Structural Analysis of the Laetiporus sulphureus HemolyticPore-forming Lectin in Complex with Sugars*

Received for publication, December 10, 2004, and in revised form, January 25, 2005Published, JBC Papers in Press, February 1, 2005, DOI 10.1074/jbc.M413933200

Jose M. Mancheno‡§, Hiroaki Tateno¶, Irwin J. Goldstein¶, Martın Martınez-Ripoll‡,and Juan A. Hermoso‡

From the ‡Grupo de Cristalografıa Macromolecular y Biologıa Estructural, Instituto Rocasolano, CSIC, Serrano, 119,28006 Madrid, Spain and the ¶Department of Biological Chemistry, University of Michigan Medical School,Ann Arbor, Michigan 48109-0606

LSL is a lectin produced by the parasitic mushroomLaetiporus sulphureus, which exhibits hemolytic andhemagglutinating activities. Here, we report the crystalstructure of LSL refined to 2.6-Å resolution determinedby the single isomorphous replacement method with theanomalous scatter (SIRAS) signal of a platinum deriva-tive. The structure reveals that LSL is hexameric, whichwas also shown by analytical ultracentrifugation. Themonomeric protein (35 kDa) consists of two distinctmodules: an N-terminal lectin module and a pore-form-ing module. The lectin module has a �-trefoil scaffoldthat bears structural similarities to those present intoxins known to interact with galactose-related carbo-hydrates such as the hemagglutinin component (HA1) ofthe progenitor toxin from Clostridium botulinum, abrin,and ricin. On the other hand, the C-terminal pore-form-ing module (composed of domains 2 and 3) exhibitsthree-dimensional structural resemblances with do-mains 3 and 4 of the �-pore-forming toxin aerolysin fromthe Gram-negative bacterium Aeromonas hydrophila,and domains 2 and 3 from the �-toxin from Clostridiumperfringens. This finding reveals the existence of com-mon structural elements within the aerolysin-like fam-ily of toxins that could be directly involved in mem-brane-pore formation. The crystal structures of thecomplexes of LSL with lactose and N-acetyllactosaminereveal two dissacharide-binding sites per subunit andpermits the identification of critical residues involvedin sugar binding.

Hemolytic lectins are sugar-binding proteins that lyse andagglutinate cells. Recently, a novel toxic lectin from the para-sitic mushroom Laetiporus sulphureus (LSL)1 that showsstrong hemagglutination and hemolytic activity has been char-

acterized (1). These activities are mediated by binding carbo-hydrates as demonstrated by the correlation found betweenboth processes in sugar inhibition assays, which reveals Lac-NAc (Gal�14GlcNAc) as a potent inhibitor. The N-terminalmodule of LSL is a carbohydrate recognition domain as de-duced from the C-terminal deletion mutant of LSL, LSLa-D1,which shows hemagglutination, but not hemolytic activity.Conversely, the C-terminal module shows structural similarityto pore-forming bacterial toxins suggesting a similar mode ofbiological activity. In this sense, LSL-induced hemolysis pro-ceeds through a mechanism involving the formation of ion-permeable membrane pores with a functional diameter smallerthan 3.8 nm, as revealed by osmotic protection experiments (1).After pore formation, erythrocytes are ruptured by a colloid-osmotic lysis mechanism. Thus, LSL appears to exhibit twofunctional modules: lectin and pore-forming. LSL is a non-covalent-linked oligomeric structure, in which the C-terminaldomains play a crucial role as deduced by the monomeric andhighly polar character of LSLa-D1.

Here we report the crystal structure of the novel eukaryoticpore-forming toxin present in the parasitic mushroom L. sul-phureus. The two-module structure of LSL described hereinsuggests the existence of mechanisms of evolution involving theassociation of discrete stable functional modules into a finalmodular protein with toxic character. Additionally, the three-dimensional similarity between LSL and aerolysin from thebacterium Aeromonas hydrophila (2) and also �-toxin fromClostridium perfringens (3), indicates a mechanism of poreformation in which amphipathic �-hairpins associate into atransmembrane �-barrel, as recently proposed for �-toxin ofClostridium septicum (4), a homologous toxin to aerolysin.

EXPERIMENTAL PROCEDURES

Materials—Lactose (Gal�14Glc) was purchased from Sigma. LacNAcwas synthesized and available from other studies. The lectin from themushroom L. sulphureus was prepared as described by Tateno andGoldstein (1).

Analytical Ultracentrifugation—Equilibrium ultracentrifugation ex-periments were performed at 10,500 and 12,500 � g at 20 °C, using aBeckman XL-A ultracentrifuge with an An-50Ti rotor and standarddouble sector centerpiece cells. Solvent density (1.005 mg/ml) and thepartial specific volume of LSL (0.723) were calculated from the buffercomposition (137 mM NaCl, 3 mM KCl, 10 mM Na2HPO4, and 2 mM

KH2PO4) and from the predicted amino acid composition, respectively,with SEDNTERP (5). Data from sedimentation velocity and equilib-rium experiments were analyzed using the programs Sedfit (6) andSedphat (7), respectively.

Data Collection and Processing—Initial diffraction data were col-lected on hexagonal LSL crystals obtained as previously described (8).For derivatization, we soaked the crystals for 24 h in crystallizationsolution containing 20 mM K2Pt(CN)4. The crystals were then trans-ferred to the cryoprotectant solution (soaking solution plus 30% (v/v)glycerol) for �5 s and flash-frozen in the cold nitrogen stream (100 K).

