2
F. Zhu et al.
The reaction was carried out at room temperature and
monitored by ultrafast liquid chromatography. The reac-
tion solution was diluted with water (50 mL) and
extracted with diethyl ether (50 mL) in a separating
funnel. After the diethyl ether layer was collected and
washed with water and 5% NaHCO3 aqueous solution,
the organic layer was evaporated to dryness. The pro-
pyl and hexyl esters were synthesized as described
above, using 1-propanol and 1-hexanol in place of
methanol.
RA, CA, Cou, FA, Cin, and their methyl esters sup-
pressed β-hexosaminidase release in a dose-dependent
manner up to the concentration of 2.0 mM, as shown
in Fig. 1. The IC50 values for all methyl esters were
below 0.4 mM (RA-me, 0.06 mM; CA-me, 0.12 mM;
Cou-me, 0.18 mM; FA-me, 0.14 mM; and Cin-me,
0.33 mM), and thus, smaller than the IC50 values of all
free acids (RA, 1.31 mM; CA, 1.46 mM; Cou,
1.04 mM; FA, 0.98 mM; and Cin, 1.50 mM). RA-me
exerted the greatest β-hexosaminidase release suppres-
sion (IC50, 0.06 mM), followed by CA-me (IC50,
0.12 mM) among all of the methyl esters. CA-me sup-
pressed β-hexosaminidase release similar to RA-me,
and then similar suppression activity was observed
between RA (1.31 mM) and CA (1.46 mM). It is indi-
cated that β-hexosaminidase release suppression of RA
and its methyl ester is mainly due to its CA moiety,
which may interact with a cross-linked IgE/FcεRI
receptor of RBL-2H3 cells. This is partly supported by
the results that 3-(3,4-dihydroxyphenyl)lactic acid (a
similar moiety of RA partial structure) related com-
pounds did not suppress β-hexosaminidase release at
all.
As a greater suppression of β-hexosaminidase release
was achieved by methyl esterification of such phenyl-
propanoids was observed, we investigated the effects of
longer alkyl chain esters of RA and its related phenyl-
propanoids on β-hexosaminidase release suppression.
Furthermore, RA-me, RA-pro, and RA-hex suppressed
β-hexosaminidase release to a greater extent than the
corresponding esters of CA and Cou (Table 1). Among
all free acids and esters, rosmarinic acid propyl ester
(RA-pro, IC50, 23.7 μM) exhibited the greatest
suppression on β-hexosaminidase release, followed by
rosmarinic acid hexyl ester (RA-hex, IC50, 34.5 μM),
caffeic acid propyl ester (CA-pro, IC50, 55.3 μM), and
rosmarinic acid methyl ester (RA-me, IC50, 57.3 μM),
as shown in Table 1. The activity of rosmarinic acid
propyl ester was only 9.1 times lower than luteolin
(IC50, 2.6 μM), a flavonoid with high antiallergic
activity that was tested in a parallel experiment of our
laboratory,19) but increased 55.4 times compared with
the activity of RA.
In addition, it was recently found that RA can be
easily extracted from natural resources in a large scale
with a newly established isolation method from Perilla
leaves using a supramolecular technique,17) implying
that RA, the precursor of RA esters, can be easily sup-
plied from natural resources such as Perilla leaves as
well as other Lamiaceae family plants that are rich in
RA. Furthermore, for applications in the food indus-
tries, we suggest that the ethyl esters might be a good
choice as their synthesis can be carried out with etha-
nol, regarded as a safe ingredient in food manufactur-
ing. Alcoholic beverages or other processed foods
containing Perilla leaves may bring great benefit for
alleviating allergy symptoms.
The octanol–water partition coefficient (LogP), a
measurement of molecular hydrophobicity, is a vital
parameter for membrane permeability, bioavailability,
and hydrophobic drug–receptor interactions. Regarding
the interaction between polyphenols and receptors, iso-
flavones are known to bind estrogen receptors and the
hydrophobic aglycone binds more than their
The antiallergic activity of the free acids and synthe-
sized esters was assessed by their ability to inhibit β-
hexosaminidase release from RBL-2H3 cells, using the
following method.17,18) In brief, the RBL-2H3 cells
were precultured in a 24-well plate (2.5 × 105 cells/
well) at 37 °C under a humidified 5% CO2 atmosphere
overnight. The cells were washed with PBS and incu-
bated with 500 μL antibody solution (mouse monoclo-
nal anti-dinitrophenyl antibody, 50 ng/mL) for 2 h’
sensitization. A test sample solution of 490 μL was
added to each well after washing with modified
Tyrode’s buffer (MT buffer). The MT buffer was used
as control instead of the sample. Albumin dinitrophenyl
(10 μL, final concentration = 50 ng/mL) was added to
each well to evoke allergic reactions (degranulation) of
the cells for 30 min. The supernatants (50 μL) were
transferred into wells (96-well microplate) and incu-
bated with 100 μL substrate (3.3 mM p-nitrophenyl-2-
acetamide-2-deoxy-β-D-glucopyranoside) at 37 °C for
25 min. The absorbance (OD) was measured at 405 nm
using a microplate reader. The gaining OD reflects β-
hexosaminidase release. The calculation was performed
using Equations (1) and (2) below. In “blank,” neither
antibody solution nor sample was added to cells to
account for spontaneous β-hexosaminidase release from
cells. In “control,” MT buffer instead of samples was
added to cells to account for β-hexosaminidase release
from cells in conditions without a sample. In “total,”
cells were lysed by 0.1% Triton X-100 in MT buffer to
measure the total amount of β-hexosaminidase con-
tained in the cells. In “sample,” both antibody solution
and samples were added to cells to measure β-hexosa-
minidase release from cells in the presence of our test
compounds.
ratio of b ꢀ hexosaminidase release ð%Þ
ðODcontrol or ODsample ꢀ ODblank
Þ
¼
ꢁ 100
(1)
ðODtotal ꢀ ODblank
Þ
The ratio of β-hexosaminidase release (%) of “con-
trol” against “total” (Triton X-100) ranged from 29 to
50%. Relative rate of β-hexosaminidase release of
“sample” against “control” was expressed as β-hexosa-
minidase release (%) as shown below,
b-hexosaminidase release ð%Þ
ODsample ꢀ ODblank
¼
ꢁ 100
(2)
ODcontrol ꢀ ODblank
The data are presented as mean standard derivation
(SD) of quadruplicate wells. Statistical differences were
assessed using one-way ANOVA, followed by the
New–Keuls test. p values below 0.05 were considered
statistically significant. Statistical analysis was carried
out by GraphPad PRISM 5.