Forbeside C, a saponin from Asterias forbesi. Complete structure by nuclear magnetic resonance methods1

JOHN A. FINDLAY AND MAHESH JASEJA Department of Chemistry, University of New Brunswick, Fredericton, N.B., Canada E3B 6E2

AND

JEAN-ROBERT BRISSON Division of Biological Sciences, National Research Council of Carzada, I00 Sussex Drive, Ottawa, Ont., Canada K I A OR6

Received March 23, 1987 This paper is dedicated to the rnernory of Professor Karel Wiesner

JOHN A. FINDLAY, MAHESH JASEJA, and JEAN-ROBERT BRISSON. Can. J . Chem. 65, 2605 (1987). The structure of forbeside C, a saponin from the starfish Asterias forbesi, has been deduced totally by nuclear magnetic

resonance methods applied to the undegraded molecule. Using principally 2D-COSY, HCORR, RCT, and 1D-nOe experiments, forbeside C is shown to be the hydrate of the monosodium salt of 6a-0-{~-~-fucopyranosy~(1+2)-O-~-~-fucopyranosyl(1+4)- O-[~-~-quinovopyranosyl(1+2)]-O-~-~-quinovopyranosyl(1+3)-O-6-deoxy-~-~-xylo-hexos-4-ulopyranosyl}-2OS-hydroxy- 3~-sulfo-oxy-5a-cholest-9(11)-en-23-one, the structure reported previously for ovarian asterosaponin 1.

JOHN A. FINDLAY, MAHESH JASEJA et JEAN-ROBERT BRISSON. Can. J. Chern. 65, 2605 (1987). Appliquant des rnCthodes de la rCsonance magnCtique nucldaire 5 la rnolCcule qui n'a pas CtC dCgradCe, on a determink la

structure du forbeside C, une saponine provenant de 1'Ctoile de rner Asterias forbesi. Faisant principalernent appel 2 des expdriences de COSY-2D, HCORR, RCT et d'eOn-ID, on a dCrnontrC que le forbeside C est l'hydrate du rnonosel de sodium de la O-{~-~-fucopyrannosy1(1+2)-O-~-~-fucopyrannosyl(1+4)-O-[~-~-quinovopyrannosyl(l+2)]-O-~-~-quinovopyranno- syl( l j3)-0-dCsoxy-6 P-D-xylo-hexosulo-4 pyrannosy1)-6a hydroxy-20(S) sulfoxy-3P 5a-cholestkne-9(11) one-23, la structure qui avait CtC rapportCe anttrieurement pour 1'astCrosaponine ovarienne 1 .

[Traduit par la revue]

Recently, we reported (1) the structures of the major saponins, forbeside A and forbeside B, from the starfish Asterias forbesi (Desor). We have now assigned the structure 1 to a third saponin, forbeside C, from the same source. As in the case of the major saponins, the structure of forbeside C has been deduced entirely by nmr methods and corroborated, in part, by FAB mass spectrometric data.

The 'H nrnr spectrum of forbeside C in pyridine-d5/D20 (5: 1) at 500 MHz shows, in the downfield region, five anomeric sugar proton signals as doublets at 6 5.108 (7.9 Hz), 6 5.057 (7.9 Hz), 6 4.997 (8.0 Hz), 6 4.947 (7.7 Hz), and 6 4.838 (7.7 Hz). The high coupling constant values are consistent only with the p configuration for all anomeric linkages. The region 6 3.5-4.5 pprn shows the overlapping resonances of the

oligosaccharide ring protons. Five sugar methyl doublets at 6 1.828 (6.1 Hz), 6 1.770 (6.5 Hz), 6 1.724 (6.1 Hz), 6 1.482 (6.2 Hz), and 6 1.445 (6.2 Hz) attest to the presence of five deoxy sugars. The rest of the signals in the upfield region 6 0.90-2.90 pprn are assigned to the aglycone moiety. The signals of the five aglycone methyl groups are easily recog- nizable, three of them are tertiary (6 0.99 s, 6 0.95 s, 6 1.67 s) and are assigned to C 18, C 19, and C2 1 protons, respectively. Isopropyl doublets appear at 6 0.94 and 6 0.92, corresponding to the C26 and C27 methyls. In addition, doublets are observed at 6 2.95 and 6 2.80 for the C22 methylene protons and two overlapping AB quartets are observed at 6 2.53, corresponding to the C24 protons. A broad signal for the olefinic proton at C11 is clearly seen at 6 5.20 and a rnultiplet at 6 4.89 for the C3a

1 Forbeside C

'NRCC No. 28067.

