A.J. Pérez et al. / Phytochemistry Letters 13 (2015) 59–67
63
Consistent with the above mentioned ESI-MS/MS analysis, the
signal at 4.97 revealed a typical spin system of a -glucopyr-
anosiduronic acid unit, which was confirmed by the presence of a
carboxylic carbon signal in 13C NMR spectrum at
172.5 (C-6GlcA
and by its long-range HMBC correlations with H-4GlcA 4.40, dd,
J = 9.6, 9.7 Hz) and H-5GlcA 4.61, d, J = 9.7 Hz). Inefficient
magnetization transfer beyond H-2Rha 4.72, dd, J = 1.6, 3.4 Hz)
in 1D-TOCSY (120 ms) for the anomeric signal at 6.14 was
observed. This finding, together with the presence of a methyl
doublet at 1.70 (J = 6.2 Hz), is characteristic of a rhamnopyranosyl
unit. A further selective TOCSY experiment on this methyl (H-6Rha
allowed a sequential assignment of the signals from H-6Rha to
H-2Rha. The anomeric configuration of the rhamnose unit was
deduced from the H-1Rha/C-1Rha direct coupling value
(J = 173.7 Hz), measured from F1-coupled HSQC spectrum
(Agrawal, 1992). Based on the correlations observed in HSQC
and HSQC-TOCSY spectra of 1, the 13C signals for each sugar unit
were assigned. The sequence and linkage sites into the sugar chain
and with the aglycone moiety were established by inter-glycosidic
HMBC/ROESY correlations (Fig. 2), which were observed between
577 [M – H]–) showed the most abundant fragments at m/z 487 and
457, deriving from the loss of 90 and 120 Da, which is typical for the
fragmentation of a hexose unit attached to the aglycone through a
CꢀꢀC bond (Vukics and Guttman, 2010). Minor fragments at m/z
503 and 473 were also observed, corresponding to the loss of
74 and 104 Da due to the cleavage of a deoxyhexose unit CꢀꢀC
bound to the aglycone (Vukics and Guttman, 2010). On the other
hand, the ESI-MS/MS spectrum of 4 (m/z 577 [M – H]–) showed
almost the opposite behavior. The most abundant fragments were
observed at m/z 503, 473 and 457. The latter ([M – H – 120]–), is also
typical in the fragmentation of a deoxyhexose unit (Vukics and
Guttman, 2010), and was the base ion peak due to the contribution
of both sugar units. Moreover, other important ions that could be
observed in both 3 and 4, were those due to the aglycone (A = 270,
apigenin) plus the residues of the sugars that remain linked to it,
i.e. A + 113 (m/z 383) and A + 83 (m/z 353) from the loss of 194 and
224 Da, respectively (Ferreres et al., 2003; Vukics and Guttman,
2010). Similarly, the ion [M – H – 18]–was detected in both
compounds, but it was more abundant in 4. Therefore, it can be
hypothesized that the structure of 3 corresponds to 6-C-hexosyl-8-
C-deoxyhexosyl apigenin, while 4 could be the positional isomer of
3, namely 6-C-deoxyhexosyl-8-C-hexosyl apigenin.
d
b
d
)
(
d
(d
(d
d
d
)
a
H-1Rha
and C-2GlcA
90.1)/H-3 (
ROESY correlations between H-1Glc
3.76) of the aglycone moiety, confirmed that 1 is a bidesmosidic
(d
6.14) and C-2Gal
78.4)/H-2GlcA
3.22) of the aglycone moiety. In addition, the HMBC/
4.79) and C-22 ( 82.4)/H-22
(
d
76.4)/H-2Gal
(
d
4.51), H-1Gal
(d 5.55)
(d
(d
4.46), H-1GlcA
(d
4.97) and C-3 (d
d
The 1H NMR spectrum of 3 and 4, in C2D6SO (Table 2), showed
(d
d
signals at
6.68 (s, H-3), which are characteristics of the apigenin aglycone.
This, together with their carbon signals at 181.9/181.8 (C-4),
d
7.96/7.95 (d, H-20,60), 6.92/6.95 (d, H-30,50), and 6.75/
(d
saponin. A d configuration for glucuronic acid, galactose and
glucose units and l for rhamnose unit were established after the
acid hydrolysis of 1, followed by derivatization of sugars and UPLC-
UV-SRM/MS analysis (Tanaka et al., 2007; Pérez et al., 2014). Based
on the evidence outlined above, the structure of 1 was finally
d
158.9/159.2 (C-5), 161.1/160.7 (C-7), and 161.1/160.7 (C-40)
confirmed the apigenin skeleton bearing a carbonyl and three
hydroxyl groups. Substitution at C-6 and C-8 was inferred, as no
corresponding aromatic proton signals were observed at these
positions. In addition, 1H NMR spectrum of 3 displayed two
determined as 3-O-[
a
-l-rhamnopyranosyl-(1 !2)-
b-d-galacto-
-d-glucopyr-
pyranosyl-(1 !2)-
b
-d-glucuronopyranosyl]-22-O-
b
anomeric signals at
HSQC spectrum with carbons at
1D TOCSY, DQF COSYand 1D ROESY spectra, they were identified as
being due to a -glucopyranosyl unit and a 6-deoxy- -glucopyr-
anosyl unit ( -quinovose). Long-range correlations in the HMBC
spectrum from H-1Glc 4.76) to C-6 ( 107.9) and C-5 ( 158.9) and
from H-1Quin 4.82) to C-8 ( 103.1) and C-9 ( 154.8) indicated
d
4.76 and 4.82, which were correlated in the
anosyl sophoradiol and it was named abyssaponin A.
