Phytochemistry, Vol. 31, No. 3. pp. 1009 1011, 1992 Printedin Great Britain.
6
KAEMPFEROL-3-O-GLUCOSYL(l-2)RHAMNOSIDE AND A REAPPRAISAL OF OTHER GLUCO(l-2, STRUCTURES KENNETH
R. MARKHAM,
HANS GEIGER*
0031-9422/92 $5.00+0.00 1992 PergamonPress plc
FROM GINKGO BILOBA l-3 AND l-4)RHAMNOSIDE
and H.
JAGGYt
DSIR-Chemistry, Private Bag, Petone, New Zealand; *FB13, Botanik, Universitiit des Saarlandes, D-6600, Saarbriicken, Germany; tFirma Dr Willmar Schwabe, 7500 Karlsruhe 41, Germany (Received 11 June 1991)
Key Word Index --Ginkgo biloba; Ginkgoaceae; Epimedium wushanense; E. pubescens; Selaginelh doederleinii; flavonol-3-0-glucopyranosyl(l-2, 1-3 and l+hamnopyranosides; structure revision; kaempferol-3-0-a-L-[Bglucopyranosyl(l-2)rhamnopyranoside].
Abstract-A
kaempferol-3-O-glucorhamnoside from Ginkgo biloba is defined as the 3-O-a-L-[fl-D-glucopyranosyl(lZ)rhamnopyranoside] on the basis of 2D NMR evidence. Complete assignments of the ‘H and 13C NMR spectra of this compound and of its known p-coumaroyl derivative are presented for the first time. The NMR distinctions of l-2, l-3 and l-4 linked glucopyranosylrhamnopyranosides are discussed and indicate(i) that the 13C NMR assignments for one published gluco(l-3)rhamnoside are in need of modification, (ii) that the published structure of hordenine-0-[6-0t-cinnamoyl-fi-glucosyl(l-4)+rhamnoside] from Selaginella doederleinii is not distinguished from the l-3 linked glucorhamnoside structure, and (iii) that the 8-prenylkaempferol-3-O-[~ucosyl(1-4)rhamnoside]-7-O-glucoside and the equivalent 4’-0-methylated xylosyl( l-4)rhamnoside from Epimedium pubescens and E. washanense, respectively, are (1-2)-linked.
INTRODUCTION
In the course of recent research which necessitated the determination of the interglycosidic linkage in a new compound, quercetin-3-glucorhamnoside-7-glucoside from Latkyrus chrysanthus petals [ 11, the NMR support data cited previously for known flavonol glucorhamnosides were re-examined. In all, seven examples were found. Originally proposed structures included, one 1-2 linked [2], two l-3 linked [3,4] and four l-4 linked [S-9] glucorhamnosides, excluding the L. ckrysanthus compound. Two of these l-4 linked compounds (both isolated from Ginkgo biloba and identified [S, 73 as kaempferoland quercetin-3-0-a-L-[6-p-coumaroyl+D-glucopyranosyl( 1-4)rhamnopyranoside]), have recently been shown to contain 1-2 interglycosidic linkages [IO]. This conclusion was based in part on the NMR spectra of the peracetylated derivatives which gave clearer distinction of the rhamnose proton signals for decoupling experiments. In practice however, there is often insufficient material available for the preparation of such derivatives in a spectroscopically pure state. The present paper describes an NMR study and structure determination of an underivatized kaempferol-3-O-glucorhamnoside from Ginkgo biloba for the first time. In addition, the structures published for other glucorhamnosides are re-evaluated and the NMR criteria for distinguishing between 1-2, l-3 and l-4 linked glucorhamnosides are discussed. IWWLTS AND DISCUSSION
