EXTRACTION OF ALKALOIDS

Technical Field

This invention relates to alkaloids produced by the Catharanthus roseus plant.

Background Art The low levels at which vinblastine and vincristine are produced naturally by Catharanthus roseus has prompted researchers to investigate the feasibility of using in vitro techniques to produce these pharmaceutically active compounds.

As a result of research in various areas, it is now generally understood that the monomeric alkaloids catharanthine and vindoline are the components which couple, in vivo, to form an intermediate compound 3' ,4'- anhydrovinblastine which is converted ultimately to form vinblastine. In turn, vincristine is generated from vinblastine. The overall reaction scheme can thus be represented as follows:

vindoline + catharanthine 3' ,4'-anhydrovinblastine vinblastine vincristine

To permit economical in. vitro production of vinblastine and vincristine, efficient production of 3' ,4'-anhydrovinblastine (hereinafter referred to as AVLB) is necessary. The present invention is therefore concerned with a method of recovering AVLB for use particularly in the application of in vitro methods for producing the more valuable dimeric alkaloids vinblastine and vincristine.

Canadian Patent 1,094,552 issued January 27, 1981

describes a process for isolating vindoline, catharanthine and AVLB which comprises extracting dried leaves with organic solvent such as methanol or toluene or a mixture thereof and aqueous acid solution, purifying the extract using a phase-change method and then precipitating with sulphate addition. Thereafter, individual alkaloids or groups thereof are isolated from the mother liquor using chromatography, solvent gradients and/or pH differentials.

In Physiol. Veg, 1985, 23 _^ (A) , 381-388 Renaudin describes an alkaloid extraction process in which suspension cells of Catharanthus roseus are extracted with 0.01% acetic acid, and collected in organic solvent fractions using a reverse phase cartridge after the pH of the acid extract is raised to pH 7.3-7.5 by NaOH addition. This particular extraction method is particularly suited to assay for alkaloid content using HPLC with fluorimetric detection.

Those familiar with alkaloid extraction techniques will appreciate that extraction processes vary depending on the chemical nature of the compound to be extracted. In this regard, AVLB demands special consideration given its propensity for oxidation to less important or less valuable alkaloids such as leurosine. Use of dried leaves in an extraction process as disclosed for example in the Canadian patent cited above is undesirable when AVLB is selected for recovery since the drying process is oxidative in nature and could reduce this available AVLB in the starting tissue.

In this same vein, pH elevation of an AVLB crude extract, as taught by Renaudin, such as by addition of sodium hydroxide may also contribute to AVLB oxidation and therefore reduce the AVLB available for recovery.

The oxidation of AVLB by prior art processes is reflected perhaps by the very modest yield of AVLB extracted

by the process disclosed in the Canadian patent (0.145 grams from 1 kg dried leaves) . Accordingly, it is an object of the present invention to provide a method suitable for extracting AVLB from C^ roseus tissue.

Disclosure of Invention

As initial extraction agent, the present invention utilizes an aqueous medium of acidic pH. Use of aqueous, acidic medium rather than organic solvent in the initial extraction procedμre offers several advantages. Primarily, the aqueous medium is selective in the sense that AVLB and other basic alkaloids are collected in the initial step rather than with a larger number of plant products in an organic solvent. In addition, cell debris including chloroplasts recovered in the aqueous medium can be removed by simple centrifugation thereby removing chlorophyll which otherwise could contribute to AVLB oxidation and may inter¬ fere with purification process. Removing chlorophyll from organic solvent requires a more complex technique.

After separating cell debris from soluble components such as by centrifugation, the soluble component is extracted with organic solvent. In accordance with a preferred embodiment, a reducing agent, preferably sodium borohydride, is added to the soluble component prior to organic solvent extraction. Addition of a reducing agent at this stage of the extraction procedure results in enhanced AVLB yield possibly by compensating oxidative agents or by converting oxidized AVLB derivatives to AVLB i.e. the iminium product of catharanthine and vindoline coupling is reduced by sodium borohydride to AVLB.

