Geraniol-Derived Monoterpenoid Glucosides from Rhodiola rosea: Resolving Structures by QM-HifSA Methodology

Abstract

Monoterpenoids are integral to the chemical composition of the widely used adaptogenic dietary supplement, Rhodiola rosea. The present study expands the chemical space and stereochemical information about these taxon-specific constituents from the isolation and characterization of five geraniol-derived glucosides, 1–5. While 1 and 2 exhibited almost identical NMR spectra and shared the same 2D structure ascribed to the 4-hydroxygeraniolglucoside previously described as rosiridin, the NMR-based Mosher ester method revealed the enantiomeric nature of their aglycone moieties. This marks the first report of enantiomeric aglycones among geraniol derivatives. These findings also resolve the long-standing dispute regarding the absolute configuration of rosiridin and congeneric C-4 hydroxylated geraniols, and may help explain incongruent bioactivity reports of R. rosea extract. Moreover, the three previously undescribed geranioloids 3–5 were full characterized by extensive spectroscopic analysis. Quantum mechanics-driven ¹H iterative functionalized spin analysis (QM-HifSA) was performed for all isolates and provides detailed NMR spin parameters, with adequate decimal pre, which enable distinction of such close congeners exhibiting near identical NMR spectra with high specificity. The outcomes also reinforce the importance of reporting chemical shifts and coupling constants with adequate decimal place precision as a means of achieving specificity and reproducibility in structural analysis.

Graphical Abstract

Preparations from the rhizomes and rotos of Rhodiola rosea L. (Crassulaceae) have a history of use as adaptogens, especially in situations of stress and/or decreased performance such as fatigue and depression.^1–3 The genus Rhodiola is native to the Arctic regions of Europe, Asia, and North America and comprises over 200 species.^1,3 Phytochemical studies report R. rosea to be rich in cinnamyl alcohol glycosides, salidroside and its aglycone, monoterpenoid glycosides, benzyl alcohol glycosides, cyanogenic glycosides, and flavonoids. While cinnamyl alcohol glycosides and salidroside have been comprehensively investigated and are considered the bioactive constituents, as some investigations suggest that geraniol-type monoterpenoids might play a critical role in the treatment of specific conditions, such as in depression and senile dementia.^1,5

Previous studies of R. rosea monoterpenoids led to the first report of myrtenol glycosides with enantiomeric aglycones in the Rhodiola genus.⁶ To gain a better understanding of the possibly unique structures of Rhodiola monoterpenoids and their purported biological effects, five geranioloid glucosides, 1–5, were isolated from the rhizomes/roots of R. rosea through a combination of repeated centrifugal partition chromatography (CPC) and HPLC separations. The structures, including their absolute configurations, were determined by a combination of 1D/2D NMR spectroscopy, quantum-mechanics driven analysis of the ¹H NMR spectra (QM-HifSA)⁷ the advanced Mosher’s ester method, and LC-HR-MS. Unexpectedly, 1 and 2 possessing the same molecular weight and near superimposable ¹H and ¹³C NMR spectra turned out to be glucosides of enantiomeric aglycones. Moreover, 1 and 2 share the 2D structure with that of rosiridin, a C-4 hydroxylated geraniolglucoside for which a long-standing controversy exists regarding its absolute configuration at C-4. Compound 3 is a new natural product exhibiting a shifted double bond (Δ²→Δ³) relative to 1/2. Monoterpenoid 4 is a new cyclic monoterpenoid aglycone, and 5 the O-methyl ether analogue of 1/2. The findings show that stereochemical variations are the likely cause for the strong spectroscopic and chromatographic overlap of the monoterpenoid metabolome of R. rosea.

RESULTS AND DISCUSSION

Reinvestigation of the Absolute Configuration of Rosiridin.

