CBMs (carbohydrate-binding modules) are a class of polypeptides usually associated with carbohydrate-active enzymatic sites. We have characterized a new member of the CBM40 family, coded from a section of the gene NanI from Clostridium perfringens. Glycan arrays revealed its preference towards α(2,3)-linked sialosides, which was confirmed and quantified by calorimetric studies. The CBM40 binds to α(2,3)-sialyl-lactose with a Kd of ∼30 μM, the highest affinity value for this class of proteins. Inspired by lectins' structure and their arrangement as multimeric proteins, we have engineered a dimeric form of the CBM, and using SPR (surface plasmon resonance) we have observed 6–11-fold binding increases due to the avidity affect. The structures of the CBM, resolved by X-ray crystallography, in complex with α(2,3)- or α(2,6)-sialyl-lactose explain its binding specificity and unusually strong binding.

INTRODUCTION

Sialic acids are the most abundant terminal carbohydrate residues present on glycoconjugates at the surface of eukaryotic cells [1,2]. The N-acetylneuraminic acid (Neu5Ac) monosaccharide is the most common form of sialic acid on mammalian cells. Sialylation is required for the structural stabilization of cells [3] and is especially important in early mammalian development [4,5]. In cancer, altered surface sialylation is a common hallmark of the malignant phenotype of tumour cells [6]. The analysis of total serum and lipid-bound Neu5Ac is valuable in cancer staging, prognosis and progression, and elevated levels have been associated with cancer development [7,8]. An increase in the negative charges associated with Neu5Ac also contributes to the cell mobility by helping dispersion from the main tumour, influencing the metastatic potential of those cells [9].

Owing to the importance of Neu5Ac, specific receptors for this epitope are of interest as biomarkers. Currently, only a handful of Neu5Ac binders are commercially available [10]. This includes lectins purified from plants, such as WGA (wheatgerm agglutinin), MAA (Maackia amurensis agglutinin) and SNA-I (Sambucus nigra agglutinin I) [11,12], mushrooms, such as PSL-I (Polyporus squamosus lectin I) [13,14], or invertebrates such as LFA (Limax flavus agglutinin) [15]. Whereas SNA-I and PSL-I are specific for α2,6-linked Neu5Ac [namely terminal 6′-SLN (6′-sialyl-N-acetyl-lactosamine)], the other lectins have a broader range of interactions, for example displaying cross reactivities with N-acetylglucosamine (GlcNAc) (WGA and LFA) or with 3′-sulfolactose (MAA). As a consequence, the analysis of complex samples requires the use of several of these lectins. Additionally, these commercial lectins come with restrictions that arise from seasonal depletion of the natural sources, undesired protein glycosylation or batch inconsistency. Recombinant proteins have proven to be a valid and effective alternative, compensating for the aforementioned complications [16].

Sialidases, a group of enzymes that cleave sialosides, are secreted by microbial organisms and mammalian cells. These include gut bacteria, either commensals or pathogens [17,18], viruses such as influenza [19] and human cells [20]. The catalytic domain of these enzymes is often linked to one or more CBMs (carbohydrate-binding modules) that reversibly binds the glycan substrate to increase its local concentration, consequently improving the catalytic efficiency of the enzyme [21]. CBMs are small monovalent domains that can generally be easily expressed in recombinant systems [22] and may provide an interesting alternative to lectins. Developed by pathogens for binding and invading human tissues, these receptors could provide specific Neu5Ac binders. In addition, CBMs can be easily produced, and their low-affinity binding can be compensated for through avidity by fusing multiple CBMs together [23].

CBMs are organized into 73 families on the basis of their amino acid sequence similarity (http://www.cazy.org) [24]. Most interact with plant cell wall glycans, but five CBM families have been categorized as mammalian glycan-binding domains: CBM32, CBM40, CBM47, CBM51 and CBM57. Within these, the members of CBM40 are known to bind to Neu5Ac, which was first verified with the CBM located at the N-terminus of GH33 (glycoside hydrolase 33) sialidase from Vibrio cholerae [25]. At present, there are only six structurally characterized CBM40s which are associated with the sialidases from Clostridium perfringens NanJ [26], V. cholerae NanH [23], Macrobdella decora NanL [27] and Streptococcus pneumoniae NanA [28], NanB [29] and NanC [30]. They all share a conserved β-sandwich fold consisting of two antiparallel β-sheets, with the exception of V. cholerae NanH which shows a different architecture, which led researchers to suggest its integration into a new CBM40 subfamily [26].

In the present paper, we report the successful characterization of the CBM40 from C. perfringens (ATCC 13124) which is associated with NanI [31], for which only the catalytic domain has been structurally characterized [32]. The recombinant CBM40_NanI demonstrates unusual micromolar affinity towards sialylated oligosaccharides. The structural analysis of the complexes with 3′-SL (3′-sialyl-lactose) and 6′-SL (6′-sialyl-lactose) provides insights into the molecular details responsible for this affinity. A pseudo-dimeric version of the CBM40 was also engineered so as to enhance the affinity from multiple interactions between the CBM and glycans. The present study describes a new probe towards sialoconjugates, expanding the repertoire of sialoside binders.

EXPERIMENTAL

Materials

Genomic DNA of C. perfringens ATCC® 13124™ was bought from the A.T.C.C. (#13124D-5™). 3′-SL sodium salt (CAS #128596-80-5) and 6′-SL (#35890-38-1) were purchased from Carbosynth. 3′-SLN (3′-sialyl-N-acetyl-lactosamine) (#81693-22-3) and 6′-SLN (#174757-71-2) were from Dextra UK and sialic acid (#131-48-6) was from AppliChem. Biotinylated 3′-SL (#0060-BM) and biotinylated 6′-SL (#0063-BM) were obtained from Lectinity.

Propargyl sialic acid synthesis

All chemical reagents were purchased from Aldrich or Acros. Moisture-sensitive reactions were performed under an argon atmosphere by using oven-dried glassware, and reactions were monitored by TLC using silica gel 60 F254 pre-coated plates (Merck). Spots were inspected by UV light and visualized by charring with 10% H2SO4 in ethanol. Silica gel 60 (0.063–0.2 mm or 70–230 mesh; Merck) was used for column chromatography. 1H and 13C NMR spectra were recorded on Bruker Avance 400 MHz or Bruker Avance III 500 MHz spectrometers and chemical shifts (δ) were reported in p.p.m. Spectra were referenced to the residual proton solvent peaks relative to the signal of CDCl3 (δ 7.27 and 77.0 p.p.m. for 1H and 13C) and 2H2O (4.79 p.p.m. for 1H), assignments were made by gCOSY (gradient-selected COSY) and gHMQC (gradient-selected HMQC) experiments.

A route for the synthesis of the propargyl sialic acid (2-propynyl 5-acetamido-3,5-dideoxy-D-α-D-galacto-2-nonulopyranosidonic acid) starting from N-acetylneuraminic acid is shown in Scheme 1. All reaction conditions, mass and NMR spectra and their assignments are in the Supplementary Online Data (SI 1).

Synthesis of 2-propynyl 5-acetamido-3,5-dideoxy-D-α-D-galacto-2-nonulopyranosidonic acid 4: a) MeOH, Amberlite 120 H+; b) AcCl, HCl(g); c) Propargyl alcohol, AgOTf, 4 Å MS; d) NaOMe, MeOH; e) LiOH, H2O.

