Serrulatane diterpenoids from the leaves of Eremophila glabra and their potential as antihyperglycemic drug leads

diterpenoids


Introduction
Natural products (NPs) continue to be a treasure chest in the hunt for new drug leads.NPs are characterized by a high diversity and structural complexity, and the traditional use of plant extracts can be helpful in the determination of which plants to investigate (Atanasov et al., 2021).As reviewed by Newman and Cragg (2020), around 38% of all approved small-molecule drugs in the period 1981-2019 were either of natural origin, derived from NPs, or synthetic drugs based on NP pharmacophores, showing that NPs continue to directly contribute to and to inspire the development of new drug leads.
The traditional use of plants in the Eremophila genus has led to the search for bioactive compounds in the more than 200 identified species (Ghisalberti, 1994).Eremophila belongs to the family Scrophulariaceae (figworts) and is endemic to Australia.Eremophila exhibits an immense morphological diversity throughout the Eremean (arid) biome, which covers approximately 70% of the Australian continent, being tolerant to extreme weather conditions, such as drought, fire or cold.Eremophila glabra (R.Br.)Ostenf.(tar bush) is widespread and grows in a wide range of soils and vegetation associations, although only in the drier areas of the continent (Brown and Buirchell, 2011).
Diabetes is a multiparametric metabolic disorder, and the global prevalence of diabetes in the 20-79-year age group was estimated to be around 463 million in 2019 (International Diabetes Federation, 2019).The number of cases is estimated to increase to around 700 million by 2045, causing an estimated similar increase in the 760 billion USD global health care expenditures encountered in 2019.These figures serve to emphasise the future challenges of providing adequate health care worldwide.Approximately 90% of all diabetes cases are caused by type 2 diabetes (T2D).T2D is characterized by insufficient insulin production by the pancreatic β-cells and/or insulin resistance in target organs like the liver, muscles, and fat tissue (Chatterjee et al., 2017).This leads to hyperglycaemia, which, if left untreated, results in long-term complications like nephropathy, neuropathy, retinopathy, and cardiovascular diseases.T2D is strongly associated with a sedentary lifestyle and obesity.Currently available treatment of T2D includes, in addition to medication, a combination of increased physical exercise and changed dietary habits, aiming at maintaining a lowered and stable blood glucose.α-Glucosidase is an enzyme present in the small intestine, where it catalyses hydrolysis of (1 → 4)-linked α-D-glucose residues from the terminal non-reducing ends of primarily disaccharides, trisaccharides, and oligosacchrides into absorbable monosaccharides.Inhibition of α-glucosidase therefore lowers the postprandial blood glucose levels, and acarbose, voglibose, and miglitol have been approved as drugs inhibiting α-glucosidase (Padhi et al., 2020), but are associated with a series of side effects.Therefore, new and improved α-glucosidase inhibitors are in demand.Protein tyrosine phosphatase 1B (PTP1B) is a protein tyrosine phosphatase catalyzing dephosphorylation of tyrosine residues of the insulin receptor kinase, and hereby downregulating insulin signalling (Galic et al., 2005).Studies have furthermore shown that PTP1B affects peptidic hunger hormon leptin signalling and that mice lacking PTP1B are resistant to obesity as well as being more sensitive to insulin (Zabolotny et al., 2002).Inhibition of PTP1B thus appear to be a promising way of treating T2D and potentially also to prevent obesity.No drugs with PTP1B inhibitory activity are on the market, but several studies have identified diterpenoids as effective PTP1B inhibitors in vitro (Hjortness et al., 2018;Wubshet et al., 2016;Kjaerulff et al., 2020), and thus future potential drugs for treatment of T2D and obesity.E. glabra is positioned in the resin-accumulating phylogenetic clade H14, known to accumulate serrulatane diterpenoids (Gericke et al., 2021), and was therefore selected as our target species in our search for new drug leads with potential antidiabetic properties.

Results and discussion
As part of our ongoing investigation of plants for antihyperglycaemic activity, a crude acetonitrile leaf resin extract of E. glabra (R.Br.)Ostenf.
was assessed for α-glucosidase and PTP1B inhibitory activity and provided IC 50 values of 19.3 ± 1.2 μg/mL and 11.8 ± 2.1 μg/mL, respectively (Supplementary data Fig.S36).The extract was therefore further investigated to pinpoint the individual constituents responsible for the observed inhibitory activities.

