Structure Elucidation of Prenyl- and Geranyl-Substituted Coumarins in Gerbera piloselloides by NMR Spectroscopy, Electronic Circular Dichroism Calculations, and Single Crystal X-ray Crystallography.

Crude ethyl acetate extract of Gerbera piloselloides (L.) Cass. was investigated by dual high-resolution PTP1B/α-glucosidase inhibition profiling and LC-PDA-HRMS. This indicated the presence of a series of unprecedented prenyl- and geranyl-substituted coumarin derivatives correlated with both α-glucosidase and PTP1B inhibitory activity. Repeated chromatographic separation targeting these compounds led to the isolation of 13 new compounds, of which ten could be isolated as both enantiomers after chiral separation. The structures of all isolated compounds were characterized by HRMS and extensive 1D and 2D NMR analysis. The absolute configurations of the isolated compounds were determined by comparison of experimental and calculated electronic circular dichroism spectra. Compound 6 features a rare furan-oxepane 5/7 ring system, possibly formed through addition of a geranyl unit to C-3 of 5-methylcoumarin, representing a new type of geranyl-substituted coumarin skeleton. Compounds 19 and 24 are the first examples of dimeric natural products consisting of both coumarin and chromone moieties.


Introduction
Diabetes mellitus is a multifactorial disease characterized by insufficient regulation of glucose metabolism, with severe long-term micro-and macrovascular complications such as retinopathy, neuropathy, nephropathy and cardiovascular diseases. More than 463 million people were affected by diabetes worldwide in 2019, and this figure is estimated to increase to 700 million by 2045 [1]. Type 2 diabetes (T2D) comprises more than 90% of all diabetes cases and is characterized by decreased insulin sensitivity in target organs like muscle and adipose tissue as well as reduced pancreatic insulin secretion [2,3]. Thus, management of a relatively stable blood glucose in the interval 5-10 mmol/L

Structure Elucidation of Prenyl-and Geranyl Coumarin Derivatives
Based on information from the two biochromatograms shown in Figures 1 and 2, material correlated with bioactivity as well as material suggested by LC-PDA-HRMS to be prenyl-or geranyl Material eluted with peaks 1-27 and fractions 1 and 2 were collected manually by preparativescale HPLC. Fraction 2 was subsequently separated using an analytical-scale pentafluorophenyl (PFP) HPLC column, and two successive separations were microfractionated. The dual highresolution inhibition profile of F2 is shown in Figure 2. This shows that the constituents eluting as peaks 13, 14 and 16-21 are correlated with α-glucosidase and/or PTP1B inhibitory activity. Thus, the extract seems to contain multiple new compounds correlated with inhibitory activity towards both PTP1B and α-glucosidase -one of many unique features of natural products.

Structure Elucidation of Prenyl-and Geranyl Coumarin Derivatives
Based on information from the two biochromatograms shown in Figures 1 and 2, material correlated with bioactivity as well as material suggested by LC-PDA-HRMS to be prenyl-or geranyl
This suggested 2 to be the 9,10-dihydroxylated analogue of bothrioclinin (7) (Figure 3), and COSY, ROESY, HSQC and HMBC correlations, of which selected correlations are shown in Figure 4, confirmed this and allowed the full assignment of 1 H and 13 C NMR signals provided in Tables 1 and 2. The relative configuration of 2 was established to be 9S*,10S* based on the coupling constant ( 3 J = 4.1 Hz) between H-9 and H-10, which suggests a cis configuration as also reported for cis-khellactone [35]. Finally, the absolute configuration of 2 was assigned 9S,10S by comparing the ECD spectrum of 2 ( Figure S8, Supplementary Material) with ECD data of cis-khellactone [35]. Compound 2 is a new compound for which the name gerbeloid 1 is proposed.         