Enzymatic Modification of Natural Compounds with Pharmacological Properties

PDFlib PLOP: PDF Linearization, Optimization, Protection

Page inserted by evaluation versionwww.pdflib.com – [email protected]

Enzymatic Modification of NaturalCompounds with Pharmacological Propertiesa

SERGIO RIVA,b DANIELI MONTI,b MONICA LUISETTI,c

AND BRUNO DANIELIc

bIstituto di Chimica degli OrmoniCNR

20131 Milano, Italy

cDipartimento di Chimica Organica ed Industrialeand

Centro CNR di Studio sulle Sostanze Organiche NaturaliUniversità di Milano20133 Milano, Italy

ABSTRACT: Glycosides of various classes of natural products are widely distrib-uted in nature, where they are often present esterified with aliphatic and aromaticacids at specific OH’s of their sugar moieties.

Many of these compounds are pharmacologically important molecules or possessother interesting properties. For instance, ginsenosides (e.g., 3) are therapeuticdammarane-type oligoglycosides isolated from the water-soluble portion of thedried roots and leaves of Panax ginseng C.A. Meyer (Araliaceae), a plant widelyused in traditional Chinese medicine.

In recent years, we have exploited the regioselectivity of lipases and proteases inorganic solvents for the synthesis of specific esters of ginsenosides as well as the se-lectivity of the �-1,4-galactosyltransferase from bovine colostrum to obtain new gly-cosyl derivatives of these compounds.

The application of these two enzymatic methodologies has also been exemplifiedwith other natural compounds with pharmacological properties: digitonin (5),colchicoside (6), and flavonoid glycosides.

INTRODUCTION

Glycosides of various classes of natural products are widely distributed in nature,where they are often present esterified with aliphatic and aromatic acids (mainlyacetic, malonic, p-coumaric, and ferulic) at specific OH’s of their sugar moieties.Many of these compounds are pharmacologically important molecules or possessother interesting properties.

In recent years, we have exploited the regioselectivity of lipases and proteases inorganic solvents for the synthesis of specific esters of natural glycosides as well asthe selectivity of the �-1,4-galactosyltransferase from bovine colostrum to obtainnew glycosyl derivatives of these compounds.

In this report, the application of these two enzymatic methodologies will be exem-

70

aFinancial support was provided by MURST (40% and 60%) and CNR (“Progetto Strategi-co Tecnologie Chimiche Innovative”) to B. Danieli and M. Luisetti.

plified with different compounds: ginsenosides, digitonin, flavonoid glycosides, andcolchicoside.

REGIOSELECTIVE ACYLATION OF NATURAL GLYCOSIDESIN ORGANIC SOLVENTS

Several esters of glycosides can be found in nature. The formation of these estersis the last step in the biosynthetic pathway and it is catalyzed by different enzymescalled acyltransferases. Enzymes of this class show relative flexibility towards theacyl groups, but strict selectivity for the substrate to be esterified. However, even ona lab-scale synthesis, acyltransferases are not very convenient as they require stoi-chiometric amounts of the corresponding acyl-coenzyme A. On the other hand, directselective chemical acylation of glycosides is still a distant target because of the pre-sent lack of suitable reagents and protocols to discriminate among primary OH’s ofvarious saccharide units present in the same molecule as well as to regioselectivelyacylate one over several secondary OH’s.

During the last years, hydrolytic enzymes and more specifically esterases, lipases,and proteases have become valuable tools in organic synthesis due to their largeavailability, low cost, wide substrate spectrum, and no need of added cofactors. Wehave found that glycosides and polyhydroxyl compounds can be selectively acylatedat specific OH’s of the molecules by the action of an activated ester in the presence ofa suitable hydrolase in anhydrous organic solvent. This is the subject of the first partof this communication.

Enzymatic Acylation of Flavonoid Glycosides

Flavonol glycosides and their esters are an important group of natural compoundswidely distributed in the plant kingdom. For instance, the p-coumarates of somekaemferol and quercetin disaccharide monoglycosides are active components (to-gether with ginkgolides) of the very popular medicinal extract of the leaves of Gink-go biloba, which is used to increase peripheral and cerebral blood flow.

