Total Synthesis of the Natural Products Ulmoside A and (2R,3R)-Taxifolin-6-C-β-d-glucopyranoside

^# Both authors contributed equally.

Abstract

An efficient first total synthesis of highly polar ulmoside A and (2R,3R)-taxifolin-6-C-β-d-glucopyranoside, useful for the prevention of metabolic disorders, has been described. Key elements of the synthesis include a Sc(OTf)₃-catalyzed regio- and stereoselective C-glycosidation on taxifolin in 35% yield with d-glucose and chiral semipreparative reverse-phase high-performance liquid chromatography (HPLC) for the separation of both taxifolins and the diastereomeric mixture of taxifolin-6-C-β-d-glucopyranosides. Correlation of the analytical data of synthetic ulmoside A and its diastereomer with a natural ulmoside A sample confirmed the assigned absolute stereochemistry of the natural products.

Key words

flavonoids - C-glycosidation - chiral HPLC - metal triflates - aglycone

Glycosyl flavonoids are ubiquitously distributed in nature and commonly found as secondary metabolites of plant or bacteria. They exhibit a wide range of very interesting biological activities. The glycosyl flavonoids are mostly present in nature as O-glycosides and rarely as C-glycosides.[1] In general, C-glycosides show excellent and different bioactivity when compared to the corresponding O-glycosides and aglycons. The C–C bond in these glycosides that links the sugar to aglycon appears to be of great importance, because the C-glycosyl bond is stable to enzymatic hydrolysis under physiological conditions, unlike the corresponding O-glycosides.[1c] Therefore, some aryl C-glycosides are on the market as SGL₂ inhibitors for treating diabetes.[1d]

Mourya and co-workers, in 2009, discovered and identified three new C-glycosylated flavonoids that were considered as stable glycosides in the biological system, namely (2S,3S)-taxifolin-6-C-β-d-glucopyranoside (or) K058 (or) ulmoside A, (2S,3S)-aromadendrin-6-C-β-d-glucopyranoside, and quercetin-6-C-β-d-glucopyranoside from Ulmus wallichiana plant[2`] [b] [c] (Figure [1]). The bark of this plant is known in practice in some parts of India to treat bone fractures.[2b] Mourya and co-workers evaluated their activity in stimulating osteoblast differentiation, which is a bone anabolic function needed for osteoporosis therapy. They also developed a composition for osteoporosis therapy with these compounds. Among them, ulmoside A (1) shows excellent bone deflecting activity and efficacy in treating steroid-induced metabolic disorders.[2d] Interestingly, it has shown excellent adiponectin agonist activity, useful for the cure and prevention of diabetes, and also antioxidative and antiapoptotic activity.[2e] Stark et al. also isolated ulmoside A (1) and (2R,3R)-taxifolin-6-C-β-d-glucopyranoside (2) and other flavonoid C-glycosides from the bark of medicinal plant Garcinia buchananii.[2c] [3a] The H₂O₂ scavenging and oxygen radical absorbance capacity (ORAC) assays demonstrated that compounds 1 and 2 show extraordinary antioxidative power. Earlier, (2R,3R)-taxifolin-6-C-β-d-glucopyranoside (2) was also isolated from Garcinia epuntata (Figure [1]).[3b]

Figure 1 Structures of some natural flavonoid C-glycosides

The above compounds can only be obtained from natural sources in very small quantities. Moreover, evaluation of their biological activity and their development into a drug requires more quantities of the compounds. Therefore, a simple, short, and scalable strategy is indeed essential. To meet the above requirements we have undertaken the total synthesis of those compounds. Interestingly, these two compounds show similar chemical shifts in their NMR spectra, and both have a positive optical rotation. They are only distinguished by their CD spectra. So far, there are no reports on the synthesis of these compounds in the literature. Hence, their total synthesis also helps confirming the assigned stereochemical structures. The important issue to be addressed for designing a synthesis for this class of ­compounds is the C–C bond formation between the aglycon and carbohydrate moiety to give corresponding aryl C-glycosides.

