Results and Discussion
To get a first insight into E. anomala metabolites, the ethanolic extract of
its leaves was profiled by HPLC/HRMS². These analytical fingerprints
were then preprocessed following the feature-based molecular networking workflow
[8 ]. The dereplication of the extract
against the spectral libraries hosted by the GNPS [9 ] afforded a dozen of hits, spreading across different phytochemical
classes, which were overall consistent with former reports related to Euphorbiaceae
plants. At first, seven hits related to C -glycosylflavones, comprising
entries with one or two sugar moieties, could be observed in the molecular network
(See S1 Supplementary Material ). The soundness of these hits and their
consistency with tandem mass spectrometric data will be discussed in a dedicated
section, but the detection of such widespread metabolites is not surprising in
Euphorbiaceae which were already known for producing such phytochemicals [10 ]. The second structural series tentatively
identified corresponded to a collection of peptide derivatives related to
N -benzoylphenylalanine and/or N -benzoylphenylalaninol
differing in their acetylation status and in their intermonomeric bonding. Although
uncommon in plants and most often reported from fungal source, all the dereplicated
compounds were already reported from Euphorbiaceae or Phyllanthaceae plants:
N -benzoylphenylalaninol, aurantiamide,
O -Acetyl-N -(N ′-benzoyl)-phenylalanylphenylalaninol,
asperphenamate, and asperglaucide. The last tentative assignment corresponded to
bisresorcinol. Although euphorbiaceous plants are not known as alkylresorcinol
producers, it is worth underlining that Wagner et al. reported on orcinol glycosides
in Phyllanthus klotzschianus Müll.Arg. [11 ]. Notwithstanding the rather scarce
taxonomic support to this last tentative tagging, the structures isolated during the
subsequent phytochemical investigation provide, to a certain extent, some credit to
this annotation.
Considerable efforts were drawn towards deciphering the mass spectrometric
fragmentation of flavonoid glycosides and perhaps more saliently yet to
C -glycosylflavones [12 ], which are not
amenable to sugar hydrolysis and are thus more difficult to elucidate, especially
prior to the wide scale deployment of 2D NMR sequences [13 ]. The differentiation between C -6 and
C -8 monoglycosylated flavone isomers proved especially tricky, and was
formerly based on the product ions specific ratio obtained following tandem mass
spectrometry, in relation to differential hydrogen bonding instigated between the
hydroxyl groups of flavonoid ring A and the hydroxyl groups of the sugar. In the
current study, these MS/MS landmarks seem to be taken into account by the
molecular networking processing since two isomeric nodes, distinguished thanks to
MZmine 2 data preprocessing [14 ], were tagged
as vitexin (C -8 glucosyl apigenin) and isovitexin (C -6 glucosyl
apigenin) as depicted in the cluster in [Fig.
1 ]. The MS/MS spectra recorded from these two precursors were
examined to determine whether these assignments were supported by comparison with
formerly reported tandem mass spectrometric data. At first, the different intensity
ratios of the
0,2 X+ /0,1 X+ signals
in the low energy tandem mass spectrum (Collision Energy=30 eV) of
the [M+H]+ ion of the two molecules of this isomeric pair
were strongly reminiscent of those reported by Waridel and co-workers [15 ], supporting the tentative assignment
proposed by the molecular networking workflow. Likewise, the inspection of the
MS² spectra of these protonated molecules obtained at higher collision
energy (CE=50 eV) highlighted formerly reported differences between
C -6 and C -8 glycosylated flavones in the intensity ratio of the
fragment ions at m/z 283.06 and 284.07 which again validated the
GNPS-provided tagging [16 ].
Fig. 1 Cluster disclosing the nodes annotated as vitexin and
isovitexin by the GNPS and analytical scores of their relative matches.
a and b . Low and high-energy (CID = 30 and 50 eV,
respectively) MS/MS spectra of the ion at m/z 433.1145 tentatively
tagged as vitexin (retention time: 20.4 min). c and d . Low and
high-energy MS/MS spectra of the ion at m/z 433.1145 tentatively
tagged as isovitexin (retention time: 21.0 min). The emphasized peaks refer
to key signals as outlined in the discussion.
The same cluster revealed two further nodes that were tentatively identified as
swertisin (C -6 glucosyl 7-O -methylapigenin), hinting that a similar
isomeric pair of C -6 and C -8 isomers comprising this
O -methylated aglycone occurred in the plant. These two nodes instigated edges
with both vitexin and isovitexin, but each of them displayed a higher cosine with a
specific member of the vitexin/isovitexin pair. Gratifyingly, the node, at
m/z 447.1303 ([Fig. 2 ]) having the
more elevated cosine score and the bigger number of shared peaks with the swertisin
MS² spectrum, comprised in the GNPS libraries, disclosed a higher similarity
with isovitexin (m/z 433.114, cosine ~ 0.70)
compared to vitexin (m/z 433.114,
cosine ~ 0.64), supporting its likely C -6
glycosylated status. Conversely, the other swertisin-tagged node (m/z
447.1302,) disclosed a higher degree of similarity with vitexin, hinting its
plausible C -8 glycosylation and therefore rather identifying it as
isoswertisin, a compound not uploaded to the GNPS repositories at the time of this
study. The examination of the MS/MS spectra, related to this isomeric pair,
validated this assumption which revealed the same diagnostic patterns of peaks,
shifted by 14 atomic mass units. Within a Wessely-Moser isomeric pair, it is worth
underlying that the C -8 glycosylated derivative was eluted before its
C -6 analogue, in our chromatographic conditions, as found for vitexin and
isovitexin.
