PREPARATION OF HIGHLY PURE TOXIN FRAGMENTS
This invention relates to derivatives, such as fragments, of toxins, particularly clostridial neurotoxins. It also relates to preparations containing those derivatives and to methods of obtaining the derivatives and the preparations.
The clostridial neurotoxins are proteins with molecular masses of the order of 1 50 Da. They are produced by various species of bacterium of the genus Clostridium, most importantly C. tetani and several strains of C. botυlinum. There are at present eight neurotoxins known: tetanus toxin, and botulinum neurotoxin in its serotypes A, B, C D, E, F and G, and they all share similar structures and modes of action. The clostridial neurotoxins are synthesised by the bacterium as a single polypeptide that is modified post-translationally to form two polypeptide chains joined together by a disulphide bond. The two chains are termed the heavy chain (H), which has a molecular mass of approximately 1 00 kDa, and the light chain (L), which has a molecular mass of approximately 50 kDa.
The clostridial neurotoxins bind to an acceptor site (Black, J. D. & Dolly, J . O. , Neuroscience, 23, 767-779, 1 987 and Dolly et al. in Cellular and Molecular Basis of Cholinergic Function, ed. Dowdall, M. J. & Hawthorne, J . N., Chapter 60, 1 987) on the cell membrane of the motoneurone at the neuromuscular junction and are internalised by an endocytotic mechanism
(Montecucco et al. , Trends Biochem. Sci. , 1 1 , 31 4, 1 986) . It is believed that the clostridial neurotoxins are highly selective for motoneurons due to the specific nature of the acceptor site on those neurones. The binding activity of clostridial neurotoxins is known to reside in a carboxy-terminal region of the heavy chain component of the dichain neurotoxin molecule, a region known as Hc. The N-terminal region of the H-chain (HN domain) is thought to be of central importance in the translocation of the L-chain
into the cytosol and has been demonstrated to from channels in lipid vesicles (Shone et al, Eur. J. Biochem. 167, 1 75-1 80) .
Clostridial neurotoxins possess a highly specific zinc-dependent endopeptidase activity that is known to reside in the L-chain. Each toxin serotype hydrolyses a specific peptide bond within one of three proteins of the SNARE complex; VAMP (synaptobrevin), syntaxin or SNAP-25. Proteolytic cleavage of one of these proteins leads to instability of the SNARE complex and consequent prevention of vesicular release. The enzymatic activity of the light chain of the neurotoxin leads to inhibition of neurotransmitter release, which results in a prolonged muscular paralysis.
The central role of the SNARE proteins in regulated secretion has been convincingly established (e.g. Niemann et al. , ( 1 994) Trends Cell Biol. , 4, 1 79-1 85) . However, the correlation of SNARE protein involvement with the release of specific hormones, peptldes, transmitters and other signalling molecules remains to be established in the majority of cases. The range of highly specific endopeptidase activities of clostridial neurotoxin serotypes provides a unique approach to the understanding of SNARE-mediated events. Unfortunately, the use of native clostridial toxins for the study of such events is limited by at least two important aspects. Firstly, the expression of the requisite toxin receptor is restricted to a limited population of cells, thereby limiting the range of cell types in which SNARE- mediated events can be studied without cellular disruption. Secondly, the significant hazards associated with working with potent neurotoxins lead to restrictions on the range of applications and experimental design. Clostridial neurotoxins are the most potent neuroparalytic toxins known and must be manipulated in specialised laboratory conditions by specially trained and, preferably, vaccinated staff. The ability to produce highly purified non-toxic fragments of clostridial neurotoxins possessing the enzymatic activity of the clostridial neurotoxins and capable of delivery to the cytosol of selected cells would therefore provide a valuable tool for
studying secretory mechanisms.
