The structural ground underlying the pH-dependency of the dimer–tetramer transition of Diocleinae lectins was investigated by equilibrium sedimentation and X-ray crystal structure determination of wild-type and site-directed mutants of recombinant lectins. Synthetic genes coding for the full-length α-chains of the seed lectins of Dioclea guianensis (termed r-αDguia) and Dioclea grandiflora (termed r-αDGL) were designed and expressed in Escherichia coli. This pioneering approach, which will be described in detail in the present paper, yielded recombinant lectins displaying carbohydrate-binding activity, dimer–tetramer equilibria and crystal structures indistinguishable from their natural homologues. Conversion of the pH-stable tetrameric r-αDGL into a structure exhibiting pH-dependent dimer–tetramer transition was accomplished through mutations that abolished the interdimeric interactions at the central cavity of the tetrameric lectins. Both the central and the peripheral interacting regions bear structural information for formation of the canonical legume lectin tetramer. We hypothesize that the strength of the ionic contacts at these sites may be modulated by the pH, leading to dissociation of those lectin structures that are not locked into a pH-stable tetramer through interdimeric contacts networking the central cavity loops.

INTRODUCTION

Lectins are a structurally heterogeneous group of proteins of non-immune origin possessing at least one non-catalytic domain that binds reversibly to a specific mono- or oligo-saccharide [15]. Lectins are ubiquitous in all kingdoms of life and play biological roles in many cellular processes, such as cell communication, host defence, fertilization and development, parasitic infection, tumour metastasis, plant defence against phytopredators and pathogens, etc. [3], by deciphering the glycocodes encoded in the structures of glycans attached to soluble and integral cell membrane glycoconjugates [4]. Mechanisms for sugar recognition have evolved independently in a small number of protein domains [1,5], and lectins sharing a common sugar-binding fold may depart in their carbohydrate-recognition specificities [6]. Multivalence is a mechanism frequently used by lectins for increasing their glycan-binding avidity and specificity [7,8]. In addition, most lectins are oligomeric or multidomain proteins, and variability in their quaternary associations represents a common evolutionary mechanism for creating structural and functional diversification among lectins exhibiting the same tertiary fold [5,913].

Although recent years have witnessed the elucidation of a large variety of novel structures of lectins [9,10 and http://www.cermav.cnrs.fr/lectines], plant lectins from the Leguminosae family remain the most thoroughly studied group of sugar-binding proteins. In particular, lectins isolated from legume seeds have been regarded as paradigmatic molecules for investigating the structural basis and thermodynamics of selective sugar recognition [58,1517]. Legume lectin subunits display a high degree of tertiary-fold conservation. The architecture of the so-called ‘legume lectin fold’ is structurally related to a jelly-roll motif characterized by a sandwich formed by a curved seven-stranded front β-sheet and a nearly flat six-stranded back β-sheet, bridged on one side by a top five-stranded β-sheet [5]. The tetrameric assembly of seed lectins from the Dioclea, Cratylia and Canavalia genera of the Diocleinae subtribe of the Papilionoideae subfamily of legume lectins involves a crosswise back-to-back association of two dimers built by the side-by-side, anti-parallel alignment of the subunit six-stranded back β-sheets [5], first characterized in the crystal structure of ConA (concanavalin A) [1820]. Diocleinae lectins exhibit pH-dependent dimer–tetramer equilibria [2124]. The physiological relevance of this phenomenon remains elusive. However, it is worth noting that the 25×8 Å2 (1Å=0.1 nm) water-filled central cavity of the tetramer of ConA and other leguminous lectins has been reported to bind various non-polar compounds, such as the plant auxin β-indoleacetic acid and cytokinins [25,26]. Auxins and cytokinins function as plant growth hormones and thus the lectins could use dimer–tetramer equilibrium as a mechanism for storage or transport of hormones.

Structural evidence gathered from comparison of the crystal structures of the pH-independent tetrameric DGL (Dioclea grandiflora lectin; PDB accession number 1DGL) [27] and those of Dguia (Dioclea guianensis lectin; PDB accession numbers 1H9P and 1H9W) [28] and CFL [Cratylia floribunda lectin; crystallized at acidic (PDB accession number 2D3R) and basic (PDB accession number 2D3P) pH] [29], which exhibit pH-dependent dimer–tetramer transition, revealed that interdimer contacts in the ‘canonical ConA tetramer’ are made between the homologous regions encompassing residues 53–78 of monomers A and D and B and C at the periphery of the dimers, and between the loop comprising residues 117–123 of each monomer located in the water-filled cavity at the centre of the tetramer. Interactions at the peripheral contact sites are essentially conserved in the structures of CFL crystallized in conditions where it exists as a dimer (pH 4.6, aCFL) or as a tetramer (pH 8.5, bCFL) in solution [29]. These interface regions of aCFL and bCFL can be superimposed with an rmsd (root mean square deviation) of 0.42 Å. Similarly, the peripheral interdimeric sites of Dguia (pH-dependent tetramer) and DGL (pH-independent tetramer) are essentially conserved [28]. The structural data also pointed to the participation of two histidine residues, His131 and His51, in stabilizing the conformation of the central cavity loops comprising residues 117–123 from each subunit. However, in contrast with loop ordering through a His51-dependent mechanism observed in CFL, which does not yield a pH-stable tetrameric association, His131 in DGL establishes a network of interdimer interactions bridging the four central loops into a pH-independent tetramer. Asn131 in CFL is unable to reproduce these intersubunit interactions. In addition, the number of interdimer contacts in the central cavity are drastically reduced in the aCFL compared with the bCFL structure [29]. The reduced number of interdimer contacts yielding a ‘weakened’ tetrameric association of aCFL is due to increased distances between atoms which are hydrogen-bonded in bCFL as a consequence of the distinct spatial orientation of residues His51, Ser66, Ser110, Thr117, Asn118, Ser119 and Thr120 in aCFL and bCFL [29].

The apparent paradox that a histidine residue (pKa 5.5–7.5) determines pH independence might be explained assuming that the network of intermonomer contacts centered at His131 cannot be disrupted by altering the pH value of the solution in the range 4.5–8.5. Current evidence suggests that although both the central and the peripheral interacting regions may bear information for formation of the canonical legume lectin tetramer, the absence of the stabilizing intermonomeric interactions (i.e. when an asparagine residue occupies position 131) allows pH-dependent destabilization of the tetrameric lectin structure through a hitherto elusive mechanism. This view reinforces our hypothesis of the key role of the side chain of the residue at position 131 for determining the pH-dependence or -independence of the tetrameric association of the seed lectins from Diocleinae.

To gain a deeper insight into the structural determinants underlying Diocleinae lectin pH-dependent quaternary association, in the present study we have designed and expressed in Escherichia coli synthetic genes coding for the full-length α-chains of Dguia and DGL. This pioneering approach, which will be described in detail, yielded recombinant proteins (r-αDguia and r-αDGL), which are biochemically indistinguishable from the naturally occurring lectins. Conversion by site-directed mutagenesis of residues at positions 123, 131 and 132 of the pH-independent tetrameric r-αDGL for the corresponding residues of Dguia endowed the triple mutant with pH-dependent dimer–tetramer equilibrium. The structural basis of this transition was investigated by solving the crystal structures of the wild-type (r-αDguia, r-αDGL) and mutated (r-αDguia S131H and r-αDGL E123A/H131N/K132/Q) recombinant lectins.

EXPERIMENTAL

PCR amplification of the genes and mRNAs coding for the precursors of Dguia and DGL

D. guianensis and D. grandiflora seeds were harvested from plants at the Federal University of Ceara campus, Fortaleza, Brazil. For germination (at the Instituto de Biomedicina de Valencia, Valencia, Spain), seeds were pretreated by immersion in concentrated sulfuric acid for 20 min followed by washing in running tap water for 5 min. The sulfuric-acid-treated seeds were placed between two filter paper sheets moistened with distilled water, rolled-up and kept in a water-saturated chamber in the dark. Genomic DNAs (referred to as gDNA) from D. guianensis and D. grandiflora were extracted from young leaves using the CTAB (cetyltrimethylammonium bromide) procedure [30]. For total RNA isolation, young buds in liquid nitrogen were immediately ground to a powder with a pestle.

