Bitter Taste Impact and Thermal Conversion of a Naringenin Glycoside from Cyclopia genistoides

Article abstract of DOI:10.1021/acs.jnatprod.8b00710

A naringenin derivative, isolated from Cyclopia genistoides, a bitter tasting herbal tea, especially when in green (unoxidized) form, was identified as (2S)-5-[α-l-rhamnopyranosyl-(1a?2)-β-d-glucopyranosyloxy]naringenin (1). The compound partially epimerizes to (2R)-5-[α-l-rhamnopyranosyl-(1a?2)-β-d-glucopyranosyloxy]naringenin (2) when heated at different temperatures (80, 90, 100, 110, and 120 °C) for a prolonged period in a phosphate buffer at pH 5. The fractional conversion model predicted the decrease in the concentration of compound 1 the best. The activation energy of the conversion reaction was calculated as 99.16 kJ mol^-1. Prolonged heating resulted not only in formation of compound 2 but eventually a decrease in its concentration and the formation of another conversion product, (E)-2′-[α-l-rhamnopyranosyl-(1a?2)-β-d-glucopyranosyloxy]-4′,6′,4-trihydroxychalcone (3). In contrast, naringin, glycosylated at C-7, remained stable when heated under the same conditions (100 °C for 6 h at pH 5). The bitter intensity of compound 1 was substantially less than that of naringin, both tested at 0.04 mM, a concentration typical of compound 1 in an herbal tea infusion of green C. genistoides. This comparison indicates that the position of the sugar moiety plays an important role in determining both bitter intensity and heat stability of naringenin glycosides.

Full text of DOI:10.1021/acs.jnatprod.8b00710

Article
pubs.acs.org/jnp
Cite This: J. Nat. Prod. XXXX, XXX, XXX−XXX
Bitter Taste Impact and Thermal Conversion of a Naringenin
Glycoside from Cyclopia genistoides
Ombeline Danton,† Lara Alexander,‡,§ Cindy Hunlun,‡ Dalene de Beer,‡,§ Matthias Hamburger,*,†
and Elizabeth Joubert*,‡,§
†Pharmaceutical Biology, Pharmacenter, University of Basel, Klingelbergstrasse 50, 4056 Basel, Switzerland
‡Plant Bioactives Group, Post-Harvest and Agro-Processing Technologies, Agricultural Research Council (ARC),
Infruitec-Nietvoorbij, Private Bag X5026, Stellenbosch 7599, South Africa
§Department of Food Science, Stellenbosch University, Private Bag X1, Matieland, Stellenbosch 7602, South Africa
S
* Supporting Information
ABSTRACT: A naringenin derivative, isolated from Cyclopia
genistoides, a bitter tasting herbal tea, especially when in green
(unoxidized) form, was identified as (2S)-5-[α-L-rhamnopyr-
anosyl-(1→2)-β-D-glucopyranosyloxy]naringenin (1). The
compound partially epimerizes to (2R)-5-[α-L-rhamnopyrano-
syl-(1→2)-β-D-glucopyranosyloxy]naringenin (2) when heated
at different temperatures (80, 90, 100, 110, and 120 °C) for a
prolonged period in a phosphate buffer at pH 5. The fractional
conversion model predicted the decrease in the concentration
of compound 1 the best. The activation energy of the
conversion reaction was calculated as 99.16 kJ mol−1.
Prolonged heating resulted not only in formation of compound
2 but eventually a decrease in its concentration and the
formation of another conversion product, (E)-2′-[α-L-rhamnopyranosyl-(1→2)-β-D-glucopyranosyloxy]-4′,6′,4-trihydroxychal-
cone (3). In contrast, naringin, glycosylated at C-7, remained stable when heated under the same conditions (100 °C for 6 h at
pH 5). The bitter intensity of compound 1 was substantially less than that of naringin, both tested at 0.04 mM, a concentration
typical of compound 1 in an herbal tea infusion of green C. genistoides. This comparison indicates that the position of the sugar
moiety plays an important role in determining both bitter intensity and heat stability of naringenin glycosides.
itter taste plays an important role in the consumer
acceptance of food and beverages. When inherent to the
naringenin derivative present in grapefruit.2 Beelders et al.6
tentatively identified a naringenin-hexosyloxy-deoxyhexosyloxy
derivative (compound 1) with the same molecular mass as
naringin in C. genistoides. Different elution times on a reversed-
phase HPLC column indicated structural differences between
naringin and compound 1. Furthermore, high-temperature
oxidation, performed at 90 °C for 16 h according to
recommended conditions for optimum aroma and flavor
development of traditional honeybush tea,5 lowered the
concentration of 1 in the plant material and also led to
formation of an isomer (compound 2).6 The objectives of the
present study were therefore to unambiguously elucidate the
structure of the two compounds, herewith identified as the
diastereomers (2S)-5-[α-L-rhamnopyranosyl-(1→2)-β-D-
glucopyranosyloxy]naringenin (1) and (2R)-5-[α-L-rhamno-
pyranosyl-(1→2)-β-D-glucopyranosyloxy]naringenin (2). An-
other conversion product, E-2′-[α-L-rhamnopyranosyl-(1→2)-
β-D-glucopyranosyloxy]-4′,6′,4-trihydroxychalcone (3), is re-
ported here for the first time. Additionally, the bitter intensity
B
sensory quality of the product, it could be one of the drivers of
liking,1 but mostly bitterness influences consumer response
negatively.2 The aversion to bitter taste is considered to be a
defense mechanism against the ingestion of potential poisons.3
The presence of bitter taste receptors in the gut and their role
in the release of gut hormones involved in the control of food
intake, however, emphasize a new role for bitter compounds in
the fight against obesity.4
Cyclopia genistoides Vent. is one of several Cyclopia species
used to produce the herbal tea honeybush. This herbal tea has
a notable bitter taste, contrary to the slightly sweet tasting
herbal teas prepared from other commercially important
Cyclopia species. The bitter taste is especially prominent in
infusions prepared from green C. genistoides, a product
produced by excluding the high-temperature oxidation step
in the manufacture of traditional honeybush tea.5 Insight into
the phenolic composition of C. genistoides and the stability of
its phenolic constituents during thermal processing may help
to elucidate their role in the bitter taste of this herbal tea.
