Quality assessment of Japanese knotweed (Fallopia japonica) grown on Prince Edward Island as a source of resveratrol.

Japanese knotweed (Fallopia japonica , also known as Polygonum cuspidatum) is a common invasive plant species on Prince Edward Island (PEI), Canada, whereas it has been used in Chinese medicine and more recently as a raw material for extracting resveratrol. This paper reports on the quantification of resveratrol, polydatin, emodin, and physcion in roots, stems, and leaves of Japanese knotweed samples from PEI and British Columbia (BC), Canada, and nine provinces of China, by ultraperformance liquid chromatography (UPLC). The results showed that the root contains a much higher level of resveratrol than the stem and leaf, and it is accumulated in its highest level in October. PEI-grown knotweed contains similar levels of resveratrol and polydatin compared to Chinese samples collected in the month of October, but the contents of the other anthraquinones (emodin and physcion) are different. As such, Japanese knotweed grown in PEI could be a commercially viable source of raw material for resveratrol production; however, caution has to be taken in harvesting the right plant species.


■ INTRODUCTION
Fallopia japonica (Houtt.) Ronse Decraene (previously known as Polygonum cuspidatum Sieb. & Zucc.), a well-known Chinese herb and officially listed in the Chinese Pharmacopoeia, has been traditionally used for treatment of various inflammatory diseases, hepatitis, tumors, and diarrhea in Eastern Asian countries such as China, Korea, and Japan. 1−6 In North America, the young stem of F. japonica (Japanese knotweed) is edible and was consumed as a vegetable in the past. 7 Japanese knotweed is listed as one of the "World's Worst" invaders, and the farming and landscaping sectors have to control its growth and invasive nature. 8 Over the past decades, the chemical composition 9−11 and biological activity 4,5,12,13 as well as quality control 2,9 of F. japonica have been widely reported, and resveratrol, polydatin, emodin, physcion, chrysophanol, rhein, and emodin-8-O-β-D-glucoside were revealed as some of the main constituents.
Resveratrol, or trans-3,5,4′-trihydroxystilbene, is a naturally occurring antioxidant compound typically associated with grapes and red wine 14,15 and is also present in F. japonica. 16 Extensive research on resveratrol over the past decade has demonstrated significant beneficial effects including the promotion of heart health, neuroprotection, anti-inflammation, and anticancer and antiaging effects, as well as diabetes and obesity prevention. 17 −20 These documented health benefits have raised the profile and interest in resveratrol as an active ingredient in the growing nutraceutical and cosmeceutical industries. Due to the relatively low level of resveratrol present in grape skin, 14,21 other plants have been investigated as potential and commercially viable natural sources of resveratrol for the growing nutraceutical market. The rhizome or root of F. japonica, also known as Japanese knotweed, is currently a major source for the natural resveratrol ingredients sold in dietary supplement markets worldwide. 22,23 On Prince Edward Island (PEI), and other places in Canada, Japanese knotweed is a common invasive plant species. However, until now no information was available on the level of resveratrol in knotweed grown in PEI.
As the demand for natural resveratrol is expected to grow rapidly in the future with increasing applications in the nutraceutical, therapeutic, and cosmeceutical markets, assessing PEI-grown knotweed as a commercial source could present new market opportunities for agriculture and bioresource sectors in the region. In this study, more than 50 knotweed samples collected from 4 different sites from PEI over various seasons have been analyzed for their levels of resveratrol and other components. Knotweed samples from China, which have been used as raw material for resveratrol production, were also studied in parallel.