* This work was supported by Ministerio de Educacion y CienciaGrant BFU2004-01554/BMC (to J. M. M., M. M.-R., and J. A. H.) andNational Institutes of Health Research Grant GM29477 (to I. J. G.). Thecosts of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked “adver-tisement” in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

The atomic coordinates and structure factors (codes 1W3A, 1W3F, and1W3G) have been deposited in the Protein Data Bank, Research Collabo-ratory for Structural Bioinformatics, Rutgers University, New Bruns-wick, NJ (http://www.rcsb.org/).

§ To whom correspondence should be addressed: Grupo de Crista-lografıa Macromolecular y Biologıa Estructural, Inst. Rocasolano,CSIC, Serrano, 119, 28006 Madrid, Spain. Tel.: 34-915619400; Fax:34-915642431; E-mail: xjosemi@iqfr.csic.es.

1 The abbreviations used are: LSL, lectin from Laetiporus sulphureus;LSLa-D1, C-terminal deletion mutant of LSL; LacNAc, N-acetyllac-tosamine; PFM, pore-forming module; HA1, hemagglutinin 1.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 280, No. 17, Issue of April 29, pp. 17251–17259, 2005© 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org 17251

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

All data were collected at beamline ID14-1 (� � 0.934 Å) at the ESRF(Grenoble, France), using an ADSC Q4 CCD detector. Data were in-dexed, integrated, and scaled with the CCP4 suite (9). Data statisticsare summarized in Table I. Diffraction data from crystals ofLSL�LacNAc complexes were collected in-house on an Kappa 2000 CCDdetector with Cu K� X-rays generated by a Nonius FR-591 rotating-anode generator equipped with Montel mirrors and operated at 45 kVand 100 mA. Data were indexed, integrated, and scaled with Denzo andScalepack (10). Subsequent data manipulation was carried out with theCCP4 software suite (9).

Structure Determination and Refinement—The structure of LSL withbound Lac (see below) was determined by the method of single isomor-phous replacement using the anomalous scatter signal of a platinumderivative. The platinum atoms were located on both isomorphous andanomalous difference Patterson maps. Refinement of heavy atom pa-rameters and phase calculations to 2.6-Å resolution were performedwith SOLVE (11), MLPHARE (9), and SHARP (12). Solvent flatteningand histogram matching were performed with DM (9) assuming asolvent content of 40%, i.e. two LSL molecules in the asymmetric unit,rendered a low-quality electron density map. On the contrary, theelectron density map obtained considering one LSL molecule in theasymmetric unit (70% solvent content), was of excellent quality, andenabled the incorporation of a complete protein model using O (13).Only two long loops in the C-terminal module did not show interpret-able density. After refinement, only two regions of the atomic model,those comprising residues 180–186 and 258–263, are not defined in theelectron density map. The initial model had a crystallographic R factorof 40% for all reflections in the resolution range of 50–2.6 Å. Structurerefinement was performed using the programs CNS (14) and REFMAC5(15). Model building was done with O (11). The final model has an Rfactor of 22.7% and a free R factor of 27.9%, and included 312 amino

acid residues, 1 Lac, 6 glycerols, and 77 solvent molecules. The struc-tures of the LSL�LacNAc complexes were determined by the molecularreplacement method with MOLREP (9) using the structure of LSL assearch model. The quality of the models were checked using the pro-gram PROCHECK (16). Phasing and refinement statistics are summa-rized in Table I.

RESULTS AND DISCUSSION

Overall Description of the Structure—The monomeric LSLmolecule consists of two different structural modules corre-sponding to the previously reported functional modules, lectinand pore-forming (Fig. 1). The lectin module (residues 3–150)has a globular structure, consisting of a �-trefoil with approx-imate dimensions 39 � 32 � 32 Å3. The structure of the com-plexes with Lac and LacNAc reveals two sugar binding sites.The pore-forming module (residues 151–314) consists of anelongated lobe with dimensions 72 � 23 � 23 Å3. This modulecan be effectively divided into two regions: domain 2 is formedby a twisted five-stranded anti-parallel �-sheet together with along amphipathic loop located in one face of the sheet, anddomain 3 consisting of a �-sandwich.

The N-terminal Lectin Module—The N-terminal module ofLSL consists of a �-trefoil (Fig. 2), a characteristic fold previ-ously described in other lectins (17–19). It is formed by asix-stranded anti-parallel �-barrel capped on one end by threetwo-stranded hairpins (strand pairs �2 and �3, �6 and �7, �10and �11) (Fig. 2A). These last three hairpins, the hairpin tri-

FIG. 1. Three-dimensional structure and topology diagram of LSL. A, ribbon model of monomeric LSL. The figure is colored from the Nto C terminus in a progression from blue to red via green. Numbers 1–23 indicate the corresponding �-strands and H1–H7 are the single-turn 310helices. B, topology diagram of LSL. �-Strands and 310 helices are represented by arrows and cylinders, respectively. The starting and endingsequence numbers for each secondary structural element are given. The lectin module is represented in green and the pore-forming module inbrown. A was prepared with BOBSCRIPT (52) and RASTER3D (53).

Hemolytic Lectin from the Mushroom L. sulphureus17252

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

plet, arrange in a triangular array that gives rise to a pseudo3-fold symmetry. The global structure of the �-trefoil is basedon tandem repetition of a basic motif (called �, �, and �, re-spectively) composed of four �-strands separated by threeloops, the third one containing a single-turn 310 helix (Fig. 2B).In this regard, it has been previously suggested that the exist-ence of a primordial galactose-binding peptide of �40 residueswould be the ancestor of the basic motifs of the �-trefoil scaffold(20). Analysis of the primary (Fig. 2C) and tertiary structure(Fig. 2D) of the three motifs reveals that in effect they arehomologous with each other, but also reveal the existence ofclear divergences. Thus, although pairwise sequence compari-sons between motifs reveal identities around 20%, only fourconserved residues are contributed by each motif (Fig. 2C); yet,despite the primary degeneration, the three-dimensional struc-ture of the peptide backbone of the motifs is essentially con-

served (Fig. 2D). Doubtless, the interplay between three-di-mensional conservation and primary departures shouldprovide the structural basis for the different sugar bindingproperties of each motif (see below).

