PART1: Chemical Diversity And Prediction Of Potential Cultivation Areas Of Cistanche Herbs

Contact: Audrey Hu Whatsapp/hp: 0086 13880143964 Email: audrey.hu@wecistanche.com

Owing to hostile growth environments and increasing related production, Cistanche plants have decreased in number. The aim of the present study was to evaluate the quality of and to predict potentially suitable regions for two offcial species and two nonofcial species (C. salsa and C. sinensis) by high-performance liquid chromatography and the MaxEnt model. The results indicated that 2′-acetylacteoside was present only in Cistanche deserticola. The compound can be used as a potential chemical marker to discriminate Cistanche deserticola from the three other Cistanche plants. Anthocyanin A and carotenoid F were the common constituents of the two ofcial species only and can thus be used as chemical markers to diferentiate between ofcial and nonofcial species. The prediction results of a potentially suitable distribution indicated that C. sinensis has much wider regions for potential distribution than the other species. Finally, the echinacoside content in Cistanche deserticola was signifcantly diferent between the two suitable potential distributions, and the contents of samples from Inner Mongolia were signifcantly higher than those from Gansu Province. This is the frst application of the combination of the contents of chemical components and the results of MaxEnt models for the quality assessment of herbal medicine. Our results may provide a reference for the sustainable utilization of endangered Cistanche species.

Since the 72nd World Health Assembly, traditional medicine has been included in the International Classification of Diseases (11th version), which mainly dates from traditional Chinese medicine and is also widely accepted in Southeast Asian countries such as Japan, Korea, and India. The conference indicated that traditional medicine plays an important role in health regulation. Medicinal herbs are the material base of traditional medicine in addition to acupuncture and manipulation. The accurate usage and combination of different herbs determine the ultimate efficacy of traditional medicine1–3. In addition, the content and type of active components in herbs fluctuate with their botanical origins and growth environments. Therefore, proper botanical origins and suitable growth environments are vital for clinical efficacy and security.

Cistanche plants are utilized as precious tonic and edible herbs for men’s care in China, Japan, and some Southeast Asian countries4,5. Four species of the genus are known for their potential medicinal value in China, namely, Cistanche deserticola, Cistanche tubulosa, Cistanche salsa, and Cistanche Sinensis. The dried succulent stems of the first two species are recorded in the Chinese Pharmacopoeia (2015 edition), named Cistanches Herba. These species are beneficial to the kidneys and intestinal tract. The other two species are consumed in some locations or are utilized as adulterants of Cistanches Herba. C. salsa was recorded in the local herbal standards in Gansu (1992 edition) and Xinjiang (1987 edition) provinces. C. Sinensis is a unique species in China. Modern chemical and pharmacological studies have shown that chemical components from these herbs have several effects, such as brain function improvement, aphrodisiac effects, and immune-boosting effects.

Published studies indicate the content and type diference of chemical constituents among diferent Cistanche plants, in which betaine, Krebs cycle intermediates, phenylethanoid glycosides, and iridoids were regarded as four chemical markers responsible for discrimination analysis between Cistanche deserticola and Cistanche tubulosa 7,8. A detailed study indicated that the isomers of campneoside II, cistanoside C, and cistanoside A were three potential

chemical markers to distinguish the two species mentioned above9. Further research showed that eight phenylethanoid glycosides could be chosen as chemical markers for discriminating the Cistanche species, which mainly include Cistanche deserticola, Cistanche tubulosa, and C. sinensis10. In addition, the contents of chemical constituents vary with geographical origin. Zhou and his co-authors indicated that the total contents of seven index components (cas- tanoside A, echinacoside, isoacteoside, 2′-actylacteoside, castanoside C, and tubluoside B) of Cistanche tubulosa from south of Xinjiang were approximately six times those of Kuitun and Hami in China11. Among these chemical components, acteoside and echinacoside are regarded as index components for the quality control of herbal medicine. The two components have been reported to improve brain function, and acteoside mainly contributes to the aphrodisiac effects.

