Introduction

Lectins are carbohydrate-binding proteins of non-immune origin, which play an important role in biological recognition process. Many of them have been experimentally used to investigate the structure and function of carbohydrate chains present in glycoproteins and glycolipids and this property of the lectins has been utilized in various diagnostic tools. Lectins show ubiquitous occurrence in nature, their presence ranging from microorganisms to plants, insects and animals. More than 300 lectin structures (native and sugar bound form) have been reported in the 3D Lectin Database (http://www.cermav.cnrs.fr/lectins/) belonging to different fold families. Structural folds of lectins have been grouped into legume lectin fold, C-type lectin, I-type lectin, P-type lectin, β-prism I and II, β-trefoil, five-β-propeller fold, etc. (Vijayan and Chandra 1999; Wright 1997; Rini 1995). Many amongst these plant lectin folds contain β-strands in the core structure. The most widely observed is the legume lectin fold, which consists of a two-layered β-sheet structure forming the typical “jellyroll motif”. The overall subunit structure and the principles of sugar recognition are nearly identical within a particular lectin family. However, in case of legume and monocot mannose-binding lectin families variations have been observed in the mode of sugar recognition; further, new lectins showing specificity for complex sugars and glycoproteins have been reported, purified and characterized (Kolberg et al. 1983; Kamemura et al. 1993; Ray and Chatterjee 1995; Kaneda et al. 2002; Shangary et al. 1995; Wright et al. 1999; Singh et al. 2004; Guzmán-Partida et al. 2004; Singh Bains et al. 2005; Dhuna et al. 2005; Kaur et al. 2006a, b; Dhuna et al. 2007).

Amongst the plant lectins, the legume lectins showing specificity for simple sugars are well characterized and crystal structures of many of them are available. Similarly, biochemical and functional characterization of plant lectins showing specificity for complex sugars also have been characterized. The crystal structures in the latter category include those of a fetuin-binding lectin from Scilla campanulata (SCAfet), (PDB code 1DLP; Wright et al. 2000), two lectins from the seeds of legume tree Maackia amurensis (MAH and MAL) exhibiting specificity towards sialylated glycans (Imberty et al. 2000) and lectin IV from Griffonia simplicifolia (GSIV) in complex with methyl glycoside of the Lewis b human blood group determinant (Delbaere et al. 1993). X-ray crystallographic studies of Lathyrus ochrus isolectin (LOl I) in complex with monomer sugars provided insights into its fine monosaccharide specificity (Bourne and Cambillau 1990). Structural studies of lectins such as Cicer arietinum lectin (CAL), Moringa oleifera lectin (MOL), monocot lectins from Sauromatum guttatum, Arisaema tortuosum and Arisaema curvatum lectin (ACL) showing specificity for complex sugars were carried out in our laboratory to get a detailed insight into differential mode of sugar recognition by these lectins (Katre et al. 2005, 2008; Dharker et al. 2009; Urvashi et al. 2011). Cicer arietinum lectin (CAL), originally identified as a PA2 albumin, showed hemagglutination of enzymatically treated erythrocytes (Kolberg et al. 1983; Vioque et al. 1998; Katre et al. 2005; Wakankar et al. 2013) and unlike lectins commonly found in legumes the hemagglutination of this one was inhibited only by complex sugars and glycoproteins (Wakankar et al. 2013). In contrast to common legume lectins, CAL is neither glycosylated nor requires metal ions for agglutination. In this communication, we report for the first time the three-dimensional structure of a plant albumin presumably a lectin (CAL) owing to its hemagglutination activity which is inhibited only by complex sugars (Kolberg et al. 1983; Katre et al. 2005; Wakankar et al. 2013). Interestingly, unlike other known legume lectin structures, the CAL structure has resemblance to the hemopexin fold observed in plant albumins from Lathyrus sativus (LS24) and Cowpea (CP4) (Gaur et al. 2010, 2011).