Structural comparison of the N-terminal module of LSL,with known folds using the DALI algorithm (21), revealed ahigh homology with proteins known to interact with sugars, forexample, the hemagglutinin component (HA1) of the progenitortoxin from Clostridium botulinum (22) (1qxm-A, Z � 17.6),basic fibroblast growth factor (23) (1bgf-A, Z � 16.8), Amaran-thus caudatus agglutinin (24) (1jly-A, Z � 15.4), inositol 1,4,5-triphosphate receptor (25) (1n4k-A, Z � 14.7), mannose recep-tor fragment (26) (1dqg-A, Z � 13.7), hisactophilin (27) (1hce-A,Z � 13.2), the toxin abrin (28) (1abr-B, Z � 12.8) and thetetanus neurotoxin fragment (29) (1a8d, Z � 10.4). Perhaps,the most notable of the DALI results is the remarkable resem-

FIG. 2. Structure and pseudo-symmetry of the LSL lectin module. A, ribbon diagram of the �-trefoil module of LSL (residues 1–148)showing the �-barrel (blue) and hairpin triplet (magenta) substructures. B, ribbon diagram of the �-trefoil in which the three basic motifs aredepicted in different colors: �-motif, blue; �-motif, green; �-motif, orange. C, structure-based sequence comparison of the individual motifs of theLSL �-trefoil. Background color coding: pairwise identities, yellow; conserved residues in the three motifs, pink. Character color coding: residuescontacting the bound carbohydrate in the �-motif, blue; residues contacting the bound carbohydrate in the �-motif, red. The �-strands are indicatedby black arrows, and the 310 helices by a red cylinder. D, superimposition of the C� traces of the three individual motifs. Calculated root meansquare deviations with the program O (13) were 0.85 (� and �; 40 C�s), 1.4 (� and �; 39 C�s), and 1.2 (� and �; 39 C�s).

Hemolytic Lectin from the Mushroom L. sulphureus 17253

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

blance of the N-terminal module of LSL with sugar-bindingdomains of toxins that exert their cytotoxic action by bindingglycoproteins; particularly, the hemagglutinin component(HA1), which has been shown to increase the toxicity ofC. botulinum neurotoxin by binding to oligosaccharides liningthe intestine (30, 31), or abrin and the closely related deadlyplant toxin ricin (20), which bind their target membranesthrough a �-trefoil motif. The emergence of this structuralmotif in toxins from diverse sources could suggest a mechanismof evolution involving the association of discrete stable func-tional modules into a final modular protein with toxiccharacter.

Sugar-binding Sites—Previous work demonstrated that Lac,LacNAc, and other galactose-related saccharides inhibited thehemagglutination and hemolytic activity of LSL (1). The corre-lation between both processes indicates that sugar binding isinvolved in both molecular mechanisms. Because experiments

of osmotic protection of erythrocytes showed that hemolysisresults from the formation of discrete pores in the membranes,it can be inferred that the molecular mechanism of pore forma-tion by LSL involves the initial, specific recognition ofcarbohydrates.

The x-ray crystal structures of LSL complexed with Lac orLacNAc show sugar binding at two (� and �) of the threepossible sites. Moreover, the soaking experiments consideredherein reveal significantly different sugar affinities betweenthe above two binding sites. The electron density map of theLSL�Lac complex was of so excellent a quality that it waspossible to model a Lac molecule into the � motif of the �-trefoil.The Fo � Fc map initially calculated without including thedisaccharide molecule showed clear density around Asp-141.The Lac molecule could be modeled in this difference densitymap (Fig. 3A). Failure to remove all the Lac in the final affinitychromatographic step in the purification of LSL is the likely

FIG. 3. Lactose and N-acetyllactosamine binding to the lectin module of LSL. A, structure of the Lac molecule bound to the �-motif ofLSL. B, three-dimensional structure of LacNAc bound to the �-motif of LSL. The models show the Fo � Fc electron density maps contoured at 2.5�around the Lac molecule initially calculated without including the disaccharide. C, LacNAc bound to the �-motif. The Fo � Fc electron density mapcontoured at 3� calculated without including the disaccharide is shown in blue around the bound carbohydrate. Broken lines indicate hydrogenbonds between sugar and protein residues. Color coding of secondary structure elements: �-motif, blue; �-motif, red, �-motif, green. D, molecularsurface of the �-trefoil scaffold, with the LacNAc molecules (as stick models) bound to the �- and �-sites. The surface is colored according to theelectrostatic potential, blue for positive and red for negative. E, structure of Lac bound to one binding site in ricin B-chain (Protein Data Bank code2AAI; Ref. 40). A–C were prepared with MOLSCRIPT (54) and RASTER3D (53); the molecular surface was built with the program GRASP (55).

Hemolytic Lectin from the Mushroom L. sulphureus17254

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

source of this molecule (1). Additionally, a well ordered glycerolmolecule is found in each of the other two motifs, these mole-cules coming from the short soak in the cryoprotectant solutionthat contained 30% (v/v) glycerol. In this sense, it has beensuggested that this compound together with ethylene glycol isgood at mimicking carbohydrate binding (22).

