Synthesis and Detailed Examination of Spectral Properties of (S)- and (R)-Higenamine 4′-O-β-d-Glucoside and HPLC Analytical Conditions to Distinguish the Diastereomers

Higenamine is a tetrahydroisoquinoline present in several plants that has β-adrenergic receptor agonist activity. Study of the biosynthesis of higenamine has shown the participation of norcoclaurine synthase, which controls the stereochemistry to construct the (S)-isomer. However, when isolated from nature, higenamine is found as the racemate, or even the (R)-isomer. We recently reported the isolation of higenamine 4′-O-β-d-glucoside. Herein, its (R)- and (S)-isomers were synthesized and compared to precisely determine the stereochemistry of the isolate. Owing to their similar spectral properties, determination of the stereochemistry based on NMR data was considered inappropriate. Therefore, a high-performance liquid chromatography method was established to separate the isomers, and natural higenamine 4′-O-β-d-glucoside was determined to be a mixture of isomers.

Biologically, higenamine is synthesized from dopamine and 4-hydroxyphenylacetic acid by the catalytic action of norcoclaurine synthase (NCS) [12,13]. The stereochemistry of higenamine is regulated by NCS, with its (S)-isomer known as the enzymatic product [12,13]. However, higenamine is often isolated from the plant as the racemate [3,14]. Furthermore, in the case of lotus, the (S)-enantiomer was isolated from the leaves [15], while the (R)-enantiomer was found in the seed embryo located near the plumule [16].
We previously reported the isolation of higenamine 4 -O-β-D-glucoside (1) from lotus plumule as a glucose uptake inducer against muscle cells, and indicated that its stereochemistry is (S) [17]. However, during a following research related to the structure-activity relationship study of 1, a question arose about the stereochemistry of natural 1 isolated from lotus plumule, and the above reports have inspired us to precisely examine the stereochemistry of natural 1. As the correct assignment of stereochemistry in natural products is important, in this article, we chemically synthesized diastereomers 1R and 1S and compared their spectral properties. Careful examination of the spectral properties of the synthetic diastereomers indicated that distinguishing them using spectral methods was difficult. Therefore, we established high-performance liquid chromatography (HPLC) analysis conditions to separate the diastereomers, and used it to analyze the stereochemistry of natural 1 isolated from lotus plumule.

Results and Discussion
To distinguish the diastereomers of 1 and determine their stereochemistry unequivocally, samples of 1R and 1S were required. Therefore, 1R was selectively synthesized using a previously reported procedure and purified using reverse-phase HPLC [18], while 1S was prepared using the same procedure by employing RuCl[R,R-TsDPEN(p-cymene)] as the catalyst for enantioselective reduction (Scheme 1). examination of the spectral properties of the synthetic diastereomers indicated that distinguishing them using spectral methods was difficult. Therefore, we established high-performance liquid chromatography (HPLC) analysis conditions to separate the diastereomers, and used it to analyze the stereochemistry of natural 1 isolated from lotus plumule.

