Transport of the botulinum neurotoxin–associating protein, nontoxic nonhemagglutinin, across the rat small intestinal epithelial cell monolayer


Correspondence: Toshihiro Watanabe, Department of Food and Cosmetic Science, Faculty of Bioindustry, Tokyo University of Agriculture, 196 Yasaka, Abashiri 099-2493, Japan. Tel.: +81 152 48 3843; fax: +81 152 48 2940; e-mail:


Botulinum neurotoxin (BoNT) associates with nontoxic nonhemagglutinin (NTNHA) yielding a complex in culture. BoNT and NTNHA have similar domain organizations, implying that they share common functions, although this remains unclear. Here, we examined cell monolayer transport of serotype D NTNHA in the rat intestinal epithelial cell line IEC-6. NTNHA and BoNT both bound to the cell and were transported across the cell layer. NTNHA contains a QXW motif and a β-trefoil fold, both common in sugar chain–recognizing proteins, whereas the QXW motif is absent in all BoNT serotypes. This could explain the distinct sugar chain–recognizing properties of NTNHA and BoNT.


Foodborne botulism is a highly fatal disease caused by the botulinum neurotoxin (BoNT; 150 kDa) produced by Clostridium botulinum. BoNT may be classified into seven immunologically distinct serotypes, termed A–G. Serotypes A, B, E, and F BoNTs are causative agents of the human botulism, whereas serotypes C and D are primarily responsible for animal and avian botulism (Montecucco & Schiavo, 1994; Li & Singh, 1999). BoNT that is ingested along with polluted foods travels through the digestive system and reaches the small intestine, where it passes through the intestinal wall and enters the bloodstream. Finally, BoNT penetrates into nerve cells at neuromuscular junctions, inhibiting the release of neurotransmitters and resulting in muscular paralysis that can result in death in the worst cases.

In BoNT culture supernatants and in foods contaminated with the bacteria, BoNT forms a toxin complex (TC) in conjunction with auxiliary nontoxic proteins. Specifically, BoNT combines with a 130-kDa nontoxic nonhemagglutinin (NTNHA) to create M-TC (serotypes A–F). Further association of the M-TC with hemagglutinins (HAs: HA-70, HA-33, and HA-17; 70, 33, and 17 kDa, respectively) leads to formation of the L-TC (serotypes A–D and G; Kouguchi et al., 2002). The nontoxic proteins protect BoNT under harsh digestive conditions (Niwa et al., 2007; Miyata et al., 2009). Furthermore, the HA component, especially HA-33, facilitates transport of the TC across the small intestinal epithelial cell monolayer (Ito et al., 2011).

We recently revealed that NTNHA and BoNT have a similar domain architecture comprised of metalloprotease-like, coiled-coil, and concanavalin A-like domains, implying that both proteins evolved from a common ancestor molecule (Inui et al., 2012). This structural similarity is further supported by their respective crystal structures, which were determined recently (Gu et al., 2012; Sagane et al., 2012). BoNT, when free from auxiliary nontoxic components, binds to and is transported across the cell monolayer of intestinal epithelial cell lines (Niwa et al., 2007; Ito et al., 2011; Thirunavukkarasusx et al., 2011). On the other hand, NTNHA has not yet been shown to have similar cell binding and cell layer transport properties. Based on the structural similarity between these two proteins, we speculated that NTNHA may have similar cell transport properties to BoNT. This study examines the cell binding and cell monolayer transport properties of serotype D NTNHA using the rat small intestinal cell line IEC-6.

Materials and methods

Production and purification of botulinum toxins

Clostridium botulinum serotype D strain 4947 was cultured using the dialysis tube method (Hasegawa et al., 2004). The culture supernatant was brought to 60% saturation with ammonium sulfate, centrifuged, and the resulting precipitate dissolved and dialyzed against 50 mM acetate buffer (pH 4.0) containing 0.2 M NaCl. The dialysate was then applied to a TOYOPEARL SP-650S (Tosoh, Tokyo, Japan) cation exchange column (1.6 × 40 cm) equilibrated with the dialysis buffer. The adsorbed protein was eluted with a linear NaCl gradient (0.2–0.8 M). The fractions containing M-TC, identified by SDS-PAGE and native PAGE, were collected, and the protein precipitated with 80% saturation ammonium sulfate. The precipitated M-TC was dissolved in 50 mM acetate buffer (pH 5.0) containing 0.15 M NaCl and then applied to a HiLoad 16/60 Superdex 200 pg gel filtration column (GE Healthcare, Little Chalfont, UK) equilibrated with the same buffer.