Only two reference compounds were available to us in the structural investigation of the Lathyrus glucorhamno-
side [l]. These were the kaempferol-3-O-[6-p-coumaroylfi-glucosyl( 1-2)-a-rhamnoside] and a kaempferol-3-0glucorhamnoside which had both been isolated earlier by one of us (H. J.) from Ginkgo biloba. NMR studies were undertaken on both samples in order to establish the interglycosidic linkage in the latter. In the 13CNMR spectra of these compounds, the rhamnose carbon bonded to the glucosylated hydroxyl resonated between 681 and 82. The ‘H-‘H COSY spectra for both compounds were used to assign chemical shift values to each proton on the rhamnose carbon chain (Table 1), and ‘H-“C COSY studies enabled correlation of the identified protons with the carbon signals. The glucosylated carbons (at 681.5 and 81.8) in these compounds correlated unambiguously with the C-2 linked rhamnose protons at 64.10 and 4.14, respectively, indicating that the interglycosidic linkage is to the 2-hydroxyl of the rhamnose in both cases. This was supported also by the near identity of the relevant 13C NMR data of the unacylated Ginkgo glycoside with those published [2] for epimedin-A, 8-prenyl-4’0-methylkaempferol-3-0-a-L-[p-D-glucosyl( 1-2)rhamnoside]-7-0-fl-D-glucoside, in DMSO-d, solution (Table 1). The ‘HNMR data (Table 2) also clearly indicate a l-2 linkage, in that the rhamnose H-2 signal appears at ca 0.1 ppm downfield from the equivalent proton in an unsubstituted 3-0-rhamnoside, while the H-3 and H-4 signals are unchanged (Table 2). Even more noticeable is the apparent effect of 2-0-glycosylation on the chemical shift of the rhamcose H-l signal. In 3-0-rhamnosides lacking 2-0-glycosylation, this signal appears in the 65.1-5.3 range [l, 3, 11, 121, whereas in the Ginkgo compounds (and in other 2-0-glycosylated flavonol-3-0rhamnosides-see below) the rhamnose H-l is found
1009
K. R. MARKHAMet al.
1010
Table 1. 13CNMR chemical shifts for the glycosidic moieties of various glucorhamnosides (in DMSO-d,)
R-l R-2 R-3 R-4 R-5 R-6
101.9 70.1 70.4 71.5 70.6 17.3
G-l G-2 G-3 G-4 G-5 G-6
101.1 81.4 70sa 71.7 70.2” 17.4
98.0 69.2” 81.2 70.3” 69.1” 17.4
101.9 70.4” 69.9” 81.9 69.0” 17.5
100.9 81.5 70.3 71.9 70.5 17.6
100.8 81.8 70.3 71.9 70.6 17.6
106.2 73.8 76.7b 69.4” 76.6 b 60.4
105.8 73.8 76.0 70.1” 73.8 63.4
104.7 74.5 76.7 b 69.9” 76.9 b 61.1
106.3 74.1 76.5 69.4 76.9 60.7
106.3 73.9 76.2 70.0 73.9 63.9
98.6 70.1’ 70.4”*’ 80.9’ 69.6” 17.6 104.6 73.7b 76.0 70.3” 73.6b 640
101.1 81.4 70.6” 71.9 70.2” 17.6 105.6 74.0 76.6b 69.5” 76.3’ 60.5
8~cAssignment~ bearing the same superscript may be reversed. *A number of signals have been re-assigned -see Discussion. tChemical shifts as measured (at 75 MHz) and assigned in the present work; all signals were assigned unambiguously by ‘H-‘H and ‘H-13C COSY studies.
