WO1990011767A1 - Purified trichosanthin and method of preparation - Google Patents

Abstract

A substantially pure, hemagglutinin-free composition of trichosanthin, and a method of producing the composition is disclosed. A plant extract from Trichosanthes kirilowii is contacted with an anionic exchange resin to remove contaminating hemagglutinins, and trichosanthin is further purified from the extract by cation exchange chromatography to a purity of greater than 95 %. Also disclosed is a stable trichosanthin composition suitable for parenteral injection and a method of producing the stable composition. The stable composition is formed by adding a non-ionic surfactant and sucrose, to an aqueous solution of purified trichosanthin. This composition can be stored in aqueous form for several days at between 2-8°C without significant protein precipitation. Alternatively, the stable composition can be stored in lyophilized form, and reconstituted with substantially complete resolubilization of the protein.

Description

PURIFIED TRICHOSANTHIN AND METHOD OF PREPARATION 1. Field of the Invention

The present invention relates to a purified trichosanthin (TCS) composition and to a novel method of producing the composition. 2. References

Armstrong, W.H., Presented at the International Conference on AIDS, 23-25 June, 1986, Paris.

Calderwood, S.B., et al., Proc Nat Acad Sci USA, 84:4364 (1987).

Chayt, K.J., et al., JAMA, 256:2356 (1986).

Coleman, W.H., et al., Biochem Biophys Acta, 696:239 (1982).

Gartner, S., Science, 233:215 (1986a).

Gartner, S., et al., JAMA, 256, 2365 (1986b).

Gasperi-Campani, et al., FEBS Lett, 76(2):173 (1977).

Grasso, S., et al., Agric Biol Chem, 52(4):1095 (1988).

Gu, Zi-wei, et al., Acta Chemica Sinica, 43:943 (1984).

Ho, D.D., et al., J Clin Invest, 77:1712 (1986).

Hsu, K.J., et al., Acta Zool Sin, 22:149 (1976).

Hwang, Y.N., Chinese J Integrated Trade and Western Medicine, 7:154 (1987).

Irvin, J., Arch Biochem Biophys, 169:5221 (1985).

Kao, H., et al., Acta Biol Exp Sin, 11:253 (1978).

Kezhan, Pan, et al., Supplement of Proc of China-Japan Bilateral Symposium on Biophys, Wuxi, China (May 1985).

Koenig, S., et al., Science, 233:1089 (1986).

Kuo-Fen, C., et al., Obs and Gyn, 59(4):494 (1982). Law, L.K., et al., J Reprod Fert, 69:597 (1983).

Lifson, J.D., et al., Science, 232:1123 (1986a).

Lowry, O.H., et al., J Biol Chem, 193:265 (1951).

Maraganove, J.M., et al., J Biol Chem, 262 (24) :11628 (1987).

Motola, et al., "Preformulation Research."

Olnes, S., Nature, 328:474 (1987).

Olnes, S., et al., in Molecular Action of Toxins and Viruses, (Elsevier, 1982), Chapter 3.

Steicher, H.Z., et al., JAMA, 256:2390 (1986).

U.S. Pharmacopia, XXI, p. 1105 (1985).

Wang, Yu, et al., Int Symposium on Org Chem of Medicinal Natural Products, Shanghai China (November 1985) .

Xiong, Y.Z., et al., Acta Zool Sin, 11:236 (1976).

Xuejan, Z., et al., Nature, 321:477 (1986).

Yeung, H.W. et al., Int J Peptide Protein Res., 27:325 (1986).

3. Background of the Invention

TCS is a plant protein which is obtained from the

Trichosanthes kirilowii root tuber. The protein, which is also known as alpha-TCS (Law) and Radix trichosanthis (Kuo- Fen), is a basic, single-chain protein which has been reported to containing about 224 (Gu) to 234 (Xuejan) amino acid residues, and reported to have a molecular weight of about 24,000 daltons. The protein sequence of TCS has been completed (Gu; Wang), and a molecular model has been derived from cytofluorographic X-ray analysis (Kezhan).

It has been shown that TCS is a potent inhibitor of protein synthesis in a cell-free lysate system (Maraganove). This activity is consistent with the observed homology in amino. acid sequence between TCS and the A chain of ricin, a ribosome-inactivating protein (RIP) which shows amino acid homology with a number of other RIPs, including abrin A chain (Olnes, 1982, 1987) and modeccin (Olnes, 1982), and various single-chain ribosome-inactivating proteins, such as pokeweed anti-viral protein (PAP) (Irvin), RIPs from a variety of other plants (Coleman; Grasso; Gasperi-Campani) and the A subunit of Shiga-like toxins from E. coli (Calderwood).

TCS, or plant extracts containing TCS, have been used in China as an abortifacient agent for inducing abortion in humans, particularly during midtrimester (14 to 26 weeks). As such, the drug has been administered by intramuscular, intravenous, or intraamniotic routes. The phenomenon of mid-term abortion has been attributed to the selective destruction of placental villi. Other studies indicate that the syncytiotro- phoblast is preferentially affected (Hsu; Kao) and that secretion of hCG may be impaired (Xiong). TCS has also been shown to have a suppressive effect on human choriocarcinoma, and the protein appears to be able to pass the blood/brain barrier (Hwang).

Co-owned U.S. Patent No. 4,795,739 for "Method of Inhibiting HIV" describes the use of TCS for inhibiting HIV (Human Immunodeficiency Virus) proliferation in infected human T cells and macrophages, for treating HIV infection in humans. In view of the significant therapeutic use of TCS, it would be valuable to obtain the protein in substantially pure form by preparation methods which are suited to large-scale processing.

Methods of preparing TCS have been reported (Yeung).

These methods typically involve an initial aqueous extraction of fresh root tubers of T. kirilowii, and lyophilization of the extract supernatant. The resulting powder is dissolved in water medium, followed by addition of solvent, such as an aqueous acetone mixture, to yield a precipitate containing TCS. After removal of the precipitate by centrifugation, additional solvent is added to the supernatant, yielding a second TCS-containing precipitate fraction on centrifugation. Both precipitates are re-extracted with aqueous medium, giving soluble protein fractions containing partially purified TCS.

The soluble fraction from the second-solvent precipitate can be further purified by cation exchange column chromatography, using a NaCl gradient to release the protein. The eluate profile from the column indicates the presence of contaminating proteins on both sides of the protein peak identified as TCS.

Alternatively, the soluble fraction from the first-or from the second-solvent precipitate can be purified by an initial pass through an anion exchange resin column, and the eluate further purified by cation exchange column chromatography. Although the resulting protein is somewhat more pure than that from the method involving cation exchange chromatography alone, the cation exchange elution profile still shows the presence of prominent side peaks, as evidenced by size-exclusion HPLC.

Studies carried out in support of the present application, and discussed below, indicate that the TCS protein isolated by the above methods contains readily detectable contaminating proteins. Based on the observed hemagglutination activity of the purified protein preparation, at least one of the contaminants is a hemagglutinin. The above-described purification methods are also difficult to adapt to large scale, in view of the need for volatile organic solvent extraction.

A further limitation of known methods of handling and storing TCS compositions is the difficulty of maintaining the protein in soluble form on storage. Preliminary stability studies conducted in support of the present application indicate that a solution of purified TCS turns cloudy on storage at refrigerator temperature for several days. If, on the other hand, the purified protein is stored in lyophilized form, the protein is difficult to bring into solution on rehydration.

4. Summary of the Invention

One object of the present invention is to provide a substantially pure TCS composition, and in particular, a composition which is substantially free of hemagglutinins.