Due to the presence of several reactive groups even on their aglycone moieties,these compounds seemed to be particularly challenging and interesting substrates forenzymatic esterification. We examined first the behavior of several enzymes towardsthe acylation of flavonol monoglycosides (i.e., isoquercitrin, 1) and of more complexflavonol disaccharide monoglycosides (i.e., rutin, 2) (see FIGURE 1) with the activatedester trifluoroethyl butanoate. The protease subtilisin and the so-called lipase “B”from Candida antarctica gave interesting results in terms of selectivity, and the cor-responding esters (1a, 1b, and 2a) were isolated in good yields.1–3

Then, a chemoenzymatic approach to some 6��-O-(3-arylprop-2-enoyl) deriva-tives (cinnamate, p-coumarate, feruloate) of isoquercitrin 1 was explored to over-come the inability to directly introduce these acyl moieties by an enzyme-catalyzedreaction of 1 with the corresponding activated ester. This approach was based on theregioselective introduction of a methyl malonate residue at the CH2OH of the sugarmoiety by subtilisin catalysis to form the mixed diester 1c. It was impossible to hy-

RIVA et al.: ENZYMATIC MODIFICATION 71

drolyze chemically the methoxycarbonyl function and therefore 1c was subjected toan enzymatic chemoselective hydrolysis with a particular enzyme, biophine esterase.The thus-obtained malonic monoester 1d was made to react in a Knoevenagel-typecondensation with the appropriate aromatic aldehydes to afford the esters 1e, 1f, and1g.4

In a more recent report,5 we have shown that the malonic monoester 1d could bemore efficiently prepared by another two-step chemoenzymatic approach. Aftersome experimentation, we found that dibenzylmalonate is a good substrate for the li-pase from Candida antarctica, even in an acetone solution containing 10% pyridineto dissolve the flavanone glucoside. Under these conditions, isoquercitrin (1) wastransformed into the mixed diester isoquercitrin 6��-O-benzylmalonate (1h) in ac-ceptable yield (74%) and with high regioselectivity. Catalytic hydrogenation of 1h onPd/C in THF solution afforded pure 1d in quantitative yield after catalyst filtrationand solvent evaporation at room temperature, without need of further purification.

Enzymatic Acylation of Terpene Glycosides

Ginsenosides are an important class of dammarane-type triterpene oligoglyco-sides, which are isolated from the water-soluble portion of the dried roots and leavesof Panax ginseng C.A. Meyer, a plant widely used in traditional Chinese and Koreanmedicine. Recently, a careful examination of white ginseng extracts has revealed thatsome ginsenosides are present as monoesters of malonic acid, with the acylation siteoccurring invariably at one of the primary OH’s of the sugar moiety. These carboxy-acetyl ginsenosides behave as acidic saponins and, besides being more soluble in wa-ter than the ordinary glycosides, they also cause a remarkable increase of the solubil-ity of the other ginsenosides. In spite of these interesting properties, which candeeply influence the absorption of these drugs in humans, no information is available

ANNALS NEW YORK ACADEMY OF SCIENCES72

FIGURE 1. Examples of 1 and 2.

on the synthesis of malonyl ginsenosides or, more generally, on the synthesis of spe-cific esters of ginsenosides with aliphatic carboxylic acid.

Therefore, we decided to examine the behavior of some of these glycosides to-wards enzymatic acylation. Using the ginsenoside Rg1 (3) as a substrate (FIGURE 2),the best results were obtained with the lipase B from Candida antarctica (novozym435) in t-amyl alcohol using vinyl acetate as the acyl donor. A complete conversionto only two products in a 22:1 ratio took place. The two products were identified, onthe basis of an extensive analysis of their 1H-NMR spectrum at 600 MHZ, as 6�-O-acetyl- and 6�,6��-O-diacetyl-ginsenoside Rg1 [(3a) and (3b)], respectively.6

Excellent results were also obtained with the more complex disaccharide digluco-side ginsenosides Rb1 (4).7 The monoacetate 4a at the primary OH of the externalglucose of the disaccharide moiety at C-20 was mainly obtained, accompanied by thediacetate 4b (FIGURE 3).