The ability of the clostridial neurotoxins' enzymatic activity to destabilise SNARE complex formation and thereby inhibit vesicle fusion at the plasma membrane also has therapeutic potential. A number of therapeutic applications have been proposed (e.g. WO 96/33273 & WO 94/21 300) that are dependent on the successful retargeting of clostridial neurotoxin fragments. These approaches require a source of non-hazardous neurotoxin fragment that is suitable for the synthesis of non-toxic conjugates, since the side effect profile of a therapeutic contaminated with neurotoxin would be unacceptably high. In addition to retargeting of clostridial toxin fragments, there are further applications for non-toxic clostridial derivatives. For example, as an immunogen for vaccine preparation, as a source of material from which highly purified neurotoxin- related fragments can be prepared, and as a non-toxic endopeptidase standard in diagnostic kits (e.g. WO 95/33850) .
Since the cell binding function of clostridial neurotoxins resides in the Hc domain of the heavy chain, generation of a fragment in which the binding capability of the Hc has been deleted but the properties of the HN domain are retained (LHN) is potentially a suitable method for the production of a non-toxic derivative.
It is known to prepare these fragments by proteolytic treatment of toxin and then separation of toxin from fragments by anion exchange chromatography, and such methods successfully yield fragments that are 99.99% pure.
A central or recurrent problem associated with obtaining or using products from toxins made by these methods is the risk of residual toxicity in those products. It would hence be desirable to provide a method of removing toxin from such products. However, existing protocols have reached the limits of their abilities in this respect.
For example, it has been observed that the known fragments often exhibit a high inhibition of neurotransmitter release by neuronal cells in vitro. This has hampered investigation into the properties of conjugates in which a toxin fragment is combined with a ligand providing a specific targeting function, because of difficulty in providing controls against which to judge the conjugate activity.
It is an object of the present invention to provide a method of preparing a toxin derivative preparation. A further object of the invention is to provide a method of removing toxin from a toxin derivative preparation.
It has been discovered in accordance with the present invention that existing toxin derivative preparations, though considered to be pure, and though containing toxin at an extremely low level, nevertheless contain sufficient residual toxin to interfere with the applications of the fragment, conjugate or other toxin derivative.
Thus, in a first aspect of the invention, there is provided a method of reducing toxicity of a toxin derivative preparation, comprising contacting said preparation with a ligand which selectively binds to the toxin but not to the toxin derivative.
In a use of the invention, the ligand binds to and effectively neutralises residual toxin which is contaminating the toxin derivative preparation. The ligand preferentially binds to the toxin compared with its binding to the
derivative. Thus in use there may be some small loss of derivative at the same time as toxin binds to the ligand. It has been found that prior art preparations of toxin derivatives may contain toxin at levels of one toxin molecule per 1 0,000 toxin fragments, and even at this low level the toxin adversely effects the uses of the fragment. According to the invention, an antibody that preferentially binds to the toxin, but not to the derivative, can be used to reduce or remove toxicity associated with the toxin, thereby enabling the effects and applications of the derivative to be examined and used without any masking effect of the residual toxin.
In an embodiment of the invention, described in more detail below, a toxin fragment prepared according to prior art methods significantly inhibited substance P release from dorsal root ganglia. This inhibition was reduced almost entirely by the combination of the fragment with an antibody that specifically bound to toxin.
Antibodies for use in the present invention can be prepared using polyclonal or monoclonal techniques. Suitable methodology is found for example in "Antibodies: A laboratory manual, by Ed Harlow and David Lane, 1 988" . Monoclonal antibodies can be prepared by immunising mice against toxin, harvesting lymphocyte cells from the spleens of immunised mice and fusing these with myeloma cells. Antibodies secreted by the resulting hybridomas are screened for binding to toxin and positive clones selected and propagated. Monoclonal antibodies are harvested from cultured hybridomas and purified using chromatographic methods - see for example
Pharmacia handbook on "Monoclonal antibody purification" . As an alternative to immunising mice with the toxin itself, the mice can be immunised with a different source of a Hc domain, whether obtained from native material or expressed in an alternative, non clostridial, host. Alternatively, mice can be immunised with toxoided preparations of intact neurotoxins and the anti Hc antibodies selected. This may be achieved in at least 2 ways for example: specific Hc antisera may be bound to
immobilised Hc and subsequently eluted for use. Conversely, specific Hc antisera may be obtained by adsorbing non-He antisera onto LHN.
