Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Biochemistry. Author manuscript; available in PMC 2010 June 23.
Published in final edited form as:
PMCID: PMC2709598

Glycosylated SV2 and Gangliosides as Dual Receptors for Botulinum Neurotoxin Serotype F


Botulinum neurotoxin causes rapid flaccid paralysis through inhibition of acetylcholine release at the neuromuscular junction. The seven BoNT serotypes (A-G) have been proposed to bind motor neurons via ganglioside- protein dual receptors. To date, the structure-function properties of BoNT/F host receptor interactions have not been resolved. Here we report the crystal structures of the receptor binding domains (HCR) of BoNT/A and BoNT/F and the characterization of the dual receptors for BoNT/F. The overall polypeptide fold of HCR/A is essentially identical to the receptor binding domain of the BoNT/A holotoxin, and the structure of HCR/F is very similar to that of HCR/A, except for two regions implicated in neuronal binding. Solid phase array analysis identified two HCR/F binding glycans: ganglioside GD1a and oligosaccharides containing an N-acetyllactosamine core. Using affinity chromatography, HCR/F bound native synaptic vesicle glycoproteins as part of a protein complex. Deglycosylation of glycoproteins using α(1-3,4)- fucosidase, endo-β-galactosidase and PNGase F disrupted the interaction with HCR/F, while the binding of HCR/B to its cognate receptor, synaptotagmin I, was unaffected. These data indicate that the HCR/F binds synaptic vesicle glycoproteins through the keratan sulfate moiety of SV2. The interaction of HCR/F with gangliosides was also investigated. HCR/F bound specifically to gangliosides that contain α2, 3-linked sialic acid on the terminal galactose of a neutral saccharide core (binding order: GT1b = GD1a [dbl greater-than sign] GM3; no binding to GD1b and GM1a). Mutations within the putative ganglioside binding pocket of HCR/F decreased binding to gangliosides, synaptic vesicle protein complexes and primary rat hippocampal neurons. Thus, BoNT/F neuronal discrimination involves recognition of ganglioside and protein (glycosylated SV2) carbohydrate moieties, providing a structural basis for the high affinity and specificity of BoNT/F for neurons.

Botulinum neurotoxins (BoNTs) inhibit neurotransmitter release by cleaving one or more components of the vesicular fusion machinery. BoNTs are single chain AB toxins that comprise an N-terminal light chain protease (LC) that is disulfide linked to a C-terminal heavy chain that includes a translocation domain (HCT) and a receptor binding domain (HCR). BoNTs are divided into seven serotypes (termed A-G) based on the lack of anti-toxin cross neutralization. While experimental evidence suggests humans are sensitive to all serotypes, natural intoxications are associated with serotypes A, B, E, and F (1-4). In addition, BoNT/A, BoNT/B, and BoNT/F have also been used as therapeutic agents for the treatment of neurological disorders in humans (1, 3).

The prototype strain of Clostridium botulinum type F was first isolated as a source of human botulism on the Danish Island, Langeland (5). BoNT/F Langeland is a 1274-amino acid protein possessing a toxic potency similar to other BoNT serotypes (6). Subsequent studies demonstrated that strains of C. barati isolated from cases of infant botulism also produced a BoNT/F neurotoxin (7-10). The delayed recognition of BoNT/F as a toxin of humans was probably linked to misidentification of type F, due to cross-neutralization by type E antitoxin (9). The molecular basis for cross-neutralization is not currently understood, but is suggested to be linked to the high sequence homology between BoNT/E and BoNT/F (11). To further understand the mechanisms of cross-neutralization, a detailed structure-function analysis of BoNT/F is required.

BoNTs target the presynaptic membranes of α-motor neurons through molecular mechanisms that are only now being defined. A ‘dual receptor’ model for BoNT intoxication was proposed by Montecucco and coworkers (12) where BoNTs initially interact with glycolipids such as gangliosides, concentrating the toxin on the presynaptic membrane. Subsequent to initial capture, BoNTs interact with a second glycolipid and/ or protein co-receptor triggering receptor-mediated endocytosis. Recent studies employing structural, biochemical and genetic approaches have supported this model for BoNT serotypes B and G (13-17).

Gangliosides are complex lipids with a strong amphiphilic character due to the large oligosaccharide headgroup and the double-tailed hydrophobic moiety. The lipid moiety of gangliosides is constituted by the long-chain alcohol sphingosine, connected to a fatty acid by an amide linkage. The oligosaccharide chain of gangliosides is variable because of the sugar structure, content, sequence, and linkages. Sialic acid is the sugar that differentiates gangliosides from neutral glycosphingolipids and sulfatides (18-20). Numerous reports have demonstrated that BoNTs bind directly to complex gangliosides (15, 21-27) and that prior incubation of BoNTs with excess b-series gangliosides (GD2, GD1b, GT1b and GQ1b) reduced toxicity both in vitro and in vivo (22). Mice unable to express complex gangliosides display lower sensitivity to BoNT/A, /B, /C, /E and /G; while binding and entry of BoNT/A, /B, /E and /G into cultured neurons lacking gangliosides is also diminished (15, 22, 28). Similarly, BoNT/F has been reported to bind GD1a, GD1b, and GT1b under conditions of low ionic strength, but a physiologic role for gangliosides has not been demonstrated (24).

In addition to ganglioside, recent studies have identified protein co-receptors for BoNTs, most completely defined for BoNT/B. Initial studies by Nishiki and coworkers showed that BoNT/B bound synaptotagmin II (23, 29-31). Observing that the entry process of BoNT/B involved synaptic vesicle exocytosis, Chapman and colleagues proposed that synaptic vesicle proteins functioned as the physiological receptor for BoNT/B (13-15). Using luminal domain fragments of synaptic vesicle proteins fused to GST, synaptotagmin I (Syt I) and synaptotagmin II (Syt II) were shown to mediate the binding and entry of BoNT/B into cultured neuroendocrine cells and hippocampal neurons (13). Similar approaches demonstrated that Syt I and Syt II also mediate the binding and entry of BoNT/G (17), whereas SV2 mediates the binding and entry of BoNT/A and BoNT/E into cultured neurons (28, 32, 33). At present, the identity of the BoNT/F protein co-receptor remains unknown.

