Archives of Biochemistry and Biophysics 543 (2014) 31–39

Contents lists available at ScienceDirect

Archives of Biochemistry and Biophysics

journal homepage: www.elsevier .com/ locate/yabbi

Vasorelaxant activity of Canavalia grandiflora seed lectin: A structuralanalysis

0003-9861/$ - see front matter � 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.abb.2013.12.006

⇑ Corresponding author. Address: BioMol-Lab, Departamento de Bioquímica eBiologia Molecular, Universidade Federal do Ceará, P.O. Box 6043, 60.455-970Fortaleza, Ceará, Brazil. Fax: +55 85 3366 9818.

E-mail address: bscavada@ufc.br (B.S. Cavada).1 Abbreviations used: ConGF, Canavalia grandiflora; CRD, carbohydrate recognition

domain; ConM, Canavalia maritima; ConA, Canavalia ensiformis; EDRF, endothelium-derived relaxing factors; DVL, Dioclea violacea; DRL, Dioclea rostrata; DvirL, Diocletavirgata lectin; CGL, Canavalia gladiata lectin; LLG, log-likelihood-gain; RFZ, rotationfunction z-score; TFZ, translation function z-score; DDA, data dependent acquisitionfunction; CID, collision-induced dissociation; NOS, nitric oxide synthase; L-NAME, N-nitro-L- arginine methyl ester; RMSD, root mean square deviation.

Ito Liberato Barroso-Neto a, Rafael Conceição Simões a, Bruno Anderson Matias Rocha a,Maria Julia Barbosa Bezerra a, Francisco Nascimento Pereira-Junior a, Vinicius José Silva Osterne a,Kyria Santiago Nascimento a, Celso Shiniti Nagano a,b, Plinio Delatorre c, Maria Gonçalves Pereira d,Alana Freitas Pires d, Alexandre Holanda Sampaio a,b, Ana Maria Sampaio Assreuy d,Benildo Sousa Cavada a,⇑a Laboratório de Moléculas Biologicamente Ativas, Universidade Federal do Ceará, Fortaleza, Brazilb Departamento de Engenharia de Pesca, Universidade Federal do Ceará, Fortaleza, Brazilc Departamento de Biologia Molecular, Universidade Federal da Paraíba, João Pessoa, Brazild Instituto Superior de Ciências Biomédicas, Universidade Estadual do Ceará, Fortaleza, Brazil

a r t i c l e i n f o

Article history:Received 30 October 2013and in revised form 2 December 2013Available online 17 December 2013

Keywords:Canavalia grandifloraLectinConA-likeVasorelaxationStructure

a b s t r a c t

Lectins are comprised of a large family of proteins capable of the specific and reversible recognition ofcarbohydrates. Legume lectins, the most studied plant lectins, show high structural similarity, butwith modifications that imply a variation in the intensity of some biological activities. In this work,the primary and tertiary structures of Canavalia grandiflora (ConGF) were determined. ConGF, a lectinisolated from C. grandiflora seeds, is able to induce relaxant activity in rat aortic rings. The completesequence of ConGF comprises 237 amino acids. This particular protein has primary sequence varia-tions commonly found in lectins from Dioclea and Canavalia genera. The protein structure was solvedat 2.3 Å resolution by X-ray crystallography. An X-Man molecule was modeled into the carbohydraterecognition domain. Still, ConGF (30 and 100 lg mL�1) elicited 25% of vasorelaxation (IC50 =34.48 ± 5.07 lg mL�1) in endothelialized aortic rings. A nonselective inhibitor of nitric oxide blockedConGF relaxant effect, showing mediation by nitric oxide. Key distances between ConGF carbohydraterecognition domain residues were determined in order to explain this effect, in turn revealing somestructural aspects that could differentiate lectins from the Canavalia genera with respect to differentefficacy in vasorelaxant effect.

� 2013 Elsevier Inc. All rights reserved.

Introduction

Lectins are proteins able to specifically and reversibly recognizecarbohydrates without modifying them through the carbohydrate-recognition domain (CRD)1 [1]. Closely phylogenetically related lec-tins from the Legume subtribe Diocleinae present small differences in

crucial positions of the primary structure, and the impact of thesechanges has been shown through several biological assays [1–7].CRDs are highly conserved in Diocleinae lectins, and, thus far, onlyCPL, the lectin of Camptosema pedicellatum seeds, has presented apoint mutation on the hydrophobic subsite from the carbohydrate-binding domain. Importantly, key distances between residues thatcomprise the CRD affect its shape and, hence, the volume of eachdomain, characteristics that directly affect carbohydrate recognitionand the ability to elicit biological effects [3,4,7].

Besides the shape and volume of the CRD, the ratio between thedimeric and tetrameric forms of lectins in solution changes byalterations in pH, i.e., pH-dependent oligomerization, reflectingdifferences in the association of monomers affecting the biologicalactivity of lectins. More specifically, the relative orientation of theCRD amino acids in tetramers affect the spatial arrangement ofcarbohydrate binding sites and determine the ability of lectins todistinguish saccharides and bind to glycoconjugates of cellsurface [1,8].

32 I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39

Structural analysis of lectins can further clarify their mecha-nisms of action and the structure–function relationship. An impor-tant example is the substitution of Pro202 by Ser202 in the lectin ofCanavalia maritima (ConM). This modification leads to conforma-tional alterations in the CRD at position 12 (tyrosine). This is a un-ique alteration in this Diocleinae lectin sequence, and it has highsimilarity with Canavalia ensiformis lectin (ConA). In comparisonwith ConA, this modification contributes to the extended CRD ofConM, allowing the approximation of Tyr12 in the carbohydrate-binding site and the emergence of a recurrent H-bond betweenthe Tyr12 hydroxyl oxygen and the hydrogen of the disaccharideouter glucose [9,10]. Thus, the structural features of ConM CRD ex-plain its high affinity for disaccharides [8].