47, s). This confirms that 5 possesses a 2,2-dimethyl-2H-chromene core skeleton with an acetyl at C-6 and a second substituent at C-8, which based on the residual molecular formulas, is in agreement with a hydroxyisoprenyl unit (C 5 H 9 O). The COSY spectrum revealed a CH 2 -CH-O spin system, and HMBC correlations from H-1 to C-8 and C-8a as well as from H-2 , H-4 and H-5 to C-3 confirmed a 2-hydroxy-3-methylbut-3-en-1-yl unit at C-8. Compound 5 gradually degraded in solution, and it was therefore not possible to determine the absolute configuration at C-2 . The structure of 5 is shown in Figure 3, selected COSY, ROESY and HMBC correlations are shown in Figure 4, 1 H NMR, COSY, ROESY, HSQC and HMBC spectra are provided in Figures  ) as also seen for 2 and 7, vide supra; leaving a residual molecular formula of C 10 H 16 O 3 and three hydrogen deficiency index numbers for the substituents at C-3 and C-4. COSY correlations were observed between oxymethines H-9 ↔ H-10 and between H-12 ↔ H-13 ↔ H-14, and with HMBC correlations from H-9 to C-14 and from H-14 to C-9 as well as from H-10 and H-12 to the oxymethine C-11, this established the oxepane ring. Furthermore, the prop-1-en-2-yl unit at C-14 was established by HMBC correlations from H-14 to C-15, C-16 and C-17, and the methyl group at C-11 was established by HMBC correlations from H-19 to C-11. Finally, the ether linkage between C-4 and C-10 to form the 2,3-dihydrofuran moiety was inferred from the downfield shift of C-4 (δ 171.0) in 6 compared to 2 (δ 164.7) and 7 (δ 161.6), as also reported by Qiang and co-workers for gerbelin 3 [13], as well as the need to establish a ring closure as determined from the hydrogen deficiency index. Thus, the structure of 6 contains a rare furan-oxepane 5/7 ring system formed by a geranyl unit.
The relative configuration 9R*,10S*,11R*,14R* was established based on the cis coupling constants between H-9 and H-10 ( 3 J = 7.6 Hz) as well as the ROESY correlations H-9 ↔ H-14, and between H-10 ↔ CH 3 -19 ( Figure 5A). An ECD spectrum of 6 indicated the material to be an enantiomeric mixture, and subsequent chiral separation showed two enantiomers (6a and 6b) in a 1:1 ratio ( Figure 5B). The two enantiomers were isolated, and their experimental ECD spectra ( Figure 5D) were as expected completely opposite. After a conformational search, the 10 lowest-energy conformations of 6a were obtained by energy minimization at the B3LYP/6-311G(d,p) level in CH 3 CN ( Figure S4 and Tables S2 and S3, Supplementary Material) and their ECD spectra calculated. The total Boltzmann-averaged ECD spectrum is shown in Figure 5D, confirming 6a to be 9R,10S,11R,14R and 6b to be 9S,10R,11S,14S. isobutyrate [37]. Furthermore, the experimental ECD spectrum of 8 shows positive Cotton effects around 195 and 265 nm and a negative Cotton effect around 225 nm ( Figure S8, Supplementary Material), whereas the experimental and calculated ECD spectra of (−)-2S-6,12-dihydroxytremetone show negative Cotton effects around 205 and 280 nm and a positive Cotton effect around 240 nm [37].
HMBC and COSY correlations support the structures of 14 and 17, with HMBC correlations from H-18 to C-4, from H-9 to the carbonylic C-2 and the olefinic C-3, and from H-14 to the oxygenated C-15 confirming the pyrano[3,2-c]coumarin skeleton ( Figure 6). Compounds 14 and 17 are diastereomers, and the vicinal coupling 3 JH9,H14 = 6.8 Hz for 14 and a ROESY correlation between H-9 and H-14, support these two protons to be cis in 14; whereas the vicinal coupling 3 JH9,H14 = 10.9 Hz for 17 as well as ROESY correlations between H-9 and CH3-18 and between H-14 and CH3-19 support these two protons to be trans in 17. Both 14 and 17 are racemic mixtures, as seen from the 1:1 peak ratio upon chiral separation (Figures S1 and S2, Supplementary Material), and separation of the two