The � site binds Lac at its nonreducing galactose (Gal) moi-ety (Fig. 3A). The aromatic ring of Phe-139 makes van derWaals interactions with the plane formed by the C-3, C-4, C-5,and C-6 carbon atoms of the Gal residue in a similar fashion tothat seen in other protein-carbohydrate complexes (32, 33). Theaxial C4 hydroxyl group can form hydrogen bonds with the ringN-� of His-125 and with the carboxyl oxygen atoms of Asp-141.On the other hand, the C-6 hydroxyl group hydrogen bonds toN�-2 of Gln-142. Furthermore, the ring O-5 coordinates a watermolecule that in turn can form hydrogen bonds with the car-bonyl oxygen of Asn-132 and the ring N� of His-133. The onlyinteraction involving the glucose residue is that observed be-tween the C3� hydroxyl group and the NH1 and NH2 atoms ofArg-123. Consistent with this pattern of toxin-Lac interactions,the x-ray structures of some bacterial toxins complexed withLac show that interaction with the disaccharide is mediatedessentially by the galactose (34–36).

Recent studies revealed the higher efficiency of LacNAc rel-ative to Lac in inhibiting both hemagglutination and hemolysis(1). Quick soaking experiments using cryoprotectant solutioncontaining 20 mM LacNAc results in the incorporation of thismolecule into the � site of LSL as revealed by the unambiguousdifference electron density (Fig. 3B). The structure of the com-plex was determined at 2.58-Å resolution (Table I) by molecularreplacement (see “Experimental Procedures”). Whereas the Lacportion of LacNAc can perfectly be superposed to the Lac ligand

present in the LSL�Lac complex, the presence of the acetamidomoiety promoted several new interactions between sugar andprotein. Thus, the carbonyl oxygen of the N-acetyl moiety ofLacNAc can form hydrogen bonds with the ring N�-1 atom ofTrp-131 and with nitrogen atoms NH1 and NH2 of Arg-76;additionally, a water molecule mediates a hydrogen bond be-tween the nitrogen atom of the acetamido moiety and thepeptidic oxygen of Gln-127 (Fig. 3B). No additional sugar bind-ing was observed in the other two potential LSL sites apartfrom the presence of a glycerol molecule in each motif.

In addition to the � site of LSL, the � site also binds LacNAcas revealed by overnight soaking experiments in the presenceof 0.1 M LacNAc. Under these experimental conditions, LSLbinds two LacNAc molecules, one at the � site and one at the �site. In this second binding site, the aromatic ring of Tyr-91 andthe side chains of Asp-93 and Asn-94 play similar roles to thoseof Phe-139, Asp-141, and Gln-142 present in the � motif, re-spectively (Fig. 3C). In this case, however, the axial C4 hy-droxyl group can only form hydrogen bonds with the carboxyloxygen atoms of Asp-93, as there is no amino acid residueequivalent to His-125 in the � site. An important differencewith respect to the � site is the absence of water moleculesmediating interactions between sugar and protein presumablybecause of critical amino acid changes. Thus, the presence ofIle-85 (equivalent to His-133) and the absence of a residueequivalent to Arg-76 would prevent the stabilization of a watermolecule that similarly to the � site would coordinate the O-5atom of the galactose residue. On the other hand, the bulkyside chain of Phe-73 (equivalent to Val-121), which interactswith Arg-75 through cation-� interactions, sterically hindersthe presence of a solvent molecule that would stabilize theacetamido moiety as observed in the � motif. Finally, no further

TABLE IData collection, phasing, and refinement statistics

Data setsLSL

LSL:Lac LSL:Lac Pt-derivative LSL:Nal LSL:(Nal)2

Data collection statisticsSpace group P6322 P6322 P6322 P6322Wavelength (Å) 0.934 0.934 1.542 1.542Resolution range (Å) 50–2.6 50–2.6 40–2.5 50–2.6Measured reflections 164,769 578,057 170,624 264,427Unique reflections 17,585 18,325 19,024 17,326Completenessa (%) 99.0 (99.0) 99.8 (99.8) 98.1 (98.1) 99.9 (99.7)Redundancya 9.4 (6.9) 31.5 (31.4) 9.0 (9.7) 15.3 (5.1)Rsym (%)a,b 9.3 (40.8) 13.6 (40.4) 9.0 (43.9) 14.7 (45.5)I/�(I)a 6.1 (1.5) 4.3 (1.6) 7.6 (1.4) 15.6 (2.2)

Phasing statisticsNumber of Pt sites 2Rcullis (%)c 81.4Figure of meritBefore density modification 0.32After density modification 0.81

Refinement statisticsResolution range (Å) 15–2.6 15–2.5 15–2.6No. of protein residues/atoms 312/2467 312/2467 312/2467No. of Lac residues/atoms 1/23No. of LacNAc residues/atoms 1/26 2/52No. of glycerol residues/atoms 6/36 5/30 4/24No. of solvent atoms 77 127 124

Ramachandran plotNo. of residues in core regions 312 312 312No. of residues in disallowed regions 0 0 0Rfac (%) 22.6 23.0 22.8Rfree (%) 27.9 27.2 28.2

Root mean square deviationsBond lengths (Å) 0.010 0.007 0.007Bond angles (°) 1.4 1.4 1.4

a Values in parentheses are for the highest resolution shell.b Rsym � �h�jh�hj � �Ih��/�h�j�Ihj�, where h represents a unique reflection and j symmetry-equivalent indices. I is the observed intensity and �I�

is the mean value of I.c Rcullis � �h��FPH � FH� � FH�/�h�FPH � FH�, summed over centric reflections.