Species distribution models (SDMs) are statistical models established with existing environmental variables to infer species’ ecological requirements. Such models can map a target species’ potential distributions on the basis of the observed distributional data12. To date, SDMs have been successfully applied to the prediction of the distribution tendency of endangered and ecological plants during climate change13,14. Among these models, the MaxEnt model is commonly used as a simple means of predicting the habitat suitability distribution with presence-only data and performs well with incomplete data, small sample sizes, and gaps14. Combined with some software (such as ArcGIS), research can extract environmental and climate variables of collection sites of samples. Furthermore, a potentially suitable location can be divided into three or four levels15–17. However, no study has reported combined strategies that incorporate potential suitable locations after division and chemical component contents for the quality control of herbal medicine.

In the present study, four endangered Cistanche plants were collected for quality assessment with high-performance liquid chromatography (HPLC), and seven chemical components were used as indices. Furthermore, a MaxEnt model equipped with 26 environmental variables was utilized to predict potentially suitable areas for investigating optimal species for further usage. Finally, the combination of the index component contents and suitable areas was systematically analyzed for better development and utilization of these endangered Cistanche plants in China and other countries.

Results

The contents of the seven chemical components were calculated on the basis of their calibration curves. The seven equations of the calibration curves were Y = 20.5400X+0.1616 (2′-acety- lacteoside, R = 0.9992), Y = 18.5640X+0.3724 (acteoside, R = 0.9821), Y = 15.6350X+0.0090 (isoacteoside, R = 0.9999), Y = 12.7460X+0.0103 (tubuloside A, R = 0.9999), Y = 10.638X+0.1615 (cistanoside A, R = 0.9997),Y = 21.533X−0.0006 (cistanoside F, R = 1), and Y = 13.0740X+1.1381 (echinacoside, R = 0.9821); the concentration (X) is the horizontal axis, and the peak area (Y) is the vertical axis. R is the correlation coefcient and indicates an excellent linear correlation between these calibration curves. A precision experiment (six sequential injections of the same sample extractives) showed that the precision was good, and the relative standard deviation (RSD) of the peak area was between 0.3% and 0.8%. Similarly, the repeatability (six sequential injections of the same sample extractives) and stability tests (six injections of the same sample extractives after 0, 2, 4, 8, 12, and 24 h) yielded RSDs of 0.56–1.46% and 1.06–3.60%, respectively.

Further content analyses of the seven index constituents were displayed in grouped horizontal boxes con- taining visual comparison results and signifcant diferences in lowercase. The recovery test results of the seven constituents varied from 99.25% to 104.10% with RSDs between 0.69% and 3.45%. Therefore, the method was accurate.

The contents of the seven index constituents in the four species and the geographical origins of the same species are displayed in grouped horizontal boxes. The contents of some chemical components were under the detection limit. Thus, only a portion of the fgures contained four Cistanche species. The chemical component 2′-acetylacteoside was detected only in Cistanche deserticola, whereas it was undetected in the other species by the present method. However, a content diference was observed among the various geographical origins focused on 2′-acetylacteoside. The contents of the component from Alashanzuoqi in Inner Mongolia were signifcantly higher than those in the three other provinces, whereas the contents in the species from Tingtuhu of Minqinin, Gansu Province, were signifcantly lower than those in the other locations (Fig. 1A). Tubuloside A was detected in Cistanche deserticola and Cistanche tubulosa. No remarkable difference was observed between the two species and even among plants of the same species from different geographical origins (Fig. 1B). The four species contained the same chemical component (acteoside). The content of acteoside in C. salsa collected from Jianga’erhan of Tacheng in Xinjiang was signifcantly higher than that of the other species and the same species from diferent growing areas. In contrast, the content of acteoside did not difer signifcantly among the various origins of Cistanche deserticola (this fnding was the same as that for tubuloside A; Fig. 1C). Cistanoside A was detected in Cistanche deserticola and C. salsa. The content of the C. salsa samples from Jianga’erhan of Tacheng in Xinjiang was signifcantly higher than that of Cistanche deserticola samples from other places. Moreover, the variation in the location did not afect the content of the component. This condition was the same as that for tubuloside A and acteoside (Fig. 1D). For cistanoside A, the content of the index component did not fuctuate with the variations in species and geographical origins (Fig. 1E). The echinacoside content of the C. salsa samples from Hejiaoke of Tuoli in Xinjiang was signifcantly higher than the content of the samples from Jianga’erhan of Tacheng in Xinjiang. The component contents of the samples from Cistanche deserticola and Cistanche tubulosa, regardless of where they were collected, were signifcantly lower than in other locations (Jianga’erhan of Tacheng in Xinjiang). In addition, the chemical component could not be detected in C. sinensis (Fig. 1F). Isoacteoside was detected in all Cistanche species except C. salsa; the isoacteoside content in Cistanche tubulosa was signifcantly higher than that in the two other species regardless of the source provinces in China (Fig. 1G). Herein, all of the signifcant diferences were less than P < 0.5. The total contents of these chemical components are stacked in Fig. 2 for improved comparison of the seven styrene glycosides. Herein, we hypothesized that geographical origins exerted a small infuence on the Potentially suitable areas in the world and in China.

