposite extremities of the toxins, constiThe CD shows that these two proteins have similar structures (Fig. 2). The analysis of the CD spectra according to the tuting two positive poles (Fig. 4b). These structural homologies method of Manavalan and Johnson (13) gives for these two could explain the interactionof both toxins with the same site on the sodium channel. The 2 acidic residues from Bg 11,Asp-7 toxins: 7 residues (14%) in a-helix structure, 17 residues in antiparallel P-sheet (36%), 13 residues in p-turn (27%), and 11 and Asp-9, have no equivalent in theAah I1 structure, which is residues in other structure(22%). Bg I1 and Bg I11 have a high illustrated by the different location of the negative electrostatic content in antiparallel and p-turn structure, which is rather potential (Fig. 4b). Trp-32 for Bg I1 and Tyr-47 for Aah I1 are in usual in sea anemone toxins, nevertheless no helix structure similar locations. No other functional similarities were found was detected in seaanemone two-dimensionalNMR structures among Bg I1 and Aah 11. Arg-14 is conserved among type 1 and type 2 sea anemone known actually (3). Two models of Bg 11, Bg IIb, and Bg IIa were builtfrom the toxins and waspointed out as highly important for toxicity (19). two-dimensional NMR structures of the sea anemones BDS I Amazingly, Fig. 3 shows that the Arg-14 of ATX Ia does not superimpose with the Arg-12 of BDS I. Bg IIb has the Arg-5 (14)and ATX Ia (15), respectively, bothavailable inthe Brookhaven data bank(16).ATX Ia has65% sequence homology superimposing with the ATX Ia Arg-14, whereas the Arg-14 with Bg I1 and belongs to typeI sea anemone group(31,whereas superimposes with the BDS I Arg-12. Bg IIa has Arg-5 and BDS I has only 32% sequence homology with Bg I1 and seems Arg-14 close t o the ATX Ia Arg-14 (Fig. 3). Arg-36 is replaced by to belong neither to type I or type I1 (Fig. 1).The pharmacology a well conserved histidine in type1 sea anemone toxins. k g - 5 of these two toxins is different. ATX Ia is ratheractive on crus- has no equivalent in type 1 toxins, but corresponds to a well conserved lysine in type2 sea anemone toxins.No equivalent is taceans and shows a poor activity on mammals with a LD,, of found for Arg-27. Arg-62, Lys-58, Lys-50 and Arg-18 are well 236 pgkg on mice (intracerebroventricularly) and a Kd of 7 in binding affinity to rat brain synaptosomes (3). BDS I is not conserved residues among a-typescorpion toxins. From these 4 toxic on mammals but appears bind to to the same receptor site residues, Lys-58 and Arg-18 are missing in Bot XI, an a-type as the a-scorpion toxins(17). Although the sea anemone proteintoxin homologous to Aah11, but almostnontoxic (20). Chemical

Similarities in Sea Anemone and Scorpion Toxin Structures

16787

TABLEI Energies for the Bg II models Energies were minimized by using the consistent valence force field from DISCOVER (BIOSYM)to a maximum derivative of 0.1 k c d A in the final steps. Dynamic steps were performeda t 300 K for 10 picoseconds. The final energies of Bg IIa and Bg IIb are compared with their respective starting models ATX Ia (15) and BDS I (14). Energies kcallmol

ATX Ia

Bg IIa

Van der Waals 523.5 717.1 164.8 1312 133.7 165.1 Coulombic 688.7 Total

-264.6 452.5

Bg IIb

withBg ub

201.3 -278.4 1513.5 -113.6

-299.9 -166.2

BDS

No a-helix structures were foundin the final structure of Bg IIa and Bg IIb, although the CD data analysis of Bg 11in solution at pH 7 shows that this protein has a two-turn a-helix. Bg IIb had the sequence 5-12 in helicoidal structure, and this sequence was easily transformed in a-helix in putting residues 8 and 11closer to form a hydrogen bond.

