NMR studies on the interaction of sugars with the C-terminal domain of an R-type lectin from the earthworm Lumbricus terrestris

Authors

  • Hikaru Hemmi,

    1.  National Food Research Institute, National Agriculture and Food Research Organization (NARO), Tsukuba, Japan
    Search for more papers by this author
  • Atsushi Kuno,

    1.  Research Center for Medical Glycoscience, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan
    Search for more papers by this author
  • Shigeyasu Ito,

    1.  Research Center for Medical Glycoscience, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan
    2.  Department of Material and Biological Chemistry, Yamagata University, Yamagata, Japan
    Search for more papers by this author
  • Ryuichiro Suzuki,

    1.  Research Center for Medical Glycoscience, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan
    2.  Department of Material and Biological Chemistry, Yamagata University, Yamagata, Japan
    Search for more papers by this author
  • Tsunemi Hasegawa,

    1.  Department of Material and Biological Chemistry, Yamagata University, Yamagata, Japan
    Search for more papers by this author
  • Jun Hirabayashi

    1.  Research Center for Medical Glycoscience, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan
    Search for more papers by this author

H. Hemmi, National Food Research Institute, National Agriculture and Food Research Organization (NARO), 2-1-12 Kannondai, Tsukuba, 305-8642 Ibaraki, Japan
Fax: +81 298 387996
Tel: +81 298 388033
E-mail: hemmi@affrc.go.jp

Abstract

The R-type lectin EW29, isolated from the earthworm Lumbricus terrestris, consists of two homologous domains (14 500 Da) showing 27% identity with each other. The C-terminal domain (Ch; C-half) of EW29 (EW29Ch) has two sugar-binding sites in subdomains α and γ, and the protein uses these sugar-binding sites for its function as a single-domain-type hemagglutinin. In order to determine the sugar-binding ability and specificity for each of the two sugar-binding sites in EW29Ch, ligand-induced chemical-shift changes in EW29Ch were monitored using 1H–15N HSQC spectra as a function of increasing concentrations of lactose, melibiose, d-galactose, methyl α-d-galactopyranoside and methyl β-d-galactopyranoside. Shift perturbation patterns for well-resolved resonances confirmed that all of these sugars associated independently with the two sugar-binding sites of EW29Ch. NMR titration experiments showed that the sugar-binding site in subdomain α had a slow or intermediate exchange regime on the chemical-shift timescale (Kd = 10−2 to 10−1 mm), whereas that in subdomain γ had a fast exchange regime for these sugars (Kd = 2–6 mm). Thus, our results suggest that the two sugar-binding sites of EW29Ch in the same molecule retain its hemagglutinating activity, but this activity is 10-fold lower than that of the whole protein because EW29Ch has two sugar-binding sites in the same molecule, one of which has a weak binding mode.

Abbreviations
Ch (C-half)

the C-terminal domain

EW29

earthworm 29 kDa lectin

α-Me-Gal

methyl α-d-galactopyranoside

β-Me-Gal

methyl β-d-galactopyranoside

STD

saturation transfer difference

Sugar-binding proteins, known as lectins, exist ubiquitously in both animals and plants, but lectins from the annelid phylum have rarely been reported. A 29 kDa lectin (EW29) was isolated from the earthworm Lumbricus terrestris using affinity chromatography on asialofetuin–agarose in the screening of galectin-like proteins. The protein consists of two homologous domains (14 500 Da), i.e. N- and C-terminal domains, which show 27% identity with each other [1]. Both domains of EW29 form a tandem-repeat type structure and contain triple-repeated QXW motifs [2,3]. This short motif has been found in many carbohydrate-recognition proteins including the plant lectin ricin B-chain [4]. The 3D structures of R-type lectins have been determined [2,5–19], and these proteins possess common β-trefoil fold structures, although their sugar-binding affinities differ depending on their ligand specificities.

