Anomer-Specific Recognition and Dynamics in a Fucose-BindingLectinPaweł M. Antonik,†,‡,# Alexander N. Volkov,§,∥,# Ursula N. Broder,† Daniele Lo Re,†

Nico A. J. van Nuland,§,∥ and Peter B. Crowley*,†

†School of Chemistry, National University of Ireland Galway, University Road, Galway, Ireland‡Department of Food BioSciences, Teagasc Food Research Centre, Ashtown, Dublin 15, Ireland§Jean Jeener NMR Centre, Structural Biology Brussels, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium∥Structural Biology Research Centre, VIB, Pleinlaan 2, 1050 Brussels, Belgium

*S Supporting Information

ABSTRACT: Sugar binding by a cell surface ∼29 kDa lectin(RSL) from the bacterium Ralstonia solanacearum wascharacterized by NMR spectroscopy. The complexes formedwith four monosaccharides and four fucosides were studied.Complete resonance assignments and backbone dynamicswere determined for RSL in the sugar-free form and whenbound to L-fucose or D-mannose. RSL was found to interactwith both the α- and the β-anomer of L-fucose and the “fucoselike” sugars D-arabinose and L-galactose. Peak splitting wasobserved for some resonances of the binding site residues. Theassignment of the split signals to the α- or β-anomer was confirmed by comparison with the spectra of RSL bound to methyl-α-L-fucoside or methyl-β-L-fucoside. The backbone dynamics of RSL were sensitive to the presence of ligand, with the proteinadopting a more compact structure upon binding to L-fucose. Taking advantage of tryptophan residues in the binding sites, weshow that the indole resonance is an excellent reporter on ligand binding. Each sugar resulted in a distinct signature of chemicalshift perturbations, suggesting that tryptophan signals are a sufficient probe of sugar binding.

Carbohydrate recognition is a topic of far-reachingimportance1 that continues to challenge the protein

scientist2−6 and the synthetic chemist7−10 alike. The principalchallenge lies in understanding the balance of noncovalentinteractions that favors receptor binding over solvation inwater. Complex formation in the case of monosaccharides hasthe added feature of selection between the α and the βanomeric forms. NMR spectroscopy offers the powerful tools ofchemical shift perturbation (Δδ) analysis to characterizeprotein−ligand binding. Here, we have applied this techniqueto study the fucose-binding lectin11 (RSL) from the bacteriumRalstonia solanacearum. Knowledge of fucose recognition bylectins is expected to aid the development of new strategies tocombat bacterial disease. For example, host−pathogeninteractions may be mediated by lectins that bind to fucosylatedoligosaccharides present in plant cell walls or mammalianepithelial tissue.12 In the case of R. solanacearum, extracellularfucose-binding lectins likely contribute to infection andprogression of wilt disease in potato and tomato.11

RSL is a ∼29 kDa trimer with a six-bladed β-propeller fold.11

Each monomer comprises two blades, corresponding to the N-and C-terminal halves of the protein, which share ∼40%sequence identity (Figure 1). There are two sugar-binding sitesper monomer. Site 1 is intra-monomeric and is nestled betweenthe N- and C-terminal blades of the monomer. Site 2 is inter-

monomeric and occurs between the blades of adjacentmonomers (Figure 1). Each site contains a WiXGXGWi+5motif. The indole ring of Wi forms CH-π bonds13,14 with themonosaccharide, while the indole of Wi+5 donates a hydrogenbond to hydroxyl O3 of fucose.11 The occurrence of two almostidentical binding sites makes RSL an interesting model proteinfor Δδ analysis. The protein is exceptionally stable (Tm = 86 °Cin the sugar-free form),15 which favors NMR studies. There arenumerous crystal structures (including PDB 2BT9 at 0.94 Åresolution) of RSL11 and related lectins15−18 in sugar-free and-bound forms. Furthermore, ligand binding to RSL has beencharacterized by isothermal titration calorimetry11,14 andcomputational methods.14,19,20

The interactions of RSL with four monosaccharides and fourfucosides were investigated by chemical shift perturbationstudies on 15N-labeled RSL. Remarkably, we observed anomer-specific effects with resonance splitting in the presence of L-fucose or related monosaccharides. While many lectin-carbohydrate complexes have been characterized by HSQCspectroscopy,4−6,21−30 we have not found evidence of suchanomer-specific effects in the literature. NMR binding data for

Received: November 9, 2015Revised: February 4, 2016Published: February 4, 2016

Article

pubs.acs.org/biochemistry

© 2016 American Chemical Society 1195 DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

Dow

nloa

ded

via

VRI

JE U

NIV

BRU

SSEL

on

Sept

embe

r 4, 2

018

at 0

8:47

:00

(UTC

). Se

e ht

tps:/

/pub

s.acs

.org

/shar

ingg

uide

lines

for o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blish

ed a

rticl

es.

α- and β-methyl- or -nitrophenyl-L-fucosides together withanalysis of the RSL crystal structure suggest that a Tyr sidechain that flanks the intra-monomeric site may act as a stericimpediment to the binding of β-L-fucosides. The backbonedynamics of the sugar-free and sugar-bound forms of RSL werecharacterized in the nanosecond to second time scales. Ourdata indicate an increased rigidity of the RSL structure uponsugar binding, consistent with a ligand-induced fit. We showalso that the tryptophan indole −NH resonance is an excellentprobe of ligand binding.31,32 In the case of RSL with three Trpside chains per binding site, sugar-specific signatures wereobtained.

