Journal of Medicinal Plant Research Editor - in - Chief Editorial Board Hippokrates Verlag E

Home , Triterpene

Journal of Medicinal Plant Research Editor - in - Chief Editorial Board Hippokrates Verlag E. Reinhard, Univ. Tiibingen H.P.T. Ammon. Tubingen Stuttgar? Pharmazeutisches lnslitut W. Barz, Munsler Auf der Morgenstelle E. Reinhard. Tubingen 7400 Tubingen 0. Sticher, Zurich Vol. 39 H. Wagner. Miinchen M. H. Zenk, Bochum May 1980 NO. 1

Droplet Counter-Current Chromatography and its Application to the Preparative Scale Separation of Natural Products

Kurt Hostettmann

Pharmazeutisches lnstitut der ETH, ETH-Zentrum CH-8092 Ziirich, Switzerland

Key Word Index: Droplet Counter-Current Chromatography; Natural Products Isolation; Saponins; Polyphenolics; lridoid Glycosides; Cardenolides; Alkaloids; Peptides; Lipids; Monosaccharides; Cyclitols.

Abstract column of surrounding stationary phase [I].It is particularly indicated for the Droplet counter-current chromatogra- preparative scalc separation of polar phy (DCCC) is a recently developed all- compounds and has been applied to the liquid separation technique based on the isolation of various types of naturnl pro- partitioning of solutes between a steady ducts. In many cases, the separations we- stream of droplets of mobile phase and a re achieved far more readily than by con- Hostettmann ventional chromatography and with a this method and have applied it to the smaller solvent consumption. In the pre- isolation of pure compounds from crude sent paper, the principle of DCCC is medicinal plant extracts [4]. The separa- described and a quick method based on tion time was very short, the resolution thin-layer chromatography (TLC) for high, but a preliminary purification of the selection of the solvent system is pro- the sample prior to injection was neccssa- posed. An up to date literature survey is ry in order to avoid column contamina- given and the problems specific to the tion. different classes of compounds are dis- All the techniques mentioned above cussed. Advantages and limits of DCCC, use a solid stationary ~haseor a liquid as well as its position among the various stationary phase fixed on an "inert" available methods for the preparative sca- support. The presence of this solid pac- le separation, are discussed. king material in a column may cause irreversible adsorption when dealing with very polar compounds or cause rnodifi- Introduction cation of the solute at the liquid solid in- terface. The purification and the separation of In order to avoid complications arising relatively large amounts of compounds is from solid supports, various supporr- often a vely tedious and time-consuming free liquid-liquid partition techniques operation, especially when dealing with have been developed. [5, 6, 71 complex mixtures such as plant extracts When two immiscible solvents contai- and other biological material. ning solutes are shaken in a separatory Various chromatography methods for funnel and then separated, the solutes are separating compounds in a semi-prepara- partitioned between the two phases. The tive and .preparative . scale have been de- ratio of solute concentration in the upper veloped. Column chromatography using phase to the lower phase is called the par- various solid stationary phases (silica gel, tition coefficient. If the partition coeffi- alumina, polyamide, cellulose, Sephad- cients of two compounds differ greatly, ex, etc.) is very popular and employed in the only process needed for separation is most laboratories. This technique has a a one-step operation called extraction. high capacity, but the separation time is As the nature of the substances becomes long and the resolution power relatively more similar, the difference of their par- low. In recent years, several preparative tition coefficient decreases, hence requi- systems have evolved which reduce sepa- ring multistep extraction. CRAIC[S] desi- ration times to several hours. Among gned machines in which many distribu- them, we have to mention shorr-column tion steps, up to hundreds in number, chromatography [2] and flash chromato- could readily be achieved by liquid-phase graphy [3] which give very rapid separa- transfers In all the stages simultaneously. tions but with moderate resolution. Pre- This technique, called counter-current parative liquid chromatography is a more distribution (CCD), was used with out- efficient method and is becoming very standing success for the purification of popular. We have been successful with natural polypeptide antibiotics. [9] Droplet Counter-Current Chromatography

Its applications in various fields of or- Descriptionof theMethod and Choice ganic chemistry and biochemistry have of the Solvent System been reviewed by HIMKI-R [lo]. CCD has a high capacity, but the resolution power DCCC grew out of the observation is limited. In addition, the apparatus is that a light phase with low wall surface complex, thc volumes of solvents used affinity formed discrete droplets that ro- large and the separation time relatively se through the heavy phase with visible long. evidence of very active interfacial mo- I tion. Under ideal conditions each droplet A simplc and very efficient all-liquid could become a "plate" if kept moreof less separation technique called droplet coun- discrete throughout the system [5]. The ter-current chromatography (DCCC) method developed by TANIMUHAet al. [I] has been developed by TA~IAI[HA et al. exploits these ideas and consists basically of [I]. 200'to 600 long vertical columns (20 to 60 It is less cumbersome than CCD and cm) of narrow bore glass silanized tubing has a higher resolution power. DCCC is (1.5 to 2 mni, i.d.) interconnected in se- carried out by passing droplets of a mo- ries by capillary Teflon tubcs. A first step bile phase through a column of surroun- consists to fill the whole system with sta- ding stationary liquid phase. Recently, tionary phase followed by the injection this technique became commercially of the sample (dissolved either in light