FIG. 3. Three-dimensional structure of the sea anemonesATX Ia (yellow) and BDS I (green) superimposed with the Bg IIa (blue) and Bg IJh (purple) models. In red are shown the lateral chains ofArg-14 and Arg-37 fromATX Ia, Arg-12 from BDS I, and Arg-5, Arg-14, Arg-27,and Arg-36 from BgIIa and Bg 1%.The structures were superimposed according to their three disulfide bridges. Bg IIa was built from sequence homologies with ATX Ia (15).The coordinatesof the ATX Ia atoms amino acid residues were directly used to build Bg IIa when there was a strict sequence homology of 3 residues or more, whereas only the Ca was used as a coordinate in case of partial homology (Fig. 1).Coordinates of sequences foundin other proteins available in the protein data bank (16) were used for nonhomologous sequence between BgI1 and ATX Ia. Then, after a splice-repairing step and creation of the disulfide bridges, a steepest descent algorithm was used in a fist step of energy minimization, followed bydynamics a t 300 K. A final step of energy minimization was donewith a conjugate gradients algorithm. The use of cross-term calculations was not possible in this last step of minimization. TheBg IIb starting model was BDS-I(14).Bg IIb was built in a same manner than Bg IIa; however, the dynamics step led to a structure very low in energy that made it possible to use cross-term calculations in thelast step of energy minimization with the conjugate gradients algorithm. The sequence 5-12 had a helicoidal structure, and this sequence was easily transformed in a two-turn a-helix by a dynamic step of 300 K with a tethering of 2 A between the amide oxygen of residue 11and the amide hydrogen of the residue 8 (Table I).

modifications show that basic amino acid residues are important for the activity of both scorpion toxins (21, 22) and sea anemone toxins (1-3). Mutation experiments show that basic residues are important for the activity of scorpion toxinsacting on potassium channels, whereas acidicresidues are important in thepotassium channel binding site for interacting with scorpion toxins (23, 24).Potassium and sodium channels have sequence homologies(25-28); in particular, the extracellular loop, where the sea anemone and scorpion toxins binding site (26, 28) is located, contains well conserved acidic amino acid residues. Our work agrees with this observation, showingthat

FIG.4. Cornpanson or me rrg UD model and the an u &ray structure. In a, Bg IIb is in blue and Aah I1 in yellow. The Bg 11 residues 5, 14,27,and 36 are inpurple, and the Aah 11residues 18,50, 56, and 58 are in red. The polypeptidic backboneis represented with a ribbon. In b, Bg IIb is on the left and Aah I1 on the rigfit, represented, at pH 7,in blue for the positive (+1KcaVmol) and in red for the negative (-1 Kcdmol) electrostatic equipotential surface.

4 basic residues could be essential for the sea anemone and scorpiontoxins interaction with the sodium channel of the mammals. The mutation of the asparagine 16 in aspartic acid decreases the toxicity. This mutation has no effect onthe structure,since the two toxins have similar CD spectra (Fig. 2). The shift from Asp-16 to Asn-16 in Bg IIb induced an increase of 10 kcaVmol in the Van der Waals energies of the Bg IIb models and showed no detectable difference in structure (datanot shown).The decrease in toxicity is probably due to the presence of one additional negative charge at the surface of Bg 111.This observation confirms indirectly the role of the positive poles, sincethe decrease of the toxicity observed forBg I11 can be explained by a modification of the electrostatic interactions with the sodium channel of one ofthe positive poles, where the Arg-14 is located.

16788

Similarities in Sea Anemone and Scorpion Toxin Structures

Positively charged amino acid residues could be expected to play a role with a receptor located on a channel dedicated to the transfer of positively charged ions. Moreover, sea anemone and scorpion toxins are basic proteins. Scorpion toxins have an hydrophobic surface, and it was proposed that scorpion toxins interact with the sodium channel by hydrophobic contacts (29). There is no hydrophobic surface in sea anemone toxins. Furthermore, electrostatic interactions are more specific compared with hydrophobic interactions, mainly because there is no real hydrophobic bondingbut rather hydrophobic effectsin protein interactions (30).The aim of the Bg I1 structural studywas to show that among different possibilities, the basic residues that are located similarly with Aah I1 are in conserved structural positions in sea anemone toxins (Fig. 3). From an evolutionary point of view, it is interesting to see how two very distantly related family of proteins were able to end up with structures completely different but having the same pharmacological target. Acknowledgments-We thank Mesdames ThBrSse Brando and Maryse Alvitre for technical assistance. We thank Dr. J. C. FontecillaCamps for providingus with the Aah I1 crystal structure data and Prof. Francis Miranda and Dr. Claude Granier for fruitful discussions. REFERENCES 1. Kem, W. R. (1988)in The Biology ofNemataysts (Hessinger, D., and Lenhoff, H., eds) pp. 375-405,Academic Press, New York 2. Beress, L. (1988)in Poisonous and Venomous Marine Animals of the World (Halstead, B. W., ed) 2nd Ed., pp. 152-161,Darwin Press, Princeton, NJ 3. Norton, R. S. (1991)Tbxicon 29, 1051-1084 4. Loret, E.P. & Hammock, B. K. (1994)in Scorpion Biology and Research (Brownell, P., and Polis, G., eds) Oxford University Press, Oxford, in press 5. Rochat, H., Bernard, C. & Couraud, F. (1979)in Advances In Cytopharmacology (Cecarelli, B., and Clementi, F., eds) Vol. 3, pp. 325-334,Raven Press, New York