Rosiridin was first reported from R. rosea rhizomes by Kurkin et al. in 1985 as a C-4 hydroxylated geraniolglucoside, exhibiting a negative specific rotation [α]_D²⁰ (−32.7 c 1.1, acetone).⁸ In 2001, Fan et al. isolated a geraniol derivative from R. sachalinensis and concluded it to be rosiridin due to an identical 2D structure derived from 1D and 2D NMR data and a negative specific rotation value, [α]_D²⁵ −16.9 (c 0.087, MeOH). The authors went on to determine the absolute configuration of C-4 in the isolate to be R by employing the Mosher ester method.⁹ However, in 2008, Yoshikawa et al. contradicted Fan’s conclusion by assigning the absolute configuration of C-4 as S also by employing the Mosher ester method, and working with an isolate from R. sachalinensis that had the same 2D structure as rosiridin.¹⁰ One notable difference to the 2001 report was that the 2008 study reported specific rotation values that matched the 1985 report very closely, even in different solvents {[α]_D²⁵ (−32.1 c 2.06, MeOH) and [α]_D²² (−31.1 c 4.65, acetone)}. Interestingly, the Yoshikawa et al. compound showed almost identical NMR chemical shifts with those of that identified by Fan et al.⁹ Of note, bioactivity tests carried out by van Diermen et al. in 2009 revealed that rosiridin showed significant inhibitory activity against monoamine oxidase B,⁵ inhibitors of which could potentially be used as therapeutic agents in senile dementia treatment.¹¹ The small but notable differences in the specific rotation of the isolates implied potential differences in the optical purity of the isolates.

In the present study, two monoterpenoids obtained after repeated pre-fractionation and final semi-HPLC purification also showed close specific rotation values {[α]_D²⁰ (−38 c 0.1, MeOH) for 1 and [α]_D²⁰ (−35 c 0.1, MeOH) for 2} as well as near superimposable ¹H and ¹³C NMR data (Table 1 and Figure 1). Subtle differences were recognized for H-1a, H₃-8, and H-1′, with Δδ_H (δ₁-δ₂) values around −0.01 and 0.02 ppm, respectively, which appeared to be negligible, but could also be potentially characteristic of near identical molecules.¹² The relevance of this hypothesis for rosiridin has to be considered more seriously when the NMR and MS data jointly revealed that 1 and 2 share the same 2D structure with those of the asserted rosiridins isolated by Fan et al. and Yoshikawa et al.^9,10 Using the Mosher ester method, the absolute configurations at C-4 of the monoterpenoid moieties were subsequently determined as R for 1 and S for 2 (Figure 2). Supported by more detailed NMR analysis, the results show that naturally occurring rosiridin consists of two C-4 epimers being derived from a single enantiomeric aglycone. Upon closer inspection assisted by Quantum mechanics-driven ¹H iterative functionalized spin analysis (QM-HifSA), most Δδ_H values fell in the lower ppb range, with the largest being Δδ_H-1′ at just 20 ppb (Figure 3). This once more reinforces the importance of reporting ¹H NMR chemical shifts with four decimal place precision for more specific compounds dereplication.^6,12,13 It is also noteworthy that very close specific rotation values go hand in hand with near identical chemical shift values of 1 and 2: Δδ_H-1a,H-1′ is 105.5 ppb for 1 and 70.3 ppb for 2, and these differences are sufficient to distinguish these epimers unequivocally (Figure 1).

Table 1.

¹H and ¹³C NMR Data of 1 and 2^a,b

position	δH (J in Hz)		δ C
	1	2	1	2
1a	4.3862, ddddd (−12.34, 6.04, 0.73, 0.57, 0.44)	4.3706, dddd (−12.35, 5.94, 0.85, 0.51)	66.00	66.05
1b	4.3056, ddd (−12.34, 7.38, 0.14)	4.3090, dddd (−12.35, 7.46, 0.66, 0.31)
2	5.5720, dddd (7.38, 6.04, −1.34, −0.88)	5.5715, dddd (7.46, 5.94, −1.32, −0.90)	122.91	122.90
3			142.95	142.91
4	3.9740, ddddd (6.86, 6.79, −0.88, 0.44, 0.14)	3.9684, dddddd (7.46, 6.10, −0.90, 0.51, 0.31, −0.22)	78.04	78.01
5a	2.2525, ddddd (−14.64, 7.08, 6.79, 1.05, 0.62)	2.2439, ddddd (−15.26, 6.68, 6.10, 1.20, 0.58)	34.76	34.77
5b	2.2396, ddddd (−14.64, 7.03, 6.86, 1.10, 0.65)	2.2418, ddddd (−15.26, 7.46, 7.46, 1.11, 0.57)
6	5.1070, dddd (7.08, 7.03, −1.53, −1.33)	5.1037, dddd (7.46, 6.68, −1.49, −1.37)	121.61	121.58
7			134.08	134.07
8	1.7002, dddd (−1.53, 1.10, 1.05, −0.21)	1.6913, dddd (−1.49, 1.20, 1.11, −0.34)	26.07	26.01
9	1.6257, dddd (−1.33, 0.65, 0.62, −0.21)	1.6223, dddd (−1.37, 0.58, 0.57, −0.34)	18.03	18.02
10	1.6659, ddd (−1.34, 0.73, 0.57)	1.6676, dddd (−1.32, 0.85, 0.66, −0.22)	12.03	12.00
1′	4.2807, d (7.84)	4.3003, d (7.85)	102.76	102.81
2′	3.1823, dd (9.26, 7.84)	3.1830, dd (9.25, 7.85)	75.05	75.10
3′	3.3409, dd (9.26, 8.92)	3.3442, dd (9.25, 8.84)	78.16	78.13
4′	3.2789, dd (9.75, 8.92)	3.2812, dd (9.56, 8.84)	71.67	71.66
5′	3.2303, ddd (9.75, 5.81, 2.34)	3.2448, ddd (9.56, 5.89, 2.29)	78.04	78.00
6′a	3.8700, dd (−11.92, 2.34)	3.8713, dd (−11.92, 2.29)	62.78	62.77
6′b	3.6666, dd (−11.92, 5.81)	3.6681, dd (−11.92, 5.89)