High-resolution inhibition profiling of the crude resin extract and isolation of selected compounds
To identify individual constituents responsible for the observed bioactivities, the crude resin extract was fractionated using analyticalscale high-performance liquid chromatography (HPLC) and the eluate from 20 to 60 min was microfractionated into two sets of two 96-well microplates.The content in each well was assayed, and the results expressed as percentage α-glucosidase and PTP1B inhibition were plotted at the respective retention times to provide the dual highresolution α-glucosidase/PTP1B inhibition profile shown in Fig. 1.
Likewise, ROE correlations between H-5 and H-11, between H-4 and H 3 -18, and between H3α and H 2 -12 suggest the 11S* configuration.The hydroxyl group at C-2 was assigned the β-orientation based on a strong ROE correlation between H-1 and H-2 and weaker ROE correlations between H-2 and H 3 -20, H-3α, H-3β, and H 2 -12.Finally, H 3 -16 and H 3 -17 were assigned based on ROE correlations to H-14 and H 2 -13, respectively.The absolute configurations at C-1, C-4, and C-11 in serrulatanes isolated from Eremophila spp.have previously been established by comparison of experimental and calculated ECD spectra (Kumar et al., 2018 and unpublished result from our own lab).In addition to that, it has recently been proposed that serrulatane diterpenoids originate from the same 8,9-dihydroserrulat-14-ene precursor containing a basic serrulatane skeleton derived from nerylneryldiphosphate (Gericke et al., 2020), and it can therefore be assumed that the stereochemistry at  C1, C-4, and C-11 is preserved in naturally occurring serrulatanes produced by plants of the Scrophulariaceae family.Compound 7 is thus tentatively assigned the 1S,2R,4S,11S configuration.Comparison of calculated and experimental ECD spectra was subsequently used to substantiate this configurational assignment.Thus, experimental ECD spectra of all isolated serrulatane diterpenoids were grouped according to structural similarities, with ECD spectra of compounds with a carboxylic acid at C-19 and a hydroxyl group at C-8 in group A, compounds with hydroxyl groups at both C-7 and C-8 in group B, and compounds with a carboxylic acid at C-19 in group C (Fig. 4A-C).ECD calculations were subsequently performed using one model structure for each group (Fig. 4D-F), with a model structure for 5a from group A, a model structure for 8a from group B, and a model structure for 7 from group C.
To reduce the computational cost due to the high flexibility of the side chain at C-4, model compounds without C-14 to C-17, which has no significant effect on the ECD spectrum, were used for the quantum chemical calculations.Thus, the 7-model shown in Fig. 4F were used for calculation of ECD spectra of 7. Conformational search for the 3D structure of (1S,2R,4S,11S)-7 model led to 64 possible conformers, among which the 12 dominant ones with more than 1% of the Boltzmann population were used for calculation of ECD spectra (Supplementary data Tables S11 and S12).The experimental ECD spectrum of 7, characterized by a negative and positive band around 215 and 235 nm, respectively, overall matched the calculated spectrum (Fig. 4F).However, it should be noted that the intensity of the negative Cotton effect at 215 nm in the calculated ECD spectrum is by far weaker than that in the experimental spectrum, which could be explained using a truncated structure of 7 for the performed ECD calculations.Thus, compound 7 was identified as (1S,2R,4S,11S)-2-hydroxyserrulat-14-en-19-oic acid, which has not previously been reported and for which the name eremoglabrane C is suggested. 1H and 13 C NMR data are provided in Table 1, selected 2D NMR correlations are shown in Fig. 3, 1 H and 13 C NMR data and all correlations from 2D NMR experiments are provided in Supplementary data Table S4, and 1 H NMR, COSY, HSQC, HMBC and ROESY spectra are provided in Supplementary data Figs.S11-S15.