To dissociate the NTNHA and BoNT, the M-TC was dialyzed against 20 mM Tris-HCl buffer (pH 8.8) containing 0.4 M NaCl. The dialysate was applied to a HiLoad 16/60 Superdex 200 pg gel filtration column equilibrated with the dialysis buffer. To obtain highly purified BoNT and NTNHA, the BoNT and NTNHA fractions were desalted separately by dialysis against 20 mM Tris-HCl buffer (pH 8.8) and then applied to a Mono Q HR 5/5 (GE Healthcare) equilibrated with dialysis buffer. Elution of the adsorbed protein was performed with a linear NaCl gradient concentration (0–0.4 M).

Cell culture

The rat small intestine epithelial cell line IEC-6 was obtained from RIKEN BioResource Center (Tsukuba, Japan). Cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS), penicillin (100 IU mL−1), and streptomycin (100 μg mL−1). Cells were maintained in a humidified environment of 5% CO2 at 37 °C. The culture medium was renewed every 2–3 days.


Rabbit polyclonal antibodies were raised against BoNT and NTNHA, respectively (Niwa et al., 2007).

Fluorescent microscopy and confocal analysis

NTNHA and BoNT were labeled with Cy3 (GE Healthcare), according to the manufacturer's recommendation. Labeled proteins were separated from free dye by column chromatography, and the degree of labeling was determined based on the measurements of the Cy3 (absorbance at 552 nm) and protein concentration (absorbance at 280 nm). IEC-6 cells were first seeded on a glass chamber slide (Nunc A/S, Roskilde, Denmark). For binding assays, cells were incubated with Cy3-labeled proteins (80 nM) diluted in Krebs–Henseleit buffer solution (KHB: 10 mM HEPES, pH 7.0, 130 mM NaCl, 1 mM NaH2PO4, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 5 mM glucose) for 1 h at 4 °C, then rinsed with cold KHB three times and fixed with 4% paraformaldehyde for 20 min. The cells were then rinsed with KHB three times and incubated with KHB containing DY-490 Phalloidin (10 U mL−1; Dyomics GmbH, Jena, Germany) for 1 h. Cells were then observed under a fluorescent microscope (Axiovert 40 CFL; CarlZeiss, Jena, Germany) with standard FITC excitation/emission filters for DY-490 and rhodamine filters for Cy3.

For the endocytosis assays, cells were incubated with Cy3-labeled proteins (200 nM) diluted in KHB for 30 min on ice, then rinsed with cold KHB three times. The cells were incubated at 37 °C for 10–20 min to allow incorporation of the proteins into the cells and then fixed with 4% paraformaldehyde for 20 min. After rinsing with KHB three times, they were incubated with KHB containing DY-490 Phalloidin (10 U mL−1) and DAPI (15 μg mL−1; Invitrogen, Carlsbad, CA) for 1 h. Cells were analyzed with a Leica TCS laser scanning confocal microscope (Plan Apo 63× oil immersion 1.25 NA objective) using leica v2.5 software.

Western-blot-based toxin binding and transport assay

NTNHA binding to IEC-6 cells and transport through the cell monolayer were assayed, as described previously (Niwa et al., 2007) with minor modifications. IEC-6 cells were prepared in 24-well dishes (Corning, Corning, NY) and grown to confluence. Each protein was suspended in 300 μL of KHB at the indicated concentrations and added to culture dishes at 4 °C. Cells were then incubated with NTNHA for 1 h at 4 °C, rinsed three times with cold KHB and lysed with 150 μL SDS buffer. Proteins bound to cells in 20 μL of the treated sample were separated on SDS-PAGE and detected by Western blot analysis.

For protein transport assays, cells were grown in Transwell culture inserts comprised of a two-compartment culture system separated by a polycarbonate membrane with a 0.4-μm pore size (Corning). Cells were seeded at a confluent density (5.0 × 105 cells cm−2) on the bottom membrane of the culture insert. The volumes of culture medium inside and outside of the culture insert were 200 and 900 μL, respectively. Cells were cultured for 5 days to allow formation of tight connections. NTNHA or BoNT was suspended in 200 μL of DMEM containing 5% FBS. Cells were incubated in a 5% CO2 incubator at 37 °C. The culture medium in the outer chamber was collected and treated with 100 μL of 3× SDS buffer. Samples were then subjected to SDS-PAGE and Western blot.