Hemolytic Lectin from the Mushroom L. sulphureus 17255

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

stabilization of this last group is observed as there is no homol-ogous side chain to Trp-131. As a result of the above amino acidchanges between both sugar-binding sites, the electrostaticproperties of the surface of both sites are different (Fig. 3D).These results clearly indicate the reduced number of stabilizinginteractions between the disaccharide LacNAc in the � sitewith respect to the � site that may explain their differences inaffinity for this sugar. This correlates well with the averagevalues of the atomic B-factors of the bound ligands in theLSL�(LacNAc)2 complex that indicate a lower occupancy of the� site as compared with the � site (62 versus 38 Å2; averagevalue for protein atoms in both sites �40 Å2).

Finally, it is worth noting that although conserving the samestructural framework of the �- and �-site, the �-site does notbind sugars in our experimental conditions. In this regard, thepresence of Ala-43 instead of an aromatic residue equivalent toTyr-91 (�-site) or Phe-137 (�-site) may be critical for explainingthe lack of sugar binding.

The structure of the above described complexes between LSLand Lac or LacNAc present close structural similarities toothers previously reported for lectin-lactose complexes (33, 37–39). Thus, the existence of stacking interactions between theGal ring and an aromatic side chain, together with hydrogenbonds between the axial C4 hydroxyl group of Gal and an acidiclectin side chain, are strictly conserved. This finding is inagreement with the lack of sugar binding by LSL in the �-site.In the particular case of the ricin B-chain (20, 40), which iscomposed of two �-trefoils each one with a galactose-bindingsite, Gal ring hydrophobically stacks against Trp-37 or Tyr-248, and also interacts through its C-4 hydroxyl group with thecarboxyl group of Asp-22 and Asp-234. For comparison pur-poses, Fig. 3E shows the structure of the Gal-binding site of theC-terminal �-trefoil of the ricin B-chain.

The C-terminal Pore-forming Module—The pore-formingmodule (PFM) of LSL (residues 149–314) has an elongatedshape, 70 Å long and 20–40 Å in thickness (Figs. 1 and 4).Apart from a single-turn 310 helix, the �-structure is the onlyregular secondary structure in the module, consisting of 11

�-strands (�13 to �23), the last two strands crossing it from tipto tip. The PFM of LSL can be suitably divided into two do-mains (Fig. 4). Domain 2 is a highly twisted five-strandedanti-parallel �-sheet capped on one side by an amphipathic loop(residues 212–241). Besides, domain 3 consists of a �-sandwichformed by a two- and a three-stranded anti-parallel sheets,respectively. In contrast to the results obtained with the lectinmodule, structural comparison of the PFM of LSL using theDALI algorithm rendered no positive hits. Still, LSL sequencecomparison searches showed structural similarity to mosquito-cidal toxin (MTX2) and � toxin from C. septicum (1), this lastbelonging to the aerolysin-like family of pore-forming toxins (4,41). Interestingly, comparison of the structure of LSL withthose of aerolysin (2) and �-toxin from C. perfringens (3), theonly members of the family with known three-dimensionalstructure, reveals structural similarities between domains 2and 3 of LSL with domains 3 and 4 of aerolysin, and domains 2and 3 of �-toxin, respectively (Fig. 4A). Markedly, this compar-ison permits the inference for the presence of common struc-tural elements that could be directly involved in membranepore formation. On the other hand, these common structuralfeatures are distinct from the aerolysin-like family of toxins: aset composed of a five-stranded �-sheet and an amphipathicloop, and a distal �-sandwich. According to the low sequenceidentity between these proteins (20%), an evident feature ofthese elements is their versatility in the sense that they ac-commodate very different sequences without significant struc-tural departures (Fig. 4A). Furthermore, it is interesting tonote that in contrast to aerolysin and �-toxin, the PFM of LSLconstitutes a structural module that is composed of a continu-ous stretch of the protein sequence. Yet, it is remarkable thatthe topology or connectivity between the above common struc-tural elements and the other structural elements is identicalfor the three proteins.

Regarding the specific role of the above mentioned structuralelements in pore formation, an exhaustive analysis carried outrecently on C. septicum � toxin has shown that the equivalentregion to the amphipathic loop becomes an amphipathic trans-

FIG. 4. Three-dimensional similari-ties between aerolysin, �-toxin, andLSL. A, ribbon diagrams of the bacterial�-pore-forming toxins aerolysin fromA. hydrophila (2), �-toxin C. perfringens(3), and LSL from L. sulphureus. The five-stranded �-sheets are depicted in blue,and the �-sandwiches in red; amphipathicloops are indicated in yellow. d1–d4 arefor the corresponding domains. Domains1 and 2 from aerolysin, and domain 1 fromboth �-toxin and LSL are shown in gray.B, sequence alignment of the putativetransmembrane domains of the aerolysin-like family (LSL, aerolysin, �-toxin,�-toxin, and enterolobin). Residues con-served in at least three members areshown in red. Positions with a negativevalue of average free energy of transfer(G kcal/mol) from water to a membraneinterface according to Wimley et al. (42)are indicated with the blue background.

Hemolytic Lectin from the Mushroom L. sulphureus17256

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

membrane �-hairpin when the C. septicum � toxin oligomerwas inserted into the membrane (4). An analysis of the putativetransmembrane domains of the aerolysin-like family in termsof position-dependent average free energy of transfer (G kcal/mol) from water to either n-octanol or to a membrane interfacecalculated according to Wimley et al. (42) demonstrates thepresence of an almost perfectly conserved alternating patternof hydrophobic residues that is characteristic of amphipathictransmembrane �-hairpins. The crystal structure of LSL showsthat the size of the amphipathic �-hairpin, �30 residues, issimilar to Staphylococcus aureus �-hemolysin (43), the anthraxprotective antigen (44), and the two transmembrane hairpinsof perfringolysin O (45, 46). In summary, these results indicatethat LSL is a � pore-forming toxin (47) capable of forming atransmembrane �-barrel by contributing individual �-hairpins.