■ EXPERIMENTAL SECTIONMaterials. Carbohydrates were purchased from SIGMA (D-

arabinose, L-fucose, L-galactose, D-mannose,) or Carbosynth(methyl-α-L-fucoside, methyl-β-L-fucoside, 4-nitrophenyl-α-L-fucoside and 4-nitrophenyl-β-L-fucoside). 13C-Labeled D-glucose and 15(NH4)2SO4 were purchased from Cortecnet.Protein Production. 15N- and 13C/15N-RSL samples were

produced in Escherichia coli BL21 transformed with the plasmidpET25rsl.11 Cultures were grown to mid-log phase on LB andthen transferred to minimal medium (supplemented with 75mg/mL carbenicillin). For 15N-labeling the medium contained1 g/L 15(NH4)2SO4 as the sole nitrogen source. For 13C/15N-labeling 2 g/L 13C-labeled glucose was added as the sole carbonsource. RSL was purified by affinity chromatography (D-mannose-agarose resin, SIGMA) on an AKTA FPLC, asdescribed previously. Extensive dialysis was required toremoved D-mannose.11 The protein concentration wasdetermined by UV spectroscopy using an extinction coefficientε280 = 44.6 mM−1cm−1.NMR Assignments. The samples contained 2 mM

13C/15N-RSL in 20 mM potassium phosphate, 50 mM NaClpH 6.1 and 6% D2O for the lock. The sugar-bound sampleswere prepared by the addition of D-mannose or L-fucose to afinal concentration of 24 mM (i.e., 2 equiv). NMR experimentswere performed at 303 K on Varian NMR Direct-Drive System600 and 800 MHz spectrometers, the latter equipped with a

salt-tolerant 13C-enhanced PFG-Z cold probe. All NMR datawere processed in NMRPipe33 and analyzed in CCPN.34 Theassignment of the backbone resonances were determined froma set of two-dimensional (2D) 1H−15N HSQC and 3DHNCACB, CBCA(CO)NH, and HNCO experiments.35 Thetryptophan indoles were assigned from a combination of 2D(HB)CB(CGCD)HD36 and three-dimensional (3D) 15N- and13C-NOESY-HSQC spectra, that provide Cβ-Hδ correlationsand strong intraresidue Hδ-Hε NOEs, respectively. Finally, theside chain NH2 resonances of asparagine and glutamineresidues were assigned from the CBCA(CO)NH spectrum bycorrelating with the corresponding Cα/Cβ (Asn) or Cβ (Gln)chemical shifts. The assignments for RSL have been depositedin the BMRB with accession numbers 25950 (bound to D-mannose), 25951 (bound to L-fucose) and 25952 (sugar-free).

Sugar Binding and Chemical Shift Perturbations.Samples contained 0.25 mM 15N-RSL in 20 mM potassiumphosphate, 50 mM NaCl pH 6.0 and 10% D2O. Mono-saccharides or fucosides were added to a final concentration of3 mM. Spectra were acquired on sugar-free RSL and sugar-bound RSL samples at 303 K on an Agilent 600 MHz NMRspectrometer equipped with a HCN cold probe. Thedifferences in chemical shifts (Δδ) between the sugar-freeand the sugar-bound RSL were measured in CCPN,34 and theaverage perturbations were calculated as Δδavg = {[ΔδH2 + (0.2× ΔδN2)]/2}1/2.37

NMR Dynamics. Data were acquired on samples of sugar-free RSL and RSL bound to L-fucose or D-mannose. Thebackbone 15N R1, R2, CPMG R2ex relaxation dispersion, and15N ZZ-exchange experiments were recorded on a 600 MHzspectrometer at 303 K. The 15N R1 and R2 relaxation rates wereobtained from a series of 2D experiments with coherenceselection achieved by pulse field gradients.38 The relaxationparameters and corresponding errors were extracted withCCPN,34 and the transverse relaxation rates were corrected forthe R2ex contribution (see below). The diffusion tensors of thefree and bound RSL were obtained from the experimental R2/R1 values using the program r2r1_diffusion (A. Palmer).Residues with a 15N R2 higher than one standard deviation fromthe average <R2> (obtained for the entire set of the backboneamides) likely exhibit significant internal motions39 and wereexcluded from the analysis. The model selection was performedusing F statistics as described elsewhere.40

The CPMG relaxation dispersion experiments41 wererecorded with 0, 25, 50 (in duplicate), 75, 125, 175, 225,350, 550, 750 (in duplicate), and 1000 Hz pulse repetitionrates. The peak intensities at each repetition rate weredetermined in CCPN, and the subsequent analysis wasperformed in NESSY,42 providing the R2ex values and theirerrors.The 15N ZZ-exchange experiments43 were acquired as a

series of interleaved 2D spectra with mixing times of 0 (induplicate), 20, 50 (in duplicate), 90, 150, 230, 340, 500, 720,1030, and 1500 ms. The ZZ exchange intensity build-up curveswere analyzed using a composite ratio of four peak intensities(two autopeaks, AA and BB, and two cross-peaks, AB and BA)according to eq 1:44

ζΞ = − = ≅tI t I t

I t I t I t I tt k k t( )

( ) ( )( ) ( ) ( ) ( )

AB BA

AA BB AB BA

2on off

2

(1)

Figure 1. (A) The primary structure of RSL comprises an N- and C-terminal half, which share ∼40% sequence identity. Colons and dotsindicate identical and similar residues, respectively, as revealed by (B)a structural alignment. Residues belonging to the intra- andintermonomeric binding sites are shown in gray and black,respectively. The panel on the right shows how the methyl substituentof the sugar is exposed to the solvent. Side chains shown as sticks(except Thr82) make at least one noncovalent bond with the sugar.11

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1196

where I(t) is the corresponding peak intensity at the mixingtime t and kon and koff are, respectively, the association anddissociation rate constants for the exchange process. A good fit(χ2red = 0.00069) was obtained for the A85 data, with ζ = 1.235± 0.022 s−2. For a system in a two-state A ↔ B equilibrium, theequilibrium constant is given by K = [A]/[B] = k1/k−1, wherek1 and k−1 are first-order association and dissociation rateconstants, respectively, and the relative populations of the twospecies, [A]/[B], can be estimated from the ratio of thecorresponding peak intensities in the fully relaxed HSQCspectrum.44 With K = [A]/[B] = k1/k−1 = 0.90 ± 0.12 for theA85 backbone amide resonances of RSL bound to the α- or β-anomers of L-fucose and ζ = k1k−1 = 1.235 ± 0.022 s−2, weobtain k1 = 1.05 ± 0.16 s−1 and k−1 = 1.17 ± 0.20 s−1, yieldingthe final value of the exchange rate constant kex = k1 + k−1 =2.22 ± 0.25 s−1.

