ABSTRACT
Tetanus neurotoxin (TeNT) and botulinum neurotoxin (BoNT) are clostridial neurotoxins (CNTs) responsible for the paralytic diseases tetanus and botulism, respectively. CNTs are AB toxins with an N-terminal zinc-metalloprotease light chain that is linked by a disulfide bond to a C-terminal heavy chain that includes a translocation domain and a receptor-binding domain (HCR). Current models predict that the HCR defines how CNTs enter and traffic in neurons. Recent studies implicate that domains outside the HCR contribute to CNT trafficking in neurons. In the current study, a recombinant, full-length TeNT derivative, TeNT(RY), was engineered to analyze TeNT cell entry. TeNT(RY) was atoxic in a mouse challenge model. Using Neuro-2a cells, a mouse neuroblastoma cell line, TeNT HCR (HCR/T) and TeNT(RY) were found to bind gangliosides with similar affinities and specificities, consistent with the HCR domain containing receptor binding function. Temporal studies showed that HCR/T and TeNT(RY) entered Neuro-2a cells slower than the HCR of BoNT/A (HCR/A), transferrin, and cholera toxin B. Intracellular localization showed that neither HCR/T nor TeNT(RY) localized with HCR/A or synaptic vesicle protein 2, the protein receptor for HCR/A. HCR/T and TeNT(RY) exhibited only partial intracellular colocalization, indicating that regions outside the HCR contribute to the intracellular TeNT trafficking. TeNT may require this complex functional entry organization to target neurons in the central nervous system.
INTRODUCTION
Tetanus neurotoxin (TeNT) and botulinum neurotoxin (BoNT) are clostridial neurotoxins (CNTs) that are the most toxic proteins for humans (1). TeNT and BoNT share ∼35% identity and ∼65% similarity and overall structure-function properties (2). BoNT intoxication results in flaccid paralysis through the inhibition of acetylcholine release by motor neurons, while TeNT intoxication yields a spastic paralysis due to inhibition of glycine release by inhibitory neurons (3). TeNT and BoNT are expressed as ∼150-kDa single-chain proteins that are cleaved to form dichain proteins linked by a disulfide bond (2). The N-terminal 50-kDa light chain (LC) is a zinc-metalloprotease that cleaves neuron-specific soluble NSF attachment protein (SNAP) receptor (SNARE) proteins (4). TeNT and BoNT serotype B cleave the same residue within vesicle-associated membrane protein 2 (VAMP2), a SNARE protein of synaptic vesicles (SVs). The C-terminal 100-kDa heavy chain (HC) contains two structurally distinct domains with separate functions. The translocation domain (HCT) facilitates LC translocation from the SV lumen into the cell cytosol, and the receptor binding domain (HCR) binds dual host receptors.
BoNT/A binds a ganglioside and synaptic vesicle protein 2 (SV2) and enters neurons upon SV recycling from the plasma membrane (5). Upon SV acidification within the periphery of motor neurons, the HCT undergoes a pH-dependent conformational change and inserts into the SV membrane, forming a channel that allows the LC to escape into the cytosol. Within the periphery of the motor neuron, the LC cleaves SNARE proteins, resulting in loss of stimulatory signaling between neurons and muscles, yielding flaccid paralysis. The LC of BoNT/A localizes to the plasma membrane to target synaptosomal-associated protein 25 (SNAP25) for cleavage within neurons (6).
TeNT binds two gangliosides as functional receptors (7). TeNT can bind a glycophosphatidylinositol (GPI)-anchored protein (8) or SV2 (9), but the significance of these interactions has not been defined (10) or reproduced (11), respectively. TeNT enters motor neurons upon endocytosis (12) and traffics through motor neurons associated with Rab7-enriched endosomes that are of neutral pH (13, 14). Retrograde trafficking proceeds from the axon to the soma, where TeNT transcytoses from the motor neuron into an inhibitory neuron of the central nervous system (CNS). Upon vesicle acidification, the LC is translocated into the cytosol and cleaves VAMP2. The block in signaling between the inhibitory neurons and motor neurons leads to the spastic paralysis characteristic of tetanus. The molecular mechanism responsible for the unique entry of BoNT and TeNT is not clearly understood.
The modular structural domains of the CNTs have permitted the study of individual domains to assess protein structure-function in vitro; the LC is a functional SNARE protease (15), the HCT inserts into lipid bilayers and forms a channel (16), and the HCR binds host receptors (17). The tetanus HCR (HCR/T) can retrograde traffic in both cultured spinal cord motor neurons and in vivo in rats (14, 17). Thus, the unique pathologies associated with tetanus and botulism have been attributed to receptor binding and intracellular trafficking of the HCR domains of the respective toxins (18). Contrary to the established model, recent studies reported that, in mice, the HCR domain of TeNT is not sufficient to cause retrograde trafficking of a BoNT-TeNT fusion protein (19). While the cellular basis for the trafficking patterns of this fusion protein needs further resolution, these observations question whether or not the HCR domain is necessary and sufficient to traffic the CNTs to their respective physiological substrates.