Can

. J. C

hem

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV O

F L

OU

ISV

ILL

E o

n 10

/08/

13Fo

r pe

rson

al u

se o

nly.

2606 CAN. 1. CHEM. VOL. 65, 1987

TABLE 1. The nrnr assignments of aglycone moiety of forbeside C in pyridine-d5/D20 (5: 1)

Carbon no. I3c shifta 'H shift(^)^

24 53.9 2.53, 2.53 25 24.6 2.22 26 22.6 0.94 27 22.5 0.92

"From comparison with forbeside A and forbeside B data (1).

bFrom 'H-I3C correlation using CHORTLE technique.

proton. These findings indicate the presence of a thomasterol A aglycone moiety (1, 2). This was confirmed by 13C nmr spectroscopy. The 13C signals (Table 1) of the aglycone of forbeside C are virtually coincident with those of forbeside A ( 1 ) and consistent with those of thomasteroside A (3). The 13C nrnr spectrum also shows five anomeric carbon signals at 106.4, 105.8, 104.5, 103.5, and 102.3 ppm and five sugar methyls at 18.0, 18.0, 16.9, 16.6, and 13.4 ppm, consistent with the presence of five deoxy sugars in the oligosaccharide moiety. Distortionless enhancement by polarization transfer (DEPT) (4) analysis showed the absence of methylene carbons (-CH2-) in the region 60-65 ppm. Thus, a preliminary review of nmr data indicated that forbeside C features a pentasaccharide chain attached to thomasterol A at position C6. With the aid of 'H-13C correlation spectra using "CHORTLE (5) technique, all the proton resonances of the thomasterol A aglycone could be identified in the 'H nrnr spectrum of forbeside C. The assignments are listed in Table 1.

The complete structure of the oligosaccharide chain was determined by two-dimensional nuclear proton correlation (2D-COSY) (6), relayed coherence transfer (2D-RCT) (7), one-dimensional nuclear Overhauser enhancement (nOe) (8) difference spectroscopy, spin simulation, and heteronuclear 1 3 C - ' ~ correlation (HCORR) spectroscopy. The carbon and proton chemical shifts and coupling constant values of the oligosaccharide moiety of forbeside C are summarized in Table 2.

The anomeric doublet at 6 4.838 is designated as F' 1, and the

FIG. 1 . 2D-nuclear proton correlation (COSY) spectrum ( 6 0.95- 5.15) of forbeside C.

others are designated as F" 1, Z1, Q 1, and Q' 1 sequentially downfield.

From the 2D-COSY spectrum (Fig. l) , the anomeric proton signal for F' 1 at 6 4.838 is coupled to the F'2 proton signal at 6 4.408 and, continuing this J-connectivity path, it was possible to locate F'3 at 6 4.182, and F'4 at 6 4.081. These assignments are confirmed by the 2D-RCT spectrum (Figs. 2a,b), which also shows that the signal at 6 4.081 (F'4) correlates with the signals at 6 3.927 and 6 1.482. Since the doublet at 6 1.482 belongs to the F'6 methyl, the signal at 8 3.926 is assigned to F'5. In the 2D-COSY spectrum the correlation between the F'4 and F'5 protons is not apparent since the J,,, coupling value is very small. Furthermore, the irradiation of the F' 1 proton signal (6 4.838) in the nOe difference spectrum (Fig. 3) induces substantial enhancements at 6 4.182 (Ff3), 6 3.927 (F' 5) , and 6 3.565, the site of interglycosidic linkage. From the COSY spectrum (Fig. 1) and nOe spectra (Fig. 3), it is clear that the J3 ,4

coupling value is small. From the J values in Table 2, which were refined by spin simulation (Fig. 4), it is evident that F ' must be a 6-fucose unit.