■ RESULTS AND DISCUSSIONBackbone Resonance Assignments. The 1H−15N

HSQC spectrum, with a total of 105 well-resolved cross-peaks, confirmed that RSL is a symmetric trimer in solution.Despite the relatively high molecular weight (29.2 kDa) theprotein yielded an excellent HSQC comparable to that of a 10kDa protein. Almost complete 1HN, 15N, C′, Cα, and Cβ

assignments were determined for sugar-free RSL. Figure S1Ashows the assigned 1H−15N HSQC spectrum. Pairs ofhomologous residues, due to the sequence and structuralsimilarity of the N- and C-terminal subdomains, had resonanceswith closely similar chemical shifts and line widths (Figure S1A;S9/S52, I16/I61, R17/R62, W31/W76, D32/D77, W36/W81,and A40/A85, F41/Y86, see also Figure 2). Interestingly, no

peaks were observed for the glycines that occur in theWiXGXGWi+5 motifs (residues 31−36 and 76−81, Figure 1),suggestive of exchange broadening for these turn residues.

Chemical Shift Perturbation Due to Sugar Binding.Monosaccharide/fucoside binding to RSL was in slow-exchangeon the NMR time scale. The presence of D-arabinose, L-fucose,methyl-α-L-fucoside, methyl-β-L-fucoside, nitrophenyl-α-L-fuco-side, nitrophenyl-β-L-fucoside, L-galactose or D-mannose (withKd values ranging from <1−100 μM11,19), resulted in distinctperturbations of the backbone 1HN, 15N resonances (Figure 3).Complete backbone assignments were determined for RSLbound to L-fucose or D-mannose (see Figure S1B,C,respectively for the assigned HSQCs). In the case of theother sugars, only the 1H−15N HSQC spectrum was assigned.The sugars D-arabinose, L-fucose, and L-galactose have

identical stereochemistry and differ only in the nature of theC5 substituent; −H, −CH3, and −CH2OH, respectively(Figure 3). In the crystal structure of RSL bound to methyl-α-L-fucoside11 the C6 methyl is inserted into a shallowhydrophobic pocket formed by the side chains of Trp10,Ile59, Ile61, and Trp76 in the intra-monomeric site or Trp53,Pro14, Ile16, and Trp31 in the inter-monomeric site (Figure 1).In addition to interactions with these side chains the methylsubstituent is also in van der Waals contact with two watermolecules. The nature of this portion of the binding site, inparticular, the water accessibility suggests a molecular basis forthe broad specificity of RSL. In the case of D-arabinose theabsence of the C6 methyl is likely compensated by the insertionof a water molecule (Figure 4B). When L-galactose binds, itshydroxyl substituent can orient toward the solvent and thereare no unfavorable clashes. The tolerance of this hydrophobicpocket to polar substituents is evidenced by the binding of D-mannose. By virtue of their similar stereochemistry, L-fucoseand D-mannose can bind to the same lectins.11,16,45 Specifically,the protein-sugar contacts mediated by hydroxyls O2, O3 andO4 of L-fucose can be established likewise by the O4, O3, andO2 groups of D-mannose (Figure 3). Consequently, when D-mannose binds to RSL its anomeric hydroxyl is inserted intothe hydrophobic pocket. And considering the stereochemistryof C5 in L-fucose it is apparent that RSL must bind β-D-mannose. One might also speculate that the ∼50 fold loweraffinity for D-mannose (compared to L-fucose) is the result of aweakened hydrophobic effect at this site.In the presence of D-arabinose, L-fucose or L-galactose, the

number of resonances in the 1H−15N HSQC spectrum of RSLincreased by ∼10%. The assignment of RSL bound to L-fucoserevealed that the backbone amide resonances of six residues(Y37, F41, K83, G84, A85, and Y86) were split into two clearlyresolved peaks. These pairs of resonances were attributed to thebinding of the α and β forms of L-fucose, which is present insolution in a ratio of 0.3 α-L-fucose to 0.7 β-L-fucose.46 Whenbound to RSL, L-fucose is positioned with the anomerichydroxyl exposed to the solvent.11,15 Thus, the α and β formscan bind with minimal steric clashes between the protein andthe anomeric hydroxyl. Crystallographic evidence for this effectwas reported in a RSL crystal structure (PDB 4I6S), whereboth anomers of L-fucose were bound.15 As such RSL is suitedto binding fucose-terminated oligosaccharides. While the lack ofα/β anomer discrimination or selection is known forlectins,16,17,47 we have not found NMR evidence for this effectin the literature. The assignment of the split peaks to thebinding of the α and β forms was confirmed by comparisonwith the spectra of RSL in complex with methyl-α-L-fucoside or

Figure 2. Spectral regions from the overlaid 1H−15N HSQC spectra ofsugar-free RSL (black contours) and RSL bound to L-fucose (red), tomethyl-α-L-fucoside (navy) or methyl-β-L-fucoside (orange). (A) Anexample of structurally homologous residues, R17 and R62, withresonances of similar chemical shift. (B) Example of a resonance, A85,that is sensitive to both anomers of L-fucose. (C) The seventryptophan indoles are useful reporters of sugar binding (refer toFigure 4).