^aThe ¹H and ¹³C NMR data were acquired in methanol-d₄ at 400 and 100 MHz, respectively.

^bThe δ_H (in ppm) and J (in Hz) values were determined by QM-HifSA analysis.

Figure 1.

¹H (upper part) and ¹³C (lower part) NMR data (methanol-d₄, 400 MHz, 298 K) comparison of 1 (blue) and 2 (red). 1 and 2 showed very similar chemical shifts, with slight differences observed for H-1a, H-8, and H-1′ in ¹H NMR spectra and near identical spectra for ¹³C NMR spectra. The inset showed that the most observable evidence for the differentiation of 1 and 2 is the Δδ_H-1a,H-1′.

Figure 2.

Δδ_S-R values of MTPA esters of 1–3 used for the determination of absolute configuration.

Figure 3.

Graphical representation of the small chemical shift differences (Δδ_H, in ppm) between 1 and 2 (see Table 1 for chemical shifts of 1 and 2). While the chemical shift congruence is high, in line with the near identical structures of 1 and 2, the largest difference of can be observed for Δδ_H‑1′ at 20 ppb.

Structure Elucidation of 3–5.

Based on its negative ion HRESIMS data, 3 was recognized as an isomer of the rosiridins. Its ¹H and ¹³C NMR data also closely resembled those of 1 and 2. However, differences were observed in the ¹H-¹H COSY correlations: both the two olefinic hydrogens in 3 were found to have correlations with the same methylene moiety, suggesting the relocation of one of the double bonds in 3 (Figure 4). Furthermore, the HMBC data revealed that, unlike the 3,7-dimethyl-2,6-octadiene-1,4-diol backbone in 1 and 2, 3 possessed the 3,7-dimethyl-3,6-octadiene-1,2-diol skeleton with a shifted double bond and hydroxy group (Figure 4). The absolute configuration at C-2 of this skeleton was then unambiguously determined to be S using the Mosher ester method (Figure 2). Collectively, the structure of 3 was determined as the new geranioloid glucoside, (3E,2S)-2-hydroxy-3,7-dimethyl-3,6-octadien-1-yl β-d-glucopyranoside.

Figure 4.

¹H-¹H COSY and key HMBC correlations of 3–5.

Compound 4 again shared the same molecular formula, C₁₆H₂₈O₇, with 1–3 based on the (−)-HRESIMS data. In contrast to 1–3, 4 displayed only one olefinic hydrogen and one pair of olefinic carbon signals in its ¹H and ¹³C NMR spectra, respectively, indicating that 4 consists of one additional ring system to satisfy the observed degree of hydrogen deficiency. Moreover, two additional oxygenated carbon resonances appeared at δ_C 84.53 and 82.78 in the ¹³C NMR spectrum of 4. All these observations indicated that 4 is similar to the sachalinols A–C,⁹ representing geranioloids that contain a tetrahydrofuran ring formed via intramolecular cyclization between carbon C-4 (shifted to higher frequency around 84 ppm vs. 78 ppm in 1 and 2) and the oxygenated carbon C-7 (~82 ppm). After detailed interpretation of the 2D NMR data (Figures 4 and S-16 to S-18, Supporting Information), the structure of 4 was deduced as the new geranioloid, 3-(5,5-dimethyltetrahydrofuranyl)-2E-buten-1-yl β-d-glucopyranoside.