This invention thus provides, in specific examples, preparation of specific, defined fragments from native clostridial neurotoxins. Purified clostridial neurotoxin obtained from Clostridia sp. using previously reported techniques (for example, Shone, C. C. and Tranter, H. S. ( 1 995) in "Clostridial Neurotoxins - The molecular pathogenesis of tetanus and botulism", (Montecucco, C, Ed.), pp. 1 52-1 60, Springer) can be fragmented by proteolytic or chemical cleavage to yield a crude mixture of derivatives that possess elements of the light chain, heavy chain or both (Gimenez, J. A. & DasGupta, B. R. ( 1 993) J. Protein Chem. , 12, 351 -363; Shone, C. C, Hambleton, P. and Melling, J . ( 1 987) Eur. J. Biochem. 167, 1 75-1 80) . Classical chromatographic techniques are used to separate the crude mixture into partially purified fragments, the residual toxicity of which would make them unsuitable for many applications, and the fragments are then combined with a neutralizing ligand.
During testing by the inventors of a conjugate or a toxin fragment with a targeting ligand, it was not possible in vitro to determine the selectivity of an LHN fragment, though the reasons for this were not known to the inventors at the time. According to the invention, it has surprisingly been found that very low levels of toxins present were masking the LHN fragment activity, and now advantageously it is possible further to reduce this toxin content so that pure LHN activity can be measured and assessed .
In the particular case of in vivo uses of fragments and conjugates and other derivatives, a specific embodiment of the invention, described in more detail below, determined that fragments according to the present invention exhibited a level of toxicity that was more than 1 0 times lower than that of the prior art fragments, which prior art fragments were hitherto considered as being pure and toxin-free.
It is a preferred additional step in the method of the invention to separate the ligand from the toxin derivative preparation.
In one embodiment of the invention, a specific binding event between an immobilised matrix and a domain present on the neurotoxin, but absent from the fragment, is used to separate neurotoxin from the fragment. Such a method is used specifically to bind neurotoxin and other fragments possessing the requisite binding domain from a crude neurotoxin fragment mixture. Non-binding fragments that are free of neurotoxin are simply isolated from the column flow through. Binding fragments including neurotoxin can be isolated by altering the conditions of binding, for instance by altering the chemical environment (e.g. pH, ionic strength) or incorporation of a substance that competes for the binding site (e.g. peptide, sugar moiety) . One suitable example is a column containing the natural receptor, or a version of the receptor, for which the neurotoxin has an affinity. This receptor may be purified and immobilised to a matrix for use in a column or free in solution, or present in a preparation of cells or cell membrane, and is of use for purification of LHN. Typically, a crude preparation of clostridial neurotoxin fragments is applied to the receptor preparation to bind neurotoxin and other fragments that possess receptor binding properties. Non-binding fragments, which include LHN, are recovered from the receptor preparation by simple elution. Binding fragments and neurotoxin are released and harvested as described above.
Purification of clostridial endopeptidases is also suitably achieved according to the invention using metal ion chromatography. Clostridial endopeptidases are characterised as metalloendopeptidases due to the co¬ ordination of a metal ion at the active site of the enzyme. Given this specific metal ion binding capacity of clostridial neurotoxins, it might be predicted that both neurotoxin and endopeptidase fragments bind to immobilised metal resin via this catalytic site interaction. However, it is found that the LHN does not bind to the chelating column whereas the
neurotoxin does. In an example, set out in more detail below, a method for the purification of LHN/A utilises immobilised zinc ions to bind BoNT/A and purify LHN/A. The low toxicity of LHN/A purified by this method is also confirmed below.