The constituent collected as peak 8a showed an [M+H] + ion at m/z 333.2066 (calcd.333.2060,ΔM − 1.7), suggesting a molecular formula of C 20 H 28 O 4 , which corresponds to a hydrogen deficiency index of 7. The 1D and 2D NMR spectra indicated that this compound exhibited the same serrulatane core structure as the other compounds isolated in this work, but with altered substitution patterns of the aromatic ring and at C-1.The singlet aromatic proton H-5 (δ H 6.51) suggested substitutions at aromatic C-6, C-7, and C-8, and a singlet methyl resonance (δ H 2.17, H 3 -19) with HMBC correlations to C-5, C-6, and C-7 confirmed a 6-methylation of the aromatic ring.The molecular formula and the chemical shift of C-7 and C-8 (δ C 141.7 and δ C 144.7, respectively) confirmed the 7,8dihydroxylation of the aromatic ring.An HMBC correlation from H-1 to the carbonylic signal (δ C 180.5, C-20) confirmed the carboxylic acid positioned at C-1.The 2D ROESY spectrum furthermore showed correlations between α-oriented H-1, H-2α, and H-3α, and between β-oriented H-4, H-3β and H-2β, whereas no correlation was observed between H-1 and H-4, indicating that these are on opposite sides of the plane, as also observed for all other compounds identified in this work, supporting a 1R,4S configuration (Fig. 3).Based on biosynthetic considerations as stated above, 8a was tentatively assigned the 1R,4S,11S configuration.Similar to 7 and 5a, ECD calculation was used to confirm the absolute configuration of 8a.Conformational search for the 3D structure of (1R,4S,11S)-8a model (Fig. 4E) led to 33 possible conformers, among which 9 dominant ones were used for calculation of ECD spectra (Supplementary data Tables S13 and S14).As shown in Fig. 4E, both the calculated and experimental ECD spectra were characterized by two positive Cotton effects around 220 and 245 nm, and two negative bands near 230 and 290 nm, even though the relative intensities were not perfectly reproduced, likely due to the truncated side chain of the model structure.Thus, compound 8a was identified as the previously unreported (1R,4S,11S)-7,8-dihydroxyserrulat-14-en-20-oic acid, for which the name eremoglabrane D is suggested. 1H and 13 C NMR data are provided in Table 1, selected 2D NMR correlations are shown in Fig. 3, 1 H and 13 C NMR data and all correlations from 2D NMR experiments are provided in Supplementary data Table S5, and 1 H NMR, COSY, HSQC, HMBC and ROESY spectra are provided in Supplementary data Figs.S16-S20.
The 2D ROESY spectrum showed correlations between the α-oriented H-1, H-2α, and H-3α, and between β-oriented H-4, H-3β and H-2β (Fig. 3).Based on biosynthetic considerations, compound 8b was tentatively assigned the 1R,4S,11S configuration.The experimental ECD spectrum of 8b (Fig. 4A), has the same overall curvature as the other spectra in class A, albeit with a small blue shift as it is the only compound with carboxylic acid at C-1, and 8b displays a negative Cotton effect around 220 nm, and a positive Cotton effect around 230 nm (in addition to a small positive Cotton effect around 245 nm.Thus, based on the experimental ECD spectrum and biosynthetic considerations, vide supra, 8b was identified as the previously unreported (1R,4S, 11S)-20-methoxycarbonyl-8-hydroxyserrulat-14-en-19-oic acid, for which the name eremoglabrane E is suggested. 1H and 13 C NMR data are provided in Table 2, selected 2D NMR correlations are shown in Fig. 3, 1 H and 13 C NMR data and all correlations from 2D NMR experiments are provided in Supplementary data Table S6, and 1 H NMR, COSY, HSQC, HMBC and ROESY spectra are provided in Supplementary data Figs.S21-S25.
The constituent collected as peak 10b showed an [M+H] + ion at m/z 359.2225 (calcd.359.2230,ΔM − 2.3), suggesting a molecular formula of C 22 H 30 O 4 , corresponding to a hydrogen deficiency index of 8. Compound 10b showed a 1 H-NMR spectrum very similar to that of 7, with identical resonances for most of the hydrogens.However, a change in chemical shift from δ H 4.20 to δ H 5.31 for H-2 and an additional methyl resonance at δ H 2.01 ppm (H 3 -2 ′ ) with an HMBC correlation to a carbonylic signal (δ C 170.7 ppm, C-1 ′ ) was observed, confirming the acetoxylation of C-2.This change correlated with the molecular formula observed for this compound.The 2D ROESY spectrum accordingly also showed ROE correlations very similar to that of 7, indicating the same relative stereochemistry for compound 10b, and the absolute configuration of 10b was tentatively assigned the 1S,2R,4S,11S configuration.The experimental ECD spectrum for 10b (Fig. 4C) showed a negative Cotton effect at 215 nm and a positive cotton effect at 235 nm, which is very similar to that of 7, further supporting that the absolute configuration of 10b is the same as that for 7. Thus, compound 10b was identified as the previously unreported (1S,2R,4S,11S)-2-acetoxyserrulat-14-en-19-oic acid, for which the name eremoglabrane F is suggested. 1H and 13 C NMR data are provided in Table 2, selected 2D NMR correlations are shown in Fig. 3, 1 H and 13 C NMR data and all correlations from 2D NMR experiments are provided in Supplementary data Table S7, and 1 H NMR, COSY, HSQC, HMBC and ROESY spectra are provided in Supplementary data Figs.S26-S30.