Recent biochemical data on EW29 and its truncated mutants showed it to be a single-domain type of hemagglutinin, differing from other tandem repeat-type proteins in the R-type lectin family, such as ricin [2], abrin [20] and Sambucus sieboldina agglutinin [21], the exception being the ricin B1 domain which has two sugar-binding sites in the same molecule [22,23]. Based on these structural features, R-type lectins generally contain one sugar-binding site per domain, suggesting that the truncated mutant, which comprises a single domain, may have no hemagglutinating activity [2,5,20]. However, the C-terminal domain (Ch; C-half) of EW29 (EW29Ch) bound to asialofetuin–agarose as strongly as the whole protein and retained its hemagglutinating activity, although at a level 10-fold lower than that of the whole protein [1]. The crystal structures of the complex between EW29Ch and lactose or N-acetylgalactosamine (PDB: 2ZQN or 2ZQO) reported recently indicate that the overall structure of EW29Ch resembles the characteristic pseudo-threefold symmetry of the three subdomains designated α, β and γ, and that the protein has two sugar-binding sites in subdomains α and γ [24,25]. Therefore, determination of the sugar-binding ability and specificity for each of the two sugar-binding sites in EW29Ch is expected to elucidate the molecular basis of the carbohydrate cross-linking properties of the lectin.

In this study, the sugar-binding ability and specificity of each of the two sugar-binding sites in EW29Ch for certain sugars were determined using NMR titration experiments in combination with recent crystallographic studies [25]. The NMR titration experiments showed that the α sugar-binding site has a much tighter sugar-binding mode than the γ sugar-binding site. Furthermore, saturation transfer difference (STD)–NMR experiments for a mixture of the protein with sugar revealed the epitope of the sugar for the sugar-binding protein. Thus, our results suggest that the two sugar-binding sites of EW29Ch in the same molecule retained its hemagglutinating activity, but this activity was 10-fold lower than that of the whole protein because EW29Ch has two sugar-binding sites in the same molecule, one of which has a weak binding mode.

Results

Resonance assignments

Complete resonance assignments for EW29Ch have been reported elsewhere [26]. In this study, we observed chemical shifts for some residues in subdomain α as a pair of resonance signals in the unbound state and the bound state. Furthermore, the resonance signal of residues Gly21 and Asn23 in subdomain α, which was assigned because of lactose contamination in the previous study [26], disappeared in the completely sugar-free state (Fig. S1). The resonance signals of EW29Ch in the completely sugar-free state were therefore reassigned using multidimensional and multinuclear NMR spectroscopy, as described elsewhere [26]. Figure 1 shows the 1H–15N HSQC spectrum for EW29Ch in the completely sugar-free state. The assignment data previously deposited in BMRB under accession number 6226 [26] were corrected and re-deposited under the same accession number.

Figure 1.

1H–15N HSQC spectrum of the C-terminal domain of EW29 in the sugar-free state. A 600 MHz 2D 1H–15N HSQC spectrum of the 0.9 mm C-terminal domain of EW29 at pH 6.1 and 298K in the sugar-free state. Cross-peaks are labeled based on an analysis of through-bond connectivities. The side chains of NH2 resonances of asparagines and glutamines are connected by horizontal lines. The side chains of NH resonances of tryptophan are marked by ‘sc’.

Identification of sugar-binding sites

The interaction of 15N-labeled EW29Ch with lactose, melibiose, galactose, methyl α-d-galactopyranoside (α-Me-Gal) and methyl β-d-galactopyranoside (β-Me-Gal) was monitored using 1H–15N HSQC spectroscopy. An overlay of 10 1H–15N HSQC spectra showed progressive chemical-shift changes for some amide resonances of EW29Ch upon the addition of lactose. Overlaid spectra showed two types of chemical exchange (slow and fast) on the chemical-shift timescale (Fig. 2A). Figure 2B shows the overall effect of lactose binding by mapping the observed main- and side-chain 15N chemical-shift changes on the crystal structure of EW29Ch. Residues showing a slow exchange regime in EW29Ch were located in subdomain α, whereas those showing a fast exchange regime were located in subdomain γ (Fig. 2B). Larger chemical-shift changes in the fast exchange regime were observed for backbone amides, as well as side-chain amide and indole groups, of residues within or adjacent to the sugar-binding site of subdomain γ, identified from the crystal structure of lactose-liganded EW29Ch. In the case of other sugars used in this study, chemical exchanges in subdomain α showed a slow exchange regime for β-Me-Gal and an intermediate exchange regime for melibiose, galactose and α-Me-Gal, whereas those in subdomain γ showed a fast exchange regime for all sugars used. No chemical-shift changes were observed for any sugars in subdomain β. These results showed that each of the two sugar-binding sites (α and γ) of EW29Ch had a distinct chemical exchange on the chemical-shift timescale.