Another object of the invention is to provide a method for producing purified TCS which is easily adapted to large-scale processing.

In practicing the TCS purification method of the invention, a clarified plant extract from the roots of T. kirilowii is contacted with an anion exchange material for removal of extract contaminants which bind to the material. According to one aspect of the purification method, it has been found that carrying out the anion exchange material binding step in a low-conductivity buffer is effective to remove substantially all hemagglutinins from the extract. The extract is also contacted with a cation exchange material, under conditions which separate TCS from extract contaminants which bind more weakly to the cation exchange material than TCS at a low buffer conductivity, and from contaminants which bind more strongly to the cation exchange material than TCS at a higher conductivity.

In a preferred embodiment, clarified extract is first contacted with the anion exchange material, and the eluate from the anion exchange material is treated by diafiltration to remove UV-absorbing contaminants having molecular weights of less than about 10,000 daltons. The diafiltration step also serves to bring the eluate to a pH and conductivity suitable for the subsequent exchange step. The eluate is now contacted with the cation exchange material, which is washed extensively to remove loosely bound contaminants. The TCS protein is released from the cation exchange material, in substantially, pure form, by washing the column with a selected, higher conductivity buffer.

The purity of the TCS protein, as judged by HPLC chromatography and SDS gel electrophoresis, is greater than about 95% pure, and preferably greater than 98% pure. The protein composition shows little or no hemagglutination activity.

In another aspect, the invention includes a TCS composition produced by the above purification method, a substantially pure TCS composition (having a purity of greater than about 95%), and a TCS composition which is substantially free of hemagglutinins. In one embodiment, the purified composition contains a non-ionic detergent, such as Tween-20™ and a sugar, such as sucrose and/or mannitol, in an amount which is effective in combination to prevent protein precipitation when the composition is stored avoer a several-day period in solution at 2°-8°C.

According to another aspect of the invention, it has been found that a purified TCS composition containing the above detergent and sugar components can be stored in lyophilized form, and reconstituted with water to solution form with little or no evidence of precipitate.

These and other objects and features of the invention will become more fully apparent when the following detailed description of the invention is read in conjunction with the accompanying drawings.

Brief Description of the Drawings

Figure 1 is a flow diagram of steps in the TCS-purification method of the invention;

Figures 2A-2D show size-exclusion HPLC profiles of (2A) protein extract from T. kirilowii; (2B) extract material after removal of hemagglutinins and other contaminants which bind to an anion-exchange material under low-condtictivity conditions; (2C), partially purified extract after concentration and diafiltration a using low-molecular weight cut-off membrane; and (2D), after final TCS purification by cation-exchange material chromatography;

Figure 3 shows SDS polyacrylamide gel electrophoresis of purified trichosanthin prepared according to invention and stained with Coomassie blue, at loaded protein amounts ranging from 0.25 μg to 40 μg/gel; and

Figure 4 shows SDS polyacrylamide gel electrophoresis of the trichosanthin preparation applied in the Figure 3 gels, and stained with silver, at loaded protein amounts ranging from .005 μg to 10 μg.

Figure 5 shows the percent turbiity (open squares) and percent protein recovery (closed diamonds) produced in each of 12 samples (trial numbers) examined for Plackett-Burman analysis.

Figure 6 shows the average protein recovery effect determined from the Figure 5 data for each of 4 variables.

Figure 7 shows the average turbidity effect determined from the Figure 5 data for each of 4 composition variables. Detailed Description of the Invention

I. Method of Purifying TCS

Figure 1 outlines the general method for producing purified TCS according to the method of the invention. A clarified extract of roots of T. kirolowii is obtained, conventionally, using peeled and sliced root tubers obtained from fresh or frozen tubers. The tubers are conveniently processed in a juicer or blender, and filtered to remove pulp. The pH is adjusted to 8.0 and the material is allowed to stand over- night, then clarified by centrifugation, following the general method of Example 1. The volume of aqueous medium used for processing is not critical, but is preferably just large enough to facilitate homogenization. The medium will usually be a salt solution, such as normal saline (0.9% w/v NaCl), phosphate buffer or phosphate buffered saline at a preferred pH of between about 7-9.

A typical size-exclusion high pressure liquid chromatography (HPLC) profile (using BioSil TSK125 column) of the clarified extraction mixture is shown in Figure 2A. The 11.3 min. retention time (RT) peak in this profile corresponds to TCS.

In the following discussion, the peak retention time values referred to were obtained by size-exclusion HPLC, using either Biosil TSK250 or Biosil TSK125. In particular, TSK250 was used to obtain better resolution of high-molecular weight species. Those retention-time values which were obtained using TSK125 and specifically designated in parenthesis and correspond to peaks shown in Figures 2A-2D. All other retention-time values were obtained by TSK250 HPLC. These retention time values are not absolute values, but will vary from system to system according to flow rate, column length, temperature and system configuration. Following clarification, the extract is diluted with water to the desired conductivity and contacted with an anion exchange material. This is preferably done by passing the diluted clarified extract through a column packed with the ion exchange material. The anion exchange material preferably has a quaternary ethylamino ethyl (QAE) functional group, as exemplified by QAE Zetaprep™ which is supplied commercially in cartridge form. Other known anion exchangers, such as diethylamino ethyl (DEAE) type anion exchange resin may also be employed.

According to one aspect of the method of the invention, it has been discovered that the degree of purification, and particularly the degree of removal of hemagglutinins from the extract, can be enhanced substantially by carrying out the anion exchange binding under low conductivity conditions. Studies presented in Example 2 compare the effectiveness of the anion exchange binding to QAE Zetaprep™ at 20 mM, 40 mM, and 50 mM phosphate buffer, pH 8. The conductivity of the 20 mM buffer is about 3.1 mS/cm, and that of 50 mM buffer, about 6.5 mS/cm.

As discussed in the example, the effectiveness of the low conductivity medium is due in part to precipitation of a protein contaminant having an HPLC peak of about 6.8 min RT. This contaminant does not bind to the anion exchange material, but in precipitated form, is filterable before applying the solution to the anion exchange material. HPLC analysis using TSK250 shows that the 6.8 min RT peak present in the extract is substantially removed after the anion exchange step carried out at low ionic strength.

The low ionic strength also produces enhanced removal of contaminants associated with 9 min and 10.3 min RT peaks (obtained from TSK250 and seen as a combined 9.75 min RT peak on TSK125 in Figure 2A). As noted in Example 2, greater contamination of the flowthrough with the 9 min RT peak occurs at high protein loading and/or at buffer conductivities greater than about 5.5 mS/cm. The amount of 10.3 min RT peak in the flowthrough decreased steadily with decreased buffer concentration, and at 3 mS/cm buffer no 10.3 min RT peak was observed in the. flowthrough.

According to another important feature of the anion exchange step, the low ionic strength eluate containing the desired TCS is substantially free of hemagglutinins. This is seen in Example 3, which describes a hemagglutination study carried out with total extract and three fractions from the initial eluate wash from the anion exchange column, at 100 mM (about 11.92 mS/cm conductivity) and 20 mM buffer concentrations (about 3.1 mS/cm). In each case, three eluate fractions were assayed for hemagglutination activity. As seen from Table 2 in the example, the later eluate fractions (F5 and F9) at 100 mM buffer contained readily detectable levels of hemagglutinins. By contrast, using 20 mM buffer, none of the eluate fractions showed any detectable hemagglutination. The "P.A." fraction shown in the figure is a TCS preparation purified according to prior art methods (Yueng). As seen, the fraction contains readily detectable hemagglutinin contaminants.

