Synthesis of prenylated flavonols and their potents as estrogen receptor modulator.

Prenylated flavonols are known as phytoestrogen and have good bioactivties. However, their abundances in nature are pretty low. It is required to find an efficient synthesis technique. Icariin is a prenylated flavonol glycoside with low cost. It can be used to synthesize different prenylated flavonols. A combination of cellulase and trifluoacetic acid hydrolysis could effectively remove rhamnose and glucose from icariin. Icaritin, anhydroicaritin and wushanicaritin were the leading prenylated flavonol products. Their affinities to estrogen receptors α and β were predicted by docking study. The weak affinity of wushanicaritin indicated that prenyl hydroxylation impaired its affinity to estrogen receptor β. The prenyl cyclization led to a loss of affinity to both receptors. The interactions between icaritin and ligand binding cavity of estrogen receptor β were simulated. π-π stacking and hydrophobic forces were predicted to be the dominant interactions positioning icaritin, which induced the helix (H12) forming an activated conformation.


Synthesis of icaritin, anhydroicaritin and wushanicaritin. Enzyme hydrolysis to remove gluco-
side. Icariin (100 mg) were added into 25 ml of water. Five milligrams of enzymes were added to start the hydrolysis reaction at 37 °C for 2.5 h. The reaction products were extracted by acetyl acetate. The extract was dried by a vacuum rotary evaporator and redissolved in methanol. The products were qualified and quantified by ultra-performance liquid chromatography.
The effects of temperature, time and enzyme type were investigated. Three enzymes, including β-dextranase, β-glucosidase and cellulase, were tested to compare their hydrolysis efficiencies. When analysing the effect of temperature, 20, 37, 60 and 80 °C were used. The time were set as 1.5, 2.5, 3.5 and 4.5 h, respectively.
Acid hydrolysis to remove rhamnoside. Icariin (100 mg) were suspended in 25 ml of water. Trifluoacetic acid (TFA) was added to a final concentation of 2 M. The hydrolysis was conducted at 60 °C for 1 h. The reaction products were extracted by acetyl acetate. The extract was dried by vacuum rotary evaporator and redissolved in methanol. The products were qualified and quantified by ultra-performance liquid chromatography.
Combination of cellulase and TFA hydrolysis. Hydrolysis by cellulase first and then TFA: Icariin (100 mg) were added into 25 ml of water. Five milligrams of cellulase were added to start the hydrolysis reaction at 37 °C for 2.5 h. Then TFA was added to a final concentation of 2 M. The hydrolysis was conducted at 60 °C for 1 h. The reaction products were extracted by acetyl acetate. The extract was dried by vacuum rotary evaporator and redissolved in methanol. The products were qualified and quantified by ultra-performance liquid chromatography.
Hydrolysis by TFA first and then cellulose: Icariin (100 mg) were added into 25 ml of water. TFA was added to a final concentation of 2 M. The hydrolysis was conducted at 60 °C for 1 h. The hydrolysates were dried by vacuum rotary evaporator. Deionized water was added and dried again. This step was repeated for four times to completely remove acid. The dry products were redissolved in 25 ml of water. Five milligrams of cellulase were added to start the hydrolysis at 37 °C for 2.5 h. The reaction products were extracted by acetyl acetate. The extract was dried by vacuum rotary evaporator and redissolved in methanol. The products were qualified and quantified by ultra-performance liquid chromatography.