Compared to 1 and 2, 5 contained an additional methyl group, corresponding to the molecular formula C₁₇H₃₀O₇, which was gleaned from the (−)-HRESIMS ion at m/z 391.1964 [M + HCOO]⁻. Analysis of its ¹H and ¹³C NMR spectra (Table 2) suggested that 5 is still structurally closely related to 1 and 2, but exhibited an O-methyl group as key difference, as a result of methylation of the OH group at C-4. The HMBC correlations (Figures 4 and S-23, Supporting Information) from the O-methyl hydrogens resonating at δ 3.1950 to the oxygenated carbon C-4 (δ 88.33) explicitly confirmed the structure of 5 to be (2E)-4-methoxy-3,7-dimethyl-2,6-octadien-1-yl β-d-glucopyranoside. In this research, the R. rosea was extracted with methanol. In order to preclude the possibility that 5 is an artifact due to solvolysis, the ¹H NMR spectra of 1 and 2 dissolved in methanol-d₄ were recorded over time. As shown in Figure 5, no changes were noticed during re-analysis after three and 15 months, which is supporting evidence for 5 being a naturally occurring metabolite.

Table 2.

The ¹H and ¹³C NMR Data of 3–5^a,b

position	3		4		5
	δH (J in Hz)	δ C	δH (J in Hz)	δ C	δH (J in Hz)	δ C
1a	3.9359, dd (−10.44, 2.87)	74.50	4.3916, ddd (−12.30, 6.09, 0.75)	66.12	4.3936, dddd (−12.48, 5.69, 0.91, 0.12)	65.93
1b	3.4423, dd (−10.44, 9.09)		4.2914, ddd (−12.30, 7.36, 0.62)		4.3460, dddd (−12.48, 7.40, 0.59, 0.46)
2	4.1903, ddddd (9.09, 2.87, −1.00, −0.34, 0.21)	77.21	5.6788, dddd (7.36, 6.09, −1.30, −1.12)	122.50	5.5768, dddd (7.40, 5.69, −1.37, −0.59)	125.97
3		134.89		141.27		139.69
4	5.4552, dddd (7.24, 7.20, −1.34, −1.00)	127.28	4.3827, ddd (7.70, 6.58, −1.12)	84.53	3.5127, dddddd (6.97, 6.91, 0.46, −0.37, −0.37, 0.12)	88.33
5a	2.7378, ddddddd (−15.41, 7.24, 7.20, 1.17, 0.58, 0.43, 0.21)	27.56	2.0918, dddd (−12.78, 6.78, 6.58, 6.37)	32.10	2.2810, ddddd (−14.70, 6.97, 6.91, 1.30, 0.82)	33.46
5b	2.7296, ddddd (−15.41, 7.24, 7.19, 0.97, 0.23)		1.8185, dddd (−12.78, 9.97, 7.70, 5.59)		2.1710, dddd (−14.70, 7.20, 6.97, 1.10)
6a	5.1028, dddd (7.24, 7.19, −1.51, −1.37)	123.58	1.8056, ddd (−13.65, 9.97, 6.37)	39.45	5.0684, ddddd (7.20, 6.97, −1.52, −1.34, −0.37)	121.38
6b			1.8132, ddd (−13.65, 6.78, 5.59)
7		132.71		82.78		134.02
8	1.6823, ddd (−1.51, 1.17, 0.97)	25.81	1.2859, s	28.95	1.6803, dddd (−1.34, 1.30, 1.10, −0.20)	25.97
9	1.6358, ddd (−1.37, 0.58, 0.23)	17.76	1.2639, s	28.30	1.6042, ddd (−1.52, 0.82, −0.20)	17.98
10	1.6549, ddd (−1.34, 0.43, −0.34)	12.64	1.6666, ddd (−1.30, 0.75, 0.62)	12.25	1.6039, dddd (−1.37, 0.91, 0.59, −0.37)	11.14
1′	4.2856, d (7.77)	104.76	4.2919, d (7.90)	103.00	4.3012, d (7.90)	102.94
2′	3.2208, dd (9.33, 7.77)	75.25	3.1769, dd (9.31, 7.90)	75.05	3.1964, dd (9.35, 7.90)	75.02
3′	3.3666, dd (9.33, 8.77)	77.82	3.3431, dd (9.31, 8.35)	78.10	3.3359, dd (9.35, 8.59)	78.15
4′	3.2808, dd (9.03, 8.77)	71.59	3.2804, dd (9.32, 8.35)	71.66	3.2880, dd (9.39, 8.59)	71.64
5′	3.2714, ddd (9.03, 5.96, 2.37)	78.00	3.2467, ddd (9.32, 5.38, 1.94)	78.00	3.2393, ddd (9.39, 5.36, 2.15)	78.07
6′a	3.8650, dd (−11.88, 2.37)	62.69	3.8674, dd (−11.92, 1.94)	62.78	3.8743, dd (−11.92, 2.15)	62.75
6′b	3.6596, dd (−11.88, 5.96)		3.6667, dd (−11.92, 5.38)		3.6737, dd (−11.92, 5.36)
OMe					3.1950, s	56.40

