Affinities of human histo-blood group antigens for norovirus capsid protein complexes

Abstract

The binding profiles of many human noroviruses (huNoVs) for human histo-blood group antigens have been characterized. However, quantitative-binding data for these important virus–host interactions are lacking. Here, we report on the intrinsic (per binding site) affinities of HBGA oligosaccharides for the huNoV VA387 virus-like particles (VLPs) and the associated subviral P particles measured using electrospray ionization mass spectrometry. The affinities of 13 HBGA oligosaccharides, containing A, B and H epitopes, with variable sizes (disaccharide to tetrasaccharide) and different precursor chain types (types 1, 2, 3, 5 and 6), were measured for the P particle, while the affinities of the A and B trisaccharides and A and B type 6 tetrasaccharides for the VLP were determined. The intrinsic affinities of the HBGA oligosaccharides for the P particle range from 500 to 2300 M⁻¹, while those of the A and B trisaccharides and the A and B type 6 tetrasaccharides for the VLP range from 1000 to 4000 M⁻¹. Comparison of these binding data with those measured previously for the corresponding P dimer reveals that the HBGA oligosaccharides tested exhibit similar intrinsic affinities for the P dimer and P particle. The intrinsic affinities for the VLP are consistently higher than those measured for the P particle, but within a factor of three. While the cause of the subtle differences in HBGA oligosaccharide affinities for the P dimer and P particle and those for the VLP remains unknown, the present data support the use of P dimers or P particles as surrogates to the VLP for huNoV-receptor-binding studies.

Introduction

Human noroviruses (huNoVs), which are the predominant cause of acute gastroenteritis outbreaks worldwide (Patel et al. 2008), is a genus of non-enveloped, single-stranded +RNA viruses in the Caliciviridae family. As there are no in vitro cell culture systems or suitable animal models available for huNoVs, the characterization of their structures and receptor interactions has relied on recombinant forms of their major capsid protein (VP1). For example, recombinant VP1, expressed using the baculovirus system, spontaneously assembles into a virus-like particle (VLP) that is devoid of genomic RNA for infection and replication but is structurally and antigenically indistinguishable from the authentic huNoV (Jiang et al. 1992). X-ray crystallography performed on the Norwalk VLP revealed that the intact particle is composed of 180 copies of VP1, which form a T = 3 icosahedral virion (Prasad et al. 1999). The formation of a smaller T = 1 viral capsid, consisting of 60 copies of VP1, has also been reported (White et al. 1997).

Production of VLPs using the baculovirus cell system is expensive and time consuming. Consequently, alternative protein complexes that can act as surrogates to VLPs are desirable. VP1 consists of two major domains bound by a flexible peptide linker, the N-terminal shell (S) domain and the protrusion (P) domain at the C-terminus (Prasad et al. 1999). The interior S domain is critical to maintaining the icosahedral structure of the virion, whereas the P domain forms a dimeric structure that is located on the outer surface and is implicated in the virus-receptor recognition process and, thus, cell entry. Expression of the P domain in E. coli has been shown to produce homodimers, called P dimers (Tan et al. 2004). The P dimers can also assemble into larger complexes, a 12-mer small P particle (Tan et al. 2011) and a 24-mer P particle (Tan and Jiang 2005b; Tan et al. 2008). A recent native electrospray ionization mass spectrometry (ESI-MS) study revealed that, in 100 mM ammonium acetate (pH 7.4), the P particle is made up of ∼85% of 24-mer and 15% of 18-mer (Bereszczak et al. 2012). Importantly, both the subviral particles and the P dimer are believed to retain the authentic antigenicity and receptor-binding capability of the VLP (Tan et al. 2004; Tan and Jiang 2005b; Tamminen et al. 2012) and are, therefore, seen as attractive substitutes to VLPs for investigating the nature of huNoV–host cell interactions and discovering potential inhibitors. However, to our knowledge, a quantitative comparison of the receptor-binding properties of a huNoV VLP and its corresponding P particle and P dimer has yet to be carried out.

It is well established that many huNoVs recognize human histo-blood group antigens (HBGAs), which are found on the surfaces of red blood cells and mucosal epithelial cells in the form of glycoproteins and glycolipids (Oriol 1990; Ravn and Dabelsteen 2000), as cellular receptors or attachment factors (Hutson et al. 2002, 2003; Huang et al. 2005; Tan and Jiang 2005a). The HBGAs are divided into four types, namely A, B, H and Lewis, based on the carbohydrate structure at the non-reducing end. Additionally, each HBGA is further divided into six subtypes (types 1–6) based on the carbohydrates structure at the reducing end. To date, there have been few quantitative-binding studies performed on the capsid proteins of huNoVs. Using saturation-transfer difference nuclear magnetic resonance (STD-NMR) spectroscopy, Peters and coworkers (Fiege et al. 2012) estimated the apparent association constants (K_a,app) of VLP from huNoV Ast6139 (GII.4 strain) for a variety of HBGA oligosaccharides (including H disaccharide, A and B trisaccharides, H types 1, 2 and 6 trisaccharides, Lewis a, b, x, y and sialyl-Lewis a and x oligosaccharides) to be in the ∼10⁴ M⁻¹ range. Based on these results, the intrinsic (per binding site) association constants (K_a,int) are predicted to be in the ∼10² M⁻¹ range. Using surface plasmon resonance spectroscopy, Belliot and colleagues (de Rougemont et al. 2011) analyzed BSA-conjugated type H, A and B carbohydrates binding to VLPs isolated from six GII.4 strains of huNoV. However, affinities could not be accurately determined due to the uncertainty in the number of active-binding sites in each VLP, as well as the number of immobilized (on the chip) carbohydrates that participated in interactions. Quantitative K_a,app values have also been reported for huNoV P dimers. Using STD-NMR spectroscopy, Kwong and co-workers (Hansman et al. 2012) measured K_a,app values of 2200 and 2600 M⁻¹ for fucose and H type 2 trisaccharide, respectively, binding to a GII.10 huNoV P dimer. More recently, the K_a,int values of a library of 42 HBGA oligosaccharides for the huNoV VA387 P dimer were measured using the direct ESI-MS assay (Han et al. 2013). The results of this study revealed that the P dimer exhibits broad specificity and binds to A, B, H and Lewis type antigens, although with low affinities (≤3000 M⁻¹).