Fig. 2 Cluster disclosing the two nodes annotated as swertisin by the
GNPS and analytical scores of their relative matches. a and b .
Low and high-energy MS/MS spectra (CID = 30 and 50 eV, respectively)
of the first ion at m/z 447.1302 tentatively tagged as swertisin, but
rather identified as isoswertisin based on manual inspection (retention
time: 22.5 min). c and d . Low and high-energy MS/MS spectra
(CID = 30 and 50 eV, respectively) of the second ion at m/z
447.1302 tentatively tagged as swertisin (retention time: 24.0 min). The
emphasized peaks refer to key signals as outlined in the discussion.
The molecular network revealed a second flavone glycoside-containing cluster that
included nodes with higher molecular masses that were evocative of flavone
derivatives with two sugar moieties. One of these nodes was tentatively identified
as apigenin 6,8-digalactoside through the molecular-networking based dereplication
process. During the phytochemical work-up of E. anomala , this compound could
be isolated and high resolution mass spectrometry and NMR analyses identified this
molecule as vicenin-II, an apigenin-6,8-di-C- hexoside derivative differing
from the GNPS assignment [17 ]
[18 ] by the nature of the sugar units. Further
nodes could correspond to C -6,8-diglycosylated flavones with different sugar
units at these two sites. In such cases, the type of sugar substituted at the
C -6 and C -8 positions could be determined based on fragmentation
differences in the product ion spectra of isomeric pairs in which the sugars are
swapped, in a comparable way to what is observed with
mono-C -glycosylflavones. Two such precursors pairs could be observed in the
HPLC/DAD/MS² run: a first one at m/z 565.16
and another at m/z 579.17 ([Fig.
3 ]). Notwithstanding an important degree of overall similarity between the
MS² spectra obtained from two isomeric precursors, some diagnostic signals
could here as well lead to assign a tentative identity to these compounds.
Fig. 3 Low-energy MS/MS spectra (CE = 30 eV) of the ions at
m/z 565.16 tentatively identified as a 6-C-pentosyl-8-C-
hexosyl-apigenin (retention time: 19.0 min) (a) and a
6-C-hexosyl-8-C-pentosyl-apigenin (retention time: 20.5 min) (b) and
possible candidates for the signals at m/z 579.17 (c) (retention
time: 21.2 min) and (d) (retention time: 22.7 min). Note that the
methylation site of the flavone component is depicted arbitrarily.
Regarding the precursor ions at m/z 565.16, the fragment ions at m/z
451.11, 457.11 and 469.12 were reported to be diagnostic of schaftoside
(i. e.
6-C -β-D-glucopyranosyl-8-C -α-L-arabinopyranosylapigenin)
as these were not observed throughout the fragmentation of isoschaftoside
(i. e.
6-C -α-L-arabinopyranosyl-8-C -β-D-glucopyranosylapigenin)
[19 ]. These key signals were observed in
the tandem mass spectrum of the isomer eluted at 20.5 min ([Fig. 3b ]), while none of these appeared in the
MS² spectrum of the isomer detected at 19.0 min ([Fig. 3a ]), which supported the identification
of the former as a 6-pentosyl-8-hexosyl apigenin, and of the latter as a
6-hexosyl-8-pentosyl apigenin derivative. A further MS/MS landmark of
interest to discriminate there between is related to the trend of the C -6
sugar substituent to undergo a greater degree of fragmentation compared to the sugar
residue located at C -8 [20 ]
[21 ]. For this purpose, when coping with di
C -glycosylflavones comprising both a pentose and a hexose, the
distinction of the isomers consists in comparing the relative abundance of the two
different 0,2 X+ signals arising at m/z
445.11 and m/z 475.11 in both MS/MS spectra. Following this
approach, the comparatively higher 0,2 X+ signal refers
to the sugar located at C -6. In our case, the m/z 445 to
m/z 475 intensity ratio appeared higher in the spectrum of the
compound eluting at 20.5 min ([Fig.
3b ]), which was evocative of a hexosyl residue at C -6 and therefore
further supported a schaftoside-type substitution pattern, in line with our former
assumptions.
The second isomeric pair appearing in this cluster was shifted by 14 atomic mass
units compared to the first one (m/z 579.17), for a determined
elemental composition of C27 H30 O14 . Interestingly,
these compounds displayed pronounced similarities in their MS² spectra with
a specific member of the former isomeric pair, consistently with their relative
retention time, with the first eluted compound (retention
time=21.2 min) resembling that of isoschaftoside while the second
eluted molecule (retention time=22.7 min) disclosed more pronounced
similarities with schaftoside spectral data. The aforementioned diagnostic signals
were now shifted by 14 atomic mass units to be detected at m/z
465.12, 471.12 and 483.13 in the MS² spectrum of the compound eluted at
22.7 min ([Fig. 3d ]), in line with
the 0,2 X+ signals now being detected at
m/z 459.12 and m/z 489.12 in both MS/MS
spectra, with compared intensity ratios similar to those outlined in the tandem mass
spectrometric data of the first isomeric pair. These spectroscopic features hinted
that these compounds should correspond to schaftoside and isoschaftoside derivatives
having their apigenin component substituted by a methyl group. When searching this
molecular formula in the Dictionary of Natural Products (DNP), 135 hits were
found among which more than 100 entries corresponded to flavonoids! Manually
refining these different flavonoids led to discard most of these propositions, so
that only one compound displayed the structural elements outlined so far,
viz. a 6,8-C -diglycosylflavone with a methylated apigenin core:
6-C -glucosyl-8-C -arabinosylgenkwanin [22 ]. We therefore decided to focus the isomeric
species deemed to correspond to the 6-pentosyl-8-hexosyl methylated flavone.