d
74.0 and 74.0, respectively. Using
The molecular formula of compound 2 was determined to be
the same as that of 1, by HRESI-TOFMS analysis (m/z 1087.5664
[M ꢀ H]–, calcd for C54H87O22, 1087.5689). It also contains the same
aglycone moiety as 1 (sophoradiol), according to their NMR
spectroscopic data (Table 1), which were almost superimposable.
Although the ESI-MS/MS spectrum of 2 showed fragmentation
pattern identical to 1, their NMR data for the sugar portion
(Table 1) showed differences. The 1H NMR spectrum of 2 exhibited
b
b
b
(
d
d
d
(d
d
d
that the C-6 and C-8 positions of the apigenin were bonded to a
glucose and a quinovose, respectively (Fig. 2). Thus, the structure of
3 was identified as 6-C-
syl apigenin and was named abyssinoside A.
Likewise, 1H NMR spectrum of 4 showed two anomeric signals
4.78 and 4.94, which were identified as being due to a
-quinovopyranosyl unit and a -glucopyranosyl unit, respective-
ly. Consistent with the aforementioned MS/MS analysis, the long-
range HMBC correlations from H-1Quin 4.78) to C-6 ( 107.9) and
C-5 ( 159.2) and from H-1Glc 4.94) to C-8 ( 103.4) and C-9 (
154.1) suggested a connectivity opposed to that of 3. Therefore, the
structure of 4 was elucidated as 6-C- -quinovopyranosyl-8-C-
-glucopyranosyl apigenin and was named abyssinoside B.
The molecular formula of compound 5 was determined to be
b-glucopyranosyl-8-C-b-quinovopyrano-
four anomeric signals at
d 4.79, 4.96, 5.67 and 6.24. However, the
latter two appear approximately at 0.1 ppm downfield shifted
compared with 1. The selective TOCSY and ROESY experiments on
at
d
these two signals, revealed the presence of a
5.67) instead of -galactopyranosyl and a
unit ( 5.67). A rigorous study of the HMBC/ROESY correlations led
to the sequence of the sugar chain as being the same as in 1. Thus,
the structure of 2 was identified as 3-O-[ -l-rhamnopyranosyl-
-d-glucuronopyranosyl]-
-d-glucopyranosyl sophoradiol and was named abyssapo-
b
-glucopyranosyl unit
b
b
(d
b
a
-rhamnopyranosyl
d
(d
d
d
(d
d
d
a
(1 !2)-
b
-d-glucopyranosyl-(1 !2)-
b
b
22-O-
b
b
nin B.
Compounds 3 and 4 were found to be structural isomers, giving
identical cationized molecules [M + Na]+ in their HRESI-TOFMS
corresponding to the molecular formula C27H30O14Na. Their UV
spectra showed absorption maxima at 225, 270 and 335 nm,
characteristic for flavones and similar to those of apigenin (Mabry
et al., 1970). Their ESI-MS/MS spectra showed typical fragmenta-
tion patterns of di-C-glycosyl flavones (Ferreres et al., 2003; Vukics
and Guttman, 2010), however certain differences were observed.
Taking into account that the fragmentation of 6,8-di-C-glycosyl
flavones occurs preferentially at position 6 in relation to the sugar
unit at C-8 (Ferreres et al., 2003), a comparative study of the
relative abundance of the fragment ions in both isomers allowed us
to draw important conclusions regarding the type of sugar and
position of the C-glycosylation. The ESI-MS/MS spectrum of 3 (m/z
C27H30O13 according to the HRESI-TOFMS spectrum (m/z 585.1573
[M + Na]+, calcd for C27H30O13Na, 585.1584). Its UV spectrum
showed absorption maxima at 220 and 335 nm, characteristic of
7,40-dihydroxyflavone (Mabry et al., 1970). The ESI-MS/MS
spectrum of 5 (m/z 561 [M – H]–) showed a fragmentation pattern
typical for
a mono-C-glycosylated flavone. The presence of
fragment ions at m/z 325 [(M – H – 146) – 90] and 295 [(M – H –
146) – 120]– indicated that 5 is a glycoside with hexose attached
through a CꢀꢀC bond to aglycone core and a deoxyhexose unit
connected through O-glycosidic bond to hexose. The fact that ion
[(M – H – 146) – 120]– was more abundant (base peak) than [(M –
H – 146) – 90] –implied that C-8 is the glycosylation position
(Ferreres et al., 2003). An additional fragment ion at m/z 267
[(M – H – 146) – 148]– was observed, which belong to the aglycone