■ MATERIALS AND METHODS
Instrumentation and Reagents. UPLC analysis was carried out on an Agilent 1200 liquid chromatography system, equipped with a vacuum     small, acuminate, entire leaves with blades showing truncate bases. Limbs are slightly longer than wide, no longer than 11 cm on the voucher specimen. Veins on the abaxial surface are devoid of hairs or trichomes; they show irregular knobs as shown in Zika and Jacobson. 26 The presence or absence of hairs or knobs is the best character that differentiates Fallopia sachalinensis and F. × bohemica from F. japonica var. japonica. The inflorescences are loose and longer than the subtending leaf. All of the other specimens were identified as Fallopia × bohemica Inflorescences are compact and short (much shorter than the subtending leaf). On the abaxial surface, short hairs are conspicuous along the veins, which is characteristic of the hybrid. The last specimen (INH-378-382) shows characters from the species and the hybrid. Leaves are shorter than the latter specimens and oscillate around 12 cm long. Bases are again truncate (or slightly cordate), and blades terminate with an acuminate apex. The presence of short hairs instead of knobs and the fact that the inflorescences are slightly shorter than the subtending leaf indicate that we have the hybrid. Although shorter, inflorescences are less compact than the other F. × bohemica specimens.
Sample Preparation. Fresh samples were cleaned with water and then milled into powder by a versatile plant grinder after being air-dried and then oven-dried (50−60°C) right before grinding. Before weighing out, all samples were further dried on a freeze-dryer to remove moisture.
Extraction Method. A 0.4 g aliquot of dry sample was accurately weighed out, and 10 mL of methanol/water (50:50, v/v) was added; the sample was then placed into an ultrasonic cleaner. The mixture was extracted for 30 min, and then the extraction solution was centrifugally separated and filtered through 0.45 μm Millipore nylon membrane, and an aliquot of the filtrate was used for UPLC analysis.
Preparation of Reference Substance Solutions. Resveratrol, polydatin, emodin, and physcion are the common components reportedly present in F. japonica, so they were commercially sourced and used as the reference substances in this study. Stock solutions of resveratrol (0.20 mg/mL), polydatin (0.60 mg/mL), emodin (0.25 mg/ mL), and physcion (0.10 mg/mL) were prepared by dissolving the pure compounds in methanol. After filtering through a 0.45 μm Millipore membrane, the reference substance solution was directly used for UPLC analysis.
UPLC Analysis. A reverse phase SB-C18 column (2.1 × 100 mm, 1.8 μm) from Agilent was used with the mobile phase consisting of

■ RESULTS AND DISCUSSION
Optimization of UPLC-DAD Method. To obtain optimal conditions for separation and quantification, the mobile phase and its flow rate, conditions for elution, column temperature, and detection wavelength were investigated in this study.
The chromatographic conditions were optimized on the basis of conditions given in the literature. 27,28 In the gradient optimization, gradient time, shape, and initial composition of the mobile phase were taken into consideration, and the optimized gradient elution was presented above under UPLC Analysis. Column temperatures of 45, 50, and 55°C and flow rates of 0.4, 0.5, 0.8, and 1.0 mL/min were also investigated. Considering system back pressure and running time, the flow rate was set at 0.5 mL/min when the column temperature was kept at 50°C. Also, the detection wavelength was examined. When the wavelength was at 303 nm, most components showed strong responses; thus, the DAD detection was performed at 303 nm.
Under these optimal conditions, resveratrol, polydatin, emodin, physcion, and other minor components in the extracts of F. japonica samples were well separated. All method validation   tests below were carried out on F. japonica extracts prepared as described above. The injection precision was determined by repeating UPLC injection of the same sample solution six times per day. The sample stability was determined with measurements from a single sample solution stored at room temperature for 2, 4, 6, 8, and 12 h. The repeatability was assessed by analyzing six separate samples. The relative peak area (RPA) of each characteristic peak was calculated for the estimation of injection precision, stability, and repeatability and the results were as follows: for injection precision, the relative standard deviations (RSD) of the RPA < 0.56%; for sample stability, RSD < 1.68%; for repeatability, RSD < 1.71% for six independent samples. Thus, the results indicated that the UPLC-DAD method was suitable for this quantitative analysis work.