Figure 2.

 Chemical-shift changes in NMR titration of EW29Ch with lactose. (A) Ten 1H–15N HSQC spectra of 15N-labeled EW29Ch in the presence of protein/lactose molar ratios of 0 (red), 0.5 (green), 1.0 (blue), 2.0 (magenta), 4.0 (gold), 6.0 (orange), 10.0 (pink), 20.0 (purple), 40.0 (coral) and 80.0 (cyan) are overlaid. Overlay spectra show that free and bound forms of some residues in the protein are in a slow exchange on the NMR time scale and those of other residues are in a fast exchange. Representative residues for slow and fast exchanges are labeled. The arrow indicates the direction in which amide 1H–15N peaks shift with the adding of sugar. (B) Mapping of the 1HN and 15N chemical-shift changes upon the addition of excess lactose on a ribbon diagram of the crystal structure of EW29Ch (PDB: 2ZQN) generated by molmol [53]. Spheres represent 15N atoms of the main chain and side chains of each residue in the protein. Residues showing a slow exchange regime are in red and those showing a fast exchange regime and Δavmax > 0.2, where Δav is the normalized weighted average of the 1H and 15N chemical-shift changes and Δmax is the maximum observed weighted shift difference (0.549 p.p.m. for side chain amide cross peak of N124), are in green. Residues showing a fast exchange regime and 0.1 ≤ Δavmax ≤ 0.2 are in light green, and others are in gray. Key residues showing chemical-shift changes in slow and in fast exchange regimes are labeled.

Site-specific dissociation constants determined by NMR

The site-specific binding constants and chemical exchange regimes of EW29Ch with sugars used in this study are given Table 1. Upon the addition of sugars, the chemical-shift changes in subdomain α were in a slow exchange regime or an intermediate exchange regime, as described above. For lactose and β-Me-Gal, the second signal corresponding to the bound state clearly appeared at [sugar]/[EW29Ch] ∼ 0.3; the first signal corresponding to the unbound state disappeared completely at [sugar]/[EW29Ch] ∼ 1.3 and only the second signal was observed (Fig. 3A). This phenomenon indicated a stoichiometric interaction between EW29Ch and lactose at a ∼ 1 : 1 ratio. The dissociation constants (Kd) of the α sugar-binding site for lactose and β-Me-Gal were calculated in a similar way using NMR titration experiments. From the theoretical calculations of Kd for the ratios of sugar-free to sugar-bound peak intensities as a function of sugar concentration, the protein concentration was lowered to 0.05 mm to obtain the Kd more precisely for the sugar-binding site in the slow exchange regime. However, the Kd of the α sugar-binding site could not be calculated using nonlinear regression fitting to the binding isotherm because the protein concentration was too low to detect the peak intensities accurately (Fig. S2). Therefore, the Kd values of the α sugar-binding site (residues Asp18, Ser28, Trp33 and Gln44) were approximated to 0.01–0.07 mm for lactose and 0.02–0.08 mm for β-Me-Gal. This was similar to the previously reported Kd value of 0.016 mm for lactose using total binding constants of the two sugar-binding sites in EW29Ch, measured by frontal affinity chromatography analysis [24].

Table 1.   Average site-specific dissociation constants calculated for EW29Ch with sugar ligands. Data obtained at 25 °C and pH 6.1 in 50 mm of potassium phosphate and a 10% D2O/90% H2O mixture. The chemical exchange regime in parentheses is based on the observed alterations in NMR signal positions and intensities.
SugarKd (mm)
α Siteaγ Siteb
  1. a Dissociation constants could not be calculated accurately due to the slow or intermediate exchange regime, so the Kd is shown as an approximation. b The reported Kd values are the average of the those determined from the 15N and HN chemical shift perturbations of Ile102, Ile104, Cys115, Trp117, Lys118, Gly122 and Asn124. The error range is the standard deviation.