^aThe ¹H and ¹³C NMR data were acquired in methanol-d₄ at 400 and 100 MHz, respectively.

^bThe δ_H (in ppm) and J (in Hz) values were determined by QM-HifSA analysis.

Figure 5.

¹H NMR spectra (methanol-d₄, 400 MHz, 298 K) of 1 and 2 in methanol-d₄ over time. The experiment revealed that both 1 and 2 showed no conversion in methanol-d₄.

Owing to the lack of a secondary hydroxy group in the aglycone portions of 4 and 5, absolute configurations could not be determined by the Mosher ester method. Attempts to use electronic circular dichroism (ECD) to determine their absolute configuration showed flat curves indicating a lack of usable chromophores, which could not be applied with confidence for further analysis. In addition, owing to the volatile characteristics of monoterpenes, methodologies aimed at removal of the sugar units to release the aglycone moieties for the absolute configuration determination were not considered in this study. Thus, the absolute configurations of C-4 in 4 and 5 remain unknown until, e.g., confirmatory synthesis can be performed. The assignment of the glucose to the D series of sugars was based on the very closely matching ¹³C NMR resonances relative to corresponding resonance in previously investigated monoterpenoids, for which the absolute configurations were determined via enzymatic hydrolysis and subsequent specific rotation value comparison,⁶ experiments also performed with the title plant. When combined with chiral aglycones, the ¹³C chemical shifts of a sugar become highly indicative of its absolute configuration due to the diastereotopic nature of the glycoside. Consequently, the isolates in the present study were considered optically pure and, therefore, were not subjected to chiral phase HPLC.

CONCLUSION

In conclusion, while the biological activity and potential phytotherapeutic role of the geraniol-type monoterpenoids require further study, the present report resolves the long-standing controversy regarding the absolute configuration of C-4 hydroxylated geraniol-type monoterpenoids. While previous studies must have obtained geraniol derivatives with either S or R C-4 configuration, this is the first report of the coexistence of both epimers. These findings not only reveal that Rhodiola can generate both C-4S and 4R rosiridins, representing major components in R. rosea, but could potentially explain why some of the in vitro and in vivo biological studies regarding the validation of the therapeutic claims of extracts of R. rosea showed positive results, while others found no evidence for the claims.^1,14

Representing structurally distinct constituents, geraniol derivatives have the potential to serve as chemotaxonomic and possibly bioactive marker compounds of R. rosea, a widely used adaptogenic dietary supplement. The role of the ratio of 4R vs 4S hydroxylated geraniol in this regard remains to be explored. The present study could provide inspiration for future studies on R. rosea materials and correlation of their biological activity with the chirality of (C-4 hydroxylated) geranioloids, which now can be reproducibly identified via NMR analysis as exemplified by 1 and 2 above. In this regard, precise δ parameter sets generated from 1D HNMR are critical elements of NMR-involved structural dereplication and elucidation.¹⁷ Especially when performed in conjunction with computer-assisted spectroscopic simulation and iteration, this approach extracts the maximum amount of structural information from the data sets and facilitates rapid and accurate structural distinction. As demonstrated in the present study, the reporting of chemical shifts to at least four decimal places (1 ppb) and inclusion of HifSA profiles analysis are essential for specific and rapid dereplication, as well as support the identification of analogous monoterpene glycosides that likely occur in R. rosea and related plants.

EXPERIMENTAL SECTION

General Experimental Procedures.