The constituent collected as peak 13c showed an [M+H] + ion at m/z 415.2464 (calcd.415.2479,ΔM +3.6), suggesting a molecular formula of C 25 H 34 O 5 , which corresponds to a hydrogen deficiency index of 9.The 1D and 2D NMR data suggested an 8-hydroxyserrulat-14-ene-19-oic acid structure, as described previously for compound 8b.Comparing the NMR data of compound 13c with that of 8b and 9 revealed a different substitution at C-1.Thus, an olefinic proton at δ H 6.97 (qq, J H3',H4' = 7.0, J H3',H5' = 1.2, H-3 ′ ) with COSY correlations to H 3 -4' (δ H 1.83, dq, J H3', H4' = 7.0, J H4',H5' = 1.2) and H 3 -5' (δ H 1.87, br t J H3',H5' = J H4',H5' = 1.2) and an HMBC correlation to a carbonylic signal (δ C 169.9, C-1 ′ ) revealed the presence of an angelic acid.Furthermore, HMBC correlations from H-5 ′ to C-1 ′ and from H-20 to C-1 ′ confirmed the angelic acid moiety being attached to C-20.The 2D ROESY spectrum showed correlations consistent with the other serrulatanes isolated in this work, with H-4 correlating with H-5, H-11, and H-18, and no correlations between H-1 and H-4, supporting that the configuration is the same throughout all serrulatanes isolated here.The experimental ECD spectrum of 13c showed positive Cotton effects around 210 and 250 nm and negative Cotton effects around 225 and 300 nm (Fig. 4A), as do the remaining spectra in class A, and in agreement with that of the calculated spectrum of 5a (except for the negative band at 300 nm as discussed above).Compound 13c was therefore identified as the previously unreported (1R,4S,11S)-20-angeloyloxy-8-hydroxyserrulat-14-en-19-oic acid, for which the name eremoglabrane G is suggested. 1H and 13 C NMR data are provided in Table 2, selected 2D NMR correlations are shown in Fig. 3, 1 H and 13 C NMR data and all correlations from 2D NMR experiments are provided in Supplementary data Table S8, and 1 H NMR, COSY, HSQC, HMBC and ROESY spectra are provided in Supplementary data Figs.S31-S35.
Eremophila is the major genus in the tribe Myoporeae.Recent extensive phylogenetic studies based on nuclear ribosomal DNA sequencing divided the Myoporeae into eight clades labelled A to H (Gericke et al., 2021).Clade H is divided into 14 subclades, H1 to H14, of which the H14 subclade is characterized by species having resinous leaves and a unique chemical signature indicating a large unknown chemical space of serrulatane chemistry with substructural motif blooms (Gericke et al., 2021).E. glabra, the focus species of our current study, belongs to the H14 subclade and our isolation of twelve serrulatane-type diterpenoids of which seven were previously unknown natural products serves to start filling this unknown chemical space with known structures.
The initial steps in the biosynthesis of serrulatane-type diterpenoids have recently been studied in E. drummondii and E. denticulate subsp.trisulcata (Gericke et al., 2020).These two species are also positioned within the H14 subclade indicating that the same biosynthetic intermediates would be involved in E. glabra.From both species, a cis-prenyltransferase was identified catalyzing the formation of nerylneryl diphosphate from the C5 isoprenoids IPP and DMAPP (Fig. 5).The nerylneryl diphosphate is then converted to the serrulatane backbone by the activity of a class I terpene synthase to produce 8,9-dihydroserrulat-14-ene, which is rapidly aromatized to serrulat-14-ene in a non-enzymatic reaction (Gericke et al., 2020).The final decoration of the serrulatane backbone is expected to involve the action of cytochrome P450s as well as acetyl-, angelate-, and methyl transferases.Thus, multiple P450 oxygenations of the C1-and C19-methyl groups are envisioned to give rise to the introduction of carboxyl group in 5a, 5b, 5c, 6, 7, 8b, 9, 10b, 11, and 13c (Hansen et al., 2021), whereas single oxygenation reactions are observed at C-2, C-7, C-8, C-14, and C-20 dependent on the compound, likewise envisioned to be P450-catalysed (Hansen et al., 2021).Esterification with acetic acid and angelic acid, observed at C-2 and/or C-20 in 5a/5c, 9, 10b, 11, and 13c, involves acetyl transferases and a putative angelic acid transferase (Luo et al., 2016;Callari et al., 2018;Wang et al., 2021).Additional modification of 8b involves methyl ether formation catalysed by a methyltransferase.
The biosynthesis of Eremophila diterpenoids, i.e., serrulatane diterpenoids, from the cisoid precursor nerylneryl diphosphate rather than from the transoid precursor geranylgeranyl diphosphate is manifested in the unique diterpenoid structures discovered in the present study.