UPLC and UPLC-MS/MS analyses of reaction products.
The analyses of reaction products and determination of conversion rate were performed on an Agilent 1260 Infinity UPLC system (Agilent Technologies, Germany) 17 . The analyses were performed on an Agilent ZORBAX SB-C18 column (3.0 × 100 mm, 1.8 μm). The flow rate was 0.3 ml/min, the column temperature was set at 40 °C, and the injection volume was 10 μl. The chromatogram was monitored at 280 nm. The mobile phases comprised solvents A (ultrapure water) and B (methanol). The elution program was as follows: 0-30 min, 5-100% B; 30-40 min, 100% B 18 .
UPLC-MS/MS was analysed on a maXis LC-ESI-QTOF-MS system (Bruker, Germany) equipped with an Agilent ZORBAX SB-C18 column (3.0 × 100 mm, 1.8 μm). The elution program was the same as UPLC analysis. Ionization of the analytes was achieved by using electron spray ionization interface in negative mode. The collision voltage was 10 eV. Mass scan was set in the range of m/z 50-1000. The daughter ions were monitored at a collision voltage range of 28-42 eV 19 .
Binding affinity prediction by molecular docking. The X-ray structure of estrogen receptor α and β in complex with agonist (2 R,3 S,4 R)-(4-hydroxyphenyl)-6-hydroxy-cyclopentyl[c]3,4-dihydro-2H-1-benzopyran (PDB id: 2i0j and 2i0g) were used as the templates 20 . Autodocktools (version 1.5.6rc3) was used to dock estrogen receptor α/β and the tested chemicals 21 . The protein was checked for any misssing atoms, removed water and added hydrogen. The ligand was drawed and saved as a mol2 file. It was opened in autodocktools and saved as PDBQT file. A grid was generated for protein with the ligand centered. Genetics algorithm was used for docking 22 . The binding affinity was recorded to evaluate the potential to be estrogen receptor agonist.