Here, we report K_a,int values of variety of HBGA oligosaccharides for the P particle and VLP of huNoV VA387. Application of the direct ESI-MS assay, which was used for the P dimer measurements, to these large capsid protein complexes is not feasible due to the difficulty in resolving the free and ligand-bound forms of the protein complexes. Consequently, an adaptation of the proxy protein ESI-MS method (El-Hawiet et al. 2012) was used to carry out the measurements. The proxy protein ESI-MS method combines direct ESI-MS-binding measurements and competitive protein binding to evaluate protein–ligand affinities. Specifically, a proxy protein (P_proxy), which binds to the ligand of interest with known affinity and can be detected directly by ESI-MS, is used to quantitatively monitor the extent of ligand binding to the protein of interest—the P particle or VLP in the present case. Using this method, the K_a,int values of 13 HBGA oligosaccharides, containing A, B and H epitopes, with variable sizes (disaccharide to tetrasaccharide) and different precursor chain types (types 1, 2, 3, 5 and 6), for the P particle were measured; affinities for the A and B trisaccharides and A and B type 6 tetrasaccharides for the VLP were also quantified. Comparison of these values with binding data recently reported for the P dimer of huNoV VA387 provides a unique opportunity to assess the similarity of the HBGA-binding sites in the P dimer, P particle and VLP.

Experimental section

Proteins

For VLP production, the gene encoding the capsid protein (VP1) of huNoV VA387 (GII.4, GenBank accession no. AY038600) was cloned and expressed through the Bac-to-Bac baculovirus expression system (Invitrogen Life Technologies, Grand Island, NY) as described elsewhere (Jiang et al. 2002; Huang et al. 2003). VLPs, comprise 180 copies of VP1 [which has a molecular weight (MW) of 58,887 Da], assembled spontaneously. For P particle and P dimer production, the gene fragments encoding the P domain (residues 222–539) of VP1 with and without a C-terminus-fused peptide CDCRGDCFC, respectively, were cloned and expressed in bacteria through the GST-Gene Fusion System (GE-Healthcare Life Sciences, Piscataway, NJ) following a protocol described elsewhere (Tan et al. 2004; Tan and Jiang 2005b). The resulting P dimer (MW 69,312 Da) or P particle (24-mer, MW 865,036 Da) that assembled spontaneously was purified using glutathione affinity chromatography, followed by gel-filtration chromatography.

The N-terminal family 51 carbohydrate-binding module (CBM; expected MW 20,735 Da) from Streptococcus pneumonia SP3-BS71 GH98 was recombinantly produced in Escherichia coli and purified by Ni²⁺ immobilized metal affinity chromatography (GE-Healthcare Life Sciences) using procedures described elsewhere (Higgins et al. 2011). A recombinant fragment of the C-terminus of human galectin-3 (Gal-3C; MW 16,330 Da) was a gift from Prof. C. Cairo (University of Alberta). A recombinant soluble fragment of α-(1→3)N-acetyl galactosaminyltransferase (GTA; homodimer, MW 69,040 Da), which contains a full C-terminal and catalytic domain, as well as a truncated N-terminal domain, was expressed in bacterial cells and purified by ion exchange chromatography using an SP-Sepharose FF resin (GE-Healthcare Life Sciences), followed by affinity purification using a UDP-hexanolamine resin (Seto et al. 1997). Bovine ubiquitin (Ubq; MW 8565 Da), which was purchased from Sigma-Aldrich Canada (Oakville, Canada), and recombinant single chain fragment (scFv; MW 26,539 Da) of monoclonal antibody Se155-4, which was produced using procedures described before (Zdanov et al. 1994), served as reference proteins (P_ref) for the binding measurements. Each protein was dialyzed and concentrated against 50 mM aqueous ammonium acetate (pH 7) using Vivaspin 0.5 mL centrifugal filters (Sartorius Stedim Biotech, Göttingen, Germany) with a 10-kDa MW cutoff. The concentrations were measured by UV spectroscopy. Protein stock solutions were stored at −80°C until used.

Carbohydrates

The oligosaccharides (L1–L14) were a gift from Prof. T. Lowary (University of Alberta) (Meloncelli and Lowary 2009, 2010; Meloncelli et al., 2011). Their structures are shown in Supplementary data, Figure S1. To prepare stock solutions, solid sample of each compound was weighed and dissolved in a known volume of ultrafiltered water (Milli-Q, Millipore, Billerica, MA) to yield a final concentration of 1 mM. These solutions were stored at −20°C until needed.