The last remaining node in this C -glycosylflavone-containing cluster
corresponded to a molecular formula of C27 H28 O16
(m/z 609.1816). Compared to the other bidesmosic flavones
described so far, the MS² spectrum obtained from this ion showed a
Yn
+ ion at m/z 447 that hinted the
existence of a O -glycoside while also exhibiting a variety of fragment ions
related to cross-ring cleavages of saccharidic residues, similar to those formerly
observed in the tandem mass spectrum of swertisin/isoswertisin, evocative of
a C -glycosidic residue. These spectroscopic features led us to surmise that
this compound should be a C ,O -diglycosylated flavone. The search of
the generated molecular formula against the DNP database [23 ] revealed 10 different hits (accessed
14th February 2021) among which nine were flavonoids. Yet, the
flavonoids matching the elemental composition query either corresponded to
O -diglycosylated flavones or to 3-methylglutaroylated
O -glycosylflavones, which seemed to be inconsistent with the putative
structural elements deduced from the cursory examination of the MS² data.
Inferring that this compound had important chances to be unknown, we decided to
target it for isolation and full structure elucidation.
The MS² landmarks outlined earlier made us gear our isolation efforts towards
some specific compounds prioritized due to their presumed structural novelty.
Accordingly, the isolation of the prioritized compounds, detected at
m/z 609.1816 and at 579.1713, could be successfully performed
using a suite of chromatographic separations. Due to this study being the first
undertaken on E. anomala , we wished to complete this targeted pipeline with a
more traditional, untargeted, phytochemical workflow. The isolation of the main
constituents of the ethanolic extract shed light on structural classes not formerly
identified in the molecular network. The structures of these compounds could be
deduced from their spectroscopic data, including 1D and 2D NMR experiments and
comparison with reported data as syringaresinol [24 ], lariciresinol [25 ],
tanarifuranonol [26 ], vitexin [27 ], trematin [28 ], isocytisoside [29 ], and
orcinol. The first three compounds were all known from Euphorbiaceae plants [26 ]
[30 ]
while orcinol seems to be a first report in this plant family. Gratifyingly, this
traditional phytochemical pipeline also resulted in the isolation of two new
compounds which both revealed an orcinol-type substitution pattern, anomaline
(3 ) and diorcinol (4 ) ([Fig.
4 ]).
Fig. 4 Structures of new compounds isolated in the present work.
Anomaloflavone (1 ) was isolated as an optically active amorphous solid with a
molecular formula of C28 H32 O15 (m/z 609.1816
[M+H]+ , calcd. for
C28 H33 O15 , 609.1814), as deduced from HRESIMS
measurements. Consistent with the tandem mass spectrometric observations which
hinted a C ,O -diglycosylated flavone, the 1 H and
13 C NMR data ([Table 1 ])
indicated a C -6 (or C -8)- substituted flavone comprising two anomeric
sites resonating at (δ
H
4.68/δ
C 72.5) and at
(δ
H 4.36/δ
C 104.1)
that were respectively diagnostic of their C -glycosylated and
O -glycosylated status, respectively [13 ]
[31 ]. The other sugar signals
appeared overlapped between δ
H 3.07 and 3.71. The
1 H NMR spectra revealed an olefinic signal at
δ
H 6.87/δ
C 103.5
which was typical of a H-3 flavone signal, an aromatic signal at
δ
H 6.55/δ
C 93.6 that
confirmed the pentasubstituted status of ring A, two doublets at
δ
H 8.04 and 7.11 (each 2H, d,
J= 9.0 Hz), typical of a A2 B2 spin
system of ring B of 4´-substituted flavones and a methoxy group at
δ
H 3.86 (3H, s). The location of the methoxy group was
established based on the correlations from the protons resonating at
δ
H 3.86 and C-4´ (δ
C
163.3) and from the aromatic protons at δ
H 7.11
(H-3´/H-5´) and at δ
C 8.04
(H-2′/H-6′) to C-4´. Altogether, these data
determined the aglycone component of anomaloflavone as acacetin [32 ]. The 13 C NMR spectrum revealed
12 sugar signals ascribable to two hexose units. The cautious analysis of the
13 C NMR chemical shifts of C-6 (δ
C 108.4)
and C-8 (δ
C 93.6) were reminiscent of those of C -6
glycosylated apigenin/acacetin derivatives such as isovitexin [31 ], paraquinins A-C [33 ], or C -6 glycosylated flavones from
Anthurium versicolor
[34 ] whereas C -8 glycosylated flavones
display 13 C chemical shift values around δ
C
99.0 and δ
C 105.0 for C-6 and C-8, respectively [34 ]
[35 ].