To obtain the low conductivity required of the anionexchange buffer, the clarified extract may be diluted with water, preferably with water for injection (WFI) i.e., pyrogen-free water. Alternatively, the extract can be dialyzed against the desired low conductivity buffer.

The flowthrough material is next adjusted to an appropriate pH and conductivity for the cation purification exchange step. The material may also be concentrated to reduce the flowthrough volume. One preferred buffer solution, or application buffer, is 50 mM phosphate buffer, pH 6.

In addition to adjusting the solution for the cation exchange step, the majority of low molecular weight contaminants are preferably removed at this stage by concentration, dialysis and/or diafiltration (ultra-filtration), using membranes with MW cutoffs about a 3,500 to 10,000 daltons. One preferred method of carrying out diafiltration of the material is given in Example 5. As seen from the HPLC profile in Figure 2C, the filtration step has removed virtually all of the low molecular weight contaminants.

In the final purification step, the partially purified material (Figure 2C) is applied to a cation exchange material, in the above application buffer. The anion exchange material is then washed extensively, until the elution profile reaches a baseline value, typically with about 15 to 20 column volumes of buffer. The extensive washing removes loosely bound material, including endotoxins and high molecular weight lipopolysaccharides (LPS), and is necessary for achieving high purity TCS.

TCS is then removed from the column in highly purified form by elution with a buffer whose salt concentration is just sufficient to release bound TCS from the anion exchange material. A preferred elution buffer, for use with SP (sulfopropyl) Zetaprep™ cation exchange material, is 50 mM phosphate buffer, pH 6.0 containing 60 mM NaCl (conductivity about 9.19 mS/cm), as described in Example 5. The eluted TCS protein is concentrated and dialyzed against a desired storage medium. As seen from the HPLC profile in Figure 2D, the eluate protein is substantially free of contaminating proteins.

A variety of cation exchange materials have been tested for their ability to purify TCS from the partially purified extract, and the results are summarized in Example 5. Briefly, each of the five different materials tested resolved TCS from the extract, with highest resolution (TCS purity) being achieved by (a) SP Zetaprep™, applied in 50 mM phosphate, pH 6.0, and eluted with 50 mM phosphate, pH 6.0, containing 60 mM NaCl; (b) Pharmacia S-Sepharose™; applied in 20 mM phosphate, pH 6.0, and eluted with 50 mM phosphate, pH 6.0, containing 60 mM NaC1; and (c) Phoenix SP13™, using the same application and elution buffers as in (b).

The TCS-containing eluent can be concentrated, ultrafiltered and sterile filtered to reduce the volume and to provide a TCS composition in a desired formulation buffer for use.

II. Purified TCS Composition

In another aspect, the present invention includes a substantially pure TCS composition (defined herein as having a purity of at least about 95%). Preferably, the composition has a purity of 98% or greater. The chemical purity of TCS protein produced as above was confirmed by HPLC, SDS gel electrophoresis, and isoelectric focusing methods. Figures 2A-2D are typical size-exclusion HPLC profiles of TCS-containing extract material at various stages of the purification method, showing the increased purity of the 11.3 min RT peak (TSK125) identified with TCS in the total clarified extract (2A), the flowthrough material from the anion exchange anion exchange material (2B), the flowthrough material following diafiltration to remove low-molecular weight contaminants (2C), and the purified TCS (2D) obtained by the final ionexchange step. As seen in Figure 2D, there are no detectable contaminant peaks present in the HPLC profile. The limit of protein detection by this method indicates that the TCS protein preparation is at least about 98% pure. A preparation of purified TCS purified as above was applied, in amounts ranging from 0.25 ug to 40 ug to a sodium doderyl sulfate (SDS) polyacrylamide slab gel. After fractionation by electrophoresis the gel was stained with Coomassie Blue stain.

Figure 3 shows the relative staining of the nine lanes with the different amounts of applied protein. Minor protein contaminant bands are seen in the gel lanes with 40 μg, 30 μg, and 20 μg protein, but not 10 μg or less. The limit of protein detection, as judged by the lowest amount of applied TCS which gave a visible band, is about 0.25 μg. It can be concluded from the gel staining patterns above that 10 μg of the purified TCS protein contains less than about .25 μg of contaminating protein. That is, the TCS is at least about 97.5% pure.

Figure 4 shows the results of a similar gel-stain study in which the above purified TCS preparation was applied to an SDS gel slab at concentrations ranging from 0.005 μg to 10 μg. After electrophoresis, the slab was stained with silver. The limit of detection, as evidenced by the lowest amount of TCS which was detectable was 0.005 μg.

As seen from Figure 4, no detectable contaminant bands were observed at 1 μg protein. Thus a 1 μg sample of purified

TCS contains no more than about 0.005 μg contaminating protein. Judged by this gel-stain result, the TCS protein is at least about 99.5% pure.

Isoelectric focusing of the above purified TCS preparation and of a TCS preparation prepared by a prior-art method (Yueng), and referred to below as "P.A.", was carried out in a PHAST™ system, supplied by Pharmacia, according to manufacturer' s instructions in pH 3-9 gradient gels . The gels were fixed and silver stained.

The positions of the bands observed for the two TCS fractions are indicated in Table 1 below. The pH values shown in the table were calculated from the relationship pH = 9.62 - 0.16 x (distance from cathode in mm). This line equation was obtained from a standard curve (pH 3-10).

The major, band in both TCS preparations is at the top of the gel, as expected, since the isoelectric point of TCS is about 9.6. In the purified TCS fraction, this band represents about 95% of the total protein. In the P.A. preparation, the upper band constitutes a significantly smaller percentage of the preparation. It is not known which, if any, of the contaminant bands in the two TCS preparations represent deamination and/or oxidation products of TCS. Clearly, however, the P.A. preparation contains many more contaminants than the TCS preparation of the present invention.

In another aspect of the invention, there is provided a purified TCS composition which is substantially free of hemagglutinins, e.g., lectins. At least three isolectins have been isolated from an acetone extract of T. kirilowii (Yeung) and studies performed in support of the present invention have identified at least one protein contaminant (a 10.3 min RT peak seen on TSK250) with hemagglutination activity. As noted above, hemagglutinin contaminants present in the original plant extract are substantially completely removed in the anion exchange purification step, when carried at a low conductivity, and preferably at a relatively low flow rate. Additional studies carried out in support of the present invention indicate that the purified TCS preparation herein shows no detectable hemagglutination activity, even at a TCS concentration of about 20 mg/mL.

The TCS composition of the invention also contains a substantially reduced amount of endotoxin. As seen from a comparison of the size-exclusion HPLC profiles in Figure 2B and 2C, virtually all of the high molecular weight contaminants (which include endotoxins) are removed by the cation exchange step. The removal of endotoxins was confirmed by limulus amoebocyte lysate (LAL) assay (U.S. Pharmacopeia). III. Stabilized TCS Composition

The present invention further provides a stabilized TCS composition in which TCS is substantially stable on storage.

In one embodiment, the TCS composition is formulated for stability on storage in solution. Preliminary experi ments conducted in support of the invention, and detailed in Example 6, show that a solution of purified TCS (15 mg/mL) in buffer alone becomes progressively more turbid on storage, as measured, for example, by light scattering at 340 nm. Clarity of trichosanthin solutions indicates a uniformly dissolved protein. Based on studies performed in support of the present invention, the clarity of trichosanthin-containing solutions is an effective indicator of the pharmaceutical value of the solution since the loss of clarity, i.e., increased cloudiness, precedes loss of biological activity. The stabilized trichosanthin solutions (described below) retain clarity, without significant loss of activity, for extended periods (e.g., four days at 4°-8°C -- see below).