Phenolic compounds elicit a bitter taste in many foods and
beverages. One such compound is naringin, a potent bitter
Received: August 22, 2018
© XXXX American Chemical Society and
American Society of Pharmacognosy
A
DOI: 10.1021/acs.jnatprod.8b00710
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
of compound 1 was determined to assess its potential impact
on the taste of honeybush tea. Heating experiments provided
kinetic data for calculation of the activation energy required for
conversion of compound 1 to 2 in a phosphate buffer at pH 5,
a pH relevant for a “natural” iced tea beverage (no acids
added). Naringin was included in sensory analysis and heat
experiments to enhance current knowledge of the effect of the
position of the sugar moiety on taste and thermal stability of
naringenin derivatives.
CE at 286 nm, the latter being indicative of the (2S) absolute
configuration of the flavanone unit.8,9
The experimental ECD spectrum of compound 2 (Figure 2)
showed two high-amplitude negative CEs at 217 and 235 nm, a
low-amplitude negative CE at 331 nm, and a low-amplitude
positive CE at 283 nm. The latter Cotton effect is reminiscent
of a (2R) absolute configuration of the flavanone moiety.8,9 In
conclusion, compound 1 was assigned as (2S)-5-[α-L-
rhamnopyranosyl-(1→2)-β-D-glucopyranosyloxy]naringenin,
and compound 2 as (2R)-5-[α-L-rhamnopyranosyl-(1→2)-β-D-
glucopyranosyloxy]naringenin. Previously, only 5-[α-L-rham-
nopyranosyl-(1→2)-α-D-glucopyranosyloxy]naringenin iso-
lated from Cyclopia intermedia has been published,10 albeit
under the incorrect name of 5-O-α-D-rutinosylnaringenin, and
identified from the peracetate derivative. Therefore, NMR
spectroscopic data of 5-[α-L-rhamnopyranosyl-(1→2)-β-D-
glucopyranosyloxy]naringenin and the absolute configuration
of both diastereoisomers are reported here for the first time.
Compound 3 also had a molecular formula of C27H32O14, as
deduced from its HR-ESIMS data (m/z 581.1879 [M + H]+,
RESULTS AND DISCUSSION
■
Structure Elucidation. Compound 1 was heated in
phosphate buffer (pH 5.0) at 90 °C for 4 h. Conversion
products were separated by semipreparative RP-HPLC to yield
compounds 2 and 3 as major and minor products, respectively.
1H and 2D NMR spectra, GC-MS sugar analysis, and HR-
ESIMS spectra of compounds 1−3 as well as UV spectra of
compound 1 are supplied in the Supporting Information
(Figures S1 to S19).
1
calcd 581.1865). H and HSQC NMR data were similar to
those of compound 1. The difference was in the presence of
signals for a 1,2-disubstituted vinylic moiety [δC 125.1, δH 7.92
(d, J = 15.6) and δC 142.1, δH 7.55 (d, J = 15.6)] instead of the
methylene and methine signals of the flavanone. An E-
configuration was indicated by the vicinal coupling of 15.6 Hz
and a correlation in the HMBC spectrum with the resonance at
δC 192.2 (Figure 1) signifying an (E)-α,β-unsaturated carbonyl
group. Since the molecular formula and, thus, the index of
hydrogen deficiency were the same as for compound 1, these
data indicated that the C-ring of 1 was cleaved during the
heating to produce naringenin chalcone glycoside 3. GC-MS
analysis confirmed the presence of D-glucose and L-rhamnose
as sugar constituents (Figures S14 and S15, Supporting
Information). Naringin chalcone, a regioisomer of 3, has
been previously isolated from grapefruit,11 but (E)-2′-[α-L-
rhamnopyranosyl-(1→2)-β-D-glucopyranosyloxy]-4′,6′,4-trihy-
droxychalcone (3) has not been reported before.
Thermal Conversion of Compound 1. The temperature
range 80 to 120 °C used to test conversion was governed by
likely conditions encountered when a Cyclopia extract is used
as a functional ingredient in a variety of food and beverage
products, such as food bars, bread, and iced tea. This will
expose the extract to high temperatures during food
processing, i.e., heat sterilization (121 °C) of “natural”
honeybush ready-to-drink iced tea (pH ∼5) to ensure stability
against microbial spoilage.