The HMBC correlations observed between both the anomeric proton detected at
δ
H 4.68 (H-1′) and the hydrogen-bonded
hydroxyl group at δ
H 13.59 (OH-5) to C-5 and C-6 confirmed
that the disaccharide was bonded by a C -glycosidic linkage at C-6 ([Fig. 5 ]). The proton sequence of the inner
sugar could be deduced from the COSY spectrum from H-1′ to H-6′ but
the multiplicities and coupling constants could not be clearly readable in the
spectra recorded in DMSO-d6 (S4, Supplementary Material ). A second
hexosyl unit could also be determined and the lack of any 13 C NMR
glycosidation shift confirmed it being the terminal unit. The 13 C NMR
data revealed a glycosidation shift at C-3′ (δ
C
90.5) that established the (1->3) linkage between the glycosyl units. A HMBC
correlation from the anomeric proton at δ
H 4.36
(H-1′) to C-3′ (δ
C 90.5) confirmed this
interglycosidic linkage. To determine the coupling constants related to the sugar
components of anomaloflavone, new NMR spectra were acquired in CD3 OD
(S9-S14, Supplementary Material ). The large coupling constants of the
anomeric protons
(J
H-1′-H-2′ =10.0 Hz and
J
H-1′-H-2′ =7.9 Hz) determined
the β-configuration of both sugar moieties. Subsequently, the axial
orientation of the remaining sugar protons H-2′ to H-5′ and
H-2′ to H-5′ could be deduced from the large magnitude of their
coupling constant values, as collated in [Table
1 ]. These data confirmed the carbohydrate part of anomaloflavone as
laminaribiose, in excellent agreement with the NMR data formerly reported by
Hamburger and co-workers in both CD3 OD and
DMSO-d
6
on closely related C -glycosylflavones
[36 ]. Accordingly, anomaloflavone was
determined as acacetin
6-C -[β-d -glucopyranosyl-(1->3)-β-d -glucopyranoside].
Fig. 5 Key COSY and HMBC correlations of anomaloflavone (1) .
Table 1
1 H and 13 C NMR Data for
Anomaloflavone (1) (500/125 MHz, at 298 K)
Anomaloflavone
a
Anomaloflavone
b
no.
δ
H , mult. (J in Hz)
δ
C , type
δ
H , mult. (J in Hz)
δ
C , type
2
163.1, C
166.0, C
3
6.87, s
103.5, CH
6.65, s
104.5, CH
4
182.0, C
184,2, C
5
160.8, C
162.3, C
6
108.4, C
109.1, C
7
163.4, C
165.1, C
8
6.55, s
93.6, CH
6.52, s
95.3, CH
9
153.6, C
158.9, C
10
103.4, C
105.2, C
1′
122.7, C
124.5, C
2′
8.04, d (7.5)
128.3, CH
7.94, d (8.4)
129.2, CH
3′
7.11, d (7.5)
114.6, CH
7.09, d (8.4)
115.6, CH
4′
162.3, C
164.5, C
5′
7.11, d (7.5)
114.6, CH
7.09, d (8.4)
115.6, CH
6′
8.04, d (7.5)
128.3, CH
7.94, d (8.4)
129.2, CH
1′′
4.68, d (8.0)
72.5, CH
4.94, d (10.0)
74.9, CH
2′′
4.32, dd (8.0)
68.6, CH
4.42, dd (10.0, 9.5)
71.6, CH
3′′
3.38, m
90.5, CH
3.65, dd (9.5, 8.8)
90.1, CH
4′′
3.25, m
69.0, CH
3.58, dd (8.8)
70.0, CH
5′′
3.25, m
81.4, CH
3.45, m
82.8, CH
6′′a
3.72 dd (11.5, 5.0)
61.1, CH2
3.90, dd (12.0, 2.1)
63.0, CH2
6’’b
3.42, m
3.74, dd (12.0, 5.4)
1′′′
4.36, d (8.0)
104.1, CH
4.62, d (7.9)
105.4, CH
2′′′
3.10, t (8.0)
73.8, CH
3.31, ov.
75.6, CH
3′′′
3.18, t (8.0)
76.1, CH
3.38, t (8.8)
78.8, CH
4′′′
3.06, t (8.0)
70.1, CH
3.31, ov.
71.5, CH
5′′′
3.22, m
76.9, CH
3.36, m
78.1, CH
6′′′
3.42, m
61.3, CH2
3.66, dd (12.1, 5.5)
62.7, CH2
3.69, dd (11.5, 3.5)
3.90 (12.1, 2.1)
4′CH3
3.86, s
55.5, CH3
3.89, s
56.2, CH3
5-OH
13.59, s
Spectra were recorded in
a
DMSO-d
6 ;
b
CD3 OD.
Anomalovicenin (2 ) was isolated as the sought-after feature with a precursor
mass at m/z 579.1713, confirming its molecular formula to be
C27 H30 O14 (calcd. for
C27 H31 O14 , 579.17083) presumably corresponding
to a 6-C -pentosyl-8-C -hexosyl based on its fragmentation pattern.
Consistent with the MS² based-hypothesis developed earlier, the
1 H NMR spectrum of 2 recorded in DMSO-d
6 (S17,
Supplementary Material ) indicated that this compound was a
C -6,8-diglycosylated flavone at the first glance. The spectroscopic data
indeed indicated the presence of five different aromatic protons, four of which
pertained to an A2 B2 spin system, with H-2´ and
H-6´ sorting out as two doublets detected at δ
H
8.15 and 8.07 (accounting for 2H overall, each d, J= 8.4 Hz),
and H-3´/H-5´ also being detected as two doublets at
δ
H 7.13 and 7.08 (2H overall, d,
J= 8.4 Hz) while the flavone signal H-3 still resonated as a
broad singlet at δ
H 6.88 (1H, br s). Likewise, a cursory
examination of the 13 C NMR data revealed a more or less obvious twinning
of most A-ring signals and of sugar-related signals, with the occurrence of four
different anomeric protons, inconsistent with the elemental composition of 2 .