available and we are using it extensively phase or in heavy. phase. or in a mixture of in our laboratories for the isolation and both phases) in a sample chamber. The purification of various natural products. mobile ~haseis than pumped through the Most of the separations could be achie- sample chamber and inserted with the ca- ved far more readily than by conventio- pillary tube into the bottom of the first nal chromatographic methods. Sample glass column of wider bore. A steady amvunts from several milligrams up to stream of ascending droplets is formed. grams can easily be handled and the sol- When a droplet reaches the top of the co- vent consumption is very small. At the lumn, it is delivered to the bottom of the present time, numerous laboratories are next column through the Teflon tubing, introducing this technique and applica- thus regenerating new droplets. Under tions of DCCC to various separation suitable conditions, the capillary tubing problems in natural products chemistry allows only the mobile phase to flow. A and in biochemisty have already been pu- small amount of stationary phase may al- blished. Some authors employed DCC as so enter the Teflon tubing initially, but abbreviation for droplet counter-current the effect is insignificant. As the mobile chromatography. In order to avoid con- phase moves through the column, turbu- fusion, it is desirable to use DCCC, since lence within the droplet promotes effi- DCC is the official abbreviation of dicy- cient partitioning of the solute between cl~hex~lcarbodiimide.Furthermore, we the two phases. noticed recently that dry column chro- Depending on the separation problem, matography was also designated by the mobile ~hasemay be either heavier or DCC. lighter than the stationary phase. When 4 Hostettmann -ca 300 tubes

phase sample,

I collector collector ascending mode descending mode li~.I. Schcniatic illu5rrarion ol rhc principle ol DCCC lighter, the mobile phase is delivered at CRAIGcounter-current distribution pro- the bottom of the column (ascending duces flow plugs which result in the dis- mode) and, when heavier, through the placement of the stationary phase by the top (descending mode) (Fig. 1). mobile phase. Another unsuitable sol- Solvent systems that form two immis- vent system was n-butanol-pyridine-wa- cible layers are usually suitable for ter which gave no droplets due to its high DCCC, however there are some limita- viscosity. We obtained good results by tions. The generation of droplets having using CHCI,/MeOH/H,O, CH,CI,/ suitable sizes and mobility are governed MeOH/HIO or CHCI,/MeOH/PrOH/ by factors such as the difference in speci- HzO mixtures in different proportions. fic gravities of the two liquid phases, the Modification of the amount of MeOH in viscosity of solvents, the surface tension, a given mixture leads to a large change in the flow rate of the mobile phase, the the polarity of the system. The solvent diameter of the inlet tip and the internal systems used so far to give a suitable size diameter of the column. In general, of droplets for efficient separation are gi- small-bore columns (less than 1.0 mm) ven in Table I. Most of them are tertiary produce flow plugs in which the entire or quaternary systems, one of the com- contents of the tube are displaced. On ponents being water. the other hand, it should be noted that a A quick way to select a solvent system large diameter of the tubes decreases the consists in checking the sample by thin- efficiency of separation. An internal co- layer chromatography (TLC) on silica lumn diameter of about 2 mm seems to gel with the water-saturated organic lay- be ideal. We have used numerous solvent er [l 1, 121. Empirically, we found that if systems, among them, some were not in- the R, values of the compounds to be se- dicated [l I]. The system methanol-hex- parated are higher than about 0.50 (less ane-water which is frequently used in the polar solutes), the less polar layer is sui- Droplet Counter-Current Chromatography 5

Table I Solvent systems employed in DCCC

Separated compounds Solvent systems References saponins (triterpenoid glycosides) CHCI3-MeOH-Hz0 (7:13:8) 11, 19,20 CHC1,-MeOH-Hz0 (5:6:4) 19 CHC13-MeOH-n-PrOH-Hz0 (9:12:1:8) 20 CHC13-C6H6-EtOAc-MeOH-Hz0(45:2:3:60:40) 25 saponins (spirostanol glycosides) CHCI3-MeOH-Hz0 (7:13:8) 18 saponins (ginsenosides) CHCI3-MeOH-n-PrOH-Hz0 (45:60:6:40) 24 CHC13-MeOH-iso-PrOH-Hz0 (5:6:1:4) 22 CHCI,-MeOH-n-PrOH-Hz0 (4:6: 1 :4) 21, 23 cardiac glycosides CHCI3-MeOH-Hz0 (5:6:4) 19 CHC13-MeOH-Hz0 (5: 10:6) 29 CHCI3-MeOH-Hz0 (5:9:7) 29 iridoid and secoiridoid glycosides CHCI3-MeOH-Hz0 (433720) 12 CHCI,-MeOH-n-PrOH-Hz0 (9: 12: 1 :8) 12 direrpenoids CHCI3-MeOH-Hz0 (7:13:8) 3 1 xanthones (glycosides) CHCI,-MeOH-n-PrOH-Hz0 (9: 12: 1 :8) I2 xanthones (aglycones) CHC13-MeOH-Hz0 (13:7:4) I2 flavonoid glycosides n-BuOH-AcOH-H20 (4:1:5) I I CHCI,-MeOH-Hz0 (7: 13:8) 19 CHCI3-MeOH-Hz0 (5:6:4) 32 CHCIj-MeOH-Hz0 (4:4:3) 35,41 CHC1,-MeOH-n-PrOH-Hz0 (5:6:1:4) 33 CHCIj-MeOH-n-BuOH-H:O (10: 10: I :6) 34 anthraquinone glycosides CHCI3-MeOH-n-PrOH-Hz0(9:12:1:8) 19, 20 chromenes, lignans CHCI3-MeOH-Hz0 (5:6:4) 32, 40 CHCL-MeOH-HD (4:4:3) 4 1 sugars (monosaccharides) CHCI,-MeOH-Hz0 (7:13:8) 20 cyclitols CHCIj-MeOH-H2O (50:57:30) 34 alkaloids CHCI,-MeOH-Hz0 (13:7:8) 4 3 CHCI3-MeOH-Hz0 (5:5:3) 44, 45, 46 CHCIj-MeOH-5 % HCI (553) 46 C6H6-CHCIJ-MeOH-Hz0 (5:5:7:2) 42 CHC13-MeOH-n-PrOH-HzO(45:60:2:40) 4 7