For the synthesis of malonyl Rg1, the ginsenoside was made to react with di-methylmalonate under the usual conditions to afford the mixed diester 3c, which wasselectively hydrolyzed with pig liver esterase to the malonate 3d. This acidic saponinproved to be very unstable during chromatographic separation and purification due toits high propensity to undergo complete hydrolysis to Rg1 or decarboxylation to 3a,and was isolated in low yield. In an attempt to improve the yield of 3d, we tested analternative chemoenzymatic approach based on the enzymatic formation of the 6��-O-(2,2,2-trichloroethyl)-malonate 3e followed by the chemical removal of 2,2,2-trichloroethanol by action of Zn/AcOH.6 Both the enzymatic and chemical reactionof this sequence were quite clean; however, in this case too, the final chromatograph-ic purification of 3d was troublesome, with the product being always contaminatedby a variable amount of ginsenoside Rg1. Not surprisingly, 3d has never been foundin ginseng extracts. Probably, the isolation of this compound is greatly hampered byits observed instability and, consequently, it is lost during the purification protocol.

We next examined digitonin 5 (FIGURE 4), a complex monodesmosidic steroidalsaponin isolated from the seeds of Digitalis purpurea, which possesses the distinc-tive characteristic to cause rupture of the membrane of the red blood cells. This he-molytic activity is shared by other natural saponins (but not by ginsenosides) and it isgreatly influenced by the structure of the aglycone and by the number and type of


FIGURE 2. Examples of 3, specifically 3a–3e.

sugars as well as by their glycosidic linkages. It is assumed that hemolysis dependson the ability to form a complex with cholesterol, a central component of the erythro-cyte membrane. For the elucidation of the structural requirements for hemolysis, a1:1 complex of digitonin and cholesterol has been prepared and extensive NMR stud-ies have been undertaken to clarify the sites of interaction between the two mole-cules.8

To support this work, we devised to modify the digitonin molecule by preparingspecific derivatives and to evaluate their aptitude to form complexes. The digitoninmolecule contains a branched pentasaccharide chain and therefore it represents achallenging target of particular interest for selective modification. Out of the several


FIGURE 3. Examples of 4.

FIGURE 4. Examples of 5.

hydroxyl groups present in this molecule, only the HO-C4 of xylose and the HO-C6of glc(II) were recognized by the lipase from Candida antarctica in t-AmOH solu-tion. In the presence of vinyl acetate and vinyl laurate, this enzyme catalyzed the ex-clusive formation of the diacetate 5a and monoacetates 5b and 5c and of the mono-laurate 5d, respectively.9 The hemolytic characteristics and ability to form complexeswith cholesterol are under study.

REGIOSELECTIVE GALACTOSYLATION OF NATURAL GLUCOSIDES

The �-1,4-galactosyltransferase from bovine colostrum (GalT), which is availablefrom natural sources, is by far among the most studied enzymes belonging to thetransferase class. Besides accepting its natural substrates D-glucose and 2-acetamido-2-deoxy-D-glucose, it has been shown that GalT is quite versatile towards substitu-tions at the pyranoside moiety of the sugar acceptor, provided that the equatorial C-4hydroxyl is always present. Recently, Kren, Augé, and coworkers have shown thatGalT can use a glucoside of an alkaloid, elymoclavine 17-O-(2-acetamido-2-deoxy)-�-D-glucopyranoside, as a substrate.10 This was an interesting observation becausenatural glycosides often possess pharmacological properties and variations of theirsugar composition might offer easy access to new compounds with increased solubil-ity, bioavailability, and biological action.

In the framework of our research on synthetic applications of this enzyme, wehave applied Augé’s protocol to other natural glycosides. In a first communication,we have shown that good results can be obtained, in terms of the degree of conver-sion and selectivity, with the sweetener stevioside and its congener steviolbioside.11

Despite the fact that stevioside and steviolbioside possess three and two glucopyra-nosides, respectively, in their molecules (and therefore three or two possible galacto-sylation sites), GalT showed an absolute regioselectivity and only one monogalacto-syl derivative was isolated with both these compounds.

The general exploitation of this enzymatic methodology is hampered, however, bythe low solubility of some glycosides due to the hydrophobic nature of their agly-cones. The use of organic cosolvents might overcome this limitation. As only limiteddata were available on the compatibility of these solvents with this transferase, wehave performed a systematic investigation on the effects of organic cosolvents on theproperties of GalT and ancillary enzymes using the alkaloid colchicoside (6) as amodel compound.12

GalT-catalyzed galactosylation of 6 was performed according to a standard proto-col (FIGURE 5), with the progress of the reaction being easily monitored by TLC andreverse-phase analytical HPLC. After 5 days, 71% conversion to a single product wasobserved. This product was isolated by flash chromatography and characterized asthe 3-O-�-lactosyl derivative of colchicine, 6a, by FAB mass spectrometry and 1H-and 13C-NMR.