Another embodiment of the invention comprises immobilising specific antibodies to a column resin. The antibodies are selected on the basis of their specificity for epitopes present on the neurotoxin but absent on the desired fragment, exemplified by LHN below. In the presence of partially purified LHN, antibodies with specificity for the Hc domain will only bind to the neurotoxin. Contaminating neurotoxin is removed from the LHN preparation by entrapment on the immobilised-antibody matrix, whereas LHN, which is not recognised by the antibodies, does not interact with the column. This method is surprisingly efficient at removing residual toxicity from the LHN preparation and affords an effective purification technique.
Entrapment of the neurotoxin contaminant by antibody binding, rather than specifically binding the LHN, enables the elution conditions to be maintained at the optimum for LHN stability. The use of harsh elution conditions e.g. low pH, high salt, chaotropic ions, which may have detrimental effects on LHN polypeptide folding and enzymatic activity, are therefore avoided.
Neurotoxin and other binding fragments may be eluted from the antibody column to release clostridial neurotoxin derivatives that are purified from derivatives deficient in the Hc domain epitopes.
An additional advantage of the methods of the invention is thus that the desired component of the mixture that is being purified, that is to say the fragment or the conjugate or the other derivative, is that portion which is eluted from the column whereas the undesired portion, the toxin, remains bound to the column. This has the benefit that the desired material is less affected by the column and that no additional step, for example to elute bound, desired material from the column, is required as part of the method.
Further embodiments of the invention use two different affinity techniques in combination. In a preferred embodiment of the invention exemplified below, combinations of antibodies with different epitope recognition properties are used. By utilising monoclonal antibodies with different recognition epitopes, small conformational changes in neurotoxin can be accommodated. In this way, a greater proportion of neurotoxin is targeted for removal from the crude starting mixture.
It is further preferred that the method of the invention comprises an additional step, after separating the ligand from the toxin derivative preparation, which ligand we will refer to in these present paragraphs as the first ligand, of contacting the toxin derivative preparation with a second ligand, which selectively binds to the first ligand but not to the toxin derivative. It is occasionally the case that ligand attached, for example, to a chromatography column, detaches and thus the toxin derivative elutes from the column in combination with-complexes of ligand and toxin. This opens the possibility of separation of the complex at a future time, releasing the toxin. It is an advantage of the preferred method of the invention that these ligand-toxin complexes, if present in the toxin derivative preparation are substantially removed by use of the further ligand.
The further, or second, ligand can suitably be an antibody that binds to the antibody used as the first ligand. A specific embodiment of the invention thus comprises:-
preparing a toxin derivative preparation which comprises toxin derivative which is contaminated by low levels of toxin;
contacting the preparation with a ligand which selectively binds to the toxin but not to the toxin derivative;
separating the ligand from the toxin derivative preparation, thereby separating toxin from the toxin derivative preparation;
contacting the preparation with a further ligand that binds selectively to the first ligand, or binds selectively to a complex of the first ligand with toxin; and
separating the second ligand from the toxin derivative preparation.
The first ligand can be an antibody which binds to an Hc portion of the toxin, and the second ligand can be a further antibody or immunoglobuiin binding domain which binds to the first antibody or to a complex of the first antibody with toxin, and in a specific example the second ligand can be protein G. In one particular example of the preferred embodiment of the invention in use, injection of 20 micrograms LHN purified according to the invention using the first ligand resulted in 0 out of 4 mice surviving, whereas use of the second ligand to remove antibody - toxin complexes resulted in 4 out of 4 survivors. This very surprising result exhibits the improved purity of the toxin derivative following application of the second purification step.