In the present study, the crystal structure of the receptor binding domain (HCR) of BoNT type F is reported. Structure-based functional studies characterized the molecular properties of the BoNT/F dual receptors, synaptic vesicle glycoprotein SV2 and gangliosides, both of which contribute to the high affinity interaction of BoNT/F with neurons.

Experimental Procedures


Unless otherwise stated, molecular biology grade chemicals and reagents were obtained from Sigma-Aldrich Co. (St. Louis, MO). Restriction enzymes were from New England Biolabs. Neuronal cell culture reagents were from Invitrogen (Carlsbad, CA). Sprague-Dawley rat embryonic day 18 hippocampal neurons were from Brainbits LLC (Springfield, IL) and cultured as described by the supplier. The monoclonal antibody against VAMP-2 was obtained from Synaptic Systems (Germany). The monoclonal antibody α-SNAP-25-C (MC-6053, specific for cleaved SNAP25) was from R and D Antibodies (USA). The SV2 monoclonal antibody developed by Kathleen M. Buckley was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences. E. coli codon optimized DNA constructs encoding HCR/A (Hall-A strain) and HCR/F (Langeland strain) were synthesized by EZBiolab (Westfield, IN). EIA 96-well flat bottom plates (Corning Costar plate 9018) were obtained from Thermo Fisher Scientific. Purified botulinum neurotoxin types A and F were provided by Dr. Eric Johnson, University of Wisconsin-Madison at levels below the limit exempted by the Centers for Disease Control and Prevention.

HCR expression and purification

DNA encoding HCR/A (residues 870-1295; Supplemental Figure 1) and HCR/F (residues 862-1278; Supplemental Figure 2) were subcloned into the pET-28 expression vector (Merck KGaA, Darmstadt, Germany). Mutated forms of HCR/F were generated using the Quikchange site-directed mutagenesis kit in accordance with the manufacturer's instructions (Stratagene). Purification of HCR/A and HCR/F from E. coli BL-21(DE3) was as described previously except that peak fractions from the S200-HR column were passed through a DEAE-sephacryl column prior to concentration with a second passage over Ni2+-nitrilotriacetic acid resin (34). A typical purification from a 1-liter culture yielded between 5-10 mg HCR/F and 15-25 mg HCR/A. The structural integrity of the mutated forms of HCR/F was estimated by performing limited proteolysis experiments measuring trypsin sensitivity.

Neutralization of BoNT activity in rat hippocampal neurons

Rat hippocampal neurons were cultured on poly-D-lysine coated glass coverslips in Neurobasal medium supplemented with 2 mM glutamine and B27 supplement for 10-14 days prior to use. Cells were treated with control solution (15 mM Hepes, 145 mM NaCl, 5.6 mM KCl, 2.2 mM CaCl2, 0.5 mM MgCl2, 5.6 mM glucose, 0.5 mM ascorbic acid, and 0.1% BSA, pH 7.4) or high K+ solution (same as control solution but adjusted to 95 mM NaCl and 56 mM KCl) for 10 min at 37°C, in the presence of 10 nM BoNTs with or without 1 μM of the indicated HCR. Cells were then washed with PBS and incubated for an additional 48 hr at 37°C in fresh Neurobasal medium: conditioned Neurobasal media (1: 1). Following treatment, cells were washed three times with PBS, fixed with 4% w/v paraformaldehyde in PBS (15 min at RT), permeabilized with 0.1% Triton X-100 / 4% formaldehyde in PBS (for 10 min at RT), and stained with either mouse α-SNAP25-C or α-VAMP-2. Bound antibodies were visualized using α-mouse IgG Alexa488 (Invitrogen). Images were captured at room temperature using a Nikon TE2000 microscope equipped with a CFI Plan Apo VC 60× Oil, N.A. 1.4 type lens and a Photometrics CoolSnap ES camera. Image acquisition and subsequent analysis were performed using Metamorph version 7. Figures were compiled using Photoshop CS3 (Adobe).

Synaptic vesicle isolation and FLAG Immunoprecipitation

The isolation of partially purified synaptic vesicles (SG-V) from rat cerebral cortex was performed as described previously (35). The SG-V pool (~20 ml) was then layered on to a 900 ml column of Sephacryl- 500HR equilibrated with buffered glycine (300 mM glycine, Hepes-OH, pH 7.4, 0.1% w/v sodium azide), overlaid with buffered glycine, and chromatography performed in buffered glycine at a flow rate of 60 ml/h, collecting 10-ml fractions. Western blotting of column fractions for the synaptic vesicle protein synaptophysin showed that synaptic vesicles were enriched approximately ~12-fold relative to initial synaptosomes. In contrast, the plasma membrane resident Na, K ATPase (α-subunit) was not enriched in the SG-V fraction relative to the initial input and not detected in the final vesicle pellet. Final standardization of vesicle preparations was made by determining total protein content.

Detergent solubilization of synaptic vesicles and subsequent immunoprecipitation of 3×FLAG tagged HCR domains from vesicle lysates were performed as described previously (36).

Deglycosylation of native synaptic vesicle proteins

Triton X-100 soluble extracts of isolated synaptic vesicles (~250 μg total protein) were incubated alone or with 0.1 units each of α(1-3,4)- fucosidase, protein N-glycosidase F (PNGase F) and endo-β-galactosidase for 16 hr at 37°C. Treated samples were clarified by centrifugation (70,000×g, 30 min) and then cooled to 4°C prior to FLAG immunoprecipitation as described above.

The efficiency of oligosaccharide release was estimated using Sambucus nigra lectin (binding sialic acid) conjugated to HRP as follows. Triton X-100 soluble extracts (400 ng total protein per well) were applied to a high-protein binding 96-well plate in sodium carbonate buffer (pH 9.6) overnight at 4°C. The plates were washed ×3 with PBS and non-specific binding sites were blocked by incubating for 1 hr in sodium carbonate buffer (pH 9.6) with 2% w/v BSA. Binding assays were performed in 100 μl of PBS- 1% w/v BSA / well for 2 hr at 4°C containing either purified HRP or at the indicated concentrations. Following incubation, plates were washed ×4 with PBS and bound lectins detected using TMB-Ultra (Pierce Biochemicals) as the substrate for HRP. The reaction was terminated by addition of 0.2 M H2SO4 and the absorbance at 450 nm was determined using a plate reader (Victor 3V, Perkin Elmer) (Supplemental Figure 3). Under the conditions described above, increased release of oligosaccharides was not observed with a further addition of enzymes or extending the incubation time.