In the first study reported in Example 6, increased turbidity was observed for all buffers, independent of pH (Table 3). Further, at pH 6, buffer concentration produced little or no effect on protein stability (Table 4).

Several solution additives were examined for their effect on TCS stability in solution, as reported in Example 6. Of these, proteins such as human serum albumin, and suspending agents, such as mannitol, sorbitol, and polyvinylpyrolidone (PVP) provided only limited protection against protein turbidity on storage (Tables 5 and 6). By contrast, TCS solutions were substantially stabilized by non-ionic surfactants, at a surfactant concentration between about 0.01 to 1 % w/v (Table 7). A preferred class of surfactants is polyoxyethylenesorbitan, as exemplified by the polyoxyethylenesorbitans sold under the trade name Tween™, such as "Tween-20" (mono- laurate), which are available commercially. Similarly, polyethylene and polypropylene glycol gave substantially complete stabilization at selected concentrations above about 5 percent w/v.

Having identified non-ionic surfactants as effective stabilizing agents, combinations of buffer concentration, pH, salt concentration, and suspending agent were tested for their effect on TCS stability. As reported in Example 6, one preferred protein solution contains 5 mg/mL TCS, 50 mM phosphate, pH 6.0, 0.01% Tween-20™, and 2% mannitol w/v, and has an osmolarity close to physiological osmolarity. The formulation was stable against turbidity formation for up to 10 days, when stored at refrigerator temperature. For use as an injectable, the composition is prepared with pyrogen-free components, and filter sterilized. The concentration of the protein in the solution formulation may be 20 mg/mL or more, and diluted if necessary before injection by physiological saline or the like.

In a second embodiment, the purified TCS protein is formulated for stability on lyophilization and reconstitution. Earlier experiments conducted in support of the present invention showed that at least a portion of the protein in buffer was denatured on lyophilization, as evidenced by substantial turbidity when the protein was rehydrated, and loss of protein which is recovered after reconstitution (rehydration). After reconstitution very little of the protein is actually resuspended and most of it remains as insoluble flakes in solution: suspensions such as these are not useful as pharmaceuticals. Several of the agents tested above for increased solution stability were tested for their ability to protect the protein against denaturation on lyophilization and reconstitution.

The effect of several composition variables was examined by a Plackett-Burman design (Motola). The variables which were tested were (a) low concentration phosphate buffer (1 mM phosphate buffer, pH 6.0, vs. WFI); (b) a non-ionic surfactant

(1% Tween-20™ vs. 0.01% Tween-20™); (c) a bulking agent (2% mannitol vs. no mannltol); (d) and a disaccharide (0.5% sucrose vs. no sucrose).

The Tween variable was selected on the basis of the solution turbidity studies above. Mannitol was selected as a sugar which has been employed as a bulking agent and/or cyroprotectant in conjunction with protein lyophilization.

The selection of sucrose was based on the discovery herein that the combination of a non-ionic detergent, such as Tween- 20™ and a disaccharide, such as sucrose, substantially increases recoverable protein, after reconstitution of the lyophilized material. In particular, the percent recovered TCS is substantially greater than that recovered in the presence of surfactant alone.

The make-rup of the test compositions in the Plackett- Burman study is given in Table 9 of Example 7, where X_x - X« are given at the bottom of the table.

Each sample was frozen, lyophilized, stored, and reconstituted as described in Example 7. Turbidity was measured at 340 nm, and the samples were normalized to the most turbid sample, which was assigned a 100% turbidity value. The percent turbidity values are given in the sixth column in Table 9 and are shown in open squares in Figure 5.

The percent protein recovery was determined by filtering the reconstituted samples through a 0.22μ filter and measuring protein concentration at 280 nm. The value obtained was expressed as a percent of the original 280 nm protein concentration in the samples prior to lyophilization. The measured values are .given in the column at the right in Table 9, and are shown as solid diamonds in Figure 5. The average protein recovery effect and average tubidity effect of the four variables were calculated by standard Plackett-Burman analysis (Motola). The calculated values are plotted in Figures 6 and 7, respectively. The two horizontal lines in each figure represent 97.5% confidence limits, calculated from the variables 5-11, which are essentially random combinations of the sets of trial results, also according to standard Plackett-Burman analysis.

As seen in Figure 6, only variable 2 (1% Tween-20™) produced a statistically meaningful (97.5 percent confidence limit) increase in percent recovery.

As seen in Figure 7, variable 2 (1% Tween-20™) produced a statistically meaningful increase in turbidity effect (decreased turbidity) and variable 3 (mannitol) produced a statistically meaningful decrease in turbidity effect (increased turbidity).

The average effect of sucrose and Tween-20™ on turbidity and protein recovery were also determined from the Table 9 data. This was done by averaging the turbidity and protein recovery values for each of the three trials among the 12 trials which have the variable composition of sucrose and

Tween-20™ indicated at the left in Table 10. Thus, for example, to determine the average turbidity value for sucrose

(-), Tween (-) trials, the turbidity values for trials 6, 8 and 12 were averaged.

The calculated average values are seen in Table 10 in Example 7. The average turbidity values indicate that Tween, and Tween plus sucrose both give about the same degree of protection against protein turbidity.

The calculated average values for protein recovery, also given in Table 10, indicate that the combination of Tween plus sucrose provides substantially greater protection against loss of protein than does Tween alone. Further, the data indicates that sucrose alone provides no appreciable protection against loss of protein. Thus, sucrose appears to have an additive effect in combination with Tween which is not observed with sucrose alone. The invention also contemplates other disaccharides, including reducing sugars such as cellobiose, maltose, and lactose, in combination with non-ionic surfactants in a stabilized TCS composition. However, a non-reducing disaccharide, such as sucrose, is preferred, because of the possibility of reaction between a reducing sugar and TCS.

It is also noted that the combination of mannitol and Tween gave only a 95.3% average protein recovery, indicating that the effect of sucrose in combination with Tween is not simply a solute or sugar effect. In fact, of the three trials containing both Tween and sucrose, the lowest protein recovery value was obtained in the mannitol-containing trial. In order to achieve effective resolubilization of trichosanthin it was necessary to include a dissacharide and a non-ionic surfactant in the buffer prior to lyophilization. The reconstituted solutions containing these additives achieve clarity within a minute or so after resuspension and maintain their clarity, for up to 8 hours at 37°C or 24 hours at 24°C.

A preferred lyophilized composition contains, after rehydration with WFI (water for injection), 5 mg/mL TCS, 0.1% Tween-20™ w/v, 5% mannitol w/v (about 50 mg/ml), and 1% sucrose w/v (about 10 mg/ml), in 27.5 mM phosphate buffer, pH 6.2. This composition also forms a preferred TCS composition for a liquid storage form of TCS.

In an ongoing stability study, lyophilized trichosanthin has been prepared in accordance with the present invention and stored at 2°-8°C for over a year. When this trichosanthin is reconstituted, for example using water or sterile saline, the resulting solutions are clear and have anti-viral and protein synthesis inhibitory activities equivalent to freshly purified trichosanthin.