DNP amino acids CHC13-AcOH-0. IN HCI (22: 1) peptides C6H6-CHCIj-MeOH-Hz0 (15:15:23:7) C6H6-CHCI3-MeOH-0. IN HC1 (1 1 :5:10:4) CHC13-MeOH-O.IN HCI (19:19:12) sec-BuOH-TFA-Hz0 (120: 1 :160) n-BuOH-AcOH-Hz0 (4:1:5) lipids heptane-n-BuOH-CHC1,-MeOH-60% AcOH (3 :2 :2:3 :5) sulf~nogl~colipids CHC13-MeOH-H20(IO:12:7) 6 Hostettmann table for use as the mobile phase. With fied as 1,3,5-trihydroxyxanthone-8-0- more polar substrates (Ri< 0.50), the 0-D-glucoside and as 1,s-dihydroxy-3- more polar layer should be used as the methoxyxanthone- 8 - 0 -0-D-glucoside mobile phase. If the Ri values are in the by comparison with authentic samples range 0140-0.60, the separatkon can be [I31 (~ig.2, top elution curve). Both gly- achieved by using either the more polar cosides were separated with a total volu- layer or the less polar layer as the mobile me of about 350 ml of mobile phase. As phase. This is illustrated in the following expected, the use of the more polar layer by the separation of a fraction (60 mg) of as the mobile phase resulted in earlier a crude extract of a Rocky Mountains elution of the more polar compound 1. Gentiana species (Gentiana strictiflora For comparison, the same mixture of 1 (R)I)I~.)A. NII~)obtained after column and 2 was submitted to DCCC and elu- chromatography on Sephadex LH-20. ted with the less polar lower layer. The On silica gel TLC, with CHC1,-MeOH- total volume of mobile phase used for PrOH-HLO(9:12:1:8), lower layer, this eluting both compounds was 800 ml fraction showed two UV active spots at (Fig. 2, bottom elution curve); however, R, = 0.35 and 0.4 1. DCCC separation the resolution was better than in the pre- by using the more polar upper layer as vious run. As expected, the less polar mobile phase yielded two pure com- glycoside 2 was eluted first. pounds 1 (32 mg) and 2 (26 mg), identi- These comparative experiments sug-

Fig. 2. DCCC separation of xanthone-0-glycosi- (Reprinted Ironi: HOSI.I.TI.MAN~., K., M. Ho- des with CHCI3-MeOH-n-PrOH-H,O (9:12:1:8). ~T~::T~IANN-KAI.I~A\~~~0.S.I.I(.HI.H: Helv. Chim. Top elution curve: thc niobilc phabc was the more Acta 62, 2079 (1979)). polar upper layer. Bottom elution curve: thc mobile phase was the less polar lower laycr. Droplet Counter-Current Chromatography

gest that in the case of difficult separa- form one of the most important group of tions, it could be better to use thh lcss naturally occurring glycosides. They ex- polar layer as mobile phase for eluting ihibit various biological- and pharmacolo- compounds with relatively low Ri values, gical activities and the interest in these and the more polar layer for eluting com- compounds is still growing. Thc me- pounds with relatively high R, values. A thods for the separation of saponins have better resolution will be obtained, but been reviewed by HI[I I R et al. [14, 151 the separation time will be much longer. and consist mainly in column chromato- If the Ri values of the compounds to be gaphy on silica gel, cellulose or Sephad- separated are either too low (< 0.3-0.2), ex. Frequently, the column chromato- or too high (> 0.8-0.7), no elution will graphic separation requires solvent gra- take place in a reasonable time, although dients which are very time consuming. the compounds can be recovered quanti- Counter-current distribution is not indi- tatively from the remaining stationary cated as the shaking device causes forma- phase. tion of foams. This is avoided in DCCC Solvent system selection based on and thus this technique is ideal for the TLC is very simple and rapid and gives isolation of saponins. During the course generally satisfactory results. The ad- of our systematic isolation and structural sorption effect is small by using the wa- studies on biologically active compounds ter-saturated organic layer as eluent and from medicinal plants, we found several can partially be eliminated by conditio- species possessing strong molluscicidal ning the plate prior to development. But properties. Plants showing this activity since DCCC depcnds solely on the parti- are currently receiving considerable att- tion coefficient, separation cannot al- ention as potential agents for the control ways be predicted from the TLC beha- of bilharzia. We used DCCC for the iso- vior. lation of the active principles of common ivy berries, Hedera helix (Araliaceae) [I'll. The fresh berries were extracted Application in Natural Products with 70 % ethanol. After concentration, Isolation this extract was partitioned berween n- DCCC has been applied up to now to butanol and water. The n-butanol layer the purification and separation of various (I .2 g) exhibiting the biological activity classes of natural products from plant was submitted to DCCC using CHCI,- MeOH-H,O (7:13:8) as solvent (mobile sources. Since there is no solid support. . which might cause irreversible adsorp- phase: the less polar lower layer) and the tion, DCCC has been used especially for fractions were monitored by TLC. This the isolation of polar plant constituents gave five fractions as shown in Fig. 3. such as glycosides which are often very Fractions 1 which was eluted with the difficult to separate. solvent front was still a mixture of the least polar components (i.e., fatty acids, Saponins chlorophylls, etc.) of the crude extract, Triterpene glycosides and spirostanol whereas four pure compounds 2-5 were glycosides are widely distributed and obtained (shaded area in Fig. 3). No 8 Hostettmann