6. Simard, M. & Watt, D. D. (1990)in The Biology of Scorpion (Polis, G., ed) pp. 415-444, Stanford University Press, San Francisco, CA 7. Catterall, W.A. (1980)Annu. Rev. Phannacol. Tbxicol. 20, 15-43 8. Possani, L. D. (1984)in Handbook of Natural Tbxins (Tu, A. T., ed) Vol. 2,pp 513450,Marcel Dekker, Inc., New York 9. Salgado, V.L. & Kem, W. R. (1992)Zbxicon 30, 1365-1381 10. Couraud, F. & Jover, E. (1984)in Handbook of Natural Zbxins (Tu, A. T., ed) Vol. 2,pp. 513-550, Marcel Dekker, Inc., New York 11. Schweitz, H., Bidard, J. N., Frelin, C., P a w n , D., Vijverberg, H. P. M., Mahasneh, D. M., Lazdunski, M., Vilbois, F. & lkugita, A. (1985)Biochemistry

24,3554-3561 12. Loret, E.P.,Martin-Eauclaire, M. F., Mansuelle, P., Sampieri, F., Granier, C. & Rochat, H. (1991)Biochemistry SO, 633-640 13. Manavalan, P. & Johnson, W.C. (1987)Anal. Biochem. 167,7645 14. Driscol, P. C., Gmnenborn, A. M., Beress, L. & Clore, G. M. (1989)Bimhemistry 28, 2188-2198 15. Widmer, H., Billeter, M. & Wiithrich, K (1989)Proteins 6,357-371 16. Bernstein, F. C., Koetzle, T. F., Williams, G. J. B., Meyer, E. F., Brice, M. D., Rodgers, J. R., Kennard, O., Shimanouchi, 0. & Tasumi, M. (1977)J. Mol. Biol. 112, 535-542 17. Liewellyn, L. E.& Norton, R. S. (1991)Biochem. Int. 24,937-946 18. Fontecilla-Camps, J. L., Habersetzer-Rochat, C. & Rochat, H. (1988)Proc. Natl. Acad. Sci. U.S.A. 85,7443-7447 19. Could, A. R., Mabbutt, B.C. & Norton, R. S. (1990)Eur. J. Biochem. 189, 145-153 20. Sampieri, E , Habersetzer-Rochat, C., Martin, M. F., Kopeyan, C. & Rochat, H. (1987)Int. J . Pept. Protein Res. 29,231-237 21. Loret, E. P., Mansuelle, P., Rochat, H. & Granier, C. (1990)Biochemistry 29, 1492-1501 22. Kharrat, R.,Darbon, H., Rochat, H. & Granier, C. (1989)Eur. J. Biochem. 181, 381-390 23. MacKinnon, R., Latorre, R. & Miller, C. (1989)Biochemistry 28,8092-8099 24. MacKinnon, R., Heginbotham, L. & Abramson, T. (1990)Neuron 5,767-771 25. Numa, S. (1989)Harvey Lect. 83,121-165 26. Catterall, W.A. (1988)Science 242, 50-61 27. Hille, B. (1991)in Ionic Channels ofExcitable Membranes (Hille, B., ed) 2nd Ed., pp. 236-258, Sinauer Associates Inc., Sunderland, MA 28. Kubo, Y., Baldwin, T. J., Jan, Y.N. & Jan, L. Y. (1993)Nature 362, 127-133 29. Fontecilla-Camps, J. C., Almassy, R. J., Ealick, S. E., Suddath, F. L.,Watt, D. D., Feldmann, R. J. & Buggs, C. E.(1981)%rids Biochem. Sci. 6,291-296 30. Mathews, C. K. & van Holde, K. E.(1990)in Biochemistry (Weisberg, s.,ed) pp. pp. 190-195, BenjamidCummings Inc., Redwood City, CA