Table 2. ‘H NMR chemical shifts for the glycosidic moieties of the Ginkgo biloba flavonol glucorhamnosides* (in DMSO& at 300 MHz) Ginkgo glucorhamnosides ~--
H
Quercetin-3-O-a-Lrhamnoside
-___-__ Acylated
Rha-1 Rha-2 Rha-3 Rha-4 Rha-5 Rha-6
5.25 d (1.1) 4.00 br d (ca 3.1) 3.53 dd (8.8, 3.1) 3.18 dd (9.3, ca 9) 3.24 dq (ca 9.5, ca 6) 0.85 d (5.8)
5.63 brs 4.14 brd (ca 3.4)
5.58 brs 4.10 brd (cn 3)
3.54 mP ca 3.2 mt
ca 3.3 mf 0.90 d (6.4)
3.56 dd (9.5, 3.3) ca 3.15P ca 3.30 dq. (9.6, 6.0) 0.86 d (6.0)
4.33 d (7.8) 3.05 rnt 3.21 mt 3.18 mt 3.34 mt 4.12-4.24 mt
4.25 d (7.8) 3.0 rn? 3.15 mt 3.15 mt 3.02 mt 3.46 brd
Glc- 1 Glc-2 Glc-3 Glc-4 Glc-5 Glc-6
Non-acylated
*Signals were assigned by ‘H-‘H and lH-‘% COSY studies; chemical shift values are in ppm from TMS and J values in Hz are presented in parentheses. Woupling not clearly defined or obscured by overlap.
significantly downfield at 65.5-5.6. On the basis of these data, the structure of the kaempferol glucorhamnoside from Ginkgo is defined as kacmpferol-3-@a-L-[B-Dglucopyranosyl(l-2)rhamnopyranosidel and the struc-
ture of the acylated glucorhamnoside is confirmed. Complete assignments are given for the carbon and proton resonances of the sugar moieties in both acylated and non-acylated compounds for the first time (Tables 1 and
Kaempferol-3-~-~ucosyl(l-2)rh~noside 2). ‘H assignments
were deduced from “H-‘H COSY measurements and these were used to identify 13C signals in the lH--13C COSY spectra. With the number of structurally proven flavonol gluco( l-2)rhamnosides now totalling at least four, including the Ginkgo compounds mentioned above and 8prenyl-4’-O-methylkaempferol-3-O-[glucoy~1-2)rhamnoside]-7-0-glucoside (epimedin-A) from Epimedium ~reunum [ZJ, reliable NMR data are now available for the identi~cation of gluc~l-2)rhamnosides. However, the 13CNMR differences between l-2, l-3 and l-4 linked glucorhamnosides are very small (see Table 1) because the C-2, C-3 and C-4 resonances of rhamnose possess similar chemical shifts. None-the-less it would appear that the 1-2 linked glucorhamnosides can be distinguished by the presence of a signal (rhamnose C-4) at 671-72, this resonance being largely unaffected by glycosylation at C-2. The l-3 and 1-4 isomers, however, give indistinguishable i3CNMR spectra and, therefore, require additional proof of structure such as “H-‘H, with or without rH-13C, connectivity data. The required additional data have been provided for the two I-3 linked examples from Asplenium ~rolo~ff~~rn [3] and Epimed~um ko~ea~um [4] giving confidence that the structures are as proposed. There is a need, however, to reassign some signals in the “C NMR spectrum of the Asplenium compound. This is to account for the effect of the 6-linked feruloyl group on the resonance of the glucose C-5 signal, and to align the rhamnose assignments with published data. The modified assignments are presented in Table 1. The l-4 linked glucorhamnosides are not distinguishable from the 1-3 linked on the basis of 1D 13CNMR spectra alone, For this reason, proposed gluco(l-4~hamnoside structures in two recent publications appear to be in need of re-examination and/or structure revision. The first of these, reports the structure ho~enine-O-~-[6-O-~~n~oyl-~-~u~syl(l~~rhamnoside] for an alkaloid isolated from Sebginella doederleinii [13j. The inter-glycosidic linkage of the saccharide portion of this compound was based on ’ 3C NMR data alone and by comparison of these data with those of the Ginkgo compound above (originally thought to be l-4 linked). Unlike the Ginkgo glyeoside with which it was aligned, this Selaginella glycoside does not appear to be a gluco(l2)rhamnoside, as its spectrum lacks a rhamnose C-4 signal at 671-72. It is probably, therefore, a gluco(l-3) or (l+rhamnoside, but distinction must await further studies. The second example of a gluco(l-4)rhamnoside assigned a suspect structure is 8-prenylkaempferol-3-O-a-L[~-D-glucos~I~)rhamnoside]-7-O-~-D-~u~side from ~p~dium w~hu~~e [8] and E. pubescens [P]. In the i “C NMR spectrum of this compound, a signal at 6 71.9 appears to be mis-assigned to rhamnose C-2. It is almost certainly representative of rhamnose C-4, and when the signals assigned to C-2 and C-4 are interchanged
from Ginkgo bifoba
1011
(Table 1) the spectrum in the sugar carbon region is near identical with those of the flavonol 3-gluco(l2)rh~nosides. Furthermore, the rhamnose H-l signal appears at $5.50 as expected (see above) for a gluco(l2)rhamnoside. On these grounds the E. ~sha~nse/~. pubescens com~und is almost certainly the demethyl equivalent of 8-prenyl-4’-0-methylkaempferol-3-0-N-L[P-D-glucosyl(l-2)rhamnoside]-7-0-j?-D-glucoside previously isolated from ~pimedium ~~ea~~ [2]. Similarly, on the basis of both ‘H and i3C NMR data, the S-prenyl4’-methylkaempferol-3-O-a-~-[a-~-xylosyl(i-4)rhamnoside]-7-0-~-~glucoside from E. w~~nense [8], is also probably the equivalent xylosyl(l-2)rhamnoside. This compound, incidentally, is not new. It has been reported previously from Epimedium koreanurn 123. In summary it is concluded that ~yco(l-2~rhamnosides can be recognized from their ‘HNMR spectra alone by the 0.3-0.4 ppm downfield shift of the readily identified rhamnose H-l signal. In the i3CNMR spectrum, a gluc~l-2)rh~noside is often identi~able from the presence of a (rhamnose C-4) signal in the range 6 71-72. The distinction between l-3 and l-4 glycosylated rhamnosides by NMR, however, requires the appIi~tion of 2D techniques and/or decoupling experiments. Acknowle&eme&-The authors thank Dr Herbert Wong (DSIR-~~~t~) for the measur~ent of the NMR spectra. REFERENCES 1. Markh~, K. R., Hammett, K. R. W. and O&mm, D. J. (1992) Pbytochemistry (in press). 2. Oshima, Y. O., Okamoto, M. and Hikino, H. (1987) Heterocycles 26,935. 3. Mizuno, M., Kyotani, Y., Iimuna, M., Tanaka, T. and Iwatsuki, K. (1990) Phyrochemisrry 29,2742. 4. Pachaly, P., ~hoa~e~-W~~~rth, C. and Sin, K.-S. (1990)
Pla?ita ?&Xl.56,277. 5. Nasr, C., Haag-Berrurier, M., Lobstein-Guth, A. and Anton, R. (1986)Phytoc~emistry2S, 770. 6. Yamasaki, K., Kasai, R., Ma&i, Y., Okihara, M., Tanaka,
7. 8. 9. 10. 11. 12.
13.
O., O&o, H., Takagi, S., Yamaki, M., Masuda, K., Nonaka, G., Tsuboi, M. and Nishioka, I. (1977) Tetrahedron Letters 1231. Nasr, C., Lobstein-Guth, A., Haag-Rem&r, M. and Anton, R. (1987) Phytochemistry 26, 2869. Li, Y.-S. and Liu, Y.-L. (1990) Phytochemistry 29, 3311. Li, Y.-S. and Liu, Y.-L. (1990) J. Nat. Prod. $3, 1337, Kang, S. S., Kim J. S., Kwak, W.-J. and Kim, K.-H. (1990) Arch. Pharm. Res. 13, 207. Takagi, S., Yamaki, M., Masuda K. and Kubota, M. (1976) Ya~g~u Zasshi 96, 1217. Mizuno, M., Iinuma, M., Tanaka, T., Sakakibara, N., Murata, J, Murata, H. and Lang, F. A. (1990) Phytochemistry 29, 1277. Chao, L. R., Seguin, E., Tillequin, F. and Koch, M. (1987) J. Nat. Prod. 50,422.