Since domain exchanges between CNTs can yield unexpected phenotypes (20), comparison of HCR trafficking relative to full-length CNT is an important question. We chose to characterize the entry of the holo-TeNT, since phenotypes of the HCs are complex, with HC/A interacting with membranes independent of a pH gradient and pH dependency ascribed to the HCR domain within holo-CNTs (21). The ability to localize full-length CNTs has been limited by the ability to produce recombinant full-length CNTs with epitope tags to detect intracellular localization. The current study characterizes a recombinant full-length, atoxic TeNT which contains Arg372Ala and Tyr375Phe mutations within the catalytic site [TeNT(RY)] and possess epitopes to allow detection of the LC and HC domains. TeNT(RY) bound gangliosides with the same specificity as HCR/T, consistent with the localization of receptor binding function within the HCR domain. Intracellular localization measurements showed that TeNT(RY) and HCR/T trafficked at similar rates but only partially colocalized within intracellular vesicles. This indicates that regions outside the HCR domain contribute to the intracellular trafficking of TeNT.
MATERIALS AND METHODS
Production and purification of recombinant proteins.Hemagglutinin (HA)-tagged and FLAG-tagged heavy chain receptor (HCR) binding domains of tetanus (HCR/T; residues 865 to 1315) and botulinum neurotoxin serotype A2 (Clostridium botulinum strain A2 Kyoto F, HCR/A, residues 870 to 1296) were produced and purified as previously described (22, 23). Briefly, DNA encoding three HA repeats (YPYDVPDYA) was subcloned downstream of the His6 epitope and upstream of the HCR sequence in pET28a (Novagen), and this construct was then confirmed by DNA sequencing. Escherichia coli BL21(DE3) was transformed with the expression plasmid. Transformants were plated overnight on LB agar containing 50 μg/ml of kanamycin (Km) and stored at −80°C in 12% (vol/vol) glycerol. For protein expression, E. coli was grown overnight on LB-Km plates that were inoculated into 400 ml of LB-Km. Cells were grown for 2 h at 30°C at 225 rpm (optical density at 600 nm [OD600] of ∼0.6), and then 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added, followed by overnight culture at 16°C (225 rpm). Cells were harvested and lysed with a French press. The lysates was centrifuged at 38,000 × g for 20 min to remove the insoluble fraction, and the supernatant was passed through a 0.45-μm-pore-size cellulose acetate filter. HCRs were purified by sequential nickel affinity chromatography, Sephadex S200HR gel filtration, and DEAE-Sephadex chromatography. Purified HCR was dialyzed into 10 mM Tris (pH 8.0), 20 to 200 mM NaCl, and 40% glycerol and stored at −20°C. HCR concentration was determined using bovine serum albumin (BSA) standards and SDS-PAGE. To produce TeNT(RY) (1,315 amino acids; accession no. P04958), E. coli codon-optimized tetanus light chain (LC) was mutated to R372A and Y375F by QuikChange (Agilent). Mutation of the desired residues was confirmed by DNA sequencing. LC(RY) was subcloned into E. coli-optimized HC in pET28a to yield pTeNT(RY), which was transformed into E. coli BL21(DE3). TeNT(RY) was purified as described above for HCRs with the exception of the buffer being 20 mM sodium phosphate (pH 7.9) instead of 10 mM Tris (pH 8.0).
Trypsin cleavage of TeNT(RY).TeNT(RY) was diluted in 20 mM phosphate and 20 mM NaCl (pH 7.9) and incubated with a 1:1,000 ratio (wt/wt) of trypsin to TeNT(RY) at room temperature for the indicated time points. To quench the reaction, a 3-fold molar excess of soybean trypsin inhibitor was added. Subsequently, SDS loading buffer was added, with or without β-mercaptoethanol, to reduce the disulfide bond linking the LC and HC. Samples were analyzed by SDS-PAGE, followed by staining with Coomassie blue or transfer to a polyvinylidene difluoride (PVDF) membrane. Blotting for the FLAG epitope was performed using mouse anti-FLAG–horseradish peroxidase (HRP) IgG (1:20,000; Sigma).