Gene cloning

The following oligonucleotide primers were used: 5′-GAC GAC GGA TCC CAT GTT AAG TTC ACT AGG AGA ATA TAA GGA TAT-3′ (45-mer) and 5′-TCC CAT CTC GAG TTA TTT TGT CTC TCC AGT CTT ACT AAG TAA ATA ATC ATC-3′ (51-mer) for the construction of pNanI. They were designed as to include BamHI and XhoI restriction sites (underlined sequences) in the CBM40_NanI gene sequence. PCRs were done using the Q5 Hot Start High Fidelity polymerase (New England Biolabs) and genomic DNA from Clostridium perfringens strain ATCC® 13124TM as a template. The same cloning protocol was followed for pNanJ, using the following oligonucleotide primers 5′-AC TAC GGA TCC AGG GTA AAT ATA ACA GGT GAT T -3′ (33-mer) and 5′-G CAT CTC GAG TTT AGT TTC TCC TGT TTT TCT TA-3′ (36-mer) which contain SacI and XhoI restrictions sites respectively. To engineer a CBM40_NanI with a multivalent presentation, we changed the 5′- and 3′-termini of the gene sequences to integrate 36 additional codons (representing a 12 amino acid long linker that will be present between two CBM copies). The PCRs were done using the oligonucleotides 1-For: 5′-GAC GAC GGA TCC CAT GTT AAG TCC ACT AGG ATA TAA GGA TAT-3′ (45-mer, BamHI); 1-Rev: 5′-G CAC GAG CTC GCT GCC GTT CAG CGC TTT TGT CTC TCC AGT-3′ (40-mer, SacI); 2-For: 5′-C GTC GAG CTC GGC AGC GGC AGC GGC TTA AGT TCA CTA GGA-3′ (40-mer, SacI), 5′-TCC CAT CTC GAG TTA TTT TGT CTC TCC AGT CTT ACT AAG TAA ATA ATC ATC-3′ (51-mer, XhoI). After the two sequences were made, the digestion by the same endonuclease allowed the ligation of the two CBM40_NanI copies in tandem. All the generated inserts were subjected to endonuclease treatment and introduced in the multiple cloning region of the pET45b(+) vector, yielding to pNanI, pNanJ and p2NanI plasmids that encode for a copy of Cp CBM40_NanI, Cp CBM40_NanJ or Cp di-CBM40_NanI, respectively, fused to a polyhistidine tag at the N-terminus. The constructs were verified by DNA sequencing (Genewiz, US or Eurofins, EU). The DNA of the positive clones was amplified in Escherichia coli XL1 blue cells and then used to transform E. coli BL21(DE3) cells for protein production.

Protein expression and purification

E. coli BL21(DE3) cells transformed with plasmids pNanI, pNanJ or p2NanI were cultured in terrific broth medium at 37°C. When reaching a D600 of 0.5–0.7, the cultures were moved to a 16°C incubator and IPTG was added to a final concentration of 1 mM. After overnight incubation, cells were pelleted down, washed and resuspended in ice-cold PBS (pH 7.4). The cells were lysed by sonication on ice three times at 20% intensity with 0.5 cycle for 2 min or with a Cell Disruption System (Constant Systems) at 1900 bars (1 bar=100 kPa). After centrifugation at 50000 g for 30 min at 4°C, the supernatants were submitted to nickel-affinity chromatography. After being allowed to bind the matrix (Ni2+-nitrilotriacetate–agarose; Qiagen), the proteins were washed with 50 mM NaH2PO4·H2O, pH 8.0, 300 mM NaCl and 10 mM imidazole, and eluted using the same buffer but with an increased 250 mM imidazole concentration. The eluted fractions were filtered, checked for absorbance at 280 nm and concentrated by centrifugation using Vivaspin (Startorius) with a 10 kDa molecular-mass cut-off filter, before being purified further by size-exclusion chromatography. A Superdex 75 prep grade column was equilibrated with PBS (pH 7.4) and used for the estimation of aggregate states and molecular masses of the purified proteins.

A calibration curve for molecular size estimation created with cytochrome c, myoglobin, ovalbumin and BSA, before the experiments, confirmed the expected molecular mass values. Purified proteins were run in denaturing protein gels in a 12% (w/v) acrylamide matrix under alkaline buffer conditions (Tris/HCl, pH 8.8).

Protein estimation

Protein concentrations were determined with a Nanodrop spectrometer ND2000 (Ozyme) with absorbance readings at 280 nm wavelengths, using molar absorption coefficient values of 11920, 21890 and 32780 M−1·cm−1 for C. perfringens CBM40_NanJ, C. perfringens CBM40_NanI and C. perfringens di-CBM40_NanI respectively, as well as their corresponding molecular masses of 23.69, 24.62 and 46.55 kDa (values for calculations that include the N-terminally tagged amino acids).

Thermostability

To examine their stability, protein solutions of C. perfringens CBM40_NanI, C. perfringens CBM40_NanJ and C. perfringens di-CBM40_NanI were assayed using a TSA (thermal shift assay), adapting a protocol described previously [33]. Samples were diluted to a final protein concentration of 0.1 mg·ml−1 in the chosen buffer and 5× SYPRO Orange (Molecular Probes/Invitrogen) was added to a final volume of 25 μl. The samples were subjected to thermal denaturation using a Real Time PCR machine (Mini Opticon, Bio-Rad Laboratories) with a temperature gradient starting at 25°C and rising to 100°C at a heating rate of 1°C/min. Protein unfolding was followed by the increase in the fluorescence values given by the SYPRO Orange probe. Different buffer effects were analysed in the pH range of 3–10.

Crystallization and data collection

Solutions of CBM40_NanI (5.5 mg·ml−1) in PBS where sent to the High Throughput Crystallization laboratory (HTXlab) in Grenoble, France, to screen for crystallization conditions. The protein/ligand mixture was tested against the JCSG (Qiagen), Wizard I and II (Rigaku Reagents), PACT (Qiagen), plate 1 (Qiagen), plates 4 and 5 (Hampton) screens. Hits were obtained with the use of the Clear Strategy™ Screens, (Molecular Dimensions) at pH 4.5. Crystals of CBM40_NanI with 3′-SL appeared in the form of plates after several days in CSSI-12 containing 0.1 M sodium acetate, pH 4.5, 0.2 M calcium acetate and 8% of both PEG550 MME (monomethyl ether) and PEG20000. In the same time period, parallelepiped crystals of CBM40_NanI in complex with 6′-SL grew in 0.1 M sodium acetate, pH 4.5, 0.3 M sodium acetate and 25% PEG2000 MME. Afterwards, larger crystals were then grown with home-made solutions. All diffraction experiment datasets were collected at the European Synchrotron Radiation Facility (ESRF), Grenoble, France. Final data for CBM/3′-SL were collected at 1.9 Å (1 Å=0.1 nm) on the BM30A beamline using an ADSC Q315r detector whereas the CBM–6′-SL complex was obtained at 2.0 Å on the ID29 beamline using a Pilatus detector. Diffraction images were integrated using program XDS [34,35] and all further processing was carried out using the CCP4 program suite [36].

Structure solution and refinement

Co-ordinates of molecule A of the CBM40 from C. perfringens ATCC® 13124™ NanJ (PDB code 2V73) were used as a model for molecular replacement with PHASER [37] in a search for six monomers per asymmetric units for the CBM40–6′-SL complex. The coordinates of molecule A of the CBM40–3′-SL were used to search for the three copies of the CBM40–6′-SL complex. A total of 5% of the observations were set aside for cross-validation analysis, and hydrogen atoms were added in their riding positions and used for geometry and structure factor calculations. The initial model was optimized using ARP/wARP [38] before restrained maximum likelihood refinement with REFMAC 5.8 [39] iterated with manual rebuilding in Coot [40]. Incorporation of the ligand was performed after inspection of the 2FoFc weighted maps. Water molecules, introduced automatically using Coot, were inspected manually. The quality of the models was assessed using the PDB validation server (http://wwpdb-validation.wwpdb.org/validservice/) and co-ordinates were deposited in the PDB under codes 5FRE and 5FRA.