Ganglioside binding assay

Purified bovine brain gangliosides (Matreya, LLC) dissolved in dimethyl sulfoxide (20 mg/ml) and diluted in methanol were applied to non-protein binding 96-well plates (1 or 10 μg gangliosides / well as indicated). The solvent was evaporated at RT and wells were washed ×3 with PBS. Non-specific binding sites were blocked by incubating for 1 hr in sodium carbonate buffer (pH 9.6) with 2% w/v BSA. Binding assays were performed in 100 μl of PBS / well for 2 hr at 4°C containing either HCR/A or HCR/F at the indicated concentrations and an α-FLAG M2 monoclonal antibody-HRP conjugate (diluted 1/10000, Sigma-Aldrich). Following incubation, plates were washed ×4 with PBS and bound HCR detected using TMB-Ultra (Pierce Biochemicals) as the substrate for HRP. The reaction was terminated by addition of 0.2 M H2SO4 and the absorbance at 450 nm was determined using a plate reader (Victor 3V, Perkin Elmer).

Synaptic vesicle binding assay

Crude synaptic vesicle membranes (LP2 fraction, 400 ng total protein per well) were applied to a high-protein binding 96-well plate in sodium carbonate buffer (pH 9.6) overnight at 4°C. The plates were washed ×3 with PBS and non-specific binding sites were blocked by incubating for 1 hr in sodium carbonate buffer (pH 9.6) with 2% w/v BSA. Binding assays were performed in 100 μl of PBS- 1% w/v BSA / well for 2 hr at 4°C containing either wild-type HCR/F or the HCR/FW1250L mutated protein at the indicated concentrations and an α-FLAG M2 monoclonal antibody-HRP conjugate (diluted 1/10000, Sigma-Aldrich). Following incubation, plates were washed ×4 with PBS and bound HCR was detected using TMB-Ultra (Pierce Biochemicals) as the substrate for HRP. The reaction was terminated by addition of 0.2 M H2SO4 and the absorbance at 450 nm was determined using a plate reader (Victor 3V, Perkin Elmer).

Protein-glycan interaction screen

HCR/F was labeled with an Alexa Fluor 488 succinimidyl ester as described in manufacture's protocol (Invitrogen). Following extensive dialysis against Dulbecco's PBS to remove unincorporated dye, the molar HCR: dye ratio was determined (~1:1). The labeled HCR/F domain was screened by core H of the Consortium for Functional Glycomics (CFG) using array version 3.0. Protein (~ 200 μg/ml) was bound to the matrix in 20mM Tris-HCL pH 7.4, 150 mM NaCl, 2mM CaCl2, 2mM MgCl2, 0.05% Tween 20, and 1% bovine serum albumin (BSA) and quantified as described previously (37). The final data set is presented in Supplemental Figure 4.

Crystallization and data collection

Purified HCR/A or HCR/F was dialyzed and concentrated to 5~12 mg/ml in 30 mM Tris-HCl, pH 7.6, and 0.5 M NaCl. The hanging drops containing 2 μl of 12 mg/ml of HCR/A or 5 mg/ml of HCR/F protein and 2μl of well solution were equilibrated against 0.5 ml well solution. For HCR/A, the well solution contained 0.1M Hepes (pH 7.5), 12% polyethylene glycol 8000, 8% glycerol, and 100 mM NaCl, while the well solution for HCR/F contained 100 mM Tris-HCl (pH 8.5), 18% ethanol, and 150 mM NaCl. HCR/A was crystallized in the orthorhombic space group, P212121 with cell dimensions a=39.8 Å, b=104.8 Å, and c=112.8 Å. The HCR/F crystals belong to the monoclinic space group P21 with unit cell parameters of a=42.3 Å, b=74.1 Å, c=74.8 Å, and β=106.4°. There is one monomer in an asymmetric unit for both crystal forms. Diffraction data for HCR/A was collected at 100 K at the SBC 19ID beamline, Advanced Photon Source, Argonne National Laboratory and data for HCR/F were collected using an R-AXIS IV++ with a MicroMax 007 generator at 100 K. HKL2000 (38) was used for data processing. Data collection and processing statistics for both HCR/A and HCR/F crystals are summarized in Table 1.

Table 1
Data Collection and Refinement Statistics.

Structure Determination and Refinement

Structures of HCR/A and HCR/F were solved by the molecular replacement method using MOLREP within the CCP4 program suite and using the structure of the HCR domain of the holotoxin A (residues 870-1295, pdb code, 3BTA) (39) as the probe for the HCR/A solution and the refined HCR/A structure as the probe for HCR/F structure determination. Initial structures from the molecular replacement results were refined using the program CNS (40). The refinement procedure consisted of rigid body and positional refinement followed by a simulated-annealing protocol. Iterative rounds of positional and temperature factor refinement followed by manual fitting and rebuilding using the graphics program TURBO-FRODO (41) with 2Fo-Fc and Fo-Fc difference Fourier maps. At later stages of refinement, water molecules were assigned where electron densities were greater than 3 σ in the Fo-Fc map and situated within 3.3 Å of a potential hydrogen bonding partner. The final models were completed with Rcrystal/Rfree of 0.225 0/.259 for HCR/A and 0.205/0.256 for HCR/F, respectively. A stereo image of HCR/A superimposed with HCR/F is shown in Figure 2. With the exception of the N-terminal 35 residues (6×His, 3×FLAG, and five N-terminal residues) the overall polypeptide fold of HCR/A (E875-L1295) is well defined and essentially identical to those observed in the BoNT/A holotoxin structure (39).