Lactose0.01–0.07 (slow)2.66 ± 0.30 (fast)
Melibiose∼ 10−1 (intermediate)5.34 ± 0.81 (fast)
Galactose∼ 10−1 (intermediate)3.89 ± 0.37 (fast)
α-Me-Gal∼ 10−1 (intermediate)4.48 ± 0.38 (fast)
β-Me-Gal0.02–0.08 (slow)2.88 ± 0.21 (fast)
Figure 3.

 Close-up of the 1H–15N HSQC regions showing the chemical exchange for Asp18 or Trp33sc with increasing amounts of some sugars. Peak movements of main chain amide and amide proton of Asp18, and side chain amide and amide proton (marked by ‘sc’) of Trp33 in EW29Ch during the titration of lactose (A, top), β-Me-Gal (A, bottom), melibiose (B, top) and α-Me-Gal (B, bottom). Shown are regions of the 1H–15N HSQC spectra corresponding to Asp18 and Trp33sc in EW29Ch at [sugar]/[EW29Ch] molar ratios indicated at the top. The behavior of Asp18 and Trp33sc during the titration of lactose and β-Me-Gal corresponds to a slow exchange regime (A) and those during titration of melibiose and α-Me-Gal correspond to an intermediate exchange regime (B).

For other sugars, the first signal corresponding to the unbound state began to broaden at [sugar]/[EW29Ch] ∼ 0.5, the center of the broadened signals shifted to the position of the second signal corresponding to the bound state during titration ([sugar]/[EW29Ch] = 0.5–3) and the resonance signal of the side-chain NH of Trp33 disappeared. The broadened signals sharpened at the position of the second signal at [sugar]/[EW29Ch] = 3–4 and the resonance signal of the side-chain NH of Trp33 appeared at the position of the second signal corresponding to the bound state (Fig. 3B). Because the signals were in an intermediate exchange, of which the Kd should have a medium value between that in the slow exchange regime and that in the fast exchange regime, titration data indicated that the interaction with sugars had a Kd of ∼ 10−1 mm. Furthermore, at the α sugar-binding site, the binding specificity for an anomer was observed; the chemical exchange for lactose and β-Me-Gal was in the slow exchange regime, whereas that for melibiose and α-Me-Gal, as anomers of lactose and β-Me-Gal, was in the intermediate exchange regime. Thus, the configuration at the hemiacetal carbon of galactose may affect the dissociation constants.

By contrast, because chemical-shift changes upon the addition of sugars at subdomain γ were in the fast exchange regime, Kd values describing the interaction of lactose, melibiose, galactose, α-Me-Gal and β-Me-Gal with EW29Ch were calculated using nonlinear least-square fitting of the chemical shift titration data to the binding isotherm [27]. A plot of the weighted average chemical-shift changes of 1H and 15N resonances for the cross-peaks of Gly122, as a function of the molar ratio of each sugar to EW29Ch, is shown in Fig. 4. The Kd values for each sugar were calculated from the titration curves measured for the main chain and amide proton groups of Ile102, Ile104, Cys115, Trp117, Lys118 and Gly122, and the side chain nitrogen and amide proton group of Asn124 in the γ sugar-binding site. These residues, which exhibited the most significant perturbations in the 15N and amide proton chemical shifts upon sugar binding, all lie within or adjacent to the sugar-binding sites of EW29Ch identified by the crystal structure of the complex. Average γ-site dissociation constants calculated for each sugar were analyzed in accordance with a simple model of each of the two sugar-binding sites in EW29Ch interacting with one sugar molecule in an independent or non-cooperative manner, because this assumption was supported by evidence (monophasic changes in chemical shifts upon the addition of each sugar and crystallographic studies of the protein-sugar complex) similar to that reported by Schärpf et al. [28]. The Kd values for the γ sugar-binding site were 2–6 mm for all sugars in this study (Table 1), so these results indicated that the α sugar-binding site of EW29Ch is a high-affinity site and the γ sugar-binding site is a low-affinity site.