The 13 C NMR assignments deduced from the HSQC spectrum for the anomeric
carbons resonated around δ
C 75 ppm confirmed their
MS²-awaited C -glycosylated status. The HMBC cross-peak of the protons
at δH 3.86 to the carbon resonating at
δ
C 163.2 located a methoxy group at C-4´, in
line with the predicted methylated status of the aglycone part of this compound.
Exchange-broadened signals at ambient temperature are often observed for
C -glycosylflavones [37 ]
[38 ]. This phenomenon is related to the
coexistence of two rotamers about the bond tethering C-6 to the anomeric carbon of
its C -linked sugar, especially in 6-C -glycosylated flavones displaying
bulky substituents at C-7 or C-8 [39 ]. In this
context, NMR acquisitions at higher temperatures can be conducted to improve
spectral resolution and simplify NMR signal patterns, guiding us to attempt Variable
Temperature (VT-NMR) experiments (S16, Supplementary Material ). NMR spectra
acquired at 373 K displayed sharp signals that allowed for the complete
assignment of the 1 H and 13 C NMR data related to the sugar
residues. COSY and HSQC sequences, in particular, determined a xylopyranose and a
glycopyranose residue, based on their diagnostic chemical shifts and on the elevated
magnitude of their readable coupling constants ([Table 2 ]). Attachment sites of the two sugars could be determined owing
to the long-range heteronuclear correlations outlined in [Fig. 6 ]. These HMBC correlations confirmed the
linkage of the xylose residue to C-6, consistent with tandem mass spectrometric data
which hinted the occurrence of a pentosyl residue at this specific position. The
NOESY crosspeaks observed between the protons of the glycosyl residue and the
aromatic proton H-2′/H-6′ validated the C-8 location of this
residue. Altogether, these spectroscopic data supported the elucidation of 2 ,
namely anomalovicenin, as disclosed in [Fig.
4 ]. The d absolute configuration of β-Glc and
β-Xyl were assumed as those naturally occurring in
C -glycosylflavones.
Fig. 6 Key COSY, HMBC, and NOESY correlations for anomalovicenin
(2) .
Table 2
1 H and 13 C NMR Data for
Anomalovicenin (2) (DMSO-d
6
at 373 K,
500/125 MHz).
(2 )
a
no.
δ
H , mult. (J in Hz)
δ
C , type
2
163.1, C
3
6.76, s
102.7, CH
4
181.9, C
5
159.3, C
6
109.1, C
7
161.1, C
8
105.1, C
9
155.3, C
10
103.0, C
1′
123.0, C
2′
8.06, d (8.3)
128.2, CH
3′
7.11, d (8.3)
114.3, CH
4′
162.1, C
5′
7.11, d (8.3)
114.3, CH
6′
8.06, d (8.3)
128.2, CH
Xyl-1′′
4.67, d (8.0)
74.1, CH
Xyl-2′′
3.95, m
70.6, CH
Xyl-3′′
3.24, t (8.2)
78.7, CH
Xyl-4′′
3.50, m
69.5, CH
Xyl-5′′
3.17, d (10.0)
70.0, CH2
Glc-1′′′
4.95, d (8.0)
74.2, CH
Glc-2′′′
3.71, m
71.0, CH
Glc-3′′′
3.43, t (8.9)
78.1, CH
Glc-4′′′
3.51, m
71.5, CH
Glc-5′′′
3.44, m
81.2, CH
Glc-6′′′
3.66, m
61.3, CH2
4′CH3
3.87, s
55.2, CH3
5-OH
13.71, s
The detailed structure elucidation of the new compounds 3 and 4 is
provided in the Supplementary Material (S23 and S30).
Some antiseptic claims in the ethnopharmacological use of E. anomala
[2 ], along with the documented antibacterial
activity of some flavones [40 ]
[41 ], guided us to evaluate the antibacterial
activities of anomaloflavone (1 ) and anomalovicenin (2 ) on different
bacteria. Anomalovicenin (2 ) was inactive on all tested bacteria.
Anomaloflavone (1 ) showed a significant antibacterial activity against the
Gram-positive S. aureus and on Mycobacterium smegmatis , with a
25 μg/mL MIC on both bacteria. However, it was inactive on
the Gram-negative Pseudomonas aeruginosa . Owing to anomaloflavone being
bactericidal against S. aureus (MBC=MIC), it is tempting to infer
that anomaloflavone (1 ) failed to be active against P. aeruginosa due
to its hydrophobic outer membrane. Nevertheless, as anomaloflavone inhibits the
growth of M. smegmatis too, notwithstanding its waxy cell wall composed of
mycolic acids, it cannot be ruled out that specific target(s) could also be missing
in P. aeruginosa . Interestingly, it is worth noting that anomaloflavone
(1 ) revealed no cytotoxicity
(IC50 >100 μg/mL) towards the human SiHa
cell line. The positive controls used were vancomycin (MIC value of
1 μg/mL against S. aureus ), cetrimide (MIC value of
0.1 μg/mL against P. aeruginosa ), rifampicine (MIC
value of 0.1 μg/mL against M. smegmatis ), and
orlistat (IC50=75 μg/mL in SiHa cell line).
The anticipation performed in this paper, is based on a MN strategy that clustered
C -glycosylated flavones as two distinct families, annotated as such
thanks to a few hits against the GNPS spectral libraries. Yet, annotation
propagation carried out here did not rely on computer-based derivatizations.