Haemagglutination and minimal concentration inhibition assays

Haemagglutination activity for C. perfringens CBM40_NanI and C. perfringens di-CBM40_NanI was determined using 2-fold serial dilutions of the protein with rabbit erythrocytes (Biomérieux). In those tests, 25 μl of the serial dilutions of the protein were incubated with 25 μl of 2% rabbit erythrocyte suspensions in 100 mM NaCl for 30 min at 37°C and 30 min at room temperature. After identifying the highest dilution of the protein that exhibited haemagglutination, a concentration 4-fold higher was used in the subsequent MIC (minimal inhibition concentration) experiments. Serial dilutions of 12.5 μl of different saccharide solutions were added to the 12.5 μl solutions of protein of identical concentrations. These 25 μl mixtures were incubated with 25 μl of 2% rabbit erythrocyte suspensions, following the same development steps as above.

Glycan array

C. perfringens CBM40_NanI and C. perfringens di-CBM40_NanI were labelled with Alexa Fluor® 488 (Molecular Probes/Invitrogen) using the manufacturer's protocols and sent to the Consortium for Functional Glycomics (CFG). Binding specificity was determined using the printed mammalian glycan array version 5.2, which contains 609 glycans [41]. The scanner response is linear to a maximum of ∼50000 RFU (relative fluorescent units). Both proteins were incubated with the arrays at 200 μg·ml−1. The extracted data was then transformed to give Z scores for each glycan on the array (Z=x-mean/S.D.). A threshold of Z=1.645 (P ≤ 0.05) was used to establish significance.

Isothermal titration calorimetry (ITC)

ITC experiments were performed using a MicroCal ITC200 microcalorimeter (Malvern Instruments). All titrations were carried out in PBS, pH 7.4, at 25°C. Aliquots of 10 μl of each carbohydrate dissolved in the same buffer were added to the protein solution present in the calorimeter cell at 2 min intervals. For the different experiments, protein concentrations were prepared in solution with concentrations that varied from 100 to 415 μM, and the carbohydrate samples ranged from 2.1 to 6.0 mM. Data were fitted using MicroCal Origin 7 software, following standard procedures. This yielded the stoichiometry (N), the association constant (Ka) and enthalpy of binding (ΔH). Other thermodynamic parameters (changes in free energy, ΔG, and entropy, ΔS), were calculated from the equation ΔGHTΔS=−RTlnKa, with R=8.314 J·mol−1·K−1.

Surface plasmon resonance (SPR)

SPR experiments were performed using a Biacore X100 biosensor instrument (GE Healthcare) at 25°C. Biotinylated 3′-SL and 6′-SL were immobilized on CM5 chips (GE Healthcare) that were coated previously with streptavidin, following the same protocol as described previously [42]. Each monovalent biotinylated sugar, 3′-SL or 6′-SL, was diluted to 1 μg·ml−1 in HBS-T (Hepes-buffered saline, pH 7.4, with 0.05% Tween 20) before being injected into one of the flow cells of the chip. Low immobilization levels of 94 and 64 response units were obtained for 3′-SL and 6′-SL respectively. A reference surface was always present in flow cell 1, thus allowing for the subtraction of bulk effects and non-specific interactions with streptavidin. The running buffer consisted of the same HBS-T. The purified proteins were injected over the flow cell surface at 30 μl·min−1 in a series of 2-fold dilutions. The dissociation of this analyte was achieved by passing running buffer for 4–6 min. Surfaces were regenerated with one or two consecutive 30-s injections of 100 mM Neu5Ac, also at 30 μl·min−1. The information on the affinity was determined by assuming Langmuir 1:1 binding, using the BIAevaluation software.

RESULTS AND DISCUSSION

Sequences identified as belonging to the family 40 CBM were selected from the CAZy (Carbohydrate-Active enZYmes) database [24] and used to build a phylogenetic tree (http://phylogeny.lirmm.fr). The topology of the tree (Supplementary Online Data SI 2), shows the CBM40s cluster into one main group with few outliers. The CBM from C. perfringens NanI clusters with the other CBM from the same organism, C. perfringens CBM40_NanJ, with a phylogenetic similarity of 0.942. As previously noted by other authors, the CBM of V. cholerae NanH is more distantly related to the other CBM40s, among those with validated Neu5Ac specificity and available structural information [26]. The BLAST protein alignment between the C. perfringens CBM40_NanI sequence and the other structurally characterized CBM40s, i.e. C. perfringens NanJ (PDB code 2V73), S. pneumoniae NanA (PDB code 4C1W), NanB (PDB code 2VW0) and NanC (PDB code 4YZ5) and M. decora NanL (PDB code 1SLI), shows that CBM40_NanI shares 56% identity with the CBM40_NanJ from the same organism (C. perfringens), 23%, 30% and 27% homology values with the S. pneumoniae proteins, and 33% with the CBM of M. decora leech. C. perfringens CBM40 NanI therefore appears sufficiently different from previously characterized CBMs of family 40 to substantiate a full characterization.

After cloning of genomic DNA of C. perfringens ATCC® 13124™, recombinant C. perfringens CBM40_NanI was purified with a yield of approximately 10 mg·l−1. The C. perfringens CBM40_NanJ was produced under the same conditions for comparison. The expressed proteins appear as one strong band on SDS/PAGE gels, with apparent molecular masses close to the 25 kDa marker, in agreement with the theoretical values of 23.7 and 24.6 kDa for C. perfringens CBM40_NanJ and C. perfringens CBM40_NanI respectively. The proteins were separated from aggregates by gel-exclusion chromatography and collected as a single peak with a retention time representative of the monomers (Supplementary Online Data SI 3). The thermal stability of the C. perfringens CBM40_NanI and C. perfringens CBM40_NanJ was assessed by TSAs. The highest stability was obtained in PBS buffer at pH 7, with melting temperatures of 54°C for C. perfringens CBM40_NanJ and 65°C for C. perfringens CBM40_NanI. One issue with the use of CBMs as glycan probes is their low affinities. One method to overcome this is to engineer constructs containing multiple CBM domains [23]. Thus a divalent version of the protein was engineered. The pseudo-dimeric C. perfringens CBM40_NanI has a 12-amino-acid-long linker (ALNGSELGSGSG) inserted between the two copies to provide flexibility between the binding modules. Using an identical expression protocol, a yield of ∼5 mg·l−1 purified di-CBM40_NanI was obtained. The expected band for a 46 kDa protein was seen in the SDS/PAGE gel (Supplementary Online Data SI 3). The engineered C. perfringens di-CBM40_NanI was stable in PBS (pH 7.0), with an observed Tm of 66°C (Supplementary Online Data SI 4).

Characterization of binding specificity demonstrates strong preference for 3-linked sialic acids

An initial evaluation of glycan binding was performed by inhibition of haemagglutination using rabbit erythrocytes. As expected, haemagglutinating activity was not observed with monomeric C. perfringens CBM40_NanI as agglutination requires a multivalent binding to cross-link the cells. In contrast, C. perfringens di-CBM40_NanI showed positive haemagglutination down to a concentration of 50 μg·ml−1. Among the glycans tested as haemagglutination inhibitors, 3′-SL was the strongest inhibitor, with an MIC of 0.78 mM. 6′-SL was also able to inhibit haemagglutination with an MIC of 1.56 mM. Lactose, galactose, glucose, mannose and fucose (tested up to 50 mM) had no inhibitory effect (Supplementary Online Data SI 5).