Conclusions
In conclusion, the use of high-resolution α-glucosidase and PTP1B inhibition profiling of the crude leaf resin extract of E. glabra in combination with HPLC-PDA-HRMS, semi-preparative-scale HPLC, and NMR spectroscopy, led to the identification of serrulatane-type diterpenoids with moderate α-glucosidase and/or PTP1B inhibitory activity.
However, seven previously unreported serrulatane diterpenoids were isolated and identified together with four previously reported flavonoids and five previously reported serrulatane diterpenoids.Two of the serrulatane diterpenoids showed weak PTP1B inhibitory activity and one exhibited dual α-glucosidase and PTP1B inhibitory activity.This study thus substantiates that species in Eremophila are a rich source of previously unreported serrulatane diterpenoids and that Eremophila is a valuable genus in the search for antidiabetic drug leads.
Eremophila spp.are culturally important plants for many of Australia's First Peoples, the Aboriginal peoples.If you use the information here provided to make commercial products, we urge you to strongly consider benefit sharing with the Aboriginal communities or groups in the areas where these species grow.We acknowledge that this work took place on the lands of Aboriginal peoples who are the custodians of this land and acknowledge and pay our respects to their Elders, past and present.

Plant material and extraction
Leaves of Eremophila glabra (R.Br.)Ostenf.(Scrophulariaceae) were collected from several individual plants in August 2017 at Lake Kerrylyn, North of Mt Methwin, Western Australia, Australia (25 02 42.3 S; 120 44 30.5 E) and identified by Dr Bevan Buirchell.The material was stored at − 20 • C until extraction.A voucher specimen has been deposited at the University of Melbourne Herbarium, Melbourne, Victoria, Australia (accession number MELUD127890a).To extract the leaf resins, the leaf material (48.0 g) was submerged in 200 mL of acetonitrile and shaken for 10 min (Ratek, Knox City, VIC, Australia).The extract was then filtered using a glass funnel, and the filtrate was dried in vacuo at 45 • C using a IKA RV10 rotary evaporator (IKA Werke, Staufen, Germany).The dried matter (2.41 g) was re-dissolved into a small volume of methanol, transferred to amber vials, dried under a stream of nitrogen gas, and then stored at − 20 • C until use.