Endothelial cells exhibit surface glycoproteins that can interactwith CRD, thereby facilitating lectin cell recognition and the re-lease of endothelium-derived relaxing factors (EDRFs) from bloodvessels [11,12]. Homologous lectins from Dioclea violacea (DVL)and Dioclea rostrata (DRL) seeds exhibit marked differences in theirrelaxation activities in vascular smooth muscle (43% for DVL and96% for DRL). Despite their otherwise remarkable similarity, differ-ences in the distance between key residues present in CRD resultsin a wider and shallower CRD in DRL that favors vasorelaxation [7].In another example, while Diocleta virgata lectin (DvirL) shows bet-ter efficacy in NO induction compared to DVL, it is inferior to thatof DRL, a phenomenon that may also be explained by the distinctlocation of the residues in CRD. Specifically, Diocleinae lectins withlarger distances between Arg228 and Leu99 and between Arg228 andTyr12 present interactions that are numerous enough to stabilize asecond unit of a disaccharide. These types of lectins do not displaythe Tyr12 pattern and have affinity for monosaccharides [6].

Different lectins of the Canavalia genus, such as ConA, ConBr(from Canavalia brasiliensis seeds), CGL (from Canavalia gladiataseeds) and ConM all exhibit relaxant effects on precontractedendothelialized rat aorta [9,13]. Bezerra and coworkers suggestedCRD as the most important factor in determining NO inductionby Diocleinae lectins. Based on this, it can be proposed that thelow ability of ConBr to induce NO production results from its smal-ler CRD volume [3]. ConGF has shown modulatory activity in acuteinflammation and antinociceptive effects [14,15]. However, the ef-fect of ConGF in the vascular smooth muscle has not been de-scribed yet.

In this work, we report the full sequencing of ConGF by massspectrometry. The 3-dimensional structure is solved by X-raycrystallography, and lectin co-crystallization is solved withX-Man (5-bromo-6-chloro-3-indolyl-a-D-mannopyranoside) inorder to analyze lectin-carbohydrate interaction and to stabilizeflexible loops in the carbohydrate recognition domain (CRD). Therelaxation activity of ConGF in rat aortic rings was additionallyperformed to analyze the structure and activity relationship.

Materials and methods

Protein purification, crystallization and data collection

ConGF was purified from powdered Canavalia grandiflora seedsby affinity chromatography using a Sephadex G-50 [16]. Thefreeze-dried, purified lectin was resuspended in Milli-Q™ contain-ing 3 mM X-Man (5-bromo-4-chloro-3-indolyl-a-D-mannose) at afinal protein concentration of 12 mg mL–1 and incubated at 310 Kfor 1 h prior to the crystallization experiments. As described by Si-mões and coworkers [15], ConGF crystals were obtained usingCrystal Screen II (Hampton Research, USA) reagent N� 34 (0.05 Mcadmium sulfate hydrate, 0.1 M HEPES buffer at pH 7.5, and1.0 M sodium acetate trihydrate) grown in 24-well Linbro platesat room temperature (293 K) by the vapor-diffusion method [17]

in hanging drops. Suitable crystals for X-ray diffraction grew in1 week to maximum dimensions of approximately0.1 � 0.3 � 0.1 mm. X-ray diffraction data were collected from asingle crystal cooled to 100 K with a cold nitrogen-gas stream.Crystals were previously soaked in a cryoprotectant aqueous solu-tion of glycerol (30%) to avoid ice formation. The X-ray diffractiondata were collected at a wavelength of 1.47 Å using a synchrotron-radiation source (MX1 station in the National Laboratory ofSynchrotron Light (LNLS), Campinas, Brazil), using a MAR CCD-detector of 165 mm (MAR Research) in 180 frames with an oscilla-tion range of 1�.

X-ray diffraction data analysis

The dataset was indexed and integrated, and the intensitieswere reduced through the XDS package [18]. To solve the phaseproblem, we performed molecular replacement using maximum-likelihood methods implemented in the PHASER program [19].The atomic coordinates used as a search model were those of C.gladiata lectin (CGL) complexed with dimannoside man1-man-OMe (PDB code 2OVU) [20]. The log-likelihood-gain (LLG) solutionobtained was 1378.09, and the Rotation Function Z-score (RFZ) andTranslation Function Z-score (TFZ) were respectively 18.0 and 32.1.Crystallographic refinement was carried out in cycles of maxi-mum-likelihood refinement using Refmac 5 [21].

ConGF models were submitted to a few cycles of rigid bodyrefinement to verify the relative position of the rigid groups, fol-lowed by restrained refinement and correction and/or substitutionof amino acid side chains using 2Fo-Fc. The electron density mapswere generated and visualized using Coot. Point mutation in thelectin sequence, followed by addition of water molecules to the re-fined ConGF structure, was performed using Coot, and all theinspection was done with different Fourier maps and stereochem-ical criteria. Finally, the structures were submitted to restrainedisotopic refinement, and the quality of the ConGF model was as-sessed with the Procheck software [22]. The final ConGF modelwas visualized with Coot [23] and PyMOL [24]. The atomic coordi-nates of ConGF complexed with X-man were deposited in the Pro-tein Data Bank labeled as 4L8Q.

Molecular mass and sequence determination by mass spectrometry

The isotopic average molecular mass of ConGF was determinedby electrospray ionization using a hybrid mass spectrometer (Syn-apt HDMS System, Waters Corp., Milford, MA, USA). Protein sus-pension (10 qmol/lL) was infused into the system at a flow rateof 10 lL/min. The capillary voltage and the cone voltage were setat 3 kV and 40 V, respectively. The source temperature was main-tained at 90 �C, and nitrogen was used as a drying gas (flow rate:150 L/h). Data were acquired with the Mass Lynx 4.1 software.Multiply charged spectra were deconvoluted using maximum en-tropy techniques [25].