Cell culture of Neuro-2a cells.Acid-etched no. 1 coverslips (12 mm) were placed into 24-well culture plates and coated with rat collagen-I (Life Technologies) overnight. Coverslips were washed twice with Dulbecco's phosphate-buffered saline (DPBS), and Neuro-2a cells (ATCC CCL-131) were plated and cultured in minimal essential medium (MEM) (Life Technologies) supplemented with 10% fetal bovine serum (FBS) (Life Technologies), sodium bicarbonate, nonessential amino acids, sodium pyruvate, and penicillin-streptomycin. Experiments were performed on cells grown to ∼80% confluence. Ganglioside-enriched Neuro-2a cells were prepared by the following protocol: GT1b, GD1b, GD1a, or GM1a (20 μg/ml final; Matreya) were sonicated in low-serum (0.5% FBS) culture medium for 20 min at room temperature. Neuro-2a cells were cultured with ganglioside-supplemented culture medium (0.5 ml) for 4 h. For binding studies, cells were washed twice in ice-cold DPBS, placed on ice for 5 min, and then incubated with 2.5 to 40 nM HA-HCR/T or TeNT(RY) in ice-cold low-K buffer (15 mM HEPES, 145 mM NaCl, 5.6 mM KCl, 2.2 mM CaCl2, 0.5 mM MgCl2, pH 7.4) for 30 min. Cells were washed twice in ice-cold DPBS and processed for immunofluorescence. For entry studies, cells were loaded with GT1b, washed twice in DPBS (37°C), and incubated with 40 nM HA-HCR/T, FLAG-HCR/T, HA-HCR/A, or TeNT(RY) in low-K buffer for the indicated times at 37°C. For entry of cholera toxin B (CTxB) and transferrin (Tfn), cells were loaded with GM1a, washed twice in DPBS (37°C), and incubated with 40 nM CTxB-Alexa 647 and 60 nM transferrin-Alexa 488 (Molecular Probes) in low-K buffer for the indicated times at 37°C. Cells were washed twice in DPBS (37°C) and processed for immunofluorescence.
Immunofluorescence methods.Cells were fixed in 4% (wt/vol) paraformaldehyde in DPBS for 15 min at room temperature, washed in DPBS (3×), permeabilized with 0.1% (vol/vol) Triton X-100 in 4% (vol/vol) formaldehyde in DPBS for 15 min at room temperature, washed in DPBS (3×), incubated with 150 mM glycine in DPBS for 15 min at room temperature, and washed in DPBS (3×). Cells were incubated in a blocking solution (10% FBS, 25 mg/ml gelatin from cold-water fish skin, 0.1% Triton X-100, and 0.05% Tween 20) for 1 h at room temperature and then incubated with primary antibodies in antibody solution (5% FBS, 10 mg/ml gelatin from cold-water fish skin, 0.1% Triton X-100, and 0.05% Tween 20) overnight at 4°C. Recombinant proteins were probed with rat anti-HA IgG (1:2,000; Roche) or mouse anti-FLAG IgG (1:10,000; Sigma). SVs were probed with rabbit anti-SV2c (1:2,000; SySy). Cells were washed for 5 min in DPBS (3×) and incubated with Alexa-labeled secondary antibodies (1:1,000 dilution, Alexa 488, 568, and 647; Molecular Probes) in antibody solution for 1 h at room temperature. Cells were washed four times for 5 min in DPBS. When appropriate, Hoechst dye (1:10,000) was added during the second wash. Cells were fixed with 4% paraformaldehyde in DPBS for 10 min at room temperature and washed in DPBS (3 times). Coverslips were mounted onto ProLong (Life Technologies) for storage and epifluorescence imaging. Epifluorescence images were captured using a Nikon TE2000 total internal reflection fluorescence microscope equipped with a Plan Apo VC 60×/1.40-numerical-aperture (NA) oil objective and an ET-Sedat Quad filter set (89000; Chroma Technology) using a Photometrics CoolSnap HQ2 camera and Nikon NIS Elements AR software. Bleed-through of fluorescent dyes was taken into consideration during acquisition, with the exposure time of each channel kept within 2-fold for each experiment (with the exception of Hoechst). In the one experiment where the exposure times were outside this parameter, control images were analyzed postacquisition, confirming undetectable crossover.
Deconvolution of epifluorescence images.Fields that underwent deconvolution were acquired using 0.4-μm z-steps. Individual cells were cropped from the field, and the channels were deconvolved with Nikon Elements AR software, using three-dimensional (3D) deconvolution, the 3D blind method, wide-field modality, and 20 iterations. The Hoechst channel was excluded from the deconvolution.
Mouse challenge model.Female ICR mice (18 to 22 g) were injected intraperitoneally with 5 μg/mouse of either the single-chain or dichain (trypsin-treated) forms of the recombinant TeNT(RY) (equivalent to ∼125,000 50% lethal doses [LD50] of wild-type TeNT) (1) and then monitored for 48 h, when survival was scored. All mice survived challenge with 5 μg of the single-chain or dichain forms of recombinant TeNT(RY) and showed no symptoms of tetanus intoxication. All animal experiments were approved by and conducted according to guidelines from the University of Wisconsin Animal Care and Use Committee.