Western blot

Samples from the binding and transport assays were electrophoresed on 10% SDS gels. The separated proteins were blotted onto a nitrocellulose membrane (GE Healthcare) and incubated with antibodies diluted 1 : 200 in TBST (20 mM Tris-HCl buffer, pH 7.5, 150 mM NaCl, and 0.1% Tween) with 5% skimmed milk at 4 °C overnight. After rinsing three times with TBST, the membrane was incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Santa Cruz Biotechnology, Dallas, TX) diluted 1 : 5000 in TBST containing 5% skimmed milk for 1 h at room temperature. After rinsing, proteins were visualized by chemiluminescence detection. Goat anti-actin antibody and donkey anti-goat IgG antibody conjugated with HRP (Santa Cruz Biotechnology) were used to detect actin.

Measurement of transepithelial electrical resistance and the paracellular tracer flux assay

To examine the integrity of the cell layer, the transepithelial electrical resistance (TER) was monitored. The electrical resistance was measured using a Millicell-ERS Epithelial Volt-Ohm Meter (Millipore). Inserts with no cell monolayers served as blanks to determine the baseline resistance. The TER (Ω cm2) was calculated using the following formula: (TERsample − TERblank) × surface area.

The paracellular flux of FITC-dextran was also assessed to examine the integrity of the cell layer. FITC-dextran with molecular mass of 70 kDa (10 μM; Sigma-Aldrich, St. Louis, MO) was poured into the apical chamber of the Transwell. After 1-h incubation, the basolateral chamber media were collected, and the amount of FITC-dextran in the medium was measured with a spectrofluorometer (FP-8200, JASCO, Tokyo, Japan; excitation 495 nm, emission 510 nm).


NTNHA and BoNT were isolated from the M-TC (a complex of BoNT/NTNHA) produced by C. botulinum D-4947. As shown in Fig. 1(a), the isolated BoNT and NTNHA displayed 150-kDa single and 130- and 115-kDa bands on SDS-PAGE, respectively. Previously, we found that the 130-kDa NTNHA isolated from the D-4947 M-TC spontaneously converted into a nicked form generating N-terminal 15-kDa and C-terminal 115-kDa fragments detectable on SDS-PAGE (Sagane et al., 2001). Thus, NTNHA in this sample solution existed as a mixture of both intact and nicked forms.

Figure 1.

Binding of NTNHA and BoNT to IEC-6 cells. (a) SDS-PAGE banding profiles of the NTNHA and BoNT proteins used in this study. The proteins were developed on a 13.6% polyacrylamide gel and stained with Coomassie Brilliant Blue. The molecular weights of the molecular-weight standards (lane 1) are labeled on the left in kilo Dalton. NTNHA (lane 2) displayed two bands with molecular weights of 130 and 115 kDa, whereas BoNT (lane 3) exhibited a single 150-kDa band. (b) Binding of Cy3-labeled NTNHA and BoNT to IEC-6 cells. The cells were incubated with the labeled proteins (80 nM) for 1 h at 4 °C. Control cells (Cont) were incubated under the same conditions without added protein. Actin filaments were visualized with DY-490 phalloidin. Original magnification: 200×. Experiments were repeated in triplicate, and representative data are shown. Scale bar indicating 50 μm at lower left panel applies to all images. (c) Fluorescent measurements of the cells treated with Cy3-labeled NTNHA and BoNT. The cells cultured in 96-well microplates (Corning) were treated with the indicated amounts of Cy3-labeled NTNHA and BoNT for 1 h at 4 °C. The cells were rinsed with KHB and lysed in KHB containing 0.4% Triton X-100. The fluorescent intensity was measured at excitation wavelength of 550 nm and emission wavelength of 570 nm. Experiments were repeated in triplicate, and error bars represent the SD. (d) Western-blot-based analysis of cell binding. Cells were incubated with NTNHA or BoNT at the indicated concentrations for 1 h at 4 °C, then lysed with SDS buffer and applied to the SDS-PAGE followed by Western blotting using an anti-NTNHA or anti-BoNT antibody. Experiments were repeated in triplicate, and representative data are shown.

Purified NTNHA and BoNT were labeled with fluorescent Cy3 dye to detect their binding to cells. The moles of dye per mole of labeled protein were estimated to be 8.5 for NTNHA and 9.2 for BoNT, based on dye/protein ratios. The labeled proteins were added to an IEC-6 cell culture and incubated for 1 h at 4 °C. Concomitantly, the cytoskeletal actin was stained with phalloidin to visualize the outline of the cell structure. As a result (Fig. 1b), fluorescence was detected from cells treated with Cy3-labeled NTNHA and with BoNT, indicating that NTNHA, as well as BoNT, bound to IEC-6 cells. The cells treated with labeled proteins were collected, and the fluorescence due to labeled proteins bound to the cells was measured using a fluorescence spectrophotometer. As shown in Fig. 1(c), fluorescence of the Cy3-labeled NTNHA-treated cells increased dose dependently. Binding of NTNHA and BoNT was further examined by a Western-blot-based method. Nonlabeled NTNHA and BoNT (20, 40, and 80 nM) were respectively added to an IEC-6 cell culture and then incubated at 4 °C for 1 h. Proteins bound to the cells were detected by Western blot using anti-NTNHA or anti-BoNT. The binding of NTNHA, as well as that of BoNT, increased dose dependently (Fig. 1d).