For primary structure determination, ConGF was subjected toSDS–PAGE, and the bands were excised and bleached with50 mM ammonium bicarbonate in 50% acetonitrile. Bands weredehydrated in 100% acetonitrile and dried in a Speedvac (Lab-Conco). For proteolytic cleavage, performed at 37 �C overnight, gelswere rehydrated with the following enzyme solutions: 50 mMammonium bicarbonate containing trypsin (Promega), chymotryp-sin (Sigma–Aldrich) and endoproteinase Asp-N (Roche). Theobtained peptides were extracted in a solution of 50% acetonitrilewith 5% formic acid and concentrated in Speedvac [26]. ConGFwas also subjected to chemical digestion with formic acid usingconditions described by Hua and coworkers [27]. The peptideswere injected into a nanoAcquity system connected to the electro-spray source of a mass spectrometer (Synapt HDMS System,

I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39 33

Waters Corp., Milford, MA, USA). The sample was applied to a re-verse phase column (75 lm � 100 mm) and eluted with an aceto-nitrile gradient of 10% to 85% containing 0.1% formic acid. The massspectrometer was operated in positive mode, using a source tem-perature of 90 �C and capillary voltage of 3.0 kV. The instrumentwas calibrated with double protonated ion of [Glu 1]-Fibrinopep-tide B (m/z 785.84). The LC-MS/MS experiment was performedaccording to the Data Dependent Acquisition (DDA) function,selecting the MS/MS doubly or triply charged precursor ions, whichwere fragmented by collision-induced dissociation (CID) using aramp collision energy that varied according to the charge state ofthe selected precursor ion. The data were processed and analyzedusing Proteinlynx (Waters), and the search parameter was the frag-mentation pattern of the peptides. The CID spectra were inter-preted manually using the Peptide Sequencing tool from MassLynx 4.0 software (Waters). Primary sequence alignments weremade with the ESPript 2.2 software [28].

Mechanical activity of isolated aorta

Male Wistar rats (250–300 g) were raised and housed (n = 6per cage) in rooms with a controlled 12/12 h light/dark cycle at25 �C with food and water ad libitum. Experimental proto-cols were approved by the Institutional Animal Care and UseCommittee of the Universidade Estadual do Ceará (CEUAn�10130208-8/40).

Thoracic aorta was quickly removed and cleaned, and ring seg-ments (3–5 mm) were mounted for tension recording (2 g) in 10-ml organ baths filled with modified Tyrode solution (136 NaCl, 5KCl, 0.98 MgCl2, 2 CaCl2, 0.36 NaH2PO4, 11.9 NaHCO3) and5.5 mM glucose, pH 7.4 (at 37 �C, 95% O2 and 5% CO2). In all exper-iments, the aorta was challenged with KCl (60 mM) after 45 min ofequilibrium to assure tissue viability. The contractile response wasmeasured using a force transducer linked to a preamplifier andcomputerized data acquisition system (DATAQ).

To evaluate the lectin relaxant effect, cumulative concentrationcurves of ConGF (1–100 lg/ml) were performed at the contractionplateau induced by phenylephrine (Phe; 0.1 lM) in either intactendothelium or denuded. Control aorta received the same volumeof Tyrode. Remotion of endothelium was assessed by mechanicalrubbing of the aorta intimal surface. Intact endothelium was con-sidered for relaxant responses to acetylcholine (ACh; 1 lM) greaterthan 75% of that induced by Phe [29]. To investigate the effect ofNO on lectin-induced relaxation, an inhibitor of nitric oxide syn-thase (NOS), N-nitro-L-arginine methyl ester (L-NAME; 100 lM),was added to tissues with intact endothelium 30 min before induc-tion with Phe. Results were expressed as mean ± S.E.M. of 5–7 ani-mals per group. Differences (P < 0.05) were analyzed using ANOVAand Student’s t-test.

Structural analysis

Primary structural analysis was accomplished using theESPript 2.2 software [28]. The 3-dimensional structures of theDiocleinae lectins (ConBr-3JU9, ConM-2CWM, ConA-1GKB andCGL-1WUV) and ConGF were analyzed, and both Coot [23] andPyMOL software [24] were used to calculate the distance of keyresidues. CRD volume was calculated by the Q-SiteFinder pro-gram [30]. Van der Waals contacts, polar contacts and hydropho-bic interaction were analyzed with the CCP4 software CONTACTimplemented in CCP4 [31], adopting the cutoff distances 3.5, 3.5and 5.0 Å, respectively. The analysis of significant distances ofatoms was performed using an error of atomic positioningcalculated by Luzatti Plot [32].

Results and discussion

ConGF primary structure

The complete sequence of ConGF lectin was reported in Fig. 1.The protein sequence data reported in this paper will appearin the UniProt Knowledgebase under the accession numberC0HJC6. The region that corresponds positions 31–39 was identi-fied by the peptide A2MS that comprises residues 28–43 and hasa theoretical mass of 1945.09 Da. Although MS/MS data are notavailable, crystallographic data of electron density clearly showthe Ile37 amino acid. We also tried to ionize A2MS by MALDI, butthe relative intensity was too low to perform a MS/MS experiment.The mass spectrometry data, together with the crystallographicdata, allowed us to perform the complete sequencing of ConGF.Table 1 shows all sequenced peptides used to complete ConGFprimary structure.

The isotopic average molecular mass determined for ConGF achain was 25,606 ± 2 Da, and for b and c, they were 12,961 ± 2and 12,663 ± 2 Da, respectively. Theoretical mass from the deter-mined sequence was 25,606, 12,961 and 12,664 Da, respectively,for a, b and c chains. The results obtained reveal the consistencybetween the experimentally determined mass and the sequenceinferred from it.

The polypeptide chain of ConGF is composed of 237 amino acidresidues and reveals four unique modifications (Ile37, Ala48, Thr164

and Thr220) when compared with ConA-like lectins (Fig. 2). Alllectins from the Canavalia genus possess a Val in position 32,except for ConGF which, similar to the lectins from the Diocleagenus, possesses an Ile in position 32. This also occurs in positions96, 121, 150 and 226, which are Ser, His, Thr and Thr, respectively,in the Canavalia genus and Thr, Ala, Ser and Gly, respectively, inthe Dioclea genus and ConGF. It should be noted that ConGFpresents some amino acid residues commonly found only inDioclea lectins.