Data analysis.Image analyses of intensity were performed using ImageJ 1.46r. The colocalization of deconvolved cells was analyzed using Pearson's colocalization coefficients in Nikon Elements AR. Graphs were created using GraphPad Prism 5, and figures were compiled using Adobe Photoshop CS3. Statistics were performed by analysis of variance (ANOVA) (see Fig. 4 to 7), using GraphPad Prism 5, or by Student's t test (see Fig. S1 in the supplemental material), using Excel.
RESULTS
Properties of TeNT(RY).Recombinant, full-length tetanus toxin [TeNT(RY)] (1,315 amino acids) was engineered with two point mutations within the light chain (R372A and Y375F). The N terminus contained His6 and 3× FLAG epitopes, while the C terminus contained HA and Strep epitopes (Fig. 1A). E. coli produced TeNT(RY) as a soluble ∼150-kDa protein as determined by SDS-PAGE (Fig. 1B). Production of TeNT(RY) in E. coli was optimized, with purified yields ranging from 2 to 8 mg/liter of batch culture. TeNT is an AB toxin which is cleaved by trypsin into an N-terminal light chain (LC) and a C-terminal heavy chain (HC) that remain linked by a disulfide bond (24). Trypsin cleavage followed by SDS-PAGE with or without β-mercaptoethanol (β-mer) showed that TeNT(RY) was cleaved to a dichain (Fig. 1B) that ran as a single ∼150-kDa band (arrow) under nonreducing conditions, indicating retention of the disulfide (Fig. 1B, top). Under reducing conditions, the LC and HC were resolved (HC, ∼100 kDa, closed arrowhead; LC, ∼50 kDa, open arrowhead [Fig. 1B, middle]). Cleavage between the LC and HC was observed at 10 min (the first time point) and complete by ∼2 h. Extended incubation resulted in additional nicking of the LC. Western blot analysis against the N-terminal FLAG epitope on the LC confirmed that the cleavage was between the LC and HC (Fig. 1B, bottom). The trypsin digestion pattern was previously reported for native tetanus toxin, with the initial trypsin cleavage at Arg455 and the second trypsin cleavage at Arg449 (24). Western blotting for the FLAG epitope additionally indicated that a fraction of the TeNT(RY) was not cleaved at the LC-HC junction, as indicated by the presence of FLAG-positive signal at ∼150 kDa, even after overnight incubation with trypsin. This fraction of TeNT(RY) may not have folded into a trypsin-sensitive conformation. Overall, these data show that TeNT(RY) retains the fundamental structure necessary for the conversion of the single-chain to the dichain protein by trypsin.
Recombinant, atoxic full-length TeNT(RY). (A) Schematics of recombinant TeNT(RY) and HCR/T. Native TeNT is composed of LC, HCT, and HCR domains; the recombinant protein contains mutations within the active site at R372 and Y375. In addition, N-terminal His6 and FLAG epitopes and C-terminal HA and Strep epitopes are present. Recombinant HCR/T contains N-terminal His6 and FLAG or HA epitopes. (B) TeNT(RY) was incubated alone (−) or with trypsin (+) at room temperature for the indicated time points. Samples were subjected to SDS-PAGE, alone (−) or with β-mer (+), and stained with Coomassie blue (top and middle) or by immunoblotting (bottom), using anti-FLAG–HRP-conjugated antibody (1:20,000).
In a mouse challenge model, intravenous delivery of single-chain or dichain (trypsin-nicked) TeNT(RY) (5 μg/mouse; equivalent to ∼125,000 LD50 doses of wild-type TeNT) did not show symptoms of tetanus intoxication over a 2-day period. Experiments in rat primary cortical neurons showed that TeNT(RY) did not cleave VAMP2 when administered at 500-fold-higher concentrations than the amount of wild-type TeNT needed to cleave VAMP2 (data not shown). TeNT(RY) was next used to probe the entry properties of full-length tetanus toxin into a neuronal cell line.