The influence of various organic cosolvents on the stability and activity of the �-1,4-galactosyltransferase from bovine colostrum (GalT) and of its ancillary enzymeUDP-galactose-4�-epimerase has been investigated. As an example, TABLE 1 showsthe degrees of conversion observed after 24 hours in the presence of 5%, 10%, and15% v/v of various water-miscible organic cosolvents.


Due to more detailed experiments with various amounts of the different cosol-vents, it has been found that some cosolvents, such as dimethyl sulfoxide andmethanol, can be used at concentrations up to 20% v/v without any influence on theperformance of these enzymes; in contrast, others, such as tetrahydrofuran, rapidlyinactivated GalT at concentrations as low as 5% v/v.

These results have been initially exploited for the galactosylation of the


FIGURE 5. Scheme of the GalT-catalyzed galactosylation of 6.

TABLE 1. Degrees of Conversion of Colchicoside into Its Corresponding Lactosidein the Presence of Various Amounts of Organic Cosolventsa

Cosolvent 5% v/v 10% v/v 15% v/v

blank 71 71 71dimethyl sulfoxide 72 74 75methanol 77 80 84ethanol 80 85 36acetone 76 76 67dioxane 72 67 25acetonitrile 69 66 55N,N-dimethylformamide 64 55 39tetrahydrofuran 55 1 0

aDetermined by HPLC after 24 h. Conditions: colchicoside, 20 mM; UDP-glucose, 150mM; Mn2+, 25 mM; �-lactalbumin, 1 mg/mL; GalT, 1 U/mL; epimerase, 1.8 U/mL; alkalinephosphatase, 5 U/mL; Tris buffer, 50 mM, pH 7.4; 30 °C. Each experiment was repeated atleast twice.

coumarinic glucoside fraxin (7), a compound that is almost insoluble in water (lessthan 0.5 mg/mL). Fraxin was solubilized up to 12.5 mg/mL (34 mM) in the reactionbuffer containing 15% v/v dimethyl sulfoxide, and a 98% conversion to the corre-sponding �-lactoside 7a was obtained in 48 h (FIGURE 6).

We have been applying the same methodology for the galactosylation of gin-senosides. As an example, FIGURE 7 shows the results obtained when ginsenosideRg1 (40 mM) was submitted to the action of GalT in the presence of 1 equivalentof UDP-glucose and of the ancillary enzyme UDP-galactose-4�-epimerase. In con-trast to the results obtained with stevioside and steviolbioside,11 it was evident thatthe GalT was able to modify both of the glucopyranosides present on the Rg1 mol-ecule. The products corresponding to the three main peaks of FIGURE 7 have beenisolated by reverse-phase preparative HPLC purification and, as might be expected,


FIGURE 6. Scheme of the galactosylation of 7 to 7a.

FIGURE 7. HPLC-chromatogram of the GalT-catalyzed galactosylation of the ginsenoside 3.

were identified by mass spectrometry as a digalactosylated derivative (3f, the lessretained peak) accompanied by the two corresponding monogalactosylated com-pounds. For an unequivocal structural assignment of the monogalactosylated com-pounds, a detailed 1H- and 13C-NMR study was undertaken, which allowed theidentification of the more retained compound as 3g and of the other one as 3h13

(see FIGURE 8).We have performed the enzymatic galactosylation with other ginsenosides that are

much less soluble in water. Exploiting the previously described results,12 these reac-tions have been run in the presence of 20% v/v of DMSO. As an example, FIGURE 9shows the HPLC-chromatogram obtained with the tetraglucoside ginsenoside Rb1.Some of the products corresponding to the peaks of this complex chromatogram havebeen isolated by preparative HPLC and their characterization is currently in progress.