Total cellular RNA from D. guianensis was isolated with Concert Plant RNA reagent (Invitrogen) according to the manufacturer's protocol. The first strand of cDNA of D. guianensis was reverse-transcribed using a standard method. Briefly, 12 μl of 0.1% DEPC (diethylpyrocarbonate)-treated water containing 0.1 μg of 5′-CCAGTGAGCAGAGTGACGAGGACTCGAGCTCAAGCT(T)16-3′ (Qt; Sigma–Genosys) and 1 μg of total RNA were heated to 70 °C for 10 min to denature any possible secondary RNA structure, cooled on ice, and mixed with 4 μl of 5× First Strand Buffer (Promega), 2 μl of 0.1 M DTT (dithiothreitol; Promega), 1 μl of the ribonuclease inhibitor RNAsin (40 units/μl) (Promega), 1 μl of dNTP (10 mM each) (Eppendorf) and 200 units of MMLV RT (Moloney Murine Leukaemia Virus Reverse Transcriptase) (RNase H Minus; Promega). The final volume was adjusted to 20 μl and the reaction mixture was incubated at 42 °C for 1 h, followed by 10 min at 50 °C to inactivate the enzyme. DNA was PCR-amplified using one-tenth of the RT reaction as a template. The forward primer, 5′-CCNGTSCACATTTGGGA-3′, and the reverse primer, 5′-WACYCTHACCCAYTCNGG-3′ (IUPAC codes are used for degeneracies) were designed according to the conserved amino acid sequences of legume lectins, PVHIWE primer sense and PEWVRV primer antisense respectively. The PCR amplification protocol included an initial denaturation step at 95 °C for 10 min, followed by 35 cycles of denaturation at 94 °C for 30 s, annealing at 50 °C for 30 s, and extension at 72 °C for 2 min. The recovered PCR products (500 bp) were cloned into a pGEM-T vector (Promega), and transformed into E. coli strain DH5α (Novagen). The white transformants were screened by PCR and the positive clones were confirmed by sequencing using an Applied Biosystems model 377 DNA sequencer. The DNA sequences obtained were used for designing specific oligonucleotides for 3′RACE (rapid amplification of cDNA ends) [31]. The reaction was carried out with the sense primer GSP (gene specific primer)-3′RACE (5′-TGGGAAGGTAGGAACTGCACACA -3′) and the antisense primer Qt-anchor (5′-CCAGTGAGCAGAGTGACG-3′). To determine the upstream sequence of the Dguia gene, 1 μg of gDNA was digested with 10 units of RsaI, ligated to an adaptor [32] with 10 units of T4-DNA ligase (Invitrogen) for 12 h at 13 °C and amplified by nested PCR using the sense primers IN (5′-AGGGCGTGGTGCGGAGGGCGGT-3′) or OU (5′-TGTAGCGTGAAGACGACAGAA-3′) and antisense specific primers GSP1-5′RACE (5′-AAGCTTGCGACCACAGCAGAT-3′) or GSP2-5′RACE (5′-CCACCAGAACCACTAGGGATTGAA-3′). Based on the amplified sequences, we designed the specific forward primer DguiaFW (5′- CCATGGATGGGTATTTCAAAAAAATC-3′, which includes a NcoI restriction site, underlined) and the reverse primer DguiaRV (5′-TCAGACGACGGATGCAAT-3′) for amplification of the complete open reading frame (using either cDNA or gDNA). For PCR amplification with specific primers, the reaction protocol was similar to the cDNA amplification with degenerate primers, but the annealing step was performed at 60 °C for 30 s. The amplified fragments were separated by 2% agarose gel electrophoresis, purified using the Perfect Pre Gel Clean Up kit (Eppendorf), and cloned into a pGEM-T vector.

For PCR-amplification of the DGL precursor, gDNA was used as the template with the forward primer 5′-CTGATAGTAGTGAGCAGGGTGAGC-3′, corresponding to the amino acid sequence L20IVVSRVS27 of the Dguia precursor protein (Figure 1), and the reverse primer 5′-ATTCGTCTTTAACTTAGAAG-3′ that corresponds to the last six amino acids of the β-chain (S275KLKTN280) (Figure 1). The reaction protocol was similar to the cDNA amplification with semi-degenerate primers. The PCR product obtained was cloned into the pGEM-T vector.

cDNA-amplified nucleotide sequences of the Dguia and DGL precursors

Figure 1
cDNA-amplified nucleotide sequences of the Dguia and DGL precursors

(A) Nucleotide and deduced amino acid sequence of the full-length open reading frame encoding the precursor of Dguia. The signal peptide sequence (residues 1–28) is underlined. The amino acid sequences of the γ- (29–147) and the β- (163–280) chains are highlighted. The eight codons coding for different amino acid residues in Dguia DNA compared with the seed-isolated lectin are highlighted in grey boxes. The degenerate nucleotide sequences of the forward (F) and reverse (R) primers, based on highly conserved amino acid sequences among legume lectins, used to amplify a specific 500 bp DNA product are shown. The nucleotide sequence (837–840) used to engineer an EcoRI restriction site is underlined. (B) Nucleotide and deduced amino acid sequence of the precursor of DGL. The forward (F) and reverse (R) primers used for amplifying the full-length DNA coding for the full-length α-chain are shown.

Figure 1
cDNA-amplified nucleotide sequences of the Dguia and DGL precursors

(A) Nucleotide and deduced amino acid sequence of the full-length open reading frame encoding the precursor of Dguia. The signal peptide sequence (residues 1–28) is underlined. The amino acid sequences of the γ- (29–147) and the β- (163–280) chains are highlighted. The eight codons coding for different amino acid residues in Dguia DNA compared with the seed-isolated lectin are highlighted in grey boxes. The degenerate nucleotide sequences of the forward (F) and reverse (R) primers, based on highly conserved amino acid sequences among legume lectins, used to amplify a specific 500 bp DNA product are shown. The nucleotide sequence (837–840) used to engineer an EcoRI restriction site is underlined. (B) Nucleotide and deduced amino acid sequence of the precursor of DGL. The forward (F) and reverse (R) primers used for amplifying the full-length DNA coding for the full-length α-chain are shown.

Construction of synthetic genes encoding the α-chains of Dguia and DGL

To produce synthetic genes coding for mature-like α (=β+γ) -chains of the Dguia and DGL, DNAs encoding the β- and γ-chains were amplified separately using primers containing restriction sites.

Amplification of the β-chains of both Dguia and DGL was performed with the forward primer β 5′- GCCATGGCCGATACTATCGTTGCTG-3′, which contains an NcoI restriction site (underlined), and a sequence coding for the first six amino acids of the polypeptide chain (ADTIVA), and the reverse primer β 5′-CCCGGGAATTCGTCTTTAACTTAGAAG-3′, which includes an EcoRI restriction site (underlined) and the last seven amino acids of the β-chain (SKLKTN).

The γ-chains were PCR-amplified with the sense primer γ 5′-CCGGAATTCAATAGCAGATG(C/A)AAAT-3′ (C for Dguia and A for DGL) that possesses an EcoRI restriction site (underlined) and the first six amino acids of the γ-chain (Dguia, SIADAN; DGL, SIADEN) and the antisense primer γ 5′-AAGCGGCCGCTCAATTTGCATCAGGAAAGAG, which includes a restriction site for NotI (underlined), a stop codon (in italics and bold), and the last six C-terminal residues of the mature lectins (LFPDAN).

The PCR protocol included denaturation at 94 °C for 2 min, followed by 25 cycles of denaturation (30 s at 94 °C), annealing (30 s at 60 °C) and extension (30 s at 72 °C), and a final extension for 7 min at 72 °C. The amplified fragments were purified using the Perfect Prep Gel Clean Up kit (Eppendorf) and were cloned into a pGEM-T vector. E. coli DH5α cells (Novagen) were transformed by electroporation using an Eppendorf 2510 electroporator following the manufacturer's instructions. Positive clones, selected by growing the transformed cells in LB (Luria–Bertani) medium containing 100 μg/ml ampicillin, were confirmed by PCR amplification using the above primers, and the PCR-amplified fragments were sequenced (using an Applied Biosystems model 377 DNA sequencer) to check the correctness of the sequences and the direction of inserts.

For generation of expression vectors comprising α-chain-thioredoxin-His6 fusion proteins, the pGEM-T plasmids containing β-chains (β-chain/pGEM-T), γ-chains (γ-chain/pGEM-T) and the pET32a vector (Novagen) were serially digested with NcoI/EcoRI, EcoRI/Not I and NcoI/NotI respectively, for 12 h at 37 °C. The 300-bp β- and γ-fragments and the pET32a vector were purified from 1.5% agarose gels using the Eppendorf Perfect Pre Gel Clean Up kit. The DNA fragment coding for the γ-chain was dephosphorylated at 37 °C for 1 h with 1 unit of calf intestinal alkaline phosphatase (Roche). After this reaction, the β- and the γ-fragments, and the open pET32a vector were ligated with 10 units of T4-DNAligase (Invitrogen) overnight at 13 °C. These constructs were used to transform electrocompetent E. coli DH5α cells. The plasmid was sequenced to confirm the correctness of the construct. The plasmid DNAs from positive clones were used to transform (by electroporation) E. coli BL21(DE3) cells (Novagen). Another pool of BL21(DE3) cells were transformed with mock pET32a plasmid and used as a negative control for the recombinant expression of α-chain-thioredoxin-His6 fusion proteins.

Generation of r-αDguia and r-αDGL mutants

Site-directed mutagenesis was performed essentially as described in the QuikChange® site-directed mutagenesis kit (Stratagene). To this end, the pET32a plasmids containing the r-αDguia and r-αDGL sequences flanked by NcoI and NotI restriction sites were used as templates for PCR amplification [initial denaturation at 94 °C for 2 min, followed by 12 cycles of denaturation (30 s at 94 °C), annealing (60 s at 55 °C), and extension (12 min at 68 °C), and a final extension for 10 min at 68 °C] using the primers described in Table 1. The mutated plasmids were sequenced to confirm the absence of undesired mutations.