HMBC and COSY correlations support the structures of 14 and 17, with HMBC correlations from H-18 to C-4, from H-9 to the carbonylic C-2 and the olefinic C-3, and from H-14 to the oxygenated C-15 confirming the pyrano[3,2-c]coumarin skeleton ( Figure 6). Compounds 14 and 17 are diastereomers, and the vicinal coupling 3 J H9,H14 = 6.8 Hz for 14 and a ROESY correlation between H-9 and H-14, support these two protons to be cis in 14; whereas the vicinal coupling 3 J H9,H14 = 10.9 Hz for 17 as well as ROESY correlations between H-9 and CH 3 -18 and between H-14 and CH 3 -19 support these two protons to be trans in 17. Both 14 and 17 are racemic mixtures, as seen from the 1:1 peak ratio upon chiral separation (Figures S1 and S2, Supplementary Material), and separation of the two enantiomers allowed comparison of their experimental and calculated ECD spectra. The absolute configurations are therefore assigned 9R,14S for 14a and 9S,14R for 14b (with ∆ ESI of 0.94), and 9R,14R for 17a and 9S,14S for 17b (with ∆ ESI of 0.86), as depicted in Figure 6. 1 H NMR, COSY, ROESY, HSQC and HMBC spectra are provided in Figures S72-S77 and Figures S88-S93, Supplementary Material, and fully assigned 1 H and 13 C NMR data are provided in Tables 2 and 3. Compounds 14a/b and 17a/b are new compounds for which the names gerbeloid 6a/gerbeloid 6b and gerbeloid 8a/gerbeloid 8b, respectively, are proposed.  Tables 2 and 3. Compounds 14a/b and 17a/b are new compounds for which the names gerbeloid 6a/gerbeloid 6b and gerbeloid 8a/gerbeloid 8b, respectively, are proposed.    Synthesis of the pyrano[3,2-c]coumarin core skeleton via the microwave-accelerated domino Knoevenagel hetero Diels-Alder reaction between 4-hydroxycoumarin and citronellal [38], also provided products with the pyrano[2,3-b]chromen-4-one core skeleton. This is because the final cyclisation of the chromane-2,4-dione intermediate can occur via two different 4 + 2 cycloadditions, and one can predict that the potential geran-1-ylidene-chromane-2,4-dione intermediate for biosynthesis of 14 and 17 (lactones) could also lead to formation of the isomeric keto products with the pyrano[2,3-b]chromen-4-one core skeleton (Scheme S1, Supplementary Material). The 13 C NMR carbonyl resonances for C-4 in both 11 and 15 in fact suggest that the two compounds are chromone (keto) products, where C-4 is at δ 182.2 in 11 and δ 181.8 in 15 (Table 2). Additionally, HMBC correlations from CH 3 -18 and H-9 to C-2, from H-9 and H-10 to C-3 and from H-8 and CH 3 (Tables 1 and 2). Thus, 10 was identified as 4-hydroxy-5-methyl-3-(3-methyl-6-(prop-1-en-2-yl)cyclohex-2-en-1-yl)-2H-chromen-2-one, the 10,11-dehydroanalog of an intramolecular domino Knoevenagel ene adduct previously reported [37]. Key COSY and HMBC correlations used for structure elucidation are shown in Figure 4, and 10 is tentatively assigned the 9S*, 14R* configuration based on the coupling constants between H-9 and H-14 ( 3 J H9,H14 = 10.5 Hz trans) as also reported for the synthetic analog [38]. However, keto-enol tautomerism did neither allow chiral separation or acquisition of ECD spectra, and 10 was gradually degraded in solution. 1 H NMR, COSY, ROESY, HSQC and HMBC spectra are provided in Figures  S56-S60 19 and 25, respectively, showed each of these to consist of a coumarin (lactone) unit and a chromen-4-one (keto) unit, whereas 18 and 25 consist of two coumarin units each. Detailed analyses of 1D and 2D NMR data for 18 and 25 show that they consist of two pyrano[3,2-c]coumarin units joined together by bonds between C-9 and C-13 and between C-10 and C-9 , and comparison with NMR data reported for dibothrioclinin I and II [11], show that 18 and 25 are diastereomers of these. This was further supported by COSY and HMBC correlations (key correlations seen in Figure 7).  