Endocytosis of Cy3-labeled NTNHA and BoNT into IEC-6 cells was analyzed using confocal microscopy. Cells cultured on glass chamber slides were exposed to Cy3-labeled NTNHA or BoNT (200 nM) on ice for 30 min to allow binding of the proteins to the cells. After rinsing with cold KHB buffer, the cells were incubated at 37 °C for 10–20 min to initiate internalization of the proteins. As shown in Fig. 2, Cy3-labeled NTNHA, as well as Cy3-labeled BoNT, was internalized. Interestingly, most of the Cy3-labeled NTNHA and BoNT proteins were adjoined to actin. This indicated that intracellular transport of NTNHA and BoNT may depend on actin fibers.

Figure 2.

Internalization of NTNHA and BoNT in IEC-6 cells. IEC-6 cells cultured on a glass chamber slide were exposed to 200 nM Cy3-labeled NTNHA or BoNT for 30 min on ice. The slide was then incubated at 37 °C for 10–20 min. The actin and chromosomes were stained with phalloidin and DAPI, respectively, and the specimens were analyzed by confocal microscopy. Experiments were repeated in triplicate, and representative data are shown. Scale bar indicates 10 μm. Images in the left and central panels display confocal sections of the cells showing the Cy3 signal (left) and a merged image of Cy3, actin, and chromosomal DNA (central). The images in the right panels show 2.7-fold zooms of regions indicated by dashed squares in the central panels.

The transport assay results for NTNHA and BoNT were also examined. Each protein (80 nM) was poured into the upper chamber of the Transwell that was separated by an IEC-6 cell monolayer, and then, protein transported across the layer was detected by Western blot (Fig. 3). Both proteins moved across the monolayer in an incubation time-dependent manner. To the best of our knowledge, this is the first demonstration that NTNHA, as well as BoNT, binds to and is internalized by IEC-6 cells and then transported across the cell monolayer.

Figure 3.

Cell layer transport of NTNHA and BoNT in IEC-6 cells. (a) Integrity of the cell layer. Cell monolayers were prepared on semi-permeable membranes of Transwell two-chamber systems. The integrity of the cell monolayer was examined based on the TER and 70-kDa FITC-dextran fluxes through the layer over a 1-h incubation (= 6, means ± SD). At day 4 of the cultivation, the TER reached a maximum and the flux of the dextran through the layer was 0.001% of the original solution. (b) NTNHA or BoNT at 80 nM was added to the culture medium on the apical side, and cells were incubated for the indicated periods of time. Proteins transported through the layer were collected from the medium in the basal side and detected by Western blot using an anti-NTNHA or anti-BoNT antibody. Experiments were repeated in triplicate, and representative data are shown. (c) TER measurements during transport of NTNHA and BoNT (= 6, means ± SD). The TER among the untreated, NTNHA-, and BoNT-treated monolayer was not significantly different at any time point. Cells without BoNT and NTNHA treatment were used as a control. (d) Concomitant FITC-dextran (70 kDa) flux during the transport of NTNHA and BoNT. In this experiment, FITC-dextran was mixed with an unlabeled NTNHA or BoNT solution (80 nM) and incubated for the indicated time at 37 °C (= 3, means ± SD). Cells without BoNT and NTNHA treatment were used as a control. The fluxes were slightly increased depending on the incubation time. The transported FITC-dextran was below 0.05% of the original solution at 24 h, whereas the transported Cy3-labeled NTNHA and BoNT during the 24-h incubation were 3.0–5.1% of the original solution (data not shown).


