White needles (methanol); melting point (mp), 195–197°C; [α]D²⁰+210° (c 0.10, methanol), exhibited a positive reaction to the ferric chloride-ferricyanatum kalium test. ¹H-nuclear magnetic resonance (NMR): δ:3.56 (1H, m, H-3), 3.63 (3H, s, OCH3–4′), 3.78 (1H, t, J=10.26 Hz, H-2a), 4.29 (1H, dd, J=10.98, 5.16 Hz, H-2a), 5.62 (1H, d, J=7.02 Hz, H-4), 6.37 (1H, d, J=2.4 Hz, H-8), 6.56 (1H, dd, J=2.22, 8.04 Hz, H-5′), 6.67 (1H, d, J=2.22 Hz, H-3′), 6.86 (1H, d, J=2.22 Hz, H-8), 6.94 (1H, dd, J=8.10, 2.58 Hz, H-6), 7.20 (1H, d, J=8.40 Hz, H-6′), 7.58 (1H, d, J=8.40 Hz, H-5). ¹³C-NMR: δ162.6 (C-4′), δ162.4 (C-2′), δ161.5 (C-7), δ158.4 (C-9), δ133.8 (C-5), δ126.3 (C-, 6′), δ121.0 (C-1′), δ112.9 (C-10), δ111.9 (C-6), δ107.5 (C-5′), δ105.2 (C-8), δ98.2 (C-3′), δ80.2 (C-4), δ67.7 (C-2), δ41.0 (C-3) and δ56.3 (C-OCH₃). Compound 1 was identified as 7,2′-dihydroxy-4′-methoxy-isoflavanol by comparison of Heteronuclear Single Quantum Correlation, H-H-Correlation Spectroscopy and Distortionless Enhancement by Polarisation Transfer data (11,12).

Colorless needles (methanol, pyridine) reacted positively to FeCl₃. The compound was purple-red when exposed to the UV lamp (254 nm), mp180–181°C (12); ¹H-NMR:δ7.53 (1H, d, J=8.82 Hz, H-1), 6.90 (1H, dd, J=8.46, 2.16 Hz, H-2), 6.84 (1H, s, H-7); δ6.83 (1H, d, J=2.22 Hz, H-4), 6.64 (1H, s, H-10), 5.91 (1H, d, J=1.14 Hz, −OCH₂O-aH); 5.88 (1H, d, J=1.14 Hz, −OCH₂O-bH), 5.57 (1H, d, J=7.32 Hz, H-11 a, 4.25 (1H, dd, J=11.04, 4.74 Hz, H-6β), 3.77 (1H, t, J=10.62 Hz, H-6α), 3.48 (1H, m, H-6a). ¹³C-NMR: δ40.5 (C-6a), 66.5 (C-6), 79.0 (C-11a), 93.7 (C-10), 101.5 (−OCH2-O), 104.0 (C-4), 105.3 (C-7), 110.7 (C-2), 111.8 (C-11b), 118.7 (C-6b), 132.5 (C-1), 141.9 (C-8), 148.3 (C-9), 154.7 (C-10a), 157.3 (C-4a), 160.4 (C-3). Mass spectrometry (MS): m/z: 307 [M+Na], m/z: 285 [M+H]. Compound 2 was identified as maackiain by comparison (¹H-NMR and ¹³C-NMR) with previous data (13,14).

White needles (methanol) that were easily soluble in methanol, mp217–219°C. 1H-NMR: δ:8.42 (1H, s, H-2), 8.04 (1H, d, J=8.76 Hz, H-5), 7.52 (2H, d, J=8.40 Hz, H-2′, 6′), 7.23 (1H, s, H-8), 7.13 (1H, d, J=8.82 Hz, H-6), 6.98 (2H, d, J=8.40 Hz, H-3′, 5′), 5.09 (1H, d, J=6.96 Hz, glc H-1′), 3.30 (3H, s, OCH₃). δ4.5–5.5: Glycosidic bond −OH, δ3–4 glycosidic bond −H. ¹³C-NMR: δ175.1 (C-4), δ161.9 (C-7), δ159.5 (C-4′), δ157.5 (C-9), δ154.1 (C-2), δ130.5 (C-2′, 6′), δ127.4 (C-5), δ124.5 (C-1′), δ123.9 (C-3), δ119.0 (C-10), δ116.1 (C-6), δ114.1 (C-3′5′), δ103.9 (C-8), δ100.5 (C-1′), δ77.7 (C-3′), δ77.0 (C-5′), δ73.6 (C-2′), δ70.1 (C-4′), δ61.1 (C-6′), δ55.6 (C-OCH₃). Compound 3 was identified as formononetin-7-O-β-D-glucoside by comparison (¹H-NMR and ¹³C-NMR) with previous data (15).

