Article Cite This: J. Org. Chem. 2017, 82, 13313−13323
pubs.acs.org/joc
Catalyst-Controlled Stereoselective Synthesis Secures the Structure of the Antimalarial Isocyanoterpene Pustulosaisonitrile‑1 Andrew M. White,† Kathy Dao,‡ Darius Vrubliauskas,‡ Zef A. Könst,‡ Gregory K. Pierens,∥ Attila Mándi,⊥ Katherine T. Andrews,∇ Tina S. Skinner-Adams,∇ Mary E. Clarke,∇ Patrick T. Narbutas,† Desmond C.-M. Sim,† Karen L. Cheney,§ Tibor Kurtán,⊥ Mary J. Garson,*,† and Christopher D. Vanderwal*,‡ †
School of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane, QLD 4072, Australia 1102 Natural Sciences II, Department of Chemistry, University of California, Irvine, California 92697-2025, United States ∥ Centre for Advanced Imaging, The University of Queensland, Brisbane, QLD 4072, Australia ⊥ Department of Organic Chemistry, University of Debrecen, Debrecen, Hungary ∇ Griffith Institute for Drug Discovery, Griffith University, Nathan, QLD 4111, Australia § School of Biological Sciences, The University of Queensland, Brisbane, QLD 4072, Australia ‡
S Supporting Information *
ABSTRACT: Three new isocyanoditerpenes (5−7) have been characterized from Australian specimens of the nudibranch Phyllidiella pustulosa. The planar structure and (3R,6S,7R) absolute configuration of pustulosaisonitrile-1 were deduced by spectroscopic analyses at 900 MHz informed by molecular modeling, DFT calculations, and computational NMR chemical shift predictions and by comparison of experimental electronic circular dichroism (ECD) data with TDDFT-ECD calculations for the truncated model compound 8. A catalyst-controlled enantio- and diastereoselective total synthesis of the two most likely diastereomeric candidates for the structure of 5 solidified its (3R,6S,7R,10S,11R,14R) absolute configuration. Three individual enantioselective methods provided stereochemical control at key positions, permitting an unambiguous final structural assignment. Isocyanide 5 and synthetic diastereomers 5a and 5c showed activity against Plasmodium falciparum malaria parasites (IC50 ∼1 μM).
■
chiral units is rendered challenging by “insulating units” devoid of stereochemical information, synthesis can be used proactively to solve these difficult problems.4 As we describe herein, this situation arose in our work to fully elucidate the absolute configuration of the novel isocyanoterpene pustulosaisonitrile-1 (5, Figure 1), which we recently isolated from nudibranchs. Nudibranchs (Mollusca, Gastropoda, Opisthobranchia, Nudibranchia) are small carnivorous molluscs that lack the protective shell of other gastropods. Instead, many nudibranchs use secondary metabolites for chemical defense purposes.5 Chemical studies of species within the Phyllidiidae family suggest that they graze on sponges, sequestering sesquiterpene or diterpene isonitriles,6−9 and related isothiocyanates,7,8 thiocyanates,7,9 and formamides8 into their own tissues.5,9,10 Herein, we report the characterization, synthesis, and in vitro antiplasmodial activity of pustulosaisonitrile-1 (5) isolated from
INTRODUCTION The large isocyanoterpene (ICT) family of natural products displays a wealth of structural diversity and biological activity, including potent in vitro activity against drug-sensitive and drug-resistant Plasmodium falciparum malaria parasites.1,2 Submicromolar anti-Plasmodium IC50 activities are demonstrated by very different ICT structural types (Figure 1); however, the most potent compounds are generally fused or bridged polycyclic molecules, with two isonitrile groups.2 In most cases, full structural elucidation of these complex molecules has been relatively routine, with some compounds appropriate for X-ray crystallographic studies, and most others amenable to topological and stereochemical revelation by modern NMR methods, owing to their rigid polycyclic scaffolds. Synthesis continues to play an important role in the structural determination of complex natural products, particularly in a retrospective way after synthesis of the target points to errors in the originally reported structure.3 In select cases, when the relay of stereochemical relationships among distinct © 2017 American Chemical Society
Received: September 25, 2017 Published: November 10, 2017 13313
DOI: 10.1021/acs.joc.7b02421 J. Org. Chem. 2017, 82, 13313−13323
Article
The Journal of Organic Chemistry
Figure 1. Structures of representative antiplasmodial isocyanoterpenes 1−4 with published IC50 values against P. falciparum lines W2, FCR-3, or Dd22 and of the newly isolated pustulosaisonitriles 5−7.