2: R= arob. 3: R- g1uc. 4: R= rham.-arab. 5: R= cj1uc.-oluc. 5 115 -0

Fig. 3. DCCC ol a crude rxrracr of Hedera helir \T~ITMA~XK.. M. H(~\TII~XI,\NN-KAI I)AI and berries (1.2 g) with CHC1,-McOH-H20(7:13:8); K. NAKASI~HI:J. Chrornatogr. 170. 355 (1979)). mobile phase. lon,cr layer (Reprinted from: HO- base line separation could be achieved for the molluscicidal principle from the me- compounds 3 and 4 with the conditions thanolic bark extract of dogwood, Cor- employed in the present experiment. Af- nus florida (Cornaceae) [I 81. ter elution of c'ompound 5, about 670 mg A very active fraction, obtained after of the most polar components (free su- gel filtration of the crude extract, was gars, polysaccharides, etc.) were recover- submitted to DCCC (same conditions as ed from the stationary phase of the appa- above) and afforded three base-line sepa- ratus. The pure compounds were identi- rated spirostanol glycosides identified as fied by means of spectroscopic studies sarsapogenin - fi -1)-galactopyranoside, (FD-MS, 'H-NMR and "C-NMR) as sarsapogenin - 0 -0-1)-xylopyranosyl- (I glycosides of hederagenin [16]. The -, 2)-fi-1)-galactop~ranosideand sarsa- structure determination of these triterpe- pogenin-0-(3-1)-glucopy ranosyl-(I + 2)- noid saponins will be published shortly (3-I)-galactopyranoside. The entire sepa- along with the results of the biological ration was achieved in less than 8h and tests [I 71. the total volume of mobile phase used for As expected by usage of the less polar the elution was 100 ml, whereas no effi- layer as mobile phase, the glycosides 2-5 cient preparative scale separation could were eluted in order of increasing polari- be made either by open column chronia- ty. In the present case, major plant con- tography or TLC. stituents were obtained directly in pure The saponins of the seeds of Zizyphus form from a crude extract with a very jujuba and the bark of Hovenia dulcis small solvent consumption. were successfully separated by T.WI>ICK,\ However, when the active principle of et al. [19, 201 who used CHCI,-MeOH- a plant is a minor constituent, it is prefe- H20 (7:13:8) and CHCI,-MeOH-n- rable to inject an enriched fraction into PrOH-H,O (9:12:1:8) as solvent sy- DCCC. We illustrated this by isolating stems in the ascending mode (mobile 1 Droplet Counter-Current Chromatography 9

phase: upper layer). In the course of the of saikosaponins in the roots of Bupleur- chemical investigation of the well-known urn falcatum (Umbelliferae) and its con- Chinese longevity drug ginseng (the root geners [25]. By using CHC1,-C,H,-EtO- of Patiax ginseng), TANAKAet al. [21,22, Ac-MeOH-H,O (45:2:3:60:40) as sol- 231 used DCCC in combination with ot- vent in the ascending mode, a base-line her chromatographic techniques for the separation of saikosaponins a, b and c isolation of closely related dammarane- was obtained. (Fig. 4) type saponins. A quantitative determina- As expected by usage of the more po- tion of ginsenosides in several ginseng lar layer as mobile phase, saikosaponin c species has been reported by O~SIJKAet was eluted before saikosaponins a and d. al. [24]. These authors fractionated the The separation of saikosaponin c with a crude butanol extracts by DCCC and trisaccharide sugar moiety from saikosa- treated each fraction with phenol/ ponins a and d possessing two sugars can

H2S0,. The characteristic orange- colour readily. be .performed by column chroma- obtained was determined photometrical- tography, but the separation of the latter ly at 490 nm. The same procedure was two saponins which differ only in the applied to the quantitative determination configurations at C-16-OH is very diffi-