Trichosanthin compositions for clinical studies are prepared in accordance with the present invention, sterile filtered, and dispensed into 10ml vials. These aliquots are lyophilized, stoppered under vacuum and crimp-sealed. The aliquots can be reconstituted using sterile water for injection, sterile normal saline, sterile dextrose 5% in water for injec- tion, or sterile dextrose 5% in normal saline. Since the reconstituted samples are isotonic and can be easily diluted to the appropriate dosage, they are immediately useful for parenteral administration. The compositions of the present invention have proven useful in clinical trials of the drug for several reasons: (i) they provide a means for reliable and accurate preparation of trichosanthin at the desired clinical dose; and (ii) there is no detectable loss of the protein (for example, by the protein sticking to the vial or syringe) during the course of resuspension and administration. This second feature allows for the consistent administration of exact dosages.

As noted above, clarity is an effective indication of pharmaceutical usefulness of trichosanthin-containing solutions. Upon resuspension of a lyophilized trichosanthin sample, clinicians are directed to examine the clarity of the solution. The absorbance at 340 nm of the trichosanthin- containing solution is measured against water. For the trichosanthin solution to be useful the solution should have an absorbance of less than 0.2 absorbance units: the trichosanthin formulations of the present invention have provided outstanding clinical reproducibility in this regard. IV. Utility

A. Abortifacient Activity

As noted above, TCS, or impure plant extracts containing TCS, have been used as an abortifacient agent for inducing abortion in humans, particularly during midtrimester (14 to 26 weeks). The protein can be administered by intramuscular, intravenous, or intraamniotic routes, typically at a single dose of between about 5-12 mg. B. Treating HIV Infection in Humans

The discovery that viral expression and cell proliferation of HIV-infected T cells and monocyte/macrophages can be selectively inhibited by TCS has been described in co-owned U.S. Patent No. 4, 795; 739 for "Method of Inhibiting HIV". An important aspect of the discovery was the finding that viral inhibition in HIV-infected cells could be achieved at concentrations of TCS which are substantially non-toxic to uninfected cells. This selective effect was evidenced by marked decreases in viral antigen associated with HIV-infected cells, and in measurable reverse transcriptase activity, several days after exposure to TCS, without a corresponding decrease in expression of non-viral proteins. Another aspect of the selective inhibitory effect of TCS is a substantial loss of cell viability in HIV-infected cells, at TCS concentrations which did not significantly reduce the viability of non-infected cells. Typically, these selective inhibition effects were seen at TCS concentrations between about 0.01-3.0 ug/ml.

The viral inhibition effects of TCS on HIV-infected T cells and monocyte/macrophages can be exploited in the treatment of HIV infection in humans. The ability of TCS to inhibit viral replication in infected cells, as evidenced by the sharp decline in reverse transcriptase activity associated with the cells, would reduce the level of infection, with the possibility of eliminating the virus "reservoir" which may be provided by the monocyte/ macrophages and other peripheral blood cells (Grove; Gartner, 1986a, 1986b; Koenig; Ho; Chayt; Armstrong; Steicher).

Secondly, recent evidence suggests that the fusion of CD4+ T cells with infected T cells or monocyte/macrophages requires the presence of the HIV gρl20 antigen, on the surface of the infected cells (Lifson, 1986a, 1986b, 1986c). One striking effect of TCS is the ability to inhibit and substantially eliminate expression of viral antigens, such as gp120, in infected cells.

TCS may also inhibit or prevent other events related to the loss of immunological competence in HIV infected individuals, through general suppression of virus levels and inhibition of viral protein synthesis in infected cells.

At the same time, and according to another important aspect of the TCS treatment, the levels of protein which are effective in preventing viral antigen production are only slightly cytotoxic to non-infected blood cells. Effective levels of TCS were in the range 0.2 to 3 μg/mL cell suspension. Assuming a uniform distribution of the drug throughout the bloodstream of a treated individual (7 liter volume), this concentration of drug is achieved with a parenteral injection of between about 1.5 to 20 mg. This is within the same range of TCS used for inducing abortion in humans (about 5-12.5 mg), a general level which can be administered without serious side effects (Kuo-Fen), as discussed above.

From the foregoing, it can be appreciated how various objects and features of the invention are met. The TCS purification method of the invention yields a highly purified TCS protein suitable for intravenous therapeutic administration. In particular, the preparation is substantially free of hemagglutinins and endotoxins.

The method is readily adapted to large scale protein production, since (a) organic-solvent extraction steps with volatile organic solvents are avoided, and (b) the anion and cation exchange steps can be carried out in a batch-wise fashion, without the need for salt-gradient or molecular sieve chromatography separation.

The following examples illustrate preferred methods according to the invention, but are in no way intended to limit the scope of the invention.

EXPERIMENTAL

Materials

T. kirilowii root tubers were obtained from the People's Republic of China.

HPLC chromatography was carried out with either a BioSil TSK125 or TSK250 column obtained from BioRad (Richmond, CA), using isocratic elution with a 1.0 mL/min flow rate and mobile phase of 0.1 M Na₂SO₄, 0.02 M NaH₂PO₄/Na₂HPO₄, pH 6.8 + 0.2.

QAE Zetaprep™ anion exchange cartridges and SP Zetaprep™ cation exchange cartridges were supplied by AMF Cuno Corp. (Meridan, CT); Phoenix SP13 cation exchange resin was obtained from Astec (Whispany, NJ); Pharmacia S-Sepharose from Pharmacia (San Diego, CA); IBF SP Sepherodex M and IBF SP Trisacryl M, from IBF Biotechnics (Savage, MD); Pellicon ultrafiltration membranes (10,000 MW cutoff), from Millipore Corp. (Bedford, MA); 0.45 and 0.22 micron pleated capsule filters from Gelman Corp. (Ann Arbor, MI). A YM5 (5,000 MW cutoff) ultrafiltration membrane was obtained from Amicon/W.R. Grace (Danvers, MA). The following "Tween" surfactants were obtained from Sigma Chemical Co. (St. Louis, MO) : "Tween 20", polyoxyethylenesorbitan monolaurate; "Tween 80", polyoxyethylenesorbitan monooleate; and "Tween 85", polyoxyethylenesorbitan trioleate.

Example 1

Preparation of Trichosanthin

This example illustrates a preferred method of producing purified TCS according to the invention. Novel features of the purification method are described in greater detail in the following examples.

Except where noted, steps of the procedure described below were carried out at room temperature.

Frozen root tubers of T. kirilowii were thawed at room temperature by placing them in water or saline for two to three hours. The tubers were peeled, sliced into small pieces and homogenized in a juicer in normal saline (0.9% NaCl) to form a homogenate. The homogenate was passed through gauze to remove pulp, adjusted to pH 8.0 with NaOH and stirred either for one hour at room temperature or overnight at 2-8°C to form an extraction.

The extraction was clarified by centrifugation at 7,000 x g for 15 min at 4°C and the pellet was discarded. Figure 2A shows a typical size-exclusion HPLC profile (TSK125) of the clarified extraction. The 11.3 min. RT peak (TSK125) in this chromatogram contains the TCS.

The supernatant was purified by elution through a QAE Zetaprep™ anion exchnnge cartridge using 20 mM sodium phosphate buffer, pH 8.0 and the flowthrough was collected. A typical size-exclusion HPLC profile of the flowthrough is shown in Figure 2B. The solution was substantially free of a 9.7 min. RT peak (TSK250) associated with hemagglutination activity.