Heating of dilute solutions of compound 1 in a 0.1 M
phosphate buffer (pH 5) at the various temperatures decreased
its concentration, with increasing temperature accelerating the
process (Figure 3). Fitting of the data to different kinetic
models revealed that the decrease in the concentration of
compound 1 could be best predicted by the fractional
conversion model (R2 ≥ 0.97). The first-order kinetic model,
commonly used to predict changes in quality parameters of
food products during thermal processing,12 gave a good fit only
when the solution was heated at 80 °C. The reaction rate
constants (k) varied from 0.36 to 6.64 h−1 (Table 2), and their
temperature-dependence complied with the Arrhenius equa-
tion. The activation energy of the reaction was calculated as
99.16 kJ mol−1 (R2 = 0.9704), which is slightly lower than that
of mangiferin (107 kJ mol−1),13 the major phenolic compound
of C. genistoides and determined under the same conditions.
Compound 1 was assigned a molecular formula of
C27H32O14, according to HR-ESIMS analysis (m/z 581.1865
1
[M + H]+, calcd 581.1865). H and 2D NMR data (Table 1)
indicated naringenin bearing a disaccharide moiety attached to
C-5 via an O-glycosidic linkage. The sugars were identified by
GC-MS analysis after derivatization with L-cysteine methyl
ester and subsequent silylation as L-rhamnose and D-glucose7
(Figures S13 and S15, Supporting Information). The coupling
constants of the anomeric protons at δH 5.16 (1H, d, J = 6.9
Hz, H-1″) and 5.19 (1H, br s, H1‴) led to assignment of β-
and α-configurations for the D-glucopyranose and L-rhamno-
pyranose moieties, respectively. HMBC correlations (Figure 1)
between δH 5.19 (1H, br s, H1‴) and δC 76.6 (C-2″) and
between δH 5.16 (1H, d, J = 6.9 Hz, H-1″) and δC 158.8 (C-5)
established the attachment positions of the two sugars, and 1
was thus identified as 5-[α-L-rhamnopyranosyl-(1→2)-β-D-
glucopyranosyloxy]naringenin.
Compound 2 was assigned a molecular formula of
C27H32O14 (HR-ESIMS m/z 581.1865 [M + H]+, calcd
581.1865). 1H and 2D NMR data (Table 1) closely resembled
those of compound 1. Given that the same COSY, HMBC, and
NOESY correlations were observed, compound 2 had to be a
diastereoisomer of 1.
The absolute configuration of compounds 1 and 2 was
determined by electronic circular dichroism (ECD), since
those compounds possess a carbonyl group and two aromatic
rings close to the C-2 stereocenter. Both compounds exhibited
the same UV spectrum with maxima at 227 and 281−282 nm
(Figure S19, Supporting Information). For compound 1, the
experimental ECD spectrum (Figure 2) exhibited positive
Cotton effects (CEs) at 216 and 331 nm along with a negative
B
DOI: 10.1021/acs.jnatprod.8b00710
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
1
Table 1. H (500 MHz) and 13C NMR (125 MHz) Spectroscopic Data for Compounds 1−3 (DMSO-d6; δ in ppm, J in Hz)
1
2
3
b
b
b
position
δC, type
δH (J in Hz)
δC, type
δH (J in Hz)
δC, type
δH (J in Hz)
1
2
3
126.6, C
130.9, CH
116.5, CH
78.0, CH
44.7, CH
5.32, dd (12.7, 2.6)
2.97, dd (15.2, 12.8)
2.51, dd (15.7, 2.6)
78.5, CH
45.2, CH
5.30, dd (12.4, 2.9)
2.91, dd (16.8, 12.5)
2.58, dd (16.6, 2.9)
7.59, d (8.5)
6.81, d (8.5)
4
5
6
7
8
9
10
1′
2′
3′
4′
5′
6′
1″
2″
3″
4″
5″
6″
187.5, C
158.8, C
95.2, CH
163.7, C
96.3, CH
163.4, C
104.6, C
129.2, C
127.7, CH
115.0, CH
157.2, C
115.0, CH
127.7, CH
97.5, CH
76.6, CH
76.8, CH
69.5, CH
76.6, CH
60.4, CH2
187.9, C
159.9, C
96.0, CH
165.0, C
97.0, CH
164.5, C
105.6, C
129.6, C
128.4, CH
115.7, CH
158.1, C
115.7, CH
128.4, CH
98.3, CH
76.7, CH
77.8, CH
70.2, CH
77.3, CH
61.1, CH2
160.5, C
116.5, CH
130.9, CH
142.1, CH
125.1, CH
192.2, C
6.81, d (8.5)
7.59, d (8.5)
7.55, d (15.6)
7.92, d (15.6)
6.19, s
5.98, s
6.22, s
6.01, s
106.7, C
160.3, C
95.5, CH
166.0, C
97.9, CH
166.0, C
98.5, CH
78.9, CH
77.4, CH
70.4, CH
77.7, CH
60.9, CH2
7.28, d (8.2)
6.79, d (7.9)
7.27, d (8.5)
6.79, d (8.2)
6.21, d (1.5)
5.95, d (1.8)
5.34, d (7.3)
6.79, d (7.9)
7.28, d (8.2)
5.16, d (7.0)
3.60, dd (7.5, 7.5)
3.43−3.56
3.32, dd (9.0, 9.0)
6.79, d (8.2)
7.27, d (8.5)
5.14, d (7.3)
3.55−3.65
3.45−3.54
a
a
3.45−3.55
a
a
a
3.45−3.55
3.32, dd (9.2, 9.2)
3.25, dd (9.0, 9.0)
a
a
a
3.35−3.41
3.65−3.73
3.43−3.56
3.34−3.41
3.65−3.70
3.45−3.54
3.33−3.38
a
a
3.68, m
3.45−3.55
a
a
a
1‴
2‴
3‴
4‴
5‴
6‴
99.4, CH
70.2, CH
70.1, CH
71.9, CH
68.2, CH
5.19, br s
99.9, CH
71.0, CH
70.9, CH
72.6, CH
68.8, CH
18.4, CH3
5.23, br s
101.3, CH
70.9, CH
71.0, CH
72.4, CH
68.9, CH
17.8, CH3
5.03, br s
a
a
3.65−3.73
3.35−3.41
3.65−3.70
3.34−3.41
3.77, m
3.33−3.38
a
a
a
3.12, dd (9.3, 9.3)
3.17, dd (9.5, 9.5)
3.09, dd (9.3, 9.3)
3.41, m
0.81, d (6.1)
a
a
3.43−3.56
3.55−3.65
17.5, CH3
0.99, d (6.1)
1.05, d (6.1)
a
b
Overlapping signals. From inverse detected HSQC and HMBC experiments.