Binding of TeNT(RY) to Neuro-2a cells.TeNT utilizes two gangliosides as functional plasma membrane receptors (7, 23, 25, 26). Dual ganglioside binding is necessary and sufficient for HCR/T entry into neurons and nonneuronal cells, such as HeLa cells (7). TeNT prefers to bind b-series gangliosides (GD1b, GT1b, and GQ1b) (27), and in solid-phase assays, HCR/T binds most avidly to GT1b, GD1b, and GD1a (23). Knockout mice that do not produce b-series gangliosides or complex gangliosides are resistant to TeNT (28, 29). To test the ganglioside binding potential of TeNT(RY) relative to HCR/T, Neuro-2a cells (ATCC CCL-131) were loaded with exogenous gangliosides. Neuro-2a cells are derived from a mouse neuroblastoma and contain minimal amounts of endogenous complex gangliosides. After incorporation of exogenous gangliosides (GT1b, GD1b, GD1a, or GM1a), the binding of HCR/T and TeNT(RY) was determined. As expected from previous experiments (7), at 4°C HCR/T bound, in a dose-dependent manner, to GT1b- and GD1b-enriched cells (Fig. 2). HCR/T bound minimally to GD1a and GM1a, except at the highest concentration (40 nM) (Fig. 2A). The amount of TeNT(RY) bound was highest with GT1b and GD1b, the same b-series gangliosides which HCR/T bound. Of note, TeNT(RY) more efficiently bound GD1a- and GM1a-enriched cells compared to HCR/T, as well as to cells without ganglioside added (Fig. 2B). Overall, TeNT(RY) and HCR/T possessed similar ganglioside binding specificity, with TeNT(RY) possessing what appeared to be a higher affinity for untreated Neuro-2a cells at higher TeNT(RY) concentrations. Ganglioside binding was dependent upon the presence of functional ganglioside binding pockets within the HCR domain, since 1 μM TeNT(RY) containing mutations R1226L and W1289A, and therefore lacking ganglioside binding (23), did not show detectable binding to cells treated with exogenous GT1b (data not shown). Thus, TeNT(RY), in addition to folding properly, exists in a conformation in which ganglioside binding pockets present in the HCR domain are functional.
Ganglioside binding profiles of HCR/T and TeNT(RY). Neuro-2a cells were loaded with exogenous ganglioside GT1b, GD1b, GD1a, or GM1a. After loading, cells were washed and cooled to 4°C in DPBS. Cells were then incubated with 2.5, 10, or 40 nM HCR/T (A) or TeNT(RY) (B) in cold buffer for 30 min. Cells were fixed and processed for immunofluorescence. The epifluorescence images shown are of GT1b-treated Neuro-2a cells with 40 nM HCR/T and TeNT(RY) (HA), and nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Intensity was quantified by averaging the intensity of HA staining in a field, subtracting background fluorescence, and normalizing to the intensity of intracellular SV2c (HA/SV2c). Intensity was then normalized to set the signal from 40 nM HCR/T bound to GT1b to 1 in both sets of data.
With the addition of exogenous GT1b, Neuro-2a cells are competent for entry of both HCR/T and the HCR of BoNT/C (30). Therefore, the entry of TeNT(RY) into these cells was measured. For the purposes of quantifying protein entry, a cell containing internalized protein was scored positive upon internalization of protein in two or more vesicles. For each protein, a time course was optimized to determine the time when 50% of the cells contained internalized protein. Controls, transferrin (Tfn) and cholera toxin B (CTxB), reached 50% internalization by <1 min and ∼1.5 min, respectively (Fig. 3A). In comparison, TeNT(RY) was internalized by 50% of the cells at ∼7 min (Fig. 3A). HCR/T was internalized at a similar rate as TeNT(RY), with 50% internalization at ∼5 min. The HCR domain of BoNT/A (HCR/A), which utilizes both a ganglioside and a defined protein, synaptic vesicle protein 2 (SV2), as receptors, exhibited 50% internalization of cells at ∼1.5 min (Fig. 3B). Entry of HCR/A was faster than entry of HCR/T or TeNT(RY), suggesting that different mechanisms are utilized by BoNT/A and TeNT to enter Neuro-2a cells.
TeNT(RY) and HCR/T are endocytosed at a similar rate into Neuro-2a cells. (A) Neuro-2a cells were treated with exogenous GT1b [for TeNT(RY) and HCR/T uptake] or GM1a (for CTxB uptake). Cells were treated with 60 nM transferrin (Tfn) or 40 nM TeNT(RY), HCR/T, or CTxB for 2, 5, 10, or 20 min at 37°C. In a separate experiment, cells were incubated with Tfn or CTxB for 0.5, 1, or 2 min at 37°C. Cells were processed for immunofluorescence analysis. TeNT(RY) and HCR/T were detected using anti-HA antibody, and CTxB and Tfn were obtained as Alexa-labeled conjugates. Individual cells were analyzed, and an internalized cell was defined as containing intracellular signal in two or more vesicles. At least three independent replicates were performed with a minimum of 200 cells analyzed per time point for each replicate. The time at which 50% of the cells contain intracellular signal (t1/2) was determined for each protein. (B) Neuro-2a cells were treated with exogenous GT1b and treated with 40 nM HCR/A and HCR/T for 0.5, 1, 2, and 5 min at 37°C. Cells were processed and analyzed as described in panel A, except the minimum number of cells analyzed was 180 per time point for each replicate.
