Botulinum neurotoxin A activity is dependent upon the presence of specific gangliosides in neuroblastoma cells expressing synaptotagmin I

Botulinum neurotoxin A (BoNT/A) is the deadliest of all known biological substances. Although its toxicity makes BoNT/A a biological warfare threat, its biologic activity makes it an increasingly useful therapeutic agent for the treatment of muscular disorders. However, almost 200 years after its discovery, the neuronal cell components required for the activity of this deadly toxin have not been unequivocally identified. In this work, neuroblastoma cells expressing synaptotagmin I, a protein shown to be bound by BoNT/A, were used to determine whether specific gangliosides were necessary for BoNT/A activity as measured by synaptosomal-associated protein of 25 kDa (SNAP-25) cleavage. Ganglioside GT1b was found to support BoNT/A activity significantly more effectively than GD1a, which was far more effective than GM1 when added to ganglioside-deficient murine cholinergic Neuro 2a or to human adrenergic SK-N-SH neuroblastoma cells. Whereas both cell lines expressed synaptotagmin I, SNAP-25 cleavage was not observed in the absence of complex gangliosides. These results indicate that 1) gangliosides are required for BoNT/A activity, 2) synaptotagmin I in the absence of gangliosides does not support BoNT/A activity, and 3) Neuro 2a cells are an efficient model system for studying the biological activity of BoNT/A.


INTRODUCTION
Botulinum neurotoxin A (BoNT/A), one of seven serotypes of BoNT, is the deadliest of all known biological substances, being ∼1 million times more poisonous than cobra toxin (1). While its toxicity makes BoNT/A a biological warfare threat, its biologic activity makes it an increasingly useful therapeutic agent (2,3). Produced as a single polypeptide chain by the bacterium Clostridium botulinum, BoNT/A (150kD) is subsequently cleaved into two functional subunits, held together by a disulfide bond. The 50kDa "light chain" is a zinc endoprotease. It catalyzes the cleavage of synaptosomalassociated protein of 25kD (SNAP-25) at the neuromuscular junction, thereby preventing neurotransmitter release, and inducing a flaccid paralysis (4). The 100kD "heavy chain" is made up of an N-terminal half, believed to mediate cytoplasmic entry of the light chain, and a C-terminal half (BoNT/A-H C ), which mediates cellular adherence (5).
While the enzymatic activity of the neurotoxin has been well defined, questions still exist regarding the molecular interactions needed for its cellular binding, internalization, and transport to its protein substrate. Identification of these interactions will provide guidance for developing potential inhibitors of its action.
It was 30 years ago that gangliosides were first mentioned as potential receptors for BoNT/A. Gangliosides, sialic acid containing glycosphingolipids, are found on the outer leaflet of cell membranes. Specifically, polysialylated gangliosides like GT1b [nomenclature according to Svennerholm (6)], were shown to be involved in BoNT/A neurotoxicity (7,8). Since then, GT1b has been shown to inhibit BoNT/A binding to synaptosomes (9), to quench BoNT/A fluorescence (10), and to be bound by BoNT/A when it was immobilized on either a thin layer chromatogram (11) or on plastic wells by guest on March 24, 2020 http://www.jbc.org/ Downloaded from Specific gangliosides are required for BoNT/A activity 5 (12). The dissociation constant for the adherence of BoNT/A-H C to GT1b-containing liposomes was recently reported to be ~10 -8 M (13). At the cellular level, cleavage of sialic acid residues by sialidase treatment of cultured cells isolated from spinal cord (14) and adrenergic chromaffin cells (15) was shown to reduce BoNT/A potency, while addition of a mixture of gangliosides to the sialidase treated chromaffin cells was found to restore it (16). In addition, a monoclonal antibody to GT1b was shown to antagonize the action of BoNT/A on rat superior cervical ganglion neurons (17). Most recently, GM2/GD2-synthase knockout mice were reported to show an increased time of survival when injected with a lethal dose of BoNT/A (18). Tetanus toxin, the other member of the clostridial toxin family, which has an affinity comparable to that of BoNT/A for gangliosides (13,19), was recently shown to use gangliosides as its cellular receptor (20).
The crystal structure of tetanus toxin C-fragment complexed with a synthetic GT1b DQDORJXHLQGLFDWHGWKHSUHVHQFHRIWZRELQGLQJVLWHVLQWKH-trefoil domain. One of these WZRVLWHVZDVDOVRIRXQGLQWKH-trefoil domain of BoNT/A and is strongly believed to be a ganglioside binding site (21,22). All of these studies indicated that gangliosides might be receptors for BoNT/A. In addition to gangliosides, a cell surface protein(s) may also be needed for BoNT/A adherence to cells. In contrast to more recent studies on chromaffin cells (16), early studies showed that BoNT/A binding to rat brain synaptosomes was trypsin sensitive (23,24). Also, a 140 kD protein in cholinergic synaptosomes prepared from Torpedo electric organs was identified as a BoNT/A binding protein. The interaction of BoNT/A with this 140 kD protein was both sialidase and trypsin sensitive (25). These studies, among others, led to the hypothesis that botulinum neurotoxins require a "double by guest on March 24, 2020 http://www.jbc.org/ Downloaded from Specific gangliosides are required for BoNT/A activity 6 receptor", comprised of protein and gangliosides (26). BoNT/B appears to follow this model. Nishiki et al. found that synaptotagmins I and II, in the presence of polysialylated gangliosides, serve as the receptors for BoNT/B (27)(28)(29)(30). Synaptotagmins, proteins that serve as calcium sensors for synaptic vesicle fusion, have also been proposed as possible receptors for BoNT/A. Studies indicated that BoNT/A adhered to synaptotagmin I solubilized from rat brain synaptosomes, as well as to purified synaptotagmin I immobilized on plastic microtiter wells (31). In contrast to the observations made with BoNT/B, this binding was not enhanced by the presence of gangliosides (31).
In this work, the question of whether specific gangliosides were needed for BoNT/A to act on SNAP-25 in synaptotagmin I expressing murine cholinergic Neuro 2a and human adrenergic SK-N-SH neuroblastoma cells was addressed. Both cell types were found to be susceptible to BoNT/A only when they contained specific gangliosides.
Interestingly, in their absence, BoNT/A activity was not observed, even though synaptotagmin I was present. in Dulbecco's modified Eagle's medium (DMEM) containing either 10% newborn bovine serum (Neuro 2a) or 10% fetal calf serum (SK-N-SH) in an atmosphere of 5%CO 2 / 95% air and 90% humidity. Human brain protein extracts were purchased from Clontech (Palo Alto, CA) and mouse brain extracts were kindly provided by the laboratory of Dr. Judith Bond (this university).