Figure 2
Binding of HCR/F to a panel of gangliosides

Results and Discussion

HCR/F competes with BoNT/F holotoxin for binding to primary hippocampal neurons

While Botulinum neurotoxin type F (BoNT/F) intoxicates the mammalian neuromuscular junction, the cellular determinants of host cell binding and internalization have not been resolved. A recombinant receptor binding domain (HCR/F) containing an N-terminal 3×FLAG epitope was engineered to analyze BoNT/F interactions with neuronal receptors. The ability of HCR/F to inhibit the intracellular activity of native BoNT/F holotoxin in rat hippocampal neurons was tested to validate this model system. Incubation of rat hippocampal neurons (DIV 10-14) with HCR/F resulted in dose-dependent association of the HCR, with similar binding properties as the HCR domain of serotype A (HCR/A) (Figure 1A). As reported previously, cleavage of SNAP25 by BoNT/A was enhanced by active synaptic vesicle recycling and was inhibited with a 100-fold molar excess of HCR/A (Figure 1B) (13, 32, 36). Cleavage of VAMP-2 by BoNT/F (42) was also enhanced by active synaptic vesicle recycling, implicating a role for synaptic vesicle protein(s) in the binding and entry of BoNT/F into neurons. Exposure of neurons to 100-fold molar excess HCR/F, but not HCR/A, reduced BoNT/F cleavage of VAMP-2 several-fold (Figure 1B). These data suggest that BoNT/F utilizes a unique receptor(s) for neuronal intoxication relative to BoNT/A and that HCR/F binds to the physiological receptor of BoNT/F, which makes HCR/F a useful tool to characterize BoNT/F-host receptor interactions.

Figure 1
BoNT HCR domains inhibit binding and entry of BoNT holotoxins into primary hippocampal neurons

HCR/F binding to a glycan array

To gain insight into the mechanism of neuronal entry, the binding of HCR/F to a glycan array was determined. HCR/F labeled with Alexa-Fluor 488 (HCR/F-488) bound maximally to ganglioside GD1a, but did not interact with other ganglioside sugar moieties present on the array (Table 2 and supplemental Figure 4). Of note, the glycan array did not include the ganglioside sugar moiety of GT1b or GD1b. In addition, HCR/F bound several glycan structures that contained N-acetylactosamine [→Galβ(1→4) GlcNAcβ(1→)] and 6O- sulfated derivatives which forms the core repeating unit of keratan sulfate (Table 2). This observation suggested that HCR/F possessed two unique sugar binding domains.

Table 2
Glycan array analysis of HCR/F.

Binding of HCR/F to gangliosides requires α2,3-linked sialic acid on the terminal galactose of a neutral saccharide core

There are reported quantitative and qualitative differences in the composition of gangliosides within the nervous system (43), which may contribute to the specificity of BoNT entry into neurons. Experiments were performed to extend the glycan array analysis to determine how HCR/F recognized gangliosides (Figure 2A, gangliosides tested GM3, GM1a, GD1a, GD1b, and GT1b). While HCR/F bound to GT1b, GD1a and GM3 (potency of binding: GT1b = GD1a [dbl greater-than sign] GM3), binding to GD1b or GM1a was not observed (Figure 2B). Increasing both the amount of immobilized GD1b or GM1a to 20 μg per well and HCR/F to 10 μM did not result in detectable binding (data not shown). Under identical conditions saturable binding of tetanus HCR was detected indicating the presence of immobilized ganglioside (data not shown). These observations demonstrate that BoNT/F binds to gangliosides through interactions with an α2, 3-linked sialic acid on terminal galactose residues and that the α2,8-linked sialic acid of GT1b (denoted sia-7, Figure 2A) is not involved in ganglioside binding. The different ganglioside binding profile of HCR/F relative to HCR/T suggests that individual BoNTs utilize unique ganglioside recognition strategies.

Structure of HCR/F

The different ganglioside binding profile of HCR/F relative to HCR/T suggests these toxins may utilize unique ganglioside recognition strategies. To begin to address these potential differences, the crystal structure of HCR/F was determined. The structure of recombinant HCR/F was solved to 2.1 Å resolution by the molecular replacement method using the HCR/A structure (residues 875-1295) as the search probe. The entire HCR/F chain was visible except that the 1155-1160 and 1204-1209 loop regions and the N-terminal 36 residues (30 tagged residues plus residues 862-867) were disordered (Figure 3). The HCR/F molecule, 411 residues from D868 to N1278, had maximum dimensions of approximately 72 × 46 × 42 Å3 and was composed of two sub-domains of approximately equal size. The N-terminal domain (HCRN, residues 868 - 1073) was formed from two anti-parallel β-sheets connected such that the HCRN domain was similar to the jelly-roll domain found in many proteins. The C-terminal domain (HCRC, residues 1085-1278) was linked to HCRN through a single α-helix formed by residues (1074-1084) and adopted a modified β-trefoil fold (Figure 3).

Figure 3
A stereo diagram of superposition of the structures of HCR/A (blue) and HCR/F (pink)

The overall structure of HCR/F was similar to those of the HCR domain of native BoNT/A (pdb code: 3BTA (39)), recombinant HCR/A (Figure 3, this study), HCR/B (pdb code: 2NM1 (16)) and tetanus HCR (HCR/T, pdb code: 1YYN (44)). The root-mean-square (r.m.s.) deviation between HCR/F and HCR/A was 1.0 Å for 377 Cα atoms, 1.2 Å for 291 HCR/B Cα atoms and 1.4 Å for 272 HCR/T Cα atoms. Thus, while the overall sequence homology is relatively low among BoNT/A, /B, and /F and TeNT, the respective HCR domains show structural similarity (45). The major differences between the HCR domains resided within the loops of the HCRC sub-domain, which are involved in protein and ganglioside dual receptor recognition (46). The structure of HCR/F is also similar to the isolated HCR domain of the recently solved BoNT/E structure (pdb code: 3FFZ (47)) with a r.m.s deviation of 0.6 Å for 324 Cα atoms (Supplemental Figure 5). The high degree of structural similarity between the two proteins is consistent with their highly related primary amino acid sequences (62% amino acid identity). In particular, the putative protein receptor binding pocket is highly conserved and suggests the two serotypes may share a common protein co-receptor (Supplemental Figure 5).

Characterization of the HCR/F ganglioside binding pocket

The interaction of BoNT serotypes A and B and TeNT with gangliosides has been extensively investigated (23-27, 48-52). The co-crystal structures of HCR/T (pdb code: 1FV2) and HCR/A (pdb code: 2VU9) bound to analogs of GT1b revealed a shared mechanism of ganglioside binding (51, 52). In each case, the GalNAc-Gal disaccharide of GT1b was found to occupy a shallow cleft on the surface of the HCR. Superposition of HCR/F with the HCR/T-GT1b and HCR/A-GT1b co-crystal structures showed a spatial coincidence in the carbohydrate binding site (Supplemental Figure 6A). Similarly, HCR/F shares a conserved H…SXWY…G motif previously identified by Binz and colleagues in HCR/A, HCR/B and HCR/T (26, 27). The corresponding motif in HCR/F is represented by His1241, Ser1248, Trp1250, Tyr1251, and Gly1263 (Supplemental Figure 6B). In addition, HCR/A (Glu1203) and HCR/B (Glu1190) contribute to ganglioside binding and is conserved in HCR/F (Glu 1195).