Similarly, from the second anomeric doublet F"l (6 4.947), the J-connectivity pathway can be traced out from the 2D-COSY spectrum (Fig. 1) identifying all the couplings belonging to the sugar network F", i.e., F"2 at 6 4.365, F"3 at S 4.036, F"4 at 6 3.997, F"5 at 6 3.639, and F 6 at S 1.455. The small couplings as observed from F 3 / F U 4 , F V 4 / F 5 cross peaks and

Can

. J. C

hem

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV O

F L

OU

ISV

ILL

E o

n 10

/08/

13Fo

r pe

rson

al u

se o

nly.

FINDLAY ET AL

TABLE 2. The nrnr assignments and coupling constant data for oligosaccharide portion of saponin forbeside C in pyridine-d5/D20 (5: 1)

'H shiftarb Coupling constant J ( H , H , ~ 13C shiftC Sugar/carbon No. (ppm) (Hz) ( P P ~ ) ( A ) e

"Assignments from 2D-COSY and 2D-RCT experiments. bParameters refined by spin simulations. 'Obtained from 'H-I3C correlation using CHORTLE technique. dAppears after equilibration with pyridine-d5/H20 (5: 1). 'A ppm in comparison with same position without interglycosidic linkage in this molecule.

the J(AB) values of each of the proton couplings derived by spin simulation show sugar F" to be a p-fucose unit.

The irradiation of the F" 1 proton signal (6 4.947) in the nOe difference spectrum (Fig. 3) causes marked enhancements of signals at 6 4.036 (F"3) and 6 3.639 (FU5), and also a substantial enhancement at 6 4.407, showing the site of interglycosidic linkage at the neighbouring sugar unit to be F'2.

From the 2D-COSY spectrum (Fig. 1) one can trace out the J-connectivity pathway starting from Q l at 6 5.057 to Q"6 at 6 1.828, thus assigning all the ring protons in sugar Q . Irradiation at 6 5.057 ( Q 1) (Fig. 3) induces enhancements at 6 4.075 (Q3) and 6 3.675 (Q5) and a marked enhancement at 6 3.961 (Q12), the site of interglycosidic linkage to the neighbouring sugar Q' .

The last anomeric doublet at 6 5.108 (Q' 1) is seen coupled to asignal at 6 3.961 (Q'2) in the 2D-COSY spectrum (Fig. 1) and, continuing this pathway, the protons Q'3, Q'4, Q'5, and Q'6 are identified at 6 3.961,64.165,6 3.565,s 3.997, and6 1.724, respectively. The nOe difference spectrum (Fig. 3) shows enhancement at 6 4.165 (Q'3) and 6 3.997 (Q'5) in agreement with th,: assigned quinovose structure.

In the 2D-COSY spectrum (Fig. 1) the anomeric doublet for sugar Z at 6 4.997 (Zl) is seen to be coupled to the signal at 6 4.328 (22) but this latter proton has no further coupling partner. Also, one can see the remaining methyl doublet at 6 1.770 (26) coupled to the signal at 6 3.993 (Z5), which in turn shows no further coupling partner. The 2D-RCT spectrum (Fig. 2) gives no evidence of proton substitution at 23 and 24.

The proton chemical shifts obtained from the 2D-COSY and 2D-RCT experiments and coupling constant information for various sugar protons were refined by spin simulation (Fig. 4) until excellent agreement between experimental and calculated spectra was obtained (Table 2). This confirmed the absence of signals due to proton substitution at 23 and 24.