Oligomeric State—The biochemical characterization of LSLhas shown that both native and recombinant LSL are oligo-meric structures (1). Estimation of the molecular mass by SDS-PAGE and gel filtration chromatography indicated that LSLexists as a tetramer of subunits of �35 kDa. Unexpectedly, thecrystal structure of LSL reveals the existence of hexameric LSLassemblies (Fig. 5). Considering the discrepancy between thedifferent experimental approaches, we carried out analyticalultracentrifugation assays (Fig. 6). The sedimentation velocityanalysis of LSL in the 0.18–1.10 mg/ml concentration rangedemonstrated that the protein is very homogeneous, with asedimentation coefficient of 8.56 S (data not shown). Addition-

ally, sedimentation equilibrium analyses indicate that the dataare well described by a monomer-hexamer association equilib-rium (Fig. 6), which is essentially displaced to the oligomericform. A fit to a monomer-dimer or monomer-tetramer equilib-rium gave much less satisfactory fits. The average molecularmass for 0.18 and 1.10 mg/ml samples were 204.8 and 204.5kDa, respectively. The hexamer-monomer equilibrium dissoci-ation constant is about 10 �M5. Thus, these results demonstratethat LSL associates as a hexamer in solution, in agreementwith the crystal structure.

The LSL monomers associate as a highly symmetrical hex-amer, in which protomers essentially interact through their�-sandwich domains forming three extended intersubunit six-stranded anti-parallel �-sheets (Fig. 5). LSL monomers areorganized around 3-fold crystallographic axes with the �14strand of each protomer contacting the C-terminal fraction ofthe �23 strand of the adjacent protomer. Tripoid-like trimersrelated by 2-fold crystallographic axes strongly interact exclu-sively through their �-sandwiches (�21 and �23 strands) toform hexamers endowed with three intersubunit large �-sand-wiches (Fig. 5B). Lattice contacts between hexamers aremainly mediated by water molecules, and with the exception ofa hydrogen bond between the carboxyl oxygen of Asp-56 andthe amide group of Gly-207, involve exclusively the N-terminaldomains of neighboring molecules. The limited number of lat-tice contacts and the 70% solvent content closely resemble asolution environment. Analyses of solvent accessible areas bur-

FIG. 5. Oligomeric structure of LSL.A, two views of the LSL hexamer, relatedby a 90° rotation about the x axis. Mono-mers are shown as ribbon models (uppertrimer) and as polypeptide traces (lowertrimer) in different colors. B, two views(related by a 180° rotation about the yaxis) of the contacting region between do-mains 3 of adjacent monomers (as ribbonmodels; red and green). This associationdetermines the formation of a large andsymmetrical �-sandwich. Residues di-rectly involved in intersubunit contactsare shown as stick models. Note the pres-ence of the crystallographic 2-fold symme-try axis (oblong solid circle). Strands �23and �21 are shown as semitransparentarrows in the left and right parts of thisfigure, respectively.

Hemolytic Lectin from the Mushroom L. sulphureus 17257

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

ied upon hexamer formation reveal some striking features.Thus, the formation of each of the three intersubunit �-sand-wiches within the hexamer involves a contact area of �680 Å2

per monomer. Although this contact area is relatively small, itis essentially hydrophobic (in fact, an important hydrophobiccore is formed upon oligomerization involving, Ala-252, Phe-254, Val-256, Ala-258, Leu-306, and Leu-310 of each protomer),which agrees well with a relatively strong and non-obligatecomplex (48, 49). Additionally, as the two �-sandwiches in-volved are related by a 2-fold crystallographic axis (i.e. isolo-gous association), this region is highly symmetric, which wouldmake the building of the oligomeric structure more economic(50). Interestingly, the polar residues contributed by the two�-strands involved in oligomerization (�21 and �23) are Thrresidues, which may indicate a possible role for small andhydroxylated residues in protomer association. In this regard,LSL has a remarkably high percentage of Ser and Thr residuesin the PFM (27.3 versus 12.9% in the lectin module). Theseresidues are mainly concentrated in the region flanking theamphipathic loop, constituting the 42% (8 of 19) and 46% (6 of13) of the total residues of �16 and �21, respectively. On theother hand, although the association of the lectin modulesinvolves a contact area of �730 Å2 per monomer, this region isessentially polar, which is typical of a very weak association(48, 49), in agreement with the monomeric character of theC-terminal LSL deletion mutant lacking the region betweenresidues 187 and 314 (LSLa-D1) (1). Finally, the oligomericcharacter of LSL in solution raises important questions regard-ing the structural rearrangements required for membrane poreformation, which are currently under study.