To further characterize the putative HCR/F ganglioside binding site, a series of mutated HCR/F proteins were generated (Figure 4A). Initially, amino acids homologous to the Gal-GalNAc binding site of BoNT/A were mutated and assessed for their ability to bind gangliosides. Replacement of Trp1250 with leucine caused a complete loss of binding to ganglioside GT1b (Figure 4B), as did replacement with alanine (data not shown). This is consistent with the requirement of the corresponding Trp residue in BoNT/A, BoNT/B and TeNT for toxicity (26). Replacement of Glu1195 and Ser1248 with alanine or Tyr1251 with phenylalanine reduced binding to GT1b between ~5 and 50-fold, while replacement of Asn1254 with alanine did not significantly affect ganglioside binding (Figure 4B). Replacement of His1241 with alanine destabilized the protein and therefore was not studied further (data not shown). The structural integrity of the remaining mutated forms of HCR/F relative to the wild-type protein was estimated by performing limited proteolysis experiments measuring trypsin sensitivity (data not shown). These data suggest that BoNT/F associates with gangliosides through a conserved mechanism and that BoNT/F contains a single ganglioside binding site.

Figure 4
Binding of mutated HCR/F to ganglioside GT1b

Protein receptor recognition by BoNT/F

Previous studies have identified synaptotagmins I and II as co-receptors for BoNT/B and BoNT/G (13, 17), while synaptic vesicle protein 2 (SV2) mediates the entry of BoNT/A and BoNT/E into cultured hippocampal neurons (28, 32, 33). At present, there is no known protein receptor for BoNT/C, BoNT/D or BoNT/F. To identify putative protein co-receptors for BoNT/F, entry of the toxin into cultured rat hippocampal neurons was investigated. The observation that intoxication of hippocampal neurons by BoNT/F was stimulated by active synaptic vesicle exocytosis (Figure 1B) suggested that BoNT/F may also utilize a synaptic vesicle protein as a co-receptor.

We recently reported the high affinity association of HCR/A and HCR/B with synaptic vesicle protein complexes. Solubilization of synaptic vesicle membranes in CHAPS resulted in the recovery of a large protein complex that included synaptic vesicle protein 2 (SV2), synaptotagmin I (Syt I), synaptophysin (Syp), synaptogyrin 3, synaptobrevin 2 (Syb 2) and multiple subunits of the vacuolar type ATPase (v-ATPase) (36). Solubilization in octylglucoside preserved a SV2 / Syt I complex, while solubilization of synaptic vesicles with Triton X-100 preserved a Syt / SV2 complex and a Syp / Syb 2 complex. These protein interactions were proposed to underlie aspects of neurotransmitter secretion, vesicle trafficking and spatial organization within the nerve terminal. A model was proposed whereby these presynaptic protein complexes form the physiologic receptor for BoNTs (36).

Utilizing the same strategy, interactions between synaptic vesicle proteins and HCR/F were analyzed. Precipitation of HCR/F from CHAPS detergent extracts resulted in the co-precipitation of four synaptic vesicle proteins as determined by Western blotting (SV2, Synaptotagmin I, synaptophysin, and synaptobrevin 2) (Figure 5). The specificity of the association of this protein complex with HCR/F was confirmed in control reactions (beads alone, 3×FLAG-GFP) where co-precipitation of the protein complex was not detected (Figure 5 and data not shown). The analysis was repeated using an extract prepared in Triton X-100. Under these conditions, HCR/F co-purified with synaptic vesicle protein 2 (SV2) and synaptotagmin I (Syt I) while other components of the neurotransmitter protein complex were not detected (Figure 5). The co-purification of SV2 and synaptotagmin I under both conditions is consistent with previous studies reporting the stable association of the two proteins (53-55). Together, these observations indicate that HCR/F can bind to a similar synaptic vesicle protein complex as previously observed for HCR/A and HCR/B (36) and is consistent with BoNT/F intoxication of neurons being enhanced by active synaptic vesicle recycling.

Figure 5
Co-immunoprecipitation of synaptic vesicle proteins with HCR/B and HCR/F

Interaction of HCR/F with glycosylated SV2

Co-purification of HCR/F with SV2/ Syt I complexes in both CHAPS and Triton X-100 extracts suggested that HCR/F may interact directly with one or both of the proteins. Indeed, BoNT/A and BoNT/B are known to directly interact with recombinant fragments of SV2 and Syt I respectively. However, under these conditions, no interaction of HCR/F with either SV2 or Syt I was observed (17, 32, 33). A recent study by Chapman and colleagues demonstrated that entry of BoNT/E into cultured hippocampal neurons was dependent on expression of SV2A and SV2B (28). The authors demonstrated that entry of BoNT/E was dependent on covalent modification of SV2, as a glycosylation deficient form of SV2 was unable to mediate BoNT/E entry. The high degree of sequence and structural homology between BoNT/E and BoNT/F suggested that the interaction of BoNT/F with SV2 or Syt I may also be dependent on covalent modification of the vesicle proteins. Moreover, the observation that HCR/F bound glycans with keratan sulfate moieties (Table 2) suggested that the interaction of HCR/F with SV2 or Syt I could be dependent on protein glycosylation (28).