The sequencing of the sugars was completed with the help of 1D nOe difference spectroscopy (Fig. 3) by irradiating each of the anomeric signals separately. The nOe data for the oligo- saccharide chain is summarized in Table 3. Thus, the positions of three interglycosidic linkages, their preferred conformations, and the position of attachment of sugar Z to the aglycone were readily revealed from inter-residue nOe. However, irradiation of signal Q'1 did not reveal the location of interglycosidic

Can

. J. C

hem

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV O

F L

OU

ISV

ILL

E o

n 10

/08/

13Fo

r pe

rson

al u

se o

nly.

2608 CAN. J. CHEM. VOL. 65, 1987

FIG. 2. (a) 2D-Relayed coherence transfer (RCT) spectrum (6 3.45- 4.5516 0.85-3.85) of forbeside C. (b) 2D-Relayed coherence transfer (RCT) spectrum (6 3.45-4.5518 4.75-5.25) of forbeside C.

linkage to sugar Z, implying that this site on Z was substituted by either a deuterium (D20) exchangeable hydrogen or no hydrogen. Alternatively, the transglycosidic nOe may be negligible for reasons of conformation. The sequencing of the sugars as deduced from nOe data can be summarized as follows:

Corroboration of the location of the interglycosidic linkages comes from the '3C shift data (Table 2) since the chemical shifts of carbons involved in the interglycosidic linkages invariably occur at lower field than when simply hydroxylated. Thus Q' is seen to be linked at C2 and C4 since their signals, 83.4 and 84.8 ppm, respectively, are substantially downfield from the

I,:, , , ,A 25

C6H

5 . 2 5 . 0 4 . 8 4 . 6 4 . 4 4 . 2 4 . 0 3 . 8 3 . 6 PPH

FIG. 3. The nOe difference spectra for forbeside C,

corresponding signals in Q" (i.e., 76.0 and 75.5, respectively). Similarly F' must be linked at C2 since its chemical shift, 81.6ppm, is 7.3 ppm lower than its counterpart in F . Thus the 13C nmr data fully support this assigned sequence and also allow insight into the'conne~tivit~ of Q' to Z. The signal at 6 92.7 for the fully substituted carbon in the 13C spectrum (pyridine- d5/D20, 5:l) of forbeside C must belong to either 2 3 or 2 4 since all other signals are satisfactorily accommodated. Further- more, it is evident that, in this medium, one of the sugar Z carbons (23 or 24) fails to give rise to a discernible I3C signal.

This dilemma was overcome by assigning the 6 92.7 signal to a gem diol carbon at C4. The inclusion of an extra oxygen atom is in full agreement with the molecular ion from positive ion FAB (fast-atom bombardment) mass spectral data (vide infra). The placement of the gem diol at C4 allows for the exchange of C3H for deuterium via the anticipated C4 gem diol S C4 ketone + water equilibrium in pyridine-d5/D20. In confirmation of these assignments, recovery of the nmr sample followed by equilibration in pyridine-d5/H20 (5: 1) for seven days afforded a 13C nmr spectrum featuring an additional rnethine signal at 90.3 ppm. Therefore the interglycosidic linkage must be at C3, the carbon that gives rise to the elusive 90.3-ppm signal and whose hydrogen is D 2 0 exchangeable. This lower field chemical shift is in accord with expectation.

To determine the relative configuration of C3H in sugar Z, the 2D-COSY spectrum of forbeside C was recorded in pyridine/H20 (5:l) and it showed the 2 3 proton at 6 4.073 coupled with a large (9.0 Hz) coupling constant to 2 2 at 6 4.328. Sugar Z is, therefore, a P-linked 6-deoxy-xylo-4-hexulose unit in hydrated form.

Can

. J. C

hem

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV O

F L

OU

ISV

ILL

E o

n 10

/08/

13Fo

r pe

rson

al u

se o

nly.