Specific antibodies or antibody fragments are optionally mixed with partially purified clostridial neurotoxin fragments in solution to form bound antibody- neurotoxin complexes. The antibody-neurotoxin complexes are then isolated from the mixture (e.g. by Protein G chromatography) and removed to yield purified agents in solution.
Immobilised monoclonal antibodies may be used specifically to bind contaminating neurotoxin and intact Hc or other contaminants. Two BoNT/A Hc specific monoclonal antibodies are thus utilised in a method that is surprisingly efficient at binding contaminating neurotoxin to prepare LHN/A of high purity and low toxicity.
The present invention is of application to any toxins which can be converted into useful fragments, conjugates and other derivatives, for example by proteolytic action, and is of particular application to clostridial neurotoxins, especially botulinum and tetanus toxins of all sub-groups.
By toxin derivative, it is intended to encompass all derivatives of a toxin that are prepared directly or indirectly from native toxin, for example by proteolytic action on the toxin, which methods can result in a preparation having residual, low levels of toxin present. Thus, a toxin derivative preparation according to the definition does not include recombinant toxin derivatives which are guaranteed to be free of toxin. The meaning of derivative thus encompasses toxin fragments and conjugates of toxin fragments with other molecules as well as variants of toxins and toxin fragments and conjugates.
In a further aspect, the invention provides an affinity chromatography column, for removal of toxin from a toxin derivative preparation, wherein the column comprises a ligand that selectively binds to toxin but not to the toxin derivative. This column is of use in separating toxin from toxin derivative, and the ligand employed is suitably an antibody or a metal ion.
A still further aspect of the invention lies in a toxin derivative preparation comprising 1 -1 00 ppm toxin per toxin derivative, preferably 1 0 - 1 OOppm. This preparation is derived from native toxin, and has a greatly reduced residual level of toxin.
A yet further aspect of the invention provides a composition comprising a derivative of a toxin and a pharmaceutically acceptable carrier, and further comprising a ligand that binds selectively to the toxin. The composition may for example comprise a conjugate of a toxin with a ligand that binds selectively to the toxin, wherein the toxin is bound non-covalently to the ligand. The toxin is thus neutralized by the ligand.
The invention is now described in specific embodiments, accompanied by drawings in which:-
Fig. 1 shows SDS-PAGE analysis of LHN/A purified by the immunoaffinity approach;
Fig. 2 shows retention of functional activity of LHN/A prepared by the invention
Fig. 3 shows inhibition of neurotransmitter release from eSCN by LHN/A and BoNT/A; and
Fig. 4 shows mouse toxicity data for purified LHN/A.
Example 1 Inhibition of Glvcine Release
Embryonic spinal cord cells were treated for one day in the presence of 30 micrograms per ml of LHN/A (prior art preparation) . This treatment resulted in an inhibition of glycine release of 64%. A parallel treatment was carried out in the presence of antibody 5BA 9.3, which binds specifically to botulinum neurotoxin A. This resulted in an inhibition of glycine release which was reduced to 44% .
Example 2 Inhibition of Substance P Release
Embryonic Dorsal route ganglia were treated for 3 days with 20 micrograms per ml LHN/A, LHN/A + antibody 5BA 9.3 and also with a conjugate of LHN/A with a targeting ligand in the presence of antibody 5BA 9.3.
The inhibition of substance P release was significant when the 20μg/ml
LHN/A fragment was used alone, and was at a level of about 26%. This level was reduced to about 4% in the presence of the specific antibody, and the level rose to about 21 % when the conjugate (also 20/vg/ml) was used in the presence of the specific antibody.
Example 3
Production of LHM/A from BoNT/A by antibody-affinity chromatography
BoNT/A was prepared according to a previous method (Shone, C. C. and Tranter, H . S. ( 1 995) in "Clostridial Neurotoxins - The molecular pathogenesis of tetanus and botulism", (Montecucco, C, Ed.), pp. 1 52- 1 60, Springer) . FPLCΦ chromatography media and columns were obtained from Amersham Pharmacia Biotech, UK. Affi-gel Hz™ matrix and materials were from BioRad, UK.