α-Glucosidase and PTP1B assays
The α-glucosidase inhibition assay was performed at 28 • C using 96 ) ⋅100% The PTP1B inhibition assay was performed at 25 • C using 96 well microplates.A buffer consisting of 50 mM Tris, 50 mM bis-tris and 100 mM NaCl was adjusted to pH 7.0 with phosphoric acid.All analytes were dissolved in 18 μL DMSO and 52 μL 3.46 mM EDTA solution, followed by the addition of 60 μL substrate solution (1.5 mM p-NPP and 6 mM DTT) to each well.The plates were incubated for 10 min at 25 • C, and 50 μL 0.001 μg/μL PTP1B solution were added to each well.The absorbance was measured at 405 nm every 30 s for 10 min, and the percentage PTP1B inhibition was calculated as described for α-glucosidase.

High-resolution inhibition profiling of crude extract
Analytical-scale HPLC separation and micro-fractionation of the crude E. glabra extract was performed using an Agilent 1200 system (Santa Clara, CA, USA) consisting of a G1367C high-performance auto sampler, a G1311A quaternary pump, a G1322A degasser, a G1316A thermostatted column compartment, a G1315C photodiode array detector and a G1364C fraction collector, controlled by Agilent Chem-Station software ver.B.03.02.The dried crude extract was dissolved in acetonitrile to a concentration of 25 mg/mL for α-glucosidase inhibition profiling and 20 mg/mL for PTP1B inhibition profiling, and 8 μL was injected and separated by a reversed phase Phenomenex Luna C 18 (2) (150 × 4.6 mm i.d, 3 μm particle size) column (Phenomenex, Torrance, CA, USA) at 40 • C. The flow rate was 0.5 mL/min and solvents A (water/ acetonitrile, 95:5, 1% formic acid, v/v/v) and B (water/acetonitrile, 5:95, 0.1% formic acid, v/v/v) were used for gradient elution: 0 min, 10% B; 10 min, 20% B; 45 min, 85% B; 55 min, 100% B; 65 min, 100% B. The 20-60 min eluate was collected into 96-well plates (2 × 176 wells), giving a resolution of 4.4 data points per min.The well contents were evaporated to dryness using a vacuum centrifuge connected to a freeze trap, and α-glucosidase and PTP1B inhibition profiling was performed for each plate using the methods described in 4.3.The inhibition percentages were plotted against respective chromatographic retention times, constructing a high-resolution α-glucosidase and PTP1B inhibition profile (biochromatogram) for the crude extract.

HPLC-PDA-HRMS analysis
All HPLC-PDA-HRMS analyses were performed using an analyticalscale Agilent 1260 HPLC system (Agilent Technologies, Palo Alto, CA, USA), consisting of a G1329B autosampler, a G1311B quaternary pump with build-in degasser, a G1316A thermostatted column compartment and a G1316A photodiode array detector.Separations were performed at 30 • C on a Phenomenex Luna C 18 (2) reversed-phase column (150 × 4.6 mm i.d., 3 μm particle size, 100 Å pore size (Phenomenex, Torrance, CA, USA)), with a flow rate of 0.5 mL/min.Eluent A consisted of water: M.J. Petersen et al. acetonitrile (95:5,v/v), and eluent B of water:acetonitrile (5:55, v/v), both acidified with 0.1% formic acid.The crude extract was separated using an elution gradient profile as described in 4.4 and pure compounds were eluted with an elution gradient profile from 0 to 100% B in 25 min.The eluate was connected to a T-piece splitter directing 1% of the eluate to a Bruker micrOTOF-Q mass spectrometer equipped with an electrospray ionization (ESI) interface (Bruker Daltonik, Bremen, Germany).Mass spectra were acquired in positive ionization 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.Chromatographic separation and mass spectrometry were controlled using Hystar ver.3.2 software (Bruker Daltonik, Bremen, Germany).

NMR acquisition and analysis
Purified compounds were dissolved in either DMSO-d 6 , methanol-d 4 or chloroform-d and transferred to 1.7-mm NMR tubes.All NMR experiments were acquired with a Bruker Avance III 600 MHz NMR spectrometer ( 1 H operating frequency 600.13 MHz) equipped with a Bruker SampleJet autosampler and a 1.7-mm TCI cryoprobe (Bruker Biospin, Karlsruhe, Germany).Acquisitions were recorded with temperature equilibration to 300 K, optimization of lock parameters, gradient shimming and setting of receiver gain, controlled by IconNMR ver.4.2 (Bruker Biospin, Karlsruhe, Germany).The 1 H NMR spectra were acquired with 30-degree pulses and 64k data points, a spectral width of 20 ppm, 1.0 s relaxation decay, and an acquisition time of 2.72 s.NOESY and DQF-COSY 2D data were acquired with a spectral width of 12 ppm and 2k × 512 data points.HSQC spectra were acquired with spectral widths of 12 ppm and 170 ppm for 1 H and 13 C, respectively, collecting 2k × 512 data points with a relaxation decay of 1.0 s.HMBC spectra were acquired with spectral widths of 12 ppm and 240 ppm for 1 H and 13 C, respectively, collecting 2k × 512 data points with a relaxation decay of 1.0 s.All NMR data were processed using Topspin (ver.3.6.0,Bruker Biospin).