Mass spectrometry

The ESI-MS measurements were carried out in positive ion mode using either a Synapt G2S quadrupole-ion mobility separation-time of flight mass spectrometer (Waters, Manchester, UK) or a 9.4T ApexQe Fourier-transform ion cyclotron resonance (FT-ICR) mass spectrometer (Bruker-Daltonics, Billerica, MA). The Synapt G2S mass spectrometer, with its high mass capabilities, was used for the ESI-MS analysis of the huNoV P particle and VLP. Both the G2S and the ApexQe mass spectrometers were used to carry out the direct and proxy protein ESI-MS measurements. Both mass spectrometers were equipped with nanoflow ESI (nanoESI) sources. To carry out ESI, the sample solution was loaded into a nanoESI tip pulled from a borosilicate capillary (1.0 mm o.d., 0.68 mm i.d.) using a micropipette puller (P-1000, Sutter Instruments, Novato, CA). To initiate ESI, a voltage of 1.0–1.4 kV was applied to a platinum wire inserted into the nanoESI tip. A detailed description of the instrumental conditions used to implement the direct and proxy protein ESI-MS-binding measurements can be found elsewhere (El-Hawiet et al. 2012; Han et al. 2013). Unless otherwise indicated, the ESI-MS measurements performed on the ApexQe mass spectrometer were carried out using 50 mM aqueous ammonium acetate solutions (pH 7 and 25°C) containing protein and ligand of interest, while those on the Synapt G2S mass spectrometer were carried out using 200 mM aqueous ammonium acetate solutions (pH 7 and 25°C) containing protein and ligand of interest. To carry out the proxy protein ESI-MS measurements, a P_proxy was also added to the solutions. In all cases, the reference protein method, which involves the addition of a P_ref to the solution, was used to correct the mass spectra for the occurrence of nonspecific protein–carbohydrate binding during the ESI process. A complete description of the correction method can be found elsewhere (Sun et al. 2006; Kitova et al. 2012).

Gel-filtration chromatography

Gel-filtration chromatography was performed using a Superdex 200 size exclusion column (HiLoad 16/60, 120 mL bed volume, GE Healthcare Life Sciences) powered by an AKTA fast-performance liquid chromatography system (model 920, GE Healthcare Life Sciences). The column was equilibrated and run in 1× phosphate-buffered saline (PBS, pH 7.4) at a flow rate of 1.0 mL min⁻¹; 1.0 mL of the VLP sample (5 mg mL⁻¹) was loaded onto the column using a manual injector. The MW of the proteins in each elution volume was calibrated with the Gel Filtration Calibration kit (GE Healthcare Life Sciences) and purified P particle as described elsewhere (Wang et al. 2013, 2014).

Determination of K_a values

Direct ESI-MS assay

The direct ESI-MS assay was used to quantify oligosaccharide affinities for CBM and Gal-3C, two of four P_proxy's used in this study, and the affinity of L14 for the P dimer. For a 1 : 1 protein-ligand complex the association constant (K_a) is calculated from the abundance (Ab) ratio (R) of the ligand bound (PL) to free protein (P) ions measured by ESI-MS and the initial concentrations of protein ([P]₀) and ligand ([L]₀), Eq. (1) (Kitova et al. 2012):

$$K_{\mathrm{a}}=\frac{R}{{[\mathrm{L}]}_0-{\displaystyle \frac{{[\mathrm{P}]}_{0}R}{R+1}}},$$ (1)

where R is taken to be equal to the equilibrium concentration ratio in the solution, Eq. (2):

$$R=\frac{\sum \text{Ab}(\text{PL})}{\sum \text{Ab}(\text{P})}=\frac{[\text{PL}]}{[\text{P}]}.$$ (2)

The abundances of free and ligand-bound proteins were calculated as the sum of the peak areas for all the charge states detected for each species. In cases where ligand binding was weak (K_a < 10⁴ M⁻¹), a titration approach was employed, whereby the protein concentration was kept constant and the ligand concentration was varied (Daniel et al. 2002). Nonlinear regression analysis of the concentration-dependence of the fraction of ligand-bound protein, [R/(R + 1)] was used to determine K_a, Eq. (3) (Kitova et al. 2012):

$$\frac{R}{R+1}=\frac{1+K_{\mathrm{a}}{[\mathrm{L}]}_0+K_{\mathrm{a}}{[\mathrm{P}]}_0-\sqrt{{(1-K_{\mathrm{a}}{[\mathrm{L}]}_0+K_{\mathrm{a}}{[\mathrm{P}]}_0)}^2+4K_{\mathrm{a}}{[\mathrm{L}]}_0}}{2K_{\mathrm{a}}{[\mathrm{P}]}_0}.$$ (3)

Proxy protein ESI-MS assay

The proxy protein ESI-MS assay was recently developed to quantify the affinities of protein–ligand complexes that could not be directly measured by ESI-MS (El-Hawiet et al. 2012). Briefly, the method involves the use of the direct ESI-MS assay to monitor the extent of ligand binding to a proxy protein (P_proxy) in the presence of the target protein (P) (El-Hawiet et al. 2012). The general expression relating the intrinsic association constant of P (K_a,int,P), with h identical and independent binding sites, to R_proxy, the abundance ratio of ligand bound to free P_proxy ions, for a P_proxy with a single-binding site is given by Eq. (4a) (El-Hawiet et al. 2012):

$$K_{\mathrm{a},\mathrm{int},\mathrm{P}}=\frac{K_{\mathrm{a},{\mathrm{P}}_{\mathrm{proxy}}}/R_{\mathrm{proxy}}}{{\displaystyle \frac{h{[\text{P}]}_0}{{[\text{L}]}_0-{\displaystyle \frac{{[{\text{P}}_{\mathrm{proxy}}]}_0{R}_{\mathrm{proxy}}}{R_{\mathrm{proxy}}+1}-\frac{R_{\mathrm{proxy}}}{K_{\mathrm{a},{\mathrm{P}}_{\mathrm{proxy}}}}}}-1}}.$$ (4a)