Instead, the cursory curation of the MS² spectra related to non-dereplicated
nodes allowed us to extend the GNPS-provided annotations based on the observation of
diagnostic tandem mass spectrometric features extrapolated from signal patterns of
dereplicated nodes.
The C -glycosylflavone emphasis of the current work seems ideally suited for
this approach as these structures, very difficult to elucidate based on NMR studies,
had their fragmentation pathways thoroughly discussed, especially through the
salient contributions of Magda Claeys group [12 ]. However, the benefits retrieved from the early MS²-based
recognition of structural details related to 1 and 2 lie way beyond a
mere facilitation of their subsequent structure elucidation. A significant feature
rendering the early curation of MS² data of peculiar interest is that the
newly reported structures have a common elemental composition, owing to the highly
isomeric nature of flavonoids (9 flavonoids displaying the same formula than
1 and more than 100 for 2 ) so that their sole elemental
composition would have deterred us from isolating them. The early in-depth curation
of their MS² data allowed us to refine these different hits to ascertain
their structural novelty and thus exerted a crucial influence on the conducted
phytochemical investigation.
The accuracy of the MS²-based predictions is worth being underlined as
anomaloflavone turned out to be a C ,O -diglycosylated flavone, as
foreseen. The extent of information retrieved from anomalovicenin prior to its
isolation is perhaps more impressive yet, as the correct location sites of the
pentosyl and hexosyl residues could have been predicted. An inherent limitation of
this strategy is the difficulty to predict the nature of the sugar components which
is, generally speaking, a vexing problem in mass spectrometry [42 ]. From a naturalistic standpoint, we herein
propose the first chemical report on a Erythrococca plant species, that also
led to describe two new phenolic structures.
Interestingly, anomaloflavone (1 ) revealed a significant antibacterial
activity against S. aureus and M. smegmatis while failing to exert a
cytotoxic activity against SiHa cell line.
Materials and Methods
General experimental procedures
The NMR spectra were recorded on a Bruker AM-400 (400 MHz), and on a
Bruker Avance III 500 MHz spectrometer (Bruker, Bremen, Germany). NMR
spectrometers were calibrated using solvent residual signals as references.
Analytical HPLC runs were carried out using an Agilent LC-MS system comprising
an Agilent 1260 Infinity HPLC hyphenated with an Agilent 6530 ESI-Q-TOF-MS
operating in positive polarity. Open-column chromatography separations were
performed on silica gel (40–63 μm, Kieselgel, Merck) at
atmospheric pressure or with Sephadex LH-20 (25–100 μm,
Pharmacia Fine Chemicals, USA). Sunfire preparative C18 columns
(150×30 mm, i. d. 5 μM, Waters) were used
for HPLC separations using a Waters Delta Prep (Waters Co., Milford, MA, USA)
consisting of a binary pump (Waters 2525) and a UV-visible diode array detector
(190–600 nm, Waters 2996). TLC analyses were carried out on
precoated silica gel 60F254 (Merck) and spots were visualized by
spraying with 1% sulfuric vanillin.
Plant Material
Erythrococca anomala (Juss. ex. Poir.) Prain whole plants were collected
in Anepe area, Adzope, Ivory Coast (GPS E: 405353.98, N: 675123.91) in May 2017.
The botanical identification was performed at the Centre National de Floristique
(CNF) – Université Félix Houphouët Boigny
– Abidjan (Côte d’Ivoire). A voucher specimen (No.
OAT-EA-2018) has been deposited at the herbarium of CNF.
Chromatographic and mass spectrometric analysis
Samples were analyzed using an Agilent 6530 Accurate Mass Q-TOF coupled with a
1260 Agilent Infinity LC system fitted with a Sunfire C18 column
(150 mm×2.1 mm; i. d. 3.5 μm, Waters,
Milford, MA, USA) with a flow rate at 0.25 mL/min. Full scan
mass spectra were acquired in the positive polarity in a m/z 100
to 1200 mass range. Capillary temperature, source voltage and sheath gas flow
rate were set at 320 °C, 3.5 kV, and 10 L/min,
respectively. Capillary, fragmentor, and skimmer voltages were set at
3500 V, 175 V and 65 V, respectively. Four scan events
were used: positive MS, mass range encompassing m/z
100–1200, and three data-dependent MS/MS scans of the five most
intense ions from the first scan event. Three collision energies
(i. e. 30, 50, and 70 eV) were used for MS/MS
data generation. Purine (C5 H4 N4 ,
m/z 121.050873) and HP-0921 (hexakis (1 H , 1 H ,
3 H -tetrafluoropropoxy)-phosphazene
C18 H18 F24 N3 O6 P3 ,
m/z 922.009798) were used as internal lock masses. Full scans
were acquired at a resolution of 10.000 (m/z 922) and 4.000
(m/z 121). A permanent MS/MS exclusion list criterion
was established to prevent oversampling of the internal calibrant.