Optical rotations at the sodium D line (589 nm) were measured with a PerkinElmer 241 digital polarimeter (Waltham, MA, USA) using a quartz cell with a path length of 100 mm in MeOH. NMR experiments were performed on a JEOL Resonance Inc. JNMR-ECZ400/L1 (Akishima, Tokyo, Japan) operating at 400 MHz and equipped with a broadband Z-gradient high-resolution SuperCool probe, or with a JEOL 600 MHz NMR spectrometer outfitted with a room temperature broadband Z-gradient high-resolution Royal probe. HifSA calculations were carried out using the CT (Cosmic Truth) software tool from NMR Solutions, Kuopio, Finland. The 3D models were constructed using Chem 3D Ultra (ChemOffice Pro. v. 18.1, PerkinElmer Informatics, Waltham, MA, USA), and the structures were energy minimized using the MM2 module.

HRESIMS analyses were carried out using a Waters 2695 (Milford, MA, USA) solvent delivery system connected to a Waters SYNAPT quadrupole/time-of-flight mass spectrometer. Semipreparative HPLC was performed with a YMC-ODS AQ semi-preparative column (10 × 250 mm, 5 μm) on a Waters 600 Delta system using MeOH-H₂O or MeCN-H₂O as the mobile phase at a flow rate of 3.0 mL/min.

Centrifugal partition chromatography (CPC) was performed on an SCPE-250 extractor from Gilson Inc (Middleton, WI, USA). Solvents and reagents were purchased from Thermo Fisher Scientific (Hanover Park, IL, USA) or Sigma Aldrich (St. Louis, MO, USA). HPLC grade solvents were purchased from Sigma Aldrich and methanol-d₄ (99.8 atom % D) were purchased from Cambridge Isotope Laboratories Inc. (Andover, MA, USA). The samples were weighed with a Mettler Toledo XS105 Dual Range analytical balance. A Pressure-Lok gas syringe (Baton Rouge, LA, USA) was used for volumetric NMR sample preparation. TLC was performed on Alugram precoated 0.2 mm thick silica gel G/UV254 10 × 20 cm aluminum plates, Macherey-Nagel GmbH & Co. (Düren, Germany). Visualization of the developed TLC plates utilized UV light (254 and 365 nm) as well as spraying the plates with a vanillin/H₂SO₄ solution followed by heating as a general-purpose detection reagent. Syringe filters (CHROMAFIL Xtra PTFE-20/13, pore size: 0.20 μm, 13 mm diameter) were used for CPC and HPLC sample filtration.

Plant Material.

The material (UIC/NIH Botanical Center code BC 872) was from Rhodiola rosea plants cultivated at the University of Alaska Experimental Farm near Palmer, AK (61 34.189′ N 149 15.336 W). The original plants were from Norway, and their seeds were sourced from Arrgo, Alberta, Canada. Collection vouchers are deposited in the University of Illinois Herbarium under DS16452-DS16460.

Extraction and Isolation.

The pulverized rhizomes/roots (2.75 kg) were defatted with hexanes and dichloromethane successively, then extracted with MeOH at room temperature four times (4 × 6 L) to afford the crude extract (680 g, semi-dry). Both the UHPLC chromatogram and the ¹H NMR spectrum of the MeOH extracts of R. rosea showed an elevated, bumpy baseline caused by its proanthocyanidin (PAC) content. In order to reduce the interference introduced by PACs during isolation as much as possible, the R. rosea crude extract (RCE) was dissolved in H₂O and partitioned with a mixture of CHCl₃/n-BuOH (1:4, v/v). In this partition, most of the PACs were retained in the lower phase along with a negligible amount of the compounds of interest. The non-PAC portion (upper phase) of the RCE partition was chromatographed over an HP-20 column, eluting with a step gradient of H₂O/MeOH to afford five fractions: 100% H₂O, 30% MeOH, 50% MeOH, 70% MeOH, and 100% MeOH labeled fr. 1–5, respectively. Fr. 3 was fractionated by CPC (264 mL rotor, S_f = 0.70, flow rate 25 mL/min, 2500 rpm) into five subfractions (fr. 3a–e) using the CHCl₃–MeOH–H₂O (ChMWat) 10:6:4 solvent system in descending mode. Subfraction 3a was purified by semi prep-HPLC (45% MeOH) to produce six tertiary fractions (frs. 3a1–3a6). Among them, fr. 3a6 yielded 5 (t_R = 21.5 min, 22.82 mg) through semi prep-HPLC (25% CH₃CN). Fr. 3b was fractionated into eight fractions (frs. 3b1–3b8) by HPLC (40% MeOH). Compound 4 (7.17 mg) was obtained from fr. 3b5. Fr. 2 was further separated by CPC (264 mL rotor, S_f = 0.60, flow rate 25 mL/min, 2500 rpm) into six fractions (frs. 2a–2f) using the CHCl₃–MeOH–H₂O (ChMWat) 10:7:3 solvent system in descending mode. Compound 3 (t_R = 28.2 min, 3.31 mg) was isolated from fr. 2a through semiprep-HPLC (23% CH₃CN). Compounds 1 (t_R = 27.2 min, 5.05 mg) and 2 (t_R = 28.6 min, 92.74 mg) were obtained from fr. 2b through semi prep-HPLC (40% MeOH).