As described in more detail in the Results and Discussion section, the huNoV P particle and VLP investigated in the present study exist in multiple forms in solution. The P particle exists predominately as a 24-mer; however, both the 18-mer and 36-mer have also been detected (Bereszczak et al. 2012). Similarly, although VLP exists predominantly as a 180-mer, the 60-mer and 80-mer, as well as dimer, are also present (Shoemaker et al. 2010). Consequently, it is more appropriate to rewrite Eq. (4a) in terms of the total concentration of ligand-binding sites ($\left({\left[P\right]}_{m,0}={{\displaystyle \sum h_i\left[{\text{P}}_{\text{i}}\right]}}_0\right)$), which is equal to the number of protomers that make up the assembly, Eq. (4b):

$$K_{\mathrm{a},\mathrm{P},\mathrm{int}}=\frac{K_{\mathrm{a},{\mathrm{P}}_{\mathrm{proxy}}}/R_{\mathrm{proxy}}}{{\displaystyle \frac{{[\text{P}]}_{\mathrm{m},0}}{{[\text{L}]}_0-{\displaystyle \frac{{[{\text{P}}_{\mathrm{proxy}}]}_0{R}_{\mathrm{proxy}}}{R_{\mathrm{proxy}}+1}-\frac{R_{\mathrm{proxy}}}{K_{\mathrm{a},{\mathrm{P}}_{\mathrm{proxy}}}}}}-1}}.$$ (4b)

It must be stressed that the K_a,int,P values determined in this way represent the weighted average of the affinities of the different assemblies present in solution.

The implementation of the proxy protein method was previously demonstrated using a P_proxy with a single-binding site. Two of the P_proxy used in the present study possess multiple (two) ligand-binding sites. In this case, K_a,P,int can be found using Eq. (5):

$$K_{\mathrm{a},\mathrm{P},\mathrm{int}}=\frac{K_{\mathrm{a},{\mathrm{P}}_{\mathrm{proxy}},\mathrm{int}}/\theta }{{\displaystyle \frac{{[\text{P}]}_{\mathrm{m},0}}{{[\text{L}]}_0-{\displaystyle \frac{g{[{\text{P}}_{\mathrm{proxy}}]}_0\theta }{\theta +1}-\frac{\theta }{K_{\mathrm{a},{\mathrm{P}}_{\mathrm{proxy}},\mathrm{int}}}}}-1}},$$ (5)

where K_a,proxy,int and g are the intrinsic affinity of P_proxy and the number of binding sites in P_proxy, respectively. The ratio of occupied-to-free-binding sites in the P_proxy (θ) is calculated using Eq. (6):

$$\theta =\frac{{[\text{occupied - binding sites]}}_{\mathrm{proxy}}}{{[\text{,free - binding sites]}}_{\mathrm{proxy}}}=\frac{{\sum }_{j=1}^{g}j{R}_{\mathrm{proxy},j}}{g\left(1+{\sum }_{j=1}^g{R_{\mathrm{proxy},}}_j\right)-{\sum }_{j=1}^{g}j{R}_{\mathrm{proxy},j}},$$ (6)

where R_proxy,j corresponds to the concentration ratio of ligand bound (to j molecules of L) to free P_proxy and is taken to be equal to the abundance (Ab) ratio of ligand bound (to j molecules of L) to free P_proxy gas-phase ions, Eq. (7):

$$R_{\mathrm{proxy},j}=\frac{\sum \text{Ab}({\text{P}}_{\mathrm{proxy}}{\text{L}}_j)}{\sum \text{Ab}({\text{P}}_{\mathrm{proxy}})}=\frac{[{\text{P}}_{\mathrm{proxy}}{\text{L}}_j]}{[{\text{P}}_{\mathrm{proxy}}]}.$$ (7)

In the case of a P_proxy with two ligand-binding sites (i.e. g = 2), θ is given by Eq. (8):

$$\theta =\frac{R_{\mathrm{proxy},1}+2R_{\mathrm{proxy},2}}{2+R_{\mathrm{proxy},1}}.$$ (8)

Given in Supplementary data is a derivation of the equations relevant to the implementation of the proxy protein method using a P_proxy with multiple ligand-binding sites.

Results and discussion

ESI-MS analysis of huNoV VA387 P particle and VLP

Representative ESI mass spectra measured for a 200 mM ammonium acetate aqueous solution (pH 7 and 25°C) containing 3 μM of P particle (which corresponds to 72 μM of monomer) or 0.2 μM VLP (corresponding to 36 μM VP1) are shown in Figure 1. As seen in Figure 1A, the P particle is present predominantly as a 24-mer (with a charge state distribution centered around +67), along with the 18-mer (charge state distribution centered around +56) at lower abundance. These observations are consistent with those reported previously (Bereszczak et al. 2012). The MWs of the 24- and 18-mer, 865,000 ± 540 Da and 648,900 ± 400 Da, respectively, are in reasonably good agreement with the expected values calculated from the protein sequence, 865,036 and 648,782 Da, respectively (Tan and Jiang 2005b). As seen in Figure 1B, the VP1 monomer (MW 58,887 Da), 60-mer (∼3.5 MDa), 80-mer (∼4.7 MDa) and 180-mer (∼10 MDa) are all present in solution. Notably, the distribution of VP1 species is reproducible over a period of weeks, Supplementary data, Figure S2. The smaller oligomers, which are believed to exist in dynamic equilibrium with the 180-mer, were previously observed by native ESI-MS, ion mobility separation MS, as well as atomic force microscopy (Shoemaker et al. 2010; Uetrecht et al. 2011). Due to the high MWs of the oligomers, it was not possible to resolve individual charge states and, thus, their identification was based on previously reported ESI-MS results (Shoemaker et al. 2010). Moreover, since the ESI-MS ionization/detection efficiencies of high MW oligomers are expected to be significantly different than those of VP1 monomer, the relative abundances of the monomer and oligomers measured by ESI-MS likely do not accurately reflect their relative concentrations in solution. In fact, gel-filtration chromatography performed on 5 mg mL⁻¹ HuNoV VA387 VLP samples (in 1× PBS, pH 7.4, 25°C) produced a single peak (Supplementary data, Figure S3), which corresponds to the void volume, suggesting that the capsid protein assembles predominantly into large complexes (>800 kDa) in solution.