Formation of the C–C bond between carbohydrate units and electron-rich aromatic moieties was first achieved by Friedel–Crafts reaction.[4] Later, Suzuki and co-workers[5] and Kometani et al.[6] developed a methodology for the C-glycosylation of phenols by a Lewis acid-catalyzed rearrangement of O-glycoside to a C-glycosyl derivative, known as a Fries-type reaction. In all these general methods, protected sugar and aromatic moieties have been used for glycosidation that, finally, have to be deprotected to obtain the target compounds. Recently, Sato and others reported the C-glycosidation of unprotected flavonoids with unprotected sugars using metal triflates.[7] [8] [9] Although the yields obtained with use of this method are lower and although the method was applied on simpler systems, we wanted to apply this method for the synthesis of chiral polyphenolic compounds 1 and 2.

On the basis of the assigned stereochemistry, the retrosynthetic route for ulmoside A (1) and (2R,3R)-taxifolin-6-C-β-d-glucopyranoside (2) has been designed (Scheme [1]). We envisioned a one-step synthesis of the target molecules 1 and 2 from corresponding taxifolins 3 and 4 by protecting-group-free C-glycosidation with d-glucose in the presence of metal triflates. The (+)-taxifolin (3) required for the synthesis of (2R,3R)-taxifolin-6-C-β-d-glucopyranoside (2) is commercially available but it is expensive; rather it can be prepared from (+)-catechin hydrate (5) by using a reported procedure.[10] In the case of (–)-taxifolin (4), the aglycon part of compound 1, to the best of our knowledge, neither its isolation nor its synthesis is reported in the literature. Only the racemic synthesis of taxifolin 6 is reported in the literature.[11] Therefore, we decided to synthesize both the taxifolins 3 and 4 from the chalcone compound 7 using epoxidation followed by a cyclization approach. Compound 7 could be obtained from commercially available phloroglucinol (8) and caffeic acid (9) (Scheme [1]).

Scheme 1 Retrosynthetic analysis

In order to prepare (+)-taxifolin (3), the hydroxy groups of (+)-catechin hydrate (5) were benzylated by treatment with BnBr and K₂CO₃ to give 5,7,3′,4′-tetra-O-benzyl-(+)-catechin, which, on further treatment with BnBr and NaH, afforded 3,5,7,3′,4′-pent-O-benzyl-(+)-catechin 10 in 73% yield (over two steps) (Scheme [2]).[10] In the previous report,[10b] oxidation of 3,5,7,3′,4′-pent-O-benzyl-(+)-catechin 10 by using 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) in a mixed solvent of CH₂Cl₂/H₂O afforded desired keto compound 11 in 15% yield. To improve the yields in the oxidation, compound 10 was subjected to excess DDQ in a mixed solvent of CH₂Cl₂/1,4-dioxane/H₂O (8:4:1), which afforded the desired keto compound 11 in 80% yield.[12] ­Further, hydrogenolysis of keto compound 11 in the presence of H₂ and 10% Pd/C in MeOH/THF afforded (+)-taxifolin (3) in 85% yield.[10]

Scheme 2 Synthesis of (+)-taxifolin (3)

Having synthesized the required (+)-taxifolin (3), we next turned our attention to a regio- and stereoselective protecting-group-free C-glycosidation between (+)-taxifolin (3) and d-glucose. Thus, experiments were carried out with different catalysts under different solvent ratios as shown in Table [1]. The highest yield (35%) of compound 2 was obtained with scandium triflate (0.4 equiv) with a 2:1 ratio of acetonitrile and water heated to reflux for 48 hours (Table [1], entry 4).[13] All analytical data including ¹H and ¹³C NMR spectra of the synthetic (2R,3R)-taxifolin-6-C-β-d-glucopyranoside (2) were in agreement with reported values.[3]

