Dx, LC and DLC were obtained from Aldrich (Milwaukee, WI, U.S.A.), Merck (Darmstadt, Germany) and the National Institute of Hygenic Sciences (Japan), respectively. ADx and β-methyldigoxin (MDx; used as an internal standard, IS) were obtained from Boehringer Mannheim (Mannheim, Germany). Acetic acid and sodium methoxide were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Methanol (HPLC grade) and acetonitrile (HPLC grade) were purchased from Kanto Chemical Co., Inc. (Tokyo, Japan).

The HPLC system comprised a model L-7110 pump (Hitachi, Tokyo, Japan), a Rheodyne injection valve (Cotati. CA, USA) with a 20-μL loop, and a model L-7405 UV detector (Hitachi) operating at UV wavelength of 220 nm. The HPLC column (Cosmosil 5C₁₈-AR-300, Nacalai Tesque, Kyoto) was 150 × 4.6 mm i.d., packed with 5 μm particles of C₁₈ material. Quantification of the peaks was performed using a Model D-2500 chromato-integrator (Hitachi). The mobile phase consisted of acetonitrile: methanol: ultrapurewater (Milli-Q water purification system, Simplicity^® UV, Millipore Corporation, Bedford, MA, USA) = 10:100:90 (v/v/v). The samples were eluted from the column at room temperature at a flow rate of 0.5 mL/min. An aliquot of 15 to 20 μL was injected into the HPLC system.

D. lanata leaves were collected during the flowering stage in June 2000 at the Medical Plant Garden (Kanazawa, Japan) of Hokuriku University. The plants were obtained from Takeda Pharmaceutical Company Ltd. in June, 1976. The color and size of flowers have not been changed morphologically. The fresh leaves were immediately freeze-dried and then dried over P₂O₅ under reduced pressure at room temperature. The dried leaves were pulverized and sifted through a sieve of mesh width 500 μm. The dried leaf powder was further dried over P₂O₅ under reduced pressure for 5 days.

LC, DLC, ADx and Dx (each ca 2.5 mg) were each accurately weighed and dissolved in methanol (100 mL). Various volumes of prepared cardiac glycoside solution were added to 2 mL of MDx (IS) solution in methanol (18.81 mg/250 mL) to prepare standard solutions. Standard curves were obtained as described below.

The previous methods were modified for sample preparation [2] [3] . Briefly, dried leaf powder (ca. 50 mg) was accurately weighed and extracted with 50% methanol (1 mL). After ultrasonication for 1 h in an ultrasonic cleaning bath at room temperature, the extract was filtered through a filter paper (No. 2, ADVANTEC, Toyo Roshi Ltd., Tokyo, Japan) on a funnel fused-in fritted glass disc (SIBATA, Soka, Saitama, Japan). The residue was extracted with 50% methanol (1 mL) three times and filtered. Methanol (40 μL), acetic acid (50 μL), water (5 mL) and IS (75.24 μg, 2 mL) were added to the combined filtrate and the mixture was applied to a Sep-Pak C₁₈ cartridge (Waters). Methanol (5 mL) and water (5 mL) was used to activate the cartridge, which was then washed with 15% acetonitrile (30 mL). Target compounds were eluted with 30% acetonitrile (10 mL). Aliquots of the eluate (15 to 20 μL) were injected into the HPLC system.

The sample (ca. 50 mg) was accurately weighed and suspended in 0.05% acetic acid 3 mL. The suspension was incubated at 2 h for 50˚C, and 50% methanol (1 mL) was added. The mixture was sonicated for 1 h. Then, the same procedure was repeated.

A sample (ca. 50 mg) was extracted with 50% methanol (1 mL) with the aid of ultrasonication three times, fol-lowed by filtration as above (“Sample Preparation”). The residue was extracted with 2.8% sodium methoxide in methanol (40 μL) at 60˚C for 2 h and filtered. Then, the same procedure was repeated.

The sample (ca. 50 mg) was accurately weighed and suspended in 0.05% acetic acid 3 mL. The suspension was incubated at 2 h for 50˚C, and 50% methanol (1 mL) was added. After ultrasonication for 1 h in an ultrasonic cleaning bath, the suspension was filtered. The residue was extracted with 50% methanol (1 mL) three times and the solution was filtered. Next, 2.8% sodium methoxide in methanol (40 μL) was added to the combined filtrate and the mixture was incubated at 60˚C for 2 h. Then, the same procedure was repeated.

Analytical data were expressed as mean ± S.D. (n = 5). S.D. =

The chemical structures and molecular weights of LC, DLC, ADx, Dx and MDx are shown in Figure 1. A typical chromatogram is shown in Figure 2. Retention times of DLC, LC, Dx, ADx and IS were 8.4, 11.9, 12.9, 20.1 and 22.9 min, respectively. Standard curves of the four cardiac glycosides were constructed by plotting five

or six data points. The regression lines were y = 0.008331x − 0.002581 for DLC with a correlation coefficient (r) of 0.9988 (range 24.74 - 123.7 μg), y = 0.005854x − 0.001715 for LC (r = 0.9996, range 26.00 - 130.0 μg), y = 0.010359x − 0.010774 for Dx (r = 0.9994, range 25.58 - 127.9 μg), y = 0.009472x − 0.001506 for ADx (r = 0.9993, range 16.16 - 161.6 μg), where y represents the peak area ratio of each cardiac glycoside to the IS and x represents the amount (μg) added in preparing the standard solutions.

A typical chromatogram of cardiac glycosides obtained from D. lanata leaves is shown in Figure 3. LC, Dx and ADx were detected, and their amounts per 100 mg of leaves were estimated to be 115.6 ± 1.8, 7.45 ± 0.40 and 23.8 ± 1.2 μg (mean ± S.D.; n = 5), respectively (Table 1). These values correspond to 117.3, 9.5 and 28.9 nmol, and total cardiac glycosides amounted to 155.7 nmol. DLC was not detected.

The powder suspended in 0.05% acetic acid as a pretreatment, which is considered to activate digilanidase present in the leaves. The amounts of Dx and ADx per 100 mg of leaves after the pretreatment were estimated to be 21.1 ± 8.0 and 52.1 ± 7.5 μg (mean ± S.D.; n = 5), respectively (Table 2). These values correspond to 27.0 and 63.3 nmol. DLC and LC were not detected (data not shown). Total recovery was 58.0% compared with the control.

Addition of sodium methoxide to the residue of the powder extracted with 50% methanol is considered to cause hydrolysis of ester. The amounts of DLC and Dx per 100 mg of leaves after the pretreatment were estimated to be 65.6 ± 1.4 and 20.9 ± 0.4 μg (mean ± S.D.; n = 5), respectively (Table 3). These values correspond to 69.6 and 20.9 nmol. LC and ADx were not detected (data not shown). Total recovery was 61.9% compared with the control.

Based on the above results, we speculated that the combination of both pretreatments would allow us to isolate digoxin alone from D. lanata leaves. Indeed, only Dx was detected after the combined pretreatments, as shown in Figure 4, and the amount of digoxin per 100 mg of the leaves was estimated to be 115.1 ± 11.7 μg (mean ± S.D.; n = 5) (Table 4). Thus, compared with the control, the Dx level was increased about 15-fold.This amount of Dx corresponds to 147.4 nmol, and the recovery was 94.7% compared with the control. Thus, the combination

of these pretreatments provides excellent recovery of Dx from the leaves. It remains to be established whether other cardiac glycosides in the leaves are also converted to Dx by these pretreatments, thereby contributing to the high recovery.