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1197

methyl-β-L-fucoside (Figure 2). There was good overlapbetween the fucose-split resonances and the single resonancein the fucoside complex. Note that the complex of RSL with D-mannose did not result in peak splitting in the HSQC. Only theβ-D-mannose can bind for reasons already discussed.Pronounced chemical shift changes (Δδavg = 0.4−0.6 ppm)

were measured for the resonances of equivalent residues Ala40and Ala85 (Figure 3). These large effects can be rationalized bycomparison with the X-ray structure.11 The amide NH of Ala40and Ala85 donate a hydrogen bond to hydroxyl O2 of methyl-α-L-fucoside (N---O2, 2.9 Å) in the intra- and inter-monomericbinding sites, respectively.Tryptophan Indole Probes of Sugar Binding. The

pattern of backbone chemical shift perturbations was similar forD-arabinose, L-fucose, L-galactose and D-mannose (Figure 3).Despite the differences in chemical structure, the maininteractions (e.g., hydrogen bonds and CH-π bonds) weremaintained by each sugar and the backbone resonances of RSLsensed similar changes to the chemical environment. Incontrast, the −NH resonance of the tryptophan indole was amore sensitive reporter,31,32 and a unique pattern of shifts wasobserved for each monosaccharide. There are seven trypto-phans in RSL, three in the intra-monomeric site (Trp10, Trp76,and Trp81), and three in the inter-monomeric (Trp53, Trp31,and Trp36). The seventh tryptophan (Trp74) does notcontribute directly to either site.Figure 4A shows the indole −NH perturbations in the

presence of the four sugars. As discussed, the differences inbinding for this set of sugars were focused on the C5

substituent (or C1 in the case of D-mannose, Figure 3). In thecrystal structure, the methyl C6 of methyl-α-L-fucoside makesvan der Waals contacts with the indoles of Trp 31/76 andTrp53/10. The shortest point of contact with C6 is to theindole −NH in the Trp31/76 pair (Figure 4B). Therefore, itcould be expected that these indole resonances will have thelargest differences in perturbation when the various sugars bind.This was clearly the case (Figure 4A, yellow bars). The Trp36/81 pair is furthest from C5, and the indole −NH of these sidechains makes a hydrogen bond to hydroxyl O3, which is likelyto dominate the chemical shift. Consistent with this structuralinterpretation, the Trp36/81 resonances had similar, largeperturbations irrespective of which sugar was bound (Figure4A, red bars). The Trp53/10 pair had similar perturbations forD-arabinose, L-fucose and L-galactose, while shifts of similar sizebut opposite sign were observed for D-mannose (Figure 4A,blue bars). Finally, the indole resonance of Trp74 hadnegligible chemical shift perturbations as this group was notinvolved in sugar binding. As with the backbone amideresonances, some of the indole resonances were also split dueto the binding of α- or β-L-fucose (Figures 2C and 4A). Againthe assignment of the split signals was confirmed bycomparison with the data for methyl-α- or methyl-β-L-fucoside.Considering the sugar-specific signatures observed for the

indole resonances (versus the similar pattern of backboneamide shifts), the use of 15N-tryptophan labeled31 samples(rather than uniform labeling) may provide a simpler probe forthe characterization of sugar binding. An effective application

Figure 3. Δδavg plots of RSL in the presence of 2 equiv of different monosaccharides. Black bars indicate resonances that were split in two due to thepresence of the α- and β-anomer. Blanks correspond to prolines (P14 and P44) and residues which lacked a resonance in the sugar-free or -boundRSL. The structure of each sugar and their Kd values

11,19 are indicated. D-Mannose is oriented to show how hydroxyls O2, O3, and O4 match thestereochemistry of L-fucose.

Figure 4. (A) Plot of chemical shift perturbations for the tryptophan indole resonance in the presence of four monosaccharides (refer to Figure 3).Filled and open circles correspond to the 1HN and 15N data, respectively. Some resonances were split in two (e.g., L-fucose data) due to the presenceof the α- and β-anomer. (B) A simplified representation of the binding site showing the homologous pairs (refer to Figure 1) of tryptophan sidechains as sticks and color-coded to match panel A. The methyl-α-L-fucoside is shown as lines and the protein backbone is represented as a ribbon.Two crystallographic water molecules are indicated by red spheres. Dashed lines indicate the distances between the Trp Nε atoms and C6 or O3 ofmethyl-α-L-fucoside.

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1198

would require two or more binding site tryptophans withstructurally distinct ligand interactions.Steric Effects in the Binding Site. The observation of

distinct chemical shift perturbations for the α- and β-anomersof L-fucose and methyl-L-fucoside prompted further analysis ofthis effect. In the atomic resolution crystal structure of RSL11

(PDB 2BT9) the methyl substituent of methyl-α-L-fucoside issolvent exposed (Figure 1B, right panel). The solvent exposureof the anomeric substituent implies that β-L-fucosides can beaccommodated. However, the side chain of Tyr37 adjacent tothe intra-monomeric binding site may impede the binding ofbulky β-L-fucosides. The corresponding residue in the inter-monomeric site is Thr82 (Figure 1B). Figures 5 and S2 showoverlaid spectral regions of RSL bound to nitrophenyl-α-L-fucoside, nitrophenyl-β-L-fucoside or the corresponding meth-yl-L-fucoside. In some cases the same chemical shiftperturbation (with respect to sugar-free RSL) was observedfor the α- and β-, methyl- or nitrophenyl-L-fucoside. For otherresonances, the β-L-fucosides gave rise to perturbations thatdiffered significantly from the α-L-fucoside shifts. For example,the R62 resonance was similarly altered in the presence of allfour fucosides while the equivalent R17 in the intra-monomericsite was not shifted by the β-anomers. While it is difficult topinpoint chemical shift changes to specific structural effects(due to the proximity of the intra- and inter-monomeric sites)it seems reasonable to assume that steric interactions betweenthe β-substituent and the Tyr37 side chain (and the lack of suchan interaction at Thr82) results in a different contribution tothe chemical shift perturbations of binding site resonances. Arole for tyrosine side chains in lectin selectivity has beenproposed previously, for example, the tyrosine gate in fimH.5,6