Materials and methods

Protein purification

Mature seeds of chickpea (Cicer arietinum) cv PUSA-256 were procured from PUSA Agriculture University, Pusa, Bihar. Purification of CAL was carried out following the procedure previously described by Katre et al. (2005).

Hemagglutination assay

Rabbit RBCs were washed five times with PBS (phosphate buffer saline, 10 mM potassium phosphate buffer pH 7.2 containing 150 mM NaCl). A 3 % (v/v) suspension of the erythrocytes in the same buffer was prepared. Hemagglutination assays were performed as described elsewhere (Gurjar et al. 1998).

Chemical modification studies of CAL

Modification of the single tryptophan present in CAL was carried out by treating the protein (50 μg) with freshly prepared N-bromosuccinimide (1–5 mM) according to the method of Spande and Witkop (1967). The residual activity was determined by hemagglutination assay for modified protein separately.

Modification of histidine residues with diethyl pyrocarbonate (DEPC) was carried out according to the method of Ovadi et al. (1967). The reagent was prepared in absolute ethanol just prior to use. The lectin (50 μg) in 50 mM phosphate buffer, pH 7.2 was treated with 2–10 mM DEPC separately for 30 min and the residual hemagglutination activity was determined in each case.

Modification of carboxylate groups with Woodward’s reagent K (WRK) was carried out as described elsewhere (Sinha and Brewer 1985). CAL (200 μg) was incubated in 50 mM citrate–phosphate buffer pH 6.0, with different concentrations (5–20 mM) of WRK. Aliquots were removed after 15 min in each case and the residual activity was determined.

Arginine residues of CAL were modified according to the method of Takahashi (1968). The lectin (50 μg) incubated in 50 mM phosphate buffer pH 8.0 was treated with phenylglyoxal (prepared in methanol) in the range of 0.5–5 mM, for 30 min at 25 °C and the residual hemagglutination activity was determined for each aliquot. The methanol concentration in the reaction mixture did not exceed 2 % (v/v) and had no effect on the activity and stability of the lectin during the incubation period.

Modification of tyrosine residues of CAL was with N-acetyl imidazole. The reaction was carried out as described by Riordan et al. (1965). To 50 μg of CAL incubated in phosphate buffer pH 7.2, N-acetyl imidazole was added in the concentration of 2–10 mM separately and hemagglutination assay was performed for the modified protein.

Modification of serine with phenylmethylsulfonyl fluoride (PMSF) was by the method of Gold and Fahrney (1964). The lectin (200 μg) in 50 mM Tris–HCl buffer, pH 8.0 was incubated with 5–10 mM PMSF, at RT, for 60 min. Aliquots were removed after 15 min and residual activity was determined by hemagglutination assay.

Modification of lysine with Trinitrobenzenesulphonic acid (TNBS) was carried out as previously described (Habeeb 1966). The reagent was dissolved in 0.1 M Tris buffer, pH 8.0 to make a stock solution of 0.1 M. 50 μg of protein in 0.1 M Tris buffer, pH 8.0 was allowed to react for 1 h with TNBS up to final concentration of 2, 4, 5 and 10 mM and the residual activity was determined in each case.

The native lectin without any reagent was used as a control for hemagglutination activity in each case of chemical modification.

Crystallization

Purified CAL of concentration 20 mg/ml was used in crystallization trials using crystallization solution consisting of 0.2 M sodium acetate, 0.1 M sodium cacodylate pH 6.5 and 30 % w/v polyethylene glycol 8,000 (PEG 8K) based on the previous crystallization report, where both orthorhombic and trigonal crystals used to grow in same condition using protein purified from chickpea seeds of cultivar BDN 9-3 (Katre et al. 2005). Same trigonal crystals were obtained using protein preparation from cv PUSA-256 variety. The quality of crystals was further improved by varying PEG concentration from 12 to 22 %. The crystals grown in the presence of 18 % PEG 8K grew to full size in 2–3 days. l-Arginine at a concentration of 20 mM was added in the crystallization solution to prevent aggregation of protein molecules.

