Adonis amurensis Is a Promising Alternative to Haematococcus as a Resource for Natural Esterified (3S,3′S)-Astaxanthin Production

Astaxanthin (AST) characteristics and pigment productivity of Adonis amurensis, one of the few AST-producing higher plants, have not yet been studied extensively. In this study, the geometrical and optical isomers of AST in different parts of the A. amurensis flower were determined in detail, followed by a separation of the all-trans AST using HPLC chromatography. AST extracted from the flower accounted for 1.31% of the dry weight (dw) and mainly existed in the di-esterified form (>86.5%). The highest concentration was found in the upper red part of the petal (3.31% dw). One optical isomer (3S, 3′S) of AST, with five geometrical isomers (all-trans, 9-cis, 13-cis, 15-cis, and di-cis) were observed in all parts of the flower. All-trans AST was the predominant geometrical isomer accounting for 72.5% of the total content of geometric isomers in total flower, followed by the 13-cis, and 9-cis isomers. The all-trans AST isomer was also isolated, and then purified by HPLC from the crude oily flower extract, with a 21.5% recovery yield. The cis-AST extracted from the combined androecium and gynoecium gives a very strong absorption in the UVA region due to a high level of cis, especially di-cis, isomers, suggesting a prospective use in the preparation of anti-ultraviolet agents. The production cost of AST from Adonis flowers can be as low as €388–393/kg. These observations together with other factors such as the low technology requirement for plant culturing and harvesting suggest Adonis has great potential as a resource for natural esterified (3S,3′S)-AST production when compared with Haematococcus culturing.


Introduction
Astaxanthin (AST), a pigment of commercial interest, imparts an attractive blue to reddish hue to the feathers, carapace, skin, and flesh of many animals, such as flamingo, shrimp, krill, crabs, salmon, and red snapper [1][2][3][4]. This lipid-soluble pigment has substantial economic value as a nutraceutical, feed additive for aquatic animals, and potential source of new pharmaceuticals for the treatment of diseases caused by oxidative stress [3][4][5][6]. AST-containing products are increasingly popular as human health food supplements, with a market size of over US$100 million in 2018, with double-figure annual growth rates [7]. These supplements are taken for many different reasons, including improving eye health and vision, skin health, and enhancing athletic performance by speeding muscle recovery after exercise. The global market for AST is rapidly increasing; however, the bulk of AST available is produced synthetically, providing an opportunity for marketing a natural product [8,9].
The de novo biosynthesis of AST occurs in certain unicellular algae, some bacteria, and fungi as well as Adonis flowers. So far, natural AST mainly comes from Haematococcus Red-flowered Adonis species are the only AST-producing higher plants [33]. Different from the culture of H. pluvialis, Adonis farming can be performed readily by farmers, based on their extensive experience growing other crops. Most species of Adonis grow best in moist soil with plenty of humus. Large-scale Adonis farming can be conducted in the traditional solar greenhouse covered by a transparent plastic film. Mature agricultural cultivation experience, especially in agricultural greenhouse cultivation, is sufficient for the cultivation of Adonis [34,35]. The cultivation of this plant has the advantages of requiring only easy, relatively low-cost, farming technology, making it a low-risk investment that is easy to decentralize [36]. So far, the biosynthetic pathway of astaxanthin in A. aestivalis flowers has been elucidated [37,38]. The carotenoids and their fatty acid esters in the petals of A. aestivalis were also identified [39]. The amount of astaxanthin that accumulates in the flower petals of A. annua is reported to be about 1% of the dry weight [40]. However, existing reports have focused on the extraction methods [40][41][42][43], biosynthetic pathway analysis [37,38], and product quality improvement [39,44] of AST in Adonis. Up to now, however, no report on the distribution and characteristics of AST and its isomers in different parts of the Adonis flower has been reported.
We extracted the AST from different parts of the A. amurensis flower, and then, analyzed the extracts using UV-visible spectroscopy and high-performance liquid chromatography (HPLC) methods. Related carotenoid substances and AST compositions, i.e., mono/diester and geometrical/optical isomers, were also identified and quantified, followed by purification of the all-trans isomer from the crude extracts of the total flower. Based on the test data, we further compared the feasibility of natural AST production comparing the microalga H. pluvialis and A. amurensis. To our knowledge, this is the first time AST production based on the quality of AST in A. amurensis flowers has been discussed. We expect that these data will help encourage high-yield production of AST from this plant.