Once extracted with organic solvent, AVLB can be purified according to techniques standard in the art e.g. chromatographically, using selective crystallization etc.

In accordance with an embodiment of the invention preferred herein, the C^_ roseus tissue is digested with enzyme to disrupt cell walls and release alkaloids to the extent possible.

It should be noted that the present invention avoids the use of oxidative agents wherever possible. Accordingly, it is of the utmost preference herein to conduct the extraction process under inert atmosphere where possible, although significant AVLB yields can be obtained in normal atmosphere.

Thus, according to one aspect of the present invention there is provided a process for extracting AVLB from C^_ roseus tissue which comprises a first step in which the tissue is extracted in aqueous medium of acidic pH and a second step in which the soluble component is extracted with organic solvent.

Best Mode for Carrying Out the Invention

The extraction process may be conducted on fresh e.g. recently grown, or frozen C_-_ roseus tissue. Tissue which has been dried is preferably avoided since the drying process is believed to have the effect of oxidizing AVLB in the tissue and therefore lowers the amount of AVLB available in dried tissue.

The fresh or frozen tissue is preferably minced prior to the first step in the extraction in order to expose cellular tissue. Fresh tissue may, for example, be homogenized in acid in a standard blender. Frozen tissue is suitably ground in liquid nitrogen. Since, however, it is always the intention herein to process the C_-_ roseus tissue under non-oxidative conditions, it is most suitable to use grinding under inert atmosphere, to prepare the tissue for extraction, whether fresh or frozen tissue is selected as

starting material.

It is believed that AVLB associates with the cell wall matrix possibly in the plant tissue or when cells are disrupted and contents mixed. It is possible therefore that while grinding of the tissue exposes a fraction of the AVLB to the initial extraction medium, a remaining fraction of the AVLB is still bound and will escape extraction. Accordingly, it is preferred herein to expose either the unground tissue or the ground tissue to enzymes capable of digesting the cell wall components, such as cellulase, peptinase, xylanase and laminarase, before the extraction is carried out. Commercial preparations useful for this purpose are abundant and include, for example Macerozyme and Driselase (both products of Yakult Honsha Co. Ltd., Japan).

Once prepared, the tissue is first Extracted with an aqueous medium having an acidic pH. Water per se may be used e.g. tap water, distilled or double distilled water, and may be preferred to keep costs at a minimum. However, experimentation reveals that lowering of the pH by adding acid to the aqueous medium can enhance AVLB extraction. For example AVLB yield when aqueous medium at around pH 2 is used has been found to be greater than the corresponding yield at around pH 6, all other conditions being equal. To lower the pH, any acid may be used, including organic acids such as acetic acid and inorganic acids such as sulfuric acid. Mineral acids such as hydrochloric acid are preferred and may be mixed with water to attain the desired pH prior to extracting the tissue therewith. The resulting pH of the aqueous medium may be within the range from 1.5 to 7.

The pH of the aqueous medium will become more basic after the C_^ roseus tissue is suspended therein. The pH elevation caused by tissue addition will vary, depending upon the amount and constitution of the tissue although a

rise of from 1 to 2 pH units can be expected under typical conditions. However, provided that the initial pH of the aqueous medium is acidic, yield of AVLB is satisfactory. Preferably, the pH of the medium after tissue addition remains acidic, for example from pH 3.5 to pH 6.5, in order to obtain preferred yields. This can be accomplished by acidifying the aqueous medium, before tissue exposure, to generate an initial pH of about 1 or 2 pH units below the pH desired after tissue extraction. Alternatively, dilute acid may be added directly to the tissue extract, if necessary. What is important to bear in mind is that base addition to correct pH should be avoided since this will impact on the oxidative state of AVLB and could lower yield.