Data collection and processing

Diffraction data of CAL crystals were collected at room temperature using Cu-Kα radiation from a Rigaku rotating anode X-ray generator operated at 50 kV and 100 mA and an R-Axis IV++ image plate detector. The different steps of data collection, processing and storage of data were carried out using Crystal clear program supplied by Rigaku-MSC. The crystal to detector distance was kept at 150 mm and each frame was collected with an oscillation width of 0.5°. For room temperature, data collection exposure time given was 2–3 min, whereas for low temperature it was 4–8 min. Data processing was carried out using DENZO and XDisplayF and data were scaled using SCALEPACK of HKL suit of programs (Otwinowski and Minor 1997).

Structure determination, model building and refinement

The crystal structure determination of CAL was carried out using molecular replacement calculation provided in program PHASER of the CCP4 suit of programs (CCP4 1994). The crystal structure of the orthorhombic form was first determined using diffraction data reported by Katre (2007) and coordinates of a PA2 albumin from Lathyrus sativus (Gaur et al. 2010) which had 75 % sequence identity with CAL. The trigonal data were then phased using the refined coordinates of orthorhombic form. Refinement of structure carried out using restrained and TLS refinement implemented in the program REFMAC 5 (Murshudov et al. 1996; 1997). The program COOT (Emsley and Cowtan 2004) was used for displaying and examining the electron density maps, interactive fitting and optimizing the geometry and also for adding water molecules. The program PROCHECK was used for checking the stereochemistry and quality of the model (Laskowski et al. 1993). The interfaces between subunits of CAL were characterized using the MSDPISA server (Krissinel and Henrick 2007). Hydrogen bonds were calculated using the CONTACT program of CCP4 suite. The program PROFACE (Saha et al. 2006) was used to analyze oligomeric interface. All cartoons prepared were generated using PyMOL (DeLano 2002). The coordinates and structure factors of CAL structure have been deposited in the Protein Data Bank (http://www.rcsb.org/pdb) with accession number 3V6N and 3S18 for orthorhombic and trigonal crystal forms, respectively.

Results and discussion

Protein purification, crystallization and structure determination

Purified CAL showed a single band corresponding to the subunit molecular mass of 25 kDa on SDS-PAGE matching the calculated mass from sequence 25.6 kDa. The lectin preparation from cultivar PUSA-256 gave only trigonal crystals (with unit cell parameters a = b = 81.9, c = 69.5 Å and β = 120ο) grown in the presence of 20 mM l-Arginine and no crystals of orthorhombic form (P21212, a = 71.2, b = 73.3, c = 87.1 Å; diffraction data from Katre 2007) previously grown using CAL purified from seeds of C. arietinum cv. BDN 9-3 could be obtained. Highly redundant diffraction data of CAL crystals soaked in the presence of 50–0.5 M KI were collected in home source since iodine was known to bind to CAL (Table 1). Despite these attempts, no second heavy atom derivative of CAL could be obtained. Phasing attempts tried using SAD approach with the reported iodine derivative in orthorhombic form (diffraction data from Katre 2007) and available phasing programs such as PHENIX (Adams et al. 2010) and CRANK (CCP4 1994) remained unsuccessful. While struggling to get a second heavy atom derivative the full sequence information of CAL became available. A BLAST search in protein data bank using full length sequence of CAL identified an albumin from Lathyrus sativus (PDB-ID: 3LP9) related to CAL by a sequence identity of 75 % (Fig. 1a).