Figure 4.

 Dissociation constants (Kd) of EW29Ch for some sugars. Kd values of EW29Ch for some sugars were determined by nonlinear regression fitting of the chemical-shift change versus the sugar concentration to the binding isotherm describing the binding of one ligand molecule to a single protein site using the Solver function of excel 2002. The weighted average of the 1H and 15N chemical-shift changes of Gly122 given by Δav = {(ΔNH2 + ΔN2/25)/2}1/2 [50] is plotted as a function of sugar/protein molar ratios of added lactose (⋄), melibiose (□), galactose (△), α-Me-Gal (×) and β-Me-Gal (○).

Interactions of sugars with EW29Ch by STD–NMR experiments

STD–NMR experiments were conducted to determine the binding epitope of lactose and β-Me-Gal to EW29Ch because NMR titration experiments of EW29Ch with sugars showed that the chemical exchange of the α sugar-binding site upon the addition of lactose or β-Me-Gal was in the slow exchange regime. The STD effect for sugar arose from the contribution of both the α and γ sugar-binding sites because of the mixture of lactose and EW29Ch at a ratio of 100 : 1. At first, 2D STD–TOCSY and 2D STD–[1H,13C] HSQC spectra were obtained for lactose with EW29Ch to assign the STD–NMR signals completely, because the proton chemical shifts of the galactose and glucose residues in lactose partly overlap. In the STD–TOCSY and STD–[1H,13C] HSQC spectra, the H1-Gal, H2-Gal, H3-Gal, H4-Gal, H5-Gal and H6-Gal resonances were assigned unambiguously (Fig. 5). In both 2D spectra, resonance signals from the glucose residue in lactose were not observed. However, the crystal structure of EW29Ch with lactose showed that the glucose residue of the lactose molecule interacted with subdomain γ of EW29Ch [25]. This interaction may be an artifact caused by the crystallization of lactose-liganded EW29Ch because: (a) in the other EW29Ch molecule of the crystal structure (each crystal contained two molecules A and B) the interaction between the glucose residue and subdomain γ of EW29Ch was not observed; (b) the B-factor of the side chain of Lys105 was high, indicating that the side chain of Lys105 is flexible; (c) in this NMR study, the Kd of the γ sugar-binding site for β-Me-Gal was the same as that for lactose; and (d) the STD–NMR data in this study showed that the epitope of lactose for EW29Ch is the galactose residue. Thus, these results showed that both sugar-binding sites of EW29Ch only interact with the galactose residue.

Figure 5.

 2D STD–TOCSY and STD–[1H,13C] HSQC spectra for the mixture of lactose (5 mm) and EW29Ch (50 μm). (A) Reference TOCSY spectrum of the mixture of lactose and EW29Ch at a ratio of 100 : 1. (B) The STD–TOCSY spectrum of the same sample was collected in an alternative fashion. (C) STD–[1H,13C] HSQC spectrum at a 10-fold higher concentration of the same sample.

1D 1H STD–NMR was conducted for the EW29Ch–β-Me-Gal complex to quantitatively analyze the epitopes of sugars interacting with the protein, because the resonance signals of the glucose residue overlapping with those of the galactose residue within the lactose affected the subtraction of the free induction decay values with on- and off-resonance protein saturation. Figure 6 shows the 1D 1H NMR spectrum of β-Me-Gal incubated with EW29Ch at a ratio of 100 : 1, and the corresponding 1D STD spectrum. The integral value of the H4 proton, the largest signal of β-Me-Gal, was set to 100%. Figure 6C shows the relative degree of saturation of individual protons normalized to that of H4. The H3, H5, H6a and H6b protons had similar STD intensities of between 41% and 54%. The H2 proton had a smaller STD intensity of 30%. The lowest intensities corresponded to the H1 proton and protons of the O-methyl group, which reached only 18% and 11%, respectively. These results indicated that EW29Ch recognizes the region from H2 to H6a/H6b, particularly H4, and barely interacts with the region of H1 and the O-methyl group. The crystal structure of EW29Ch with lactose showed that the O2, O3 and O4 atoms of the galactose residue of lactose formed hydrogen bonds with EW29Ch, and the C3, C4, C5, C6, O3 and O6 atoms of the galactose residue formed stacking interactions with both the α and γ sugar-binding sites of EW29Ch [26]. Therefore, our results confirmed that GalO2–GalO6 of the galactose residue are epitopes for binding to EW29Ch.