Figure 1. Key HMBC correlations (green arrows) of compounds 1
and 3.
Formation of compound 2 at the various temperatures showed
a progressive shift in formation/degradation (Figure 3). After 6
h at 80 °C the concentration of compound 2 was still
increasing, at 90 °C the maximum was reached and
degradation had started, and at 100, 110, and 120 °C, the
optimum was reached earlier with each higher temperature.
Degradation also became more pronounced as temperature
increased. Naringin remained stable over 6 h when heated at
100 °C, indicating that glycosylation at C-7 instead of C-5
conferred increased thermal stability. Heating of an aqueous
naringin solution at ca. pH 6.6 at 100 °C also showed little
degradation and k values of 0.0004 and 0.0016 min−1 at 110
and 120 °C, respectively.14
Figure 2. Experimental ECD spectra of compounds 1 and 2 in MeOH
(0.1 mg/mL).
The appearance of compound 3 and other minor
compounds in the reaction mixtures (Figure S20, Supporting
Information) indicates conversion of the flavanone to a
chalcone and degradation of 2 and/or 3. Flavanones are
mostly stable in acidic and neutral solutions of protic solvents,
conditions favoring cyclization.15,16 At pH 5 the reaction
C
DOI: 10.1021/acs.jnatprod.8b00710
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 4. Bitter intensity of compound 1 compared to naringin at the
same concentration (24 mg/L; 0.04 mM). Samples were prepared in
hot water (blank) and served hot (60 °C) to a trained sensory panel
(n = 8). Different letters indicate a significant difference (p < 0.05) in
mean values. Error bars indicate standard deviation.
the tongue for several minutes after tasting. Many flavanones
and flavanone glycosides are known to affect bitter taste, either
acting as agonists of bitter receptors or by masking the bitter
taste of other compounds.17,18 Molecular structure is
important, especially the substitution pattern as indicated by
Roland et al.19 for (iso)flavonoid aglycones. In this case
compound 1 and naringin differ only with regard to the
location of the sugar moiety on the aglycone. Interestingly, a
(1→6) linkage of the rhamnose to the glucose moiety instead
of a (1→2) linkage (rutinose instead of neohesperidose) also
eliminates the bitter taste.20 The C-2 configuration of the
flavanone may also play a role in the interaction with bitter
receptors and, hence, in the sensory quality of the compounds.
For example, L-enantiomers of tryptophan and phenylalanine
elicit a bitter taste and even a cellular TAS2R response,21 while
the D-enantiomers have a distinct sweet taste.22 However, the
effect of the conversion of compound 1 to 2 on taste is not
known at present, due to limited availability of 2 for sensory
analysis. Results obtained by Gaffield et al.20 for the 2S and 2R
stereoisomers of naringin were inconclusive, but suggest the
possibility that conversion of 1 to 2 could have little effect on
bitterness.
Figure 3. Change in the concentration of compounds 1 (A) and 2
(B) when 1 was heated in an aqueous phosphate buffer at pH 5 at
temperatures ranging from 80 to 120 °C. Data points for 1 were fitted
to the fractional conversion model (smooth lines). Naringin (A) was
heated under the same conditions at 100 °C. The initial concentration
of the compounds was 0.1 mM. Data indicated by markers are means
standard deviation.
equilibrium would thus be toward cyclization and not ring
opening, explaining the limited formation of compound 3.
Sensory Analysis. Descriptive sensory analysis confirmed
the discrimination in bitter taste intensity between compound
1 and naringin (Figure 4). Compound 1, tested at a
concentration similar to what could be expected in a C.
genistoides infusion, was only slightly bitter, whereas naringin,
at the same concentration, was considerably more bitter. Their
bitterness was rated 7 and 26, respectively, on a 100-point
scale. A bitter intensity of 7 is just perceptible. During panel
training panelists noted a difference in mouthfeel between
compound 1 and naringin. Compound 1 appeared “smooth”,
while naringin had a delayed bitter taste, only reaching
maximum bitterness after ∼2 s in the mouth and lingering on
EXPERIMENTAL SECTION
■
General Experimental Procedures. HPLC-grade solvents were
obtained from Macron Fine Chemicals (Avantor Performance
Materials, Phillipsburg, NJ, USA) and Merck (Darmstadt, Germany).