The sequence identity score of ConGF varies between 81.62 and90.72 when compared with other ConA-like proteins shown inFig. 2. The lectin most similar to ConGF is CGL (91.56), followedby ConM, ConA (both 90.72), and ConBR (90.3). All identity scoresare available in the legend to Fig. 2. Simões et al. (2012) showedthat ConGF is positioned between the Canavalia and Dioclea/Craty-lia groups, as an intermediate based on phylogenetic analysis,using the alignment of partial amino acid sequence of ConGF withCanavalia, Dioclea and Cratylia lectins as synapomorphy [15].

Overall structure of ConGF

The diffraction data showed that ConGF crystals belong to theorthorhombic I222 space group with cell parameters ata = 63.42 Å, b = 65.15 Å, and c = 106.71 Å. Data processing, scaling,and refinement statistics are presented in Table 2. Molecularreplacement was performed using C. gladiata (CGL) crystal struc-ture coordinates (PDB 2OVU) as the search model. Matthewscoefficient was calculated as 2.05 Å3 Da�1 based on a molecularweight of 25.5 kDa, indicating the presence of a monomer in theasymmetric unit and a solvent content of 39.54% [33].

The overall structure of ConGF complexed with X-Man(Fig. 3A) has been refined to a resolution of 2.3 Å. The refinedstructure of ConGF presents 88 water molecules, one calciumion and one manganese ion per monomer. An X-Man moleculewas modeled into the carbohydrate recognition domain (CRD)(Fig. 3). We also found a sulfate ion, two cadmium ions and threeglycerol molecules incorporated in the ConGF structure, allaccording to the Fo-Fc electron density map. After the last refine-ment of the structure, Rfactor was 0.22 and Rfree was 0.26. The

Fig. 1. Complete sequence of ConGF. The peptides starting with T originated from a Trypsin digestion, with C from a Chymotrypsin digestion, with A from an AspNendoproteinase digestion and with an F from a chemical formic acid digestion.

Table 1Peptides derived from de novo ConGF sequencing.

Peptide Experimental mass Theoretical mass Difference mass Sequence

T1 3268.6265 3268.6138 �0.01 ADTIVAVELDTYPNTDIGDPNYPHIGIDIKT2 893.4940 893.3701 �0.12 WNMQDGKT3 1371.7266 1371.7510 0.02 VATAHIIYNSVGKT4 3293.5566 3293.6091 0.05 LSAVVSYPNADSATVSYDVDLDNVLPEWVRT5 1108.5844 1108.6128 0.03 VGLSATTGLYKT6 1512.6273 1512.7460 0.12 ETNTILSWSFTSKT7 2157.0786 2157.0786 0.07 SNSTAETNALHFTFNQFTKT8 2145.0518 2145.0549 0.00 DLILQGDATTDSDGNLQLTRT9 1359.7244 1359.6378 �0.09 VSSDGTPQGNSVGRT10 2827.2444 2827.2607 0.02 SPDSDPADGITFFLSNMDSTLPSGSGGRT11 958.4844 958.5123 0.03 LLGLFPDANC1 1314.6626 1314.6931 0.03 QDGKVATAHIIYC2 1629.7576 1629.7522 �0.01 SAVVSYPNADSATVSYC3 1837.9664 1837.9574 �0.01 DVDLDNVLPEWVRVGLC4 1090.4891 1090.5659 0.08 KETNTILSWC5 1418.6790 1418.6445 0.03 KSNSTAETNALHFC6 4806.3687 4806.3809 0.01 NQFTKDQKDLILQGDATTDSDGNLQLTRVSSNGTPQGNSVGRALFC7 1598.7843 1598.8092 0.02 APVHIWESSAVVASFC8 1461.6379 1461.6987 0.06 IKSPDSDPADGITFC9 1850.7644 1850.9197 0.16 FISNMDSTLPSGSGGRLLC10 732.2814 732.3442 0.06 GLFPDANA1 1024.4844 1024.4978 0.01 DPNYPHIGIA2 1945.0771 1945.0931 0.02 DIKSIRSKKIAKWNMQA3 1696.7354 1696.8209 0.07 DSTIPSGSGGRLLGLFPDANA4 1390.6488 1390.6688 0.02 DSDGNLQLTRVSSA5 1972.9590 1972.9854 0.03 DSTIPSGSGGRLLGLFPDANF1 1367,7072 1367.7561 0.05 NVLPEWVRVGLSF2 1189,6044 1189.6099 0.00 GNLQLTRVSSDF3 1155,7644 1155.5996 0.16 GTPQGNSVGRALF4 1238,6244 1238.6547 0.03 ATFTFLIKSPDF5 1143,5040 1143.6248 0.12 TSIPSGSGGRLL

34 I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39

model presents acceptable stoichiometry based on Ramachan-dran plot and a well-defined geometric structure. Calculated rootmean square deviation (RMSD) shows bond length deviations of0.021 and bond angle deviations of 2.110. The main deviationsare in the loops, including the loop of the carbohydrate-bindingdomain.

Some Canavalia and Dioclea lectins do not display electron den-sity in unstable regions of the structure, usually encompassingloops 117-123 and 149-151. In ConGF, only loop 117-123 is unsta-ble, and coordinates of residues (S117N118S119T120) were not mod-eled. Therefore, this region was cut in the final structure. Allother regions were well defined and consistent with the aminoacid sequences.