Table 1
Forward (F) and reverse (R) primers used for generating the site-directed mutants of the α-chains of r-αDguia r-αDGL

Codons producing the mutations are in bold and underlined

Mutant Template used Primers (5′–3′) 
r-αDguia S131H r-αDguia F: AATTCACTCCATTTCAGCTTCCACCAGTTTAGCCAAAACCCAAAGG 
  R: GGGTTTTGGCTAAACTGGTGGAAGCTGAAATGGAGTGAATTTGC 
r-αDGL H131N r-αDGL F: TCACTCCATTTCAGCTTCAACAAATTTAGCCAAAACCCAAAGG 
  R: GGGTTTTGGCTAAATTTGTTGAAGCTGAAATGGAGTGAATTTT 
r-αDGL E123A/H131N r-αDGL H131N F: GACGAATTCAATAGCAGATGCAAATTCACTCCATTTCAGC 
  R: GCTGAAATGGAGTGAATTTGCATCTGCTATTGAATTCGTC 
r-αDGL H131N/K132Q r-αDGL H131N F: CTCCATTTCAGCTTCAACCAATTTAGCCAAAACCCAAAGG 
  R: CTTTGGGTTTTGGCTAAATTGGTTGAAGCTGAAATGGAG 
r-αDGL E123A/K132Q r-αDGL E123A/ F: AATTCACTCCATTTCAGCTTCCACCAGTTTAGCCAAAACCCAAAGG 
 H131N/K132Q R: GGGTTTTGGCTAAACTGGTGGAAGCTGAAATGGAGTGAATTTGC 
r-αDGL E123A/H131N/K132Q r-αDGL F: CTCCATTTCAGCTTCAACCAATTTAGCCAAAACCCAAAGG 
 E123A/H131N R: CTTTGGGTTTTGGCTAAATTGGTTGAAGCTGAAATGGAG 
Mutant Template used Primers (5′–3′) 
r-αDguia S131H r-αDguia F: AATTCACTCCATTTCAGCTTCCACCAGTTTAGCCAAAACCCAAAGG 
  R: GGGTTTTGGCTAAACTGGTGGAAGCTGAAATGGAGTGAATTTGC 
r-αDGL H131N r-αDGL F: TCACTCCATTTCAGCTTCAACAAATTTAGCCAAAACCCAAAGG 
  R: GGGTTTTGGCTAAATTTGTTGAAGCTGAAATGGAGTGAATTTT 
r-αDGL E123A/H131N r-αDGL H131N F: GACGAATTCAATAGCAGATGCAAATTCACTCCATTTCAGC 
  R: GCTGAAATGGAGTGAATTTGCATCTGCTATTGAATTCGTC 
r-αDGL H131N/K132Q r-αDGL H131N F: CTCCATTTCAGCTTCAACCAATTTAGCCAAAACCCAAAGG 
  R: CTTTGGGTTTTGGCTAAATTGGTTGAAGCTGAAATGGAG 
r-αDGL E123A/K132Q r-αDGL E123A/ F: AATTCACTCCATTTCAGCTTCCACCAGTTTAGCCAAAACCCAAAGG 
 H131N/K132Q R: GGGTTTTGGCTAAACTGGTGGAAGCTGAAATGGAGTGAATTTGC 
r-αDGL E123A/H131N/K132Q r-αDGL F: CTCCATTTCAGCTTCAACCAATTTAGCCAAAACCCAAAGG 
 E123A/H131N R: CTTTGGGTTTTGGCTAAATTGGTTGAAGCTGAAATGGAG 

Recombinant expression of r-α-thioredoxin-His6 fusion proteins

Positive E. coli BL21(DE3) clones, shown by PCR to contain the r-α-chain-thioredoxin-His6 fusion constructs, were grown overnight at 37 °C in LB medium containing 100 μg/ml ampicillin. For inducing the expression of the recombinant fusion proteins, the cell cultures were diluted 1:10 (v/v) with medium, IPTG (isopropyl β-D-thiogalactoside) was added to a final concentration of 1 mM, and the cell suspensions were incubated for a further 24 h at 14 °C. Thereafter the cells were pelleted by centrifugation (10000 g for 30 min at 4 °C), resuspended in the same volume of 20 mM sodium phosphate and 150 mM NaCl (pH 7.4), washed three times with the same buffer, and resuspended in 100 ml/l of initial cell culture of 20 mM sodium phosphate, 250 mM NaCl and 10 mM imidazole (pH 7.4). The cells were lysed by sonication (15 cycles of 15 s sonication followed by 1 min resting) in an ice bath. The lysates were centrifuged at 10000 g for 30 min at 4 °C, and the soluble and the insoluble fractions were analysed by SDS/PAGE (15% gels).

Purification of r-αDguia and r-αDGL

The α-chain-thioredoxin-His6 fusion proteins (r-αDguia or r-αDGL) were purified from the soluble fraction of positive E. coli BL21(DE3) clones by affinity chromatography using an ÄKTA Basic chromatographer equipped with a 5 ml HisTrap HP column (Amersham Biosciences) equilibrated in 20 mM sodium phosphate, 250 mM NaCl and 10 mM imidazole (pH 7.4). After the absorbance at 280 nm of the flow-through fraction reached baseline, the bound material was eluted at a flow rate of 1.5 ml/min with a linear gradient of 10–500 mM imidazole for 60 min. The purified protein fractions (checked by SDS/PAGE) were pooled, dialysed against 50 mM Tris/HCl (pH 7.4), and digested with 0.5 units of enterokinase (Invitrogen) per mg of recombinant protein. The reaction mixture was freed from enterokinase by chromatography on a 0.5 ml column of agarose-trypsin inhibitor (Sigma) equilibrated and eluted with 50 mM Tris/HCl (pH 7.4). The r-α-chain was separated from thioredoxin-His6 by affinity chromatography on a Sephadex G-75 column (2.0 cm×7.0 cm) equilibrated in 50 mM Tris/HCl (pH 7.4) containing 150 mM NaCl, 5 mM CaCl2 and 5 mM MnCl2. After exhaustive washing with equilibration buffer, the bound material (carbohydrate-binding r-α-chain) was eluted with equilibration buffer containing 200 mM D-glucose, dialysed for 30 min against 0.1 M acetic acid, overnight against Milli-Q water and freeze-dried. The purity of the isolated proteins was assessed by SDS/PAGE.

Analytical ultracentrifugation

The apparent molecular mass of the recombinant lectins in solutions of different pH were determined by analytical ultracentrifugation at 20 °C using a Beckman XL-A centrifuge with UV absorption scanner optics using an AN-50 Ti 8-hole rotor and charcoal-filled eppon 6-channel centrepieces. This set-up allows the simultaneous analysis of 21 different samples. The lectins were dissolved at 0.6–1.0 mg/ml in the following buffers containing 1 mM CaCl2, 1 mM MgCl2 and 0.1 M NaCl, and Tris/HCl (for pH 7.5 and 8.5), 20 mM sodium citrate (for pH 2.5, 3.5 and 4.5) and 20 mM Mes (for pH 5.5 or 6.5). Molar masses were determined by sedimentation–diffusion equilibrium experiments using short (approx. 3 mm) sedimentation columns. To avoid differences in apparent molecular masses due to rotor-speed-dependent weighting of apparent molecular masses, all experiments were carried out at the same speed (15000 rev./min). Sedimentation equilibrium was assumed to be attained when the measured concentration profile remained unchanged for at least 12 h. The equilibrium concentration gradient for a single species is described by the equation:

 
formula

where c(r) and c(m) are the concentrations at radius r and at the meniscus (radius m) respectively; M is the molar mass of the solute; υ is the partial specific volume of the solute (assumed to be 7.35×10 m−3×kg); ω is the angular speed of the rotor (radians×s−1); and R and T are the gas constant (8.314 J·mol−1·K−1) and the temperature (in K) respectively. Apparent molar masses were determined by fitting this function to the measured radial distribution of the concentration gradient at equilibrium using the program EQASSOC [33] provided by the manufacturer. Blank buffer absorption was determined after overspeeding (50000 rev./min) to sediment all material to the bottom of the cell.

Crystallization, data collection and model building

Crystals of r-αDguia were grown at 22 °C by the vapour diffusion method using hanging drops composed of equal volumes of protein solution [10–15 mg/ml in 0.1 M Mes (pH 6.0), containing 10 mM CaCl2, 10 mM MnCl2, 3 mM X-Man (5-bromo-4-chloro-3-indolyl-α-D-mannose, 12 mM stock solution in DMSO)], and of the reservoir solution [30% poly(ethylene glycol) 400, 100 mM Mes (pH 6.4) and 100 mM CdCl2] used as the precipitant.

Crystals of r-αDguia S131H were obtained as above except that the reservoir solution was 600 mM NaCl and 100 mM Hepes (pH 7.0).