HPTLC analysis of lipid extracts of cultured cells
Neuro 2a (N2a) and SK-N-SH cells were seeded at a density of 1X10 7 cells per 150cm 2 tissue culture flask. When ganglioside-depleted N2a cells were needed, they were grown in the presence of medium containing 7.5 µM PPMP, an inhibitor of glucosyl ceramide synthase. Forty-eight hours after seeding, the medium was replaced with serum-free medium or serum-free medium containing either GM1, GD3, GD1a, or GT1b

Western blot analysis to determine synaptotagmin expression by cultured cells
Cells were seeded at 5X10 5 cells (N2a) or 1X10 6 cells (SK-N-SH) per 25cm 2 flask and allowed to grow for 96 hrs with a change to fresh culture medium after 48 hrs.

Western blot analysis of BoNT/A activity in cultured cells
N2a cells were seeded as described for analysis of synaptotagmin I content and grown in the presence of medium containing PPMP (concentrations of PPMP tested ranged from 1 -10 µM with 7.5 µM used in most experiments). SK-N-SH cells were also seeded and grown as described for analysis of synaptotagmin I. After 48 hours, the medium on both cell types was removed and replaced with serum-free medium with or without ganglioside (GM1, GD3, GD1a, or GT1b, 50µg/ml) and in the case of N2a cells, with PPMP. Control cells were grown in the absence of added ganglioside and PPMP.
Twenty-four hours later, the medium was changed to DMEM containing BoNT/A (6 nM), 10% serum, no added ganglioside, and for N2a cells, PPMP. Forty-eight hours later, the cells were harvested, as described above. Western blot analyses were used to determine the amount of intact and cleaved SNAP-25 present in 10-20 µg of whole cell extract (36). Densitometric analyses were performed as described above. To insure linearity, multiple exposures were taken of each blot, and all bands used for analysis were exposed at less than saturated levels.

Ganglioside content of neuronal cell lines
HPTLC analyses indicated that N2a cells expressed detectable levels of gangliosides with mobilities comparable to those of the monosialoganglioside standards GM1, GM2, and GM3, and a relatively high level of a ganglioside that migrated with the disialoganglioside standard, GD1a (Figure 1). These gangliosides were previously shown to be the major ones expressed by N2a cells (37). On a molar basis, the gangliosides identified as GM1, GM2, and GM3 accounted for 62%, and that as GD1a for 38% of the gangliosides in N2a cells. PPMP treatment reduced the level of GD1a by 85 ± 3% (n=4).
In contrast, resorcinol-positive lipids that migrated with the gangliosides GM2 and GM3 accounted for 95% and with GD1a for 5% of the gangliosides in SK-N-SH cells ( Figure   2).
Added gangliosides were taken up by both cell lines (Figures 1 and 2).

Synaptotagmin I content of neuronal cell lines
Synaptotagmin I was detected by Western blot analysis of N2a and SK-N-SH cell proteins ( Figure 3). The identification of synaptotagmin I in PPMP-treated N2a cells (data not shown) verified that its expression was not altered by inhibition of ganglioside synthesis. In each analysis, the anti-synaptotagmin I antibody adhered to a protein in the appropriate control (mouse or human) brain extract that had the apparent mass of synaptotagmin I (65kD).  (Figure 1, lanes 3 and 4). The effect of PPMP on BoNT/A activity was reversible. Growth of PPMP-treated N2a cells in PPMP-minus medium for 48 hr prior to exposure to BoNT/A, restored the susceptibility of SNAP-25 to the protease activity of BoNT/A (data not shown).