ConGF binding sites

ConGF structure shows a metal-binding site containing con-served residues [10]. The structures of the Mn2+- and Ca2+-bindingsites showed similarities with those of other legume lectins(Fig. 3A). The metal-binding site present in the monomeric struc-ture is located in the vicinity of the saccharide-binding site. Themanganese binding also establishes some coordinated interactionsto stabilize the carbohydrate-binding site. Four amino acids,mostly through their side chains, and two water molecules coordi-nate each metal ion. Mn2+ is coordinated by Glu8, Asp10, Asp19 andHis24; Ca2+ is coordinated by Asp10, Tyr12, Asn14 and Asp19. In addi-tion, the water molecules indirectly connect Ile32 and Ser34 to

Fig. 2. Multiple alignment of ConGF with other ConA-like seed lectins. ConA (C. ensiformis; Uniprot accession: P02866; Identity Score: 90.72), ConBR (C. brasiliensis; Uniprot accession:P55915; Identity Score: 90.3), ConM (C. maritima; Uniprot accession: P81364; Identity Score: 90.72), CGL (C. gladiata; Uniprot accession: P14894; Identity Score: 91.52), DWL (Diocleawilsonii; Uniprot accession: P86624; Identity Score: 83.12), DVL (D. violacea; PDB accession: 2GDF; Identity Score: 83.54), rDGL (recombinant lectin from D. grandiflora; PDB accession:2JE9; Identity Score: 83.97), DSL (Dioclea sclerocarpa; Uniprot accession: B3EWJ2; Identity Score: 83.97), Dgui (Dioclea guianensis, PDB accession: 1H9W; Identity Score: 84.81), Dvir(Dioclea virgata; Uniprot accession: P58907; Identity Score: 85.65), DRL (D. rostrata; PDB accession: 2ZBJ; Identity Score: 85.65), Cmollis (Cratylia mollis; Uniprot accession: P83721;Identity Score: 81.62); CFL (Cratylia floribunda; Uniprot accession: P81517; Identity Score: 82.2), CPL (C. pedicellatum; Uniprot accession: J9PBR3; Identity Score: 83.05).

I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39 35

Table 2Data collection, refinement and structure quality.

Parameter Value

Data collectionSpace group I222Unit cell parameters (Å)a 63.42b 65.15c 106.71Total reflections 71,496Number of unique reflections 10,110Molecules per asymmetric unit 1Resolution limits (Å) 10.29–2.3Rmerge

a (%) 10.3 (29.8)d

Completeness 99.4 (100)d

Multiplicity 7.07(I)/r 5.16 (10.00)d

Molecular replacementLLG 1,378.09RFZ 18.2TFZ 32.1RefinementResolution range (Å) 19.17–2.30Rfactor

b (%) 22.41Rfree

c (%) 26.40Number of residues in asymmetric unit 237Number of water molecules 88RMS deviations from ideal valuesBond lengths (Å) 0.021Bond angles (degrees) 2.110Luzatti plot 0.301Temperature factorAverage B value for whole protein chain (Å2) 32.43Ramachandran plotResidues in most favored regions (%) 96.07Residues in additional allowed regions (%) 3.93Residues in generously allowed regions (%) 0

a Rmerge ¼P

hkl

PijIðhklÞ�ðIðhklÞiÞjP

hkl

PiðIðhklÞiÞ where I(hkl)i is the intensity of ith measurement of the

reflection h, and I(hkl) is the mean value of the I(hkl)i for all I measurements.b Rfactor ¼

PhjjFobs j�jFcalc jjP

hjFobs j

.c Calculated with 5% of the reflections omitted from refinement.d Values in parentheses represent the high resolution shell.

Fig. 3. Overall structure of ConGF. (A) Tetrameric assembly of ConGF chains are in cmanganese ions (in purple), and cadmium ions (in gold). Other molecular ligands are inatoms in red. Carbons of three glycerol molecules are colored in yellow, and carbons fromcentral cavity are represented in (B) and (C), respectively. Residues in contact are highlighreferences to color in this figure legend, the reader is referred to the web version of thi

36 I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39

manganese and the residues Asp208 and Arg228 to calcium. Thepeptide bond of Ala207 and Asp208 in the cis configuration is isom-erized by the presence of divalent metals changing the side chainorientation and stabilizing the unusual Ala207-Asp208 cis-peptidebond, in this case by a bridge formed among a calcium ion, a watermolecule and the main-chain carbonyl group of Asp208.

Legume lectins can bear ligands different from carbohydrates.The literature shows the presence of binding sites with hydropho-bic characteristics, and the association of molecules, such as hor-mones, nucleic acids and non-natural amino acids [10,34,35].Binding sites can be located at the surface of lectins, e.g., theCRD, as well as in cavities, which are sometimes formed by dimericand the tetrameric association [10]. The difference electron densitymap exhibited some features through ConGF structure. Thesecould be explained by the presence of crystallized ions and glycerolmolecules used as cryoprotectant for X-ray data acquisition. Twocadmium ions, one sulfate ion and three glycerol molecules weresolved in ConGF structure (Fig. 3A). In the tetrameric arrangementof ConGF, the central cavity possesses sulfate ions facing each otherat opposite monomers. These ions are coordinated by residuesHis127 and Thr129, two water molecules, and an additional polarcontact between each sulfate (Fig. 3B). In addition, residues pre-sented in the central cavity also coordinate one glycerol molecule(Fig. 3C). Lys114, Lys116 and Ser190 of one monomer interact withglycerol, which, in turn, interacts with Asp192, Ala193 and Thr194

from opposite monomers. Ligands that perform bridges betweenfacing monomers in tetrameric lectins are able to stabilize theoligomer.

In the ConGF structure, CRD is occupied by X-man that was fit tothe lectin using the 2Fo-Fc electron density map (Fig. 4A). Themannose residue of X-man was driven to the site and stabilizedby a network of twelve H bonds connecting residues Asn14, Gly98,Leu99, Tyr100, Asp208 and Arg228 to the mannoside oxygen atomsO-3, O-4, O-5 and O-6 (Table 3). The indolyl group of X-man is sta-bilized by interaction with the hydroxyl group of Tyr12 and severalvan der Waals interactions with the hydrophobic subsite, involvingresidues Tyr12, Leu99 and Tyr100 (Fig. 4B).

artoon representation (colored gray); spheres represent calcium ions (in green),stick representation. Sulfate in the center shows sulfur atoms in orange and oxygenX-man are colored in cyan. Details of sulfate- and glycerol-binding molecules in theted, and orange dashed lines represent the polar contacts. (For interpretation of the

s article.)