For crystallization of r-αDGL and its triple mutant E123A/H131N/K132N, the lectins were dissolved at approx. 10–12 mg/ml in 50 mM Hepes (pH 7.5), containing 3 mM X-Man and 10 mM each of MnCl2 and CaCl2. Crystals of r-αDGL were grown at constant room temperature (20 °C) by the hanging-drop vapour diffusion method using 15% poly(ethylene glycol) 4000, 100 mM Hepes (pH 7.5) and 100 mM lithium sulfate as the precipitant. Crystals of the triple mutant were obtained with the same method but using 15% poly(ethylene glycol) 8000, 100 mM cacodylate (pH 6.5) containing 200 mM zinc acetate as the precipitant.

A reservoir buffer containing 20–25% glycerol proved to be a suitable cryoprotectant. The crystals were transferred directly from the drop to the cryoprotectant solution and were allowed to equilibrate for approx. 10 s. Thereafter, single crystals were mounted in cryo-loops and rapidly transferred to the cryostream. Cryoprotected crystals were analysed using the Synchrotron Radiation Source at the ESRF (European Synchrotron Radiation Facility, Grenoble, France) on beam lines ID23-1 (for r-αDguia), BM14 (for r-αDguia S131H), ID23-2 (for r-αDGL) and BM16 (for r-αDGL triple mutant E123A/H131N/K132N).

Crystallographic data were indexed, integrated and scaled using MOSFLM [34] and SCALA [35]. The crystal structures were determined by molecular replacement using the MOLREP program [36]. The search models used were Dguia structures (PDB accession code 1H9P for r-αDguia and 1H9W for r-αDguia S131H) [28] and DGL (PDB accession code 1DGL [27] for r-αDGL). The crystal structure of the triple mutant r-αDGL E123A/H131N/K132N was determined using the refined structure of r-αDGL. The position and orientation of the molecules as a single rigid body entity were refined for 20 cycles with REFMAC [CCP4] [35]. Several steps of rebuilding, interspersed with restrained refinement using REFMAC, yielded the final models at resolutions ranging from 1.65 to 2.1 Å (Table 2). The X-Man molecule was placed by inspection of the FoFc maps. Finally, water molecules were placed in the model over several steps of refinement with Arp/Warp and inspected manually. The atomic co-ordinates, fitted with Coot [37], are accessible from the PDB under accession codes 2JDZ (r-αDguia), 2JE7 (r-αDguia S131H), 2JE9 (r-αDGL) and 2JEC (r-αDGL E123A/H131N/K132N).

Table 2
Summary of data collection and refinement statistics
 r-αDguia r-αDguia S131H r-αDGL r-αDGL E123A/H131N/K132Q 
Data collection statistics     
 Space group I4122 I222 P212121 P212121 
 Cell constants     
 a (Å) 88.2 65.6 72.2 72.2 
 b (Å) 88.2 88.0 85.2 84.5 
 c (Å) 106.1 91.5 175.7 176.2 
 α=β=γ 90° 90° 90° 90° 
 Resolution(Å) 2.1 1.65 2.1 2.0 
 Total reflections 133461 98371 295146 278161 
 Unique reflections 12834 31757 63973 73352 
 Data coverage (%) 99.9 (100.0)* 98.8 (99.3)* 99.9 (99.7)* 99.6 (100.00)* 
 Rmerge (%)a 8.2 (30.3) 4.9 (36.2) 7.9 (30.3) 7.2 (25.2) 
 Multiplicity 10.4 (10.6) 3.1 (3.0) 4.6 (4.0) 3.8 (3.8) 
I/σ<ι>: 5.6 (2.1) 8.4 (2.1) 6.9 (2.1) 12.9 (3.0) 
 Molecules in asymmetric unit Monomer Monomer Tetramer Tetramer 
Refinement statistics     
 Resolution range (Å) 34.1–2.1 44.0–1.65 72.2–2.1 28.9–2.0 
 Reflections 11848 30142 60684 69591 
 Rfactor (%)b 19.6 15.4 18.7 18.0 
 Rfree (%) 24.9c 20.1d 23.9d 23.0d 
 rmsd     
  Bond lengths (Å) 0.014 0.012 0.015 0.015 
  Bond angles(deg.) 1.6 1.5 1.6 1.5 
Temperature factors (A2    
 Main chain atoms (average) 31.0 18.7 24.0 18.2 
 X-Man (average) 42.2 21.0 37.7 24.9 
 X-Man Binding Loop atoms 34.0 20.9 26.0 19.0 
 r-αDguia r-αDguia S131H r-αDGL r-αDGL E123A/H131N/K132Q 
Data collection statistics     
 Space group I4122 I222 P212121 P212121 
 Cell constants     
 a (Å) 88.2 65.6 72.2 72.2 
 b (Å) 88.2 88.0 85.2 84.5 
 c (Å) 106.1 91.5 175.7 176.2 
 α=β=γ 90° 90° 90° 90° 
 Resolution(Å) 2.1 1.65 2.1 2.0 
 Total reflections 133461 98371 295146 278161 
 Unique reflections 12834 31757 63973 73352 
 Data coverage (%) 99.9 (100.0)* 98.8 (99.3)* 99.9 (99.7)* 99.6 (100.00)* 
 Rmerge (%)a 8.2 (30.3) 4.9 (36.2) 7.9 (30.3) 7.2 (25.2) 
 Multiplicity 10.4 (10.6) 3.1 (3.0) 4.6 (4.0) 3.8 (3.8) 
I/σ<ι>: 5.6 (2.1) 8.4 (2.1) 6.9 (2.1) 12.9 (3.0) 
 Molecules in asymmetric unit Monomer Monomer Tetramer Tetramer 
Refinement statistics     
 Resolution range (Å) 34.1–2.1 44.0–1.65 72.2–2.1 28.9–2.0 
 Reflections 11848 30142 60684 69591 
 Rfactor (%)b 19.6 15.4 18.7 18.0 
 Rfree (%) 24.9c 20.1d 23.9d 23.0d 
 rmsd     
  Bond lengths (Å) 0.014 0.012 0.015 0.015 
  Bond angles(deg.) 1.6 1.5 1.6 1.5 
Temperature factors (A2    
 Main chain atoms (average) 31.0 18.7 24.0 18.2 
 X-Man (average) 42.2 21.0 37.7 24.9 
 X-Man Binding Loop atoms 34.0 20.9 26.0 19.0 

* Data in parentheses are for the highest resolution shell.

a Rmerge=|I−〈I〉)|/〈I〉), where I is the intensity and 〈I〉 is the average I for all observations of equivalent reflections.

b R-factor=|Fobs|−|Fcalc|/|Fobs|.

c Calculated with 8%

d5% of the reflections omitted from refinement.

RESULTS AND DISCUSSION

In the present study we have primarily addressed the question of the role of interdimeric interactions at the central cavity in confering pH-dependent dimer–tetramer transition upon Diocleinae lectins. In order to do this it was necessary to produce recombinant lectins. To circumvent the inability of E. coli to post-translationally process pre-pro Diocleinae lectins as in the plant system (discussed below), we have designed and cloned synthetic genes coding for mature wild-type and mutated single-polypeptide α-chains. This novel approach, which may inspire molecular biological studies of other lectins, was successfully employed for producing recombinant wild-type and mutated proteins biochemically indistinguishable from the naturally occurring seed lectins, and is described in the following sections.

The Dguia and DGL precursors are coded for by intronless genes

PCR using primers designed from the cDNA sequences of the highly-related ConA [38] and ConBr (Canavalia brasiliensis lectin) [39] lectin precursors failed to amplify any product using cDNA or gDNA of either Dguia and DGL. However, using degenerated primers based on highly conserved amino acid sequences among legume lectins, a 500 bp product was PCR-amplified (Figure 1). This cDNA-amplified nucleotide sequence was then used to design specific primers to extend the Dguia transcript. The Dguia cDNA sequence was extended in the 3′ direction by 3′RACE. The upstream Dguia gene sequence was determined by nested PCR using RsaI-digested gDNA as a template. The 3′RACE- (cDNA) and 5′-extension- (gDNA) gathered nucleotide sequences were employed to construct specific primers, which were then used to amplify a fragment of 876 bp encoding the full-length Dguia open reading frame (Figure 1A). Identical sequences were amplified using cDNA and gDNA templates, clearly showing that the gene coding for the Dguia did not possess introns. Similarly, the seed lectins from Canavalia ensiformis [38], Canavalia brasilensis [39] and Canavalia gladiata [40] have been previously reported to be coded for by intronless genes.

The DNA-deduced amino acid sequence of the Dguia precursor (Figure 1) differs at eight positions from the primary structure of Dguia isolated from seeds [24,28]. The eight conservative changes [Dguia(DNA)/Dguia(seed): S41A, G54S, Q65E, A68K, D72S, S74D, S230T and T232S in Figure 1A) were found in every clone sequenced, ruling out a DNA sequencing error. None of these residues participated in interdimeric contacts in the crystal structure of the Dguia seed lectin [28]. Of note, the isotope-averaged molecular masses calculated for the full-length αDguia (25395 Da) and its derived β- (12832 Da) and γ- (12584 Da) fragments are in excellent agreement with those determined by electrospray-ionization MS for the Dguia seed lectin [24], suggesting that the cloned and the seed-isolated lectins may be isobaric isolectins.