Then, we decided to synthesize (±)-taxifolin (6) from chalcone 7 as shown in Scheme [3]. For this purpose, a new and biomimetic approach was developed to obtain 7. Treatment of phloroglucinol (8) with methoxymethyl chloride (MOMCl) and NaH in dry DMF gave compound 12 in 90% yield. Esterification of caffeic acid (9) with MeOH in the presence of H₂SO₄ as a catalyst and protection of the phenolic groups with MOMCl provided the ester compound. Hydrolysis of the resultant ester with NaOH in MeOH/THF afforded the acid compound 13 in 90% yield. Further, acylation of compound 12 with acid 13 in the presence of trifluoroacetic anhydride (TFAA) in CH₂Cl₂ afforded chalcone 7 in 85% yield. Treatment of compound 7 with alkaline hydrogen peroxide led to the formation of racemic epoxide 14, which was subjected to concd HCl in MeOH/THF to afford (±)-taxifolin (6) in 85% yield (Scheme [3]).[11] Our procedure gave better overall yields (72%) when compared to the earlier reported procedure[11] (40%) from MOM-protected intermediates, and also it involves cheaply available starting materials. Our approach has an additional advantage: The caffeic acid can be converted into a chiral epoxy or dihydroxy derivative, which, upon acylation on 12 followed by cyclization, can lead to chiral taxifolins.

Table 1 Optimization of C-Glycosidation Reaction Conditions

Entry	Catalyst	Equiv.	MeCN/H2O	Time (h)	Yield (%)b
1	Sc(OTf)3	0.2	1:1	24	10
2	Sc(OTf)3	0.2	1:1	48	15
3	Sc(OTf)3	0.2	2:1	48	20
4	Sc(OTf)3	0.4	2:1	48	35
5	Sm(OTf)3	0.4	2:1	48	8
6	Bi(OTf)3	0.4	2:1	48	5
7	Yb(OTf)3	0.4	2:1	48	no reactionc

^a Reaction conditions: (+)-taxifolin (3) (1 equiv), d-glucose (2.5 equiv), M³⁺(OTf)₃, solvent, reflux.

^b Isolated yield.

^c Based on TLC analysis.

Having (±)-taxifolin (6) in hand, our next step was to resolve it to obtain enantiomers 3 and 4. Chiral semipreparative reverse-phase high-performance liquid chromatography (HPLC) allowed us to successfully separate 3 and 4 (Scheme [4]). The separated enantiomers have the same physical data with opposite optical rotation values. The spectral and analytical data of 3 and 4 were in full agreement with the literature values, and also the values of the product obtained from (+)-catechin hydrate (5) in Scheme [2].[10c]

Scheme 3 Synthesis of (±)-taxifolin (6)

To accomplish the second target molecule ulmoside A (1), we initiated our synthesis from the (–)-taxifolin (4) as shown in Scheme [5]. Thus, C-glycosidation of (–)-taxifolin (4) with d-glucose in the presence of Sc(OTf)₃ in MeCN/H₂O furnished ulmoside A (1) in 35% yield.[11] Conducting the reaction for 48 hours helped us to obtain 35% yield of 1, whereas Sato’s conditions gave 25% yield of the product in glycosidation. During the column purification of the product, the unreacted starting material was also recovered. Based on the recovery the yield of the product is around 93%. All analytical data including ¹H and ¹³C NMR spectra of the synthetic ulmoside A (1) were in full agreement with the data of a natural sample of ulmoside A (1).[2a] [13] [14]

Scheme 4 Chiral HPLC separation of racemic taxifolins 6

Scheme 5 Total synthesis of ulmoside A (1)

Next, we proceeded with similar conditions as mentioned in Scheme [5], to obtain the mixture of ulmoside A (1) and (2R,3R)-taxifolin-6-C-β-d-glucopyranoside (2) from (±)-taxifolin (6) and d-glucose (Scheme [6]). Chiral semipreparative reverse-phase HPLC allowed us to separate successfully both ulmoside A (1) and its diastereomer 2 without any loss of compound. The HPLC chromatogram of synthetic ulmoside A (1) revealed a peak at 23.6 min, which was very well correlated with that of natural ulmoside A (see Supporting Information). The complete analysis of HPLC data clearly confirmed that the absolute configuration of the aglycon part of ulmoside A (1) is 2S,3S and that of its diastereomer is 2R,3R.