UV-Visible Absorption Spectra of Acetone Extracts
Lipid-soluble pigments from A. amurensis have a visible absorption spectrum in acetone similar to that of the AST standard, with a maximum absorption wavelength (λ max ) of 480 nm ( Figure 2). Peaks at 665 nm, observed in the spectra of the sepal and the combined androecium and gynoecium, indicated the presence of a trace of Chl a in these parts of the flower. Extracts of the flower, especially the androecium and gynoecium and the sepal, also showed a peak around 330-340 nm, the specific peak for cisisomers of carotenoids. Inflections in the blue region in the gynoecium and androecium and sepal fractions can also be observed.

Pigment Contents of Flower Parts
Different flower parts showed a similar level of moisture, which ranged from 67.70% to 73.85% (Table 1). The highest carotenoid content was obtained from the red part of the petal (3.31%); in contrast, the sepal and the combined androecium and gynoecium had significantly lower pigment levels than the other parts. Chl a content of both the sepal and the combined androecium and gynoecium was relatively higher than that in other flower parts. Chl b content was lower than that of Chl a in total flower. Higher contents of Chl b were obtained in the upper red part of petal and sepal, with the values of 0.027% and 0.029%, respectively. Meanwhile, the contents of these two chlorophylls were both much lower than that of carotenoids.

AST Isomers of Flower Parts
In Figure 3a, peak 4 at the retention time 8.0 min is the characteristic absorption band of all-trans AST. Peak 1 at 2.0-2.5 min of all plant samples was determined to be di-esterified AST, which is not present in the standard. The highest and lowest ratios of di-ester to free AST were obtained in the petal and the combined androecium and gynoecium, at 40:1 and 7:1, respectively. The chromatogram of AST after enzymatic hydrolysis (Figure 3b) showed that peaks 1-6 (retention time 5.7, 7.8, 8.5, 9.9, 10.4, and 10.9 min), represent astacene, di-cis, all-trans, 9-cis, 13-cis, and 15-cis AST, respectively, and these were observed in all the plant samples; however, the peak at the retention time 3.2 min (peak 7) was only observed in the spectrum for the combined androecium and gynoecium, which differed from other samples. With regard to AST optical isomers, only a peak for (3S, 3 S)-AST was found in the spectra of all samples (Figure 3c).
The levels of different AST isomers are shown in Figure 4. Among the geometrical isomers, the all-trans isomer was the dominant form in A. amurensis flowers (about 70%). Regarding all the cis isomers, the proportions of the 9-cis and 13-cis isomers were significantly higher than that of 15-cis in the total flower. Interestingly, the levels of di-cis and free AST in the combined androecium and gynoecium extract were both higher than the 9-cis, 13-cis, and 15-cis forms, which differed from the pattern in other parts of the flower.  Values are averages of four replicates ± standard error (SE). Eight randomly selected flowers were used to perform the determination.

Isolation of the All-Trans Isomer of (3S,3 S)-AST
In Figure 5a, the peak at the retention time of 34.23 min was confirmed to be the all-trans isomer of (3S,3 S)-AST. We then purified it from a crude extract of the total flower (contains 8.52% AST on the wet mass base of extract detected according to Section 4.3). The HPLC chromatograms and the UV-visible absorption spectrum of the purified all-trans isomer are shown in Figure 5b,c. As predicted, no peak was found in the 330-340 bands, consistent with the standard spectrum shown in Figure 1. Further calculations showed that the recovery of the all-trans isomer of (3S,3 S)-AST from the total carotenoids of the crude extract was 21.50 ± 0.81%. Figure 5. HPLC chromatograms of the crude extracts of the total flower (a) and the isolated all-trans isomer of (3S,3 S)-astaxanthin (b). c. the UV-visible spectrum of the purified all-trans isomer of 3S,3 S astaxanthin. The detection was performed at 478 nm.