The DNA sequence coding for αDGL (Figure 1B) was PCR-amplified using gDNA as a template and forward and reverse primers designed from the DNA sequence of the Dguia precursor. The same 783 bp product was amplified using cDNA as a template, indicating that, like the Dguia lectin, the DGL gene does not include any intronic sequence. The calculated and the experimentally determined masses of the αDGL (25607 Da), β- (12872 Da) and γ- (12752 Da) chains [24] coincide within the experimental error limit (±2 Da).

Design and cloning of synthetic genes coding for αDguia and αDGL

Maturation of Diocleinae seed lectins involves a complex mechanism of post-translational processing that includes deglycosylation-dependent proteolytic cleavage of the precursor at the carboxy site of internal asparagine residues, and religation of the resulting γ- and β-fragments in inverse DNA-coded order [38,41]. Thus the resulting active α-chain (α=β−γ) is circularly permuted in primary sequence relative to its own inactive γ−β precursor. Proteolytic cleavage and religation of the fragments to form the α-chain are both catalysed by an asparagine endopeptidase [42]. The nature of the splicing and religation events has been controversial, with the report of Yamauchi and Minamikawa [40] that pre-pro-ConA from C. gladiata expressed in E. coli underwent peptide cleavage and ligation in the same way as the lectin synthesized during seed maturation, contrasting with a lack of evidence for any processing of pre-pro-ConA in the cytoplasm [43,44] or of pre-ConA in the periplasm of E. coli [45]. Indeed, E. coli does not express asparagine endopeptidase and is therefore unable to process the legume lectin precursor as in the plant [43]. Thus to produce native-like and active Dguia and DGL, we have designed and cloned synthetic genes coding for the single polypeptide α-chains. The cloning strategy employed for constructing synthetic genes for the α-chains from the DNA-amplified precursors of the Dguia and DGL, schematically displayed in Figure 2, included PCR engineering of EcoRI restriction sites at the 3′of the β-chain and at the 5′of the γ-chain. The β/pGEM-T and γ/pGEM-T constructs were then serially digested with EcoRI/NcoI and EcoRI/NotI respectively, generating complementary cohesive ends that were subsequently ligated into a full-length α-chain flanked by 5′-NcoI and 3′-NotI restriction sites. The synthetic α-chain was cloned into an open pET32a expression vector digested with NcoI and NotI, creating an α-chain-thioredoxin-His6 fusion protein.

Strategy followed for construction of synthetic αDguia and αDGL

Figure 2
Strategy followed for construction of synthetic αDguia and αDGL

After PCR-engineering of EcoRI restriction sites at the 3′of the β-chain and at the 5′of the γ-chain, the β- and the γ-fragments were separately cloned into pGEM-T vectors containing, respectively, NcoI and NotI restriction sites. The vectors were serially digested, the fragments were purified by agarose gel electrophoresis, the γ-chain was dephosphorylated, and the β- and γ-chains were ligated through their EcoRI-created cohesive ends. Synthetic α-chains were cloned into open expression vectors pET32a, and the constructs were expressed as α-chain-thiredoxin-His6 fusion proteins in E. coli.

Figure 2
Strategy followed for construction of synthetic αDguia and αDGL

After PCR-engineering of EcoRI restriction sites at the 3′of the β-chain and at the 5′of the γ-chain, the β- and the γ-fragments were separately cloned into pGEM-T vectors containing, respectively, NcoI and NotI restriction sites. The vectors were serially digested, the fragments were purified by agarose gel electrophoresis, the γ-chain was dephosphorylated, and the β- and γ-chains were ligated through their EcoRI-created cohesive ends. Synthetic α-chains were cloned into open expression vectors pET32a, and the constructs were expressed as α-chain-thiredoxin-His6 fusion proteins in E. coli.

Recombinant expression in E. coli of synthetic and active αDguia and αDGL

Recombinant synthetic αDguia and αDGL (r-αDguia and r-αDGL) were expressed as thioredoxin-His6-tagged fusion proteins in E. coli BL21(DE3) cells. Biosynthesis of the recombinant construct was independent of the addition of the Lac operon inducer IPTG and approx. 90% of the fusion protein was recovered in the soluble cell lysate (Figure 3). Following affinity chromatography in HisTrap and digestion with enterokinase, r-αDguia and r-αDGL were purified to apparent homogeneity by affinity chromatography on Sephadex G-75. The average protein yield was similar for both lectins, approx. 20 mg/l. The r-αDguia and r-αDGL were quantitatively retained in the cross-linked dextran gel column and were desorbed with D-glucose, providing strong evidence that the lectins displayed native-like, carbohydrate-binding activity. In line with this assumption, r-αDguia and r-αDGL also exhibited identical haemagglutinating activity as the corresponding lectins isolated from seeds (results not shown).

SDS/PAGE (12.5% gel) analysis of the recombinant expression and the purification of the synthetic αDguia

Figure 3
SDS/PAGE (12.5% gel) analysis of the recombinant expression and the purification of the synthetic αDguia

The same result was obtained with the synthetic DGL construct. Lane a, cell lysates of E. coli BL21(DE3) cells transformed with the mock pET32a plasmid. Lanes b and c, 100 μg of total proteins of the soluble and insoluble fractions respectively of the cell lysate of BL21(DE3) cells expressing the αDguia-thioredoxin-His6 fusion protein. Lane d, HisTrap affinity-purified αDguia-thioredoxin-His6 construct. Lane e, protein mixture generated by digestion with enterokinase of the αDguia-thioredoxin-His6 fusion protein. Lane f, r-αDguia purified by affinity chromatography on Sephadex G75. Lane g, mature Dguia isolated seeds exhibiting the typical α- (25 kDa), β- (14 kDa) and γ- (12 kDa) mixture. Lane s, molecular-mass markers (Mark12; Invitrogen), from top to bottom: BSA (66.3 kDa), glutamic dehydrogenase (55.4 kDa), lactate dehydrogenase (36.5 kDa), carbonic anhydrase (31.0 kDa), soybean trypsin inhibitor (21.5 kDa), lysozyme (14.4 kDa) and aprotinin (6 kDa).

Figure 3
SDS/PAGE (12.5% gel) analysis of the recombinant expression and the purification of the synthetic αDguia

The same result was obtained with the synthetic DGL construct. Lane a, cell lysates of E. coli BL21(DE3) cells transformed with the mock pET32a plasmid. Lanes b and c, 100 μg of total proteins of the soluble and insoluble fractions respectively of the cell lysate of BL21(DE3) cells expressing the αDguia-thioredoxin-His6 fusion protein. Lane d, HisTrap affinity-purified αDguia-thioredoxin-His6 construct. Lane e, protein mixture generated by digestion with enterokinase of the αDguia-thioredoxin-His6 fusion protein. Lane f, r-αDguia purified by affinity chromatography on Sephadex G75. Lane g, mature Dguia isolated seeds exhibiting the typical α- (25 kDa), β- (14 kDa) and γ- (12 kDa) mixture. Lane s, molecular-mass markers (Mark12; Invitrogen), from top to bottom: BSA (66.3 kDa), glutamic dehydrogenase (55.4 kDa), lactate dehydrogenase (36.5 kDa), carbonic anhydrase (31.0 kDa), soybean trypsin inhibitor (21.5 kDa), lysozyme (14.4 kDa) and aprotinin (6 kDa).

As judged by sedimentation equilibrium, r-αDguia exhibited similar pH-dependent dimer–tetramer transition to the Dguia (Figure 4). On the other hand, r-αDGL, like DGL isolated from seeds, behaved as a pH-independent tetrameric protein (Table 3). Furthermore, r-αDguia and r-αDGL were crystallized and their structures, solved to 2.1 Å resolution (Table 2), could be superimposed on those of their natural homologues with an overall rmsd for all Cα atoms of 0.33/0.45 Å (monomer/tetramer, r-αDGL/1DGL) and 0.35/0.56 Å (monomer/tetramer, r-αDguia/1H9P).

Determination by equilibrium sedimentation analytical centrifugation of the apparent molecular masses of the recombinant lectins as a function of the pH of the solution

Figure 4
Determination by equilibrium sedimentation analytical centrifugation of the apparent molecular masses of the recombinant lectins as a function of the pH of the solution

Apparent molecular masses (Mapp) between a pure dimer (D, 51 kDa) and a pure tetramer (T, 102 kDa) correspond to Mapp=(%D×51+%T×102) kDa. The location of the peripheral (P) and central (C) interdimeric regions are indicated in the structures of the canonical dimer and tetramer of r-αDguia.

Figure 4
Determination by equilibrium sedimentation analytical centrifugation of the apparent molecular masses of the recombinant lectins as a function of the pH of the solution

Apparent molecular masses (Mapp) between a pure dimer (D, 51 kDa) and a pure tetramer (T, 102 kDa) correspond to Mapp=(%D×51+%T×102) kDa. The location of the peripheral (P) and central (C) interdimeric regions are indicated in the structures of the canonical dimer and tetramer of r-αDguia.