19 and 25, respectively, showed each of these to consist of a coumarin (lactone) unit and a chromen-4-one (keto) unit, whereas 18 and 25 consist of two coumarin units each. Detailed analyses of 1D and 2D NMR data for 18 and 25 show that they consist of two pyrano[3,2-c]coumarin units joined together by bonds between C-9 and C-13′ and between C-10 and C-9′, and comparison with NMR data reported for dibothrioclinin I and II [11], show that 18 and 25 are diastereomers of these. This was further supported by COSY and HMBC correlations (key correlations seen in Figure 7).   The core cyclohexane of 18 was identified as a chair conformation with axial-axial couplings between H-9 and H-10 and between H-9 and H-13 ax (J H9,H10 = J H9,H-13 ax = 12 Hz) and 1,3-diaxial ROEs between H-10, H-10 ax , and H-13 ax . With equatorial position of H-9 and CH 3 -14 , as seen from coupling constants and ROEs to H-10, H-10 ax , and H-13 ax , the relative configuration of 18 was established as 9R*,10S*,9 R*,11 S*. Contrary to this, the core cyclohexane of 25 was identified as a boat conformation, with very strong ROE correlation between H-9 and H-10 endo (and ROEs to H-9 , H-9 , H-13 eq and CH 3 -14 on the same side of the plane) and 1,3-diaxial ROEs between H-10 and H-13 ax . The relative configuration of 25 was therefore established as 9R*,10S*,9 S*,11 R* (Figure 7). Two pairs of enantiomers (18a and 18b, 25a and 25b) with the peak area ratios 1:1 were separated by chiral HPLC chromatography ( Figures S2 and S3, Supplementary Material). Comparison of the experimental and calculated ECD spectra (Figure 7) allowed assignment of the absolute configurations to be 9R,10S,9 R,11 S for 18a and 9S,10R,9 S,11 R for 18b (with ∆ ESI of 0.93), and 9R,10S,9 R,11 S for 25a and 9S,10R,9 S,11 R for 25b (with ∆ ESI of 0.82). 1 H NMR, COSY, ROESY, HSQC and HMBC spectra are provided in Figures S94-S99 and S118-S123, Supplementary Material, and fully assigned 1 H and 13 C NMR data are provided in Tables 4 and 5. Compounds 18a/b and 25a/b are new compounds for which the names gerbeloid 9a/gerbeloid 9b and gerbeloid 13a/gerbeloid 13b, respectively, are proposed.   (Figure 7). Furthermore, HMBC correlations from H-9 to C-11 and C-13 as well as from H-10 to C-3 and C-9 and from H-9 to C-10 and C-11 proved the bridging between the pyrano[2,3-b]chromen-4-one and the pyrano[3,2-c]coumarin skeleton as shown in Figures 3 and 7. Based on analysis of coupling constants and ROESY correlations (Figure 7), 19 was assigned the relative configuration 9R*,10R*,9 R*,11 R* and 24 was assigned the relative configuration 9R*,10S*,9 S*,11 S*. Additionally, X-ray crystal structure analysis of the racemate of 19 yielded the structure shown in perspective drawing (ORTEP-3) [39] in Figure 8. This agrees with the structure established based on the NMR data, confirming the 9R*,10R*,9 R*,11 R* configuration of 19.
After chiral separation of 19 and 24 (1:1 ratios of enantiomers 19a, 19b and 24a, 24b, see Figures  S2 and S3, Supplementary Material), ECD spectra of both pairs of enantiomers were acquired. These are shown in Figure 7, and comparison with the calculated ECD spectra of 19a and 24a allowed assignment of the absolute configurations to be 9R,10R,9 R,11 R for 19a and 9S,10S,9 S,11 S for 19b (with ∆ ESI of 0.77), and 9R,10S,9 S,11 S for 24a and 9S,10R,9 R,1 R for 24b (with ∆ ESI of 0.87) (Figure 7). 1 H NMR, COSY, ROESY, HSQC and HMBC spectra are provided in Figures S100-S105 and S112-S117, Supplementary Material, and fully assigned 1 H and 13 C NMR data are provided in Tables 4 and 5. Compounds 19a/b and 24a/b are new compounds for which the names gerbeloid 10a/gerbeloid 10b and gerbeloid 12a/gerbeloid 12b, respectively, are proposed. 