While the effect of the acid in the extraction process is not completely understood, it is believed that the acid acts by dissociating AVLB from various macromolecules such as polysaccharides, proteins, polyphenols etc. A similar effect appears to occur when salt, rather than acid, is added ^• initially to the aqueous medium in accordance with an alternative embodiment of the present invention. Although no acid is added, the salted aqueous medium is still considered herein to be an aqueous medium of acidic pH since addition of salt to water per se for example, still results in a medium having a pH of 7 or lower. The salts which are suitable for addition to the water include highly ionizable salts lsuch as ammonium sulphate, potassium chloride and especially sodium chloride.. Salt concentration in the water can be in the range from 0.01 - 5.0 M although, particularly in the case when sodium chloride is used, molar concentrations in the range from 0.9 - 3.0 M are preferred. The salt can be added directly to water (about pH 6.0) and used as an extraction medium when the molar salt concentration is as desired.

In an additional embodiment of this invention,

hydrogen peroxide is added to the acidic aqueous extract of C. roseus tissue to enhance the AVLB yields. Preferably, the amount of hydrogen peroxide added is above 8 mM, most preferably, it is 20 mM.

Once the ground and enzyme-digested tissue is suspended in the aqueous medium, whether water per se, acidified water or salted water is used, preferably with agitation, and optinally mixed with hydrogen peroxide, the soluble component containing the alkaloids is separated from cell debris i.e. the insoluble component for example by filtration or, more preferably, by centrifugation. Aqueous extraction of the insoluble component may be repeated to recover additional alkaloid in a soluble com- ponent.which may then be pooled with previously collected soluble components.

Prior to extracting with organic solvent, it is in accordance with a preferred embodiment herein that a reducing agent is added to the soluble component. In this respect, sodium borohydride is particularly suitable in amounts sufficient, for example to establish a concentration in the soluble component in the range from about 0.05 mg/mL to about 4.0 mg/mL, more preferably in the range from 0.1 to 3.0 mg/mL. The reducing agent is believed to enhance AVLB extraction yield either by reducing an iminium analogue of AVLB or by hindering oxidation of AVLB in the soluble component. Best results are obtained after separation of soluble component from insoluble component. Addition of reducing agent prior to separation i.e. during extraction appears to reduce AVLB levels in the final extract.

The aqueous medium into which the alkaloids are initially extracted may be frozen, if desired, for at least as long as about three weeks before subsequent extraction with organic solvent.

Extraction of the soluble aqueous component with organic solvent, the second extraction step, is preferably conducted in a two phase system using, for example, an organic solvent selected from benzene, toluene and ethyl acetate of which the use of ethyl acetate is preferred. A series of organic extractions may be carried out on the same aqueous extract and the organic fractions pooled for further processing if desired.

Purification of the AVLB contained in the organic fraction may be carried out using techniques which are now standard in the art such as chromatography, crystallization etc. The extract which results from the extraction process described herein will be rich in the monomers catharanthine and vindoline. Detectable amounts of vinblastine are also present. Of particular relevance hereto is the presence, in the extract, of an enhanced proportion of AVLB which with extraction procedures preferred herein, can approach 0.2% by dry weight of tissue. This can be compared with the accepted values of about 0.0003% by dry weight for the alkaloids vinblastine and vincristine which can be prepared using AVLB as a precursor.

Thus, by providing a method for enhancing the extracted yield of AVLB, processes which utilize this pre¬ cursor can become more economically attractive.

Embodiments of the invention are described here- inbelow by way of example only.

Example 1 - Broad Spectrum Analysis pH Effect

To a suspension medium consisting of 7.5ml water (pH 5.9) and 1 gram NaCl was added 2.5 grams fresh weight of C_^_ roseus leaf powder prepared by grinding the leaves in liquid nitrogen. The initial pH of the NaCl-containing suspension medium was acidified to pH 2.0 in one trial,

basified to pH 9.0 in another trial and was unaltered (pH 6.2) in a third trial, each trial being run in duplicates. The samples were sonicated for 10 minutes and then centrifuged for 30 min. at 23,000 g. The supernatant was removed and extracted 3 times with equal volumes of ethyl acetate. This was dried down and the residue was taken up in methanol for HPLC analysis. HPLC was performed on a C-8 reverse phase 5um column, using a solvent gradient of methanol and water with t-butyl ammonium phosphate as modifier.