Fig. 1.

(A) ESI mass spectrum acquired in positive ion mode for aqueous ammonium acetate (200 mM, pH 7 and 25°C) solution with (A) 3 μM P particle (corresponding to 72 μM monomer) and (B) 0.2 μM VLP (corresponding to 36 μM VP1) of huNoV VA387. The measurements were carried out using a Waters Synapt G2S mass spectrometer.

Affinities of HBGA oligosaccharides for the huNoV VA387 P particle and VLP

The proxy protein ESI-MS assay was used to evaluate the affinities of the 13 HBGA oligosaccharides, L1–L13, for the huNoV VA387 P particle. Due to the limited availability of huNoV VA387 VLP, measurements were restricted to four oligosaccharides, L1, L2, L7 and L8. In order to implement the assay, a suitable P_proxy (one that exhibits moderate/high affinity for the ligand) was required for each oligosaccharide tested. Four different P_proxy were used for these measurements, the P dimer of huNoV VA387, a truncated recombinant form of the human blood group glycosyltransferase GTA, the family 51 CBM and a recombinant fragment of the Gal-3C. Representative ESI mass spectra acquired for each of the P_proxy are shown in Supplementary data, Figure S4. Although the VA387 P dimer binds to a broad range of HBGAs, the interactions are uniformly weak (Tan et al. 2004; Han et al. 2013). Consequently, in the present study the P dimer was only used to quantify the interaction between the P particle and B type 3 tetrasaccharide (L12), which is the highest affinity HBGA ligand (K_a,int = 1500 M⁻¹) identified for the P dimer (Han et al. 2013). Truncated recombinant GTA, which forms a homodimer in aqueous solution at neutral pH, possesses two thermodynamically equivalent and independent acceptor substrate-binding sites (Patenaude et al. 2002; Shoemaker et al. 2008). Recent ESI-MS measurements revealed that GTA exhibits modest intrinsic affinities for B trisaccharide (L1, 1.6 × 10⁴ M⁻¹) and H disaccharide (L13, 3.2 × 10⁴ M⁻¹) (Shoemaker et al. 2008; Soya et al. 2009), which enabled the use of GTA as a P_proxy to quantify the interactions of L1 and L13 with the P particle. Three distinct CBM species (labeled as I, II and III) with MWs of 20,738, 20,798 and 20,916 Da, respectively, are evident from the mass spectrum shown in Supplementary data, Figure S4C. The MW measured for CBM I agrees well with the theoretical MW of 20,735, the nature of the modifications giving rise to the other two forms of CBM were not established. However, all three forms bind to HBGA oligosaccharides with identical affinities (data not shown). Consequently, for the direct and proxy protein ESI-MS-binding measurements, the abundances of all three forms were summed together to calculate the ratio of ligand-bound to free CBM. Glycan array screening carried out on CBM revealed binding to the A and B trisaccharides, as well as A and B type 2 and type 6 tetrasaccharides (Higgins et al. 2011). According to isothermal titration calorimetry (ITC), binding between CBM and the A and B type 2 tetrasaccharides is quite strong, with K_a values in the range of 10⁴–10⁵ M⁻¹ (Higgins et al. 2011). To extend the utility of CBM as a P_proxy for the current study, the direct ESI-MS assay was used to measure the affinities of L1–L14 for CBM (Supplementary data, Table S1). For a majority of the ligands (L1–L8 and L12), ESI-MS titration experiments were performed and the K_a values obtained by fitting Eq. (3) to the fraction of ligand-bound protein measured experimentally (Supplementary data, Figure S5). The K_a values for the other ligands (L9–L11 and L13 and L14) were determined from ESI-MS measurements carried out at ≥3 different ligand concentrations. The results show that CBM only binds to A and B oligosaccharides, with the following trend in affinities: A/B trisaccharides > type 2 ∼ type 6 > type 5 > type 3 > type 1 tetrasaccharides. It should be noted that the binding data for L3 ((5.3 ± 0.3) × 10⁴ M⁻¹) and L4 ((7.4 ± 0.3) × 10⁴ M⁻¹) measured by ESI-MS are in reasonable agreement with the K_a values obtained using ITC, 7.8 × 10⁴ and 6.6 × 10⁴ M⁻¹, respectively (Higgins et al. 2011). Recombinant Gal-3C contains a carbohydrate recognition domain that interacts with the β-galactoside motif (Hirabayashi et al. 2002; Collins et al. 2014). The affinities of the HBGA oligosaccharides (L1–L14) for Gal-3C were measured by the direct ESI-MS assay (Supplementary data, Table S1). For L9–L11 and L14 the K_a values were obtained from ESI-MS titration experiments (Supplementary data, Figure S6). For the remaining oligosaccharides, the K_a values were determined from ESI-MS measurements carried out at ≥3 different ligand concentrations. Notably, Gal-3C binds strongly (K_a ∼ 10⁵ M⁻¹) to A/B type 1, 2 and 6 tetrasaccharides; exhibits moderately strong binding (K_a ∼ 10⁴ M⁻¹) to B type 3, A type 5, B type 5 tetrasaccharides, H type 6 and type 2 trisaccharides and weak binding (K_a < 10³ M⁻¹) to A/B trisaccharides and H disaccharide. It can also be seen from these data that Gal-3C binding to B type oligosaccharides is consistently stronger than to the corresponding A type oligosaccharides. It should also be pointed out that the K_a values for L7 [(1.02 ± 0.04) × 10⁵ M⁻¹], L8 [(6.4 ± 0.6) × 10⁴ M⁻¹], L10 [(1.34 ± 0.02) × 10⁵ M⁻¹] and L11 [(1.38 ± 0.06) × 10⁴ M⁻¹] agree reasonably well with values measured using frontal affinity chromatography for B type 6 tetrasaccharide (1.7 × 10⁵ M⁻¹), A type 6 tetrasaccharide (7.1 × 10⁴ M⁻¹), A type 1 hexasaccharide (5.6 × 10⁵ M⁻¹) and H type 6 trisaccharide (1.1 × 10⁴ M⁻¹) (Hirabayashi et al. 2002).