Results and Discussion
To distinguish the diastereomers of 1 and determine their stereochemistry unequivocally, samples of 1R and 1S were required. Therefore, 1R was selectively synthesized using a previously reported procedure and purified using reverse-phase HPLC [18], while 1S was prepared using the same procedure by employing RuCl[R,R-TsDPEN(p-cymene)] as the catalyst for enantioselective reduction (Scheme 1). Nuclear magnetic resonance (NMR) spectra of synthetic diastereomers 1R and 1S were obtained. However, in deuterium oxide, no apparent difference was found between the diastereomers (Table 1). This result might be due to (i) the stereochemistry of the synthetic products not being constructed as intended, resulting in the same diastereomer being isolated from both procedures; or (ii) the two diastereomers being indistinguishable under these NMR conditions.
To avoid the former possibility, the stereochemistry of the higenamine moiety was confirmed by deprotection of 3S and 3R to obtain 5R and 5S, respectively, followed by HPLC purification and specific rotation measurements. The specific rotation values (5S: −20.0 (c = 0.68, methanol); 5R: +21.4 (c = 0.58, methanol)) were in good correlation with previous reports, which confirmed their stereochemistry [19,20]. Furthermore, the isomers were analyzed using a Chiral CD-Ph column (Shiseido Co., Tokyo, Japan) to determine their retention times ( Figure 1). The specific rotations of synthetic 1S (−32.4 (c = 1.5, methanol)) and 1R (−16.3 (c = 1.5, methanol)) showed distinct values, which excluded the possibility of isomerization during the synthesis of 5R and 5S. Nuclear magnetic resonance (NMR) spectra of synthetic diastereomers 1R and 1S were obtained. However, in deuterium oxide, no apparent difference was found between the diastereomers (Table 1). This result might be due to (i) the stereochemistry of the synthetic products not being constructed as intended, resulting in the same diastereomer being isolated from both procedures; or (ii) the two diastereomers being indistinguishable under these NMR conditions.
To avoid the former possibility, the stereochemistry of the higenamine moiety was confirmed by deprotection of 3S and 3R to obtain 5R and 5S, respectively, followed by HPLC purification and specific rotation measurements. The specific rotation values (5S: −20.0 (c = 0.68, methanol); 5R: +21.4 (c = 0.58, methanol)) were in good correlation with previous reports, which confirmed their stereochemistry [19,20]. Furthermore, the isomers were analyzed using a Chiral CD-Ph column (Shiseido Co., Tokyo, Japan) to determine their retention times ( Figure 1). The specific rotations of synthetic 1S (−32.4 (c = 1.5, methanol)) and 1R (−16.3 (c = 1.5, methanol)) showed distinct values, which excluded the possibility of isomerization during the synthesis of 5R and 5S.   Deuterated solvents other than deuterium oxide were then tested for their suitability to separate the NMR signals of diastereomers of 1. Methanol-d4 gave a result similar to that of deuterium oxide, Deuterated solvents other than deuterium oxide were then tested for their suitability to separate the NMR signals of diastereomers of 1. Methanol-d 4 gave a result similar to that of deuterium oxide, and was therefore considered unsuitable (Table 2). In contrast, pyridine-d 5 gave relatively separated 1 H-NMR signals for the two diastereomers (Table 3). To determine whether these 1 H-NMR signal differences in pyridine-d 5 provided adequate evidence to determine the stereochemistry of 1, a mixture of 1R and 1S was analyzed by 1 H-NMR. The mixture showed distinguishable signals (Figure 2), but a problem was observed that prevented the use of NMR for the determination of stereochemistry.     As shown in Figure 2, when the 1 H-NMR signals of the mixture were compared with those of 1R or 1S, differences in the chemical shifts (ppm) were apparent. For example, the signal of the anomeric proton in glucose moiety (Glc-1) is 5.45 ppm for 1R and 5.43 ppm for 1S; however, for the mixture of the two, the same proton appeared at 5.45 and 5.48 ppm, for which correspondence is unclear. These shifts in ppm value were presumably due to differences in the solution conditions, pH, or sample concentration during NMR analysis, which would influence the status of the cyclic amine in the tetrahydroisoquinoline moiety. They thus indicated that determining the stereochemistry of 1 through NMR experiments could potentially cause assignment errors, rendering it an inappropriate method.
Because of the fluctuation of 1 H-NMR data with analytical conditions, distinguishing the diastereomers of 1 required an alternative method. HPLC is a method capable of separating and analyzing diastereomers. In HPLC analysis, analytical conditions are adjusted by changing the eluent used, which is present in large excess, and should therefore be a reliable method for analyzing the stereochemistry of 1.
Although a frequently employed C18 column was not capable of separating the isomers, a COSMOSIL Cholester column (Nakalai Tesque, Inc., Kyoto, Japan) was found to separate 1R and 1S to a good degree. Under the HPLC condition employed using this column, 1R and 1S showed apparent separation with retention times of 29 min and 32 min, respectively, indicating its ability to efficiently distinguish these isomers ( Figure 3).  Figure 2, when the 1 H-NMR signals of the mixture were compared with those of 1R or 1S, differences in the chemical shifts (ppm) were apparent. For example, the signal of the anomeric proton in glucose moiety (Glc-1) is 5.45 ppm for 1R and 5.43 ppm for 1S; however, for the mixture of the two, the same proton appeared at 5.45 and 5.48 ppm, for which correspondence is unclear. These shifts in ppm value were presumably due to differences in the solution conditions, pH, or sample concentration during NMR analysis, which would influence the status of the cyclic amine in the tetrahydroisoquinoline moiety. They thus indicated that determining the stereochemistry of 1 through NMR experiments could potentially cause assignment errors, rendering it an inappropriate method.