The flowthrough was clarified by filtration over a 0.45 μm filter pleated capsule filter. The filtered flowthrough was concentrated by ultrafiltration using a Pellicon cartridge membrane with a 10,000 MW cutoff. The concentrated flowthrough was diafiltered using the Pellicon cartridge against 50 mM sodium phosphate, pH 6.0 to adjust the conductivity. The permeate was discarded and the retentate was clarified by filtration through a 0.45 μm filter pleated capsule filter. The retentate was substantially free from small molecular weight contaminants, as shown from the absence of peaks in the range above about 11.3 min. RT (TSK125) in the size-exclusion HPLC shown in Figure 2C.

The following steps were carried out in a controlled environment to minimize the amount of biological contaminants added to the preparation during processing. All of the buffers were filtered using a 0.22 μm pleated capsule filter prior to use.

The filtered retentate from the previous step was loaded onto an SP Zetaprep™ cation exchange cartridge equilibrated with 50 mM phosphate buffer, pH 6.0. After loading, the cartridge was washed extensively with about 15 to about 20 times the column volume of buffer to ensure removal of weakly bound contaminants (associated with endotoxin activity). TCS protein was eluted in 50 mM sodium phosphate buffer pH 6.0, 60 mM NaCl and collected in fractions and stored overnight at 4°C. The TCS-containing fractions were approximately 98% pure, as determined by size-exclusion HPLC, and SDS gel electrophoresis. The eluent is substantially free from endotoxin, as determined by Limulus Amoebocyte Lysate (LAL) analysis. The TCS-containing fractions were pooled and concentrated using a Pellicon cartridge membrane with a 10,000 MW cutoff.

The permeate was discarded. The retentate was diafiltered against a suitable phosphate buffer and filter sterilized by passage through a 0.22 μm membrane.

Example 2

Anion Exchange Procedures

Optimal Ionic Strength

All experiments were run at a flow rate of 2mL/min on a QAE Zetaprep™ 60 disk. Prior to sample loading, the QAE Zetaprep™ cartridge was equilibrated with phosphate buffer having the same conductivity as the sample. Following the procedure, the QAE Zetaprep™ was regenerated with 0.1 M acetic acid followed by l.OM NaC1. All experiments used the same clarified batch extracted in saline. The protein concentration of the clarified extract was 16.9 mg/mL as determined by the Lowry method (Lowry).

Samples (75 or 30 mL) of the extract were diluted with pyrogen-free water to yield a solution of 2.9 mS/cm conductivity at pH 8.0. When the clarified extract was diluted to about 2.9 mS/cm conductivity, a significant amount of precipitate was observed. As determined by HPLC using a Bio-Sil TSK250 column (BioRad, Richmond, CA), the precipitate was the highest molecular weight contaminant (HPLC peak about 6.8 min. RT).

The contaminant having a 6.8 min. RT peak did not bind in any significant amount to the QAE Zetaprep™ when the application buffer was greater than about 40 mM in phosphate, pH 8.0. However, the 6.8 RT peak was decreased significantly in the diluted, 20 mM phosphate buffer, pH 8.0, due to precipitation and subsequent entrapment in the anion exchange material. A 9.0 min. RT peak contaminant bound almost completely to the QAE Zetaprep™ when a 30 mL (507 mg protein) sample was loaded. However, when using a larger sample, the 9.0 min. RT peak appeared in the flowthrough after approximately 500 mg of protein had been loaded. Some of this protein peak was also seen in later fractions when using 40 mM sodium phosphate, pH 8.0. A 10.3 min. RT peak contaminant bound better to the cartridge when the buffer conductivity was below that of 50 mM sodium phosphate. No 10.3 min. RT peak was observed in the flowthrough at 20 mM phosphate, pH 8.0. The low molecular weight proteins were seen in almost all of the flowthrough fractions. The presence of low MW proteins was not of major concern since the subsequent diafiltration step efficiently removed those proteins. These results can be appreciated from the HPLC profile shown in Figure 2B which shows the substantial removal of peaks eluting ahead of the 11.3 min RT TCS peak on TSK125 HPLC.

The results of this study demonstrated that the use of solutions having low ionic strength (low conductivity) improve the effectiveness of the QAE Zetaprep anion exchange step. By diluting the sample to a conductivity approxi-mately of 20mM sodium phosphate, pH 8.0, (about 3.08 mS/cm) the 6.9 RT peak contaminant was removed by precipitation and the 9.0 and 10.3 RT peaks bound more efficiently to the anion exchange material disk.

Example 3

Removal of Hemagglutinins

As noted in Example 2, a low ionic strength medium is required for removal of a 10.3 min RT peak contaminant by anion exchange. The study described below shows that the anion exchange material-binding conditions which are effective in removing the 10.3 min RT peak also effectively remove hemagglutinins from the partially purified flowthrough material.

Clarified extracts prepared as in Example 1 were equilibrated to 100 mM or 20 mM sodium phosphate buffer, pH 6.8. The extract material was added to QAE-anion exchange material equilibrated with the same buffer, i.e., 100 mM or 20 mM sodium phosphate. In the case of protein binding at 100 mM sodium phosphate, flowthrough fractions corresponding approximately to the 6.8 (fraction # 2), 10.35 (fraction # 5) and 17.46 (fraction # 9) min. RT peaks seen in TSK250 HPLC profiles. For the case of binding at 20 mM sodium phosphate, flowthrough fractions corresponding approximately to the 6.8 (fraction # 2), 9.040 (fraction # 4) and 10.03 (fraction # 5) min. RT peaks were collected.

The collected fractions were assayed for hemagglutination activity by a standard hemagglutination test. Briefly, a 50 μl aliquot of each fraction, diluted in 0.9% NaCl as indicated in Table 2 below, was added to about =200 μ1 of a 5% human O-negative erythrocyte suspension centrifuged for 15 sec (Serofuge), and hemagglutination was recorded in a serological scale of 0 to 4+. Ten minutes after incubation at 25 ± 3°C the samples are read again for hemagglutination. The blood samples were examined visually for the presence of detectable agglutination. The tests results are expressed as number of positive agglutinations/total number of assays. For example, a result of 1/3 indicates one positive agglutination in 3 tests.

Also tested were diluted aliquots of the unpurified extract and a TCS preparation purified according to published methods.

The data show that fractions 5 and 9 from the 100 mM sodium phosphate anion exchange material wash have high hemagglutination activity. By contrast, none of the fractions collected in the 20 mM sodium phosphate wash show detectable hemagglutination activity. It is also noteworthy that the TCS protein (P.A.) purified according to a previously published method (Yeung) shows relatively high hemagglutination activity.

Example 4

Concentration and Ultrafiltration

Major low-molecular weight contaminants are removed, in accordance with the purification method, by concentration and/or dialysis (or diafiltration) using membranes with 3,500-

10,000 dalton MW cut-offs. Figure 2B shows the size-exclusion

HPLC profile of plant extract prior to concentration with a PTGC cassette 10,000 MW cut-off ultrafiltration membrane. When the retentate of the concentration procedure was run on size-exclusion HPLC, the profile seen in Figure 2C was obtained. The figure shows that low molecular weight proteins were largely removed, but that high MW proteins, including the TCS protein, were not.

Diafiltration with a 10,000 MW cut off membrane produced a similar reduction in the peaks of lower molecular weight proteins. When the two procedures are combined, the protein peaks eluting after the TCS peak (11.3 min. RT) in the size-exclusion HPLC (TSK125) profiles were virtually eliminated.