Table 3
Protein–carbohydrate interactions in the mannose-binding site of the crystal structures of wild-type and mutated r-αDguia– and r-αDGL–X-Man complexes

For comparison, the interactions between r-αDGL and the mannose residue occupying the monosaccharide-binding site in the crystal structure of the lectin–trimannoside complex (PDB accession number 1DGL) are listed.

  Interacting distance (Å) 
Sugar atom Amino acid residue atom r-αDguia (2JDZ) r-αDguia S131/H (2JE7) r-αDGL (2JE9) r-αDGL H131N/K132Q/E123A (2JEC) 1DGL 
O3 R2282.9 2.9 2.9 2.9 2.9 
O4 N14 ND2 2.8 2.9 2.9 2.8 2.8 
 D208 OD2 2.5 2.5 2.6 2.6 2.7 
O5 L993.4 3.1 3.3 3.2 2.9 
O6 Y1003.1 3.3 3.5 3.4 3.0 
 L993.3 3.1 3.2 3.3 2.9 
 Y1003.1 3.0 3.1 3.1 3.0 
 D208 OD1 2.7 2.7 2.0 2.9 3.3 
  Interacting distance (Å) 
Sugar atom Amino acid residue atom r-αDguia (2JDZ) r-αDguia S131/H (2JE7) r-αDGL (2JE9) r-αDGL H131N/K132Q/E123A (2JEC) 1DGL 
O3 R2282.9 2.9 2.9 2.9 2.9 
O4 N14 ND2 2.8 2.9 2.9 2.8 2.8 
 D208 OD2 2.5 2.5 2.6 2.6 2.7 
O5 L993.4 3.1 3.3 3.2 2.9 
O6 Y1003.1 3.3 3.5 3.4 3.0 
 L993.3 3.1 3.2 3.3 2.9 
 Y1003.1 3.0 3.1 3.1 3.0 
 D208 OD1 2.7 2.7 2.0 2.9 3.3 

Similarly to what has been reported for Dguia [28], the r-αDguia crystal structure displayed poor electronic density for the central cavity loop residues 117–123. This observation indicates that disorder of the corresponding loop in the natural lectin is not due to the incomplete ligation of the C-terminal residue of the β-chain and the N-terminal residue of the γ-fragment that leads to the mixture of α-, β- and γ-chains.

The architecture of the primary monosaccharide recognition site and the interactions established between the protein and the mannose residue of X-Man (Figures 5A–5C and Table 3) are essentially the same as in other Diocleinae lectin–sugar complexes. The quality of the electron density maps around the X-Man moiety is shown in Figure 5(C). Although the apo- (PDB accession numbers 1H9P and 1H9W) and the sugar-bound (r-αDguia, PDB accession number 2JDZ) wild-type Dguias possess identical preformed carbohydrate-recognition motifs, the average B-factor for the main chain of the mannose-binding loops of apo-Dguia (52 Å2) is significantly higher than that for the r-αDguia–X-Man complex (34 Å2) (Figures 5D and 5E), suggesting that binding of the brominated monosaccharide-derivative restrains the conformational flexibility of the sugar-binding, solvent-exposed loops. The low average B-factors of the mannose-binding loops of r-αDguia S131H–X-Man (20.9 Å2), r-DGL–X-Man (26.0 Å2) and r-αDguia E123A/H131N/K132Q–X-Man (19.0 Å2) (Table 2) further support this conclusion. A similar observation has been reported for the seed lectin from Parkia platycephala when the apo- and the X-Man-complexed structures were compared [46].

The X-Mannose-binding sites

Figure 5
The X-Mannose-binding sites

Comparison of the mannose-binding sites of (A) the r-αDGL–X-Man complex and (B) the DGL complexed with a trimannoside (PDB accession number 1DGL) [27]. Broken lines represent hydrogen bonds. For detailed interactions between lectin and sugar atoms see Table 3. (C) Shows a portion of the 2FoFc electron density map contoured at 1.0 σ showing the X-Man bound to the carbohydrate-binding site of monomer A of the Dguia S131H mutant. (D and E) Display, respectively, close-up views of the carbohydrate-recognition domains of r-αDguia with bound X-Man (PDB accession number 2JDZ; the present study) and apo-Dguia (PDB accession number 1H9P) [28]. B-factors are colour-coded from 25 Å2 (green) to 60 Å2 (red).

Figure 5
The X-Mannose-binding sites

Comparison of the mannose-binding sites of (A) the r-αDGL–X-Man complex and (B) the DGL complexed with a trimannoside (PDB accession number 1DGL) [27]. Broken lines represent hydrogen bonds. For detailed interactions between lectin and sugar atoms see Table 3. (C) Shows a portion of the 2FoFc electron density map contoured at 1.0 σ showing the X-Man bound to the carbohydrate-binding site of monomer A of the Dguia S131H mutant. (D and E) Display, respectively, close-up views of the carbohydrate-recognition domains of r-αDguia with bound X-Man (PDB accession number 2JDZ; the present study) and apo-Dguia (PDB accession number 1H9P) [28]. B-factors are colour-coded from 25 Å2 (green) to 60 Å2 (red).

Taken together, our results indicate that the recombinant lectins are structurally and biochemically indistinguishable from their natural homologues.

Conversion by site-directed mutagenesis of r-αDGL into a lectin exhibiting pH-dependent dimer–tetramer transition

The amino acid sequences of r-αDguia and r-αDGL differ in only 11 residues, of which three are located in the 123–132 region: rDguia, A121DANSLHFSFSQFS134 and rDGL, A121DENSLHFSFHKFS134 (the differing residues are underlined). Previous studies [28] have suggested that residues within this region of DGL participate in the interdimeric contacts that yield a pH-independent tetrameric structure. To gain a closer insight into the mechanism governing the pH-dependent dimer–tetramer transition, we have produced single-, double- and triple-mutated recombinant lectins and have assessed their structure–activity correlations by a combination of equilibrium sedimentation analytical ultracentrifugation and X-ray crystallography.

The crystal structures of r-αDguia, r-αDguia S131H, r-αDGL and r-αDGL E123A/H131N/H132K can be superimposed with an rmsd for Cα atoms of 0.3–0.7 Å, with interdimeric hydrogen-bonding contacts between Arg60 and Asp78 of opposed monomers at the peripheral sites essentially conserved (Figure 6), whereas the major differences appear to be associated with the central cavity loops (Figure 7A).

Interdimeric contacts at the peripheral regions

Figure 6
Interdimeric contacts at the peripheral regions

Comparison of the interactions (hydrogen bonds, broken lines) at the peripheral interdimeric regions between monomers A and D (see Figure 7A) of natural (PDB accession number 1DGL) (A) and recombinant wild-type (rDGL) (B) and triple-mutated (rDGL E123A/H131N/H132K) (C) DGL. The latter exhibits pH-dependent dimer–tetramer transition, whereas 1DGL and rDGL are pH (4.5–8.5)-independent tetrameric structures (see Figure 4). Interactions between monomers B and C are essentially the same.

Figure 6
Interdimeric contacts at the peripheral regions

Comparison of the interactions (hydrogen bonds, broken lines) at the peripheral interdimeric regions between monomers A and D (see Figure 7A) of natural (PDB accession number 1DGL) (A) and recombinant wild-type (rDGL) (B) and triple-mutated (rDGL E123A/H131N/H132K) (C) DGL. The latter exhibits pH-dependent dimer–tetramer transition, whereas 1DGL and rDGL are pH (4.5–8.5)-independent tetrameric structures (see Figure 4). Interactions between monomers B and C are essentially the same.

Crystal structures of recombinant lectins

Figure 7
Crystal structures of recombinant lectins

(A) Superposition of the Cα traces of recombinant lectins r-αDguia (black), r-αDguia S131H (red), r-αDGL (blue) and r-αDGL E123A/H131N/H132K (green). (B) Detail of the interdimeric interactions at the central cavity of r-αDGL between His131 and Iso120 from an opposite monomer that render a pH-stable tetrameric structure. (C) Detail of the intradimeric contacts at the central cavity of r-αDguia S131H. (D) Close-up view of the interdimeric region at the central cavity of r-αDGL (blue) and its triple mutant r-αDGL E123A/H131N/H132K (green). (E and F) Comparison of the intradimeric hydrogen-bond network in the crystal structures of r-αDGL (blue) and its triple mutant (green) showing the structural changes caused by the E123A/H131N/H132K mutations, which generate a 117–123 loop conformation less extended towards the centre of the central cavity than the corresponding structure in r-αDGL (see also D).

Figure 7
Crystal structures of recombinant lectins

(A) Superposition of the Cα traces of recombinant lectins r-αDguia (black), r-αDguia S131H (red), r-αDGL (blue) and r-αDGL E123A/H131N/H132K (green). (B) Detail of the interdimeric interactions at the central cavity of r-αDGL between His131 and Iso120 from an opposite monomer that render a pH-stable tetrameric structure. (C) Detail of the intradimeric contacts at the central cavity of r-αDguia S131H. (D) Close-up view of the interdimeric region at the central cavity of r-αDGL (blue) and its triple mutant r-αDGL E123A/H131N/H132K (green). (E and F) Comparison of the intradimeric hydrogen-bond network in the crystal structures of r-αDGL (blue) and its triple mutant (green) showing the structural changes caused by the E123A/H131N/H132K mutations, which generate a 117–123 loop conformation less extended towards the centre of the central cavity than the corresponding structure in r-αDGL (see also D).