Table 2. Kinetic Parameters for the Degradation of Compound 1 Determined by Fitting Data to the Fractional Conversion
a
Model
temperature
equilibrium concentration (C∞, mM)
initial concentration (C0, mM)
reaction rate constant (k, h−1
)
R2
80 °C
90 °C
100 °C
110 °C
120 °C
0.0430 0.0046 b
0.0471 0.0006 a
0.0407 0.0006 bc
0.0406 0.0009 bc
0.0374 0.0002 c
0.1047 0.0002 a
0.1027 0.0008 b
0.1018 0.0001 b
0.1005 0.0002 c
0.0951 0.0008 d
0.36 0.06 e
1.11 0.07 d
2.11 0.14 c
4.26 0.39 b
6.64 0.27 a
0.9914
0.9845
0.9920
0.9716
0.9781
a
Values are mean standard error. Different letters indicate a significant difference at p > 0.05.
D
DOI: 10.1021/acs.jnatprod.8b00710
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
̈
DMSO-d6 was purchased from Armar Chemicals (Dottingen,
Switzerland). Analytical-grade formic acid, XAD porous resin
(Amberlite, 20−60 mesh), analytical-grade EtOH, MeOH, DMSO,
anhydrous pyridine, caffeine for sensory analysis, L-ascorbic acid, L-
cysteine methyl ester hydrochloride, and hexamethyldisilazane/
trimethylchlorosilane/pyridine (3:1:9) were purchased from Merck.
L-Glucose and D-glucose were obtained from Acros Organics
(Thermo Fisher Scientific, Waltham, MA, USA), while L-rhamnose
and D-rhamnose were sourced from Carbosynth (Compton, UK).
HCl and analytical-grade EtOAc were purchased from Scharlau
(Scharlab S. L.) (Barcelona, Spain). Deionized water and ultrapure
water were prepared using an Elix and Milli-Q water purification
system (Merck).
18−25 min (23% B); 35 °C; flow rate 21.2 mL/min) to yield
compound 1 (80 mg, tR = 13.3 min; 97% purity based on LC-MS).
Isolation of Compounds after Heat Treatment. Compound 1
(15.1 mg) was dissolved in water (10 mg/mL) and heated for 4 h at
90 °C. The resulting mixture was purified by semipreparative RP-
HPLC [0.1% aq. formic acid (A), 0.1% formic acid in MeCN (B); 0−
5 min (5−15% B), 5−20 min (15% B), 20−40 min (15−60% B); flow
rate 4 mL/min; injection volume 4 × 250 μL, room temperature] to
yield compound 2 (3.1 mg, tR 14.4 min), compound 1 (5.3 mg, tR
15.9 min) and compound 3 (0.6 mg, tR = 28.1 min).
(2S)-5-[α-L-Rhamnopyranosyl-(1→2)-β-D-glucopyranosyloxy]-
naringenin (1): yellow, amorphous powder; [α]25 −68.0 (c 0.17,
MeOH); UV λmax (MeOH) (log ε) 227 (3.08), 281D(2.81) nm; ECD
(MeOH, c 1.8 × 10−4 M, 1 mm path length) λmax(Δε) 216 (+8.88),
Optical rotations were measured in MeOH on a Jasco P-2000
digital polarimeter (Tokyo, Japan) equipped with a sodium lamp (589
nm) and a 10 cm length temperature-controlled microcell. UV and
ECD spectra were recorded in MeOH on a Chirascan spectrometer
(Applied Photophysics, Leatherhead, UK) with 1 mm path precision
̈
cells 110 QS (HellmaAnalytics, Mullheim, Germany). NMR spectra
were recorded in DMSO-d6 on a Bruker AVANCE III 500 MHz
1
286 (−7.67), 331 (+3.42) nm; H and 13C NMR, see Table 1; HR-
ESIMS m/z 581.1865 [M + H]+ (calcd for C27H32O14+, 581.1865);
MSE fragment ions (negative ionization) m/z 459, 433, 313, 271*,
209, 169, 151, 145, 125.