Results;

Yield of AVLB

Initial pH (% of dry wt)

2.0 0.099

6.2 0.066

9.0 0.000

An acidic pH therefore enhances the yield of anhydrovinblastine. ^***

Example 2 - Narrow Spectrum Analysis of pH Effect

Duplicate samples of 1 gram fresh weight leaf powder as described in Example 1 were extracted initially with 3ml HC1 at various concentrations followed by cen- trifugation and organic extraction as outlined in Example 1. The results, analyzed by TLC are presented below.- The pH values are given both for the pH prior to leaf powder addition and for the pH after addition of leaf powder.

Results :

HC1 Cone. P ^H AVLB (% dry weight)

(M) Before After

1.00 nm nm 0

0.50 0.68 0.85 0

0.10 1.26 2.28 0

0.005 1.53 3.62 0.04

0.025 2.05 4.52 0.122

OcOlO 2.37 5.18 0.101

0.005 2.59 5.55 0.078

0.0011 3.18 6.05 0.065

From these results, it is evident that an initial suspension medium pH of about 1.5 is required in order to recover anhydrovinblastine. This translates to a minimum suspension medium pH, after leaf powder addition of about 3.5. It will fee further noted that a pH of from about 4 to 6 in the leaf-suspended medium provides desirable AVLB yields.

In a separate experiment, the effects of various acids on yield of AVLB were analyzed under conditions as described above. The results appear below:

Acid Cone PH AVLB (M) Before After

HC1 0.025 2.05 4.52 0.122

0.010 2.37 5.18 0.101

0.005 2.59 5.55 0.078

0.0011 3.18 6.05 0.065

H ₂ S0 ₄ 0.010 2.16 4.67 0.100

0.001 2.92 5.82 0.062

CH-.COOH 0.010 3.56 5.36 0.091

0.001 4.20 5.95 0.067

Evidently, a variety of different acids can be used successfully to control pH in the reaction medium without adversely affecting yield.

Example 3 - Sodium Chloride

When leaves were extracted by the same method described in Example 1, using water at normal pH (c. 5.9), the levels of anhydrovinblastine extracted could be increased by having sodium chloride in the water.

Results

Cone. of NaCl (M) AVLB Yield (% of dry wt)

0.00 0.041

0.23 0.049

0.91 0.059

1.14 0.050

2.28 0.066

4.56 0.021

Optimal yields of AVLB were obtained with Ig NaCl in 7.5ml water (i.e. at a concentrations of 2.28M) when there was a 61% increase above samples extracted in the absence of sodium chloride.

When dilute acid was used to extract the AVLB, as described in Example 1, no further increase occurred when sodium chloride was added.

NaCl Cone. (M) AVLB Yield (% of dry wt)

0.00 0.104

1.14 0.105

1.71 0.095

2.28 0.079

2.85 0.078

These results indicate that the lower pH may also cause some form of dissociation of the AVLB such that sodium chloride has no additional effect. The lower _?H may also impart greater stability to the AVLB. The fall in AVLB yields as the NaCl-concentration increases may be due to a salting out of the less soluble AVLB.

Example 4 - Sodium Borohydride Addition Leaves were ground in liquid nitrogen and 1 g portions were mixed in 3 ml 0.025 M HC1. 0.1 M magnesium

_3 chloride and 3 x 10 % hydrogen peroxide were both added to this mixture. After centrifugation, the supernatant was removed and sodium borohydride was added such that a range of final concentrations from 0 to 3.33 mg/ml could be tested.