Internalization of TeNT(RY) into Neuro-2a cells.Pearson's coefficients were determined to measure protein-protein colocalization. Experimental colocalization ranges were established for this system. As a measure of maximum colocalization, the colocalizations of the HA (LC) and FLAG (HC) epitopes on TeNT(RY) bound to Neuro-2a cells at 4°C were determined to have a Pearson's coefficient of 0.80 (Fig. 4, 5, 6, and 7; see also Fig. S1 in the supplemental material). The minimum colocalization, determined by measuring colocalization of CTxB bound to Neuro-2a cells at 4°C with SV2c, an intracellular protein enriched in synaptic vesicles, was determined to have a Pearson's coefficient of 0.03 (Fig. 4 to 7; see also Fig. S1). After a 20-min incubation, the Pearson's coefficient of LC and HC on TeNT(RY) was 0.77, which indicated that the LC and HC domains had not dissociated during this time course. In contrast, colocalization of TeNT(RY) with SV2c was low, at ∼0.2.
LC and HC of TeNT(RY) are endocytosed into the same pools of vesicles in Neuro-2a cells. Exogenous GT1b was incorporated into Neuro-2a cells, which were subsequently treated with 40 nM TeNT(RY) for 20 min at 37°C. Cells were probed with antibodies to HA and FLAG on TeNT(RY) and cellular SV2c. Representative deconvolved epifluorescence images are shown (top). The merge image is HA (red), FLAG (green), and SV2 (blue). The Pearson's colocalization coefficients are graphed (bottom), with the values for the control colocalization from Fig. S1 in the supplemental material shown in hatched bars. Statistics were performed by ANOVA using GraphPad Prism 5. Scale bar, 10 μm.
TeNT(RY) and HCR/T are endocytosed into similar pools of vesicles in Neuro-2a cells. Exogenous GT1b was incorporated into Neuro-2a cells, which were subsequently treated with 40 nM TeNT(RY) and HCR/T-FLAG for 20 min at 37°C. The TeNT(RY) used in this experiment was engineered without a FLAG tag, allowing for differential detection of TeNT(RY) with an HA antibody and of HCR/T with a FLAG antibody. Representative deconvolved epifluorescence images are shown (top). The merge image is TeNT(RY) (red), HCR/T (green), and SV2 (blue). Pearson's colocalization coefficients are graphed (bottom), with the values for the control colocalization from Fig. S1 in the supplemental material shown in hatched bars. Statistics were performed by ANOVA using GraphPad Prism 5. Scale bar, 10 μm.
Differentially tagged HCR/T proteins are endocytosed into the same vesicles in Neuro-2a cells. Exogenous GT1b was incorporated into Neuro-2a cells, which were treated with 40 nM HCR/T-HA and HCR/T-FLAG for 20 min at 37°C. Representative deconvovled epifluorescence images are shown (top). The merged image is HCR/T-HA (red), HCR/T-FLAG (green), and SV2c (blue). The Pearson's colocalization coefficients are graphed (bottom), with the values for the control colocalization from Fig. S1 in the supplemental material shown in hatched bars. Statistics were performed by ANOVA using GraphPad Prism 5. Scale bar, 10 μm.
HCR/A and HCR/T are endocytosed into different pools of vesicles in Neuro-2a cells. Neuro-2a cells were treated with exogenous GT1b, and cells were treated with 40 nM HA-HCR/A and FLAG-HCR/T for 20 min at 37°C. Representative deconvolved epifluorescence images are shown (top). The merge image is HCR/A (red), HCR/T (green), and SV2c (blue). Pearson's colocalization coefficients are graphed (bottom), with the values for the control colocalization from Fig. S1 in the supplemental material shown in hatched bars. HCR/A was localized to SV2c-positive vesicles and not with HCR/T. Statistics were performed by ANOVA using GraphPad Prism 5. Scale bar, 10 μm.
Partial colocalization of intracellular TeNT(RY) and HCR/T in Neuro-2a cells.TeNT(RY) was engineered with an HA epitope on the C terminus of the HC domain and incubated with FLAG epitope-tagged HCR/T on ganglioside-enriched Neuro-2A cells. At 20 min, the Pearson's coefficient for HCR/T and TeNT(RY) was 0.50, indicating a partial colocalization of the two proteins (Fig. 5). The Pearson's coefficient of differentially epitope labeled HCR/T was 0.69 (Fig. 6), statistically different from the colocalization between HCR/T and TeNT(RY). The different coefficients indicate that the intracellular localization of TeNT(RY) and HCR/T represents partial colocalization. Inspection of the merged images in Fig. 6 showed intracellular vesicles containing both TeNT(RY) and HCR/T and also vesicles which contained either TeNT(RY) or HCR/T. In both experiments, the colocalization of TeNT(RY) and HCR/T with SV2 was low, ∼0.2. This indicates that regions outside the HCR contribute to TeNT intracellular trafficking.
