Z. Hu et al.
Fitoterapia131(2018)112–118
as determined by the ratio of the EC50 for acetylcholine in the presence
of the antagonist against the EC50 for acetylcholine in the absence of the
antagonist; [B] = molar concentration of antagonist; n = the slope of
the regression. The EC50 value for the tested compound was calculated
by Graphpad Prism 5.0.
3. Results and discussion
Compound 1 was obtained as white amorphous powder. The
HRESIMS showed a positive [M + H]+ ion at m/z 430.3311 (calcd for
C
27H44NO3, 430.3316), corresponding to a molecular formula of
Fig. 3. ORTEP drawing of compound 3.
C27H43NO3. The IR spectrum exhibited absorption bands of hydroxy
(3420 cm−1) and carbonyl (1704 cm−1) groups. The 1H NMR spectrum
of 1 exhibited diagnostic signals for two methyl singlets at δH 1.09 (3H,
s, H3‐21) and 0.86 (3H, s, H3‐19) and one methyl doublet at δH 1.12
(3H, d, J = 7.2 Hz, H3‐27) (Table 1). The 13C NMR and DEPT spectra
methyls, eleven methylenes (two nitrogenated at δC 60.5, C-18; 62.6, C-
26), ten methines (one oxygenated at δC 70.8, C-3 and one nitrogenated
at δC 65.4, C-22), a ketocarbonyl carbon (δC 216.5, C-6), an oxygenated
tertiary carbon (δC 73.5, C-20), and a quaternary carbon (δC 36.2, C-10)
(Table 1). Based on the indices of hydrogen deficiency, six rings should
be present in the structure of 1. These data combined with the 1H−1H
COSY and HMBC spectra (Fig. 2) suggested a typical cevanine-type
alkaloid skeleton of 1 with two hydroxy groups at C-3 and C-20 and a
ketocarbonyl moiety at C-6 [20]. The 13C NMR data of 1 showed con-
sistence with those of imperialine (8) with the exception of signals for A
and B rings (Supplementary data, Table S1). This implied that they
possessed the same C－F rings but different A and B rings. The NOESY
correlations of H-1β/H-3, H-3/H-5, and H-5/H-1β exhibited their 1,3-
diaxial relationship, assigning a β-orientation for H-3 and H-5 (Fig. 2).
The NOESY cross-peaks of H-5/H3‐19 and H3‐19/H-8 indicated a β-
orientation for these protons. These NOESY data indicated a cis A/B
ring junction, different from the trans-fused mode in 8. Furthermore,
the half-height width of H-3 (W1/2 = 19.8 Hz) [21] and the coupling
constant of H-5 (J = 13.0, 4.8 Hz) suggested the axial orientation for H-
3 and H-5, which was consistent with the NOESY data. The relative
configurations of C−F ring moieties were confirmed by the NOESY
correlations shown in Fig. 2. The structure of 1 was thus elucidated, and
the compound was named frititorine A.
(Supplementary data, Table S1) indicated the obviously deshielded C-3
(ΔδC + 7.3) and shielded C-2 and C-4 (ΔδC −1.6 and − 3.4, respec-
tively) in 2, which further validated the glycosidation position at C-3.
Meanwhile, the sugar was identified as D-glucose by TLC analysis and
comparing its specific rotation ([α]D25 = +51.8) with those of au-
thentic
samples
(D-glucose:
[α]D25 = +52.0;
L-glucose:
[α]D25 = −51.9). Thus, the structure of 2 was elucidated as imperialine
β-N-oxide-3-O-β-D-glucoside, and the compound was given the name
frititorine B.
Compound 3 was obtained as colorless crystals from MeOH. It had a
molecular formula of
C27H41NO3, as determined by HRESIMS
(m/z 428.3159 [M + H]+; calcd for C27H42NO3, 428.3159). The 1H
NMR spectrum of 3 (Table 1) indicated the presence of two methyl
doublets at δH 0.97 (3H, d, J = 7.8 Hz, H3‐21) and 0.94 (3H, d,
J = 6.6 Hz, H3‐27) and two methyl singlets at δH 1.63 (3H, s, H3‐18)
and 0.66 (3H, s, H3‐19), characteristic of a jervine-type alkaloid. Ana-
lysis of the 1H−1H COSY, HMBC, and NOESY data of 3 (Fig. 2) and
comparison of its 13C NMR data with 5 (Supplementary data, Table S1)
revealed that they were stereoisomers. Different from 5, the chemical
shifts of C-1 and C-5 in 3 were shielded by ΔδC 4.6 and 5.0, respectively.