Perhaps, the single Tyr37 serves to modulate the bindingaffinity of the three intra-monomeric sites in RSL.Dynamics of Sugar-Free and Sugar-Bound RSL. The

dynamics of the RSL backbone on the ps-ns time scale wereassessed by using the residue-based squared-order parameter(S2).48 With an average S2 = 0.82 ± 0.08, the sugar-free RSLbackbone was rigid, except for the homologous loop regionscomprising residues 32−35 and 77−80 (Figure 6, upper panelsand compare Figure 1). These considerably flexible (S2 < 0.75)loops contain two glycines (no amide resonances detected) andconnect two of the sugar-binding tryptophans in the intra- andinter-monomeric binding sites. The flexibility of these loopswas not affected by binding to either L-fucose or D-mannose.Residues 40−44 and 64−72 exhibited minor changes in theorder parameters (|ΔS2| < 0.05), while S2 for the rest of thebackbone remained constant. These data suggest that sugarbinding did not affect the RSL backbone dynamics on the ps-nstime scale.To complement the S2 analysis, 15N R1 and R2 relaxation

parameters were also measured. Governed by the rotationaldiffusion of a molecule, the R1 and R2 relaxation rates canprovide information on the hydrodynamic properties of theprotein. A small, uniform decrease in the R2/R1 ratios wasobserved for the sugar-bound forms of RSL compared to sugar-free RSL (Figure 6, middle panels). Quantitative analysis of theR2/R1 profiles yielded isotropic diffusion tensors with rotationalcorrelation time τc = 12.42 or 11.75 ns for the sugar-free and-bound RSL, respectively.49 These data are consistent with anRSL trimer in solution and suggest that the protein adopts amore compact structure upon sugar binding.RSL dynamics were probed further by Carr−Purcell−

Meiboom−Gill (CPMG) relaxation dispersion spectrosco-

py.50,51 In this experiment, protein groups that undergoconformational exchange on the μs-ms time scale give rise torelaxation dispersion curves, which provide an exchangecontribution (R2ex) to the transverse relaxation rate R2. Insugar-free RSL, ∼40% of residues experienced R2ex > 0 (Figure6, lower panels), suggesting that the protein was in conforma-tional exchange between two or more lowly populated species.

Figure 5. Regions from the overlaid 1H−15N HSQC spectra of sugar-free RSL (black contours) and RSL bound to methyl-α-L-fucoside(navy), methyl-β-L-fucoside (dark orange), nitrophenyl-α-L-fucoside(light blue) or nitrophenyl-β-L-fucoside (light orange). Thesimilarities/dissimilarities in chemical shift perturbations due tosugar binding are evident. β substituents resulted in a distinct pattern(see C75, T82, and G84).

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1199

The clustering of these exchange effects suggests the possibilityof concerted motions. Mannose binding moderately altered thedistribution of the R2ex terms, with a reduction in the number ofresidues with R2ex contributions. In contrast, fucose bindingabolished the R2ex terms for most of the protein. Such a drasticdecrease in R2ex values suggests that the conformationaldynamics observed in sugar-free RSL were quenched uponfucose binding. These data, together with the τc results, indicatean increased rigidity of the sugar-bound protein.1,21,27

Finally, by using 15N ZZ-exchange spectroscopy,51 westudied the slow dynamics of RSL bound to the α- and β-anomer of L-fucose. In this experiment, magnetization arisingfrom species in slow exchange (1 s−1 ≤ kex ≤ 103 s−1) gives riseto symmetry-related cross-peaks. The buildup of the cross-peakintensities, followed in a series of spectra recorded with varyingmixing time, can be used to extract the kinetic parameters. Atlong mixing times (0.5−1 s), the resonances of RSL bound toα- or β-L-fucose featured distinct cross-peaks (Figure 7),indicating that the two bound forms were in exchange. Only theA85 resonances were resolved sufficiently for quantitativeanalysis (Figure 7C) and an exchange constant kex = 2.22 ±0.25 s−1 was obtained (see Experimental Section). Consistentwith the μM binding affinity11 the NMR data suggest that theα- and β-L-fucose-bound forms of RSL undergo very slowexchange.

The dynamics of sugar-binding proteins have been assessedin numerous studies, some of which have pointed to a role forconformational entropy.5,6,24,25 In some examples, it was foundthat the sugar-bound form was more dynamic than the sugar-free form, with the entropy gain being thermodynamicallyfavorable for ligand binding.25 In the case of RSL, the proteindynamics were reduced in the sugar-bound form. Upon bindingto L-fucose the protein became more compact (decreased τc)and motions on the μs-ms time scale were frozen out.Interestingly, the fast dynamics analysis revealed flexible loops(S2 < 0.75) adjacent to both the intra- and inter-monomericbinding sites. Here, the flexibility was not affected by binding tomonosaccharides. These loops extend away from the bindingsite and may play a structural role in RSL interactions with cellsurface oligosaccharides.

■ CONCLUSIONSUsing NMR spectroscopy, we have shown that the bacteriallectin RSL binds both the α- and the β-anomer of L-fucose and“fucose like” sugars. Anomer-specific chemical shift perturba-tions were observed for methyl- and nitrophenyl-substituted L-fucosides. A 15N ZZ-exchange experiment confirmed that RSLbinds to α- or β-L-fucose, which exchange on the second timescale. This appears to be the first example of such anomer-specific interactions measured by NMR. Analysis of the atomic

Figure 6. Squared order parameters (S2), fast dynamics (15N R2/R1 profiles), and exchange contribution to the transverse relaxation rates (R2ex) ofRSL in the sugar-free and -bound forms. S2 values were calculated in RCI48 from the backbone amide secondary chemical shifts. Note the flexible (S2

< 0.75) loop regions comprising residues 32−35 and 77−80. The dashed lines are the averages for the sugar-free data. The RSL secondary structureis indicated.