Preparation of an anti-BoNT/A antibody-affinity column
An antibody-affinity column was prepared with specific monoclonal antibodies essentially as suggested by the manufacturers' protocol. Briefly, monoclonal antibodies 5BA2.3 & 5BA9.3 which have different epitope recognition in the Hc domain (Hallis, B. , Fooks, S., Shone, C. and Hambleton, P. ( 1 993) in "Botulinum and Tetanus Neurotoxins", (DasGupta, B. R. , Ed.), pp. 433-436, Plenum Press, New York) were purified from mouse hybridoma tissue culture supernatant by Protein G (Amersham Pharmacia Biotech) chromatography. These antibodies represent a source of BoNT/A Hc-specific binding molecules and can be immobilised to a matrix or used free in solution to bind BoNT/A. In the presence of partially purified LHN/A (which has no Hc domain) these antibodies will only bind to BoNT/A. The antibodies 5BA2.3 & 5BA9.3 were pooled in a 3: 1 ratio and two mg of the pooled antibody was oxidised by the addition of sodium periodate (final concentration of 0.2%) prior coupling to 1 ml Affi-Gel Hz™ gel ( 1 6 hours at room temperature). Coupling efficiencies were routinely
greater than 65%. The matrix was stored at 4°C in the presence of 0.02% sodium azide.
Purification strategy for the preparation of pure LHN/A
BoNT/A was treated with 1 7μg trypsin per mg BoNT/A for a period of 72- 1 20 hours. After this time no material of 1 50kDa was observed by SDS- PAGE and Coomassie blue staining. The trypsin digested sample was chromatographed (FPLC^system, Amersham Pharmacia Biotech) on a Mono QΦ column (HR5/5) to remove trypsin and separate the majority of BoNT/A from LHN/A. The crude sample was loaded onto the column at pH 7 in 20mM HEPES, 50mM NaCI and 2ml LHN/A fractions eluted in a NaCI gradient from 50mM to 1 50mM. The slightly greater pi of BoNT/A (6.3) relative to LHN/A (5.2) encouraged any BoNT/A remaining after trypsinisation to elute from the anion exchange column at a lower salt concentration than LHN/A. LHN/A containing fractions (as identified by SDS- PAGE) were pooled for application to the antibody column.
The semi-purified LHN/A mixture was applied and reapplied at least 3 times to a 1 -2ml immobilised monoclonal antibody matrix at 20°C. After a total of 3 hours in contact with the immobilised antibodies, the LHN/A-enriched supernatant was removed. Entrapment of the BoNT/A contaminant, rather than specifically binding the LHN/A, enables the elution conditions to be maintained at the optimum for LHN stability. The use of harsh elution conditions e.g. low pH, high salt, chaotropic ions, which may have detrimental effects on LHN polypeptide folding and enzymatic activity, are therefore avoided. Treatment of the immobilised antibody column with 0.2M glycine/HCI pH2.5 resulted in regeneration of the column and elution of BoNT/A-reactive proteins of 1 50kDa.
The LHN/A enriched sample was then applied 2 times to a 1 ml HiTrap" Protein G column (Amersham Pharmacia Biotech) at 20°C. Protein G was
selected since it has a high affinity for mouse monoclonal antibodies. This step was included to remove BoNT/A-antibody complexes that may leach from the immunocolumn. Antibody species bind to the Protein G matrix allowing purified LHN/A to elute.