HCR/T and HCR/A entry into Neuro-2a cells.HCR/A and HCR/T had different rates of internalization (Fig. 3). To determine if HCR/A and HCR/T trafficked to the same vesicles, Neuro-2a cells were incubated with HA-HCR/A and FLAG-HCR/T. Figure 7 shows that HCR/A trafficked to an SV2-positive compartment in a compacted perinuclear region. The Pearson's coefficient for HCR/A and SV2 was 0.82, a coefficient expected of proteins that were completely experimentally colocalized. In contrast, the Pearson's coefficient for the colocalization between HCR/A and HCR/T was low, at ∼0.20. This Pearson's value was lower than anticipated based on a visual inspection of the colocalization in Fig. 7, which had overlap of HCR/T with HCR/A and SV2c in the perinuclear region. However, inspection of the individual images in the z-stack of the merged image in Fig. S2 in the supplemental material showed that the majority of HCR/T localized to vesicles that were peripheral to the compact SV2 vesicles. Together, the data showed that the majority of HCR/T traffic was independent of HCR/A and SV2, with only a fraction of HCR/T within SV2-positive vesicles. HCR/T-associated vesicles did not contain the endocytic markers clathrin, Rab5, EEA1, and LAMP1, while the HCR/A-associated vesicles possessed the synaptic vesicles marker proteins SV2c, synaptosomal-associated protein 25 (SNAP25), vesicle-associated membrane protein 2 (VAMP2), and acetylcholine transferase (AChT) (data not shown). Colocalization of HCR/A or HCR/T with SV2c at earlier time points showed Pearson's coefficients similar to those of the 20-min incubations (data not shown). This indicates that the HCR/A-SV2 association is an early trafficking event.
DISCUSSION
This study describes the properties of full-length tetanus neurotoxin. The toxicity of TeNT(RY) was tested in mice, in which both trypsin-treated and untreated TeNT(RY) were atoxic. Thus, TeNT(RY) appears to be a useful reagent to study neurotoxin entry, since tetanus toxin is not a select agent, and in the United States, humans are immunized with tetanus toxoid during the DT and DTaP vaccinations (31). HCR/T and TeNT(RY) displayed similar properties when assayed for binding gangliosides (Fig. 2) and for the rate of internalization (Fig. 3) but showed only partial colocalization within Neuro-2a cells (Fig. 5). This is consistent with the HCR domain defining the interaction with host cell receptor, while regions outside the HCR contribute to the fate of intracellular trafficking. The observed complexity of TeNT trafficking relative to BoNT/A, which colocalized within SV2-enriched vesicles in this study, is consistent with the necessity of TeNT to traffic through multiple vesicles to reach the CNS, while BoNT is delivered from the SV in the periphery of the neuron. The complex trafficking has been observed in cortical neurons, where HCR/T trafficked through both activity-dependent and -independent trafficking pathways, whereas HCR/A trafficked through solely activity-dependent pathways (11). One difference with the fraction of HCR/T that colocalized with HCR/A in Neuro-2a cells relative to colocalization observed in primary cortical neurons is that in Neuro-2a cells, the colocalization appears to occur from within the intracellular vesicle population, while in cortical neurons, the colocalization occurs at the plasma membrane during SV endocytosis. The significance of this differential trafficking is under investigation.
The receptor binding activity of TeNT has been defined within the HCR (32). The ganglioside preference of TeNT(RY) and HCR/T were the same (Fig. 2), with the highest binding to the b-series gangliosides GT1b and GD1b, suggesting that while there is a preferred ganglioside for each of the ganglioside binding pockets, the overall affinity for b-series gangliosides is greater than that of a-series gangliosides. This is supported by crystallographic and biochemical characterization, which showed that the two ganglioside binding sites are within the HCR (7, 23, 25, 26). TeNT(RY) binding was higher than HCR/T to untreated Neuro-2a cells, which may reflect a nonganglioside interaction located outside the HCR. The translocation domain of CNTs is sufficient for insertion into a lipid bilayer (16), indicating that the HCT is capable of interacting with lipids independent of the HCR domain. We hypothesize that the increased association of TeNT(RY) to untreated Neuro-2a cells is due to interactions between the cell membrane and the holotoxin, possibly the translocation domain. Experiments are under way to determine if a membrane-interacting site can be ascribed to regions outside the HCR domain. Recent work with chimeric botulinum and tetanus neurotoxins adds to the complexity of the trafficking of CNTs. Wang et al. found that efficient retrograde trafficking of TeNT was dependent on the presence of the entire toxin (19) and that replacing any domain of TeNT with the homologous domain of BoNT/A resulted in a toxin with induced flaccid paralysis in mice. This also suggests that the HCR domain is not the sole determinant in the entry and trafficking of CNTs. We found that TeNT(RY) and HCR/T partially colocalized within Neuro-2a cells, statistically less than HCR/T colocalized with HCR/T, supporting a role for domains outside the HCR in the trafficking of TeNT into Neuro-2a cells. Continued assessment of HCR/T and TeNT trafficking in Neuron 2a cells may resolve the components within TeNT that are responsible for the unique trafficking pattern.