(2R)-5-[α-L-Rhamnopyranosyl-(1→2)-β-D-glucopyranosyloxy]-
naringenin (2): yellow, amorphous powder; [α]25 −72.3 (c 0.25,
D
MeOH); UV λmax (MeOH) (log ε) 227 (2.91), 282 (2.63) nm; ECD
1
spectrometer (Billerica, CA, USA) operating at 500.13 MHz for H
(MeOH, c 1.7 × 10−4 M, 1 mm path length) λmax (Δε) 216 (−10.48),
and 125.77 MHz for 13C. Measurements were performed with a 1 mm
TXI probe at 18 °C. Data were processed with Bruker TopSpin 3.5
software. HR-ESIMS data were recorded in positive ion mode on a
Shimadzu LC-20A system (Kyoto, Japan) with a Thermo Scientific
Orbitrap LTQ XL detector (Waltham, CA, USA) and a Waters
SunFire C18 column (5 μm, 150 × 10 mm i.d.). Data acquisition was
recorded using Xcalibur 2.1 software. Fragmentation data (MSE using
a collision energy ramp from 20 to 60 V) were obtained by analysis in
the negative ion mode on an Acquity UPLC system coupled to a
Synapt G2 Q-TOF equipped with an electrospray ionization source
(Waters, Milford, MA, USA). The LC inlet method used a Kinetex
C18 column (2.6 μm, 150 × 4.6 mm, Phenomenex, Torrance, CA,
USA) as described by Beelders et al.6 Preparative HPLC was
performed on a Waters preparative LC, with autosampler, variable
UV−visible detector, and fraction collector, equipped with a Gemini-
NX C18 preparative column (5 μm; 150 × 21.2 mm, 100 Å;
Phenomenex), protected by a guard column with the same stationary
phase. Semipreparative HPLC was performed on an Agilent 1100
Series instrument with a DAD detector (Santa Clara, CA, USA) and
equipped with a Waters SunFire C18 column (5 μm, 150 × 10 mm
i.d.), protected by a guard column (10 × 10 mm i.d.; Waters). Data
acquisition and processing were performed using ChemStation
software. GC-MS analysis was performed on an Agilent G1530A
equipped with an RXI 5 MS column (25 m × 0.2 mm i.d., 0.3 μm).
Plant Material. Shoots of C. genistoides Vent. (genotype GK5 of
the ARC Honeybush Plant Breeding Programme) were harvested at
an evaluation site (GPS coord. −34.702, 19.618), shredded, dried (40
°C/16 h), and sieved to obtain the tea bag fraction (<1.68 mm, >0.42
mm; retention sample CGN_L0152).
1
235 (−6.57), 294 (+5.11), 329 (−4.55) nm; H and 13C NMR, see
+
Table 1; HR-ESIMS m/z 581.1865 [M + H]+ (calcd for C27H32O14
,
581.1865); MSE fragment ions (negative ionization) m/z 433, 271*,
227, 151, 145, 125, 119, 107.
E-2′-[α-L-Rhamnopyranosyl-(1→2)-β-D-glucopyranosyloxy]-
1
4′,6′,4-trihydroxychalcone (3): H and 13C NMR, see Table 1; HR-
ESIMS m/z 581.1879 [M + H]+ (calcd for C27H32O14+, 581.1865);
MSE fragment ions (negative ionization) m/z 271*, 169, 151, 145,
125, 107.
Sugar Analysis. Compound 1, 2, or 3 (0.5 mg) was heated at 105
°C for 1 h in 1 mL of 2 M HCl. After extraction with EtOAc, the
aqueous phase was lyophilized and resolubilized in 1 mL of anhydrous
pyridine. Derivatization with L-cysteine methyl ester hydrochloride
(200 μL, 60 °C, 1 h) and subsequently silylation with
hexamethyldisilazane and Me3SiH in pyridine (3:1:9; 200 μL; 60
°C, 30 min) were performed. Pyridine was evaporated prior to GC-
MS analysis. The column temperature was kept at 60 °C for 1 min
and then increased at 10 °C/min until 300 °C; L-rhamnose (tR 21.29
min), D-rhamnose (tR 21.39 min), D-glucose (tR 22.59 min), L-glucose
(tR 22.73 min).
Thermal Conversion of Compound 1. The thermal degradation
of 1 in 0.1 M phosphate buffer solution (pH 5) was assessed at five
temperatures (80, 90, 100, 110, and 120 °C). Similarly, naringin was
heated at 100 °C. The experiments were performed as previously
described.13 Aliquots (800 μL; n = 25) of the working solution of
each compound (58 μg/mL = 0.1 mM, dissolved 0.1 M phosphate
buffer solution) were transferred into 5 mL glass reaction vials
(Merck). One aliquot served as unheated control while the remaining
vials were heated in a preheated Stuart heating block with glycerin
added to the cavities to improve heat transfer. Replicate samples (n =
3) were randomly removed at predetermined time points (n = 8),
cooled, filtered (0.22 μm pore size, 4 mm diameter Millex-GV syringe
filters; Merck), and analyzed using UHPLC-DAD.13 The content of 1,
2, and naringin was quantified using the peak area at 288 nm.
Univariate analysis of variance (ANOVA) was performed on the
data sets for each temperature (SAS, version 9.4; SAS Institute, Cary,
NC, USA). The Shapiro−Wilk test was performed to test for
normality. Fisher’s least significant difference was calculated at the 5%
level (p < 0.05) to compare means across treatment times. The kinetic
data for compound 1 were fitted to zero-, first-, and second-order and
fractional conversion models by nonlinear regression, using SAS. The
fractional conversion model (eq 1) was selected based on goodness-
of-fit of predicted and actual data (R2).