The profile of the purification procedure is illustrated by the SDS-PAGE analysis in Figure 1 . Trypsin digested BoNT/A (lane 4) was applied to a Mono Q® anion exchange column and fractions harvested (lane 5) . Material pre- and post-Protein G is indicated in lanes 6 and 7 respectively. Lane 7 represents the final purified LHN/A preparation. Samples were analysed by
SDS-PAGE on 4-20% polyacrylamide and stained with Coomassie blue (Panel A), or Western blotted and probed with anti-BoNT/A (Panel B) . Molecular weight markers are indicated on the Figure with reference to the standards in lanes 2 & 8. Molecular weight markers in lane 1 are compatible with enhanced chemiluminescence and are for visualisation purposes only.
In vitro SNAP-25 peptide cleavage
The in vitro cleavage of SNAP-25 by LHN/A and other endopeptidase samples were assessed essentially as described previously (Hallis, B., James, B. A. F. and Shone, C. C. ( 1 996) J. Clin. Microbiol. 34, 1 934- 1 938) . Figure 2 clearly demonstrates the similarity in catalytic activity between purified LHN/A and reduced BoNT/A. A series of LHN/A ( • ) and BoNT/A (□) dilutions were incubated for 1 hour prior to assessment of
SNAP-25 cleavage. The data is representative of at least 7 experiments which have suggested EC50 for LHN/A and BoNT/A to be 3.8 ± 0.7pM and 3.6 ± 0.6pM respectively.
SDS-PAGE and Western Blotting
SDS-PAGE and Western Blotting were performed using standard protocols.
Proteins were resolved on a 4-20% Tris/glycine polyacrylamide gel (Novex) and either stained by the addition of Coomassie blue or transferred to nitrocellulose. Positive binding of antibodies was detected by enhanced chemiluminescence.
Example 4
Production of LHN/A from BoNT/A by immobilised metal affinity chromatography
BoNT/A was prepared according to a previous method (Shone, C. C. and
Tranter, H . S. ( 1 995) in "Clostridial Neurotoxins - The molecular pathogenesis of tetanus and botulism", (Montecucco, C, Ed .), pp. 1 52- 1 60, Springer) . FPLC51 chromatography media and columns were obtained from Amersham Pharmacia Biotech, UK. Affi-gel Hz™ matrix and materials were from BioRad, UK.
Preparation of cationic metal affinity column
An affinity column was prepared essentially as suggested by the manufacturers' protocol. Briefly, chelating Sepharose* (Amersham
Pharmacia Biotech) was washed with 50mM EDTA + 1 M NaCI ( 1 0 column volumes), washed with 1 0 column volumes ultra-high purity water, primed with 5mg/ml ZnCI2 ( 10 column volumes -neutral pH, filtered), and finally washed with purified water ( 1 0x column volumes) .
Purification strategy for the preparation of pure LHN/A
BoNT/A was treated with 1 7μg trypsin per mg BoNT/A for a period of 72- 1 20 hours. After this time no material of 1 50kDa was observed by SDS- PAGE and Coomassie blue staining. The trypsin digested sample was chromatographed (FPLCS system, Amersham Pharmacia Biotech) on a Mono Q* column (HR5/5) to remove trypsin and separate the majority of BoNT/A
from LHN/A. The crude sample was loaded onto the column at pH 7 in 20mM HEPES, 50mM NaCI and 2ml LHN/A fractions eluted in a NaCI gradient from 50mM to 1 50mM. The slightly greater pi of BoNT/A (6.3) relative to LHN/A (5.2) encouraged any BoNT/A remaining after trypsinisation to elute from the anion exchange column at a lower salt concentration than LHN/A. LHN/A containing fractions (as identified by SDS- PAGE) were pooled for application to the metal affinity column.
For the purification of 1 mg LHN/A a 1 ml column of Zn2 + primed chelating Sepharose®, prepared as described above was equilibrated to the appropriate pH (7.5-8.5) using 50mM HEPES + 1 M NaCI ('equilibration buffer') . 1 mg of LHN (post Mono Q® fractionation) was applied to the column after dialysis against the appropriate equilibration buffer. Material that did not bind was recovered and the column washed with excess equilibration buffer. Bound material was eluted by the application of equilibration buffer supplemented with 1 0mM EDTA.