2), C-3 (δ 103.9), and C-4 (δ 163.6) as also observed for 14 and 17, suggesting they share the same 5-methylcoumarin skeleton. The pyrano[3,2-c]coumarin core skeleton is supported by HMBC correlations from CH3-16 and CH3-17 to C-4, from CH3-17 to C-10, and from H-9 to C-2 and C-4 ( Figure 9) as well as correlations in the COSY spectrum between H-9 and H-10. In fact, the COSY spectrum showed correlations corresponding to the H-9 ↔ H-10 ↔ H-14 ↔ H-13 ↔ H-12 spin system, and HMBC correlations from H-12 to C-11 and C-17 revealed the cyclopentane ring system, whereas HMBC correlations from H-9, H-14, CH3-18 and CH3-19 to C-15 and from H-18 and H-19 to C-9 and C-14 revealed the dimethylated cyclobutane ring system. Thus, 23 was identified as a 5-methylcoumarin with a 6/5/4 ring system fused at C-3 and C-4. Two coumarins with similar 6/5/4 ring systems have been reported before [40], but linked to either C-5 and C-6 or C-7 and C-8 of the coumarin moiety; and the NMR data for the 6/5/4 ring system of the previously reported compounds resemble those observed for 23 (Table 2 and   2), C-3 (δ 103.9), and C-4 (δ 163.6) as also observed for 14 and 17, suggesting they share the same 5-methylcoumarin skeleton. The pyrano[3,2-c]coumarin core skeleton is supported by HMBC correlations from CH 3 -16 and CH 3 -17 to C-4, from CH 3 -17 to C-10, and from H-9 to C-2 and C-4 ( Figure 9) as well as correlations in the COSY spectrum between H-9 and H-10. In fact, the COSY spectrum showed correlations corresponding to the H-9 ↔ H-10 ↔ H-14 ↔ H-13 ↔ H-12 spin system, and HMBC correlations from H-12 to C-11 and C-17 revealed the cyclopentane ring system, whereas HMBC correlations from H-9, H-14, CH 3 -18 and CH 3 -19 to C-15 and from H-18 and H-19 to C-9 and C-14 revealed the dimethylated cyclobutane ring system. Thus, 23 was identified as a 5-methylcoumarin with a 6/5/4 ring system fused at C-3 and C-4. Two coumarins with similar 6/5/4 ring systems have been reported before [40], but linked to either C-5 and C-6 or C-7 and C-8 of the coumarin moiety; and the NMR data for the 6/5/4 ring system of the previously reported compounds resemble those observed for 23 (Tables 2 and 3  Chiral resolution of 23 gave two enantiomers (23a and 23b) in a 1:1 ratio ( Figure S2, Supplementary Material), and comparison of the ECD spectra of 23a and 23b with the ECD spectrum calculated for 23b (Figure 9), allowed assignment of the absolute configuration to be 9R,10R,11R,14S for 23a and 9S,10S,11S,14R for 23b, respectively, with ΔESI of 0.88. 1 H NMR, COSY, ROESY, HSQC and HMBC spectra are provided in Figures S106-S111, Supplementary Material, and fully assigned 1 H and 13 C NMR data are provided in Tables 2 and 3. Compounds 23a/b are new compounds for which the names gerbeloid 11a/gerbeloid 11b are proposed.
The material eluted with peaks 14-19 and 23-25 were correlated with both α-glucosidase and PTP1B inhibitory activity in the high-resolution inhibition profile (Figures 1 and 2). Due to limited amount of material isolated of the racemic mixtures as well as the individual enantiomers after chiral separation, it was not possible to make dilution series to obtain full dose-response curves. However, for the racemic mixtures of 17-19, 24 and 25, it was possible to test the inhibitory activity at a single concentration between 32.2 and 201.6 μM. The results are provided in Table 6 and show that the tested compounds show weak to moderate inhibitory activity towards both α-glucosidase and PTP1B. The reported values may be underestimated, because typically only one enantiomer of a racemic mixture is an active inhibitor of the target enzyme. Thus, isolation of more material and subsequent chiral separation are needed to obtain full dose-response curves for both enantiomers in both the αglucosidase and the PTP1B assay, as well as for being able to study the mode of inhibition of the many compounds correlated with one or both activities. The only prior study with dual inhibitory activity of PTP1B and α-glucosidase with this kind of compounds, is a recent study with a series of prenylated coumarins from Angelica decursiva, showing weak α-glucosidase inhibitory activity with IC50 values ranging from 65 to 172 μM, and PTP1B inhibitory activity with IC50 values ranging from 5 to 59 μM.