Example 5

Cation Exchange Procedures

Cation exchange purification of the above partially purified extract was carried out in the following manner. A protein preparation produced from a plant extract, after anion exchange binding (Example 2) was equilibrated with 50 mM sodium phosphate, pH 6.0 (conductivity 3.6 mS/cm) by diafiltration, using a 10,000 MW cut-off membrane as described in Example 4. Following diafiltration, the material was loaded onto the SP Zetaprep™ column, and the column was washed with 50 mM sodium phosphate, pH 6.0 washed extensively (15-20 volumes) with 50 mM sodium phosphate buffer, pH 6.0, until the chromatogram returned to base line. The TCS was eluted with 50 mM sodium phosphate, 60 mM NaCl (conductivity 9.19 mS/cm). The TCS protein which eluted from the column was examined by size-exclusion HPLC, with the results shown in Figure 2D. Following elution of TCS, the column was stripped with 50 mM sodium phosphate, 1.0 M NaC1, yielding high molecular weiggt contaminants. Several cation exchange materials were investigated for their ability to bind TCS under conditions which allow elution of TCS in pure form. Summarizing the results obtained:

(a) Phoenix SP13 was tested at initial 20 mM or 50 mM sodium phosphate buffer concentrations. The resin showed poor binding of material at 50 mM sodium phosphate buffer, pH 6.0, but gave good separation of TCS at an initial binding at 20 mM buffer, when followed by extensive washing with the same buffer and TCS elution with 50 mM sodium phosphate, 60 mM NaC1, pH 6.0.

(b) Pharmacia S-Sepharose was also tested at initial 20 mM and 50 mM sodium phosphate buffer concentrations. The resin showed strong binding of TCS at 20 mM phosphate buffer, pH 6.0, and release of a substantially pure protein with elution at 50 mM phosphate buffer, 60 mM NaCl, pH 6.0.

(c) IBF SP Spherodex M required higher salt concentration for TCS release.

(d) IBF SP Trisacryl M was tested at initial 50 mM phosphate, pH 6.0, buffer concentration. The resin showed strong binding of TCS at this concentration and eluted from the resin at 50 mM buffer, 60 mM NaC1, pH 6.0.

In summary, all of the cation exchange materials were acceptable substitutes for the SP Zetaprep™ cartridge.

Phoenix SPB and Pharmacia S-Sepharose at 20 mM sodium phosphate, pH 6.0, were most comparable in performance to SP

Zetaprep.

Example 6

Protein Stability in Solution

Stability studies on purified TCS, prepared as above, showed that the protein tends to precipitate on storage over a several-week period at refrigerator (2-8°C) and room tempera tures as evidenced by increased turbidity. Protein preparations containing about 15 mg/mL TCS and formulated to contain one or more of several different buffers, pH conditions, surfactants, solubilizing agents, and suspending agents were stored for up to several days at 2-8°C, then stored for several hours at room temperature. The turbidity of the solution was followed by increased light scattering measured at 340 nm. Turbidity was scored as either "-" (no increase in turbidity) or 1+ to 5+, indicating increasing levels of turbidity. The results are summarized in Tables 3-8 below.

The data in Tables 3 and 4 do not suggest a strong buffer or pH effect on protein stability. Additional studies discussed below indicate that a 50 mM phosphate buffer, pH 6.0 provides enhanced stability in the presence of surfactant and suspending agent components.

The data from Tables 5 and 6 indicate that both human serum albumin (HSA) and various suspending agents, including, mannitol, sorbitol, and polyvinylpyrrolidone (PVP), provide a limited degree of protection against protein denaturation on storage. The effect is not appreciably dependent on concentration of the agent in the range tested.

The data from Table 7 show that non-ionic surfactants, as exemplified by Tween-20™, Tween™-80, and Tween™-85, effectively stabilize TCS on storage. The concentration of surfactant required for optimal or near-optimal effect is between about 0.1 to 1 mg/mL, and decreasing stability may occur at concentrations above about 10 mg/mL.

As seen from Table 8, concentrations of solubilizing agents, as exemplified by polyethylene glycol and propylene glycol, in the range 5-40 percent w/v, provide moderate to strong enhancement of protein stability.

The stability of the purified TCS preparation in four formulations was also examined. These formulations were selected on the basis of stability data presented above and additional studies to confirm optimal stabilizing conditions. The formulations were: (a) 50 mM NaCl, 0.01% Tween-20™, 2% mannitol in 20 mM phosphate buffer, pH 8.0; (b) 50 mM NaCl, 0.01% Tween-20™, 2% mannitol in 50 mM phosphate buffer, pH 6.0; (c) 0.01% Tween-20™, 2% mannitol in 20 mM phosphate buffer, pH 8.0; and (d) 0.01% Tween 20™, 2% mannitol in 50 mM phosphate buffer, pH 6.0. The protein solutions were stored in glass, plastic, or siliconized vessels (coated with Sigmacote™, a siliconizing solution available commercially form Sigma Chemical Company of St. Louis, MO) for up to ten days. The formulations (c and d) which did not contain 50 mM NaCl were more stable than those (a and b) which did. Both in the presence and absence of NaCl, greater stability was seen in 50 mM phosphate, pH 6.0, than in 20 mM phosphate, pH 8.0. The solutions were most stable in plastic containers, measurably less stable in glass containers, and least stable in the siliconized vessels.

Example 7

Lyophilized TCS Composition

Purified TCS prepared as above was brought to a final concentration of about 15 mg/mL in 50 mM phosphate buffer, pH 6.0, or WFI. The buffer solutions were formulated to include Tween-20™, either 1% or 0.01%, mannitol, either 0 or 2%, and sucrose, either 0 or 0.5%, yielding the twelve formulations given below in Table 9. Each formulation was frozen on the lyophilizer shelf, and lyophilized. The lyophilized samples were stored at -20°C for several days, then reconstituted with an original volume of distilled water. After slight agitation to dissolve the dried protein, the samples were examined spectrophotometrically at 340 nm to measure turbidity. The percent turbidity was determined from the average absorbance reading of two identical samples. Protein recovery was determined by filtering the reconstituted material through a 0.22 micron Millex Millipore low-protein binding filter. The filtrate was read at 280 nm, and the percent protein recovery was determined against the known amount of material (280 nm absorbance) in the sample prior to lyophilization.

The data from Table 9 (plotted in Figure 5) were analysed by standard Plackett-Burman analysis, as described above, to give the percent protein recovery effect and percent turbidity effect plotted in Figures 6 and 7, respectively. The data in these figures show than Tween-20™ (variable 2) is the only variable which, by itself, produces a statistically meaningful increase in protein recovery (Figure 6) and decrease in turbidity.

The combined effect of Tween and sucrose was determined by averaging protein recovery and turbidity values for each of the three trials in Table 9 containing the combination of variables given below in Table 10.

The data show that (a) sucrose alone has little or no effect on protein recovery, and (b) the combination of Tween and sucrose provides substantially greater protein recovery that Tween alone.

While the invention has been described with reference to specific methods and embodiments, it will be appredated that various modifications and changes may be made without departing from the invention.

Claims

IT IS CLAIMED:

1. A method of producing trichosanthin, comprising:

(a) preparing a single-aqueous phase clarified plant extract from the roots of Trichosanthes kiri- lowii; and

(b) contacting the extract with an anion exchange material, under conditions of low conductivity which extract hemagglutinin contaminants from the extract, and

a cation exchange material, to separate trichosanthin from contaminants which bind more weakly to the cation exchange material than trichosanthin, at a relatively low ionic strength, and from contaminants which bind more strongly to the cation exchange material at a relatively high ionic strength;

(c) isolating substantially pure trichosantin.