White needles (methanol), mp257–258°C; showed a positive reaction to the ferric chloride-ferricyanatum kalium test. ¹H-NMR: δ:8.30 (1H, s, H-2), 7.95 (1H, dd, J=8.76,2.22 Hz, H-5), 7.49 (2H, d, J=8.40 Hz, H-2′, 6′), 6.96 (1H, d, J=8.76 Hz, H-3′, 5′), 6.92 (1H, dd, J=8.76, 2.22 Hz, H-6), 6.85 (2H, d, J=2.22 Hz, H-8), 3.77 (3H, s, OCH3). ¹³C-NMR: δ175.1 (C-4), δ163.0 (C-4′), δ159.4 (C-7), δ157.9 (C-9), δ153.8 (C-3), δ130.5 (C-2′, 6′), δ127.8 (C-5), δ124.7 (C-1′), δ123.6 (C-2), δ117.1 (C-10), δ115.6 (C-6), δ114.1 (C-3′5′), δ102.6 (C-8). MS: m/z: 291 [M+Na], m/z: 269 [M+H]. Compound 4 was identified as formononetin by comparison with previous data (11,16).

White needles (methanol), soluble in dimethyl sulfoxide; ¹H-NMR: δ:8.32 (1H, s, H-8), 8.12 (1H, s, H-2), 7.30 (2H, s, H-NH2), 5.86 (1H, d, J=6.24 Hz, H-1′), 5.00–5.50 (3H, OH-H), 4.59 (1H, m, H-2′), 4.12 (1H, m, H-3′), 3.94 (1H, m, H-4′), 3.65 (1H, m, H-5′), 3.53 (1H, m, H-5′). ¹³C-NMR: δ156.6 (C-6), δ152.8 (C-2), δ149.6 (C-4), δ140.4 (C-8), δ119.8 (C-5), δ88.4 (C-1′), δ86.4 (C-4′), δ73.9 (C-2′), δ71.1 (C-3′), δ62.2 (C-5′). MS: m/z: 255 [M+Na], m/z: 237 [M+H]. Compound 5 was identified as 9-(β-D-ribofuranosyl)-adenosine by comparison (¹H-NMR and ¹³C-NMR) with literature data (17,18).

The structures of compounds 1–5 are shown in Fig. 1. The structures of compounds 6 and 7 were complex and thus remain to be identified.

The detection of the compounds was performed at 310 nm. Satisfactory separation was obtained with a reverse-phase analytical column (250×4.6 mm, 5 µl, serial no. 8132964) and eluted with a methanol:water solution (70:30, v/v) at a flow rate of 1.0 ml/min. The column temperature was 25°C and the injection volume was 10 µl. Theoretical plate numbers of formononetin and maackiain were >4,000.

i) Preparation of stock standard solutions: Formononetin and maackiain stock solution were prepared to a final concentration of 0.73 mg/ml in methanol. Chromatograms are shown in Fig. 2A and B.

ii) Sample preparation. The Pi Han Yao powder (3 g) was processed with ultrasonic extraction (power, 300 W; rate, 50 Hz) for 1 h in EtOAc, and subsequently recovered EtOAc in vacuo to yield the extract. The extract was dissolved with methanol in a 10 ml volumetric flask and filtered through a 0.45-µm filter prior to injection. The chromatogram is shown in Fig. 2C.

Calibration curves were constructed in the range of 2–20 µl for formononetin and maackiain. The regression equation curves are listed in Table I.

The data showed a linear association between peak areas and concentration over the listed range for formononetin and maackiain. The detection limit was 0.03992–0.3992 µg for formononetin and 0.0292–0.292 µg for maackiain.

The precision of injection was evaluated by performing replicate injections of the sample solution, and the relative standard deviations of formononetin and maackiain were 0.26 and 0.55%, respectively. These results demonstrated that the method was highly precise.

To confirm that the standards remained stable over time, the concentrations of the two components were determined at 0, 2, 4, 8, 12 and 24 h, respectively, and the solution remained relatively stable. The RSD values of formononetin and maackiain were maintained at 1.59 and 2.57%, respectively. The stability of the sample solutions was also evaluated in the same manner by determining the concentrations of the two components at 0, 2, 4, 8 and 12 h, respectively. The test solution remained relatively stable. The RSD values of formononetin and maackiain were maintained at 2.18 and 5.98%, respectively.

The percentage of recovery was determined by adding a known concentration of the formononetin and maackiain standards to selected samples during the extraction process. The amounts added were ~50% of the actual concentration of the samples. The concentration of the formononetin and the maackiain standards in the mixture were subsequently determined using the same methods described for the sample analysis. Without exception, recovery rates of 100.31% were achieved for the compound analyzed.

The formononetin and maackiain contents of Pi Han Yao from plants collected from 10 different sources were determined in the same way as the sample analysis. The results are shown in Table II.

Analysis of the different Pi Han Yao plants from the 10 locations showed that the contents of formononetin and maackiain varied depending on the source. The content of formononetin ranged from 0.576–0.762 µg/g with an average value of 0.6544 µg/g. The content of maackiain ranged from 0.0124–0.0289 µg/g, with an average value of 0.02221 µg/g. Overall the content of the two chemicals in wild Pi Han Yao was generally higher compared to cultivated plants. However, the difference was not significant. Furthermore, the content of formononetin and maackiain in Pi Han Yao was lower compared to the plant itself.