Scheme 6 Direct C-glycosidation on (±)-taxifolin (6)

A simple and scalable first synthesis of highly polar ulmoside A (1) and (2R,3R)-taxifolin-6-C-β-d-glucopyranoside (2), useful for the prevention of metabolic disorders, has been described. Scandium triflate proved to be a suitable catalyst for the regio- and stereoselective C-glycosylation on taxifolins (3) and (4) in 35% yield with unprotected d-glucose. Despite this, an achiral semipreparative HPLC method was developed for the separation of both (±)-taxifolin (6) and the diastereomeric mixture of taxifolin-6-C-β-d-glucopyranosides 1 and 2. The absolute configuration of aglycon of ulmoside A (1) and its diastereomer (2) was confirmed as 2S,3S and 2R,3R. This strategy is also useful to build a library of compounds with different sugars and polyphenols for better therapeutic activity. Most of these flavonoid C-saccharides are present in dietary supplements and also in the extracts used in Indian traditional medicines. These compounds are less toxic and have a high probability to become drugs. Therefore, we are planning to apply this approach to the synthesis of other flavonoid C-saccharides. Further, chiral epoxidation of chalcone 7 or converting caffeic acid into chiral epoxy or hydroxy derivatives for acylation can lead to a chiral synthesis of taxifolins.

Acknowledgment

We thank Dr. T. K. Chakraborthy, former Director of CSIR-CDRI, for providing the sample of natural ulmoside A. The authors also thank the Director of CSIR-IICT for the constant support and encouragement.