The coumarin skeleton is synthesized from acetyl-CoA and malonyl-CoA through the polyketide pathway in plants [42]. Inspired by the biosynthetic pathway of tetrahydrocannabinolic acid (THCA) in Cannabis sativa [43], the monoterpenoid-coumarins identified in this study are plausibly first added a monoterpenoid chain to C-3 of the coumarin skeleton, catalyzed by prenyl Chiral resolution of 23 gave two enantiomers (23a and 23b) in a 1:1 ratio ( Figure S2, Supplementary Material), and comparison of the ECD spectra of 23a and 23b with the ECD spectrum calculated for 23b (Figure 9), allowed assignment of the absolute configuration to be 9R,10R,11R,14S for 23a and 9S,10S,11S,14R for 23b, respectively, with ∆ ESI of 0.88. 1 H NMR, COSY, ROESY, HSQC and HMBC spectra are provided in Figures S106-S111, Supplementary Material, and fully assigned 1 H and 13 C NMR data are provided in Tables 2 and 3. Compounds 23a/b are new compounds for which the names gerbeloid 11a/gerbeloid 11b are proposed.
The material eluted with peaks 14-19 and 23-25 were correlated with both α-glucosidase and PTP1B inhibitory activity in the high-resolution inhibition profile (Figures 1 and 2). Due to limited amount of material isolated of the racemic mixtures as well as the individual enantiomers after chiral separation, it was not possible to make dilution series to obtain full dose-response curves. However, for the racemic mixtures of 17-19, 24 and 25, it was possible to test the inhibitory activity at a single concentration between 32.2 and 201.6 µM. The results are provided in Table 6 and show that the tested compounds show weak to moderate inhibitory activity towards both α-glucosidase and PTP1B. The reported values may be underestimated, because typically only one enantiomer of a racemic mixture is an active inhibitor of the target enzyme. Thus, isolation of more material and subsequent chiral separation are needed to obtain full dose-response curves for both enantiomers in both the α-glucosidase and the PTP1B assay, as well as for being able to study the mode of inhibition of the many compounds correlated with one or both activities. The only prior study with dual inhibitory activity of PTP1B and α-glucosidase with this kind of compounds, is a recent study with a series of prenylated coumarins from Angelica decursiva, showing weak α-glucosidase inhibitory activity with IC 50 values ranging from 65 to 172 µM, and PTP1B inhibitory activity with IC 50 values ranging from 5 to 59 µM.
The coumarin skeleton is synthesized from acetyl-CoA and malonyl-CoA through the polyketide pathway in plants [42]. Inspired by the biosynthetic pathway of tetrahydrocannabinolic acid (THCA) in Cannabis sativa [43], the monoterpenoid-coumarins identified in this study are plausibly first added a monoterpenoid chain to C-3 of the coumarin skeleton, catalyzed by prenyl transferase (Pathway 1 in Figure 10). As mentioned above, [4 + 2] cycloaddition reactions have been utilized for the total synthesis of structures similar to 11, 14, 15 and 17 (Scheme S1, Supplementary Material) [38], and therefore cyclisations via different Diels-Alder cycloadditions between the monoterpenoid and coumarin, followed by oxidations or reductions catalyzed by oxidoreductase enzymes, could be involved in the formation of 2, 6, 7, 10, 11, 14-17 and 23 ( Figure 10).
There are no prior studies on the biosynthesis of coumarins related to the dimeric coumarin derivatives isolated in this study, but total synthesis of pyranylquinoline alkaloids [44] suggests mechanisms for the plausible biosynthetic route for 18, 19, 24 and 25 presented as pathway 2 in Figure  10. In this route, the precursor of 18, 19, 24 and 25 is suggested to be 7, which is seen as a major peak in the ethyl acetate extract of G. piloselloides (Figure 1). The key steps in the dimerization of 7 include a ring opening, a hydride shift and a 1,4-addition (Figure 10), followed by 1,4-addition reactions leading to 18 and 25 with two coumarin moieties or 19 and 24 with one coumarin moiety and one chromone moiety in the structure (Figure 10).  There are no prior studies on the biosynthesis of coumarins related to the dimeric coumarin derivatives isolated in this study, but total synthesis of pyranylquinoline alkaloids [44] suggests mechanisms for the plausible biosynthetic route for 18, 19, 24 and 25 presented as pathway 2 in Figure 10. In this route, the precursor of 18, 19, 24 and 25 is suggested to be 7, which is seen as a major peak in the ethyl acetate extract of G. piloselloides (Figure 1). The key steps in the dimerization of 7 include a ring opening, a hydride shift and a 1,4-addition (Figure 10), followed by 1,4-addition reactions leading to 18 and 25 with two coumarin moieties or 19 and 24 with one coumarin moiety and one chromone moiety in the structure (Figure 10).