R: POH Saikosaponin a 1 R: aOH Saikosaponin d ~xa

Saikosaponin c

2.011Saikosaponin c

Saikosaponin a Saikosaponin d

100 200 300 Fraction Number FIR. 4. DCCC of an exrracr of wild-growing Bu- (Reprinted from: OTS~KA,H., S. KOBAYASHIand pleurum longirudiarum with CHC1,-C6H6-ErO S. S~l8m.A:Planta med. 33, 152 (1978)). Ac-MeOH-H20 (45:2:3:60:40) by the ascending method. Hostettrnann

cult to achieve by the usual method (i.e. The crude leaf extract was first fractio- one single spot on TLC). Recently, a nated by silica gel column chromatogra- high performance liquid chromatography phy into biosides and triosides. The frac- (HPLC) study of Bupleuri Radix (Bu- tion containing the biosides was subjec- pleurum falcatum) has been reported ted to DCCC using CHCI,-MeOH- [26]. The genuine saikosaponins contai- H,O (5:10:6) by the ascending method ning the unstable ally loxide linkage were and afforded compounds I and 2. The converted into artificial diene-saponins more polar triosides 3 and 4 were separa- by mild acid treatment and separated ted with CHCI,-MeOH-H,O (5:9:7, as- by reverse-phase chromatography on a cending). column of octadecylsilylated silica gel with MeOH - H,O - AcOH - (Et),K (75:25:0.2:0.2). The separation of the stereoisomers obtained by DCCC was comparable, even better, than the one obtained by HPLC. But for quantitative analysis, HPLC is, of course, a much more rapid method. By repeated column chromatography and DCCC, ICHIIet al. [27], isolated ten closely related oleanene-type saponins from the roots of Platycodon grhdij7or- urn. Solvent systems formed of CHC1,- MeOH-H,O are well indicated for the separation of various types of saponins and can be used either in the ascending mode (very polar saponins) or in the descending mode (saponins possessing one A purification of digironin was achie- or two sugars and few free hydroxyl ved by T.AXI\ICHAet al. [I91 with CHCI,- groups). We used the relatively weakly 1MeOH-H,O (5:6:4) in the ascending polar system CHCI,-MeOH-H,O mode. (65:35:20) in the descending mode for the isolation of rriterpenes such as ole- lridoid and secoiridoid glycosides anolic acid from Gentiana species [28]. We applied DCCC to the isolation of iridoid glycosides from Ajnga pyramida- Cardiac glycosides lis (Labiatae) [I2]. The crude butanol ex- Cardenolides being structurally related tract (1.4 g) was fractionated with to saponins, the above mentioned solvent CHCI,-MeOH-H,O (43:37:20) by the systems can be used. During the course ascending method. This resulted in three of re-investigation of the cardiac glycosi- major fractions (Fig. 5). The fraction elu- des of Nerium oleander (Apocynaceae), ted with the solvent front (680 mg) was a AUIand YAMAUIHI [29] isolated new gly- mixture of the most polar constituents, cosides with an unusual framework. whereas fractions 1 (1 14 mg) and 2 (508 Droolet Counter-Current Chromatography 11

/-is I LX:CC oi ;I cruclr cxtr2c.t oi .,l;rrga pymmi- (Rcprinred troni: HO\TI.I-I.\I.\\XK.. M. HO\IITI' rLlis (1.4 g) with CH<:I,.McOH-H?O(43:37:20) \IA\S KAI~)~\.s.intl 0. SII~'III 4: Heh. Chirn. Ac.1~ by thc aaccnding 111e1Iiotl(~nobilr phase: upper lay 62. 2079 (1979)). rr). mg) showed single spots on TLC. The ned after column chromatography on Se- pure compounds were identified by me- phadex LH-20 a fraction containing mi- ans of spectroscopic studies as harpagidc nor iridoid constituents. This fraction, and 8-0-acety Iharpagide, respectively. when submitted to DCCC, gave four ne- The entire isolation process required on- arly pure glycosides. Final purification ly about Sh and the total amount of sol- was achieved by semipreparative HPLC vent used for the elution was less than on a C,, chemically bonded silica gel 150 ml. From the remaining stationary with MeOH-H,O. Sptctroscopic studies phase, 94 mg of a mixture of minor con- on the pure compounds indicate that stituents could be recovercd. they are closely rclated to loganin and di- Gentiopicrin, one of the bitter princip- hydrocornin. The structure determina- le of Gentlanu, and related secoiridoid tion is currently under investigation and glycosides have also been isolated by nlill be published shortly. DCCC [12]. The solvent consumption CHCI,-MeOH-H,O mixtures were for separation by preparative HPLC on found to be ideal for the separation of the same scale was much higher. HPLC this class of glycosides. For iridoid gly- is very fast, but requires tedious purifica- coside esters, it is desirable to use the less tions of the sample prior to injection in polar lower layer as mobile phase. order to avoid column contamination [301. Dit erpenoids From a Casrilleja species (Scrophula- Only one example of separation of we- riaceae) collected in Colorado, we obtai- akly polar compounds by DCCC has be- Hostettmann