To quantify the affinities of L1–L13 for the P particle, the proxy protein ESI-MS method was implemented using a titration format, whereby the concentrations of P_proxy, HBGA oligosaccharide and P_ref were fixed, while the concentration of the target protein (i.e. P particle or VLP) was varied. From the dependence of R_proxy (which corresponds to the abundance ratio of ligand bound to free P_proxy) on target protein concentration, K_a,int for the target protein could be determined. Shown in Figure 2 are representative ESI mas spectra acquired for aqueous 50 mM ammonium acetate solutions (pH 7, 25°C) containing CBM (12 μM), B trisaccharide (L1, 35 μM), Ubq (8 μM) with 0, 6 and 12 μM P particle (corresponding to 0, 144 and 288 μM of monomer, respectively). Visual inspection reveals that the relative abundance of L1-bound P_proxy decreased with increasing P particle concentration, indicating that L1 binds to the P particle. In Figure 2D, the measured R_proxy values are plotted vs. P particle concentration. Fitting Eq. (4b) to these data gives a K_a,int of 940 ± 90 M⁻¹. Analogous measurements were carried out using GTA as P_proxy. Representative mass spectra acquired for aqueous 50 mM ammonium acetate solutions (pH 7, 25°C) containing GTA (10 μM), L1 (60 μM), scFv (8 μM, P_ref) with 0, 6 and 12 μM P particle are shown in Figure 3. Clearly, the relative abundance of L1-bound P_proxy decreased with the addition of P particle to the solution, consistent with the occurrence of binding of L1 to the P particle. From the ratios R_proxy,1 and R_proxy,2, which correspond to abundance ratios of L1-bound GTA (to one or two L1, respectively) to free GTA, the magnitude θ was calculated (Eq. (8)). Shown in Figure 3D is a plot of the calculated values of θ vs. P particle concentration. Fitting Eq. (5) to these data gives a K_a,int of 1100 ± 100 M⁻¹. Importantly, these two values of K_a,int (which were determined using different P_proxy) are indistinguishable, within experimental error. Using an analogous strategy (Supplementary data, Figures S7–S11), K_a,int values for the interactions of L2–L13 with the P particle were determined, Table I.

Fig. 2.

Representative ESI mass spectra obtained in positive ion mode for a 50 mM aqueous ammonium acetate solution (pH 7 and 25°C) containing 12 μM CBM, 8 μM Ubq (P_ref), 35 μM L1 (B trisaccharide) with (A) 0 μM, (B) 6 μM and (C) 12 μM P particle (24-mer) of huNoV VA387. (D) Plot of R_proxy vs. concentration of monomer in the P particle. The solution condition for each measurement was same as (A), but with the addition of 0–12 μM P particle. The solid curve corresponds to the best fit of Eq. (4b) for the experimental data. The error bars correspond to 1 SD. The measurements were carried out using a Bruker ApexQe FT-ICR mass spectrometer.

Fig. 3.

Representative ESI mass spectra obtained in positive ion mode for a 50 mM aqueous ammonium acetate solution (pH 7 and 25°C) containing 10 μM GTA, 10 μM scFv (P_ref), 60 μM L1 (B trisaccharide) with (A) 0 μM, (B) 6 μM and (C) 12 μM P particle (24-mer) of huNoV VA387. (D) Plot of θ vs. concentration of monomer in the P particle. The solution condition for each measurement was same as (A), but with the addition of 0–12 μM P particle. The solid curve corresponds to the best fit of Eq. (5) to the experimental data. The error bars correspond to 1 SD. The measurements were carried out using a Bruker ApexQe FT-ICR mass spectrometer.

Table I.

Intrinsic association constants (K_a,int) for HBGA oligosaccharides (L1–L14) binding to huNoV VA387 P dimer, P particle and VLP, measured at 25°C and pH 7 using the ESI-MS proxy protein assay^a