Fig. 4. Carbohydrate-binding site of ConGF interacting with X-Man. (A) Omit map representation of the electron density contoured at 1r of X-Man. Orange dashes representpolar contacts. Amino acids that compose the CRD are labeled. (B) LIGPLOT representation of the hydrogen bonds and van der Waals contact around the X-man. (Forinterpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 3The van der Waals interaction and hydrogen bonds between ConGF and X-Man.

Amino acid X-Man Distances (Å)

van der Waals interactionsTyr12 OH C11 2.9Asn14 CG O4 3.5Gly98 CA O6 3.1Gly98 C O6 3.4Leu99 CB O5 3.2Ala207 CB O6 3.0Ala207 CB C6 3.4Asp208 CG O6 3.2Asp208 OD2 C4 3.4Asp208 OD2 C6 3.4Arg228 CA O4 3.4Arg228 CB O4 3.2Arg228 CG O4 3.3Hydrogen bondsTyr12 OH N1 3.3Asn14 ND2 O4 2.4Gly98 N O6 3.3Leu99 N O5 3.0Leu99 N O6 3.0Tyr100 N O6 3.0Tyr100 O O6 3.4Asp208 OD1 06 2.9Asp208 OD1 04 3.4Asp208 OD2 04 2.5Arg228 N 03 2.9Arg228 N 04 3.2

I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39 37

Vasorelaxant effect of ConGF

Phenylephrine induced tonic contractions in rat aortas withamplitude of 0.48 ± 0.03 g in the absence (n = 10; Fig. 5A) and0.37 ± 0.04 g in the presence of endothelium (n = 12; Fig. 5B). ConGFrelaxed precontracted endothelialized aorta at 30 and 100 lg mL�1

(IC50 = 34.48 ± 5.07 lg mL�1) by about 25% (Fig. 5B and C). However,the lectin relaxant effect was not observed in denuded endothelium(Fig. 5A and C). Moreover, ConGF did not alter tissue responsiveness

since, at the end of each experiment, KCl-contractile response wassimilar to the initial one (Fig. 5 A and B).

The involvement of NO in vasorelaxant responses has been set-tled for several lectins of the Diocleinae subtribe, such as ConBr,ConM, ConA, DVL and DvirL [6,7,9,13]. The endothelium depen-dency for the relaxant effect of ConGF suggests the involvementof NO. In aortas previously incubated with L-NAME, Phe inducedcontractions with amplitude of 0.73 ± 0.07 (n = 5). L-NAME blockedthe relaxant effect of ConGF at all concentrations (Fig. 5D), indicat-ing a potential role of NO as a mediator of lectin vasorelaxation.

Compared to other Canavalia lectins, ConGF exhibits the lowestvasorelaxant effect at 100 ug/mL (25%), while ConBr, CGL, ConMand ConA provoke, respectively, 74%, 108%, 110% and 85% of relax-ation in rat aorta segments. It is already well established thathighly homologous legume lectins display differences in biologicalactivities [1–7]; surprisingly, ConGF shows values much lowerthan similar proteins acting in vasorelaxation. We hypothesize thatthe configuration of CRD in ConGF, that accounts for this differ-ence, since this domain is the responsible for protein-cell adhesionand production of NO.

CRD structure and biological activity

Carbohydrate-binding domains in legume lectins are composedof adjacent loops stabilized by weak interactions, which are, inturn, stabilized between themselves and the metal-binding site[36,37]. The main chain residues positioned in those loop regionsare generally very similar among Canavalia lectins, with high loopflexibility occurring in side chains. Most Diocleinae lectins exhibitsimilar biological properties, but with a high degree of differencein their intensities [1]. Even so, differences between the distancesof residue atoms that comprise CRD are mostly small, but still pow-erful enough to promote differences in intensities of biologicalactivities [3].

CRD undergoes induced fit by carbohydrate binding, and here,the solved ConGF structure exhibits a CRD occupied by X-man.However, a few key distances between amino acid residues stillundergo changes, albeit insignificant, by the carbohydrate intake.Based on the structure alignment of uncomplexed and complexed

Fig. 5. Role of endothelium in the relaxant effect of ConGF on aorta precontracted with phenylephrine. Typical traces showing the effect of ConGF (1, 30 and 100 lg/mL) ondenuded (A) or endothelialized (B) aortic rings. Effect of lectin on Phe-induced contraction compared between denuded and endothelialized aorta (C) and its effect in theabsence or presence of L-NAME (100 lM) (D). Mean ± S.E.M.; ⁄p < 0.05 versus control (100% of Phe contraction); #p < 0.05 versus lectin alone. Phenylephrine (Phe; 0.1 lM);acethylcholine (ACh; 1 lM); Potassium chloride (KCl; 60 mM); ‘TN’ indicates washing of the preparation with Tyrode’s solution; Denuded aorta (E-); Endothelialized aorta(E+).

Table 4Distances (Å) between amino acid residues involved in carbohydrate recognition.