en reported up to now, namely the isola- nolic glycosides by DCCC. For flavono- tion of cytotoxic nor-diterpenoid dilac- ids and related compounds bearing few tones from Podocarpus nagi [31]. Com- hydroxyl groups and one sugar or agly- pounds I and 2 both showed very similar cones possessing several free hydroxyl chromatographic behaviors and could groups, the separation can be achieved not be separated by TLC nor by HPLC. by using the less polar lower layer as mo- Only a combined technique of adsorp- bile phase (descending mode), whereas tion chromatagraphy with DCCC gave a the more polar layer is suited for more satisfactory separation of 1 and 2. The polar monosaccharides or disaccharides. solvent system used was CHC1,-MeOH- For example, from a crude fraction of H,O (7:13:8) in the descending mode. Tecoma stuns (Bignoniaceae), we werc able to isolate within 6h a pure glycoside Polyphenols (flavonoids, xanthones, ca- identified as quercetin-3-0-glucosidr techins, anthraquinones, using the upper layer of CHCI,-MeOH- chromenes, lignans) H,O (7:13:8) as mobile phase [I I]. [:or The acidic hydroxyl groups in poly- the separation of very polar flavone gl!.- phenols often cause irreversible adsorp- cosides, we used n-BuOH-AcOH-H1O tion on the solid sationary phase during (4:1:5), but the separation time was long. the column chromatographic procedure. due to the high viscosity of this solvent This can be parrially avoided by adding system [I I]. DCCC has also been app- an acid to the mobile phase and is useful lied to the isolation of flavone-c-glycosi- for analytical work. However when de- des from various plant sources (Table 11). veloping a preparative scale separation, A remarkable separation has been re- the use of acid as one of the components cently reported by T.\>,\~Aet al. [35]. of the mobile phase is not always indica- They isolated 3 "-0-aceryl-quercitrin (I) ted. The risk of sample deterioration, i.e. and 4"-0-acetyl-quercitrin (2) from the hydrolysis of glycosides, is increased du- aerial parts of Pteris grandifolia (Pterida- ring the long operation and the recovery ceae) using CHC1,-MeOH-HIO (4:4:3) of the sample from the relatively large in the ascending mode (mobile phase: volume of eluent. upper layer). CHC1,-MeOH-H,O mixtures in dif- A base-line separation of I and 2 was ob- ferent proportions have been used for the tained. separation of various classes of polyphe- Separations of isonleric acylnted flwono- Droplet Counter-Current Chromatography 13

Table II C-gIycosides isolated by DCCC.

- - C-glycoside plant source solvent Ref. swenisin Zizyphus juiuba CHCI3-MeOH-Hz0 (7:13:8) 19 vitexin Lindsaea ensifolia CHCI3-MeOH-Hz0 (5:6:4) 32 isoorienrin Gentiana srrictiflora CHCI,-MeOH-nPrOH-Hz0 (5:6:1 :4) 33 6,s-di-C-pentosy l Lespedeza cuneata CHCIj-MeOH-n-BuOH-Hz0 34 apigenin (10:10:1:6)

strictiflora was submitted to DCCC with CHCI,-MeOH-H,O (65:35:20, descending mode) [12]. Three fractions were obtained (Fig. 6). The less polar constituents (colour pigments, fatty acids, triter- penes) were not separated under the cho- sen conditions and were eluted with the solvent front, whereas fraction 1 (29 mg) and 2 (10 mg) showed single spots on T1.C. Spectroscopic studies of the pure compounds led to the structure of belli- RO OH difolin and desmethyl-bellidifolin, re- I:R = Ac, R, = H spectively. These xanthone aglycones 2: R = H, R, = Ac were separated far more readily than by column chromatography and with a id glycosides are usually very difficult to smaller solvent consumption (the total achieve, even by HPLC. Thus, DCCC amount of mobile phase used being 160 opens new possibilities and will be of ml). great help in the future for all chemists The above mentioned solvent system is interested in flavonoids. indicated for the separation of other The structures of xanthones are related kinds of non glycosidic phenolics. We to that of flavonoids and their chromato- employed it for the isolation of the red graphic behaviour is similar. There is at colour pigments of a crude extract of Hy- present great interest in these compounds pericum perforaturn (common St. John's since SLZLRIet al. [36] demonstrated that wort) [37]. Both, the ascending and the xanthones isolated from GI:~TIANAinhibit descending modes, furnished hypericin rat brain mitochondria1 monoamine oxi- and pseudohypericin [38] in a pure form. dase (MAO) in vitro. In connection with Among other polyphenolics successfully our systematic investigation on xantho- separated by DCCC, we have to men- nes from GI:>~.I,I~A,we studied several tion the isolation of an unstable catechin Norrh-American species. The crude glycoside from Dalbergiu nitidula (Legu- chloroform extract (200 rng) of Gentiana minosae) [I I], the separation of sennosi- 14 Hostettmann

abs.

Solwent front

Trg. 6. DCCC of a crudr chloroforn~extract oC (Reprinred from: Hon-t~rrlhsx.K.. M. HQI G'rnrhna strrctrf7ora with CHCI,-MeOH-H20 (;.rl:'l-rhlr\~s-KAlI)A\ and 0. STICHI.I(:Hcl\.. Chini. (13:7:4) in the dcsccnding mode (mobile phasc: lo- Acta 61. 2079 (1979)). wcr layer). des (bisanthrone glycosides) from rhu- works to isolate biologically active sub- barb [19, 201, the purification of lucidin stances to insects from plants, Nc\I.\I.+et primeveroside from ccll cultures of Ga- al. [34] studied the feeding stimulants for lium mollugo [38] and the isolation of li- larvae of the yellow butterfly Eurema he- gnans [40] and chromenes [32, 411 from cabe mandarins. They isolated from the various Pteridaceae species. All these se- leaves of Lespedeza cuneata two active parations could be achieved by using compounds identified as myo-inositol CHCI,-MeOH-H20 or CHCIl-MeOH- and D-pinitol. n-PrOH-H:O as solvent systems.