With the combined calculations from MaxEnt and ArcGIS, the potential suitable areas for the four Cistanche species of herbs were calculated. The results are shown in Table 1. Similar to the categories in the above sections, the potentially suitable areas were divided into four classes. Regarding the high habitat suitability class, three species had approximately 30 km2 larger suitable areas worldwide (Cistanche deserticola: 29.2210 km2, C. sinensis: 31.3034 km2, C. salsa: 30.8633 km2) compared with C. tubu- losa (10.6866 km2). As shown by the comparison of the high habitat suitability class, Cistanche deserticola had the largest acreage (38.9267 km2) in terms of the moderately suitable class (Cistanche tubulosa: 12.4069 km2, C. sinensis: 32.1292 km2, C. salsa: 38.5764 km2), while the area of Cistanche tubulosa was still the smallest.

In terms of the potential areas regarded as a potential distribution of suitable cultivation locations, the high habitat suitability and moderately suitable classes are discussed as follows. Most of the suitable areas (approximately 90%) for Cistanche deserticola were mainly distributed in the north and northwest of China. In addition, a few areas with potential cultivation conditions were identifed in the south of the United States, southwestern Iran, eastern Turkey, southern Mongolia, and eastern Kazakhstan (Fig. 3). Unlike the abovementioned species, C. tub- ulosa seemed to be well distributed worldwide, covering the west of China, northwest of South America, north of Africa, and some countries between China and Africa. However, although Cistanche tubulosa had a wide potential distribution, the total area of the species was the smallest compared with the other three species, for which the total areas of the high habitat suitability and moderate classes were 23.0935 km2. Regarding the two classes, only a few potential areas suitable for the growth of the species were found in northern Egypt, in the western region of the Kingdom of Saudi Arabia, southern Yemen, northern United Arab Emirates, southern Iran, and Pakistan. The largest areas were still in China, where the total prediction areas were located in the west of the country (Fig. 4). Similar to Cistanche deserticola, C. salsa was mainly distributed in the northwest and north of China, where the specimen has records. After prediction by the MaxEnt model, there was a small prediction area that may be a suitable growth environment in the United States and eastern of Kazakhstan and Kyrghyzstan (Fig. 5). Most of the predicted locations of C. sinensis were distributed in northern and western China. Additionally, a few areas were identifed in the middle part of Morocco, southeastern Algeria, and northwestern Chad and Iran (Fig. 6).

The potentially suitable areas calculated by MaxEnt are displayed in the bottom right corner of Figs. 3–6 to show the prediction distribution of the four species in China. Herein, the smallest units of prediction distribution mapping were based on the provinces in China. Similar to the analysis for the world, the analysis in this section was based on high habitat suitability and moderate classes. The prediction areas indicated that Cistanche deserticola was mainly distributed in Xinjiang, Inner Mongolia, Ningxia, Gansu, and Qinghai provinces. Although a few distribution areas were found in other provinces, they belonged to Class 2. Therefore, these regions were poorly suitable and thus not recommended for the cultivation of the species. For Cistanche tubulosa, all potentially suitable regions of the species were predicted in Xinjiang Province in China. Interestingly, C. salsa had a wide range of potential distributions in China, including Xinjiang, Inner Mongolia, Ningxia, Gansu, Qinghai, Shanxi, Shaanxi and Hebei provinces. Moreover, the entire Ningxia Province seemed to be a possible cultivation area for the species. Similar to C. salsa, C. Sinensis also had wide regions for potential distribution excluding Xinjiang Province (but it had a prediction area in Xizang Province)

evaluation of each species model.