Mutant r-αDguia S131H exhibited a similar dimer–tetramer equilibrium to the wild-type (Figure 4). This result was unexpected since previous work strongly suggested that a histidine at position 131 was responsible for stabilizing the pH-independent tetrameric structure of DGL [28]. However, inspection of the crystal structure of the S131H mutant showed that His131 established contact with Asp122 from an adjacent monomer of the canonical dimer (Figure 7B). The absence of interactions between His131 and the central cavity loop residues, as in r-αDGL (Figure 7C) from an opposed monomer explains why the mutation did not lock r-αDguia S131H into a pH-independent tetramer.

Also unexpected, the single-mutant r-αDGL H131N and the double-mutants r-αDGL E123A/H131N and r-αDGL H131N/K132Q, all exhibited the same equilibrium sedimentation behaviour as wild-type r-αDGL (Figure 4), indicating that mutating His131 into an asparagine residue does generate a pH-dependent dissociable tetramer. On the other hand, pH-dependent tetramer dissociation was achieved with the double-mutant r-αDGL E123A/K132Q and the triple-mutant r-αDGL E123A/H131N/K132Q (Figure 4). The r-αDGL E123A/K132Q mutant is equivalent to r-αDguia S131H discussed above.

The crystal structure of r-αDGL E123A/H131N/K132Q was solved to a resolution of 2.0 Å (Table 2). This structure and that of DGL (PDB accession number 1DGL) [27] can be superimposed with an rmsd of 0.56 Å for all Cα atoms, except for the central cavity loops (residues 117–123), where significant conformational differences were noticed (Figure 7D). Thus in the triple-mutant structure the polypeptide stretch 117–123 folds into a well-defined loop which adopts a less extended projection towards the centre of the central cavity than the corresponding loop in 1DGL (Figure 7D). The reason for this different loop 117–123 conformation in the triple mutant and in 1DGL lies in the distinct intradimeric interactions established by residues 123, 131 and 132 in each structure (Figures 7E and 7F). In 1DGL, Oϵ2 of Glu123 (from monomers D and A) interacts with NZ of Lys132 (C/B), and the main-chain oxygen of Glu123 is hydrogen-bonded to the main-chain nitrogens of residues His131(C/B) and Lys132(C/B). In the triple mutant, the Cα residues of Asp122 and Ala123 are displaced 3.34 Å and 3.44 Å respectively, from the position occupied in r-αDGL, the amide oxygen of Gln132 and Asn131(C/B) contact respectively, ND2 and OD1 of Asn124(D/A), and ND2 of Asn131(C/B) makes a hydrogen bond with the main-chain oxygen of Asp122(D/A). It is worth noting that the conformation of residues D122AN124 in the r-αDGL E123A/H131N/K132Q structure mirrors that adopted by the same residues in the crystal structure of Dguia (PDB accession number 1H9P) [28]. However, the remainder of the central loop (residues 117–121) is disordered in hindering any structural comparison between the triple-mutated r-αDGL and Dguia. Nevertheless, the consequence of the structural alteration in the triple-mutated loop structure renders it unable to establish the interdimeric tetramer-stabilizing interactions observed in the wild-type DGL.

Concluding remarks

Evidence gathered from comparison of the crystal structures of lectins isolated from seeds of Diocleinae, which exhibit [ConA (PDB accession number 1NLS), ConBr (PDB accession number 1AZD), Dguia (PDB accession number 1H9P), CFL (PDB accession numbers 2D3P and 2D3R), Cratylia mollis lectin (PDB accession number 1MVQ), Canavalia gladiata lectin (PDB accession number 1WUV), Canavalia maritima lectin (PDB accession number 2CWM)] or not {DGL (PDB accession number 1DGL), D. violacea lectin [47]} dimer–tetramer transition in the pH range 4.5–8.5, have indicated that the occurrence of interdimeric interactions between residues at positions 131 and 120 of opposite monomers locks the quaternary structure of the lectin into a pH-independent tetramer. Similarly, conversion of the pH-stable tetrameric r-αDGL into a structure exhibiting pH-dependent dimer–tetramer equilibrium was accomplished through mutations that abolished the interdimeric interactions at the central cavity. In addition to these ‘central’ loop contacts, other interdimer interactions are established between His51 of monomers A and B and residues at positions 116 and 117 of monomers D and C respectively, and between homologous regions encompassing residues 53–78 of monomers A and D and B and C at the periphery of the dimers. Both the central- and the peripheral-interacting regions may bear information for formation of the canonical legume lectin tetramer (Figure 4). The ‘peripheral’ interfaces include a network of interactions orchestrated by residues Asn55, Arg60, Asp78, which are absolutely conserved in legume lectins that adopt the crosswise tetrameric association classified as type II by Brinda et al. [48]. Our working hypothesis is that the strength of the ionic contacts established by His51, Arg60 and Asp78 may be modulated by pH, leading to protonation-dependent tetramer dissociation of those lectin structures that are not stabilized through interdimeric interactions networking the central cavity loops of the two dimers.

This work has been financed in part by grant BFU2004-01432/BMC from the Ministerio de Educación y Ciencia, Madrid, Spain (to J. J. C.). C. S. N. is recipient of a Ph.D. fellowship from the Coordenação Aperfeiçoamento de Pessoal de Nivel Superior (CAPES/MEC, Brazil). B. S. C. is a senior investigator of CNPq (Brazil).