Carbohydrates and Cyclitols A satisfactory separation of naturally occurring monosaccharides (arabinose, xylose, galactose, glucose, fucose and rhamnose) was obtained by OC.III,\I~Aet a]. [20] with CHCI,-MeOH-H,O (7:13:8) in the ascending mode. The number of theoretical plates obtained in A crude fraction (1.549 g) was submit- this separation was 1200. As a part of ted to DCCC with CHC13-MeOH-HjO Droplet Counter-Current Chromatography

(50:57:30; descending mode) and affor- quick and good separation of numerous ded first D-pinitol (1.287 g) and then alkaloids . myo-inositol (0.212 g). In the course of a biosynthctic study of the alkaloids of Cephaelis ipecacuanha Alkaloids (Rubiaceae), NA(;AKLRAet al. [47] used The benzylisoquinoline alkaloid co- DCCC for the purification of ipecosidc claurin was isolated from the root bark of with CHCI,-MeOH-n-PrOH-Ha Zizyphus jujuba with a CHCI,-MeOH- (45:60:2:40) in the ascending nlodc. H,O system [42]. We used this solvent system for the separation of the main opium (Papaver somniferum) alkaloids Application in Biochemistry [43]. As the extract consisted of components with a wide range of polarity, the The first separation reported by the in- separation had to be achieved in two ventors of DCCC, TALIMLKAet al. [I], steps. A first run with the more polar up- was a high rcsvlution of dinitrophenyl per layer as the mobile phase afforded (DNP) amino acids. The solvent system crude morphine, a mixture of narceine was CHCI,-AcOH-O.1N HCI (2:2:1) and codeine and pure thebaine. After used in the ascending mode. DCCC is al- elution of thebaine, the remaining statio- so indicated for the purification of pepti- nary phase, consisting of the less polar des. Otin>~c)~oet al. [48] fractionated cru- benzylisoquinoline alkaloids was reco- de gramicidin into pure gramicidins A, B vered. A second run with the less polar and C and a partially purified tyrocidine lower layer resulted in the separation of into pure tyrocidines A, Band C with C,, narcotine and pnpaverine. The two-step HI, - CHCI, - MeOH-O.IN HCI separation has been avoided by TAXIet (1 1 :5:10:4), C,H,-CHCI,-MeOH-H,O al. [44, 45, 461 who modified the pH of (15:15:23:7) or CHC1,-MeOH-0. IN the mobile phase during the elution. HCI (19:19:12). The lower layer was With this method they obtained a good used as mobile phase (descending mode) resolution of highly complex alkaloid with a flow rateof 12-15 ml/h. NA~~\.\\I\ mixtures from Papaveraceous plants. For et al. [49] obtained pure fowl angiotensin example, from Corydalis ophiocarpa, from a partially purified peptide. Kalli- two new protoberberine alkaloids were din, bradykinin and angiotensin were se- isolated together with known alkaloids parated with sec-BuOH-trifluoroacetic [46]. Up.to 5 g of sample was submitted acid-H,O (120:1 :160), mobile phase: lo- to DCCC with CHCI,-MeOH-H,O wer layer [19]. A mixture of three similar (50:55:30; McIlvain buffer solution, peptides was separated by Takahashi et pH7) by the ascending method. In the al. [50] with n-BuOH-AcOH-H,O course of the elution, the aqueous mobile (4:1:5). It appears from their work that phase was replaced first by the upper lay- DCCC could be a suitable method for er of CHCI,-MeOH-H,O (50:55:30; the separation of synthetic oligopeptides. McIlbain buffer solution, pH3) and fi- A fractionation of human serum lipids nally by the upper layer of CHCI,-Me- has been reported by Konobu et al. [51]. OH-5 % HCI (50:55:30), resulting in a With heptane-n-BuOH-CHCII-MeOH- Hosteltmann