Previous reports demonstrated SV2 is a keratan sulfate (KS) proteoglycan (56, 57). Treatment of SV2 with either protein N- Glycosidase F (PNGase F) or endo-β-galactosidase alone resulted in modest changes in apparent molecular weight (56). This suggests that most of the KS chains are resistant to cleavage by endo-β-galactosidase or keratanases I and II (56). While the exact nature of the SV2 KS chains is unknown, modification of the core N-acetylactosamine disaccharide in KS chains with fucose and sialic acid is commonly found. To determine whether glycosylation of SV2 contributes to HCR/E and HCR/F binding, conditions were established to deglycosylate native SV2 using endo-β-galactosidase. Consistent with previous reports (56), treatment of lysates with endo-β-galactosidase removed a small amount of carbohydrate from SV2 (Figure 6). Precipitation of HCR/E or HCR/F from lysates incubated with endo-β-galactosidase reduced the co-purification of the SV2/ Syt I complex relative to control lysates as determined by Western blotting (Figure 6). Treatment of lysates with a mixture of α(1-3,4) fucosidase, endo-β-galactosidase and PNGase F caused a loss of the high molecular weight forms of SV2 and inhibited the interaction of SV2 with HCR/E and HCR/F (Figure 6). Under the conditions used, significant change in the apparent molecular mass of Syt I was not observed upon PNGase F treatment, suggesting the glycosidases were primarily targeting SV2. Preliminary studies of HCR/A interaction with deglycosylated SV2 have been performed, but the data at this point is inconclusive. In contrast to HCR/E and HCR/F, treatment of lysates with the mixture of glycosidases did not inhibit the interaction of HCR/B with the SV2/ Syt I complex (Figure 6). This is consistent with previous reports demonstrating the direct interaction of the Syt I and Syt II polypeptides with BoNT/B (13, 16, 17). Together, these data suggest that both HCR/E and HCR/F bind to synaptic vesicle protein complexes primarily through the keratan sulfate moieties of native SV2. Identification of the precise oligosaccharide structure directly binding to HCR/E and HCR/F will require a detailed characterization of the SV2 KS moiety.

Figure 6
Glycosylation of SV2 is required for interaction with HCR/E and HCR/F

Ganglioside recognition is required for SV2 interaction and neuron binding

To further elucidate the contributions of ganglioside binding to the intoxication process, the ability of HCR/FW1250L to bind to synaptic vesicle protein complexes and cultured hippocampal neurons was investigated. Under conditions where HCR/B and wild-type HCR/F bound the SV2/ Syt I complex, binding of HCR/FW1250L to the SV2/ SytI complex was not observed (Figure 7A). HCR/FW1250L did not bind crude synaptic vesicle membranes (Figure 7B) and binding to rat hippocampal neurons was reduced relative to HCR/F (Figure 7C), further supporting a role for ganglioside in protein co-receptor recognition.

Figure 7
Ganglioside binding is required for SV2 interaction and neuronal binding

The observation that BoNT/E and BoNT/F recognize the KS moieties of SV2 suggests that these toxins possess a second carbohydrate binding site in addition to the characterized ganglioside binding pocket. Several studies have demonstrated that tetanus HCR (HCR/T) possesses two carbohydrate binding pockets (27, 44, 51, 58). The first characterized ganglioside binding pocket is shared by HCR/T, HCR/A, HCR/B, HCR/E and HCR/F (Supplemental Figure 6). The second carbohydrate binding pocket of tetanus neurotoxin is located at the distal tip of the HCR and overlaps with the synaptotagmin II binding site of BoNT/B (16, 44). Arginine 1226 of HCR/T directly interacts with the di-sialic acid moieties of gangliosides GT1b and GD3 and is required for both carbohydrate binding and cellular toxicity (27, 51). Superposition of HCR/E and HCR/F with HCR/T identified Lys1176 and Lys1199 of HCR/E and HCR/F respectively as the direct equivalent residues to tetanus Arg1226 (data not shown). Thus, HCR/E and HCR/F may also coordinate an oligosaccharide at this position (Figure 8A).

Figure 8
Model of HCR/F and HCR/E bound to dual carbohydrate receptors

The data presented here indicate that the “dual receptor” model can be applied to BoNT/F with the novel observation that the sugar moiety of SV2 has replaced the protein co-receptor of BoNT/B (Figure 8A). BoNT/F may interact with ganglioside concentrating the toxin on the presynaptic membrane and subsequent to the initial capture, fusion of synaptic vesicles with the plasma membrane exposes synaptic vesicle protein luminal domains to the extracellular milieu allowing interaction of BoNT/F with the keratan sulfate moiety of SV2 and uptake through clathrin-mediated endocytosis (Figure 8B).

Supplementary Material



We acknowledge the assistance of Amanda Przedpelski for HCR production. We thank the staff at the Advanced Photon Source beamline SBC 19ID for their excellent assistance in data collection.


botulinum neurotoxin serotype A
botulinum neurotoxin serotype F
Heavy chain receptor binding domain of BoNT serotype F
Heavy chain receptor binding domain of BoNT serotype A
Heavy chain receptor binding domain of Tetanus neurotoxin
vesicle associated membrane protein 2
synaptosomal-associated protein of 25 kDa
soluble NSF attachment receptor
days in vitro
keratan sulfate


This work was supported by grants from the Great Lakes Regional Center of Excellence (GLRCE) from NIH-NIAID U54 AI057153 (JTB, and J-J K) and NIH-NINDS K99 NS061763 (MRB). Results shown in this report are derived from work performed at Argonne National Laboratory, Structural Biology Center at the Advanced Photon Source. Argonne is operated by UChicago Argonne, LLC, for the U.S. Department of Energy, Office of Biological and Environmental Research under contract DE-AC02-06CH11357.

The atomic coordinates and structure factors (pdb code: 3FUO for HCR/A and 3FUQ for HCR/F) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ.

Supporting Information Available” Supporting information showing structure based alignments of HCR/F with HCR/A, the HCR/F glycan binding array, and the isolated HCR domain of BoNT/E and HCR/T are available free of charge via the Internet at