To gain further insight into the binding motifs of C. perfringens CBM40_NanI, both monomeric and dimeric CBMs were tested on the CFG glycan microarray version 5.2. This array contains a panel of 609 mammalian glycans, including α2,3-, α2,6- and α2,8-sialosides [43]. Monomeric C. perfringens CBM40_NanI gave only a very weak signal with clear binding to only a few epitopes bearing α2,3-sialic acid on a type II N-acetyl-lactosaminyl core (Neu5Acα2,3Galβ1,4GlcNAc) (Supplementary Online Data SI 6). Previous assays involving monomeric C. perfringens CBM40_NanJ were also unsuccessful due to low affinity [26]. The signal-to-noise ratio was considerably improved when assaying the engineered dimeric protein. The C. perfringens di-CBM40_NanI screen confirms the specificity of this binding domain towards α2,3-sialosides (Figure 1). Of the 27 glycans that had significant binding (P ≤ 0.05 as determined by Z-score analysis) on the array, 24 contained an α2,3-sialoside. The remaining three were an anomalous chitin epitope, sialic acid bound directly to a linker and a single biantennary α2,6-epitope detected. Both N-glycolyl- and 9-O-acetyl-containing α2,3-sialosides acids were tolerated. Binding was enhanced by an additional nearby negative charge (sulfation or sialylation).

Glycan array profile of C. perfringens di-CBM40_NanI

Figure 1
Glycan array profile of C. perfringens di-CBM40_NanI

The Alexa Fluor® 488-fused protein was tested against a panel of 609 glycans. Some examples of α(2,3)-linked sialosides are shown (more details at CFG request number 16976, and in Supplementary Online Data file CFG_diCBM40_NanI.xls).

Figure 1
Glycan array profile of C. perfringens di-CBM40_NanI

The Alexa Fluor® 488-fused protein was tested against a panel of 609 glycans. Some examples of α(2,3)-linked sialosides are shown (more details at CFG request number 16976, and in Supplementary Online Data file CFG_diCBM40_NanI.xls).

C. perfringens CBM40_NanI displays unusually strong affinity for sialylated oligosaccharides

The glycan-binding activities of C. perfringens CBM40_NanI and di-CBM40_NanI were confirmed by ITC at 25°C. All interactions showed exothermic behaviours with a steep decrease in the exothermic heat of binding until saturation was achieved (Figure 2). The stoichiometry values (N) for sialylated compounds titrated in C. perfringens CBM40_NanI and di-CBM40_NanI are close to 1 and 2 respectively, in agreement with the expected presence of one or two binding sites (Table 1). Both proteins have similar behaviour with strong affinity for 3′-SL (Kd ∼30 μM) and weaker affinity for 6′-SL (Kd ∼60 μM). Values in the same data range were obtained for 3′- and 6′-SLN (Supplementary Online Data SI 7). Free Neu5Ac monosaccharide could not be assayed directly due to pH variations during the titration. To assay the monosaccharide, a derivative of Neu5Ac with a propargyl aglycon (SiaOPr) was synthesized and assayed for binding. The CBMs showed weaker binding (Kd values of 170 and 189 μM for monomeric and dimeric CBM respectively) to the monosaccharide conjugate than to the sialylated trisaccharides. To compare binding affinities directly, ITC experiments were also performed for C. perfringens CBM40_NanJ. As expected, lower binding affinities were obtained with Kd values of 1.48 and 2.67 mM for 3′-SL and 6′-SL respectively (Supplementary Online Data SI 7), confirming the previously published estimations of low affinity [26].

Table 1
Thermodynamics of binding of C. perfringens CBM40_NanI, C. perfringens di-CBM40_NanI C. perfringens CBM40_NanJ to different sialylated ligands, as determined by ITC at 25°C

ND, not determined. 1 kcal=4.184 kJ.

(a) 
Protein [Protein] (μM) Ligand [Ligand] (mM) N Ka (×104 M−1Kd (μM) ΔH (kcal/mol) TΔS (kcal/mol) ΔG (kcal/mol) 
C. perfringens CBM40_NanI 400 3′-SL 5.0 0.86±0.01 3.16±0.09 32 −10.49±0.14 −4.35 −6.14 
 400 6′-SL 3.6 0.95±0.01 1.48±0.05 68 −8.69±0.05 −3.01 −5.68 
 50 3′-SLN 0.9 0.98±0.05 2.68±0.20 37 −9.38±0.06 −3.34 −6.04 
 50 6′-SLN 0.8 0.92±0.08 2.23±0.25 49 −8.81±0.10 −2.88 −5.93 
 100 SiaOPr 5.0 1.00* 0.59±0.05 170 −19.95±0.05 −14.81 −5.14 
C. perfringens di-CBM40_NanI 185 3′-SL 5.0 1.94±0.01 2.84±0.08 35 −11.20±0.01 −5.16 −6.07 
 185 6′-SL 3.6 1.88±0.01 1.71±0.06 59 −9.74±0.01 −3.96 −5.78 
 50 SiaOPr 5.0 2.00* 0.51±0.06 196 ND ND ND 
(b) 
Protein [Protein] (μM) Ligand [Ligand] (mM) N Ka (×103 M−1Kd (mM) ΔH (kcal/mol) TΔS (kcal/mol) ΔG (kcal/mol) 
 
C. perfringens CBM40_NanJ 100 3′-SL 10.0 1.00* 0.67±0.04 1.48 −1.51±0.05 2.35 −3.86 
 100 6′-SL 10.0 1.00* 0.44±0.06 2.23 −2.72±0.27 0.89 −3.61 
(a) 
Protein [Protein] (μM) Ligand [Ligand] (mM) N Ka (×104 M−1Kd (μM) ΔH (kcal/mol) TΔS (kcal/mol) ΔG (kcal/mol) 
C. perfringens CBM40_NanI 400 3′-SL 5.0 0.86±0.01 3.16±0.09 32 −10.49±0.14 −4.35 −6.14 
 400 6′-SL 3.6 0.95±0.01 1.48±0.05 68 −8.69±0.05 −3.01 −5.68 
 50 3′-SLN 0.9 0.98±0.05 2.68±0.20 37 −9.38±0.06 −3.34 −6.04 
 50 6′-SLN 0.8 0.92±0.08 2.23±0.25 49 −8.81±0.10 −2.88 −5.93 
 100 SiaOPr 5.0 1.00* 0.59±0.05 170 −19.95±0.05 −14.81 −5.14 
C. perfringens di-CBM40_NanI 185 3′-SL 5.0 1.94±0.01 2.84±0.08 35 −11.20±0.01 −5.16 −6.07 
 185 6′-SL 3.6 1.88±0.01 1.71±0.06 59 −9.74±0.01 −3.96 −5.78 
 50 SiaOPr 5.0 2.00* 0.51±0.06 196 ND ND ND 
(b) 
Protein [Protein] (μM) Ligand [Ligand] (mM) N Ka (×103 M−1Kd (mM) ΔH (kcal/mol) TΔS (kcal/mol) ΔG (kcal/mol) 
 
C. perfringens CBM40_NanJ 100 3′-SL 10.0 1.00* 0.67±0.04 1.48 −1.51±0.05 2.35 −3.86 
 100 6′-SL 10.0 1.00* 0.44±0.06 2.23 −2.72±0.27 0.89 −3.61 

ITC isotherms showing the binding of (A) CBM40_NanI to 3′-SL and 6′-SL, (B) di-CBM40_NanI to 3′-SL and 6′-SL, and (C) CBM40_NanI to SiaOPr

The binding interactions were also analysed by SPR spectroscopy. Commercially available monomeric biotinylated 3′-SL and 6′-SL were immobilized on a streptavidin-coated sensorchip to obtain high-density coverage. The sensorgrams show a typical shape for fast association and dissociation events after the injection of increasing concentrations of C. perfringens CBM40_NanI or C. perfringens di-CBM40_NanI. The binding kinetics of the experiments were too fast to be reliably measured by SPR, and steady-state analysis was performed to evaluate the dissociation constants (Supplementary Online Data SI 8). The data confirmed the stronger binding of C. perfringens CBM40_NanI and C. perfringens di-CBM40_NanI with the 3′-SL- compared with the 6′-SL-functionalized surface. The monomeric CBM showed a Kd of 14.4 μM for 3′-SL, whereas 6′-SL appears to bind with 2.5-fold lower affinity, with a Kd of 36 μM, in agreement with ITC data. For the C. perfringens di-CBM40, an increase of more than one order of magnitude in affinity (11-fold) is observed for the interaction with 3′-SL and a 6-fold increase for 6′-SL, resulting in affinities of 1.3 and 6.1 μM respectively. Experiments conducted with C. perfringens CBM40_NanJ did not yield sensorgrams of sufficient quality to evaluate, probably due to the low affinity of C. perfringens CBM40_NanJ–sialoside interactions.