Results of HPLC analysis for AVLB are given below:

__>_l_!._._—11 UUJ._iyUJ._ U_ J- -liO. _ AVLB Yield

Cone, (mg/ml) (% of dry weight)

0 .102

0.067 .161

0.167 .172

0.333 .171

0.666 .207

1.667 .164

3.333 .167

Thus, addition of sodium borohydride to the acidic extract will enhance the yields of AVLB. This must be done after the centrifugation step - if aBH4 is added prior to centrifugation, no AVLB can be extracted.

Magnesium chloride and hydrogen peroxide were subsequently found to be unnecessary for the borohydride effect.

Example 5 - Spectral Analysis

Duplicate lg fresh weight samples of leaf powder (prepared by grinding in liquid nitrogen) were each mixed in 3ml of 0.03M HC1 containing 0.4g sodium chloride. These were centrifuged at 23,000g for 30 min. and the supernatant was extracted with ethyl acetate twice. The ethyl acetate was dried down and the remaining residue was taken up in 500 ul methanol for analysis by TLC and HPLC.

HPLC: Samples were analyzed on a C-8 HPLC column as previously described. 3' ,4'-anhydrovinblastine was identified by its retention time, UV spectrum and 1st derivative of the spectrum, all of which correspond with those of an authentic standard. The yield of AVLB was calculated as 0.089% of the dry weight. Vindoline, catharanthine and trace amounts of leurosine were also identified by their retention times and spectra. The calculated yields of these three alkaloids are

given below:

% Dry Weight Leurosine 0.001 Vindoline 0.148

Catharanthine 0.142

TLC: Normal phase thin layer chromatography was carried out, using silica gel and a mobile phase of toluence : acetone : methanol : ammonium hydroxide (28:10: % .5) .

Anhydrovinblastine, catharanthine and vindoline were all identified by their Rf values and UV spectra which corre¬ sponded with those of authentic standards. In addition, the spots eluted by TLC were sprayed with eerie ammonium sulphate spray, revealing the characteristic colour reactions for all three alkaloids.

Rf Values on TLC

Anhydro-VLB Catharanthine Vindoline

Sample .37 .59 .45

Standard .37 .59 .44

Example 6 - Sonication Effect 2o5g samples of fresh J.eaf powder were extracted as described in Example 1 at acidic pH using dilute HC1 (0.03M). Duplicate incubations were sonicated for various time intervals, after which they were filtered through miracloth. The filtrate was centrifuged and the resultant supernatant extracted with ethyl acetate as before.

Results

Sonication

Time (in minutes) Yield of AVLB (% of Dry Wt)

0 0.104

15 0.094

30 0.090

60 0.067

These results show that sonication is unnecessary and that if such incubation is prolonged there is actually a decrease in the yield of AVLB possibly due to aerial oxidation. Accordingly, sonication is preferably avoided in an AVLB extraction process.

Example 7 - Enzyme Digestion of Leaf Material

Powdered leaf material samples (1 gram fresh weight) were incubated with various enzyme preparations at room temperatures for one hour. Double-distilled water was used as the extractive aqueous medium which was centrifuged and extracted with organic solvent as usual. The results appear below:

Enzyme Cone. AVLB Alkaloid Cone. (% of dry wt.

Catharanthine Vindoline Macerozyme 0.2% .045 .122 .393

(a crude peptinase) 0.4% .073 .163 .382

Driselase 2.0% 049 .126 .328

(crude preparation 4.0% 039 .128 .342 of lamarinase, xylanase and cellulose)

Beta-Glucosidase 10 units .020 .068 .260 20 units .008 .044 .204

Control .010 .054 .209

The enhanced alkaloid yield which results when Macerozyme and Driselase (both products of Yakult Honsah Co. Ltd. , Japan) are used is a result most likely of the release of alkaloids from the cell wall matrix of the C _ roseus tissue.