1. Arnon SS, Schechter R, Inglesby TV, Henderson DA, Bartlett JG, Ascher MS, Eitzen E, Fine AD, Hauer J, Layton M, Lillibridge S, Osterholm MT, O'Toole T, Parker G, Perl TM, Russell PK, Swerdlow DL, Tonat K. Botulinum toxin as a biological weapon: medical and public health management. Jama. 2001;285:1059–1070. [PubMed]
2. Hatheway CL. Botulism: the present status of the disease. Curr Top Microbiol Immunol. 1995;195:55–75. [PubMed]
3. Johnson EA. Clostridial toxins as therapeutic agents: benefits of nature's most toxic proteins. Annu Rev Microbiol. 1999;53:551–575. [PubMed]
4. Johnson EA, Bradshaw M. Clostridium botulinum and its neurotoxins: a metabolic and cellular perspective. Toxicon. 2001;39:1703–1722. [PubMed]
5. Moller V, Scheibel I. Preliminary report on the isolation of an apparently new type of CI. botulinum. Acta Pathol Microbiol Scand. 1960;48:80. [PubMed]
6. Rymkiewicz D. Studies on Clostridium botulinum type F. I. Properties of selected cultures of the Langeland strain. Med Dosw Mikrobiol. 1968;20:169–174. [PubMed]
7. Barash JR, Tang TW, Arnon SS. First case of infant botulism caused by Clostridium baratii type F in California. J Clin Microbiol. 2005;43:4280–4282. [PMC free article] [PubMed]
8. East AK, Richardson PT, Allaway D, Collins MD, Roberts TA, Thompson DE. Sequence of the gene encoding type F neurotoxin of Clostridium botulinum. FEMS Microbiol Lett. 1992;75:225–230. [PubMed]
9. Eklund MW, Poysky FT, Wieler DI. Characteristics of Clostridium botulinum type F isolated from the Pacific Coast of the United States. Appl Microbiol. 1967;15:1316–1323. [PMC free article] [PubMed]
10. Harvey SM, Sturgeon J, Dassey DE. Botulism due to Clostridium baratii type F toxin. J Clin Microbiol. 2002;40:2260–2262. [PMC free article] [PubMed]
11. Smith TJ, Lou J, Geren IN, Forsyth CM, Tsai R, Laporte SL, Tepp WH, Bradshaw M, Johnson EA, Smith LA, Marks JD. Sequence variation within botulinum neurotoxin serotypes impacts antibody binding and neutralization. Infect Immun. 2005;73:5450–5457. [PMC free article] [PubMed]
12. Montecucco C, Rossetto O, Schiavo G. Presynaptic receptor arrays for clostridial neurotoxins. Trends Microbiol. 2004;12:442–446. [PubMed]
13. Dong M, Richards DA, Goodnough MC, Tepp WH, Johnson EA, Chapman ER. Synaptotagmins I and II mediate entry of botulinum neurotoxin B into cells. J Cell Biol. 2003;162:1293–1303. [PMC free article] [PubMed]
14. Chai Q, Arndt JW, Dong M, Tepp WH, Johnson EA, Chapman ER, Stevens RC. Structural basis of cell surface receptor recognition by botulinum neurotoxin B. Nature. 2006;444:1096–1100. [PubMed]
15. Dong M, Tepp WH, Liu H, Johnson EA, Chapman ER. Mechanism of botulinum neurotoxin B and G entry into hippocampal neurons. J Cell Biol. 2007;179:1511–1522. [PMC free article] [PubMed]
16. Jin R, Rummel A, Binz T, Brunger AT. Botulinum neurotoxin B recognizes its protein receptor with high affinity and specificity. Nature. 2006;444:1092–1095. [PubMed]
17. Rummel A, Karnath T, Henke T, Bigalke H, Binz T. Synaptotagmins I and II act as nerve cell receptors for botulinum neurotoxin G. J Biol Chem. 2004;279:30865–30870. [PubMed]
18. Sonnino S, Mauri L, Chigorno V, Prinetti A. Gangliosides as components of lipid membrane domains. Glycobiology. 2007;17:1R–13R. [PubMed]
19. Abad-Rodriguez J, Robotti A. Regulation of axonal development by plasma membrane gangliosides. J Neurochem. 2007;103 1:47–55. [PubMed]
20. Bullens RW, O'Hanlon GM, Wagner E, Molenaar PC, Furukawa K, Plomp JJ, Willison HJ. Roles of complex gangliosides at the neuromuscular junction. Ann N Y Acad Sci. 2003;998:401–403. [PubMed]
21. Yowler BC, Kensinger RD, Schengrund CL. Botulinum neurotoxin A activity is dependent upon the presence of specific gangliosides in neuroblastoma cells expressing synaptotagmin I. J Biol Chem. 2002;277:32815–32819. [PubMed]
22. Kitamura M, Takamiya K, Aizawa S, Furukawa K. Gangliosides are the binding substances in neural cells for tetanus and botulinum toxins in mice. Biochim Biophys Acta. 1999;1441:1–3. [PubMed]
23. Nishiki T, Tokuyama Y, Kamata Y, Nemoto Y, Yoshida A, Sato K, Sekiguchi M, Takahashi M, Kozaki S. The high-affinity binding of Clostridium botulinum type B neurotoxin to synaptotagmin II associated with gangliosides GT1b/GD1a. FEBS Lett. 1996;378:253–257. [PubMed]
24. Ochanda JO, Syuto B, Ohishi I, Naiki M, Kubo S. Binding of Clostridium botulinum neurotoxin to gangliosides. J Biochem (Tokyo) 1986;100:27–33. [PubMed]
25. Kitamura M, Iwamori M, Nagai Y. Interaction between Clostridium botulinum neurotoxin and gangliosides. Biochim Biophys Acta. 1980;628:328–335. [PubMed]
26. Rummel A, Mahrhold S, Bigalke H, Binz T. The HCC-domain of botulinum neurotoxins A and B exhibits a singular ganglioside binding site displaying serotype specific carbohydrate interaction. Mol Microbiol. 2004;51:631–643. [PubMed]
27. Rummel A, Bade S, Alves J, Bigalke H, Binz T. Two carbohydrate binding sites in the H(CC)-domain of tetanus neurotoxin are required for toxicity. J Mol Biol. 2003;326:835–847. [PubMed]
28. Dong M, Liu H, Tepp WH, Johnson EA, Janz R, Chapman ER. Glycosylated SV2A and SV2B Mediate the Entry of Botulinum Neurotoxin E into Neurons. Mol Biol Cell 2008 [PMC free article] [PubMed]
29. Nishiki T, Kamata Y, Nemoto Y, Omori A, Ito T, Takahashi M, Kozaki S. Identification of protein receptor for Clostridium botulinum type B neurotoxin in rat brain synaptosomes. J Biol Chem. 1994;269:10498–10503. [PubMed]
30. Nishiki T, Ogasawara J, Kamata Y, Kozaki S. Solubilization and characterization of the acceptor for Clostridium botulinum type B neurotoxin from rat brain synaptic membranes. Biochim Biophys Acta. 1993;1158:333–338. [PubMed]