Table 1 Data collection and refinement statistics of CAL in orthorhombic and trigonal crystal forms. Values in parentheses are for the highest resolution shell
Full size table
Fig. 1
figure 1

a Sequence alignment of CAL with PA2 albumin from pea. Alignment of CAL sequence with that of structurally homologous plant PA2 albumin from grass pea (LS24, PDB-ID-3LP9) having 75 % sequence identity. Sequence identity was established using PDB-BLAST and sequences were aligned using (CLUSTAL W (1.81). The cartoon was rendered with ESPRIPT (3.0) (Gouet et al. 2003). b The four-bladed β-propeller fold of CAL monomer unit (orthorhombic form) viewed along the pseudo four-fold axis. The four blades are numbered as I, II, III and IV. The bound calcium, chloride, sodium and iodide ions are shown in blue, cyan, light pink and magenta spheres, respectively, and arranged in a queue in the central channel. Orange sphere represents the chloride ion, near blade II which is at the dimerization interface. Bound sulfate molecule is shown in stick between blade III and IV. The cartoon was prepared using PyMOL (Delano 2002)

Full size image

The three-dimensional structure of CAL was determined using molecular replacement calculation using PHASER module of the CCP4 suit (CCP4 1994) which gave positive solutions for dataset in orthorhombic crystal form (diffraction data from Katre 2007) and structure of PA2 albumin from Lathyrus sativus (LS24) (PDB-ID: 3LP9) used as model for phasing. Further, the refined coordinates of orthorhombic crystal form of CAL were used to determine structure in trigonal form. For the final refined model, the R-factors in the orthorhombic form were: R free 22.6 % and R work 18.2 % for data in the resolution range 20–2.2 Å. In case of trigonal form, the final R-factors for the refined model were 22.3 and 17.9 % using data in resolution range of 35–2.2 Å. The data collection and refinement statistics are summarized in Table 1. In the structure of both orthorhombic and trigonal crystal forms, extra electron density in Fo-Fc map was observed near Cys163, revealing the oxidation state of this residue, may be a result of exposure to X-rays and at room temperature. The Ramachandran plot showed that most of the residues were present in the favored or allowed regions of the plot in case of orthorhombic crystal form (Table 1). However, in case of trigonal crystal form 0.3 % of residues fall in the disallowed region of the Ramachandran plot. The torsional angles in the residue range 202–208 could not be regularized despite several attempts and refinement cycles owing to the poor electron density in these regions in both orthorhombic and trigonal crystal forms. These residues are part of the sulfate-binding loop, which might also be the reason for improper geometry and high B-factor.

Structural features of CAL

CAL mostly exists as a stable homodimer (MW 50 kDa) in solution under native conditions. Crystallographic assembly of CAL is also a dimer. The polypeptide chain in each subunit of CAL is organized in four β-sheets (I–IV), which are arranged almost radial around a central pseudo fourfold axis giving the appearance of a four-bladed propeller (Fig. 1b) also designated as hemopexin fold; characterized for the first time in hemopexin structure (Paoli et al. 1999). The pseudo fourfold axis of symmetry passing through the centre of the monomer unit forms a funnel-shaped channel of approximate dimension 20 × 8 Å (Fig. 1b). This channel provides ligand-binding sites for a calcium ion, chloride ion and a sodium ion exactly in a queued fashion one followed by the other. An iodine atom is bound near the bottom of the channel in case of the iodine derivatives of orthorhombic and trigonal crystal forms. A velcro closure brings together the C- and N-terminus as part of the blade-I which is supposed to play an important role in stabilizing the circular structure (Fülöp and Jones 1999); found to be a conserved feature in several proteins containing the β-propeller motif.

Each blade in the monomer unit comprises of four-stranded twisted β-sheet. The innermost β-strands, towards the centre of the channel are most regular whereas the outer ones are slightly deviating. The adjacent blades are connected by a linker region consisting of two short α-helices. A region consisting of α-helices that extends from the blade IV toward the N-terminus of the subunit, thereby maintaining cementing interactions between blade I and blade IV similar to those at the other three interfaces. The β-strands of the adjacent blades together form a β-sandwich structure enclosing a hydrophobic cavity predominated by Phe (10, 68, 76, 78, 124, 135, 177 and 188) and Tyr residues (5, 31 and 133). These hydrophobic residues enclose the central channel towards the outermost side while the innermost part of the channel is maintained hydrophilic by the presence of various charged residues namely, Asn7, Asp65, Asp69, Asp121, Asp174, Ser67, Arg11 and Arg125. The alignments of the four blades of CAL monomer carried out using CLUSTAL W server are shown in Fig. 2. Conserved Asp residues were detected in the alignment at two places which were part of the four different blades in loops (Asn7, Asp residues 65, 121 and 174, shown in red in Fig. 2) and conserved β-turns (Asp48, 106, 158 and 215, colored blue in Fig. 2), respectively, reported to be important for the stability of such folds (Paoli 2001).