Figure 6.

 1D STD–NMR spectrum for the mixture of β-Me-Gal (5 mm) and EW29Ch (50 μm). (A) Reference NMR spectrum of the mixture of β-Me-Gal and EW29Ch at a ratio of 100 : 1. (B) STD–NMR spectrum of the same sample. Prior to acquisition, a 30 ms spin–lock pulse was applied to remove residual protein resonance. (C) Structure of β-Me-Gal and the relative degree of saturation of individual protons normalized to that of the H4 proton as determined from the 1D STD–NMR spectrum (B).

Discussion

Binding of an individual lectin site (monovalent binding) to a monosaccharide is extremely weak, with Kd values typically in the range of 0.1–10 mm [29–31]. In the R-type lectin family, the dissociation constants of ricin and RCA120 have been determined mainly by equilibrium dialysis studies and fluorescence polarization studies [32–42]. Ricin has at least two binding sites in its molecule. The Kd values of ricin for lactose at 4 °C are ∼ 0.03 mm for the high-affinity site and ∼ 0.3 mm for the low-affinity site. Recently, a third binding site has been found in ricin; thus, the ricin B1 domain of the ricin B chain has two sugar-binding sites in the same molecule [22,23]. Because sugars bound at the third sugar-binding site of the ricin B chain were not observed, it is speculated that Kd for the third sugar-binding site of the ricin B1 domain is more than one order of magnitude larger than that for the low-affinity site of ricin, like the Kd value for the γ sugar-binding site of EW29Ch. In this study, Kd of the α sugar-binding site of EW29Ch for lactose at 25 °C was 0.01–0.07 mm and that of the γ sugar-binding site was 2.66 mm (Table 1). Kd for the α sugar-binding site of EW29Ch was almost the same as that for the high-affinity sites of ricin and RCA120, whereas Kd for the γ sugar-binding site of EW29Ch is very similar to that for the third binding site of the ricin B chain and has the lowest binding ability.

Although previous structural studies using X-ray crystallography have clearly shown the mechanism of galactoside-binding to the two binding sites, it is still unclear why one site binds lactose more strongly [2]. In this study, we observed slight differences between the binding modes of the α and γ sugar-binding sites in EW29Ch from the sugar complex structure of EW29Ch [25]. The residue at the α sugar-binding site, Gln22, interacts with GalO2 of lactose, but the corresponding residue in the γ sugar-binding site was not observed. Lys36 in subdomain α is replaced by His120 in subdomain γ, and one hydrogen bond toward the O2 atom was deduced. These results suggested that the α sugar-binding site has a tighter interaction with lactose than the γ sugar-binding site because of the number of intermolecular hydrogen bonds and of residues interacting with lactose. Our results agreed well with those from the crystal structure of the complex between EW29Ch and lactose [25]. However, it remains unclear why the α sugar-binding site binds to lactose much more strongly. Future studies will aim to determine both the refined sugar-free structure and the refined complex structure of EW29Ch with lactose in a solution state by using residual dipolar coupling constants by NMR to analyze the interaction between the protein and lactose in a solution state.