Example 8 - Enzyme Digestion and Acidification

Duplicate 1 g portions of leaves (ground in liquid nitrogen) were either:

1. incubated with 0.4% macerozyme in 3 ml water for 1 hour, followed by centrifugation and extraction of the supernatant as normal, or

2. 0.1 M HC1 was added to the mixture of leaves and water (3 ml) to give final pH of 4.5 which was centrifuged and extracted as normal, or

3. Enzyme incubation was performed as in 1, followed by acid addition as in 2. - centrifuged and extracted as normal.

Alkaloid content (mg)

Treatment AVLB Catharanthine Vindoline .4% macerozyme 93.1 433.5 323.1

Acid 90.6 402.6 307.2

.4% macerozyme + acid 132.9 428.8 328.1

Thus, a combination of enzyme treatment followed by acidification enhanced the yield of AVLB by 43% over enzyme treatment per se.

Example 9 - Mass Spectrophotometry

A sample of AVLB extracted from leaves was purified by collection of the appropriate fraction off the HPLC. The

normal HPLC gradient was run with methanol, water and 0.1% triethylamine. The fraction collected was then dried down and analyzed by electron impact mass spectrophotometry. The fragmentation pattern obtained was the same as that for an authentic sample of 3' ,4'-anhydrovinblastine. High resolu¬ tion data showed that it had a mass of 792.4107 units (± 0.9% with a deviation of 0.9 milli mass units from theoretical value) .

Example 10 - Identification of Vinblastine

100 g (fresh weight) of leaves were ground in 300 ml of 0.025 M HC1 in a Waring blender. Debris was removed by centrifugation and the supernatant was extracted with ethyl acetate. The ethyl acetate was evaporated off and taken up in a small volume of methanol which was streaked on a preparative TLC plate, using toluene : acetone : methanol : NH4OH (28:10:2:0.5). A band with the same Rf value as a VLB standard was detected and scraped off the plate. This was extracted from the silica using methanol : dichloromethane (2:1) with 1.55 triethylamine. The solvent was then removed by evaporation and the remaining alkaloid was repurified on a preparative TLC plate using diethyl ether : chloroform : methanol (50:35:20). A band with the same Rf value as a VLB standard was removed and extracted as before. The solvent was evaporated off and the remaining alkaloid was dissolved in a small amount of dichloromethane, filtered and then dried again. The alkaloid was analyzed with desorption electron impact mass spectrophotometry and an ion corre¬ sponding to VLB was observed. High resolution data gave an accurate mass that was within 0.8 milli mass units of the theoretical value for vinblastine.

Example 11 - Effect of Sodium Hydroxide and Ammonium Hydroxide Addition 1 g portions of leaves (ground in liquid nitrogen) were extracted with 0.025 M HC1 in the normal manner. After

centrifugation, the pH of the supernatant was increased to 7.3 or 9.0 with either NaOH or NH4OH. In addition, leaves suspended in acid (0.025 M HCl) were treated with 0.5 g NaCl and after centrifugation the pH was increased to 7.3 with either NH4OH or NaOH. Appropriate controls (i.e. without base) were also run.

Results (from HPLC Analysis)

E__ AVLB (% dry weight)

Controls .124

+ NH4OH 7.3 .116

+ NaOH 7.3 .090

+ NH4OH 9.0 .076

+ NaOH 9.0 .084

NaCl (0.5 g) .089

NaCl + NH4OH 7.3 .070

NaCl + NaOH 7.3 .068

Thus, when AVLB is to be extracted it is preferred not to increase the pH of the aqueous medium prior to two phase extraction.

Example 12 - E fect of Hydrogen Peroxide Addition 1 g portions of leaf powder were mixed in with 3 ml of 0.025 M HCl and a range of H2O2 concentrations were added. After centrifugation, sodium borohydride was added to supernatant at a final concentration of 0.67 mg/ml, and extraction with ethyl acetate was performed as usual.

Results (from HPLC Analysis)

H2O2 Cone. AVLB

(mm) (ug)

0 372

1 359

2 360

4 375

8 445

12 461

20 444

Thus, the addition of H2O2 at a concentration of above 8 mM serves to enhance the yields of AVLB from leaves.