AcOH 60 % (3:2:2:3:5), a base-line se- not comparable with HPLC, however paration of triolein, palmitic acid, capric 1000-1500 theoretical plates can easily be acid and lecithin was achieved by the as- obtained, depending on the number of cending method. In the course of the stu- columns used. dy of the metabolites of Anthoc~daris Since a given solvent system.can be crassispina (sea urchin), KITALA~Aet al. employed either in the ascending or in [52] isolated a sulfonoglycolipid from the the descending mode, compounds with a shell by usage of DCCC in combination large range of polarity can be separated. with column chromatography on silica With buffer solution, modification of the gel. pH during the elution is possible [45, 461. Water being a component of the solvent systems employed up to now, DCCC has been applied mainly to gly- Advantages and Limits of the Method cosides and other polar compounds. An example of separation of weakly polar di- DCCC has proved to be an efficient terpene dilactones has been reported and reproducible technique for the puri- [31]. We are currently attempting to find fication and separation of natural pro- solvent systems without water which ducts. Quantities ranging from about 1 form suitable droplets. So far, a solvent mg up to 4 g can easily be handled. Since system which is suited for various kinds there is no solid support which might of separations is formed of CHCI,-Me- cause irreversible adsorption, the sample OH-H,O in different proportions (Table is recovered quantitatively. This is of I). A list of some other solvents for particular interest when isolating biologi- DCCC has been published by OCIHARAet cally active compounds because the acti- al. [20]. Systems containing butanol are vity is frequently lost during the time- working, but the separation time is in- consuming column chromatography creased due to the high viscosity of this process. In contrast to the conventional solvent. counter-current distribution (CCD) me- The operation of a DCCC instrument thod, no shaking is involved and hence, is simple and requires little training. there is no formation of foams; further- When a separation is completed, cleaning more, as there is no space for atmosphe- is made by flushing all the tubings with ric oxygen in the apparatus, aerial oxida- the stationary phase for the next analysis. tion of sensitive compounds can be avo- However, we observed that after several ided. Crude plant extracts can be injected months, by using similar solvent sy- without preliminary purification. This stems, the droplet formation became would cause column contamination if the poor. It was necessary to flush the tu- separation is achieved by preparative bings with an appropriate detergent. The HPLC. It is noteworthy that the solvent operating pressure and the flow rate de- consumption is much smaller than with pend on the solvent system and are about other chromatographic techniques such 5-15 kg/cm' and 8-25 ml/h, respectively. as CCD, preparacive HPLC or open co- Thus, the separation time is relatively lumn chromatography. The resolution is long. An operation, including the filling Droplet Counter-Current Chromatography of the instrument with stationary phase, and HPLC, it is of great help for the pre- quires at least 12-18 h depending on parative scale separation of polar natural the number of columns employed. It can products. last for several days, but should not ex- ceed one week. Solvent systems which References give relatively rapid separations should be selected in order to minimize the risk I. Tanimura, T., J. J. I'isano. Y. Ito and R. L. of sample deterioration. When the samp- Bowman: Science 169. 54 (1970). le contains UV active compounds, conti- 2. Hunt, B. J. and 1'.Rigby: Chem. Ind. (Lon- nuous-flow detection can easily be made don) 1868, (1 967). 3. Still, W. C., M. Kahn and A. Mitra: J. Org. by an appropriate detector. The use of a Chem. 43, 2923 (1978). refractive-index detector is not indicated 4. Hostettmann. K., M. J. Pcttci. I. Kubo and K. as sometimes small amounts of stationary Nakanishi: Helv. Chim. Acta 60, 670 (1977). phase are also eluted, resulting in an un- 5. Ito, Y. and R. I..Bowman: J. Chrornamgr. stable base-line. In the course of our sa- Sci. 8, 315 (1970). 6. Ito, Y. and R. L. Bowman: Anal. Chem. 43, ponins and iridoids separations, the ob- 69A (1971). tained fractions were monitored by 7. Ito. Y.. R. F. Hurst. R. L. Bowman and F,. K. TLC, a commonly used procedure for Achter: Separatton and methods 3. column chromatography analysis. 133 (1974). . Although DCCC is simple and separa- 8. Craig. L. C.. in: R. Alexander and R. J. Block. (eds.), Analytical Methods of Protein Chcmi- tes various rypes of compounds in a pre- stry, Vol. I, pp. 121- 160, Oxford 1960, Pcrga- parative scale with a good resolution, it mon Press. possesses limitations arising from the fact 9. Craig, L. C. and J. Sogn: Methods of Enzy- that the efficiency of the method depends mology. Vol. 43. pp. 32G346, New York entirely upon droplet formation. To 1975, Academic Press. 10. Hecker, E.: VerreilungsverTahren irn Labora- challenge these problems Fro et al. 6, [5, torium, WeinheimlBergctraRe 1955. [Verlag 7,531 devised new systems based only on Chemie]. liquid-liquid partition. The emerging 11. Hostettmann. K., M. Hostrttmann-Knldas technique seems to be the coil planet cen- and K. Nakanishi: J. Chrornatogr. 170, 355 trifuge [53, 541 which is a method of ( 1979). counter-current chromatography em- 12. Hostettrnann, K.. M. Hostettnlann-Kaldas and 0. Sticher: Helv. Chim. Acra 62, 2079 ploying a long column of vertical helical (1979). tubings in a centrifugal field. The separa- 13. Kaldas, M., K. Hostcttmann and A. Jacot-Gu- tion time is thus reduced to several illarmod: Hclv. Chirn. Actd 57. 2557 (1974). hours, the resolution is high, but the in- 14. .Hiller. K-., M. Keiprrr and B. Linzer: Pharma- zie 21, 713 (1966). strumentation is more complicated. 15. Hiller, K, and G. Voist: Pharnlazie 32, 365 DCCC seduces by its simplicity and ( 1977). its economical operation (no expensive 16. Cheung, H. T. 2nd M. C. Fens: J. Cllem. Soc. column packing material and very small C 1047, (1968). solvent consumption). It will not replace 17. Hostettmann, K.: Helv. Chirn. Acra, 63, 606 the usual chromatographic methods, but ( 1980). 18. Hostettmann, K.. M. Hostcttmann-Kalclas is complementary. Used alone or in com- and K. Nakanishi: Hrlv. Chin>. Acta 61. 1990 bination with column chromatography (1978).