As shown in
Because of the fluctuation of 1 H-NMR data with analytical conditions, distinguishing the diastereomers of 1 required an alternative method. HPLC is a method capable of separating and analyzing diastereomers. In HPLC analysis, analytical conditions are adjusted by changing the eluent used, which is present in large excess, and should therefore be a reliable method for analyzing the stereochemistry of 1.
Although a frequently employed C18 column was not capable of separating the isomers, a COSMOSIL Cholester column (Nakalai Tesque, Inc., Kyoto, Japan) was found to separate 1R and 1S to a good degree. Under the HPLC condition employed using this column, 1R and 1S showed apparent separation with retention times of 29 min and 32 min, respectively, indicating its ability to efficiently distinguish these isomers (Figure 3). Natural 1, isolated from lotus plumule, underwent stereochemical analysis using the established HPLC method. The results showed that natural 1 was a mixture of both isomers (1R/1S = 59/41, Figure 3). This was further confirmed by analyzing the lotus plumule extract to avoid the possibility of isomerization during isolation, which showed a similar ratio (1R/1S = 60/40, data not shown).
Other than lotus plumule, Phoebe chekiangensis has been reported to contain higenamine 4 -O-β-D-glucoside (1) [21]. The stereochemistry of 1 in P. chekiangensis was determined as (S) by Wu et al. by comparing with the NMR spectrum and specific rotation of synthetic 1R [18]. Our results indicate that this result may need to be reassessed.
In conclusion, we synthesized 1R and 1S and directly compared their spectral data. We showed that the NMR spectra of these diastereomers were similar and their chemical shifts were influenced by solution conditions. This indicated that NMR was an unreliable method for determining the stereochemistry of 1, and presumably those of other similar compounds. Therefore, an HPLC method for separating the diastereomers was established and natural 1, isolated from lotus plumule, was analyzed and shown to be a mixture of diastereomers. Natural 1, isolated from lotus plumule, underwent stereochemical analysis using the established HPLC method. The results showed that natural 1 was a mixture of both isomers (1R/1S = 59/41, Figure 3). This was further confirmed by analyzing the lotus plumule extract to avoid the possibility of isomerization during isolation, which showed a similar ratio (1R/1S = 60/40, data not shown).
Other than lotus plumule, Phoebe chekiangensis has been reported to contain higenamine 4′-O-β-D-glucoside (1) [21]. The stereochemistry of 1 in P. chekiangensis was determined as (S) by Wu et al. by comparing with the NMR spectrum and specific rotation of synthetic 1R [18]. Our results indicate that this result may need to be reassessed.
In conclusion, we synthesized 1R and 1S and directly compared their spectral data. We showed that the NMR spectra of these diastereomers were similar and their chemical shifts were influenced by solution conditions. This indicated that NMR was an unreliable method for determining the stereochemistry of 1, and presumably those of other similar compounds. Therefore, an HPLC method for separating the diastereomers was established and natural 1, isolated from lotus plumule, was analyzed and shown to be a mixture of diastereomers.

Synthesis
As the higenamine moiety with a protective group on the cyclic amine contains isomers, which were either rotamers resulting from the 4-hydroxybenzyl group or diastereomers resulting from the stereochemistry of the protective group on the cyclic amine, detailed data were not obtained for compounds 3 and 4.