Structural basis for specificity and affinity

To gain insight into the structural basis for sialoside binding of C. perfringens CBM40_NanI, crystal structures were obtained. An original screening of 600 conditions, using a high-throughput crystallization robot, led to isolation of thin crystals that diffracted at low resolution. Upon optimization with the Crystal Clear Strategy Screen (Molecular Dimensions), bigger and thicker crystals of C. perfringens CBM40_NanI complexed with 3′-SL and 6′-SL were obtained. The protein crystalized in P21 space group for both complexes, with three and six independent monomers in the asymmetric unit for crystals of the CBM40_NanI complexed with 3′-SL or 6′-SL respectively (Supplementary Online Data SI 9). The structure of the CBM–6′-SL was solved using PDB code 2V73 (C. perfringens CBM40_NanJ) as a model, and subsequently used to solve the structure of the C. perfringens CBM40_NanI–3′-SL. The monomers superimposed with RMSD values between 0.12 and 0.16 Å for C. perfringens CBM40–3′SL and between 0.15 and 0.20 Å for C. perfringens CBM40_NanI–6′-SL.

The C. perfringens CBM40_NanI adopts the characteristic β-sandwich fold of CBM40 members, formed by two β-sheets of five and six antiparallel β-strands and two α-helices, one within the sheets and another at the C-terminus, packed against β-strand 7. The carbohydrate-recognition site is located on the concave face of the five-stranded β-sheet, in a shallow depression, in which the ring of the sialic acid lies parallel to the surface of the protein (Figure 3,Scheme 1A). Examination of the electron density clearly allowed the identification of the sialic acid and galactose residues in all binding sites in the CBM–3′-SL complex and in half of the molecules in the 6′-SL, whereas only the Neu5Ac moiety could be defined with confidence for the other sites. The glucose moieties could not be modelled as they were too disordered.

3D structure of CBM40 from Clostridium perfringens NanI in complex with 3′-SL and 6′-SL and sequence comparison with characterized CBMs from the same family

Figure 3
3D structure of CBM40 from Clostridium perfringens NanI in complex with 3′-SL and 6′-SL and sequence comparison with characterized CBMs from the same family

(A) Structure of the CBM40 from C. perfringens ATCC® 13124™ NanI in complex with 3′-SL and 6′-SL (at the left and right respectively). Ligands are shown with a 2FoFc density map contoured at 1.0 σ. (B and C) Close-up view of the binding domain occupied by 3′-SL and 6′-SL. (D) Comparison of the CBM40 of C. perfringens NanI with structurally related structures. C. perfringens NanI (in red, boxed), C. perfringens NanJ (in cyan), S. pneumoniae NanA (in yellow), S. pneumoniae NanB (in green), S. pneumoniae NanC (in brown) and M. decora NanL (in purple). The loop 5–6 in CBM40_NanI is identified with an arrow. (E) Multiple amino acid sequence alignment between CBM40_NanI (shown as NanI) and the CBM40s from C. perfringens ATCC® 13124™ NanJ, S. pneumoniae NanA, NanB and NanC, and M. decora NanL. The secondary-structure regions of the CBM40_NanI are identified on top. Shaded residues show the conservation between the proteins. The CBM40_NanI amino acids making key contacts with the sialic acid are identified with blue triangles. The region between β-strands 5 and 6 is boxed within red limits and Gln85 is marked with a red star due to its particular importance as part of the binding domain. The sequence of V. cholerae NanH is not shown due to its poor alignment (Supplementary Online Data SI 8). The Figure was created with ESPript 3.0 [53].

Figure 3
3D structure of CBM40 from Clostridium perfringens NanI in complex with 3′-SL and 6′-SL and sequence comparison with characterized CBMs from the same family

(A) Structure of the CBM40 from C. perfringens ATCC® 13124™ NanI in complex with 3′-SL and 6′-SL (at the left and right respectively). Ligands are shown with a 2FoFc density map contoured at 1.0 σ. (B and C) Close-up view of the binding domain occupied by 3′-SL and 6′-SL. (D) Comparison of the CBM40 of C. perfringens NanI with structurally related structures. C. perfringens NanI (in red, boxed), C. perfringens NanJ (in cyan), S. pneumoniae NanA (in yellow), S. pneumoniae NanB (in green), S. pneumoniae NanC (in brown) and M. decora NanL (in purple). The loop 5–6 in CBM40_NanI is identified with an arrow. (E) Multiple amino acid sequence alignment between CBM40_NanI (shown as NanI) and the CBM40s from C. perfringens ATCC® 13124™ NanJ, S. pneumoniae NanA, NanB and NanC, and M. decora NanL. The secondary-structure regions of the CBM40_NanI are identified on top. Shaded residues show the conservation between the proteins. The CBM40_NanI amino acids making key contacts with the sialic acid are identified with blue triangles. The region between β-strands 5 and 6 is boxed within red limits and Gln85 is marked with a red star due to its particular importance as part of the binding domain. The sequence of V. cholerae NanH is not shown due to its poor alignment (Supplementary Online Data SI 8). The Figure was created with ESPript 3.0 [53].

The majority of the interactions between the protein and the ligand involve the sialic acid moiety. In both complexes (Figures 3B and 3C), Ile46, Tyr65 and His93 side chains form a hydrophobic pocket that accommodates the methyl of the N-acetyl group. Arg76 and Arg153 make electrostatic interactions with the carboxylate group. Asn158 establishes hydrogen bonds with the glycerol moiety. The hydroxy group of Tyr160 makes hydrogen bonds with the carboxylate group and the nitrogen of the N-acetyl group. Glu74 is involved in two hydrogen bonds with the nitrogen of the N-acetyl group and the O4 hydroxy group, and the NH1 and NH2 atoms of Arg153 make hydrogen bonds with the carboxylate group. The carboxylate makes water-mediated interactions with the main-chain nitrogen of Asn158 and the NH2 atom of Arg76. Comparison between the binding sites of C. perfringens CBM40_NanI and NanJ show that the amino acids making direct contacts with the ligand are conserved in both structures (Supplementary Online Data SI 10).

In the complex with 3′SL, the galactose does not interact directly with the protein, but is involved in a dense network of water-bridged hydrogen bonds. Six water molecules are conserved in the three independent monomers where three of them are directly involved in stabilizing the galactose residue (Supplementary Online Data SI 11). These three water molecules are responsible for locking the conformation of the galactosyl moiety and creating a further network with other waters, bridging both inter-amino acid and intra-ligand interactions. Gln85 is a key amino acid in this network, interacting with two water molecules which bridge with the O4 hydroxy group of the galactosyl moiety. Lys58 NH makes another water-mediated hydrogen bond with the OH6 group of the galactose. Also, Ser157 interacts with a chain of water molecules connecting to the carboxylate of the Neu5Ac and the O4 hydroxy group of the galactose.