Plant Material and Extraction Procedure
Aerial parts of Gerbera piloselloides (L.) Cass was collected from Sandaohe Village, Duyun City, Guizhou Province, People s Republic of China in September 2017 and authenticated by Associate Prof. Chunhua Liu (Provincial Key Laboratory of Pharmaceutics in Guizhou Province, Guizhou Medical University). A voucher specimen (accession number JX20170208) is deposited at the Guizhou Medical University. The air-dried material (1.7 kg) was milled and extracted three times with 2 L ethyl acetate (1 h of ultra-sonication for each extraction). The combined extracts were filtered and concentrated under reduced pressure to give 20 g of dark green crude extract.

LC-PDA-HRMS
LC-PDA-HRMS analyses were performed on a model 1260 analytical-scale HPLC system consisting of a G1322A degasser, a G1311A quaternary pump, a G1316A thermostatted column compartment, and a G1315A photodiode-array detector (Agilent, Santa Clara, CA, USA) and a Bruker micrOTOF-Q II high-resolution mass spectrometer (Bruker Daltonik, Bremen, Germany). The column eluate was connected to a T-piece splitter directing 1% of the flow to the mass spectrometer and 99% of the flow the photodiode-array detector. The micrOTOF-Q II mass spectrometer, equipped with an ESI source, was operated in the positive-ion mode using a drying temperature of 200 • C, a capillary voltage of 4100 V, a nebulizer pressure of 2.0 bar, and a drying gas flow of 7 L/min. A solution of sodium formate clusters was injected automatically at the beginning of each run to enable internal mass calibration. Chromatographic separations were performed at 40 • C on a Luna C 18 (2) column, 150 × 4.6 mm i.d., 3 µm particle size, 100 Å pore size (Phenomenex, Torrance, CA, USA) at a flow rate of 0.5 mL/min and using a binary gradient of mobile phase A (CH 3 CN:H 2 O 5:95 + 0.1% FA) and mobile phase B (CH 3 CN:H 2 O 95:5 + 0.1% FA). Chromatographic separation and mass spectrometry were controlled using Hystar ver. 3.2 software (Bruker Daltonik).

Targeted Isolation of PTP1B/α-Glucosidase Inhibitors and Major Metabolites
Crude ethyl acetate extract (100 mg/mL in MeOH) was separated using a preparative-scale Agilent 1100 HPLC system comprising two G1361A preparative-scale solvent delivery pumps, a G2260A autosampler, and a G1365B multiple wavelength detector, controlled by Agilent ChemStation software ver. B.01.01. Injections (900 µL) were separated at room temperature on a Phenomenex Luna C 18 column (250 mm × 21.2 mm i.d., 5 µm particle size) using the following binary gradient of mobile phase A and B, vide supra, at a flow rate of 17

Chiral Separation of Enantiomers
Chiral resolution of 6, 15, 18, 19, 23, 24 and 25 was performed at room temperature on a Dionex Ultimate instrument consisting of a LPG-3200BX pump and a MWD-3000SD UV detector, both controlled by Chromeleon software ver. 6.80 (Thermo Fisher Scientific, Waltham, Ca, USA), and with a Rheodyne 9725I injector with 10 µL loop. Enantiomers of compounds 6, 15, 18, 19, 23 and 24 were separated at a flow rate of 1 mL/min on a Chiralpak AD-H column (250 × 4.6 mm i.d., 5.0 µm particle size) (Chiral Technologies, West Chester, PA, USA) isocratically eluted with the following mixtures as well as by single X-ray crystallography of one of the isolated dimeric coumarin derivatives (19). This study demonstrates the advantage of the combined use of dual high-resolution inhibition profiling and hyphenated LC-PDA-HRMS for pinpointing new compounds with potential as dual inhibitors, and the microplate-based inhibition profiling can be extended to include multiple pharmacological targets. Further improvement in the preparative-scale chiral separation of enantiomers is however needed to allow isolation of enantiomerically pure material for further characterization of the mode of inhibition.
Supplementary Materials: The following are available online, HRMS and 1D data of known compounds, chromatograms from chiral separations, ECD spectra, details of ECD calculation, X-ray crystallographic analysis of 19, and 1D and 2D NMR spectra of new compounds identified in this study.