Residue pairs ConGF ConBr ConM ConA CGL

Tyr12 CZ-Gly227 CA 9.2 8.6 8.9 9.5 9.3Leu99 CG-Tyr12 CZ 7.2 6.8 7.6 7.4 8.1Leu99 CG-Asn14 ND2 9.5 9.1 9.0 9.3 8.6Leu99 CG-Arg228 CZ 12.3 11.8 11.1 11.5 9.9Tyr100 CZ-Tyr12 CZ 5.2 5.4 5.2 5.4 5.4Tyr100 CZ-Asn14 ND2 9.8 9.9 9.4 9.3 9.5Tyr100 CZ-Arg228 CZ 14.1 14.2 13.7 13.0 13.0

38 I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39

forms of lectins with high similarity, we found that the residue dis-tances of Tyr12-Gly227, Leu99-Asn14 and Tyr100-Arg228 pairs aremostly constant [3,9]. However, any changes in these specific ami-no acid residue distances among lectins from different species canlead to a change in NO synthase stimulation. More specifically, weare correlating NO as a mediator of such distances, and, as such,efficacy of vasorelaxation depends on NO induction. As a reference,

Table 4 lists key distances between atoms of CRD residues fromCanavalia lectins. Among these, ConM appears to induce the mostpotent vasorelaxation [9,13]. Accordingly, the CRD of ConM showssome structural characteristics, i.e., aa distances and site volume,that favor this induction, and, in this context, the distances thatseparate Tyr12CZ-Gly227CA, Leu99CG-Asn14ND2 and Tyr100CZ-Arg228CZ should be respectively smaller than 9.0 Å, smaller than9.5 Å and bigger than 12.0 Å to promote this activity [9,13].

Similarly, the distances exhibited by CRD of ConBr could alsolead to classifying this lectin as a highly efficient NO-dependentvasorelaxation inducer based on the distances reported to ConMand other Canavalia lectins previously characterized [9]. However,different from other lectins, ConBr presents such distances asTyr100CZ-Arg228CZ (14.2 Å) and Leu99CGTyr12CZ (6.8 Å) that arerespectively large and short enough to lead to a narrower CRDand, hence, a smaller volume of the binding site [3]. These modifi-cations have direct effects on biological activity and place ConBr inthe class of a vasorelaxation inducer of moderate efficacy. These re-sults show that the amino acid residue distances in CRD of legumelectins are determinants of NO production and, hence, the greater

I.L. Barroso-Neto et al. / Archives of Biochemistry and Biophysics 543 (2014) 31–39 39

or lesser induction of vasorelaxation. Based on these structuralcharacteristics, ConGF does not fulfill the requirements of a highlyefficient vasorelaxantion inducer. To explain, ConGF has two keydistances related to vasorelaxation activity (Tyr12CZ-Gly227CAand Leu99CG-Asn14ND2) set a few Ångströms away from themarked limits for a good inducer. These limits are Tyr12CZ-Gly227CA, Leu99CG-Asn14ND2 and Tyr100CZ-Arg228CZ shouldbe respectively smaller than 9.0 Å, smaller than 9.5 Å and biggerthan 12.0 Å. Accordingly, the distance of Tyr100CZ-Arg228CZ inConGF should be inserted within the limits for a potent inducersince it is larger than 12 Å (14.3 Å), but this distance fits in thesame range of ConBr and, along with the distance Leu99CG-Tyr12CZ(7.23 Å), a more narrow and deeper CRD arises with a volumesmaller than the best inducer, ConM (121 Å3 for ConGF and135 Å3 for ConM CRD). These outcomes reveal the importance ofkey distances and CRD volume in NO production induced by lec-tins, and they are considered determinants of the low capacity ofConGF to elicit vasorelaxation.

Conclusion

Mass spectrometry and X-ray crystallography give rise to pri-mary and three-dimensional structures of ConGF. This lectin wasable to induce relaxation of relatively low intensity in isolatedendothelialized rat aorta, via NO production, which seems to bedependent on key distances between residues comprising theConGF CRD and its reduced volume. In accordance with the limita-tions of key amino acid distances established in the present report,ConGF can be classified as a weak vasorelaxation inducer, in turnprovoking reduced CRD volume in a manner that justifies theobserved biological activity of ConGF in comparison to other lec-tins from the Canavalia genus, as strong or weak NO-dependentvasorelaxation inducers.

Acknowledgments

This work was supported by the Fundação Cearense de Apoio aoDesenvolvimento Científico e Tecnológico – FUNCAP, ConselhoNacional de Desenvolvimento Científico e Tecnológico – CNPq,Coordenação de Aperfeiçoamento de Pessoal de Nível Superior –CAPES, National Synchrotron Light Laboratory – LNLS, Brazil.C.S.N, P.L, A.M.S.A, K.S.N, B.A.M.R, A.H.S and B.S.C are senior inves-tigators of CNPq. We thank David Martin for helping with the Eng-lish language editing of the manuscript.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.abb.2013.12.006.

References

[1] B.S. Cavada, T. Barbosa, S. Arruda, T.B. Grangeiro, M. Barral-Netto, Curr. ProteinPept. Sci. 2 (2001) 123–135.

[2] J. Sanz-Aparício, J. Hermoso, T.B. Granjeiro, J.J. Calvete, B.S. Cavada, FEBS Lett.405 (1997) 114–118.

[3] E.H.S. Bezerra, B.A.M. Rocha, C.S. Nagano, G.D. Bezerra, T.R. Moura, M.J.B.Bezerra, R.G. Benevides, A.H. Sampaio, A.M.S. Assreuy, P. Delatorre, B.S. Cavada,Biochem. Biophys. Res. Commun. 408 (2011) 566–570.

[4] T.T.A. Cavalcante, B.A.M. Rocha, V.A. Carneiro, F.V.S. Arruda, A.S.F. Nascimento,N.C. Sá, K.S. Nascimento, B.S. Cavada, E.H. Teixeira, Molecules 16 (2011) 3530–3543.

[5] B.A.M. Rocha, P. Delatorre, T.M. Oliveira, R.G. Benevides, A.F. Pires, A.A. Sousa,L.A. Souza, A.M.S. Assreuy, H. Debray, W.F. Azevedo-Junior, A.H. Sampaio, B.S.Cavada, Biochimie 9 (2011) 806–816.

[6] R.B. Nóbrega, B.A.M. Rocha, C.A.A. Gadelha, T. Santi-Gadelha, A.F. Pires, A.M.S.Assreuy, K.S. Nascimento, C.S. Nagano, A.H. Sampaio, B.S. Cavada, P. Delatorre,Biochimie 94 (2012) 900–906.