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1200

resolution crystal structure suggests further that the accessibilityof the intra- and inter-monomeric binding sites is different. ATyr side chain (Tyr37) adjacent to the intra-monomeric mayact as a steric impediment to the binding of β-L-fucosides withbulky substituents. The relaxation data reveal that RSL adopts amore compact structure when bound to L-fucose, suggesting aligand-induced fit. The protein compaction due to sugar-binding was accompanied by decreased backbone motions onthe μs-ms time scale. Finally, in addition to the conventionalanalysis of backbone amide chemical shifts, we show that thetryptophan indole resonance is an excellent reporter of ligandbinding, with characteristic Δδ signatures observed for eachsugar.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.bio-chem.5b01212.

Assigned 1H−15N HSQC spectra of RSL in the sugar-free and -bound forms. Plots of chemical shiftperturbations due to sugar binding. 1H−15N HSQCspectra showing α- and β-anomer effects. Structuralrepresentation of RSL showing the R2ex data (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: peter.crowley@nuigalway.ie. Phone: +353 91 49 2480.Author Contributions#P.M.A and A.N.V. contributed equally.FundingThis research was supported by Teagasc (Walsh fellowship toP.M.A.), FWO (postdoctoral fellowship to A.N.V.), MarieCurie Action (Grant PIEF-GA-2011-299042 to D.L.R.) theHercules Foundation (infrastructure grant to Jean Jeener NMRCenter), NUI Galway (NMR infrastructure support andMillennium Fund to P.B.C.) and Science Foundation Ireland(Grant 10/RFP/BIC2807 to P.B.C.).NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe thank M. Wimmerova (Masaryk University) for providingthe pET25rsl plasmid and E. Tishchenko (Agilent Technolo-gies) for setting up the ZZ exchange and relaxation dispersionexperiments.

■ REFERENCES(1) Gabius, H.-J., Andre, S., Jimenez-Barbero, J., Romero, A., andSolis, D. (2011) From lectin structure to functional glycomics:principles of the sugar code. Trends Biochem. Sci. 36, 298−313.(2) Fiege, B., Rademacher, C., Cartmell, J., Kitov, P. I., Parra, F., andPeters, T. (2012) Molecular Details of the Recognition of BloodGroup Antigens by a Human Norovirus as Determined by STD NMRSpectroscopy. Angew. Chem., Int. Ed. 51, 928−932.(3) de Bentzmann, S., Varrot, A., and Imberty, A. (2014) MonitoringLectin Interactions with Carbohydrates. Methods Mol. Biol. 1149, 403−414.(4) Pederson, K., Mitchell, D. A., and Prestegard, J. H. (2014)Structural Characterization of the DC-SIGN−LewisX Complex.Biochemistry 53, 5700−5709.(5) Vanwetswinkel, S., Volkov, A. N., Sterckx, Y. G. J., Garcia-Pino,A., Buts, L., Vranken, W. F., Bouckaert, J., Roy, R., Wyns, L., and vanNuland, N. A. J. (2014) Study of the Structural and Dynamic Effects inthe FimH Adhesin upon α-d-Heptyl Mannose Binding. J. Med. Chem.57, 1416−1427.(6) Fiege, B., Rabbani, S., Preston, R. C., Jakob, R. P., Zihlmann, P.,Schwardt, O., Jiang, X., Maier, T., and Ernst, B. (2015) The TyrosineGate of the Bacterial Lectin FimH: A Conformational Analysis byNMR Spectroscopy and X-ray Crystallography. ChemBioChem 16,1235−1246.(7) Segura, M., Bricoli, B., Casnati, A., Munoz, E. M., Sansone, F.,Ungaro, R., and Vicent, C. (2003) A Prototype Calix[4]arene-BasedReceptor for Carbohydrate Recognition Containing Peptide andPhosphate Binding Groups. J. Org. Chem. 68, 6296−6303.(8) Ferrand, Y., Crump, M. P., and Davis, A. P. (2007) A syntheticlectin analog for biomimetic disaccharide recognition. Science 318 (80),619−622.(9) Jang, Y., Natarajan, R., Ko, Y. H., and Kim, K. (2014)Cucurbit[7]uril: A High-Affinity Host for Encapsulation of AminoSaccharides and Supramolecular Stabilization of Their α-Anomers inWater. Angew. Chem., Int. Ed. 53, 1003−1007.

Figure 7. Slow dynamics of RSL bound to L-fucose. Selected spectralregions of the 15N ZZ-exchange experiments acquired with tmix = 0 s(black), 0.23 s (blue), and 0.5 s (red) showing the resonances of (A)W81 and W36 (indoles) and (B) A85. Dashed rectangles connect thefour exchange cross-peaks. Note the intensity decrease of theautopeaks (AA, BB) and the concomitant increase of the cross-peaks(AB, BA) at longer mixing times. (C) Quantitative analysis of thebuild-up curve for the exchange cross-peaks of A85. See ExperimentalSection for the fit parameters.