6
KAEMPFEROL-3-O-GLUCOSYL(l-2)RHAMNOSIDE AND A REAPPRAISAL OF OTHER GLUCO(l-2, STRUCTURES KENNETH
R. MARKHAM,
HANS GEIGER*
0031-9422/92 $5.00+0.00 1992 PergamonPress plc
FROM GINKGO BILOBA l-3 AND l-4)RHAMNOSIDE
and H.
JAGGYt
DSIR-Chemistry, Private Bag, Petone, New Zealand; *FB13, Botanik, Universitiit des Saarlandes, D-6600, Saarbriicken, Germany; tFirma Dr Willmar Schwabe, 7500 Karlsruhe 41, Germany (Received 11 June 1991)
Key Word Index --Ginkgo biloba; Ginkgoaceae; Epimedium wushanense; E. pubescens; Selaginelh doederleinii; flavonol-3-0-glucopyranosyl(l-2, 1-3 and l+hamnopyranosides; structure revision; kaempferol-3-0-a-L-[Bglucopyranosyl(l-2)rhamnopyranoside].
Abstract-A
kaempferol-3-O-glucorhamnoside from Ginkgo biloba is defined as the 3-O-a-L-[fl-D-glucopyranosyl(lZ)rhamnopyranoside] on the basis of 2D NMR evidence. Complete assignments of the ‘H and 13C NMR spectra of this compound and of its known p-coumaroyl derivative are presented for the first time. The NMR distinctions of l-2, l-3 and l-4 linked glucopyranosylrhamnopyranosides are discussed and indicate(i) that the 13C NMR assignments for one published gluco(l-3)rhamnoside are in need of modification, (ii) that the published structure of hordenine-0-[6-0t-cinnamoyl-fi-glucosyl(l-4)+rhamnoside] from Selaginella doederleinii is not distinguished from the l-3 linked glucorhamnoside structure, and (iii) that the 8-prenylkaempferol-3-O-[~ucosyl(1-4)rhamnoside]-7-O-glucoside and the equivalent 4’-0-methylated xylosyl( l-4)rhamnoside from Epimedium pubescens and E. washanense, respectively, are (1-2)-linked.
INTRODUCTION
In the course of recent research which necessitated the determination of the interglycosidic linkage in a new compound, quercetin-3-glucorhamnoside-7-glucoside from Latkyrus chrysanthus petals [ 11, the NMR support data cited previously for known flavonol glucorhamnosides were re-examined. In all, seven examples were found. Originally proposed structures included, one 1-2 linked [2], two l-3 linked [3,4] and four l-4 linked [S-9] glucorhamnosides, excluding the L. ckrysanthus compound. Two of these l-4 linked compounds (both isolated from Ginkgo biloba and identified [S, 73 as kaempferoland quercetin-3-0-a-L-[6-p-coumaroyl+D-glucopyranosyl( 1-4)rhamnopyranoside]), have recently been shown to contain 1-2 interglycosidic linkages [IO]. This conclusion was based in part on the NMR spectra of the peracetylated derivatives which gave clearer distinction of the rhamnose proton signals for decoupling experiments. In practice however, there is often insufficient material available for the preparation of such derivatives in a spectroscopically pure state. The present paper describes an NMR study and structure determination of an underivatized kaempferol-3-O-glucorhamnoside from Ginkgo biloba for the first time. In addition, the structures published for other glucorhamnosides are re-evaluated and the NMR criteria for distinguishing between 1-2, l-3 and l-4 linked glucorhamnosides are discussed. IWWLTS AND DISCUSSION