Then, the extraction solvent, a key factor for UE, was evaluated. Three solvent mixtures were tested, methanol− ethanol, methanol−water, and acetonitrile−methanol, in a ratio of 80:20 (v/v) to evaluate their efficiency for extraction of resveratrol, polydatin, emodin, and physcion from F. japonica.
After ultrasonic extraction, the levels of resveratrol, polydatin, emodin, and physcion by using the three solvent mixtures were found to be 4.43, 6.24, 6.47, and 5.52 mg/g; 4.72, 6.57, 6.59, and 5.55 mg/g; and 4.36, 6.15, 6.07, and 5.31 mg/g, respectively ( Figure 1B). This showed that the extraction of resveratrol, polydatin, emodin, and physcion with methanol−water (80: 20) was the best. For dry plant materials, solvents such as methanol− water with higher polarity could penetrate plant cell membranes and enhance the extraction of bioactive components. Therefore, the methanol−water (80:20) was chosen as the solvent for further investigation.
Subsequently, the effects of different factors on ultrasonic extraction of resveratrol, polydatin, emodin, and physcion were investigated, and orthogonal experiment design was used to determine the significance of four environmental factors including solvent/sample ratio (mL/g), methanol/water ratio (v/v), extraction time (min), and times of extraction (number of times).
The experimental factors with corresponding levels are presented in Table 2A, and the results of orthogonal design L 9 (3 4 ) are presented in Table 2B. The intuitional analysis from R value in Table 2B indicates that the influence order of each variable on the extraction efficiency is A > C > D > B. It also showed that K 1 had the highest K value in factor A, which means that compared with other levels of factor A, level 1 has the highest extraction efficiency and is the most suitable level for extraction. On the basis of the same principle, level 3 of factor B, level 1 of factor C, and level 1 of factor D were chosen for extraction. To verify the data analysis of orthogonal experiment design, analysis of variance (ANOVA) was used to further determine which factor was significant in affecting the extraction yield of resveratrol, polydatin, emodin, and physcion. As shown in Table 2C, factors A, C, and D are clearly the most significant ones (P < 0.05) affecting extraction.
The orthogonal experiment design combined with statistical analysis had been applied to optimize the conditions of extraction for improving the yield for resveratrol, polydatin, emodin, and physcion in F. japonica, and the optimal extraction condition was used as described above.
Results of Analysis. Chromatograms of Reference Compounds and F. japonica Extract. F. japonica contains emodin, emodin-6-ether, emodin-8-monomethyl ether, chrysophanol, rhein, emodin-8-β-D-glucoside, physcion, resveratrol, and polydatin, with resveratrol, polydatin, emodin, and physcion  grown knotweed shows a similar composition profile in comparison with the Chinese samples except that it contains an additional, more polar analogue of resveratrol, identified as resveratroloside 29 (based on NMR and MS analysis of pure compounds obtained through preparative HPLC, data not shown).
Contents of Resveratrol, Polydatin, Emodin and Physcion in Different Plant Parts of Knotweed Grown in PEI, Canada. It is well-known that different parts of a raw herb may have somewhat different composition profiles and thus may be used for different therapeutic purposes. In this study, the contents of resveratrol, polydatin, emodin, and physcion of various parts of F. japonica grown in PEI were measured and compared. Root, stem, flower, and leaf samples of F. japonica collected from PEI were extracted and analyzed on UPLC. The contents of resveratrol, polydatin, emodin, and physcion are shown in Table 3. The contents of resveratrol, polydatin, emodin, and physcion were found to vary considerably in various parts of the plant and samples collected during different seasons (Figure 3).