Fig. 2
figure 2

Multiple sequence alignment of the amino acids present in four blades of CAL (CLUSTAL W (1.81). Conserved residues in the four blades are shown in red and blue color

Full size image

Ligand-binding site features of CAL

The identification and correct placement of metal ions has been done based on the electron density, B-factors and chemical environment surrounding them in both the orthorhombic and trigonal crystal forms. A Ca2+ ion is bound to the conserved residues Asn7 and Asp65, 121, 174 positioned in four different blades and thus plays a key role in providing structural integrity and stability to the β-propeller motif. Apart from Ca2+, other ions bound in the interior of the channel along the pseudo four-fold axis were iodide, sodium and chloride in the orthorhombic form (Fig. 3a). In case of the trigonal form, the electron density for sodium and chloride in the central channel is missing but an extra water molecule between Ca2+ and iodine atom could be located (Fig. 3b). The two subunits of the CAL homodimer showed differences in ligand binding with respect to each other in trigonal form, a NaCl molecule was bound near the Ca2+ binding site (Fig. 4a) where the Na ligated to OD1 of Asp65 which in turn bound to calcium through carbonyl oxygen (details not shown). In chain B, electron density for sodium and chloride was missing. Another Cl− is bound in the blade III of chain A, sandwiched between a beta strand and a loop. The occupancy of iodine atom was also poor in trigonal form (one-third) compared to the orthorhombic form. Interestingly, the CAL crystals grown/soaked in iodine salt diffracted better, probably due to the additional stability imparted by bound iodide which further stabilizes the β-propeller blades by ligating to four main chain peptide nitrogens of Arg11, Asp69, Arg125 and Ala178. In addition to these metal atoms, a clear electron density above 12σ was detected in the Fo-Fc map and assigned to a sulfate molecule in the loop regions connecting the β-blade III and IV. This sulfate might have got bound during the ammonium sulfate fractionation step of the purification. An additional chloride ion is present at the interface of the two subunits, probably playing role in stabilizing the homodimer in both orthorhombic and trigonal forms (Fig. 4a). These differences in ligand binding observed between the two crystal forms of CAL and between the two subunits of homodimer in the asymmetric unit may be attributed more to differences in molecular arrangement and conformation in crystals than assigning to CAL molecule itself and in turn may be influencing the iodine occupancy in two crystal forms.

Fig. 3
figure 3

Close up at the metal-binding site of CAL in the channel, shown fitted into 2Fo-Fc map. The Ca2+ ion is bound in all subunits and ligand to conserved residues. a Orthorhombic form has other ions such as iodide, sodium and chloride also bound apart from water molecules and b trigonal form has iodide bound but sodium and chloride are present only in one subunit and their position vary along with that of water molecules

Full size image
Fig. 4
figure 4

a Ligand-binding site in homodimer of trigonal form of CAL. The two subunits of CAL dimer in trigonal form differ with respect to the ligand-binding site in the central channel which occupies Ca2+, H2O molecule (instead of NaCl in orthorhombic form) and an iodide (shown in blue, light cyan and magenta spheres, respectively). Chain A has a bound NaCl near the Ca2+ binding site while no corresponding density for NaCl was seen in chain B. Also in blade III of chain A, an extra Cl− ion was bound (shown as orange sphere). b Calcium ion coordination and geometry (shown for orthorhombic crystal form), the residues ligating to the Ca2+ ion are all labeled and their respective coordination lengths in Å are also shown