Only two reference compounds were available to us in the structural investigation of the Lathyrus glucorhamno-
side [l]. These were the kaempferol-3-O-[6-p-coumaroylfi-glucosyl( 1-2)-a-rhamnoside] and a kaempferol-3-0glucorhamnoside which had both been isolated earlier by one of us (H. J.) from Ginkgo biloba. NMR studies were undertaken on both samples in order to establish the interglycosidic linkage in the latter. In the 13CNMR spectra of these compounds, the rhamnose carbon bonded to the glucosylated hydroxyl resonated between 681 and 82. The ‘H-‘H COSY spectra for both compounds were used to assign chemical shift values to each proton on the rhamnose carbon chain (Table 1), and ‘H-“C COSY studies enabled correlation of the identified protons with the carbon signals. The glucosylated carbons (at 681.5 and 81.8) in these compounds correlated unambiguously with the C-2 linked rhamnose protons at 64.10 and 4.14, respectively, indicating that the interglycosidic linkage is to the 2-hydroxyl of the rhamnose in both cases. This was supported also by the near identity of the relevant 13C NMR data of the unacylated Ginkgo glycoside with those published [2] for epimedin-A, 8-prenyl-4’0-methylkaempferol-3-0-a-L-[p-D-glucosyl( 1-2)rhamnoside]-7-0-fl-D-glucoside, in DMSO-d, solution (Table 1). The ‘HNMR data (Table 2) also clearly indicate a l-2 linkage, in that the rhamnose H-2 signal appears at ca 0.1 ppm downfield from the equivalent proton in an unsubstituted 3-0-rhamnoside, while the H-3 and H-4 signals are unchanged (Table 2). Even more noticeable is the apparent effect of 2-0-glycosylation on the chemical shift of the rhamcose H-l signal. In 3-0-rhamnosides lacking 2-0-glycosylation, this signal appears in the 65.1-5.3 range [l, 3, 11, 121, whereas in the Ginkgo compounds (and in other 2-0-glycosylated flavonol-3-0rhamnosides-see below) the rhamnose H-l is found
1009
K. R. MARKHAMet al.
1010
Table 1. 13CNMR chemical shifts for the glycosidic moieties of various glucorhamnosides (in DMSO-d,)
R-l R-2 R-3 R-4 R-5 R-6
101.9 70.1 70.4 71.5 70.6 17.3
G-l G-2 G-3 G-4 G-5 G-6
101.1 81.4 70sa 71.7 70.2” 17.4
98.0 69.2” 81.2 70.3” 69.1” 17.4
101.9 70.4” 69.9” 81.9 69.0” 17.5
100.9 81.5 70.3 71.9 70.5 17.6
100.8 81.8 70.3 71.9 70.6 17.6
106.2 73.8 76.7b 69.4” 76.6 b 60.4
105.8 73.8 76.0 70.1” 73.8 63.4
104.7 74.5 76.7 b 69.9” 76.9 b 61.1
106.3 74.1 76.5 69.4 76.9 60.7
106.3 73.9 76.2 70.0 73.9 63.9
98.6 70.1’ 70.4”*’ 80.9’ 69.6” 17.6 104.6 73.7b 76.0 70.3” 73.6b 640
101.1 81.4 70.6” 71.9 70.2” 17.6 105.6 74.0 76.6b 69.5” 76.3’ 60.5
8~cAssignment~ bearing the same superscript may be reversed. *A number of signals have been re-assigned -see Discussion. tChemical shifts as measured (at 75 MHz) and assigned in the present work; all signals were assigned unambiguously by ‘H-‘H and ‘H-13C COSY studies.