[7] M.J.B. Bezerra, N.V.F.C. Rodrigues, A.F. Pires, G.A. Bezerra, C.B. Nobre, K.L.L.Alencar, P.M.G. Soares, K.S. Nascimento, C.S. Nagano, J.L. Martins, K. Gruber,A.H. Sampaio, P. Delatorre, B.A.M. Rocha, A.M.S. Assreuy, B.S. Cavada, Int.J. Biochem. Cell Biol. 45 (2013) 807–815.

[8] M.V. Ramos, R.A. Moreira, B.S. Cavada, J.T.A. Oliveira, P. Rouge, Rev. Bras. Fisiol.Veg. 3 (1996) 193–199.

[9] C.A. Gadelha, F.B.M.B. Moreno, T. Santi-Gadelha, J.B. Cajazeiras, B.A.M. Rocha,A.M.S. Assreuy, J. Struct. Biol. 152 (2005) 185–194.

[10] P. Delatorre, B.A.M. Rocha, E.P. Souza, T.M. Oliveira, G.A. Bezerra, F.B.M.B.Moreno, B.T. Freitas, T. Santi-Gadelha, A.H. Sampaio, W.F. Azevedo-Jr, B.S.Cavada, BMC Struct. Biol. 7 (2007) 52–61.

[11] N. Sharon, H. Lis, Adv. Exp. Med. Biol. 491 (2001) 1–16.[12] J.F. Kleha, P. Devesly, S.A. John, Br. J. Pharmacol. 104 (1991) 287–288.[13] A.M.S. Assreuy, S.R. Fontenele, A.F. Pires, D.C. Fernandes, N.V. Rodrigues, E.H.S.

Bezerra, T.R. Moura, K.S. Nascimento, B.S. Cavada, N.S. Arch, Pharmakology 380(2009) 509–521.

[14] B.S. Nunes, N.S. Rensonnet, B.S. Cavada, D. Dal-Secco, S.M. Vieira, E.H. Teixeira,T.R. Moura, C.S. Teixeira, J.T. Clemente-Napimoga, F.Q. Cunha, M.H. Napimoga,N.S. Arch, Pharmakology 379 (2009) 609–616.

[15] R.C. Simões, B.A.M. Rocha, M.J.B. Bezerra, I.L. Barroso-Neto, F.N. Pereira-Junior,R.M. Moura, K.S. Nascimento, C.S. Nagano, P. Delatorre, A.F. Pires, A.M.S.Assreuy, A.H. Sampaio, B.S. Cavada, Rapid Commun. Mass Spectrom. 26 (2012)811–818.

[16] V.M. Ceccatto, B.S. Cavada, E.P. Nunes, N.A. Nogueira, M.B. Grangeiro, F.B.Moreno, E.H. Teixeira, A.H. Sampaio, M.A. Alves, M.V. Ramos, J.J. Calvete, T.B.Grangeiro, Protein Pept. Lett. 9 (2002) 67–73.

[17] J. Jancarik, S.H. Kim, J. Appl. Crystallogr. 24 (1991) 409–411.[18] W. Kabsch, Acta Crystallogr. D 66 (2010) 125–132.[19] A.J. Mccoy, R.W. Grosse-Kunstleve, P.D. Adams, M.D. Winn, L.C. Storoni, R.J.

Read, J. Appl. Crystallogr. 40 (2007) 658–674.[20] G.A. Bezerra, T.M. Oliveira, F.B.M.B. Moreno, E.P. Souza, B.A.M. Rocha, R.G.

Benevides, P. Delatorre, W.F. Azevedo-Jr, B.S. Cavada, J. Struct. Biol. 160 (2007)168–176.

[21] G.N. Murshudov, P. Skubák, A.A. Lebedev, N.S. Pannu, R.A. Steiner, R.A. Nicholls,M.D. Winn, F. Long, A.A. Vagin, Acta Crystallogr. D 67 (2011) 355–367.

[22] R.A. Laskowski, M.W. MacArthur, D.S. Moss, J.M. Thornton, J. Appl. Crystallogr26 (1993) 283–291.

[23] P. Emsley, B. Lohkamp, W.G. Scott, K. Cowtan, Acta Crystallogr. D 66 (2010)486–501.

[24] W.L. Delano, DeLano Scientific, San Carlos, CA, 2002.[25] A.G. Ferrige, M.J. Seddon, B.N. Green, S.A. Jarvis, J. Skilling, Rapid Commun.

Mass Spectrom. 6 (1992) 707–711.[26] A. Shevchenko, H. Tomas, J. Havlis, J.V. Olsen, M. Mann, Nat. Protoc. 6 (2006)

2856–2860.[27] L. Hua, T.Y. Low, S.K. Sze, Proteomics 6 (2006) 586–591.[28] P. Gouet, X. Robert, E. Courcelle, Nucleic Acids Res. 31 (2003) 3320–3323.[29] R.F. Furchgott, J.V. Zawadzki, Nature 288 (1980) 373–376.[30] A.T. Laurie, R.M. Jackson, Bioinformatics 21 (2005) 1908–1916.[31] Collaborative Computational Project Number 4. Acta Crystallog. D 50 (1994)

760–763.[32] P.V. Luzatti, Acta Crystallogr. 5 (1952) 802–810.[33] B.W. Matthews, J. Mol. Biol. 33 (1968) 491–587.[34] D.D. Roberts, I.J. Goldstein, J. Biol. Chem. 258 (1983) 13820–13824.[35] A.V. Babosha, Biochemistry 73 (2008) 1007–1022.[36] R. Loris, T. Hamelryck, J. Bouckaert, Biochim. Biophys. Acta 1383 (1998) 9–36.[37] J. Bouckaert, R. Loris, F. Poortmans, L. Wyns, Proteins Struct. Funct. Genet. 23

(1995) 510–524.