Distribution of AST and Its Isomers in Adonis Flowers
The carotenoid content (mainly AST, Figure 2) in the total flower of A. amurensis can be up to 1.31% (dw), found mainly in the petals. The highest AST content, 3.31%, was recorded for the upper red part of the petal, followed by the purple spot (1.21%), the sepal (0.32%), and the androecium/gynoecium (0.14%) ( Table 1). The AST content of the red part of the petal resembles the content in H. pluvialis (commonly from 2.5% to 3.5%, dw) [45]. The AST content in the total petal is also encouraging, with a value of 2.58%.
Because free AST is relatively unstable, most AST in nature exists in the form of an ester, i.e., mono-ester or di-ester [34,46]. In H. pluvialis cells, approximately 70% of the AST exists as mono-esters, 25% as di-esters, and only 5% is unesterified (free) [31]. Previous studies have reported that esterified AST in cultured shrimps ranged from 47.69% to 95.56% of the total AST [23]. For A. amurensis, almost all of the AST exists in the di-ester form with the lowest proportion (85%) in the androecium/gynoecium and the highest level in the petal (>97.5%, Figure 3a).
AST has a broad absorption curve with a maximum wavelength of 480 nm, as shown in Figure 2. The peak at 330 nm, a characteristic of the cis-isomer [47,48], can be observed in the pigment's UV-visible spectrum. Geometrical isomers of AST were further measured by using an HPLC method. The proportions of di-cis AST in the sepal and the androecium/gynoecium (2.10% and 15.91%, Figure 4) were both higher than in other flower parts. This result is consistent with Figure 2, in which peaks at 330 nm of the sepal and the androecium/gynoecium were significantly higher than those of other parts. However, the total cis-AST ratios in other parts of the flower, i.e., petal 29.3.0%, red spot 24.0%, and purple spot 25.9%, are also at the same level as that of androecium and gynoecium (23.2%) and sepal (28.2%) (Figure 4). But the intensities of the cis peak in the AST extracted from the petal, the upper red part of the petal, and the lower purple part of the petal are much weaker (Figure 2). High cis peak intensity cannot be simply attributed to the high cis isomer content. A previous study has shown that chlorophyll has a wide absorption band at 330 nm [49]. The relatively high content of Chl a in sepal and androecium and gynoecium (Table 1) may lead to higher cis peak intensities in these two parts, which needs further study. Another interesting note was the inflections in the blue region in the gynoecium and androecium and sepal fractions in Figure 2. This can also be explained by the relatively high content of Chl a in these two parts ( Table 1). The absorption in the blue light region was caused by Chl a.
The peak at 3.2 min for the androecium/gynoecium in Figure 3b, which differed from the other spectra, was identified as β-carotene. In A. amurensis, AST exists as the diesterified form (Figure 3a), which is consistent with the situation in A. aestivalis as reported by Kamata and Simpson [44]. Double esterification of AST with fatty acids added to 3 and 3 hydroxyl groups of astaxanthin allow the AST to have more solubility and stability in the cellular environment. High di-ester AST content in the crude extract of A. amurensis flowers will also help to prolong the storage time of the pigment products. Esterification is also important to store the pigment in the hydrophobic environment. Interestingly, the proportions of free AST in the purple part of the petal and the androecium/gynoecium are significantly higher than that of the red area, which may mean less difficulty in getting highly pure free AST from these two parts without enzymatic hydrolysis.
The (3S,3 S) form has higher bioavailability and is more suitable for the health of humans [50]. Our result shows that (3S,3 S)-AST is the only optical isomer in A. amurensis (Figure 3c), which is consistent with the previous studies of A. annua [40], A. aestivalis [39], and A. amurensis [33], respectively.
In the androecium/ gynoecium, the ratio of all-trans AST was lowest; however, free, di-cis, and astacene had their highest values here as compared to their values in other flower parts (Figure 4). Little is known about the AST cis-trans isomerization pathway. To date, an epimerase for AST isomerization has not been identified. According to the results of studies conducted in humans [51] and rainbow trout [52], the in vivo isomerization of AST may lead to a different geometrical isomer composition. In A. amurensis, the mechanism(s) leading to the differences in AST composition between the Adonis flower components remains a question that needs to be investigated further. Given that animals cannot synthesize carotenoids de novo, AST in the shrimp must be obtained from food. When the white shrimp were fed using natural AST consisting of pure (3S,3 S)isomer, the ratio of (3S,3 S): (3S,3 R): (3R,3 R) in the L. vannamei body (with the value 6:2:1) differed from that of its feed supplement [2]. In A. aestivalis, the cDNA encoding the enzyme that catalyzes the addition of the carbonyls has been identified, and the possible route to AST from β-carotene is clear, in which the addition of a carbonyl to carbon C 4 involves a keto-enol tautomerization [37]. However, only a few studies have focused on the hydroxyl to C 3 relating to the optical isomers in prawn [53][54][55]. In animals, the transformation mechanism of AST from (3S,3 S)to (3S,3 R)and/or (3R,3 R)-isomer is far from being clarified. We deduced that epimerases exist in the AST metabolic pathway both in animals and plants [25], which changes the initial optical isomer composition. A further comparative study performed on the AST pathways in animals and higher plants could provide more information about this unresolved issue.