Another suitable method is described by Rossetto et al, Biochem J., ( 1 992), vol. 285, pp 9-1 2.
Example 5
Evaluation of BoNT/A contamination levels in LHN/A prepared by immunocolumn chromatography in primary cultures of spinal cord neurons
Inhibition of [3H]-glycine release from eSCN
Spinal cord neurons are exquisitely sensitive to neurotoxin, with an IC50 of inhibition of glycine release of 0.027 ± 0.0006pM after 3 days exposure to BoNT/A (unpublished observations) . Therefore they serve as a suitably sensitive assay for screening samples for the presence of neurotoxin.
Spinal cords dissected from 1 4- 1 5 day old foetal Sprague Dawley rats were cultured for 21 days using a modification of previously described method
(Ransom, B. R., Neale, E., Henkart, M., Bullock, P. N. and Nelson, P. G. ( 1 977) J. Neurophysiol. 40, 1 1 32- 1 1 50 & Fitzgerald, S. C. ( 1 989) "A dissection and tissue culture manual of the nervous system", Alan R. Liss, Inc., New York, NY) . Cells were loaded with [3H]-glycine for 30 minutes prior to determination of basal and potassium-stimulated release of transmitter (essentially as described in Williamson, L. C, Halpern, J. L., Montecucco, C, Brown, J. E. and Neale, E. A. ( 1 996) J. Biol. Chem. 271 , 7694-7699). A sample of 0.2M NaOH-lysed cells was used to determine total counts, from which % release could be calculated.
In figure 3 there is shown a comparison of the inhibition curves obtained after 3 days incubation of spinal cord neurons with LHN/A and BoNT/A. Samples of LHN/A ( • ) or BoNT/A (□) were diluted in eSCN growth medium and 1 ml of the appropriate concentration applied. Cells were incubated for three days prior to assessment of [3H]-glycine release. The % inhibition data is calculated by relating the net stimulated [3H]-glycine release to the total uptake and then expressing this as a percentage of the release obtained from control media treated cells. The data are representative of at least three experiments. The IC50 data determined for LHN/A ( 1 06.2 ± 49.3nM) and BoNT/A can be used to estimate the ratio of BoNT/A to LHN/A in the purified material. It is estimated that a maximum of 1 BoNT/A molecule per 4x106 LHN/A molecules was present in the final purified material.
Example 6
Removal of BoNT/A from LHN/A preparations as assessed by mouse lethality
Residual BoNT/A contamination was evaluated following intraperitoneal injection of 0.5ml of test sample in gelatine-phosphate buffer ( 1 % (w/v)
Na2HPO4, 0.2% (w/v) gelatine, pH 6.5-6.6) into mice. After 4 days the number of surviving animals was counted. Literature precedents (Shone
et al. 1 987 & de Pavia, A. and Dolly, J. O. 1 990) cite the LD50 to be the concentration of test sample that killed half the animals in the test group within 4 days. By this analysis, batches of purified LHN/A were demonstrated to exhibit an LD50 of approximately 50μg/mouse i.e. approximately 20LD50/mg. This is significantly lower than previously reported LD50 data for LHN/A (6000- 1 2000LD50/mg by Shone, C. C, Hambleton, P. and Melling, J. ( 1 987) Eur. J. Biochem. 167, 1 75-1 80) and LC/A ( < 1 00LD50/mg in Shone et al. 1 987 & 10000LD50/mg in de Pavia, A. and Dolly, J. O. ( 1 990) FEBS Lett. , 277, 1 71 -1 74) .
As an alternative method of data analysis, the mean number of survivors from different batches of LHN/A was determined and is presented in Figure 4 (Abbreviations: IMAC; immobilised metal affinity column, ND; Not determined) . These data clearly demonstrate the significantly low toxicity of the LHN/A prepared by both exemplified methods of purification.