31. Nishiki T, Tokuyama Y, Kamata Y, Nemoto Y, Yoshida A, Sekiguchi M, Takahashi M, Kozaki S. Binding of botulinum type B neurotoxin to Chinese hamster ovary cells transfected with rat synaptotagmin II cDNA. Neurosci Lett. 1996;208:105–108. [PubMed]
32. Dong M, Yeh F, Tepp WH, Dean C, Johnson EA, Janz R, Chapman ER. SV2 Is the Protein Receptor for Botulinum Neurotoxin A. Science 2006 [PubMed]
33. Mahrhold S, Rummel A, Bigalke H, Davletov B, Binz T. The synaptic vesicle protein 2C mediates the uptake of botulinum neurotoxin A into phrenic nerves. FEBS Lett. 2006;580:2011–2014. [PubMed]
34. Baldwin MR, Tepp WH, Przedpelski A, Pier CL, Bradshaw M, Johnson EA, Barbieri JT. Subunit vaccine against the seven serotypes of botulism. Infect Immun. 2008;76:1314–1318. [PMC free article] [PubMed]
35. Huttner WB, Schiebler W, Greengard P, De Camilli P. Synapsin I (protein I), a nerve terminal-specific phosphoprotein. III. Its association with synaptic vesicles studied in a highly purified synaptic vesicle preparation. J Cell Biol. 1983;96:1374–1388. [PMC free article] [PubMed]
36. Baldwin MR, Barbieri JT. Association of botulinum neurotoxin serotypes A and B with synaptic vesicle protein complexes. Biochemistry. 2007;46:3200–3210. [PubMed]
37. Bochner BS, Alvarez RA, Mehta P, Bovin NV, Blixt O, White JR, Schnaar RL. Glycan array screening reveals a candidate ligand for Siglec-8. J Biol Chem. 2005;280:4307–4312. [PubMed]
38. Otwinowski Z, Minor W. Processing of X-ray Diffraction Data Collected in Oscillation Mode. Methods in Enzymology. 1997;276:307–326.
39. Lacy DB, Tepp W, Cohen AC, DasGupta BR, Stevens RC. Crystal structure of botulinum neurotoxin type A and implications for toxicity. Nat Struct Biol. 1998;5:898–902. [PubMed]
40. Brunger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges M, Pannu NS, Read RJ, Rice LM, Simonson T, Warren GL. Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr D Biol Crystallogr. 1998;54:905–921. [PubMed]
41. Roussel A, Inisan AG. Turbo-Frodo, Version OpenGL:1. Marseille, France: CNRS/Universitite - Marseille; 1999.
42. Schiavo G, Benfenati F, Poulain B, Rossetto O, Polverino de Laureto P, DasGupta BR, Montecucco C. Tetanus and botulinum-B neurotoxins block neurotransmitter release by proteolytic cleavage of synaptobrevin. Nature. 1992;359:832–835. [PubMed]
43. Ogawa-Goto K, Abe T. Gangliosides and glycosphingolipids of peripheral nervous system myelins--a minireview. Neurochem Res. 1998;23:305–310. [PubMed]
44. Jayaraman S, Eswaramoorthy S, Kumaran D, Swaminathan S. Common binding site for disialyllactose and tri-peptide in C-fragment of tetanus neurotoxin. Proteins. 2005;61:288–295. [PubMed]
45. Umland TC, Wingert LM, Swaminathan S, Furey WF, Schmidt JJ, Sax M. Structure of the receptor binding fragment HC of tetanus neurotoxin. Nat Struct Biol. 1997;4:788–792. [PubMed]
46. Lalli G, Herreros J, Osborne SL, Montecucco C, Rossetto O, Schiavo G. Functional characterisation of tetanus and botulinum neurotoxins binding domains. J Cell Sci. 1999;112(Pt 16):2715–2724. [PubMed]
47. Kumaran D, Eswaramoorthy S, Furey W, Navaza J, Sax M, Swaminathan S. Domain Organization in Clostridium botulinum Neurotoxin Type E Is Unique: Its Implication in Faster Translocation. J Mol Biol 2008 [PubMed]
48. Yowler BC, Schengrund CL. Botulinum neurotoxin A changes conformation upon binding to ganglioside GT1b. Biochemistry. 2004;43:9725–9731. [PubMed]
49. Kozaki S, Kamata Y, Watarai S, Nishiki T, Mochida S. Ganglioside GT1b as a complementary receptor component for Clostridium botulinum neurotoxins. Microb Pathog. 1998;25:91–99. [PubMed]
50. Kitamura M, Igimi S, Furukawa K, Furukawa K. Different response of the knockout mice lacking b-series gangliosides against botulinum and tetanus toxins. Biochim Biophys Acta. 2005;1741:1–3. [PubMed]
51. Fotinou C, Emsley P, Black I, Ando H, Ishida H, Kiso M, Sinha KA, Fairweather NF, Isaacs NW. The crystal structure of tetanus toxin Hc fragment complexed with a synthetic GT1b analogue suggests cross-linking between ganglioside receptors and the toxin. J Biol Chem. 2001;276:32274–32281. [PubMed]
52. Stenmark P, Dupuy J, Imamura A, Kiso M, Stevens RC. Crystal structure of botulinum neurotoxin type A in complex with the cell surface co-receptor GT1b-insight into the toxin-neuron interaction. PLoS Pathog. 2008;4:e1000129. [PMC free article] [PubMed]
53. Schivell AE, Batchelor RH, Bajjalieh SM. Isoform-specific, calcium-regulated interaction of the synaptic vesicle proteins SV2 and synaptotagmin. J Biol Chem. 1996;271:27770–27775. [PubMed]
54. Schivell AE, Mochida S, Kensel-Hammes P, Custer KL, Bajjalieh SM. SV2A and SV2C contain a unique synaptotagmin-binding site. Mol Cell Neurosci. 2005;29:56–64. [PubMed]
55. Lazzell DR, Belizaire R, Thakur P, Sherry DM, Janz R. SV2B regulates synaptotagmin 1 by direct interaction. J Biol Chem. 2004;279:52124–52131. [PubMed]
56. Scranton TW, Iwata M, Carlson SS. The SV2 protein of synaptic vesicles is a keratan sulfate proteoglycan. J Neurochem. 1993;61:29–44. [PubMed]
57. Sinouris EA, Skandalis SS, Kilia V, Theocharis AD, Theocharis DA, Ravazoula P, Vynios DH, Papageorgakopoulou N. Keratan sulfate-containing proteoglycans in sheep brain with particular reference to phosphacan and synaptic vesicle proteoglycan isoforms. Biomed Chromatogr 2008 [PubMed]
58. Chen C, Baldwin MR, Barbieri JT. Molecular basis for tetanus toxin coreceptor interactions. Biochemistry. 2008;47:7179–7186. [PubMed]