Full size image

The kind of arrangement of calcium-binding site observed in the interior of the tunnel has also been found in other proteins containing four-bladed β-propeller motif such as, hemopexin (Paoli et al. 1999), human gelatinase A (MMP2) (Gohlke et al. 1996), porcine synovial collagenase (Li et al. 1995) and PA2 albumins from Lathyrus sativus and cowpea (Gaur et al. 2010, 2011). The proteins having β-propeller motif use their central tunnel or the entrance to that tunnel to coordinate a ligand or to carry out some catalytic function (Baker et al. 1997). However, unlike legume lectins, there is no evidence for involvement of metal–ligand in sugar binding of CAL, since hemagglutination activity of the lectin remains unaffected by treatment with chelating agents, such as EDTA. This indicates the essential requirement of Ca2+ in maintaining the structural integrity of the β- propeller fold in CAL. A comparison of Ca2+ binding site of CAL with those reported in other calcium-binding proteins was carried out (Pidcock and Geoffrey 2001; Deng et al. 2006). The Ca2+ binding site in CAL belongs to type III where coordination groups are supplied by Asn7, Asp65, 121 and 174 from four different blades of β-propeller fold. The geometry of Ca2+ binding sites in CAL is pentagonal bipyramidal type, with Ca2+ coordination number 7.0 in both the crystal forms (Fig. 4b).

Subunit association and interface analysis

The crystallographic asymmetric unit of CAL is a homodimer. The oligomerization has been analyzed in the solution state as well. Although CAL migrated as a single band corresponding to a subunit molecular weight ~25 kDa in SDS-PAGE, the gel filtration chromatography with sephacryl S-200 showed two peaks corresponding to the monomer and dimer (data not shown) predominated by the latter. Oligomeric state of CAL monitored at acidic pH 4.0 was further dominated by dimer. The analysis of the interface was carried out by submitting the structure to MSDPISA server which suggested the existence of a highly stable dimer in solution associated with it a high free energy value of interface 105.0 (orthorhombic) and 91.9 kcal/mol (trigonal). The H-bonds and salt bridges stabilizing the quaternary structure of CAL in orthorhombic and trigonal crystal forms were analyzed using CONTACT and these were similar except in few places. The various physicochemical parameters at the dimeric interface of CAL in orthorhombic and trigonal form were analyzed using PROFACE, summarized in Table 2. The analysis revealed that the total interacting surface area is same in both the forms (Table 2). Interestingly, the crystal packing of CAL shows a tetrameric association like that found in Concanavalin A (Edelman et al. 1972; Reeke et al. 1975). Thus, there might be possibility for the existence of a weak tetramer with poor interface interaction and stability. The oligomerization observed in the case of CAL makes it clearly distinct from the tetrameric legume lectins possessing specificity for mono/oligosaccaharides. A comparison between the two subunits of the dimer in the asymmetric unit gave rms deviation 0.47 and 0.35 Å for C-alpha atom positions in orthorhombic and trigonal forms, respectively. A similar comparison between the two subunits across the two crystal forms gave average rmsd 0.8 Å with significant deviations in blade I and IV. Large deviations and higher B-factors were observed for the loops comprising residues 33–37 in blade I and residues 201–208 in blade IV as well as residues 211–222 in C-terminal helix (Fig. 5).