Potential Uses of Adonis Flower AST
UVA can penetrate the dermis layer of the skin and damage the elastic fibers and collagen fibers, causing the skin to lose its elasticity. At the same time, it darkens the epidermis layer, while inflicting damage to the skin [56]. AST is commonly used in the cosmetics industry for its well-known antioxidant, anti-inflammatory, and antitumor properties [5]. The androecium and gynoecium of the A. amurensis flower appear to be a potential source of AST for cosmetic purposes if a process for sorting flower components can be made economically viable. In addition, AST has been approved as a dietary supplement and is widely used in the development of novel foods [31,57]. Improving the yield of AST from natural sources, such as A. amurensis, to satisfy the demands for human applications, especially the anti-ultraviolet agent, deserves further study.
The function and application of specific carotenoids depend on their molecular structure [58]. From the thermodynamic point of view, the stability of all-trans AST is better than cis isomers but they may be isomerized from one form to another, and natural AST is easily oxidized and converted to astacene [28]. We have summarized the pigment content and isomer composition of most AST-containing organisms, in Table 2. Cis isomers together in A. amurensis flowers account for a higher proportion of AST (20.0%), in comparison with those of other natural sources. Compared with the all-trans isomer, cis AST, and especially 9-cis AST, has been shown to have higher antioxidant activity in vitro [28]. The highest concentration of cis isomers found so far is in the A. amurensis flower, which might make this a more suitable source for extraction of antioxidant material than other sources, e.g., Haematococcus, shrimp, and P. rhodozyma. There have been attempts to utilize AST from another Adonis species, A. aestivalis, as the pigmentation for fish [36,59], although its safety for humans has not been established.

Adonis Flower as a Promising Source of AST
The high production cost of natural AST is still the primary factor impacting the availability of this pigment. A previous study showed that the production cost of natural AST can be as low as €632/kg, which is lower than the cost for synthetic AST at €880/kg [45]. This encouraging price is not too surprising considering the low labor, land, and utility costs in China a decade ago. A recent assessment has shown that the production cost of natural AST is unable to compete with the synthetic alternative (as summarized in Table 3). The costs of natural AST in Livadeia and Amsterdam are €1536/kg and €6403/kg, respectively, both much higher than that of synthetic AST [32]. According to our results and published data, we estimate that the cost of producing natural AST from total Adonis flowers can be as low as €388-393/kg (Table 3), suggesting it is a very promising source for producing natural AST. As far as we know, the commercially available all-trans AST standard is a mixture of (3S,3 S), (3S,3 R) and (3R,3 R)-AST, and a pure all-trans isomer of (3S,3 S) AST is still lacking in the market. Our results suggest that Adonis flowers would be particularly suitable as raw material for obtaining pure all-trans (3S,3 S) AST. Note: N.D.: not detected; dw: dry weight; ww: wet weight. The data in shrimp are for the muscle, cephalothorax, and shell while the data in crab are for the ovaries and carapace. Note: a. the production was estimated according to the data provided by Huang [65]. b. the price was calculated according to the data provided by Li and Xiu [66], which covers the cost of mechanical operation, fertilizer, labor, watering, and land lease. The rent of film greenhouses ( Figure 6) for planting flowers is €255/year.

Plant Materials and Reagents
A. amurensis was cultivated by sowing in Shandong province, May 2017. Fresh flowers were handpicked from the plants and then sorted into different flower parts, i.e., total flower, the petal including the red spot and the purple spot, the sepal, and the remainder (mainly the combined androecium and gynoecium) ( Figure 7). All materials were weighted, then stored in liquid nitrogen until further analysis. An all-trans AST standard for AST quantification was obtained from Sigma Chemical Co., while cholesterol esterase for hydrolysis came from Wako Pure Chemical Industries, Ltd. HPLC-grade methanol, n-hexane, acetonitrile, methyl tert-butyl ether, and 2-propanol were obtained from Adamas Reagent Ltd. Acetone used as extracting reagent was prefiltered through 0.22 µm GF/C filters (Whatman International Ltd., Maidstone, UK).