Results and Discussion
Synthesis of anhydroicaritin, icaritin and wushanicaritin. Enzyme hydrolysis of icariin. Three enzymes, including cellulase, β-glucosidase and β-dextranase, were used to hydrolyse icariin, respectively. There was only one product generated after enzyme hydrolysis (Fig. 1). It was purified and subjected to NMR analysis. The 1 H and 13 C chemical shifts of this chemical are listed in Table 1S and Fig. 2S. When comparing with icariin (1, Fig. 1S), the missing of glucosyl signals indicated the loss of this moiety. An upfield shift of H-6 was observed at 6.26 ppm. The chemical shifts of C-7 and C-8 were changed accordingly. The rhamnosyl with anomeric signals (5.40/103.7 ppm) was detected 23 . It confirmed that these enzymes could not degrade this unit. The above information confirmed the presence of baohuoside I (2).
UPLC-QTOF-MS is used to determine the precise molecular weight and fragmentation pattern of baohuoside I in negative mode (Table 2S) Though β-glucosidase has been reported to be effective in hydrolysis of icariin 25 , its effectiveness was much lower than cellulase in this work (Fig. 10S). The icariside I yield reached 94.8% by cellulase hydrolysis, while the yield was lower than 50% when β-glucosidase or β-dextranase was used. Temperature influenced the glucosyl cleavage to a certain extent. The highest yield of icariside I was obtained when 37 or 60 °C were applied. It indicated a broad temperature tolerance of cellulase. Out of this range made the hydrolysis efficiency decreased sharply. Hydrolysis time showed a weak effect on icariside I yield. 2.5 h was the optimal value. Further extension of reaction time led to a slight decrease of icariside I, which might be due to the degradation.
Acid hydrolysis of icariin. Icariin was hydrolysed by TFA and four products were detected (Fig. 2). They were purified by C-18 column and identified by UPLC-MS/MS and NMR. Please see Table 2S     The four chemicals generated by TFA hydrolysis of icariin were compared. 2 and 5 contained rhamnosyl, but no glucosyl, while 3 and 4 contained glucosyl but no rhamnosyl. Rhamnosyl was easier to be cleaved than glucosyl by acid. In acidic conditions, the prenyl was unstable and readily to be hydrated or cyclized. No products without both rhamnosyl and glucosyl were detected. It indicated that TFA hydrolysis could not directly generate icaritin. Therefore, combination of cellulase and TFA hydrolysis was carried out in the following work.
Combination of cellulase and acid hydrolysis. When icariin was hydrolysed by cellulase firstly and then by TFA, five main products were detected ( Figure 11S). The reaction route and products are shown in Fig. 3. Chemical 6 was the dominant product which had a parent ion [M-H] − at m/z 367.1142. No rhamnosyl and glucosyl signals were observed in the NMR signals (Fig. 6S). Their losses led to the upfield movements of H-6, C-2, C-4 and downfield shift of C-3. The fragment ion at m/z 352.0974 was due to the loss of methyl. m/z 311.0601 was produced by the cleavage of prenyl between C-1″ and C-2″. These information confirmed that chemical 6 was icaritin 10 (Table 1S and Figure 8S). It was called wushanicaritin 28 . Chemical 9 was identified to be 3-O-rhamnosyl wushanicaritin by the parent ion [M-H] − at m/z 531.1246 and NMR signals ( Figure 9S).
When icariin was hydrolysed by cellulase firstly and then TFA, baohuoside I (2) was not detected in the products. It was cyclized into anhydroicaritin 3-O-rhamnoside (5). The yields of icaritin, anhydroicaritin and wushanicaritin were calcalated to be 61.1%, 12.2% and 3.7%, respectively. Therefore, hydrolysis by cellulase firstly and then TFA would be a good choice for icaritin preparation.
Docking study. Molecular docking simulation was performed for icaritin, anhydroicaritin and wushanicaritin, in order to predict the estrogen receptor α/β-agonist interactions. The most favorable docking conformation was retrieved from calculation. As shown in Table 3S, in the docking model, wushanicaritin showed high binding affinities to estrogen receptors α (−8.48 kcal/mol) and β (−7.97 kcal/mol). Icaritin only exhibited a good binding affinity to estrogen receptor β (−9.46 kcal/mol). Anhydroicaritin showed poor binding affinities to both estrogen receptors. Icaritin was predicted to have a good selectivity due to its specific binding affinity to estrogen receptor β. Figure 5 shows the docking conformation of estrogen receptor α and wushanicaritin. The ligand binding domain of estrogen receptor α comprises of three anti-parrallel α-helices layers, including a central layer and two outside layers 29 . The residues Met343, Leu346, Leu349, Ala350, Trp383, Leu384, Leu391, Leu387, Met421, Ile424, Leu428 and Leu525 are predicted to be involved in the positioning of wushanicaritin by hydrophobic forces. Phe404 was predicted to interact with wushanicaritin by π-π stacking. His524 was predicted to generate a hydrogen bond and π-π stacking interaction to stabilize wushanicaritin. The conformation of H12 in the complex determines the binding possibility of coactivator to activate function domain (AF-2). In the estrogen receptor α/ wushanicaritin complex, H12 was predicted to seal the ligand-binding cavity and to generate a competent AF-2 that could interact with coactivator. This could be helpful to explain how wushanicaritin acted as an agonist. Figure 6 shows thne predicted conformation of estrogen receptor β in complex with icaritin. Estrogen receptor β is present similiarly to estrogen receptor α as a sandwich made by a central layer and two outside layers. H12 forms a lid on the binding cavity. π-π stacking interaction was observed by residues His475 and Phe356. No hydrogen bonds were detected to position icaritin in the predicted conformation. Hydrophobic forces from Leu298, Ala302, Leu476, Met479, Ile373, Ile376, Leu380, Phe377, Met340, Met336 and Leu339 were predicted to be involved in stabilizing icaritin.

Conclusions
A combination of cellulase and TFA hydrolysis was confirmed to be an efficient technique to synthesize icaritin, wushanicaritin and anhydroicaritin. This technique extended the application of cellulase in icariside preparation and made production of three aglycones in one pot possible 30 . It was easy to carry out and to produce at a large scale when comparing synthesis from kaempferol 31 . Icaritin, wushanicaritin and anhydroicaritin are good drug candidates due to their good performance against cancer and osteoporosis. Developing a cost effective technqiue is of great interest for medicine industry. As the prenylation pattern produced a significant effect on the activity performance, it is required to conduct more bioactivity evaluations, including the estrogen receptor modulatory mechanisms in vitro and in vivo, to understand the potents of these chemicals in medicines. Some bioactivites, like anti-breast cancer, anti-osteoporosis or memory improving capabilities, should be evaluated, as estrogen receptor modulatory behaviours are involved in the mechanism of action.