Table 2 Analysis of the physiochemical parameters of dimerization interface in orthorhombic and trigonal crystal forms of CAL using program PROFACE
Full size table
Fig. 5
figure 5

Structural variation in the subunits of CAL in orthorhombic and trigonal crystal forms. Structural superposition of four independent subunits of CAL dimers on each other is shown for blade I and IV revealing large structural deviations in regions marked in the cartoon. The skeleton drawn in red and in blue is for chains A and B, respectively, for orthorhombic form, while the skeleton colored in green and orange shows chains A and B of trigonal form of CAL

Full size image

Identification of the putative sugar-binding site

Since CAL showed hemagglutination inhibition only with complex glycoproteins such as fetuin and desialylated fetuin and fibrinogen, the saccharide-binding specificity of the lectin could not be identified which imposed a major hurdle in complexing the lectin with mono/disaccharides to get a sugar bound structure for characterization of the binding site. Attempts to co-crystallize/soak CAL with commonly available mono/disaccharides were also not successful. As an alternative method, chemical modification studies on CAL were carried out to trace the sugar-binding site. Chemical modification studies on CAL were carried out to get an idea about the residues involved in sugar recognition. The amino acid modification was carried out according to the standard protocol as mentioned in the methods section and effect of modification at the sugar-binding site was estimated by hemagglutination assay. Two probable sugar-binding sites could be possible as the protein had two histidines, modification of which resulted in complete loss of hemagglutinating activity (Table 3). However, out of the four blades the lysine and carboxylate groups affecting the activity to a significant extent were located near His180 found in the loop regions of blade IV and I, respectively. Thus, it might be possible that the putative sugar-binding site involves the short loop of blade IV containing His180, Lys181 and carboxylate group of Asn15 and 16 which are the part of the loop from blade I (Fig. 6). As already mentioned, the loops in blade I and IV in the vicinity of above residues show higher B-factors indicating the flexibility of these regions (Fig. 5) favoring sugar binding. In blades II and III, all these residues are not present in the vicinity. The single tryptophan 217 in CAL is part of blade IV and found imbedded between the β-sheets thus, upon modification of this residue the hemagglutination of the lectin remained unaffected. Therefore, it seems that possibly there is only a single sugar-binding site per subunit of CAL dimer and the binding site is very shallow and towards the solvent exposed region on the surface of the protein (Fig. 7a). Since the sugar-binding site is away from the interface of the subunits in dimer and also the calculated solvent accessibility is high for residues N15, K181, H180, N16 in that order, the quaternary structure will not affect sugar binding (Fig. 7b). However, it may be premature to conclude about the details of sugar binding. More insight into the binding can be achieved only with a sugar bound structure of CAL.

Table 3 Summary of the results obtained upon chemical modification studies of CAL. The residues present within 5.0 Å distances of two histidine residues of CAL in the four blades are listed. The residues shown in bold and italics are those which on modification affected the hemagglutination activity of CAL while those in bold are found near the putative-binding site but have no effect on activity upon treatment with modifier
Full size table
Fig. 6
figure 6

Putative residues involved in sugar recognition by CAL inferred using chemical modification studies. The His180 and Lys181 are from blade IV while the carboxylate groups are contributed by Asn15 and 16 from blade I. The solvent accessibility in the decreasing order for these residues is N15, K181, H180 and N16. Also shown are the bound metal ions in the central channel. The cartoon was generated using PyMOL (Delano 2002)

Full size image
Fig. 7
figure 7

a Space filling model of CAL showing the surface representation of putative sugar-binding site. As can be seen from the structure the binding site is shallow, similar to that generally observed in lectins. b The disposition of the plausible sugar-binding site in each subunit of the dimer and its disposition with respect to interface of quaternary association. Drawing using orthorhombic structure and the trigonal structure is very similar