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1201

(10) Chandramouli, N., Ferrand, Y., Lautrette, G., Kauffmann, B.,Mackereth, C. D., Laguerre, M., Dubreuil, D., and Huc, I. (2015)Iterative design of a helically folded aromatic oligoamide sequence forthe selective encapsulation of fructose. Nat. Chem. 7, 334−341.(11) Kostlanova, N., Mitchell, E. P., Lortat-Jacob, H., Oscarson, S.,Lahmann, M., Gilboa-Garber, N., Chambat, G., Wimmerova, M., andImberty, A. (2005) The fucose-binding lectin from Ralstoniasolanacearum - A new type of beta-propeller architecture formed byoligomerization and interacting with fucoside, fucosyllactose, and plantxyloglucan. J. Biol. Chem. 280, 27839−27849.(12) Esko, J. D., and Sharon, N. (2009) Microbial Lectins:Hemagglutinins, Adhesins, and Toxins, in Essentials of Glycobiology(Varki, A., Cummings, R. D., Esko, J. D., Freeze, H. H., Stanley, P.,Bertozzi, C. R., Hart, G. W., and Etzler, M. E., Eds.) 2nd ed., pp 489−500, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.(13) Asensio, J. L., Arda, A., Canada, F. J., and Jimenez-Barbero, J.(2013) Carbohydrate-Aromatic Interactions. Acc. Chem. Res. 46, 946−954.(14) Wimmerova, M., Kozmon, S., Necasova, I., Mishra, S. K.,Komarek, J., Koca, J. (2012) Stacking Interactions betweenCarbohydrate and Protein Quantified by Combination of Theoreticaland Experimental Methods. PLoS One 7.e4603210.1371/journal.-pone.0046032(15) Arnaud, J., Claudinon, J., Trondle, K., Trovaslet, M., Larson, G.,Thomas, A., Varrot, A., Romer, W., Imberty, A., and Audfray, A.(2013) Reduction of Lectin Valency Drastically Changes GlycolipidDynamics in Membranes but Not Surface Avidity. ACS Chem. Biol. 8,1918−1924.(16) Wimmerova, M., Mitchell, E., Sanchez, J. F., Gautier, C., andImberty, A. (2003) Crystal structure of fungal lectin - Six-bladed beta-propeller fold and novel fucose recognition mode for Aleuria aurantialectin. J. Biol. Chem. 278, 27059−27067.(17) Fujihashi, M., Peapus, D. H., Kamiya, N., Nagata, Y., and Miki,K. (2003) Crystal Structure of Fucose-Specific Lectin from Aleuriaaurantia Binding Ligands at Three of Its Five Sugar Recognition Sites.Biochemistry 42, 11093−11099.(18) Audfray, A., Claudinon, J., Abounit, S., Ruvoen-Clouet, N.,Larson, G., Smith, D. F., Wimmerova, M., Le Pendu, J., Romer, W.,Varrot, A., and Imberty, A. (2012) Fucose-binding Lectin fromOpportunistic Pathogen Burkholderia ambifaria Binds to Both Plantand Human Oligosaccharidic Epitopes. J. Biol. Chem. 287, 4335−4347.(19) Mishra, S. K., Adam, J., Wimmerova, M., and Koca, J. (2012) InSilico Mutagenesis and Docking Study of Ralstonia solanacearum RSLLectin: Performance of Docking Software To Predict SaccharideBinding. J. Chem. Inf. Model. 52, 1250−1261.(20) Mishra, S. K., Calabro, G., Loeffler, H. H., Michel, J., and Koca,J. (2015) Evaluation of Selected Classical Force Fields for AlchemicalBinding Free Energy Calculations of Protein-Carbohydrate Com-plexes. J. Chem. Theory Comput. 11, 3333−3345.(21) Canales-Mayordomo, A., Fayos, R., Angulo, J., Ojeda, R.,Martin-Pastor, M., Nieto, P. M., Martin-Lomas, M., Lozano, R.,Gimenez-Gallego, G., and Jimenez-Barbero, J. (2006) Backbonedynamics of a biologically active human FGF-1 monomer, complexedto a hexasaccharide heparin-analogue, by N-15 NMR relaxationmethods. J. Biomol. NMR 35, 225−239.(22) Li, J., Wu, H., Hong, J., Xu, X., Yang, H., Wu, B., Wang, Y., Zhu,J., Lai, R., Jiang, X., Lin, D., Prescott, M. C., Rees, H. H. (2008)Odorranalectin Is a Small Peptide Lectin with Potential for DrugDelivery and Targeting. PLoS One 3.e238110.1371/journal.-pone.0002381(23) Vakonakis, I., Langenhan, T., Promel, S., Russ, A., and Cambell,I. D. (2008) Solution structure and sugar-binding mechanism ofmouse latrophilin-1 RBL: a 7TM receptor-attached lectin-like domain.Structure 16, 944−953.(24) Nesmelova, I. V., Ermakova, E., Daragan, V. A., Pang, M.,Menendez, M., Lagartera, L., Solís, D., Baum, L. G., and Mayo, K. H.(2010) Lactose Binding to Galectin-1 Modulates Structural Dynamics,Increases Conformational Entropy, and Occurs with ApparentNegative Cooperativity. J. Mol. Biol. 397, 1209−1230.