MZmine data processing and Molecular Networking Parameters
The MS/MS data files were converted from the .d in-house Agilent data
format to .mzXML using the MS Convert software, as included in the ProteoWizard
package [43 ]. The .mzXML files were then
processed using the MZmine 2 v.52 workflow. The mass detection was
performed using a threshold of 1.0E4. The ADAP chromatogram builder was obtained
using a minimum group scan size of 4, a group intensity threshold of 1.0E4, a
minimum highest intensity of 1.0E5 and m/z tolerance of
0.05 Da or 5 ppm [44 ]. The
ADAP wavelets deconvolution algorithm was performed using the following
settings: S/N threshold=10, minimum feature height=3,
coefficient/area threshold=3, peak duration
range=0.02–4.0 min, RT wavelet
range=0.02–0.6. MS/MS scans were paired using a
m/z tolerance of 0.02 Da and a RT tolerance of
1.0 min. The chromatogram was deisotoped using the isotopic peak grouper
algorithm with a m/z tolerance of 0.02 Da or 10 ppm and a
RT tolerance of 1.0 min. The .mgf preclustered data file and the
associated .csv feature quantification table (for RT, areas, and formula
integrations) were further exported using the dedicated built-in options. A
molecular network was finally created with MS-Cluster Off, a parent ion mass
tolerance of 0.05 Da and a fragment ion mass tolerance of
0.05 Da to generate consensus spectra. Consensus spectra containing less
than one spectrum were discarded. Edges were filtered to have a minimal cosine
score of 0.6 with at least six matched fragment peaks and further maintained if
both nodes appeared in each other top 10 most similar nodes. Matches against the
GNPS spectral libraries required a minimal cosine of 0.6 and at least six
matched fragment peaks. The generated molecular networks were visualized using
Cytoscape 3.5.1 [45 ].
Extraction and purification of compounds
Dried leaves of E. anomala (3.0 kg) were milled and extracted by
maceration with EtOH (6×10 L, 24 h each, room
temperature, atmospheric pressure). The EtOH extract was concentrated in
vacuo at 40°C to afford 494 g of dry residue. An aliquot
of 200 g of this dry residue was dissolved in distillated water
(1000 mL) and this solution was submitted to extraction with
n -hexane (400 mL×7) to afford a n -hexane fraction
(EAH) (109 g). The aqueous phase was then alkalinized using 6 M
NH4 OH (until a pH ca. 12 was reached) prior to being re-extracted
with methylene chloride (300 mL×4) to afford 110 g of
dry residue (subfraction EAD1) and finally with EtOAc (300 mL×6)
(subfraction EAE1, 879 mg). A relatively similar extraction process was
repeated on another aliquot of the crude ethanolic extract (160 g),
differing from the former by lacking the alkalinization step. In brief, this new
aliquot of crude ethanolic extract was dissolved in distillated water
(1000 mL) to be extracted with DCM (400 mL×3) to afford
a second DCM fraction (EAD2) (112 g). The aqueuous phase was
subsequently reextracted with EtOAc (250 mL×5) (EAE2,
2 g) and at last with n -BuOH (150 mL×4) (EAB,
7.1 g).
The dichloromethane subfraction EAD1 was fractionated by silica gel column
chromatography using a gradient mobile phase consisting of a DCM/EtOAc
solvent system (stepwise gradient 10:0 to 9:1) to afford 12 fractions
EAD1A-EAD1L. Fraction EAD1C (11.5 g) was selected for further
purification by normal-phase silica gel column chromatography using an isocratic
regime consisting of n -hexane/acetone (6 :4) to afford
subfractions EAD1C1-EAD1C5. Subfraction EAD1C2 (1.2 g) was selected for
chromatography on Sephadex LH-20 to be isocratically eluted with a
DCM/MeOH (2:1) mobile phase to afford compound 3
(3.2 mg). Likewise, subfraction EAD1C5 (2.3 g) was selected for
chromatography on Sephadex LH-20 to be isocratically eluted with a
DCM/MeOH (2:1) mobile phase to afford syringaresinol (13.4 mg).
Fraction EAD1G (9 g) further fractionated by normal-phase silica gel
column chromatography using an isocratic regime consisting of
n -hexane/acetone (6:4) to afford subfractions EAD1G1–4.
Subfraction EAD1G4 (1.7 g) was subjected to column chromatography on
Sephadex, and isocratically eluted with a n -hexane/acetone (6:4)
mobile phase to afford lariciresinol (1.9 mg). Fraction EAD1I
(2.9 g) was selected for further purification by normal-phase silica gel
column chromatography using an isocratic regime consisting of
n -hexane/acetone (6:4) to afford subfractions EAD1I1–6.
Subfraction EAD1I2 (1.5 g) was subsequently purified by chromatography
on Sephadex LH-20 using an isocratic regime consisting of a
n -hexane/EtOAc (7:3) mobile phase to afford tanarifuranonol
(1.1 mg).
The subfraction EAE1 (879 mg) was separated by normal-phase silica gel
column, eluted with a solvent system of DCM/MeOH (stepwise gradient,
99:1 to 9:1) to afford 12 fractions (EAE1A-EAE1L). The subfraction EAE1F
(250 mg) was subsequently fractionated by normal-phase silica gel column
chromatography using an isocratic regime consisting of
n -hexane/acetone (5:5) to afford subfractions EAE1F1-EAEF4. Among
these, EAE1F4 (120 mg) was finally chromatographed on Sephadex LH-20
using an isocratic DCM/MeOH (2:1) mobile phase to afford orcinol
(60 mg). The subfraction EAE1H (200 mg) was selected for further
purification by normal-phase silica gel column chromatography using an isocratic
regime consisting of n -hexane/acetone (6:4) to afford
subfractions EAE1H1-EAEH3. Among these, EAE1H2 (51 mg) was selected for
further purification by normal-phase silica gel column chromatography using an
isocratic regime consisting of n -hexane/acetone (7:3) and was
fractionated by HPLC separation using a gradient system of ACN-H2 O
with 0.1% HCOOH (9:1 to 7:3) to afford trematine (61 mg) and
isocytisoside (1.7 mg).