In the CBM–6′-SL complex, two of the three resolved 6′-SL molecules have the same conformation, whereas the third shows different dihedrals angles (Supplementary Online Data SI 12). This is an effect of the highest solvent exposure and absence of contacts between the galactose ring of the ligand and the protein. It can be seen in Figures 3B and 3C that the galactose ring is oriented outside the recognition domain of the protein.

Comparison with the other CBM40 proteins

The overall macromolecular architecture of the C. perfringens CBM40_NanI is well conserved when comparing with other members of CBM40 family, apart NanH which belongs to another branch of the phylogenetic tree as discussed above. Sequence alignments show that the recognition domains of the CBM40s share common key residues that provide the protein its binding specificity (Figure 3E and Supplementary Online Data SI 13). Within these binding domains, Glu74 and Arg153 (C. perfringens CBM40_NanI numeration) are particularly important, being spatially conserved in all CBMs (with the exception of M. decora NanL, which has an aspartic acid instead of glutamic acid), making a total of four hydrogen bonds with the Neu5Ac (two with the N-acetyl and O4 hydroxy group, and two with the carboxylate group respectively). A highly variable region between all of the CBM40s is the loop between β-strands 5 and 6 (Figures 3D and 3E, see starred box). In C. perfringens CBM40_NanI, this loop is longer by six to eight amino acids and has a unique spatial arrangement compared with the other CBM40 members. In fact, this is a major point of difference between C. perfringens CBM40_NanI and CBM40_NanJ, which share considerable sequence identity and bind sialic acid in virtually the same position on their protein surfaces. Despite this, C. perfringens CBM40_NanI displays a much higher affinity for sialylated oligosaccharides (∼100-fold) than its counterpart. This difference in affinity appears to be due to an increased water-mediated hydrogen-bond set-up inside the binding site of C. perfringens CBM40_NanI.

As seen in the crystal structures, the loop between β-strands 5 and 6, exclusive to C. perfringens CBM40_NanI, introduces an additional surface in the recognition domain. In this loop, Gln85 creates water-mediated interactions with both the sialyl carboxylate and galactosyl hydroxy groups. In contrast with NanI, the crystal structure of C. perfringens CBM40_NanJ does not show such a water-mediated hydrogen-bond network (Supplementary Online Data SI 10).

Sialylated oligosaccharides are flexible epitopes, due to the possibility of their conformation at the α2-3 and α2-6 linkages [44]. It is therefore of interest to compare whether different CBMs recognized the same epitope. S. pneumoniae CBM40_NanA and NanC were also successfully co-crystalized with the 3′-SL and 6′-SL ligands (PDB codes 4C1W and 4C1X, and 4YZ5 and 4YW2 respectively), among others [28,30]. Probably due to the exposure of the ligands and the few contacts made with the protein, different conformations are seen for 3′-SL and 6′-SL when bound to C. perfringens CBM40_NanI, S. pneumoniae CBM40_NanA and S. pneumoniae CBM40_NanC. 3′-SL shares similar conformations in NanI and NanA [in NanC, the galactose ring is in a full inverted position relative to the others, whereas the 6′-SL structure is alike in NanI and NanC (Supplementary Online Data SI 14)].

Conclusion

Although the use of CBMs in biotechnology is fairly recent, they have been described in the paper, textile and food industries [45]. Their potential application in health sciences has also been proved in different scenarios as molecular probes [46], in microarrays [47], as tools for the immobilization of proteins and cells [48] or assisting in vitro glyco-catalysed reactions leading to enhanced yields [45,49,50]. Recent studies have shown the therapeutic potential of multivalent CBMs in the prevention of infection by the influenza virus, through the masking of sialic acid receptors [51,52].

The present study describes a new family 40 CBM that binds specifically to Neu5Ac with a preference for α2,3-sialosides. It was easily engineered and expressed in multimeric fashion by fusing consecutive copies of its gene, leading to a substantial increase in its affinity (11-fold) over the monomeric version. Although the previously described C. perfringens NanJ is highly similar to NanI, the binding affinity of NanI to sialosides acid was much stronger, making it a better candidate to develop as a glycan-binding reagent. Our studies suggest that the variable loop between β-strands 5 and 6 may be responsible for this difference in affinity and may contribute to the unusual ligand specificity observed. The binding of C. perfringens CBM40_NanI to sialosides is considerably more restricted than any other published CBM40, with a distinct preference for α2,3-sialosides, and its interactions at micromolar values are comparable only with the CBM of Vibrio cholerae NanH, which shows a broad specificity for all sialosides [23]. The binding of α2,3-sialosides, but not 3′-O-sulfated ones, makes this reagent more specific for sialic acid residues than the commonly used MAL-1 (Maackia amurensis lectin I) [41].

The present study unravels the molecular basis for the difference in affinity and specificity for two closely related CBM40s from the same organisms. In addition, the high affinity of C. perfringens CBM40_NanI for 3′-SL and its availability as monomer or as engineered dimer makes it an excellent tool for detection and characterization of sialylation for biotechnology use.

AUTHOR CONTRIBUTION

Lara Mahal conceived the project. João Ribeiro, Lara Mahal and Anne Imberty co-ordinated the paper. João Ribeiro and William Pau performed cloning, protein expression and purification. João Ribeiro ran all of the biophysical experiments and provided data analysis. João Ribeiro and Annabelle Varrot resolved the crystal structures. Carlo Pifferi and Olivier Renaudet synthesized and characterized the SiaOPr. João Ribeiro analysed data, generated the Figures and Tables and wrote the body of the paper. Annabelle Varrot, Lara Mahal and Anne Imberty revised the paper leading to the final version. All authors reviewed the results and approved the paper for submission.

Crystal data collection was performed at the European Synchrotron Radiation Facility, Grenoble, France, and we are thankful for the access to beamlines BM30A-FIP and ID29 as well as for the technical support of David Cobessi on BM30A.

FUNDING

The research leading to these results has received funding from the European Community's 7th Framework Programme (FP7/2007–2013) under the Marie Curie International Outgoing Fellowship for Career Development [grant number PIOF-GA-2011-298910], BioStruct-X [grant number 283570] and ERASynbio program SynGlycTis [grant number ANR-14-SYNB-0002-02]. We are also grateful to the Labex Arcane for the financial support [grant number ANR-11-LABX-0003-01]. We are also grateful for the participation of the Protein–Glycan Interaction Resource of the Consortium for Functional Genomics (CFG) [supporting grant number R24 GM098791].