In a previous study (Inui et al., 2012), we demonstrated that NTNHA and BoNT share a similar domain organization. BoNT may be divided into three domains, each of which has a distinct function, for example, the N-terminal catalytic domain (light chain; Lc), the central channel-forming domain (N-terminal half of the heavy chain; HcN), and the C-terminal binding domain (C-terminal half of the heavy chain; HcC). The C-terminal HcC binding domain of BoNT binds to nerve and intestinal epithelial cells (Maksymowych & Simpson, 2004; Stenmark et al., 2008). Based on in silico analyses (Inui et al., 2012) and crystal structures (Gu et al., 2012; Sagane et al., 2012), NTNHA and BoNT share very similar three-dimensional structures, and thus, NTNHA can also be divided into three domains: N-terminal nLc, central nHcN, and C-terminal nHcC. However, the functions of these NTNHA domains have not yet been clarified. As shown in Fig. 4(a), the nHcC domain contains both N-terminal jelly-roll and C-terminal β-trefoil domains, similar to the HcC domain of BoNT (Ginalski et al., 2000). Tetanus neurotoxin (TeNT) and serotypes A and B BoNT share the consensus sequence SXWY in the C-terminal β-trefoil domain of the HcC, which appears to be responsible for ganglioside-receptor binding on nerve cells (Rummel et al., 2004). This SXWY motif is not conserved in NTNHA, nor is it conserved in serotypes C and D BoNT (Rummel et al., 2004). These proteins share a conserved Trp residue, which is probably responsible for carbohydrate binding by the ganglioside-binding loop in the β-trefoil domain (Kroken et al., 2011). NTNHA does not possess a Trp residue in the loop region corresponding to the ganglioside-binding loop in the HcC-C domain of BoNT. On the other hand, the serotype B–F NTNHA possesses a single consensus QXW motif in its nHcC-C domain, as shown in Fig. 4(b). Additionally, serotypes A–D, and F have a conserved sequence that is a variant of the known QXW motif, QXY, 50 residues upstream from the QXW motif. On the other hand, no serotypes of BoNT possess these sequences. The QXW motif is typified by the carbohydrate-binding subunit of ricin (Hazes, 1996). Previously, Arndt et al. (2005) explained that NTNHA does not bind carbohydrates because it lacks the ganglioside-binding site that is shared by TeNT and BoNT serotypes A and B. Therefore, the function of NTNHA has been considered to be only protection of BoNT against digestive conditions. However, we found that the serotype D NTNHA molecule binds to IEC-6 cells. NTNHA also retains the β-trefoil domain and the QXW motif, which are both found in a lectin family, and the botulinum HA-33 protein also retains this motif (Hazes, 1996). Very recently, models of the three-dimensional structure of the serotypes A and B L-TC (M-TC/HAs complex) were proposed based on transmission electron microscopy images and the crystal structures of its individual components (Benefield et al., 2013). This model resembles our model for the three-dimensional structure of serotype D L-TC (Hasegawa et al., 2007), and it indicates that the C-terminal region of NTNHA, which contains candidates for the sugar-binding sites, is exposed on the surface of the molecule. Similar to the same site in NTNHA, the sugar-binding sites in HA-33, HA-70, and BoNT are fully accessible in the L-TC model (Benefield et al., 2013). The cell binding and transport through the IEC-6 cell layer of the L-TC (M-TC/HA-70/HA-17/HA-33) depend on the number of the HA-33 molecule in the TC, implying that the HA-33 enhances toxin transport through the intestinal wall (Ito et al., 2011). On the other hand, the NTNHA and HA-70 proteins do not facilitate the efficiency of the toxin transport (Ito et al., 2011). However, these multiple cell-binding sites, which have distinct binding interactions in the TC, may enhance the chances for the toxins to access different types of the intestinal epithelial cells. It is still unknown which specific residues are responsible for cell binding and transport through the intestinal epithelial cell layer. This will be addressed in future studies in which amino acid substitutions are engineered to elucidate the pathogenic mechanisms of botulinum toxin delivery in the gastrointestinal tract. Such studies should also lead to the eventual design of stable oral delivery systems for proteolytically unstable drugs.

Figure 4.

The C-terminal third domain (nHcC) of the NTNHA protein. (a) The nHcC domain from the crystal structure of D-4947 NTNHA (PDB code 3VUO). The nHcC domain consists of an N-terminal jelly-roll and a C-terminal β-trefoil domain. The barrel and cap structures in the β-trefoil domain are indicated. Each β-trefoil repeat is colored with red, green, or blue. (b) Multiple alignment of the amino acid sequences of serotype A–F NTNHA proteins. Numbers to the left of each column indicate the original residue numbering from the N-termini of the proteins. The β-sheets comprising the β-trefoil repeats in a are indicated by red, green, and blue arrows, respectively. Residues highlighted in black, dark gray, and light gray indicate 100%, 80%, and 60% sequence identity among the serotypes. Closed red circles indicate the location of the QXW sequence motif or a variety of the QXW motif.


We thank Mr. Yusuke Nakamura and Shunta Matsuura for his technical assistance. This work was supported by KAKENHI provided by MEXT/JSPS (for T.W., Grant No. 22590405).

Authors’ contribution

S.-I.M. and Y.S. contributed equally to this project.