Isolation of Compound 1. A freeze-dried, hot water extract (120
g), prepared from the plant material (600 g) as described by Beelders
et al.,23 was sonicated in EtOH (1:10 m/v) for 60 min. The EtOH-
soluble fraction was recovered and freeze-dried, and 15 g of the
fraction, suspended in 50 mL of HPLC-grade water, was further
fractionated on a dechlorinated XAD (Amberlite, 20−60 mesh)
column (68 × 500 mm). An EtOH/water gradient at a flow rate of 38
mL/min was employed: 0% EtOH (4.5 L), 5% EtOH (4.5 L), 10%
EtOH (6 L), 20% EtOH (4.5 L), 30% EtOH (4.5 L), 100% EtOH
(4.5 L). Column fractions (750 mL) were monitored by HPLC-DAD,
and those containing flavanones (fractions 29 to 38) were pooled,
vacuum-evaporated, and freeze-dried, yielding a 4 g fraction. An
aliquot of the XAD fraction (2 g) was dissolved at 5 mg/mL (as
limited by solubility) in a 23% MeOH/water solution containing 10%
DMSO (v/v), and 5000 μL (maximum capacity of injection loop)
was repeatedly injected on preparative HPLC, employing a gradient
solvent program (solvent A: 0.1% formic acid(aq); solvent B: 100%
MeOH): 0−2 min (23% B), 2−15 min (23−24% B), 15−16 min
(24−90% B), 16−17 min (90% B), 17−18 min (90−23% B), and
C = C∞ + (C0 − C∞)exp−kt
(1)
where C, C0, and C∞ are the concentration of compound 1 (mM) at
time t, time 0, and equilibrium conditions, respectively, t is the time in
h, and k is the reaction rate constant (h−1).
E
DOI: 10.1021/acs.jnatprod.8b00710
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
1H and 2D NMR spectra of compounds 1−3; GC-MS
sugar analysis of compounds 1−3; HR-ESIMS spectra of
compounds 1−3; UV spectra of compound 1; HPLC
chromatograms of compound 1 heated at 80, 90, 100,
110, and 120 °C for 2 h (PDF)
The values of k, C0, and C∞ were compared over the different
treatment temperatures (ANOVA) to establish significant differences
at the 5% significance level. By fitting a linear regression for ln(k) on
1/T (T = absolute temperature in K) it was confirmed (R2 = 0.9768, p
< 0.05) that k was related to T according to the Arrhenius law (eq 2).
Ea
RT
i
y
j
z
z
z
j
k = A exp −
j
(2)
k
{
AUTHOR INFORMATION
■
where A is the pre-exponential factor, Ea is the activation energy, and
R is the universal gas constant (8.314 J mol−1 K−1).
To estimate Ea, a global model that includes both time and
temperature effects, combining eq 1 with the reparametrized form of
eq 2, was used:24
Corresponding Authors
*Tel: +41 61 207 14 25. Fax: +41 61 207 14 74. E-mail:
matthias.hamburger@unibas.ch.
*Tel: +27 21 809 3444. Fax: +27 21 809 3430. E-mail:
joubertL@arc.agric.za.
[−E /R(1/T−1/T )]t
ref
a
e
]
C = C∞ + (C0 − C∞)e[−k
ref
ORCID
(3)
Matthias Hamburger: 0000-0001-9331-273X
Elizabeth Joubert: 0000-0002-9717-9769
where kref is the reaction rate at the reference temperature (Tref). Tref
was the average value of the temperature range evaluated (i.e., Tref
100 °C or 373.15 K).
=
Notes
The model parameters were estimated by nonlinear regression
analysis using the NLIN procedure of SAS, solving the nonlinear least-
squares problem, using the Marquart algorithm. The parameter’s
precision was evaluated by standard errors, and the quality of
regression was assessed by the coefficient of determination (R2) and
randomness and normality of residuals, thus allowing best fit model
parameters.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Thanks are due to O. Fertig and M. Oufir (Division of
Pharmaceutical Biology, University of Basel) for their technical
assistance. The work was funded by the National Research
Foundation of South Africa under the Swiss/SA Joint Research
Programme (grant 107805 to E.J.) and the Research and
Technology Fund (grant 98695 to E.J.). L.A. received an
NRF/DST-PDP Ph.D. scholarship (grant 96711), and C.H. an
NRF/DST-PDP Postdoctoral fellowship (grant 104908). The
grant holders acknowledge that opinions, findings, and
conclusions or recommendations expressed in any publication
generated by the NRF-supported research are those of the
authors and that the NRF accepts no liability whatsoever in
this regard.
Sensory Analysis. The bitter taste of compound 1 was compared
to that of naringin at a concentration similar to what could be
expected in a C. genistoides infusion, by taking into account its
concentration in the hot water extract (1.21 g/100 g) and the typical
soluble solids content of a hot water infusion (2 g/L).25 Samples
(0.04 mM) were dissolved in hot water by stirring the solution for ca.
20 min in a jacketed flask connected to a circulating bath controlled at
70 °C. The three samples (hot water blank, compound 1, and
naringin) were presented to each panelist in three-digit blind coded
40 mL screw-cap amber vials (ca. 10 mL/serving). The vials were
placed in metal racks and kept in a heated water bath (60 °C) during
tasting to prevent precipitation. Samples were randomized in each
session for each panelist. Panelists were also supplied with two
marked reference samples, namely, a blank (hot water, 0 bitter
intensity) and a bitter sample (caffeine, 0.4 g/L, bitter intensity of
45). Descriptive sensory analysis (DSA) was conducted under
standardized conditions by a panel of eight judges experienced in
the sensory analysis of green C. genistoides honeybush tea, specifically
bitter taste. Panelists were assigned individual booths, and each
sample was scored for bitter taste intensity on an anchored,
unstructured 100-point line scale (0 = not bitter, 100 = extremely
bitter), using Compusense Five software (Compusense, Guelph,
Canada). Each sample set was presented to each panelist four times
during four 20 min sessions with 10 min breaks scheduled between
sessions to prevent panel fatigue. Distilled water, water biscuits, and
dried apple pieces were provided between samples as palate cleansers,
and a 2 min delay was scheduled between samples to minimize bitter
taste carryover. Panel performance was monitored by applying
Panelcheck software (version 1.4.0; Nofima, Ås, Norway) to the
DSA data. Panel reliability was tested using an ANOVA model that
included panelist, replicate, sample, and interaction effects.26 The
residuals were tested for non-normality, using the Shapiro−Wilk test,
and outliers were identified and removed in the event of deviation
from normality (p ≤ 0.05). Subsequent ANOVA was conducted on
means over judges, and Fisher’s LSDs were calculated to compare
treatment means (p ≤ 0.05).