FINDLAY ET AL. 2609

'H NHR TABLE 3. The nOe data for forbeside C obtained from nOe difference

spectroscopy

nOe enhancements at 'H shift A

Simulated spin Irradiation Intraresidual Axial-axial Interresidual through

at 'H ( P P ~ ) 1-3 1-5 glycosidic bond to

d -

,I, 4.997 (Zl) 3.993 (C6 of aglycone) 3.93 5.108 (Q' 1) - 3.997 -

5.057 ( Q 1) 4.075 3.675 3.961 (Q'2)

Sugar F' ' 4.838 (F' 1) 4.182 3.927 3.565 (Q'4) 4.947 (F" 1) 4.036 3.639 4.408 (F'2)

Sugar F '

Sugar Z , 2 5

- T I I I I I I

4 . 4 0 4 . 3 0 4 . 2 0 4 . 1 0 4 . 9 9 3 . 9 0 3.BR 3 . 7 R 3 . 6 0 P P H

FIG. 4. Spin simulation of the 'H nmr spectra for the sugar moiety ~f forbeside C.

The positive ion FAB mass spectrum (Fig. 5) of forbeside C exhibits prominent ions at m / z 1303.3 (M + Na), 1281.2 (M + H), 1201.5 (M + H + Na - NaSO,), and 1183.5 (M + H + Na - NaS03 - H20), confirming the elemental composition C57H93Na028S, while other ions provide corroboration of aspects of the oligosaccharide chain as shown in Scheme 1. Thus cleavage of the oligosaccharide moiety furnishes an ion m / z 769, which is subject to further fragmentation corres- ponding to loss of one or two sugars by fission on each side of the glycosidic oxygens.

Thus, assuming D-family sugars, forbeside C possesses structure 1, the hydrate of the monosodium salt of 6a-0-{P- ~-fucopyranosyl(l+2)-O-~-~-fucopyranosyl(l+4)-0-[~-~- quinovopyranosyl(l+ 2)]-0-p-D-quinovopyranosyl(l+3)-0- 6-deoxy-p-~-xylo- hexos-4-ulopyranosy1)-20s -hydroxy -3 p- sulfo-oxy-5a-cholest-9(11)-en-23-one. This same structure has been assigned to ovarian asterosaponin 1 from the starfish Asterias amurensis by Ikegami and co-workers (9) and was deduced principally from examination of various hydrolysis and degradation products. It has also been found by Itakura and

Aglycone - NaHS04

[5 331 --------, 4. - H20

SCHEME 1. FAB+ fragmentation of forbeside C.

Can

. J. C

hem

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV O

F L

OU

ISV

ILL

E o

n 10

/08/

13Fo

r pe

rson

al u

se o

nly.

CAN. J . CHEM. VOL. 6 5 , 1987

Can

. J. C

hem

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV O

F L

OU

ISV

ILL

E o

n 10

/08/

13Fo

r pe

rson

al u

se o

nly.

FINDLAY ET AL. 261 1

Komori (10) in Asterias amurensis (cf.) versicolor Sladen. Very recently we have also isolated 1 from the starfish Asterias vulgaris, erri ill.

Experimental Forbeside C

Forbeside C (30 mg) was obtained via fractionation of crude saponin (3 g) obtained from specimens of Asterias forbesi (Desor) (collected in Passamaquoddy Bay, August 1984) as previously described (1). This saponin was obtained as a glassy solid (mp 180-182°C uncorrected).

Nuclear magnetic resonance spectra The nmr spectra were recorded on a Bruker AM-500 spectrometer at

300 K using forbeside C (25 mg) in 0.5 mL of pyridine-d5/D20 (5: 1) unless otherwise indicated.

All one-dimensional and two-dimensional nmr experiments were performed using the standard pulse sequences provided by Bruker (DISNMRP) . One-dimensional nuclear Overhauser erlhancemerlt difference experiments

The 1D-nOe difference experiments were performed as previously described (1 1). However, instead of irradiation at a single frequency of the whole multiplet, each line in the multiplet was sequentially irradiated 100 ms for a total time of 1 s per multiplet (12). A line broadening factor of 2 Hz was used prior to Fourier transformation to improve signal-to-noise ratio.