Full size image

Comparison of CAL structure with other β-propeller fold containing lectins

The three-dimensional structure of CAL consists of four-bladed β-propellers, which is unique amongst the known plant lectin folds. The lectins containing β-propeller motif are few. They have been reported from sources other than plants, such as bacteria (six-bladed Ralstonia solanacearum lectin, RSL (Kostla´nova et al. 2005), fungi [six-bladed Aleuria aurantia lectin, AAL (Wimmerova et al. 2003)] and eukaryotes [five-bladed Tachylectin-2 from horse shoe crab (Beisel et al. 1999)]. The specificity of both RSL and AAL is towards fucose, while that of tachylectin is for GlcNAc. In case of RSL and AAL, the conserved tryptophans were important for binding fucose (Kostla´nova´ et al. 2005), however, in case of CAL the tryptophan happened to be buried in the interior region and had no influence on hemagglutination. The structure of these lectins at the sugar-binding site was compared with individual blades of CAL but the results were ambiguous and no useful information could be obtained. Large deviations in C-alpha carbons were observed in all the cases with rmsd >3.0 Å. The β-propeller fold containing proteins are ubiquitous and they show diversity at functional, phylogenetic as well as at the level of their sequences (Paoli 2001) thus imposing difficulty in protein structure and function assessment by sequence comparisons. Sequence and structural comparison of CAL showed no similarity to fetuin-binding monocot family lectin SCAfet from the bulbs of Scilla campanulata (Wright et al. 2000) which exhibits β-prism II fold. Although CAL shows high hemagglutinating activity, the sequence and structure of CAL resembles more of PA2 albumin from legumes (Gaur et al. 2010, 2011) than of the known legume lectins.

Conclusions

The β-propeller fold structures show extreme diversity of function. Till date only three lectins in PDB are reported to have β-propeller architecture, they are from bacteria: six-bladed RSL (Kostla´nova et al. 2005), from fungi: six-bladed AAL (Wimmerova et al. 2003) and from eukaryotes: five-bladed Tachylectin-2 (Beisel et al. 1999). Recently, a protein with hemopexin fold from oyster mushroom (Pleurotus ostreatus), structurally resembling plant PA2 albumins from pea (LS-24) and chickpea (CAL) and exhibiting hemin-binding property, has been reported (Ota et al. 2013). CAL will be the first amongst the plant hemagglutinins comprising of a four-bladed β-propeller fold which resembles PA2 albumins from plants than to the known lectins. Unlike legume lectins, comprising of the typical “jellyroll motif” structure CAL has four-bladed β-propeller structure. The metal-binding site and oligomerization mode observed in CAL was also different from that observed in legume lectins. Thus, the differences in the structural organization, oligomerization pattern and metal ion requirement of CAL might be contributing towards complex sugar specificity of CAL, compared to other known simple sugar specific legumes lectins. Plants might have utilized the β-propeller fold for dual function: in sugar recognition and as albumins for carrying small ligands. Similar to other two PA2 albumins LS24 (Gaur et al. 2010) and CP4 (Gaur et al. 2011) there are evidences that CAL also binds hemin (Pedroche et al. 2005), however, during this study, despite several attempts at crystallization, no hemin bound structure could be obtained. The sequence and structural fold of CAL closely resembles pea hemopexin which has been well characterized for its physiological role in polyamine metabolism in plants and in regulation of abiotic stress (Vigeolas et al. 2008; Gaur et al. 2010; Gill and Tuteja 2010). PA2 albumins from chickpea and pea also bind to thiamine and thus play role in vitamin storage (Adamek-Swierczynska and Kozik, 2002). Functional characterization and homology modeling studies have recently been reported for rice plant (Oryza Sativa) hemopexin fold protein (OsHFP), structurally resembling to plant PA2 albumins from pea and chickpea with respect to the central four-bladed β-propeller architecture but not related to the flanked N- and C-terminal domains (Chattopadhyay et al. 2012). Apart from the hemin-binding property known for CAL, the rice plant OsHFP also plays role in chlorophyll degradation linked with programmed cell death in plants (PCD) (Chattopadhyay et al. 2012). The four-fold pseudo-symmetry offered by the β-propeller architecture in proteins provides multivalent-binding sites which is suitable for lectin function as well as could be important for binding hemin, plant hormones and hydrophobic groups, which might be playing role in plant metabolism (Barondes 1988).

Author contribution

US and UK crystallized CAL and collected data. US designed experiments and solved the crystal structure of CAL. CGS envisaged the problem, analyzed the results and wrote the manuscript with US. All the authors read and approved the manuscript.