As mentioned above, one of two sugar-binding sites of EW29Ch, the α sugar-binding site, has a tight binding mode, but the γ sugar-binding site has a weak binding mode. This manner of binding reflected the dissociation constants of EW29Ch in previous frontal affinity chromatography analysis and corresponds to that of the higher binding site by NMR, even if the sugar-binding ability of one of the two binding sites is much higher than that of the other. However, the hemagglutinating activity of lectin is related to its multivalency for the cross-linking of cells. This means that the hemagglutinating activity depends on the weaker of the two sugar-binding sites because both must bind to sugars on the surface of cells to cross-link and agglutinate cells. Consequently, EW29Ch retains its hemagglutinating activity but at level 10-fold lower than that of the whole protein, whereas EW29Ch binds to asialofetuin–agarose as strongly as the whole protein.

A common feature of lectins is their multivalent binding properties. As a consequence, lectin binding to cells leads to cross-linking and aggregates of glycoprotein and glycolipid receptors. Thus, the carbohydrate cross-linking properties of lectins are a key feature of their biological activities [30,43,44]. R-type lectins are reported to have physiological functions such as enzyme targeting and glycoprotein hormone turnover [45]. The physiological function of EW29, however, remains unknown. Clarification of the sugar-binding ability and specificity of the two sugar-binding sites, which was determined by NMR titration studies and STD–NMR experiments, is expected to provide clues to understand the precise physiological function of EW29.

Experimental procedures

Sample preparation

15N-labeled or 13C,15N-labeled EW29Ch was expressed and purified using 15N-labeled or doubly labeled CHL medium (Chlorella Industry Co., Tokyo, Japan) as described elsewhere [26]. In this study, purified EW29Ch was dialyzed in distilled water many more times than had been done previously, because a pair of resonance signals in the sugar-free state and the sugar-bound state were observed for some residues in the α subdomain owing to lactose contamination from affinity chromatography using lactose–agarose. The final product contains the full-length 131 amino acid EW29Ch sequence of Lumbricus terrestris (residues Lys130−Glu260 in EW29) [1], with an N-terminal methionine residue (total length of 132 amino acids). Residues are numbered from the N-terminal methionine residue (Met1–Lys2–Pro3…). Residue Lys2 of EW29Ch in this study corresponds to residue Lys130 of EW29 or residue Lys130 of the crystal structure of EW29Ch (PDB: 2ZQN or 2ZQO) [25].

NMR spectroscopy

Purified EW29Ch was dissolved in 50 mm of potassium phosphate buffer (pH 6.1) and a protease inhibitor cocktail (Sigma Chemical Co, St Louis, MO, USA) in either a 90% H2O/10%2H2O mixture or 99.96%2H2O. The final concentration of the protein was 0.9 mm. NMR spectra were acquired at 25 °C on Bruker DRX600 and Avance 800 NMR spectrometers. All 1H dimensions were referenced to internal 4,4-dimethyl-4-silapentane-1-sulfate, and 13C and 15N were indirectly referenced to 4,4-dimethyl-4-silapentane-1-sulfate [46]. All multidimensional NMR spectra were acquired in the phase-sensitive mode using the States–time-proportional phase increment method [47] or the echo-antiecho mode [48]. Shifted sine-bell window functions were applied to NMR data prior to zero-filling and Fourier transformation. NMR data were processed using felix2000 software (Accelrys, San Diego, CA, USA) or the nmrpipe package [49], and analyzed using sparky software (Goddard and Kneller, sparky 3, University of California, San Francisco, CA, USA). 1H, 13C and 15N assignments were obtained from standard multidimensional NMR methods as described elsewhere [26].