2D-COSY and 2D-RCT experimerlrs 2D-COSY and 2D-RCT experiments were performed using the

standard pulse sequences. The initial matrix of 256 X 2048 data points over a sweep width of 5.4 ppm in both dimensions was zero-filled to 1024 X 2048 points for a final resolution of 2.6 Hz/point. Sine-bell functions were used for resolution enhancement. Magnitude calcula- tions and symmetrization about the diagonal were used to represent the data. For the two-step RCT experiment, a delay of 32 ms was used.

The heterorluclear ';'c-'H correlation The heteronuclear I3C-'H correlation was done by a least-squares

analysis of the I3C intensities as a function of proton evolution times according to Pearson (5). Five evolution times of 0.16, 0.4, 1 .O, 2.4, and 3.2 ms were used.

Spin simulatiorls The spin simulations were done using the Bruker programme

PANIC. A linewidth of 0.5 Hz was used in all cases.

2Larry Calhoun and John A. Findlay, unpublished results

Fast-atom bombardtnerlt (FAB) mass spectra The FAB mass spectra were obtained on VG 7070E and VG ZAB-SE

mass spectrometers employing xenon as an ionizing gas, accelerating voltages of 5 and 10 kV, respectively, and "MAGIC BULLET" was used as matrix (1 3).

Acknowledgements The authors are grateful to Professor K. L. Rinehart,

University of Illinois at Urbana-Champaign, for providing mass spectra and to Larry Calhoun, Department of Chemistry, University of New Bmnswick, for helpful discussions on aspects of 2 D nmr. This study was funded with a grant from the Natural Sciences and Engineering Research Council of Canada.

1. JOHN A. FINDLAY, MAHESH JASEJA, D. JEAN BURNELL, and JEAN-ROBERT BRISSON. Can. J . Chem. 65, 1384 (1987).

2. M. HONDA and T. KOMORI. Tetrahedron Lett. 27, 3369 (1986). 3. Y. ITAKURA, T. KOMORI, and T. KAWASAKI. Liebig's Ann.

Chem. 1983,2079-2091 (1983). 4. D. M. DODDRELL, D. T. PEGGY, and M. R. BANDALL. J. Magn.

Reson. 48, 323 (1982). 5. G. A. PEARSON. J . Magn. Reson. 64, 487 (1985). 6. A. BAX. Two-dimensional NMR in liquids. Riedel, Boston.

1982; A. BAX and R. FREEMAN. J . Magn. Reson. 44,542 (1981); A. BAX, R. FREEMAN, and G. MORRIS. J . Magn. Reson. 42, 164 (1981).

7. G. EICH, G. BODENHAUSEN, and R. R. ERNST. J . Am. Chem. Soc. 104,3731 (1982); A. BAX andG. DROBNY. J. Magn. Reson. 61, 306 (1985).

8. J. K. M. SANDERS and J . D. MERSCH. Progress in nuclear magnetic resonance spectroscopy. Vol. 15. Edited by J. W. Emsley, J. Feeney, and L. H. Sutcliffe. Pergamon Press, Oxford. 1983. p. 353.

9. K. OKANO, T. NAKAMURA, Y. KAMAIYA, and S. IKEGAMI. Agric. Biol. Chem. 45, 805 (1981).

10. Y. ITAKURA and T. KOMORI. Liebig's Ann. Chem. 1986, 359 (1986).

11. J.-R. BRISSON and J. P. CARVER. J. Biol. Chem. 258, 1431 (1983).

12. D. MENHAUS. J. Magn. Reson. 53, 109 (1983); M. KINNS and J. K. M. SANDERS. J. Magn. Reson. 56, 518 (1984).

13. J. L. WHITTEN, M. H. SCHAFFER, M. O'SHEA, J. C. COOK, M. E. HEMLING, and K. L. RINEHART, JR. Biochem. Biophys. Res. Commun. 124, 350 (1984).

Can

. J. C

hem

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV O

F L

OU

ISV

ILL

E o

n 10

/08/

13Fo

r pe

rson

al u

se o

nly.