In cultured cells, Tfn is endocytosed via a well-defined pathway into early endosomes and recycling endosomes prior to recycling back to the plasma membrane. The entry of Tfn observed here is in line with the rates of entry in other cell types (33). CTxB utilizes GM1a as a receptor and enters cells through many mechanisms, all of which converge into early endosomes, prior to retrograde trafficking to the trans-Golgi membrane and subsequently to the endoplasmic reticulum (ER). The faster entry of CTxB than HCR/T and TeNT(RY) was unexpected, given the use of gangliosides as receptors for each protein, but may be due to the active entry of CTxB via clustering of GM1a molecules (34), compared to passive entry of HCR/T and TeNT(RY). Alternatively, this may be due to slower entry of HCR/T and TeNT(RY) due to the utilization of an accessory factor, possibly a host protein, in addition to gangliosides (35).
Since HA-tagged HCR/T and FLAG-tagged HCR/T possessed similar rates of entry that were different from that of HA-tagged HCR/A, the epitope could influence the absolute rate of entry, but the relative differences in the rates of entry of HCR/A and HCR/T were not due to the epitope. The epitope tags on HCR/T and HCR/A were located on the N terminus, spatially distant from the known receptor binding regions. Two epitope tags on TeNT(RY) were present on the C terminus and could have interfered with binding and internalization. However, as the intensities of bound HCR/T and TeNT(RY) were not statistically significantly different (Fig. 2), the C-terminal epitopes on TeNT(RY) did not appear to affect binding. In CNTs, the N terminus of the HCR is a jelly roll domain, and the C terminus is a beta-trefoil domain containing the known ganglioside binding sites. The extreme C terminus of the HCR folds back toward the N terminus, suggesting that a C-terminal tag would be spatially distant from the known receptor binding sites.
While Neuro-2a cells appear to be useful to measure the entry of tetanus toxin, other cell lines are also capable of internalizing tetanus toxin. Adrenal chromaffin cells, after the addition of exogenous ganglioside, are intoxicated by TeNT, which blocks evoked release of catecholamine (36). Differentiated PC12 cells, a rat pheochromocytoma cell line, internalize HCR/T and TeNT (7, 37) mediated by the C-terminal subdomain of the HCR (38). The intoxication of PC12 cells by TeNT is dependent on the presence of cholesterol (10), but a further mechanism for TeNT entry has not been defined. The observation that HCR/T is functional for entry into ganglioside-loaded HeLa cells confirms that TeNT is capable of being internalized via an SV-independent mechanism (7), likely in a manner similar to the entry of TeNT into motor neurons (12). While our earlier studies (11) observed that HCR/T entry into cortical neurons was primarily membrane depolarization independent (independent of SVs), intracellular localization of HCR/T was difficult to resolve due to the limited ability to resolve intracellular trafficking patterns within overlapping axons in culture. The advantage of the Neuro-2a cells used in this study was the ability to resolve the intracellular movement of BoNT and TeNT derivatives from the plasma membrane to intracellular vesicles. Fischer and Montal (39) also utilized Neuro-2a cells for patch clamp experiments. Neuro-2a cells are derived from a neuroblastoma, making these cells more similar to the sympathetic nervous system instead of the central nervous system. Neuro-2a cells release catecholamines when stimulated (as opposed to acetylcholine from motor neurons or glycine from inhibitory neurons) and have been shown to incorporate gangliosides to induce surface activity (40).
The two-receptor model proposes that BoNTs bind dual receptors: a protein and ganglioside (41). Protein receptors for the majority of botulinum neurotoxin serotypes have been defined, but the identity of a protein receptor for tetanus has remained elusive (18). In addition, TeNT entry is complicated by the requirement for two entry mechanisms, one into motor neurons and subsequently into inhibitory interneurons (42). TeNT entry into motor neurons has been characterized using the HCR domain. HCR/T does not colocalize with VAMP2 and is able to enter BoNT/A- and BoNT/D-treated neurons (12). The retrograde movement of HCR/T is dependent on Rab5 and Rab7 and occurs in Rab7 vesicles which are also positive for low-affinity neurotrophin receptor, TrkB, and brain-derived neutrotrophic factor (BDNF) (14). These vesicles are of neutral pH and vATPase excluded (13). Little is known concerning the mechanism by which TeNT is transcytosed, but TeNT intoxication of inhibitory interneurons requires entry into a vesicle, which will undergo acidification, making endosomes an attractive mechanism of entry. The complexity of TeNT trafficking may be required to enter multiple neuronal cell types; this is facilitated by functions within multiple domains of the toxin.
ACKNOWLEDGMENTS
This study was supported by funds from NIH R01 AI-031062 and NIH U54-AI-057153. J.T.B. and E.A.J. are members of the Great Lakes Regional Center of Excellence (GLRCE) and acknowledge partial support by the GLRCE for these studies.
FOOTNOTES
- Received 2 December 2013.
- Accepted 2 December 2013.
- Accepted manuscript posted online 9 December 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.01539-13.
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.