(25) Diehl, C., Engstrom, O., Delaine, T., Hakansson, M., Genheden,S., Modig, K., Leffler, H., Ryde, U., Nilsson, U. J., and Akke, M. (2010)Protein Flexibility and Conformational Entropy in Ligand DesignTargeting the Carbohydrate Recognition Domain of Galectin-3. J. Am.Chem. Soc. 132, 14577−14589.(26) Miller, M. C., Ribeiro, J. P., Roldos, V., Martin-Santamaria, S.,Canada, F. J., Nesmelova, I. A., Andre, S., Pang, M., Klyosov, A. A.,Baum, L. G., Jimenez-Barbero, J., Gabius, H. J., and Mayo, K. H.(2011) Structural aspects of binding of alpha-linked digalactosides tohuman galectin-1. Glycobiology 21, 1627−1641.(27) Ortega, G., Castano, D., Diercks, T., and Millet, O. (2012)Carbohydrate Affinity for the Glucose−Galactose Binding Protein IsRegulated by Allosteric Domain Motions. J. Am. Chem. Soc. 134,19869−19876.(28) Ermakova, E., Miller, M. C., Nesmelova, I. V., Lopez-Merino, L.,Berbis, M. A., Nesmelov, Y., Tkachev, Y. V., Lagartera, L., Daragan, V.A., Andre, S., Canada, F. J., Jimenez-Barbero, J., Solis, D., Gabius, H. J.,and Mayo, K. H. (2013) Lactose binding to human galectin-7 (p53-induced gene 1) induces long-range effects through the proteinresulting in increased dimer stability and evidence for positivecooperativity. Glycobiology 23, 508−523.(29) Probert, F., Whittaker, S. B. M., Crispin, M., Mitchell, D. A., andDixon, A. M. (2013) Solution NMR analyses of the C-typecarbohydrate recognition domain of DC-SIGNR protein revealdifferent binding modes for HIV-derived oligosaccharides and smallerglycan fragments. J. Biol. Chem. 288, 22745−22757.(30) Canales, A., Mallagaray, A., Berbís, M. A., Navarro-Vazquez, A.,Domínguez, G., Canada, F. J., Andre, S., Gabius, H.-J., Perez-Castells,J., and Jimenez-Barbero, J. (2014) Lanthanide-Chelating CarbohydrateConjugates Are Useful Tools To Characterize CarbohydrateConformation in Solution and Sensitive Sensors to DetectCarbohydrate−Protein Interactions. J. Am. Chem. Soc. 136, 8011−8017.(31) Rodriguez-Mias, R. A., and Pellecchia, M. (2003) Use ofSelective Trp Side Chain Labeling To Characterize Protein−Proteinand Protein−Ligand Interactions by NMR Spectroscopy. J. Am. Chem.Soc. 125, 2892−2893.(32) Bista, M., Kowalska, K., Janczyk, W., Domling, A., and Holak, T.A. (2009) Robust NMR Screening for Lead Compounds UsingTryptophan-Containing Proteins. J. Am. Chem. Soc. 131, 7500−7501.(33) Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., andBax, A. (1995) NMRPipe: A multidimensional spectral processingsystem based on UNIX pipes. J. Biomol. NMR 6, 277−293.(34) Vranken, W. F., Boucher, W., Stevens, T. J., Fogh, R. H., Pajon,A., Llinas, P., Ulrich, E. L., Markley, J. L., Ionides, J., and Laue, E. D.(2005) The CCPN data model for NMR spectroscopy: Developmentof a software pipeline. Proteins: Struct., Funct., Genet. 59, 687−696.(35) Sattler, M., Schleucher, J., and Griesinger, C. (1999)Heteronuclear multidimensional NMR experiments for the structuredetermination of proteins in solution employing pulsed field gradients.Prog. Nucl. Magn. Reson. Spectrosc. 34, 93−158.(36) Yamazaki, T., Forman-Kay, J. D., and Kay, L. E. (1993) Two-dimensional NMR experiments for correlating carbon-13.beta. andproton.delta./.epsilon. chemical shifts of aromatic residues in 13C-labeled proteins via scalar couplings. J. Am. Chem. Soc. 115, 11054−11055.(37) Williamson, M. P. (2013) Using chemical shift perturbation tocharacterise ligand binding. Prog. Nucl. Magn. Reson. Spectrosc. 73, 1−16.(38) Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal, S. M., Kay,C. M., Gish, G., Shoelson, S. E., Pawson, T., Formankay, J. D., andKay, L. E. (1994) Backbone Dynamics of a Free and aPhosphopeptide-Complexed Src Homology 2 Domain Studied by15N NMR Relaxation. Biochemistry 33, 5984−6003.(39) Rule, G. S., and Hitchens, T. K. (2006) Fundamentals of proteinNMR Spectroscopy, in Focus on Structural Biology, Springer,Dordrecht, Netherlands.(40) Mandel, A. M., Akke, M., and Palmer, A. G., III (1995)Backbone Dynamics of Escherichia coli Ribonuclease HI: Correlations

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1202

with Structure and Function in an Active Enzyme. J. Mol. Biol. 246,144−163.(41) Palmer III, A. G., Kroenke, C. D., and Loria, J. P. (2001)Nuclear magnetic resonance methods for quantifying microsecond-to-millisecond motions in biological macromolecules, in Methods inEnzymology (Thomas, L. James, V. D., and Uli, S., Eds.) pp 204−238.Academic Press, New York.(42) Bieri, M., and Gooley, P. (2011) Automated NMR relaxationdispersion data analysis using NESSY. BMC Bioinf. 12, 421.(43) Farrow, N., Zhang, O., Forman-Kay, J., and Kay, L. (1994) Aheteronuclear correlation experiment for simultaneous determinationof 15N longitudinal decay and chemical exchange rates of systems inslow equilibrium. J. Biomol. NMR 4, 727−734.(44) Miloushev, V. Z., Bahna, F., Ciatto, C., Ahlsen, G., Honig, B.,Shapiro, L., and Palmer, A. G., III (2008) Dynamic Properties of aType II Cadherin Adhesive Domain: Implications for the Mechanismof Strand-Swapping of Classical Cadherins. Structure 16, 1195−1205.(45) Lis, H., and Sharon, N. (1998) Lectins: Carbohydrate-specificproteins that mediate cellular recognition. Chem. Rev. 98, 637−674.(46) Park, D., Ryu, K.-S., Choi, D., Kwak, J., and Park, C. (2007)Characterization and role of fucose mutarotase in mammalian cells.Glycobiology 17, 955−962.(47) Loris, R., Tielker, D., Jaeger, K.-E., and Wyns, L. (2003)Structural Basis of Carbohydrate Recognition by the Lectin LecB fromPseudomonas aeruginosa. J. Mol. Biol. 331, 861−870.(48) Berjanskii, M. V., and Wishart, D. S. (2005) A Simple MethodTo Predict Protein Flexibility Using Secondary Chemical Shifts. J. Am.Chem. Soc. 127, 14970−14971.(49) Jarymowycz, V. A., and Stone, M. J. (2006) Fast time scaledynamics of protein backbones: NMR relaxation methods, applica-tions, and functional consequences. Chem. Rev. 106, 1624−1671.(50) Loria, J. P., Berlow, R. B., and Watt, E. D. (2008)Characterization of enzyme motions by solution NMR relaxationdispersion. Acc. Chem. Res. 41, 214−221.(51) Kleckner, I. R., and Foster, M. P. (2011) An introduction toNMR-based approaches for measuring protein dynamics. Biochim.Biophys. Acta, Proteins Proteomics 1814, 942−968.

Biochemistry Article

DOI: 10.1021/acs.biochem.5b01212Biochemistry 2016, 55, 1195−1203

1203