Interestingly, Uchida et al. (51) have recently reported thecrystal structure of the Ca2�-dependent hemolytic lectin CEL-

III from Cucumaria echinata. CEL-III is organized in threedomains: two N-terminal �-trefoil domains (domains 1 and 2),and a C-terminal pore-forming domain (domain 3). As is thecase for LSL, this modular design is consistent with evolutionprocesses involving the association of discrete stable functionalmodules into a final modular protein with toxic character.Although sequence comparison analysis (not shown) reveals nosignificant similarity between �-trefoils from CEL-III and LSL,their three-dimensional structure is highly conserved, withroot mean square C� deviations of 1.53 Å (for 103 atoms) and1.61 Å (for 102 atoms) when comparing LSL domain 1 withCEL-III domains 1 and 2, respectively. Nevertheless, despitethis structural similarity a different mechanism of �-galacto-side binding can be expected between these proteins as theenvironment of the sugar-binding motifs are clearly dissimilar.Thus, although an acidic residue (equivalent to Asp-93 or Asp-141 in LSL) is present in all except one of the CEL-III motifs(Asp-43, Asp-141, Asp-188, Asp-229, Asp-276), there are noaromatic residues equivalent to Tyr-91 or Phe-139, which inLSL stack against the Gal ring. This comparison in turn re-veals two additional aspects: first, it is precisely in the motif ofCEL-III that lacks the acidic residue equivalent to that of LSLwhere no Ca2� binding is observed (51) and second, the abovementioned equivalent acidic residues are accompanied by twoadditional aspartic residues (except motif 1�, which has onlyone) and an aromatic residue (Tyr-36, Tyr-133, Tyr-181, Tyr-222, and Trp-269), which may be involved in sugar binding assuggested (51). Finally, regarding the PFM of both hemolyticlectins despite the absence of three-dimensional similarity be-tween them, it is remarkable that similarly to LSL the regionproposed by the above authors as responsible for pore forma-tion (�-helices H-8 and H-9) is very rich in Ser and Thr residues(�34% for the sequence stretch comprising residues 318–360),indicating a potential functional role for these residues in thepore formation mechanism.

Acknowledgments—We thank Dr. J. L. Saiz-Velasco for analyticalultracentrifugation data analysis and the ESRF (Grenoble, France) forprovision of synchrotron radiation facilities. In particular we thankSofıa Macedo for assistance in using the beamline ID14-2.

REFERENCES

1. Tateno, H., and Goldstein, I. J. (2003) J. Biol. Chem. 278, 40455–404632. Parker, M. W., Buckley, T., Postma, J. P., Tucker, A. D., Leonard, K., Pattus,

F., and Tsernoglou, D. (1994) Nature 367, 292–2953. Cole, A. R., Gibert, M., Popoff, M., Moss, D. S., Titball, R. W., and Basak, A. K.

(2004) Nat. Struct. Mol. Biol. 11, 797–7984. Melton, J. A., Parker, M. W., Rossjohn, J., Buckley, J. T., and Tweten, R. K.

(2004) J. Biol. Chem. 279, 14315–143225. Laue, T. M., Shah, B. D., Ridgeway, T. M., and Pelletier, S. L. (1992) Biochem-

istry and Polymer Science, Royal Society of Chemistry, London6. Schuck, P. (2000) Biophys. J. 78, 1606–16197. Schuck, P. (2003) Anal. Biochem. 320, 104–1248. Mancheno, J. M., Tateno, H., Goldstein, I. J., and Hermoso, J. A. (2004) Acta

Crystallogr. Sect. D Biol. Crystallogr. 60, 1139–11419. Collaborative Computational Project, Number 4 (1994) Acta Crystallogr. Sect.

D Biol. Crystallogr. 50, 760–76310. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307–32611. Terwilliger, T. C., and Berendzen, J. (1999) Acta Crystallogr. Sect. D Biol.

Crystallogr. 55, 849–86112. de la Fortelle, E., and Bricogne, G. (1997) Methods Enzymol. 276, 472–49413. Jones, T. A., Zou, J-Y., Cowtan, S. W., and Kjelgaard, M. (1991) Acta Crystal-

log. Sect. A 47, 110–11914. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-

Kunstleve, R. W., Jiang, J.-S., Kuszewski, J., Nilges, M., Pannu, N. S.,Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) ActaCrystallogr. Sect. D Biol. Crystallogr. 54, 905–921

15. Murshudov, G. N., Vagin, A. A., and Dobson, E. J. (1997) Acta Crystallogr.Sect. D Biol. Crystallogr. 53, 240–255

16. Laskowski, R., MacArthur, M., Moss, D., and Thornton, J. M. (1993) J. Appl.Crystallogr. 26, 283–291

17. McLachlan, A. D. (1979) J. Mol. Biol. 133, 557–56318. Murzin, A. G., Lesk, A. M., and Chothia, C. (1992) J. Mol. Biol. 223, 531–54319. Hazes, B. (1996) Protein Sci. 5, 1490–150120. Rutenber, E., Ready, M., and Robertus, J. D. (1987) Nature 326, 624–62621. Holm, L., and Sander. C. (1993) J. Mol. Biol. 233, 123–13822. Inoue, K., Sobhany, M., Transue, T. R., Oguma, K., Pedersen, L. C., and

Negishi, M. (2003) Microbiology 149, 3361–3370

FIG. 6. Analytical equilibrium ultracentrifugation analysis ofLSL. Lower, sedimentation equilibrium analysis of LSL (9.37 �M) at10,500 rpm (open circles) and 12,500 (open squares). Absorbance at 280nm is plotted against the radial position from the center of the rotor.The best fit to the data set (solid line curves) is a monomer-hexamerassociation equilibrium with a dissociation constant about 10 �M5.Best-fit weight average molecular mass is 212.6 kDa. Residuals fromthis fit are shown in the top two panels.