BoNT/A activity in SK-N-SH cells
Western blot analysis indicated that exposure of SK-N-SH cells to 6 nM BoNT/A for 48 hrs did not result in cleavage of SNAP-25 ( Figure 7). However, if cells were maintained for 24 hr in serum-free medium containing either GD1a or GT1b (50 µg/ml) and then exposed to BoNT/A for 48 hrs, cleavage of SNAP-25 was observed. Exposure

DISCUSSION
Botulinum neurotoxin was first described in the early 1800's (39). Now, almost 200 years later, the molecules necessary for the neurotoxin to enter neurons and find its substrate have yet to be unequivocally identified. This lack of knowledge hampers development of treatments for botulism or improvement of its use as a therapeutic agent.
If the components needed for the interaction of BoNT with neurons could be identified, inhibitors could be designed to prevent its uptake and thus limit its deadly activity.
Inhibitors that could be easily administered, perhaps taken orally, would be of particular interest to victims of bioterrorism or military personnel who may be exposed to BoNT, especially in areas far from modern medical facilities.
Many of the early studies of BoNT/A implicated gangliosides as possible receptors. Gangliosides were shown to reduce the toxicity of BoNT/A in vivo (7)(8)(9), as well as to adhere to BoNT/A in a variety of in vitro experiments (9)(10)(11)(12)(13). Tetanus toxin, the other member of the clostridial toxin family, which has an affinity comparable to that of BoNT/A for gangliosides (13,19), was recently shown to recognize gangliosides as its cell surface receptor (20). In addition, the crystal structure of the tetanus toxin Cfragment complexed with a synthetic GT1b analogue indicated the presence of two ELQGLQJVLWHVLQWKH-WUHIRLOGRPDLQ2QHRIWKHVHWZRVLWHVLVDOVRIRXQGLQWKH-trefoil domain of BoNT/A and is strongly believed to bind a ganglioside (21,22). However, the affinity of the interaction of BoNT/A with gangliosides has been traditionally believed to be weak in relation to other receptor/ligand interactions and therefore, unable account for the high potency of the toxin. For example, cholera toxin binds to its cell surface receptor, ganglioside GM1, with a K D in the range of 10 -9 to 10 -12 M (13,19,40,41).
However, the recent report that the apparent dissociation constant for the adherence of BoNT/A to GT1b was ~10 -8 M (13) supports the hypothesis that GT1b could function as its receptor.
In this work, BoNT/A activity was observed in N2a cells when it was added to the medium at concentrations similar to those reported for its dissociation constant. The observations that PPMP treatment of N2a cells resulted in an 85% decrease in GD1a synthesis and eliminated observable BoNT/A activity, while growing the PPMP-treated cells in the presence of GD1a or GT1b restored activity, indicate that specific gangliosides are needed for BoNT/A activity. These observations plus the fact that SK-N-SH cells, expressing primarily GM3 and GM2, were not susceptible to BoNT/A, unless they were grown in medium containing a gangliotetraose ganglioside, (GT1b > GD1a > GM1), indicate that specific gangliosides must be present in the cells in order for added BoNT/A to catalyze the cleavage of SNAP-25. The need for polysialylation was corroborated by the fact that GD1a or GT1b were more effective at restoring BoNT/A activity than GM1. This observation is in agreement with studies indicating that BoNT/A is neutralized by and has a greater affinity for GD1a and GT1b than GM1 (9,11,12,16).
The need for polysialylated gangliosides may explain the observation that in order to see levels of SNAP-25 cleavage in SH-SY5Y human neuroblastoma cells that were comparable to those observed in this study with control murine N2a cells, a 50-fold greater concentration of BoNT/A and a 24-hr longer exposure time were used (42). SH-SY5Y cells may be less susceptible to BoNT/A than N2a cells due to their lower expression of polysialylated gangliosides. The monosialylated gangliosides, GM1, GM2, and GM3 were shown to account for 71.5% of the total ganglioside sialic acid content of by guest on March 24, 2020 http://www.jbc.org/ Downloaded from the SH-SY5Y cells, while GD1a accounted for the rest (43). In the N2a cells used in this study, GD1a comprised 56 ± 3% (n=4) of the total ganglioside sialic acid. The higher content of polysialylated gangliosides in N2a cells may account for why activity was observed at a BoNT/A concentration below 1 nM. Combined, these observations indicate that N2a cells provide an effective model for studying the biological activity of BoNT/A.
Although addition of GD1a to N2a cells resulted in a cell-associated level of GD1a that was ~40% that of controls (Fig. 1)      µg/ml of GT1b. Following a 24 hr incubation, the medium was replaced with medium containing BoNT/A (6nM). After a 24 hr exposure, the cells were harvested in RIPA buffer. Twenty micrograms of protein in each whole cell extract were separated by SDS-PAGE under reducing conditions on a 5% stacking/13% running gel. Following transfer to PVDF, the immunoblot was probed with an anti-SNAP-25 monoclonal antibody followed by a HRP-conjugated goat anti-mouse antibody. Bands were visualized using SuperSignal West Pico Chemiluminescent Substrate™ and exposure to film.