These changes were ascribed to the γ-gauche effect of HO-3, indicating
that HO-3 was axial and α-oriented. Additionally, H-3 showed a broad
singlet at δH 4.19 (W1/2 = 9.6 Hz) [22], which verified its equatorial β-
orientation. These data showed that 3 was a 3-epimer of 5. The quali-
fied crystals of 3 were obtained from MeOH, and an X-ray diffraction
experiment with Ga Kα radiation [Flack parameter: −0.08 (13)] was
applied to assign its absolute configuration (Fig. 3). Thus, the structure
of 3 was characterized, and the compound was named frititorine C.
The known compounds were identified as imperialinol (4) [18],
peimisine (5) [23], peimisine-3-O-β-D-glucoside (6) [24], ebeinine (7)
those reported. Compound 4 is a new natural cevanine-type alkaloid,
which had been synthesized previously [18].
Compound 2 was assigned the molecular formula C33H53NO9 by the
HRESIMS ion at m/z 608.3791 [M + H]+ (calcd for C33H54NO9,
608.3793). The 13C NMR and DEPT spectra exhibited 33 carbon signals
for three methyls, twelve methylenes, fifteen methines, a ketocarbonyl
carbon, an oxygenated tertiary carbon, and one quaternary carbon. Its
13C NMR data were similar to those of imperialine β-N-oxide except for
some additional signals of a hexose moiety [17]. A set of NMR re-
sonances (Table 1) were observed at δH 4.39–3.14 (one anomeric: δH
4.39, d, J = 7.8 Hz, H-1′) and δC 102.3 (C-1′), 75.1 (C-2′), 78.1 (C-3′),
71.7 (C-4′), 77.9 (C-5′), and 62.8 (C-6′), assignable to a β-D-glucopyr-
anosyl motif. The HMBC correlation of H-1′/C-3 (Fig. 2) indicated that
the sugar moiety was attached to C-3. The half-height width of H-3
(W1/2 = 20.4 Hz) [21] demonstrated its axial orientation, and the sugar
was thus equatorial and β-oriented. Furthermore, the NOESY cross-
peaks of H-13/H-17, H-13/H-18β, H-18α/H-22, H-22/H3‐21, H-22/H-
26α, and H-26β/H3‐27 suggested a β-orientation for H-13, H-17, and
H3‐27 and an α-orientation for H3‐21 and H-22 (Fig. 2). In order to
oxide was synthesized using imperialine (8) as starting material. Ad-
ditionally, the enzymatic hydrolysis of 2 produced the aglycone and the
sugar. The aglycone showed the same HPLC retention time as the
synthetic imperialine β-N-oxide, and its 13C NMR data (Supplementary
data, Table S1) were also consistent with those of the synthetic com-
pound and the literature [17]. The aglycone was therefore proved to be
imperialine β-N-oxide. Comparison of the 13C NMR data of 2, the syn-
thetic imperialine β-N-oxide, and the hydrolytic aglycone
The plausible biogenetic pathway for compound 1 is proposed
(Supplementary data, Scheme S1). It would be derived from solanidine,
a steroidal alkaloid that was isolated from the same plant previously
[13]. Solanidine would transform into isosteroidal alkaloid via proce-
vine [29]. Subsequently, through epoxidation, serial proton migration,
and epoxide ring opening [30], a 5β-cevanine alkaloid with a cis A/B
ring junction would be produced, which after reduction and oxidation
would generate 1.
Compounds 1–7 and 13 were evaluated for their relaxant effects
against Ach-induced tracheal contraction. Compound 1 showed re-
laxation activity, causing a rightward shift of the Ach concentration-
response curves (Fig. 4). The pA2 and EC50 values of
1
(pA2 = 5.97
4.90 μM) were equivalent to
0.82;
those of the positive control aminophylline (pA2 = 5.39
EC50 = 8.50
7.50 μM). Other compounds were not active. Imperia-
line (8) has been reported to have significant potency in relaxing the
isolated tracheas [6]. Compound 4, differing structurally from 8 only in
the C-6 moiety, was inactive in this test. The carbonyl group at C-6
116
Hu, Zhu