Table 2. ‘H NMR chemical shifts for the glycosidic moieties of the Ginkgo biloba flavonol glucorhamnosides* (in DMSO& at 300 MHz) Ginkgo glucorhamnosides ~--
H
Quercetin-3-O-a-Lrhamnoside
-___-__ Acylated
Rha-1 Rha-2 Rha-3 Rha-4 Rha-5 Rha-6
5.25 d (1.1) 4.00 br d (ca 3.1) 3.53 dd (8.8, 3.1) 3.18 dd (9.3, ca 9) 3.24 dq (ca 9.5, ca 6) 0.85 d (5.8)
5.63 brs 4.14 brd (ca 3.4)
5.58 brs 4.10 brd (cn 3)
3.54 mP ca 3.2 mt
ca 3.3 mf 0.90 d (6.4)
3.56 dd (9.5, 3.3) ca 3.15P ca 3.30 dq. (9.6, 6.0) 0.86 d (6.0)
4.33 d (7.8) 3.05 rnt 3.21 mt 3.18 mt 3.34 mt 4.12-4.24 mt
4.25 d (7.8) 3.0 rn? 3.15 mt 3.15 mt 3.02 mt 3.46 brd
Glc- 1 Glc-2 Glc-3 Glc-4 Glc-5 Glc-6
Non-acylated
*Signals were assigned by ‘H-‘H and lH-‘% COSY studies; chemical shift values are in ppm from TMS and J values in Hz are presented in parentheses. Woupling not clearly defined or obscured by overlap.
significantly downfield at 65.5-5.6. On the basis of these data, the structure of the kaempferol glucorhamnoside from Ginkgo is defined as kacmpferol-3-@a-L-[B-Dglucopyranosyl(l-2)rhamnopyranosidel and the struc-
ture of the acylated glucorhamnoside is confirmed. Complete assignments are given for the carbon and proton resonances of the sugar moieties in both acylated and non-acylated compounds for the first time (Tables 1 and
Kaempferol-3-~-~ucosyl(l-2)rh~noside 2). ‘H assignments
were deduced from “H-‘H COSY measurements and these were used to identify 13C signals in the lH--13C COSY spectra. With the number of structurally proven flavonol gluco( l-2)rhamnosides now totalling at least four, including the Ginkgo compounds mentioned above and 8prenyl-4’-O-methylkaempferol-3-O-[glucoy~1-2)rhamnoside]-7-0-glucoside (epimedin-A) from Epimedium ~reunum [ZJ, reliable NMR data are now available for the identi~cation of gluc~l-2)rhamnosides. However, the 13CNMR differences between l-2, l-3 and l-4 linked glucorhamnosides are very small (see Table 1) because the C-2, C-3 and C-4 resonances of rhamnose possess similar chemical shifts. None-the-less it would appear that the 1-2 linked glucorhamnosides can be distinguished by the presence of a signal (rhamnose C-4) at 671-72, this resonance being largely unaffected by glycosylation at C-2. The l-3 and 1-4 isomers, however, give indistinguishable i3CNMR spectra and, therefore, require additional proof of structure such as “H-‘H, with or without rH-13C, connectivity data. The required additional data have been provided for the two I-3 linked examples from Asplenium ~rolo~ff~~rn [3] and Epimed~um ko~ea~um [4] giving confidence that the structures are as proposed. There is a need, however, to reassign some signals in the “C NMR spectrum of the Asplenium compound. This is to account for the effect of the 6-linked feruloyl group on the resonance of the glucose C-5 signal, and to align the rhamnose assignments with published data. The modified assignments are presented in Table 1. The l-4 linked glucorhamnosides are not distinguishable from the 1-3 linked on the basis of 1D 13CNMR spectra alone, For this reason, proposed gluco(l-4~hamnoside structures in two recent publications appear to be in need of re-examination and/or structure revision. The first of these, reports the structure ho~enine-O-~-[6-O-~~n~oyl-~-~u~syl(l~~rhamnoside] for an alkaloid isolated from Sebginella doederleinii [13j. The inter-glycosidic linkage of the saccharide portion of this compound was based on ’ 3C NMR data alone and by comparison of these data with those of the Ginkgo compound above (originally thought to be l-4 linked). Unlike the Ginkgo glyeoside with which it was aligned, this Selaginella glycoside does not appear to be a gluco(l2)rhamnoside, as its spectrum lacks a rhamnose C-4 signal at 671-72. It is probably, therefore, a gluco(l-3) or (l+rhamnoside, but distinction must await further studies. The second example of a gluco(l-4)rhamnoside assigned a suspect structure is 8-prenylkaempferol-3-O-a-L[~-D-glucos~I~)rhamnoside]-7-O-~-D-~u~side from ~p~dium w~hu~~e [8] and E. pubescens [P]. In the i “C NMR spectrum of this compound, a signal at 6 71.9 appears to be mis-assigned to rhamnose C-2. It is almost certainly representative of rhamnose C-4, and when the signals assigned to C-2 and C-4 are interchanged
from Ginkgo bifoba
1011
(Table 1) the spectrum in the sugar carbon region is near identical with those of the flavonol 3-gluco(l2)rh~nosides. Furthermore, the rhamnose H-l signal appears at $5.50 as expected (see above) for a gluco(l2)rhamnoside. On these grounds the E. ~sha~nse/~. pubescens com~und is almost certainly the demethyl equivalent of 8-prenyl-4’-0-methylkaempferol-3-0-N-L[P-D-glucosyl(l-2)rhamnoside]-7-0-j?-D-glucoside previously isolated from ~pimedium ~~ea~~ [2]. Similarly, on the basis of both ‘H and i3C NMR data, the S-prenyl4’-methylkaempferol-3-O-a-~-[a-~-xylosyl(i-4)rhamnoside]-7-0-~-~glucoside from E. w~~nense [8], is also probably the equivalent xylosyl(l-2)rhamnoside. This compound, incidentally, is not new. It has been reported previously from Epimedium koreanurn 123. In summary it is concluded that ~yco(l-2~rhamnosides can be recognized from their ‘HNMR spectra alone by the 0.3-0.4 ppm downfield shift of the readily identified rhamnose H-l signal. In the i3CNMR spectrum, a gluc~l-2)rh~noside is often identi~able from the presence of a (rhamnose C-4) signal in the range 6 71-72. The distinction between l-3 and l-4 glycosylated rhamnosides by NMR, however, requires the appIi~tion of 2D techniques and/or decoupling experiments. Acknowle&eme&-The authors thank Dr Herbert Wong (DSIR-~~~t~) for the measur~ent of the NMR spectra. REFERENCES 1. Markh~, K. R., Hammett, K. R. W. and O&mm, D. J. (1992) Pbytochemistry (in press). 2. Oshima, Y. O., Okamoto, M. and Hikino, H. (1987) Heterocycles 26,935. 3. Mizuno, M., Kyotani, Y., Iimuna, M., Tanaka, T. and Iwatsuki, K. (1990) Phyrochemisrry 29,2742. 4. Pachaly, P., ~hoa~e~-W~~~rth, C. and Sin, K.-S. (1990)
Pla?ita ?&Xl.56,277. 5. Nasr, C., Haag-Berrurier, M., Lobstein-Guth, A. and Anton, R. (1986)Phytoc~emistry2S, 770. 6. Yamasaki, K., Kasai, R., Ma&i, Y., Okihara, M., Tanaka,
7. 8. 9. 10. 11. 12.
13.
O., O&o, H., Takagi, S., Yamaki, M., Masuda, K., Nonaka, G., Tsuboi, M. and Nishioka, I. (1977) Tetrahedron Letters 1231. Nasr, C., Lobstein-Guth, A., Haag-Rem&r, M. and Anton, R. (1987) Phytochemistry 26, 2869. Li, Y.-S. and Liu, Y.-L. (1990) Phytochemistry 29, 3311. Li, Y.-S. and Liu, Y.-L. (1990) J. Nat. Prod. $3, 1337, Kang, S. S., Kim J. S., Kwak, W.-J. and Kim, K.-H. (1990) Arch. Pharm. Res. 13, 207. Takagi, S., Yamaki, M., Masuda K. and Kubota, M. (1976) Ya~g~u Zasshi 96, 1217. Mizuno, M., Iinuma, M., Tanaka, T., Sakakibara, N., Murata, J, Murata, H. and Lang, F. A. (1990) Phytochemistry 29, 1277. Chao, L. R., Seguin, E., Tillequin, F. and Koch, M. (1987) J. Nat. Prod. 50,422.