The subfraction EAE1J (87 mg) was selected for further purification by
normal-phase silica gel column chromatography using an isocratic regime
consisting of n -hexane/acetone (6:4) to afford subfractions
EAE1J1-EAEJ5. Among these, EAE1J3 (40 mg) was subsequently
chromatographed on Sephadex LH-20 using an isocratic DCM/MeOH (2:1)
mobile phase to afford 4 (3 mg).
The subfraction EAB (7.1 g) was subjected to silica gel column
chromatography using a EtOAc/MeOH (99:1 to 9:1) giving 11 fractions
EAB1–11. Fractions EAB9 (930 mg), EAB10 (890 mg), and
EAB11 (730 mg) were all fractionated by HPLC separation using a gradient
system of ACN-H2 O with 0.1% HCOOH (9:1 to 7:3) to
respectively afford vitexine (25.2 mg), and 1
(17.2 mg).
Anomaloflavone (1 ). Yellow amorphous solid;
[α]D
25 –33.3 (c 0.031, in MeOH);
UV (MeOH) λ
max (log ε ) 349 (4.31), 272
(4.17), 243 (4.15) nm ; 1 H NMR (500 MHz, MeOH-d4
and DMSO-d6 ) see [Table 1 ],
13 C NMR (125 MHz, MeOH-d4 and
DMSO-d6 ) see [Table 1 ],
HRESIMS m/z 609.1816 [M+H]+ (calcd. for
C28 H33 O15 , 609.1814). MS/MS
spectrum was deposited in the GNPS spectral libraries under the identifier
CCMSLIB00005788115.
Anomalovicenin (2 ). Yellow amorphous solid;
[α]D
25 +22.7 (c 0.044, in MeOH);
UV (MeOH) λ
max (log ε) 333 (4.29), 273 (4
.16), 227 (4.12) nm; 1 H NMR (400 MHz, DMSO-d6 at
373 K) see [Table 2 ],
13 C NMR (125 MHz, MeOH-d4 and
DMSO-d
6
) see [Table 2 ], HRESIMS m/z 579.1713
[M+H]+ (calcd. for
C27 H31 O14 , 579.17083). MS/MS
spectrum was deposited in the GNPS spectral libraries under the identifier
CCMSLIB00005788116.
Anomaline (3 ). White amorphous solid. UV (MeOH)
λ
max (log ε ) 280 (3.71) nm;
1 H NMR (400 MHz, MeOH-d4 ) see Table S1 ,
13 C NMR (100 MHz, MeOH-d4 ) see Table S1,
HRESIMS m/z 183.1025 [M+H]+ (calcd. for
C10 H15 O3 , 183.1016). MS/MS spectrum
was deposited in the GNPS spectral libraries under the identifier
CCMSLIB00005788117.
Diorcinol (4 ). White amorphous solid. UV (MeOH)
λ
max (log ε ) 280 (3.71) nm;
1 H NMR (400 MHz, MeOH-d4 ) see Table S2 ,
13 C NMR (100 MHz, MeOH-d4 ) see Table S2,
HRESIMS m/z 261.1134 [M+H]+ (calcd. for
C15 H17 O4 , 261.1121). MS/MS spectrum
was deposited in the GNPS spectral libraries under the identifier
CCMSLIB00005788118.
Antibacterial activity
A broth microdilution method was used to determine the MIC values of the isolated
compounds (1 –2 ) against bacteria. The MIC values of the
samples were determined using Staphylococcus aureus (ATCC 6538),
Pseudomonas aeruginosa (ATCC 15442) and Mycobacterium
smegmatis (CIP 7326). In brief, 100 µL of the pure
compounds (1 mg/mL) dissolved in DMSO were added to
100 µL of bacterial culture in Mueller Hinton medium in
triplicate and serially diluted twofold in 96-well microtiter plates. The plates
were incubated overnight at 37°C, and the MIC value was determined as
the lowest concentration inhibiting bacterial growth. To further discriminate a
bactericidal effect from a bacteriostatic effect, 10 µL of the
bacterial suspension were inoculated onto a on Tryptic Soya Agar plate, and the
bacterial growth was assessed after 24 h of incubation. Positive control
drugs were vancomycin hydrochloride (Cayman Chem, purity≥95%)
for Staphylococcus aureus , cetrimide (Sigma, purity≥95%)
for Pseudomonas aeruginosa , and rifampicin (Sigma,
purity≥97%) for Mycobacterium smegmatis .
Cytotoxicity assessment
Cytotoxicity of the compounds was investigated on the human cervical cancer cell
line, SiHa, in a MTT assay. Briefly, cells (5.104 /mL) were
seeded in a 96-well plate at 100 µl/well and incubated
at 37 °C with 5% C02 for 24 h. Two-fold
serial dilutions of compounds (in same culture DMEM-10% bovine serum)
were added (100 µl/well) to the adherent SiHa cells and
further incubated for 3 days at 37 °C. MTT reagent
(0.5 mg/mL final concentration) was added to each well and
4 h later plate was washed twice with PBS (phosphate buffer saline).
Formazan crystals were dissolved in 100 µl/well DMSO and
absorbance was measured at 570 nm and 630 nm using microplate
reader (Bio-Rad 680). The results were expressed as the percentage of optical
density (OD) values (% of cell viability) relative to the optical
density (OD) values from the control cells, without compounds (set to
100% viability). Positive control drug was orlistat (Cayman Chem,
purity≥98%).