Preparation of Pigment Extracts from Adonis
Chl a and carotenoids from various flower parts were extracted by acetone using the modified method of Arnon [67] and Johnston et al. [68]. The tissue (about 0.5 g fresh weight) was placed in 10 mL centrifuge tubes in ice, then 6.0 mL of pre-cooled 100% acetone was added. The flower parts were blended, soaked for 30 min in acetone, and then homogenized in an ice-water bath. The pigment extracts were collected by centrifugation at 2500× g for 8 min at 4 • C. The cell debris was repeatedly extracted with the same solvent until it was colorless. All the pigment extracts for individual samples were merged into a 50 mL volumetric flask and then filtered using 0.22 µm GF/C filters (Whatman International Ltd., Maidstone, UK).

Quantification of Pigments
The acetone extracts were dehydrated with sodium sulfate. Then, the absorption spectra of pigment extracts were measured using a spectrophotometer (U-2900, HITACHI Co., Ltd., Tokyo, Japan). The optical path length was 10 mm. Chl a, Chl b, and total carotenoid concentrations of the acetone extract were calculated using the following equations [69]: where M o is the wet weight of the fresh tissue; M e is the dry weight of the biomass after drying for 24 h at 80 • C.

Identification of Carotenoids by HPLC
The hydrolysis of the carotenoid esters and the analysis of carotenoids followed the methods described in previous reports [69,70]. An Agilent 1200 HPLC system (Agilent Technologies Inc., Santa Clara, CA, USA) with a Luna Silica column (3 µm, 150 mm × 4.6 nm, Phenomenex, Torrance, CA, USA) was used for qualitative and quantitative analysis of carotenoids. Hydrolysis of astaxanthin was performed with cholesterol esterase for 60 min to remove esters according to our previous study [69] before identification and quantification by conventional HPLC. Identification of carotenoids without standards, such as di-cis-AST, astacene, and 15-cis-AST, was performed by comparing their spectra or retention times with those of published data [27,69,[71][72][73]. Optical isomers of AST were further separated according to the method described by [29]. The levels of different optical isomers, i.e., (3S,3 S), (3R,3 R), and (3S,3 R)-AST, were calculated according to the peak areas and by comparing the retention time of sample peak to synthetic AST standard (CaroteNature, Lupsingen, Switzerland), respectively.

Preparation of the All-Trans Isomer of (3S,3 S)-AST
The all-trans isomer of (3S,3 S)-AST was prepared from the crude extracts of the total flower on an Agilent 1200 HPLC system (Agilent Technologies Inc., Santa Clara, CA, USA) with an InertSustain C18 column (5 µm, 250 mm × 10 mm, GL Science, Tokyo, Japan) [27]. The oily flowers extract is provided by the manufacturer settled in Weifang, China. The binary mobile phase consisted of A: 0.1% trifluoroacetic acid in water (v/v) and B: 95% methanol mixed with 5% acetonitrile (v/v). The solvent gradient was as follows: 0-5.8 min, 60-80% B; 5.8-32.6 min, 80-100% B; 32.6-35 min, 100% B; 35-47 min, 100-60% B; 47-50 min, 60% B. Peaks were detected at 470 nm. The flow rate was set at 3.0 mL/min. Under these conditions, the fraction containing the all-trans isomer of (3S,3 S)-AST eluted at a retention time of 34 to 35 min and was collected and dried under a nitrogen stream.

Statistics
SPSS 16.0 software was used for statistical analyses. One-way analysis of variance (ANOVA) and Tukey test (p < 0.05) was used to analyze differences for the multiple comparisons.

Conclusions
One optical isomer (3S, 3 S) of AST, with five geometrical isomers (all-trans, 9-cis, 13-cis, 15-cis, and di-cis) were observed in all parts of the flower, i.e., the petal including the red spot and the purple spot, the sepal, and the remainder (mainly the combined androecium and gynoecium). The highest carotenoid content was obtained from the red part of the petal (3.31%, dw). All-trans AST was the predominant geometrical isomer accounting for 72.5% of the total content of geometric isomers in total flower, followed by the 13-cis (9.8%), 9-cis (7.8%), di-cis (1.8%), and 15-cis (0.6%) isomers. The cis-AST extracted from the combined androecium and gynoecium can be used to prepare the anti-ultraviolet agents. The production cost of AST from Adonis flowers can be as low as €388-393/kg, which is far lower than that from Haematococcus culturing. Together with other factors such as the low technology requirement for plant culturing and harvesting, suggest Adonis has great potential as a resource for natural esterified (3S,3 S)-AST production.  Data Availability Statement: All relevant data can be found within the manuscript and its supporting materials.