REFERENCES
■
(1) Lee, J.; Chambers, E.; Chambers, D. H.; Chun, S. S.;
Oupadissakoon, C.; Johnson, D. E. J. Sens. Stud. 2010, 25, 109−132.
(2) Drewnowski, A.; Gomez-Carneros, C. Am. J. Clin. Nutr. 2000, 72
(6), 1424−1435.
(3) Meyerhof, W.; Behrens, M.; Brockhoff, A.; Bufe, B.; Kuhn, C.
Chem. Senses 2005, 30, 14−15.
(4) Janssen, S.; Depoortere, I. Trends Endocrinol. Metab. 2013, 24,
92−100.
(5) Erasmus, L. M.; Theron, K. A.; Muller, M.; Van der Rijst, M.;
Joubert, E. S. Afr. J. Bot. 2017, 110, 144−151.
(6) Beelders, T.; De Beer, D.; Stander, M.; Joubert, E. Molecules
2014, 19, 11760−11790.
(7) Severi, J. A.; Fertig, O.; Plitzko, I.; Vilegas, W.; Hamburger, M.;
Potterat, O. Helv. Chim. Acta 2010, 93, 1058−1066.
(8) DeAngelis, G. G.; Wildman, W. C. Tetrahedron 1969, 25, 5099−
5112.
(9) Li, X.-C.; Ferreira, D.; Ding, Y. Curr. Org. Chem. 2010, 14,
1678−1697.
(10) Kamara, B. I.; Brandt, E. V.; Ferreira, D.; Joubert, E. J. Agric.
Food Chem. 2003, 51, 3874−3879.
(11) Miles, C. O.; Main, L. J. Chem. Soc., Perkin Trans. 2 1988, 195−
198.
(12) Van Boekel, M. A. J. S. Compr. Rev. Food Sci. Food Saf. 2008, 7,
144−158.
ASSOCIATED CONTENT
■
(13) Beelders, T.; De Beer, D.; Ferreira, D.; Kidd, M.; Joubert, E.
Food Chem. 2017, 233, 412−421.
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jnat-
prod.8b00710.
́
(14) Chaaban, H.; Ioannou, I.; Chebil, L.; Slimane, M.; Gerardin, C.;
Paris, C.; Charbonnel, C.; Chekir, L.; Ghoul, M. J. Food Process.
Preserv. 2017, 41, e13203.
F
DOI: 10.1021/acs.jnatprod.8b00710
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
(15) Cisak, A.; Mielczarek, C.; Chemistry, O. J. Chem. Soc., Perkin
Trans. 2 1992, 2, 1603−1607.
́
(16) Gonzalez, E. A.; Nazareno, M. a.; Borsarelli, C. D. J. Chem. Soc.,
Perkin Trans. 2 2002, 2052−2056.
(17) Ley, J. P.; Krammer, G.; Reinders, G.; Gatfield, I. L.; Bertram,
H.-J. J. Agric. Food Chem. 2005, 53, 6061−6066.
(18) Ley, J. P. Chemosens. Percept. 2008, 1, 58−77.
(19) Roland, W. S. U.; Van Buren, L.; Gruppen, H.; Driesse, M.;
Gouka, R. J.; Smit, G.; Vincken, J. P. J. Agric. Food Chem. 2013, 61,
10454−10466.
(20) Gaffield, W.; Lundin, R. E.; Gentili, B.; Horowitz, R. M. Bioorg.
Chem. 1975, 4, 259−269.
(21) Kohl, S.; Behrens, M.; Dunkel, A.; Hofmann, T.; Meyerhof, W.
J. Agric. Food Chem. 2013, 61, 53−60.
(22) Solms, J.; Vuataz, L.; Egli, R. Experientia 1965, 21, 692−694.
(23) Beelders, T.; Brand, D. J.; De Beer, D.; Malherbe, C. J.;
Mazibuko, S. E.; Muller, C. J. F.; Joubert, E. J. Nat. Prod. 2014, 77,
2694−2699.
(24) Pinheiro, J.; Alegria, C.; Abreu, M.; Gonca̧ lves, E. M.; Silva, C.
L. M. J. Food Eng. 2013, 114, 338−345.
(25) Alexander, L.; De Beer, D.; Muller, M.; Van der Rijst, M.;
Joubert, E. Food Chem. 2019, 276, 626−635.
(26) Næs, T.; Brockhoff, P.; Tomic, O. Statistics for Sensory and
Consumer Science; Wiley: New York, USA, 2010.
G
DOI: 10.1021/acs.jnatprod.8b00710
J. Nat. Prod. XXXX, XXX, XXX−XXX