Titration of EW29Ch with sugars monitored by NMR

The binding of each of sugar; lactose, melibiose, galactose (all from Wako Chemicals, Tokyo, Japan), α-Me-Gal and β-Me-Gal (both from Seikagaku Co., Tokyo, Japan), to EW29Ch at 25 °C (pH 6.1) was measured quantitatively using 1H–15N HSQC NMR spectroscopy. Each sugar stock solution used in this study was prepared by weight in a sample buffer of 50 mm of potassium phosphate (pH 6.1). Aliquots of these solutions (starting protein concentration of 300–350 μm) were added directly to uniformly 15N-labeled EW29Ch contained in an NMR tube. For each titration, 20 1H–15N HSQC spectra were recorded consecutively with increasing concentrations of each sugar. For the progressive chemical-shift changes of EW29Ch under conditions of fast exchange on the chemical-shift timescale, 15N and 1HN chemical-shift changes in EW29Ch were calculated using the equation Δav = {(ΔNH2 + ΔN2/25)/2}1/2, where ΔNH is the chemical-shift change of the amide proton and ΔN that of the nitrogen [50]. Sugar-binding constants (Kd) were calculating using the Solver function of excel 2002 (Microsoft, Redmond, WA, USA) for the γ sugar-binding site by nonlinear regression fitting of the chemical-shift change versus sugar concentration to the binding isotherm describing the binding of one ligand molecule to a single protein site [27]. Similarly, assuming that sugar binding to EW29Ch is a reversible single-step transition under conditions of slow exchange on the chemical-shift timescale, the dissociation constant, Kd, is given by

image

Here, [P], [L] and [PL] are the respective concentrations of free EW29Ch, free sugar and the EW29Ch–sugar complex. [P]/[PL] ratios were determined as a function of [L] from free and bound peak intensities [51], because the two signals in sugar-free and sugar-bound forms were observed separately under the slow exchange regime. Kd values were also calculated using the Solver function of excel 2002 for the α sugar-binding site by nonlinear regression fitting of the ratio of free and bound peak intensities versus sugar concentration to the binding isotherm. Throughout the titration for calculating Kd values under the slow exchange regime, the concentration of EW29Ch was maintained at 0.05 mm and lactose or β-Me-Gal was added incrementally from 0 to 0.15 mm.

STD–NMR experiments

Non-labeled EW29Ch was expressed using Luria–Bertani medium and purified using the same method as the preparation of labeled EW29Ch [26]. To a sample of EW29Ch in phosphate-buffered solution (50 mm of potassium phosphate buffer pH 6.1 and a protease inhibitor cocktail in 99.96%2H2O) was added lactose or β-Me-Gal. The final sugar concentration was 5 mm at a sugar-to-protein ratio of 100 : 1. 1D and 2D STD–NMR spectra were obtained as described previously [52]. The time dependence of the saturation transfer was determined by recording 1D STD spectra with 1000 scans and saturation times from 0.25 to 6.0 s. The irradiation power in all STD–NMR experiments was set to ∼ 0.15 W. Relative STD values were calculated by dividing STD signal intensities by the intensities of the corresponding signals in a reference spectrum of the same sample recorded with 64 scans. All STD–NMR spectra for epitope mapping were acquired using a series of equally spaced 50 ms Gaussian-shaped pulses for saturation with 1 ms intervals and a total saturation time of ∼ 2 s. On-resonance irradiation of the protein was conducted at a chemical shift of –0.4 p.p.m. and off-resonance at a chemical shift of 30 p.p.m. where no protein signal was present. Free induction decay values with on- and off-resonance protein saturation were recorded in an alternative fashion. Subtraction of the 1D STD spectra was achieved via phase cycling. Protein resonance was suppressed by the application of a 30 ms spin–lock pulse before acquisition. 2D STD–TOCSY and STD–[13C,1H] HSQC spectra at natural abundance with on- and off-resonance protein saturation were recorded with 128 scans or 512 scans per t1 increment in an alternative fashion. The 2D spectra were acquired with spectra widths of 10 p.p.m. in 1H and 80 p.p.m. in 13C, and 128 (t1) and 2048 (t2) complex points or 64 (t1) and 1024 (t2) complex points for STD-TOCSY or STD-[13C, 1H] HSQC spectra. An MLEV mixing time of 100 ms was applied in STD–TOCSY spectra.

Acknowledgements

We thank Drs Rintaro Suzuki and Toshimasa Yamazaki (National Institute of Agrobiological Sciences) and Dr Chojiro Kojima (Nara Institute of Science and Technology) for help in calculating sugar-binding constants from NMR titration data. We also thank Ms Sachiko Unno (AIST) for the preparation of the proteins. This work was supported in part by Grant-in-Aid for Scientific Research (C) (18580342 and 20580373) from the Japan Society for the Promotion of Science (to HH and AK).

Ancillary