Abbreviations

     
  • CFL

    Cratylia floribunda lectin

  •  
  • ConA

    concanavalin A

  •  
  • DGL

    Dioclea grandiflora lectin

  •  
  • Dguia

    Dioclea guianensis lectin

  •  
  • gDNA

    genomic DNA

  •  
  • IPTG

    isopropyl β-D-thiogalactoside

  •  
  • RACE

    rapid amplification of cDNA ends

  •  
  • r-αDGL

    recombinant α-chain from DGL

  •  
  • r-αDguia

    recombinant α-chain from Dguia

  •  
  • rmsd

    root mean square deviation

  •  
  • RT

    reverse transcription

  •  
  • X-Man

    5-bromo-4-chloro-3-indolyl-α-D-mannose

References

References
1
Van Damme
E. J. M.
Peumans
W. J.
Barre
A.
Rougé
P.
Plant lectins: a composite of several distinct families of structurally and evolutionary related proteins with diverse biological roles
Crit. Rev. Plant Sci.
1998
, vol. 
17
 (pg. 
575
-
692
)
2
Peumans
W. J.
Van Damme
E. J.
Barre
A.
Rougé
P.
Classification of plant lectins in families of structurally and evolutionary related proteins
Adv. Exp. Med. Biol.
2001
, vol. 
491
 (pg. 
27
-
54
)
3
Gabius
H.-J.
Gabius
S.
Glycoscience, Status and Perspectives
1997
Weinheim (Germany)
Chapman & Hall
4
Gabius
H.-J.
Siebert
H. C.
André
S.
Jiménez-Barbero
J.
Rüdiguer
H.
Chemical biology of the sugar code
Chembiochem
2004
, vol. 
5
 (pg. 
740
-
764
)
5
Loris
R.
Principles of structures of animal and plant lectins
Biochim. Biophys. Acta
2002
, vol. 
1572
 (pg. 
198
-
208
)
6
Elgavish
S.
Shaanan
B.
Lectin–carbohydrate interactions: different folds, common recognition principles
Trends Biochem. Sci.
1997
, vol. 
22
 (pg. 
462
-
467
)
7
Drickamer
K.
Multiplicity of lectin–carbohydrate interactions
Nat. Struct. Biol.
1995
, vol. 
2
 (pg. 
437
-
439
)
8
Rini
J. M.
Lectin structure
Annu. Rev. Biophys. Biomol. Struct.
1995
, vol. 
24
 (pg. 
551
-
577
)
9
Bouckaert
J.
Hamelryck
T.
Wyns
L.
Loris
R.
Novel structures of plant lectins and their complexes with carbohydrates
Curr. Opin. Struct. Biol.
1999
, vol. 
9
 (pg. 
572
-
577
)
10
Vijayan
M.
Chandra
N.
Lectins
Curr. Opin. Struct. Biol.
1999
, vol. 
9
 (pg. 
707
-
714
)
11
Prabu
M. M.
Suguna
K.
Vijayan
M.
Variability in quaternary association of proteins with the same tertiary fold: a case study and rationalization involving legume lectins
Struct. Funct. Genet.
1999
, vol. 
35
 (pg. 
58
-
69
)
12
Chandra
N.
Prabu
M. M.
Suguna
K.
Vijayan
M.
Structural similarity and functional diversity in proteins containing the legume fold
Prot. Engineering
2001
, vol. 
14
 (pg. 
857
-
866
)
13
Srinivas
V. R.
Reddy
G. B.
Ahmad
N.
Swaminathan
C. P.
Mitra
N.
Surolia
A.
Legume lectin family, the ‘natural mutants of the quaternary state’, provide insights into the relationship between protein stability and oligomerization
Biochim. Biophys. Acta
2001
, vol. 
1527
 (pg. 
102
-
111
)
14
Reference deleted
15
Loris
R.
Hamelryck
T.
Bouckaert
J.
Wyns
L.
Legume lectin structure
Biochim. Biophys. Acta
1998
, vol. 
1383
 (pg. 
9
-
36
)
16
Chervenak
M. C.
Toone
E. J.
Calorimetric analysis of the binding of lectins with overlapping carbohydrate-binding ligand specificities
Biochemistry
1995
, vol. 
34
 (pg. 
5685
-
5695
)
17
Dam
T. K.
Brewer
C. F.
Thermodynamic studies of lectin–carbohydrate interactions by isothermal titration calorimetry
Chem. Rev.
2002
, vol. 
102
 (pg. 
387
-
429
)
18
Edelman
G. M.
Cunningham
B. A.
Reecke
G. N.
Becker
J. W.
Waxdal
J. M.
Wang
J. L.
The covalent and three-dimensional structure of concanavalin A
Proc. Natl. Acad. Sci. U.S.A.
1972
, vol. 
69
 (pg. 
2580
-
2584
)
19
Hardman
K. D.
Ainsworth
C. F.
Structure of concanavalin A at 2.4 Å resolution
Biochemistry
1972
, vol. 
11
 (pg. 
4910
-
4919
)
20
Reecke
G. N.
Becker
J. W.
Edelman
G. M.
The covalent and three-dimensional structure of concanavalin A. IV. Atomic coordinates, hydrogen bonding, and quaternary structure
J. Biol. Chem.
1975
, vol. 
250
 (pg. 
1525
-
1547
)
21
Huet
M.
Claverie
J. M.
Sedimentation studies of the reversible dimer-tetramer transition kinetics of concanavalin A
Biochemistry
1978
, vol. 
17
 (pg. 
236
-
241
)
22
Senear
D. F.
Teller
D. C.
Thermodynamics of concanavalin A dimer–tetramer self-association: sedimentation equilibrium studies
Biochemistry
1981
, vol. 
20
 (pg. 
3076
-
3083
)
23
Senear
D. F.
Teller
D. C.
Effects of saccharide and salt binding on dimer–tetramer equilibrium of concanavalin A
Biochemistry
1981
, vol. 
20
 (pg. 
3083
-
3091
)
24
Calvete
J. J.
Thole
H. H.
Raida
M.
Urbanke
C.
Romero
A.
Grangeiro
T. B.
Ramos
M. V.
Rocha
I. M. A.
Guimarães
F. N.
Cavada
B. S.
Molecular characterization and crystallization of Diocleinae lectins
Biochim. Biophys. Acta
1999
, vol. 
1430
 (pg. 
367
-
375
)
25
Edelman
G. M.
Wang
J. L.
Binding and functional properties of concanavalin A and its derivatives. III. Interactions with indoleacetic acid and other hydrophobic ligands
J. Biol. Chem.
1978
, vol. 
253
 (pg. 
3016
-
3022
)
26
Roberts
D. D.
Goldstein
I. J.
Adenine binding sites of the lectin from lima beans (Phaseolus lunatus)
J. Biol. Chem.
1983
, vol. 
258
 (pg. 
13820
-
13824
)
27
Rozwarski
D. A.
Swami
B. M.
Brewer
C. F.
Sacchettini
J. C.
Crystal structure of the lectin from Dioclea grandiflora complexed with core trimannoside of asparagine-linked carbohydrates
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
32818
-
32825
)
28
Wah
D. A.
Romero
A.
Gallego del Sol
F.
Cavada
B. S.
Ramos
M. V.
Grangeiro
T. B.
Sampaio
A. H.
Calvete
J. J.
Crystal structure of native and Cd/Cd-substituted Dioclea guianensis seed lectin. A novel manganese-binding site and structural basis of dimer-tetramer association
J. Mol. Biol.
2001
, vol. 
310
 (pg. 
885
-
894
)
29
Gallego del Sol
F.
Cavada
B. S.
Calvete
J. J.
Crystal structures of Cratylia floribunda seed lectin at acidic and basic pHs. Insights into the structural basis of the pH-dependent dimer-tetramer transition
J. Struct. Biol.
2007
, vol. 
158
 (pg. 
1
-
9
)
30
Steenkamp
J.
Wiid
I.
Lourens
A.
van Helden
P.
Improved method for DNA extraction from Vitis vinifera
Am. J. Enol. Vitic.
1994
, vol. 
45
 (pg. 
102
-
106
)
31
Frohman
M. A.
Martin
G. R.
Rapid amplification of cDNA ends using nested primers
Techniques
1989
, vol. 
1
 (pg. 
165
-
170
)
32
Siebert
P. D.
Chenchik
A.
Kellogg
D. E.
Lukyanov
K. A.
Lukyanov
S. A.
An improved PCR method for walking in uncloned genomic DNA
Nucleic Acids Res.
1995
, vol. 
23
 (pg. 
1087
-
1088
)
33
Minton
A. P.
Schuster
M.
Laue
T. M.
Conservation of signal: a new algorithm for elimination of the reference concentration as an independently variable parameter in the analysis of sedimentation equilibrium
Modern Analytical Ultracentrifugation
1994
Boston
Birkhauser
(pg. 
81
-
93
)
34
Leslie
A. G. W.
Mosflm User's Guide: Mosflm Version 5.50
1997
Cambridge, U.K.
MRC Laboratory of Molecular Biology
35
Collaborative Computational Project, Number 4
Acta Cryst.
1994
, vol. 
D50
 (pg. 
760
-
763
)
36
Vagin
A.
Teplyakov
A.
MOLREP: an automated program for molecular replacement
J. Appl. Cryst.
1997
, vol. 
30
 (pg. 
1022
-
1025
)
37
Emsley
P.
Cowtan
K.
Coot: model-building tools for molecular graphics
Acta Crystallogr. Sect. D Biol. Crystallogr.
2004
, vol. 
D60
 (pg. 
2126
-
2132
)
38
Carrington
D. M.
Auffret
A.
Hanke
D. E.
Polypeptide ligation occurs during post-translational modification of concanavalin A
Nature
1985
, vol. 
313
 (pg. 
64
-
67
)
39
Grangeiro
T. B.
Schriefer
A.
Calvete
J. J.
Raida
M.
Urbanke
C.
Barral-Netto
M.
Cavada
B. S.
Molecular cloning and characterization of ConBr, the lectin of Canavalia brasiliensis seeds
Eur. J. Biochem.
1997
, vol. 
248
 (pg. 
43
-
48
)
40
Yamauchi
D.
Minamikawa
T.
Structure of the gene encoding concanavalin A from Canavalia gladiata and its expression in Escherichia coli cells
FEBS Lett.
1990
, vol. 
260
 (pg. 
127
-
130
)
41
Chrispeels
M. J.
Hartl
P. M.
Sturm
A.
Faye
L.
Characterization of the endoplasmic reticulum-associated precursor of concanavalin A. Partial amino acid sequence and lectin activity
J. Biol. Chem.
1986
, vol. 
261
 (pg. 
10021
-
10024
)
42
Min
W.
Jones
D. H.
In vitro splicing of concanavalin A is catalyzed by asparaginyl endopeptidase
Nat. Struct. Biol.
1994
, vol. 
1
 (pg. 
502
-
504
)
43
Min
W.
Jones
D. H.
Stability and detection of recombinant pre-pro-concanavalin A after cytoplasmic expression in Escherichia coli
FEBS Lett.
1992
, vol. 
301
 (pg. 
315
-
318
)
44
Dinkturk
H. B.
Dunn
A. J.
Jones
D. H.
Recombinant pre-pro-concanavalin A (jack bean) is stable but of low solubility
J. Biosci.
2001
, vol. 
26
 (pg. 
635
-
640
)
45
Min
W.
Dunn
A. J.
Jones
D. H.
Non-glycosylated recombinant pro-concanavalin A is active without polypeptide cleavage
EMBO J.
1992
, vol. 
11
 (pg. 
1303
-
1307
)
46
Gallego del Sol
F.
Nagano
C. S.
Cavada
B. S.
Calvete
J. J.
The first crystal structure of a Mimosoideae lectin reveals a novel quaternary arrangement of a widespread domain
J. Mol. Biol.
2005
, vol. 
353
 (pg. 
574
-
583
)
47
Gallego del Sol
F.
Estudios Estructurales de Lectinas Vegetales
Ph.D. Thesis
2006
University of Valencia
48
Brinda
K. V.
Mitra
N.
Surolia
A.
Vishveshwara
S.
Determinants of quaternary association in legume lectins
Protein Sci.
2004
, vol. 
13
 (pg. 
1735
-
1749
)

Author notes

The nucleotide sequences of the synthetic gene coding for the α-chains of the lectins from Dioclea guianensis (termed r-αDguia) and Dioclea grandiflora (termed r-αDGL) have been deposited with the EMBL Nucleotide Sequence Database and are accessible under accession codes AM701772 and AM701773 respectively.