Hemolytic Lectin from the Mushroom L. sulphureus17258

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

23. Faham, S., Hileman, R. E., Fromm, J. R., Linhardt, R. J., and Rees, D. C.(1996) Science 271, 1116–1120

24. Transue, T. R., Smith, A. K., Mo, H., Goldstein, I. J., and Saper, M. A. (1997)Nat. Struct. Biol. 4, 779–783

25. Bosanac, I., Alattia, J. R., Mal, T. K., Chan, J., Talarico, S., Tong, F. K., Tong,K. I., Yoshikawa, F., Furuichi, T., Iwai, M., Michikawa, T., Mikoshiba, K.,and Ikura, M. (2002) Nature 420, 696–700

26. Liu, Y., Chirino, A. J., Misulovin, Z., Leteux, C., Feizi, T., Nussenzweig, M. C.,and Bjorkman, P. J. (2000) J. Exp. Med. 191, 1105–1116

27. Habazettl, J., Gondol, D., Wiltscheck, R., Otlewski, J., Schleicher, M., andHolak, T. A. (1992) Nature 359, 855–858

28. Tahirov, T. H., Lu, T. H., Liaw, Y. C., Chen, Y. L., and Lin, J. Y. (1995) J. Mol.Biol. 250, 354–367

29. Knapp, M., Segelke, B., and Rupp, B. (1998) Am. Crystallogr. Assoc. Abstr.Papers (Annu. Meeting) 25, 90

30. Fujinaga, Y., Inoue, K., Watanabe, S., Yokota, K., Hirai, Y., Nagamachi, E.,and Oguma, K. (1997) Microbiology 143, 3841–3847

31. Fujinaga, Y., Inoue, K., Nomura, T., Sasaki, J., Marvaud, J. C., Popoff, M. R.,Kozaki, S., and Oguma, K. (2000) FEBS Lett. 467, 179–183

32. Vyas, N. K. (1991) Curr. Opin. Struct. Biol. 1, 732–74033. Rini, J. M. (1995) Annu. Rev. Biophys. Biomol. Struct. 24, 551–57734. Sixma, T. K., Pronk, S. E., Kalk, K. H., van Zanten, B. A. M., Berghuis, A. M.,

and Hol, W. G. (1992) Nature 355, 561–56435. Sixma, T. K., Kalk, K. H., van Zanten, B. A., Dauter, Z., Kingma, Z., Witholt,

B., and Hol, W. G. (1993) J. Mol. Biol. 230, 890–91836. Merritt, E. A., Sarfaty, S., van der Akker, F., L’Hoir, C., Martial, J. A., and Hol,

W. G. (1994) Protein Sci. 3, 166–17537. Cioci, G., Mitchell, E. P., Gautier, C., Wimmerova, M., Sudakevitz, D., Perez,

S., Gilboa-Garber, N., and Imberty, A. (2003) FEBS Lett. 555, 297–301

38. Walker, J. R., Nagar, B., Young, N. M., Hirama, T., and Rini, J. M. (2004)Biochemistry 43, 3783–3792

39. Poget, S. F., Legge, G. B., Proctor, M. R., Butler, P. J., Bycroft, M., andWilliams, R. L. (1999) J. Mol. Biol. 290, 867–879

40. Rutenber, E., and Robertus, J. D. (1991) Proteins 10, 260–26941. Ballard, J., Bryant, A., Stevens, D., and Tweten, R. K. (1992) Infect. Immun.

60, 784–79042. Wimley, W. C., and White, S. H. (1996) Nat. Struct. Biol. 3, 842–84843. Song, L. Z., Hobaugh, M. R., Shustak, C., Cheley, S., Bayley, H., and Gouaux,

J. E. (1996) Science 274, 1859–186644. Benson, E. L., Huynh, P. D., Finkelstein, A., and Collier, R. J. (1998) Biochem-

istry 37, 3941–394845. Shepard, L. A., Heuck, A. P., Hamman, B. D., Rossjohn, J., Parker, M. W.,

Ryan, K. R., Johnson, A. E., and Tweten, R. K. (1998) Biochemistry 37,14563–14574

46. Shatursky, O., Heuck, A. P., Shepard, L. A., Rossjohn, J., Parker, M. W.,Johnson, A. E., and Tweten, R. K. (1999) Cell 99, 293–299

47. Heuck, A. P., Tweten, R. K., and Johnson, A. E. (2001) Biochemistry 40,9065–9073

48. Nooren, I. M., and Thornton, J. M. (2003) EMBO J. 22, 3486–349249. Nooren, I. M., and Thornton, J. M. (2003) J. Mol. Biol. 325, 991–101850. Blundell, T. L., and Srinivasan, N. (1996) Proc. Natl. Acad. Sci. U. S. A. 93,

14243–1424851. Uchida, T., Yamasaki, T., Eto, S., Sugawara, H., Kurisu, G., Nakagawa, A.,

Kusunoki, M., and Hatakeyama, T. (2004) J. Biol. Chem. 279, 37133–3714152. Esnouf, R. M. (1999) Acta Crystallogr. Sect. D Biol. Crystallogr. 55, 938–94053. Merrit, E. A., and Bacon, D. J. (1997) Methods Enzymol. 277, 505–52454. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946–95055. Nicholls, A., Sharp, K. A., and Honig, B. (1991) Proteins 11, 281–296

Hemolytic Lectin from the Mushroom L. sulphureus 17259

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from

A. HermosoJosé M. Mancheño, Hiroaki Tateno, Irwin J. Goldstein, Martín Martínez-Ripoll and Juan

Complex with Sugars Hemolytic Pore-forming Lectin inLaetiporus sulphureusStructural Analysis of the

doi: 10.1074/jbc.M413933200 originally published online February 1, 20052005, 280:17251-17259.J. Biol. Chem. 

  10.1074/jbc.M413933200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/280/17/17251.full.html#ref-list-1

This article cites 54 references, 9 of which can be accessed free at

by guest on January 3, 2019http://w

ww

.jbc.org/D

ownloaded from