In conclusion, we have shown that complex polyhydroxylated natural compoundscan be selectively esterified under the catalysis of the enzymes subtilisin and Candi-da antarctica lipase in organic anhydrous solvents, or galactosylated by action of the�-1,4-galactosyltransferase from bovine colostrum. These enzymatic methodologiesseem to be highly valuable from the point of view of the preparation of derivativeswith defined structure for the investigation of their biological activity. However, nopredictions can be presently made with confidence a priori on the site of acylation orgalactosylation.


FIGURE 8. Examples of 3f, 3g, and 3h.

ACKNOWLEDGMENTS

We thank Manfred Schubert-Zsilavecz, J. W. Goethe Universität, Frankfurt, Ger-many, for the NMR spectra. We also thank INDENA S.p.A. for a generous gift of thevarious natural glycosides described in this paper.

REFERENCES

1. DANIELI, B., P. DEBELLIS, G. CARREA & S. RIVA. 1989. Enzyme-mediated acylation offlavonoid monoglycosides. Heterocycles 1989: 2061–2064.

2. DANIELI, B., P. DEBELLIS, G. CARREA & S. RIVA. 1990. Enzyme-mediated regioselectiveacylation of flavonoid disaccharide monoglycosides. Helv. Chim. Acta 73: 1837–1884.

3. DANIELI, B., M. LUISETTI, G. SAMPOGNARO, G. CARREA & S. RIVA. 1997. Regioselectiveacylation of polyhydroxylated natural compounds catalyzed by Candida antarctica li-pase B (Novozym 435) in organic solvents. J. Mol. Cat. B: Enzymatic 3: 193–201.

4. DANIELI, B., A. BERTARIO, G. CARREA, B. REDIGOLO & S. RIVA. 1993. Chemo-enzymaticsynthesis of 6��-O-(3-arylprop-2-enoyl) derivatives of the flavonol glucoside iso-quercitrin. Helv. Chim. Acta 76: 2981–2991.

5. RIVA, S., B. DANIELI & M. LUISETTI. 1996. A two-step efficient chemoenzymatic synthesisof flavonoid glycoside malonates. J. Nat. Prod. 59: 618–621.

6. DANIELI, B., M. LUISETTI, S. RIVA, A. BERTINOTTI, E. RAGG, L. SCAGLIONI & E. BOM-BARDELLI. 1995. Regioselective enzyme-mediated acylation of polyhydroxy natural com-pounds: a remarkable, highly efficient preparation of 6�-O-acetyl and 6�-O-carboxy-acetyl ginsenoside Rg1. J. Org. Chem. 60: 3637–3642.

7. DANIELI, B., M. LUISETTI, E. RAGG & S. RIVA. Unpublished results.8. MUHR, P., W. LIKUSSAR & M. SCHUBERT-ZSILAVECZ. 1996. Structure investigation and pro-


FIGURE 9. HPLC-chromatogram of the GalT-catalyzed galactosylation of the ginsenoside 4.

ton and carbon-13 assignments of digitonin and cholesterol using multidimensionalNMR techniques. Magn. Reson. Chem. 34: 137–142.

9. DANIELI, B., S. RIVA, S. STEUER, W. LIKUSSAR & M. SCHUBERT-ZSILAVECZ. Unpublished re-sults.

10. KREN, V., C. AUGÉ, P. SEDMERA & V. HAVLICEK. 1994. �-Glucosyl and �-galactosyl transfercatalyzed by �-1,4-galactosyltransferase in preparation of glycosylated alkaloids. J.Chem. Soc. Perkin Trans. I, p. 2481–2484.

11. DANIELI, B., M. LUISETTI, M. SCHUBERT-ZSILAVECZ, W. LIKUSSAR, S. STEUER, S. RIVA, D.MONTI & J. REINER. 1997. Regioselective enzyme-mediated galactosylation of naturalpolyhydroxy compounds (part I): galactosylation of stevioside and steviolbioside. Helv.Chim. Acta 80: 1153–1160.

12. RIVA, S., B. SENNINO, F. ZAMBIANCHI, B. DANIELI & L. PANZA. 1998. Effect of organic co-solvents on the stability and activity of the �-1,4-galactosyltransferase from bovinecolostrum. Carbohydr. Res. 305: 525–531.

13. DANIELI, B., S. RIVA, D. MONTI, W. LIKUSSAR & M. SCHUBERT-ZSILAVECZ. Unpublished re-sults.


Documents

Enzymatic Modification of Natural Compounds with Pharmacological Properties