(4R)-rosiridin (1): colorless solid; [α]²⁰_d-38 (c 0.1, MeOH); NMR (400 MHz, methanol-d₄) see Table 1.

(4S)-rosiridin (2): colorless solid; [α]²⁰_d-35 (c 0.1, MeOH); NMR (400 MHz, methanol-d₄) see Table 1.

(3E,2S)-2-hydroxy-3,7-dimethyl-3,6-octadien-1-yl β-d-glucopyranoside (3): colorless solid; [α]²⁰_d-18 (c 0.1, MeOH); NMR (400 MHz, methanol-d₄) see Table 2; HRESIMS m/z 377.1817 [M + HCOO]⁻, calcd for C₁₇H₂₉O₉ (1.3 ppm), 377.1812.

3-(5,5-Dimethyltetrahydrofuranyl)-2E-buten-1-yl β-d-glucopyranoside (4): colorless solid; [α]²⁰_d-23 (c 0.1, MeOH); NMR (400 MHz, methanol-d₄) see Table 2; HRESIMS m/z 377.1817 [M + HCOO]⁻, calcd for C₁₇H₂₉O₉ (1.3 ppm), 377.1812.

(2E)-4-methoxy-3,7-dimethyl-2,6-octadien-1-yl β-d-glucopyranoside (5): colorless solid; [α]²⁰_d-12 (c 0.1, MeOH); NMR (400 MHz, methanol-d₄) see Table 2; HRESIMS m/z 391.1964 [M + HCOO]⁻, calcd for C₁₈H₃₁O₉ (−1.0 ppm), 391.1968.

Preparation of the (R)- and (S)-MTPA Ester Derivatives of 1–3.

Two aliquots each of the compounds 1–3 (0.3 mg each for 1 and 2 and 0.2 mg each for 3 in 200 μL) were transferred into separate NMR tubes and dried under vacuum overnight at room temperature. Then, either 6 μL of (R)- or (S)-MTPA chloride along with 500 μL of pyridine-d₅ were added. The NMR reaction tubes were immediately sealed, shaken vigorously to ensure even mixing, and stored in a desiccator overnight until the reaction was complete. ¹H NMR spectra were used to monitor the reaction. The ¹H NMR spectra of the final (R)- and (S)-MTPA adducts were recorded directly after each reaction, and the chemical shifts were assigned based on ¹H–¹H COSY NMR experiments. Ambiguous signals were excluded from the calculation of Δδ_S–R values.¹⁵

¹H NMR data of the (R)-MTPA ester of 1 (600 MHz, pyridine-d₅) (Figure S-24, Supporting Information): δ 5.9121 (1H, br t, J = 6.62, 6.35 Hz, H-2), 5.5909 (1H, br t, J = 7.79, 7.47, H-4), 4.9624 (1H, br t, J = 7.15, 6.73 Hz, H-6), 4.5158 (1H, dd, J = 12.28, 6.35 Hz, H-1a), 4.2767 (1H, dd, J = 12.28, 6.62 Hz, H-1b), 2.4486 (1H, m, H-5a), 2.2711 (1H, m, H-5b), 1.6563 (3H, br s, H-10), 1.5211 (3H, br s, H-8), 1.4033 (3H, br s, H-9).

¹H NMR data of the (S)-MTPA ester of 1 (600 MHz, pyridine-d₅) (Figure S-26, Supporting Information): δ 5.7153 (1H, br t, J = 6.31, 6.02 Hz, H-2), 5.5233 (1H, br t, 7.15, 6.31, H-4), 5.0879 (1H, br t, J = 6.59, 6.31 Hz, H-6), 4.3659 (1H, dd, J = 12.33, 6.01 Hz, H-1a), 4.0918 (1H, dd, J = 12.33, 6.31 Hz, H-1b), 2.5141 (1H, m, H-5a), 2.3057 (1H, m, H-5b), 1.6076 (3H, br s, H-8), 1.5173 (3H, br s, H-10), 1.4895 (3H, br s, H-9).