ECD calculation and ECD data acquisition
To reduce the computational cost, the conformational search and DFT calculations of 5a, 7, and 8a were conducted for model structures 5a-, 7-and 8a-model (Fig. 2) for each of them by cutting off the terminal structural unit from C-14 to C-17 of the side chain.Conformational searches were performed with the Molecular Merck force field static (MMFFs) using MacroModel interfaced to Schrödinger Maestro 11.9 (Schrödinger, LLC, USA) (Habgood et al., 2020).The resultant conformers within 5.02 kcal/mol of relative energy were further optimized using Density Functional Theory (DFT) by locating minima on the potential energy surface as provided by the B3LYP functional with the 6-31G (d,p) basis set (Pescitelli and Bruhn, 2016).The effect of acetonitrile solvent was incorporated using the polarizable continuum model in its integral equation formalism (IEF-PCM) (Tomasi et al., 2005).Subsequent DFT calculations were carried out for each optimized conformer accounting for more than 1% of the Boltzmann population using the Gaussian 16 program package (Frisch et al., 2016).Oscillator strengths and rotatory strengths of the first 60 excited states of each conformer were calculated using time-dependent (TD)DFT with the CAM-B3LYP functional and the 6-31G (d,p) basis set, and IEF-PCM for the solvent effects of acetonitrile.The final calculated ECD spectra were obtained after Boltzmann averaging the spectra of all identified conformers.The final calculated ECD spectra were exported using SpecDis software ver.1.71 (Berlin, Germany) (Bruhn et al., 2017) with a half-bandwidth of 0.20 eV for model structures of 5a and 7 and 0.25 eV for that of 8a.Experimental ECD spectra of 5a, 7 and 8a were acquired on a JASCO J-1500 CD spectrometer (JASCO, Tokyo, Japan) in acetonitrile in a quartz cuvette (1 cm path length).Each acquisition was performed at 25 • C using the following parameters: wavelength 190-400 nm, scan speed 50 nm/min, digital integration time 4 s, bandwidth 5.0 mm, 5 accumulations.

Fig. 1 .
Fig. 1.Dual high-resolution α-glucosidase/PTP1B inhibition profile of the crude extract of E. glabra leaf resin.The analytical-scale HPLC profile recorded at 254 nm is shown at the top with isolated fractions/compounds marked by numbers.The corresponding high-resolution α-glucosidase (red) and PTP1B (green) inhibition profiles are shown below.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3 .
Fig. 3. Key COSY (bold), HMBC (black, H to C), and ROESY (red) correlations used for structure elucidation of the previously unreported serrulatane-type diterpenoids isolated from E. glabra.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
a Spectral data acquired in methanol-d 4 .bSpectral data acquired in chloroform-d.c13 C NMR shifts and assignments were based on 2D HSQC and HMBC.M.J.Petersen et al.

Table 3
PTP1B and α-glucosidase inhibitory activities of the isolated compounds at 200 or 100 μM.microplates.A phosphate buffer consisting of 0.066 M Na 2 HPO 4 and 0.034 M NaH 2 PO 4 •H 2 O with 0.02% NaN 3 was adjusted to pH 7.5 with NaOH.All analytes were dissolved in buffer containing 10% DMSO, and 5 μl of a 2 U/mL α-glucosidase were added to each well together with 75 μL phosphate buffer.The microplates were incubated for 10 min at 28 • C, followed by the addition of 20 μL substrate solution (10 mM p-NPG) to a final volume of 200 μL pr.well.The absorbance at 405 nm was well