Model performance was divided into fve categories: fail, poor, fair, good, and excellent (0.9 < AUC ≤ 1). The model performance in terms of the AUC value and ROC curve among the four models is displayed in Fig. 7. The results indicated that the four models’ performance in the training data was excellent, with AUC values between 0.997 and 0.998 and a top left corner ROC curve close to 1. Moreover, the AUC value in the test data also showed good model performance, with a high value between 0.979 and 0.996 and a top left corner ROC curve close to 1. In general, these models have a strong capability to predict potential species distribution.

Variables’ response analysis to suitability.

Response curves show the quantitative relationship between habitat suitability and environmental variables (also known as the logistic probability of presence)14. In the present study, the common variables between the jackknife test and permutation importance (percentage contribution) were used to analyze the variables’ response to suitability. In terms of Cistanche deserticola, bio09 and wind10 were common variables. The suitable mean temperature of the driest quarter (bio09) was between −12.71227957 °C and −1.296478534 °C. The optimal wind speed in October (wind10), which is benefcial for the distribution of the species, was between 0.5841 and 3.2581 m·s−1 (Fig. S5). Two common variables were identifed for the potential distribution in the Cistanche tubulosa model. The first one was solar radiation in May, with a suitable range exceeding 24661.0800 kJ·m−2·day−1. The other was precipitation in October (pre10), with an optimal value lower than 5.4640 mm (Fig. S6). The results of the jackknife test showed that four variables were the same as those of the two other evaluation methods for the contribution of the bioclimatic factors in the C. salsa model. The frst two variables comprised solar radiation in June and November (srad06 and srad11), with ranges of 23298.9691–25137.457 and 6709.6219–10687.2852 kJ·m−2·day−1, respectively. The third was the mean temperature of the driest quarter (bio09), with an optimal range of -12.6159–2.8479 °C. The fourth variable was precipitation in September (pre09), with a suitable range of 8.3762–32.4312 mm (Fig. S7). The MaxEnt model of C. Sinensis had three important variables that mostly contributed to the potential distribution of the species. The suitable range of solar radiation in December (srad12) was 6138.3161–9488.8316 kJ·m−2·day−1, and that of maximum temperature in November (tmax11) was -1.9115–9.6048 °C. Furthermore, the suitable range of wind speed in August (wind08) was 2.1005– 3.3054 m·s−1 (Fig. S8). The other important variables contributing to the potential distribution of the four species are summarized in Figs. S5–S8.

Correlation between chemical components and bioclimatic variables. The most popular species from the main distribution provinces (Cistanche deserticola) was compared to fnish the correlation analysis and to investigate the content fuctuation of seven index components with the environmental condition. After analysis of the suitable potential distribution of the species, these potential distribution regions were divided into four categories. Herein, six collection sites including 22 samples were mapped on the China map extracted from the results of the MaxEnt model of the species (Fig. 8). The results indicated that two collection sites were situated in Class 1 (high habitat suitability class), and two collection locations belonged to Class 2 (moderately suitable class). The two fnal sites were classifed as Class 3 (low suitable class). A significant difference in the 2′-acetyl acteoside content was observed between CD-TG (Tingtuhu, Minqin, Gansu Province) and CD-AI (Alashanzuoqi, Inner Mongolia), which belonged to Classes 2 and 1, respectively. In addition, the content of the component in CD-AX (Aibi Lake region, Tacheng, Xinjiang) and CD-CG (Changcheng, Liangzhou, Gansu Province), which belonged to Class 3, did not significantly differ from those of the four other sites, which belonged to Class 1 and Class 2. For the six other chemical components, no signifcant diference was identifed among the diferent geographical origins in terms of the content of these constituents.

Discussion

Herbal medicine is a complex mixture of many chemical components that contribute to the systematic targeting of disease. In addition, one type of herbal material may be collected from diferent botanical species that have similar characteristics or efective components. For instance, Coptidis Rhizoma is a herbal medicine with high-frequency usage; it is obtained from three congeneric plants (Coptis chinensis, C. deltoidea, and C. teeta) and used to cure diarrhea induced by dampness retention with the help of various berberines18. In addition, Polygonatum Rhizoma is commonly utilized as a functional food in daily life in China for immunoregulation; the crude materials are rhizomes of Polygonatum kingianum, P. sibiricum, and P. cyrtonema19. The present study included four species, two of which are ofcially recorded in the China Pharmacopoeia (Cistanche deserticola and Cistanche tubulosa in the 2015 version).