¹H NMR data of the (R)-MTPA ester of 2 (600 MHz, pyridine-d₅) (Figure S-28, Supporting Information): δ 5.7958 (1H, br t, J = 6.70, 6.38 Hz, H-2), 5.5113 (1H, br t, J = 7.01, 6.07 Hz, H-4), 5.1015 (1H, br t, J = 7.33, 6.59 Hz, H-6), 4.4709 (1H, dd, J = 12.77, 6.38 Hz, H-1a), 4.2860 (1H, dd, J = 12.77, 6.70 Hz, H-1b), 2.5249 (1H, m, H-5a), 2.2937 (1H, m, H-5b), 1.5915 (3H, br s, H-8), 1.5371 (3H, br s, H-10), 1.4772 (3H, br s, H-9).

¹H NMR data of the (S)-MTPA ester of 2 (600 MHz, pyridine-d₅) (Figure S-30, Supporting Information): δ 5.8034 (1H, br t, J = 6.49, 6.14 Hz, H-2), 5.5541 (1H, br t, J = 7.01, 6.75 Hz, H-4), 4.9847 (1H, br t, J = 6.49, 6.14 Hz, H-6), 4.4078 (1H, dd, J = 12.27, 6.14 Hz, H-1a), 4.1175 (1H, dd, J = 12.27, 6.49 Hz, H-1b), 2.4609 (1H, m, H-5a), 2.2587 (1H, m, H-5b), 1.6341 (3H, br s, H-10), 1.5367 (3H, br s, H-8), 1.4218 (3H, br s, H-9).

¹H NMR data of the (R)-MTPA ester of 3 (600 MHz, pyridine-d₅) (Figure S-32, Supporting Information): δ 5.8614 (1H, br dd, J = 6.92, 5.43 Hz, H-2), 5.7400 (1H, br t, J = 7.31, 7.24 Hz, H-4), 5.0593 (1H, br t, J = 7.09, 7.01 Hz, H-6), 4.0503 (1H, dd, J = 11.50, 5.43 Hz, H-1a), 4.0427 (1H, dd, J = 11.50, 6.92 Hz, H-1b), 2.6862 (1H, m, H-5a), 2.6497 (1H, m, H-5b), 1.6964 (3H, br s, H-10), 1.5719 (3H, br s, H-8), 1.4875 (3H, br s, H-9).

¹H NMR data of the (S)-MTPA ester of 3 (600 MHz, pyridine-d₅) (Figure S-34, Supporting Information): δ 5.8312 (1H, dd, J = 8.72, 2.93 Hz, H-2), 5.4914 (1H, br t, J = 7.29, 7.09 Hz, H-4), 4.9949 (1H, br t, J = 7.68, 7.29 Hz, H-6), 4.0943 (1H, dd, J = 11.26, 2.93 Hz, H-1a), 3.8369 (1H, dd, J = 11.26, 8.72 Hz, H-1b), 2.6283 (1H, m, H-5a), 2.6132 (1H, m, H-5b), 1.5784 (3H, br s, H-8), 1.4917 (3H, br s, H-9), 1.4807 (3H, br s, H-10).

Acquisition of qHNMR Spectra.

Samples were dissolved in 200 μL of methanol-d₄ then transferred into 3 mm NMR tubes (Landisville, NJ, USA). All NMR experiments were performed at 298 K (25 °C) using standard JEOL pulse sequences. Chemical shifts (δ) are expressed in ppm with reference to the residual solvent signals (3.3100 ppm for ¹H and 49.0000 ppm for ¹³C). The qHNMR spectra were acquired using established quantitative parameters,¹⁶ specifically: a 60 s relaxation delay, ensuring fully quantitative conditions for all the signals; GARP ¹³C-decoupling; 64 scans and 64k acquired data points at 400 MHz; and a 90° flip angle. NMR data were processed and analyzed using MestReNova 14.1.2 software from Mestrelab Research S.L. (Santiago de Compostela, Spain). For qHNMR analysis, the following processing scheme was used: a mild Lorentzian-to-Gaussian window function (−0.3 Hz line broadening and 0.05 (5%) asas Gaussian factor) was applied, followed by zero filling the original 64k to 128k data points before Fourier Transformation, manual phasing, and fifth order polynomial baseline correction.

Chemical Formulas.