2. The method of claim 1, wherein said anion exchange material has a quaternary ethylamino ethyl (QAE) functional group, and the extract contacted with the anion exchange material has a conductivity of less than about 3- 3.5 mS/cm and a pH of between about 7-9.

3. The method of claim 1, wherein the cation exchange material has a sulfopropyl (SP) functional group.

4. The method of claim 3, wherein said contacting with the cation exchange material includes binding extract material with the cation exchange material at a conductivity of about 3-4 mS/cm, washing the cation exchange material extensively to remove weakly bound contaminants, and releasing trichosanthin from the cation exchange material by washing with a solution of a conductivity of about 9.0 ms/cm.

5. The method of claim 1, which further includes removing small molecular weight contaminants by membrane ultrafiltration or dialysis.

6. The method of claim 1, wherein the trichosanthin is substantially free of hemagglutinins.

7. The method of claim 1, wherein the protein purity of trichosanthin is greater than 98%.

8. A method of producing trichosanthin, comprising:

(i) preparing a single-aqueous phase clarified plant extract from the roots of Trichosanthes kirilowii, said extract having a conductivity of less than about 3.5 mS/cm;

(ii) contacting the extract with an anion exchange material, having quaternary ethylamino ethyl (QAE) functional groups, to produce a substantially hemagglutinin-free eluent;

(iii) treating said eluent by membrane ultra-filtration or dialysis to produce a retentate that is substantially free of contaminants having molecular weights less than about 10,000 daltons;

(iv) contacting said retentate with a cation exchange material, having sulfopropyl functional groups, at a conductivity of between about 3-4 ms/cm;

(iv) washing the cationic exchange material extensively to remove weakly bound contaminants;

(vi) treating the cationic exchange material with a solution whose ionic strength is just sufficient to release trichosanthin from the cationic exchange material; and

(vii) collecting the released trichosanthin having a purity of greater than about 95% and which is substantially free of hemagglutinins.

9. The method of claim 8, wherein the trichosanthin is released from the cation exchange material with a buffer having a conductivity of about 9.0 mS/cm.

10. The method of claim 8, wherein the trichosanthin produced has a purity of at least about 98%.

11. The method of claim 8, which further comprises adding between about 0.01 to 1.0 % w/v non-ionic surfactant to an aqueous solution of the released trichosanthin.

12. The method of claim 11, which further includes adding between about 0.5-5% disaccharide to the aqueous solution, lyophilizing said solution for storage, and reconstituting the lyophilized material.

13. The method of claim 12, wherein said disaccharide is sucrose.

14. The method of claim 13, wherein the surfactant is a polyoxyethylenesorbitan, at a concentration before lyophilization of between about 0.01 and 1.0 % w/v, and the sucrose is at a final concentration prior to lyophilization of about 0.5-5% w/v.

15. A trichosanthin composition prepared from the roots of Trichosanthes kirilowii

(a) having a trichosanthin purity of greater than 95%, and said trichosanthin having a pi of about 9.6;

(b) being substantially free of hemagglutination activity as determined by the composition's inability to cause agglutination of a human O-negative erythrocyte suspension; and

(c) being substantially free of endotoxin activities as determined by the limulus amoebocyte lysate assay.

16. The composition of claim 15, produced by

(i) preparing a single-aqueous phase clarified plant extract from the roots of Trichosanthes kirilowii, said extract having a conductivity of less than about 3.5 mS/cm;

(ii) contacting the extract with an anion exchange material, having quaternary ethylamino ethyl (QAE) functional groups, to remove substantially all hemagglutination activity from the eluent; (iii) treating said eluent by membrane ultra-filtration or dialysis to yield a retentate that is substantially free of contaminants having molecular weights less than about 10,000 daltons;

(iv). contacting said retentate with a cation exchange material, having sulfopropyl functional groups, at a relatively low ionic strength;

(iv) washing the cationic exchange material extensively to remove weakly bound contaminants;

(vi) treating the cationic exchange material with a solution whose ionic strength is just sufficient to release trichosanthin from the cationic exchange material; and

(vii) collecting the released substantially pure trichosanthin.

17. The composition of claim 15, having a trichosanthin purity of greater than 98%.

18. The composition of claim 15, which further comprises a biocompatible, non-ionic surfactant chosen from the group consisting of polyoxyethylenesorbitans; and,

said composition characterized by substantially complete trichosanthin solubility, when stored in an aqueous form over a four-day period at 2-8°C.

19. The composition of claim 18, wherein the concentration of the surfactant is between about 0.01 to 1.0 % w/v in said aqueous form.

20. The composition of claim 19, wherein the surfactant is a polyoxyethylenesorbitan.

21. The composition of claim 18, wherein the surfactant is selected from the group consisting of the following surfactants: "Tween-20", polyoxyethylenesorbitan monolaurate; "Tween 80", polyoxyethylenesorbitan monooleate; and "Tween 85", polyoxyethylenesorbitan trioleate.

22. The composition of claim 18, which further includes a disaccharide and is further characterized by substantially complete protein resolubilization, when reconstituted from a lyophilized form to an aqueous form.

23. The composition of claim 22, wherein the disaccharide is sucrose, at a concentration between about 0.5-5% w/v in said aqueous form.

24. The composition of claim 23, which further ineludes mannitol, at a concentration of between about 1-10% w/v in said aqueous form.

25. The composition of claim 18, wherein the protein purity of. trichosanthin is at least about 98%.

26. The composition of claim 18, wherein the concentration of trichosanthin is between about 1-10 mg/mL in an aqueous form.

27. The composition of claim 22, wherein said surfactant is "Tween 20", polyoxyethylenesorbitan monolaurate; at a weight ratio of trichosanthin to "Tween 20", polyoxyethylenesorbitan monolaurate, of about 5:1, and said disaccharide is sucrose at a weight ratio of trichosantin to sucrose of about 1:2.

28. The composition of claim 27, which further includes mannitol, at the weight ratio of trichosanthin to mannitol of about 1:10.

29. A sterile, injectable solution of trichosanthin, prepared from the roots of Trichosanthes kirilowii, which is characterized by:

(a) said trichosanthin having a trichosanthin purity of greater than 98%, and said trichosanthin having a pI of about 9.6;

(b) said solution being substantially free of hemagglutination activity as determined by the solution's inability to cause agglutination of a human O-negative erythrocyte suspension;

(c) said solution being substantially free of endotoxin activities as determined by the limulus amoebocyte lysate assay;

(d) said solution containing a biocompatible, non-ionic surfactant chosen from the group consisting of polyoxyethylenesorbitans; and,

said solution characterized by substantially complete trichosanthin solubility, when stored in an aqueous form over a four-day period at 2-8°C.

30. The solution of claim 29, wherein said surfactant is "Tween 20", polyoxyethylenesorbitan monolaurate, at a concentration of between about 0.1 and 10 mg/ml, and sucrose or mannitol, at a concentration of between about 2- 10%.

31. A lyophilized trichosanthin preparation, which when reconstituted in an aqueous solution:

(a) has a trichosanthin purity of greater than

98%, and said trichosanthin has a pi of about 9.6;

(b) is substantially free of hemagglutination activity as determined by the preparations' s inability to cause agglutination of a human O-negative erythrocyte suspension;

(c) is substantially free of endotoxin activities as determined by the limulus amoebocyte lysate assay;

(d) contains a biocompatible, non-ionic surfactant chosen from the group consisting of polyoxyethylene- sorbitans; and,

said solution is characterized by substantially complete trichosanthin solubility, when stored in an aqueous form over a four-day period at 2-8°C.