Abbreviations

     
  • CBM

    carbohydrate-binding module

  •  
  • CFG

    Consortium for Functional Glycomics

  •  
  • GlcNAc

    N-acetylglucosamine

  •  
  • HBS-T

    Hepes-buffered saline with Tween 20

  •  
  • ITC

    isothermal titration calorimetry

  •  
  • LFA

    Limax flavus agglutinin

  •  
  • MAA

    Maackia amurensis agglutinin

  •  
  • MIC

    minimal inhibition concentration

  •  
  • MME

    monomethyl ether

  •  
  • Neu5Ac

    N-acetylneuraminic acid

  •  
  • PSL-I

    Polyporus squamosus lectin I

  •  
  • SiaOPr

    Neu5Ac derivative with a propargyl aglycon

  •  
  • 3′-SL

    3′-sialyl-lactose

  •  
  • 6′-SL

    6′-sialyl-lactose

  •  
  • 3′-SLN

    3′-sialyl-lactosamine

  •  
  • 6′-SLN

    6′-sialyl-lactosamine

  •  
  • SNA-I

    Sambucus nigra agglutinin I

  •  
  • SPR

    surface plasmon resonance

  •  
  • TSA

    thermal shift assay

  •  
  • WGA

    wheatgerm agglutinin

References

References
1
Varki
A.
Biological roles of oligosaccharides: all of the theories are correct
Glycobiology
1993
, vol. 
3
 (pg. 
97
-
130
)
[PubMed]
2
Traving
C.
Schauer
R.
Structure, function and metabolism of sialic acids
Cell. Mol. Life Sci.
1998
, vol. 
54
 (pg. 
1330
-
1349
)
[PubMed]
3
Varki
A.
Essentials of Glycobiology
1999
Cold Spring Harbor
Cold Spring Harbor Laboratory Press
4
Schnaar
R.L.
Gerardy-Schahn
R.
Hildebrandt
H.
Sialic acids in the brain: gangliosides and polysialic acid in nervous system development, stability, disease, and regeneration
Physiol. Rev.
2014
, vol. 
94
 (pg. 
461
-
518
)
[PubMed]
5
Wang
B.
Brand-Miller
J.
The role and potential of sialic acid in human nutrition
Eur. J. Clin. Nutr.
2003
, vol. 
57
 (pg. 
1351
-
1369
)
[PubMed]
6
Wang
P.-H.
Altered glycosylation in cancer: sialic acids and sialyltransferases
J. Cancer Mol.
2005
, vol. 
1
 (pg. 
73
-
81
)
7
Pearce
O.M.
Laubli
H.
Sialic acids in cancer biology and immunity
Glycobiology
2016
, vol. 
26
 (pg. 
111
-
128
)
[PubMed]
8
Dall'Olio
F.
Chiricolo
M.
Ceccarelli
C.
Minni
F.
Marrano
D.
Santini
D.
β-Galactoside α2,6 sialyltransferase in human colon cancer: contribution of multiple transcripts to regulation of enzyme activity and reactivity with Sambucus nigra agglutinin
Int. J. Cancer
2000
, vol. 
88
 (pg. 
58
-
65
)
[PubMed]
9
Schultz
M.
Swindall
A.
Bellis
S.
Regulation of the metastatic cell phenotype by sialylated glycans
Cancer Metastasis Rev.
2012
, vol. 
31
 (pg. 
501
-
518
)
[PubMed]
10
Ito
S.
Hayama
K.
Hirabayashi
J.
Enrichment strategies for glycopeptides
Methods Mol. Biol.
2009
, vol. 
534
 (pg. 
195
-
203
)
[PubMed]
11
Rogerieux
F.
Belaise
M.
Terzidis-Trabelsi
H.
Greffard
A.
Pilatte
Y.
Lambré
C.
Determination of the sialic acid linkage specificity of sialidases using lectins in a solid phase assay
Anal. Biochem.
1993
, vol. 
211
 (pg. 
200
-
204
)
[PubMed]
12
Nicholls
J.
Bourne
A.
Chen
H.
Guan
Y.
Peiris
J.
Sialic acid receptor detection in the human respiratory tract: evidence for widespread distribution of potential binding sites for human and avian influenza viruses
Respir. Res.
2007
, vol. 
8
 (pg. 
73
-
82
)
[PubMed]
13
Tateno
H.
Winter
H.
Goldstein
I.
Cloning, expression in Escherichia coli and characterization of the recombinant Neu5Acα2–6Galβ1–4GlcNAc-specific high-affinity lectin and its mutants from the mushroom Polyporus squamosus
Biochem. J.
2004
, vol. 
382
 (pg. 
667
-
675
)
[PubMed]
14
Kadirvelraj
R.
Grant
O.
Goldstein
I.
Winter
H.
Tateno
H.
Fadda
E.
Woods
R.
Structure and binding analysis of Polyporus squamosus lectin in complex with the Neu5Acα2–6Galβ1–4GlcNAc human-type influenza receptor
Glycobiology
2011
, vol. 
21
 (pg. 
973
-
984
)
[PubMed]
15
Knibbs
R.N.
Osborne
S.E.
Glick
G.D.
Goldstein
I.J.
Binding determinants of the sialic acid-specific lectin from the slug Limax flavus
J. Biol. Chem.
1993
, vol. 
268
 (pg. 
18524
-
18531
)
[PubMed]
16
Hsu
K.L.
Gildersleeve
J.C.
Mahal
L.K.
A simple strategy for the creation of a recombinant lectin microarray
Mol. Biosyst.
2008
, vol. 
4
 (pg. 
654
-
662
)
[PubMed]
17
Lewis
A.L.
Lewis
W.G.
Host sialoglycans and bacterial sialidases: a mucosal perspective
Cell. Microbiol.
2012
, vol. 
14
 (pg. 
1174
-
1182
)
[PubMed]
18
Juge
N.
Tailford
L.
Owen
C.D.
Sialidases from gut bacteria: a mini-review
Biochem. Soc. Trans.
2016
, vol. 
44
 (pg. 
166
-
175
)
[PubMed]
19
Matrosovich
M.
Herrler
G.
Klenk
H.D.
Sialic acid receptors of viruses
Top. Curr. Chem.
2015
, vol. 
367
 (pg. 
1
-
28
)
[PubMed]
20
Miyagi
T.
Yamaguchi
K.
Mammalian sialidases: physiological and pathological roles in cellular functions
Glycobiology
2012
, vol. 
22
 (pg. 
880
-
896
)
[PubMed]
21
Bourne
Y.
Henrissat
B.
Glycoside hydrolases and glycosyltransferases: families and functional modules
Curr. Opin. Struct. Biol.
2001
, vol. 
11
 (pg. 
593
-
600
)
[PubMed]
22
Ficko-Blean
E.
Boraston
A.B.
Insights into the recognition of the human glycome by microbial carbohydrate-binding modules
Curr. Opin. Struct. Biol.
2012
, vol. 
22
 (pg. 
570
-
577
)
[PubMed]
23
Connaris
H.
Crocker
P.R.
Taylor
G.L.
Enhancing the receptor affinity of the sialic acid-binding domain of Vibrio cholerae sialidase through multivalency
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
7339
-
7351
)
[PubMed]
24
Lombard
V.
Golaconda Ramulu
H.
Drula
E.
Coutinho
P.M.
Henrissat
B.
The carbohydrate-active enzymes database (CAZy) in 2013
Nucleic Acids Res.
2014
, vol. 
42
 (pg. 
D490
-
D495
)
[PubMed]
25
Moustafa
I.
Connaris
H.
Taylor
M.
Zaitsev
V.
Wilson
J.C.
Kiefel
M.J.
von Itzstein
M.
Taylor
G.
Sialic acid recognition by Vibrio cholerae neuraminidase
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
40819
-
40826
)
[PubMed]
26
Boraston
A.B.
Ficko-Blean
E.
Healey
M.
Carbohydrate recognition by a large sialidase toxin from Clostridium perfringens
Biochemistry
2007
, vol. 
46
 (pg. 
11352
-
11360
)
[PubMed]
27
Luo
Y.
Li
S.C.
Chou
M.Y.
Li
Y.T.
Luo
M.
The crystal structure of an intramolecular trans-sialidase with a NeuAcα2→3Gal specificity
Structure
1998
, vol. 
6
 (pg. 
521
-
530
)
[PubMed]
28
Yang
L.
Connaris
H.
Potter
J.A.
Taylor
G.L.
Structural characterization of the carbohydrate-binding module of NanA sialidase, a pneumococcal virulence factor
BMC Struct. Biol.
2015
, vol. 
15
 pg. 
15
 