Contents of Resveratrol, Polydatin, Emodin, and Physcion in the Root of F. japonica Harvested at Different Times in PEI, Canada. As shown above, different plant parts contain various levels of resveratrol, polydatin, and emodin, and the root was found to contain higher levels of these compounds. Therefore, we studied the effect of harvesting time on the levels of resveratrol, polydatin, emodin, and physcion various in root samples of F. japonica, and the results are shown in Figure 4. The results indicated that polydatin in F. japonica increases from May to July, reaches its highest in August, and then begins to decrease. Resveratrol, on the other hand, gradually increased from May to October, and the level was 2.5 times higher in October than in May. However, emodin in F. japonica appeared not to have a particular trend of accumulation during the year, whereas physcion seemed to decrease gradually from May to October.
Comparison on Contents of Polydatin, Resveratrol, Emodin, and Physcion among Root Samples of F. japonica from China and PEI, Canada. It is well-known that different growth environments have an enormous impact on the content of active ingredients in plants. 30 China is in eastern Asia and Canada is located in North America, so there is a great difference of climate and environmental factors related to plant growth. It is thus very interesting to compare the levels of these key components among the F. japonica root samples.
As shown in Table 4 and Figure 5, polydatin level was found to be within a similar range in both Chinese and PEI samples; the average content of polydatin in the root of F. japonica grown in PEI is about 11.04 mg/g, slightly higher than the average of Chinese samples (9.27 mg/g). The level of resveratrol in the Chinese and PEI samples differs considerably, with the average content in Chinese samples being 4.30 mg/g, 1.6 times higher than that of the PEI samples. However, if we choose only the samples harvested in October for comparison, the contents of resveratrol appeared to be at similar levels. Chinese samples also contain a higher level (1.3 times) of emodin and less physcion than the PEI samples (6.72 mg/g emodin and 10.31 mg/g physcion in Chinese samples; and 5.15 and 12.58 mg/g in PEI samples).
Comparison of Contents of Polydatin, Resveratrol, Emodin, and Physcion among Root Samples of F. japonica, F. × bohemica, and Polygonum polystachyum Collected at Different Sites from PEI, and Acquired from BC, Canada, and China. In this study, we compared the contents of polydatin, resveratrol, emodin, and physcion in all root samples, including 20 samples collected at three different locations (4 sites) in PEI, Canada, 3 samples acquired from BC, Canada (as shown in Table  1), and 14 samples from China. The results indicated that the contents of resveratrol and polydatin were higher in Japanese knotweed, F. japonica,g r o w ni nP E Io rf r o mC h i n a .I n comparison, we observed much lower levels of these compounds in other related species, F. × bohemica and P. polystachyum, with a difference of >10 times in some samples.
This difference of key metabolite abundances was further demonstrated by using a multivariate data analysis tool, which showed a fairly clear separation of F. japonica from other related species, based on levels of the four compounds analyzed in root samples. As shown in Figure 6, the score and loading plots (A and B) of the partial least-squares discriminant analysis (PLS-DA) demonstrated that F. japonica contains distinctly higher levels of resveratrol, polydatin, emodin, and physcion. One sample (no. 72), acquired from BC, Canada, as F. japonica, seems to contain lower levels of resveratrol and other compounds, indicating the effect of environmental factors.
In summary, this study showed that the content of resveratrol in Japanese knotweed (F. japonica) grown in PEI, Canada, varies in different parts of the plant and that the root contains a higher level of resveratrol than stem and leaf. The level of resveratrol in root samples of F. japonica from PEI is comparable to that of Chinese samples, whereas a significant difference was found between the common Japanese knotweed (F. japonica) and its closely related species F. × bohemica. Japanese knotweed contains >10 times the level of resveratrol as F. × bohemica. Also, time of collection was found to have a great influence on the contents of resveratrol; resveratrol level is at its highest in October. Additionally, PEI-grown knotweed contains a similar levels of resveratrol and polydatin as the Chinese samples collected in the month of October, but the contents of emodin and physcion are statistically different (lower for emodin and higher for physcion). As such, we conclude that Japanese knotweed grown in PEI, Canada, is a commercially viable source of raw material for resveratrol production.