	HBGA	Structures	Ka,int (M−1) P dimerb	Ka,int (M−1) P particle	Ka,int (M−1) VLP
L1	B trisaccharide	α-D-Gal-(1→3)-[α-L-Fuc-(1→2)]-β-D-Gal- O(CH2)8COOC2H5	800 ± 100	940 ± 90c 1100 ± 100d	2300 ± 250c
L2	A trisaccharide	α-d-GalNAc-(1→3)-[α-l-Fuc-(1→2)]- β-d-Gal-O(CH2)8COOC2H5	500 ± 50	840 ± 90c	1400 ± 150c
L3	B type 2 tetrasaccharide	α-d-Gal-(1→3)-[α-l-Fuc-(1→2)]-β-d-Gal- (1→4)-β-d-GlcNAc-O(CH2)6CH=CH2	410 ± 45	870 ± 60c	n.d.
L4	A type 2 tetrasaccharide	α-D-GalNAc-(1→3)-[α-L-Fuc-(1→2)]-β-D- Gal-(1→4)-β-D-GlcNAc-O(CH2)6CH=CH2	290 ± 30	710 ± 90c	n.d.
L5	B type 5 tetrasaccharide	α-d-Gal-(1→3)-[α-l-Fuc-(1→2)]-β-d-Gal- (1→3)-β-d-Gal-O(CH2)6CH=CH2	700 ± 100	930 ± 80c	n.d.
L6	A type 5 tetrasaccharide	α-d-GalNAc-(1→3)-[α-l-Fuc-(1→2)]-β-d- Gal-(1→3)-β-d-Gal-O(CH2)6CH=CH2	560 ± 40	900 ± 100c	n.d.
L7	B type 6 tetrasaccharide	α-d-Gal-(1→3)-[α-l-Fuc-(1→2)]-β-d-Gal- (1→4)-β-d-Glc-O(CH2)6CH=CH2	600 ± 45	650 ± 30c	1000 ± 160c
L8	A type 6 tetrasaccharide	α-d-GalNAc-(1→3)-[α-l-Fuc-(1→2)]- β-d-Gal-(1→4)-β-d-Glc-O(CH2)6CH=CH2	1200 ± 50	1400 ± 120c	3900 ± 260c
L9	B type 1 tetrasaccharide	α-d-Gal-(1→3)-[α-l-Fuc-(1→2)]-β-d-Gal- (1→3)-β-d-GlcNAc-O(CH2)6CH=CH2	700 ± 100	1530 ± 40e	n.d.
L10	A type 1 tetrasaccharide	α-d-GalNAc-(1→3)-[α-l-Fuc-(1→2)]-β-d- Gal-(1→3)-β-d-GlcNAc-O(CH2)6CH=CH2	600 ± 65	1500 ± 100e	n.d.
L11	H type 6 trisaccharide	α-l-Fuc-(1→2)-β-d-Gal-(1→4)-β-d-Glc- O(CH2)6CH=CH2	330 ± 25	880 ± 50e	n.d.
L12	B type 3 tetrasaccharide	α-d-Gal-(1→3)-[α-l-Fuc-(1→2)]-β-d-Gal- (1→4)-α-d-GalNAc-O(CH2)6CH=CH2	1500 ± 150	2300 ± 110f	n.d.
L13	H disaccharide	α-l-Fuc-(1→2)-β-d-Gal-O(CH2)8COOC2H5	240 ± 40	520 ± 100d	n.d.
L14	Type 2 trisaccharide	α-d-Gal-(1→3)-β-d-Gal-(1→4)-β-d- GlcNAc-O(CH2)6CH=CH2	NB	NB	n.d.

NB, no binding detected. n.d., not determined.

^aUncertainties correspond to 1 SD.

^bValues are adapted from Han et al. (2013).

^c–fValues are measured using CBM, GTA, Gal-3C and P dimer as P_proxy, respectively.

The aforementioned measurements were carried out using 50 mM aqueous ammonium acetate solutions. To ensure that the affinity measurements were not sensitive to ionic strength, ESI-MS measurements were repeated using substantially higher concentrations of ammonium acetate. Representative mass spectra acquired for solutions of CBM (12 μM), L1 (35 μM), Ubq (8 μM) and 6 μM P particle in either 200 or 800 mM aqueous ammonium acetate solution (pH 7, 25°C) are shown in Supplementary data, Figure S12A and B, respectively. Notably, the K_a,int values of 940 ± 60 and 900 ± 110 M⁻¹, respectively, are indistinguishable from the value measured in 50 mM aqueous ammonium acetate (940 ± 90 M⁻¹). From these results, it is concluded that the affinities of the HBGA ligands for the P particle and, presumably, VLP are relatively insensitive to the ionic strength of the solution.

Control experiments were also carried out to rule out the possibility that the aglycone of the HBGA oligosaccharides used in the study, –(CH₂)₈COOC₂H₅ for L1, L2 and L13 and –(CH₂)₆CH=CH₂ for L3–L12, promotes nonspecific binding to the P particle or VLP. The affinities of a type 2 trisaccharide (L14), which also has –(CH₂)₆–CH=CH₂ at the reducing end but does not bind to GII.4 huNoVs (Fiege et al. 2012) for both the VA387 P dimer and the P particle, were measured using direct and the proxy protein ESI-MS assay, respectively. Shown in Supplementary data, Figure S13 is a representative ESI mass spectrum acquired for an aqueous 200 mM ammonium acetate solutions (pH 7, 25°C) of P dimer (12 μM), L14 (80 μM) and scFv (8 μM). It can be seen from the normalized distributions of L14 bound P dimer (after correction for nonspecific ligand binding) that L14 does not exhibit any detectable binding to the P dimer. Shown in Supplementary data, Figure S14A–C are representative ESI mass spectra acquired for aqueous 200 mM ammonium acetate solutions (pH 7, 25°C) containing Gal-3C (5 μM), L14 (40 μM), Ubq (3 μM) with 0, 4 and 8 μM P particle (corresponding to 0, 96 and 192 μM of monomer, respectively). Notably, it can be seen from Supplementary data, Figure S14D that R_proxy is independent of P particle concentration. This result confirms that the –(CH₂)₆CH=CH₂ aglycone does not bind nonspecifically to the P dimer, the P particle or, presumably, the VLP.

According to the results of the proxy protein ESI-MS measurements, the 13 HBGA oligosaccharides investigated (L1–L13) exhibit binding to the P particle, with K_a,int values ranging from 500 to 2300 M⁻¹. From these data, the apparent affinities of these HBGA oligosaccharides for the P particle (24-mer) can be estimated to be between 1 × 10⁴ and 6 × 10⁴ M⁻¹. Notably, the trend in measured affinities of L1–L13 for the P particle mirrors that found for the P dimer, with the B type 3 tetrasaccharide and H disaccharide being the strongest and weakest binders, respectively. Additionally, the H type 6 trisaccharide (L11, which is 2′-fucosyllactose), which is abundant in human milk and was recently shown to inhibit VA387 P particles from binding to other HBGAs (Shang et al. 2013), exhibits a K_a,int of 880 ± 50 for the P particle. Overall, the K_a,int values for the P dimer and P particle are within a factor of 2. This finding suggests that the binding sites of the P dimer and P particle are structurally identical, or nearly so.