[PubMed]
29
Xu
G.
Potter
J.A.
Russell
R.J.
Oggioni
M.R.
Andrew
P.W.
Taylor
G.L.
Crystal structure of the NanB sialidase from Streptococcus pneumoniae
J. Mol. Biol.
2008
, vol. 
384
 (pg. 
436
-
449
)
[PubMed]
30
Owen
C.D.
Lukacik
P.
Potter
J.A.
Sleator
O.
Taylor
G.L.
Walsh
M.A.
Streptococcus pneumoniae NanC: structural insights into the specificity and mechanism of a sialidase that produces a sialidase inhibitor
J. Biol. Chem.
2015
, vol. 
290
 (pg. 
27736
-
27748
)
[PubMed]
31
Myers
G.S.A.
Rasko
D.A.
Cheung
J.K.
Ravel
J.
Seshadri
R.
DeBoy
R.T.
Ren
Q.
Varga
J.
Awad
M.M.
Brinkac
L.M.
, et al. 
Skewed genomic variability in strains of the toxigenic bacterial pathogen, Clostridium perfringens
Genome Res.
2006
, vol. 
16
 (pg. 
1031
-
1040
)
[PubMed]
32
Newstead
S.
Chien
C.H.
Taylor
M.
Taylor
G.
Crystallization and atomic resolution X-ray diffraction of the catalytic domain of the large sialidase, NanI, from Clostridium perfringens
Acta Crystallogr. D Biol. Crystallogr.
2004
, vol. 
60
 (pg. 
2063
-
2066
)
[PubMed]
33
Dupeux
F.
Rower
M.
Seroul
G.
Blot
D.
Marquez
J.A.
A thermal stability assay can help to estimate the crystallization likelihood of biological samples
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
915
-
919
)
[PubMed]
34
Leslie
A.G.W.
1992
 
Automated data collection and processing for macromolecular crystallography. In Joint CCP4 and ESF-EAMCB Newsletter on Protein Crystallography, No. 26
35
Kabsch
W.
Xds
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
125
-
132
)
[PubMed]
36
Winn
M.D.
Ballard
C.C.
Cowtan
K.D.
Dodson
E.J.
Emsley
P.
Evans
P.R.
Keegan
R.M.
Krissinel
E.B.
Leslie
A.G.
McCoy
A.
, et al. 
Overview of the CCP4 suite and current developments
Acta Crystallogr D Biol Crystallogr
2011
, vol. 
67
 (pg. 
235
-
242
)
[PubMed]
37
McCoy
A.J.
Grosse-Kunstleve
R.W.
Adams
P.D.
Winn
M.D.
Storoni
L.C.
Read
R.J.
Phaser crystallographic software
J. Appl. Crystallogr.
2007
, vol. 
40
 (pg. 
658
-
674
)
[PubMed]
38
Langer
G.
Cohen
S.X.
Lamzin
V.S.
Perrakis
A.
Automated macromolecular model building for X-ray crystallography using ARP/wARP version 7
Nat. Protoc.
2008
, vol. 
3
 (pg. 
1171
-
1179
)
[PubMed]
39
Murshudov
G.N.
Skubak
P.
Lebedev
A.A.
Pannu
N.S.
Steiner
R.A.
Nicholls
R.A.
Winn
M.D.
Long
F.
Vagin
A.A.
REFMAC5 for the refinement of macromolecular crystal structures
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
355
-
367
)
[PubMed]
40
Emsley
P.
Lohkamp
B.
Scott
W.G.
Cowtan
K.
Features and development of Coot
Acta Crystallogr. Sect. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
486
-
501
)
41
Wang
L.
Cummings
R.D.
Smith
D.F.
Huflejt
M.
Campbell
C.T.
Gildersleeve
J.C.
Gerlach
J.Q.
Kilcoyne
M.
Joshi
L.
Serna
S.
, et al. 
Cross-platform comparison of glycan microarray formats
Glycobiology
2014
, vol. 
24
 (pg. 
507
-
517
)
[PubMed]
42
Lameignere
E.
Malinovská
L.
Sláviková
M.
Duchaud
E.
Mitchell
E.P.
Varrot
A.
Sedo
O.
Imberty
A.
Wimmerová
M.
Structural basis for mannose recognition by a lectin from opportunistic bacteria Burkholderia cenocepacia
Biochem. J.
2008
, vol. 
411
 (pg. 
307
-
318
)
[PubMed]
43
Blixt
O.
Head
S.
Mondala
T.
Scanlan
C.
Huflejt
M.E.
Alvarez
R.
Bryan
M.C.
Fazio
F.
Calarese
D.
Stevens
J.
, et al. 
Printed covalent glycan array for ligand profiling of diverse glycan binding proteins
Proc. Natl. Acad. Sci. U.S.A.
2004
, vol. 
101
 (pg. 
17033
-
17038
)
[PubMed]
44
Imberty
A.
Pérez
S.
Structure, conformation, and dynamics of bioactive oligosaccharides: theoretical approaches and experimental validations
Chem. Rev.
2000
, vol. 
100
 (pg. 
4567
-
4588
)
[PubMed]
45
Oliveira
C.
Carvalho
V.
Domingues
L.
Gama
F.M.
Recombinant CBM-fusion technology: applications overview
Biotechnol. Adv.
2015
, vol. 
33
 (pg. 
358
-
369
)
[PubMed]
46
von Schantz
L.
Hakansson
M.
Logan
D.T.
Walse
B.
Osterlin
J.
Nordberg-Karlsson
E.
Ohlin
M.
Structural basis for carbohydrate-binding specificity: a comparative assessment of two engineered carbohydrate-binding modules
Glycobiology
2012
, vol. 
22
 (pg. 
948
-
961
)
[PubMed]
47
Ofir
K.
Berdichevsky
Y.
Benhar
I.
Azriel-Rosenfeld
R.
Lamed
R.
Barak
Y.
Bayer
E.A.
Morag
E.
Versatile protein microarray based on carbohydrate-binding modules
Proteomics
2005
, vol. 
5
 (pg. 
1806
-
1814
)
[PubMed]
48
Shoseyov
O.
Shani
Z.
Levy
I.
Carbohydrate binding modules: biochemical properties and novel applications
Microbiol. Mol. Biol. Rev.
2006
, vol. 
70
 (pg. 
283
-
295
)
[PubMed]
49
Volkov
I.
Lunina
N.A.
Velikodvorskaia
G.A.
[Prospects for practical application of substrate-binding modules of glycosyl hydrolases (a review)]
Prikl. Biokhim. Mikrobiol.
2004
, vol. 
40
 (pg. 
499
-
504
)
[PubMed]
50
Codera
V.
Gilbert, Harry
J.
Faijes
M.
Planas
A.
Carbohydrate-binding module assisting glycosynthase-catalysed polymerizations
Biochem. J.
2015
, vol. 
470
 (pg. 
15
-
22
)
[PubMed]
51
Govorkova
E.A.
Baranovich
T.
Marathe
B.M.
Yang
L.
Taylor
M.A.
Webster
R.G.
Taylor
G.L.
Connaris
H.
Sialic acid-binding protein Sp2CBMTD protects mice against lethal challenge with emerging influenza A (H7N9) virus
Antimicrob. Agents Chemother.
2015
, vol. 
59
 (pg. 
1495
-
1504
)
[PubMed]
52
Connaris
H.
Govorkova
E.A.
Ligertwood
Y.
Dutia
B.M.
Yang
L.
Tauber
S.
Taylor
M.A.
Alias
N.
Hagan
R.
Nash
A.A.
, et al. 
Prevention of influenza by targeting host receptors using engineered proteins
Proc. Natl. Acad. Sci. U.S.A.
2014
, vol. 
111
 (pg. 
6401
-
6406
)
[PubMed]
53
Robert
X.
Gouet
P.
Deciphering key features in protein structures with the new ENDscript server
Nucleic Acids Res.
2014
, vol. 
42
 (pg. 
W320
-
W324
)
[PubMed]