• References and Notes

• 1a Bililign T, Griffith BR, Thorson JS. Nat. Prod. Rep. 2005; 22: 742
• 1b Hultin PG. Curr. Top. Med. Chem. 2005; 5: 1299
• 1c Rauter AP, Lopes RG, Martins A. Nat. Prod. Commun. 2007; 11: 1175
• 1d Wang Q, Shuang A, Deng Z, Zhu W, Huang Z, Gang H, Chen G. Nat. Catal. 2019; 2: 793
• 1e Xiao J, Capanoglu E, Jassbi AR, Miro A. Crit. Rev. Food Sci. Nutr. 2016; 56: S29
• 2a Rawat P, Kumar M, Sharan K, Chattopadhyay N, Maurya R. Bioorg. Med. Chem. Lett. 2009; 19: 4684
• 2b Maurya R, Rawat P, Sharan K, Siddiqui JA, Swarnkar G, Mishra G, Manickavasagam L, Jain GK, Arya KR, Chattopadhyay N. PCT Patent PCT/IN2009/000146, 2009; (WO2009110003A1, WO200911000, 2009
• 2c Stark TD, Germann D, Balemba OB, Wakamatsu J, Hofmann T. J. Agric. Food Chem. 2013; 61: 12572
• 2d Sabyasachi S, Chattopadhyay N, Maurya R, Gayen JR, Bhadauria S, Trivedi AK, Singh AK, Mishra JY, Kumari R, Sharan K, Khan MP, Khan K, Singh N, Dwivedi SK. D, Yadav M, Dixit P, Mishra DP, Sharma S, Kamal RA. PCT Patent PCT/IN2014/000464, 2014; (US2016/019324, 2014
• 2e Gupta P, Singh A, Tiwari S, Mishra A, Maurya R, Singh S. Neurotoxicology 2019; 73: 100
• 3a Stark T, Matsutomo T, Losch S, Boakye P, Balemba O, Pasilis S, Hofmann T. J. Agric. Food Chem. 2012; 60: 2053
• 3b Mbafor JT, Fomum ZT, Promsattha B, Sanson DR, Tempesta MS. J. Nat. Prod. 1989; 52: 417
• 4a Jaramilo C, Knapp S. Synthesis 1994; 1
• 4b Levy DE, Tang C. The Chemistry of C-Glycosides. Pergamon; Oxford: 1995
• 4c Postema MH. D. C-Glycoside Synthesis. CRC Press; London: 1995
• 4d Du Y, Linhardt RJ. Tetrahedron 1998; 54: 9913
• 4e Toshima K, Matsuo G, Nakata M, Matsumura S. J. Synth. Org. Chem. Jpn. 1998; 56: 841
• 4f Schmidt RR, Hoffmann M. Tetrahedron Lett. 1982; 23: 409
• 5 Matsumoto T, Katsuki M, Suzuki K. Tetrahedron Lett. 1988; 29: 6935
• 6 Kometani T, Kondo H, Fujimori Y. Synthesis 1988; 1005
• 7a Sato S, Akiya T, Nishizawa H, Suzuki T. Carbohydr. Res. 2006; 341: 964
• 7b Van Der Westhuizen JH, Ferreira D, Joubert E, Bonnet SL. WO2011064726A1, 2011
• 7c Sato S, Koide T. Carbohydr. Res. 2010; 345: 1825
• 8 Santos RG, Xavier NM, Bordado JC, Rauter AP. Eur. J. Org. Chem. 2013; 1441
• 9a Kitamura K, Ando Y, Matsumoto T, Suzuki K. Chem. Rev. 2018; 118: 1495
• 9b Yang Y. Yu B. 2017; 117: 12281
• 10a Park KD, Cho SJ. Eur. J. Med. Chem. 2010; 45: 1028
• 10b Es-Safi N.-E, Ducrot P.-H. Lett. Org. Chem. 2006; 3: 231
• 10c Kikuchi T, Nishimura M, Hoshino A, Mortia Y, Saito N, Honda T. Heterocycles 2003; 60: 1469
• 11 Yang LX, Huang KX, Li HB, Gong JX, Wang F, Feng YB, Tao QF, Wu YH, Li XK, Wu XM, Zeng S, Spencer S, Zhao Y, Qu J. J. Med. Chem. 2009; 52: 7732
• 12 Wu Z, Wei G, Lian G, Yu B. J. Org. Chem. 2010; 75: 5725
• 13 Rao BV, Lingamurthy M, Raju G, Sarma VU. M. US20160362401A1, 2016; (US9611255B2)
• 14 Ulmoside A (1)A solution of (–)-taxifolin (4) (0.3 g, 0.98 mmol) and d-glucose (0.44 g, 2.46 mmol) was stirred in a mixture of solvents MeCN/H₂O (2:1, 10 mL) heated to reflux in an oil bath for 48 h in the presence of Sc(OTf)₃ (0.19 g, 0.39 mmol). Then, the reaction mixture was subjected to silica gel column chromatography (acetone/EtOAc/H₂O/AcOH 15:30:2:0.1) to afford ulmoside A (1) as a white amorphous powder yield: 0.16 g (35%); [α]_D ²⁵ = +1.56 (c = 0.1, MeOH){lit^2a[α]_D ²⁵ = +1.33 (c = 0.098, MeOH)}. IR (neat): 3325, 2945, 2833, 1646, 1450, 1413, 1017 cm^–1. ¹H NMR (500 MHz, DMSO-d ₆): δ = 12.47 (s, 1 H, HO-C(5)), 9.03 (s, 1 H, HO-C(4')), 8.98 (s, 1 H, HO-C(3')), 6.84 (s, 1 H, H-2'), 6.71 (s, 2 H, H-5', 6'), 5.87 (s, 1 H, H-8), 5.78 (s, 1 H, HO-C(3)), 4.92 (d, J = 10.9 Hz, 1 H, H-2), 4.82 (br s, 2 H, OH), 4.58 (br s, 1 H, OH), 4.46–4.47 (m, 1 H, H-6''b), 4.45 (d, J = 10.0 Hz, 1 H, H-1''), 4.43 (d, J = 10.9 Hz, 1 H, H-3), 3.96 (t, J = 9.1, 9.3 Hz, 1 H, H-2''), 3.63 (d, J = 10.9 Hz, 1 H, H-6''a), 3.04–3.15 (m, 3 H, H-3'', 4'', 5''). ¹³C NMR (125 MHz, DMSO-d ₆): δ = 198.0 (C-4), 166.0 (C-7), 162.6 (C-5), 161.3 (C-9), 145.8 (C-3'), 145.0 (C-4'), 128.0 (C-1'), 119.4 (C-6'), 115.3 (C-2'), 115.1 (C-5'), 106.0 (C-6), 100.2 (C-10), 94.7 (C-8), 82.9 (C-2), 81.6 (C-5''), 79.1 (C-3''), 72.9 (C-1''), 71.6 (C-3), 70.7 (C-2''), 70.3 (C-4''), 61.6 (C-6''). FAB-MS: m/z 467 [M + H]⁺.