The affinities of four HBGA oligosaccharides (L1, L2, L7 and L8) for the VLP were also measured using CBM as the P_proxy. Shown in Figure 4A–C are representative ESI mass spectra acquired for aqueous 200 mM ammonium acetate solutions (pH 7, 25°C) containing CBM (12 μM), L1 (25 μM) and Ubq (4 μM) with 0, 380 and 760 nM VLP (corresponding to 0, 68 and 136 μM of monomer, respectively). From visual inspection, it is seen that the relative abundance of L1-bound P_proxy decreased with the addition of the VLP to the solution, consistent with binding between L1 and the VLP. Shown in Figure 5 are plots of R_proxy vs. VLP concentrations determined for L1, L2, L7 and L8. By fitting Eq. (4b) to these data, the corresponding K_a,int values were determined, Table I. The K_a,int values, which range from 1000 to 4000 M⁻¹, are consistently larger than those measured for the P particle, but are within a factor of 3. From the measured K_a,int values, the apparent affinities of these ligands for the VLP (180-mer form) are estimated to be between 2 × 10⁵ and 7 × 10⁵ M⁻¹.

Fig. 4.

Representative ESI mass spectra obtained in positive ion mode for a 200 mM aqueous ammonium acetate solution (pH 7 and 25°C) containing 12 μM CBM, 4 μM Ubq (P_ref), 25 μM L1 (B trisaccharide) with (A) 0 nM, (B) 380 nM and (C) 760 nM VLP (180-mer) of huNoV VA387. The measurements were carried out using a Waters Synapt G2S mass spectrometer.

Fig. 5.

Plots of R_proxy vs. concentration of monomer in the VLP measured for aqueous ammonium acetate solutions (200 mM, pH 7, and 25°C) containing CBM (12 μM), Ubq (P_ref, 4 μM), VLP (180-mer) of huNoV VA387 (0–760 nM) with (A) L1 (B trisaccharide, 25 μM), (B) L2 (A trisaccharide, 20 μM), (C) L7 (B type 6 tetrasaccharide, 40 μM) and (D) L8 (A type 6 tetrasaccharide, 25 μM). The solid curves correspond to the best fit of Eq. (4b) to the experimental data for each ligand. The error bars correspond to 1 SD. The measurements were carried out using a Waters Synapt G2S mass spectrometer.

Taken together, the binding data measured in the present study and those reported previously for the P dimer indicate that the interactions between the HBGA ligands and the P dimer, P particle and VLP of huNoV VA387, while similar, are not identical and exhibit the following trend: P dimer ≈ P particle < VLP. This finding suggests that there exist subtle differences in the structure of the carbohydrate-binding sites presented by the P dimer and P particle and those of the VLP. It is interesting to note, however, that the crystal structures of the VLP (Prasad et al. 1999) and P dimer (Bu et al. 2008; Choi et al. 2008) of Norwalk virus (GI.1) do not reveal a distinct structural difference between the P dimer of VLP and the one formed from the isolated P domain. Therefore, our data may imply that such subtle structural difference may not be easily recognized from crystal structures or, alternatively, that such subtle difference may occur between the VLP and the P dimers/P particle of VA387 (GII.4) but not between those of Norwalk virus. Nevertheless, the differences in K_a,int values for the VA387 P dimer, P particle and VLP are small and support the use of P dimers and P particles as surrogates to the VLP.

Conclusion

In summary, the interactions between the huNoV VA387 P particle and VLP and a series of HBGA oligosaccharides were quantified for the first time. The measured K_a,int values of the 13 HBGA oligosaccharides for the P particle range from 500 to 2300 M⁻¹; those of the A and B trisaccharides and the A and B type 6 tetrasaccharides for the VLP range from 1000 to 4000 M⁻¹. Comparison of the binding data with those measured previously for the P dimer reveals that the HBGA oligosaccharides tested exhibit similar intrinsic affinities for the P dimer and P particle. The intrinsic affinities exhibited for the VLP are consistently higher than those measured for the P particle, by a factor of 3. Based on these data, the apparent affinities of the HBGA oligosaccharides tested for the P particle and VLP were estimated to be in the 10⁴–10⁵ and 10⁵–10⁶ M⁻¹ range, respectively. While the cause of the subtle differences in HBGA oligosaccharide affinities for the P dimer and P particle and those for the VLP remains unknown, the present data support the use of P dimers and P particles as substitutes to the VLP for NoV-receptor-binding studies.

Supplementary data

Supplementary data for this article are available online at http://glycob.oxfordjournals.org/.

Funding

This work was supported by the Alberta Glycomics Centre (J.S.K.); the Natural Sciences and Engineering Research Council (J.S.K.; DG 205047); the Canadian Institutes of Health Research (A.B.B.; FRN 86610) and National Institutes of Health of the United States of America (R01 AI089634 to X.J. and R21 AI092434 to M.T.). A.B.B. was also supported by an EWR Steacie Memorial Fellowship, a Canada Research Chair in Molecular Interactions and a Michael Smith Foundation for Health Research Career Scholar Award.

Conflict of interest statement

None declared.

Abbreviations

Ab, abundance; CBM, carbohydrate-binding module; ESI-MS, electrospray ionization mass spectrometry; FT-ICR, Fourier-transform ion cyclotron resonance; Gal-3C, C-terminus of human galectin-3; GTA, α-(1→3)N-acetyl galactosaminyltransferase; HBGAs, histo-blood group antigens; huNoVs, human noroviruses; ITC, isothermal titration calorimetry; MW, molecular weight; nanoESI, nanoflow ESI; PBS, phosphate-buffered saline; scFv, single-chain variable fragment; STD-NMR, saturation-transfer difference nuclear magnetic resonance; Ubq, ubiquitin; VLP, virus-like particle