specimens of Phyllidiella pustulosa collected in South East Queensland, Australia. The structure was examined by highfield NMR with the relative configuration probed by 2D NOESY spectra informed by computational studies, including NMR chemical shift calculations, and by comparison with data for two minor diterpenes 6 and 7. The absolute configuration of the 3,3,6-trisubstituted cyclohexene unit of 5 was determined by comparison of experimental electronic circular dichroism (ECD) data with TDDFT-ECD calculations. Since the low amount of available sample and considerable conformational freedom of 5 did not allow a vibrational circular dicroism (VCD) study, verification of the absolute configuration of 5 required a fully stereocontrolled total synthesis of two candidate diastereomers via application of several methods of asymmetric catalysis. The antiplasmodial activity of 5 was compared to that of other ICTs, including the synthetic diastereomer of 5.
diterpene isolated with a nitrogenous substituent. Isonitriles 6 and 7 are variations of the amitorine (prenylbisabolene) skeleton recently reported from a Lithistid sponge.18 Structural Determination via Spectroscopic and Computational Methods. Pustulosaisonitrile-1 (5) was isolated as a colorless oil from the P. pustulosa extract by NP HPLC (10% EtOAc in hexanes). The compound was identified as an oxygenated diterpene isonitrile from the HRESIMS quasimolecular ion (m/z 338.2452 [M + Na]+) corresponding to the molecular formula C21H33NO. The 1H NMR spectrum (Table 1) identified the presence of two alkene signals at δH 5.63 (dd) and 5.66 (br d) as well as an oxymethine at δH 3.71 (d). Five methyl groups were observed with a methyl adjacent to the isonitrile at δH 1.47 (t, 1JNC = 1.9 Hz), a methyl attached to an oxygenated carbon at δH 1.31 (s), two methyl singlets at δH 1.04 (s) and 0.98 (s), and finally a methyl doublet at δH 0.81. The 13C NMR spectrum (Table 2) identified 21 signals including alkene signals at δC 133.7 (d) and 130.0 (d), an isonitrile carbon at δC 152.7, and two carbons adjacent to an oxygen at δC 86.9 (q) and 86.2 (d). We found that 900 MHz NMR spectra were essential for characterization work, but even at this very high-field strength, overlapped signals were still present in the proton spectra and complicated data interpretation. The gradient-selected COSY (gCOSY) spectrum identified the methyl-substituted propylene chain linking the cyclohexene ring to the 1,3,3-trimethyl-7oxabicyclo[2.2.1]heptane ring (Figure 2a). The planar structure of the cyclohexene ring was established by the following HMBC correlations: Me-19 to C-7 and C-6; H-6 to C-7, C-1, C-2, and C-5; and Me-20 to C-2, C-3, and C-4, while the oxabicycloheptane ring was established by HMBC correlations from Me-18 to C-10, C-11, and C-12, from Me-16 and Me-17 to C-10, C-14, and C-15, and from H-14 to C-12 and C-13 (Figure 2a). The relative configuration of the oxabicycloheptane ring was informed by 1H NMR coupling values (3JH‑14/H‑13b 5.4 Hz, 3JH‑13a/H‑12a 8.9 Hz, and 3JH‑13a/H‑12b 4.8 Hz) that were all typical for a bicyclo[2.2.1]heptane ring system.19 In a NOESY spectrum, correlations between H-10/H-12a, H-12a/H-13a, and H-13a/Me-17 positioned these protons on the same face of the bicyclic ring, while on the opposite face there were
■
RESULTS AND DISCUSSION Isolation of Pustulosaisonitriles and Related Isocyanoterpenes from P. pustulosa. Specimens of P. pustulosa were collected on SCUBA from Gneerings Reef (Mooloolaba, Queensland) in March 2013. Seven individuals were finely chopped and extracted with acetone, and the extracts were concentrated under a vacuum. The aqueous suspension was partitioned with Et2O, and the resulting extract was further partitioned by normal-phase (NP) flash column chromatography (hexane/CH2Cl2/EtOAc/MeOH), after which the fractions were subjected to GC/MS and 1H NMR analysis. Subsequent NP HPLC revealed the new compound pustulosaisonitrile-1 (5) together with six known terpenes (acanthene B,11 halichonadin C,12 axisonitrile-3,13 1-isocyanoaromadendrane,14 10-isocyano-4-amorphene,15 and 2-isocyanopupukeanane16) and some known ketosterols.7b A second batch of four nudibranchs from the same location provided a further sample of 5 together with pustulosaisonitrile-2 (6), while a third collection provided pustulosaisonitrile-3 (7) in addition to 5. Pustulosaisonitrile-1 (5) represents a structural variation of the obtusane skeleton, so far found only in metabolites from algae and their sea hare predators,17 and is the first such 13314
DOI: 10.1021/acs.joc.7b02421 J. Org. Chem. 2017, 82, 13313−13323
Article
The Journal of Organic Chemistry Table 1. 1H NMR (CDCl3) Data of Compounds 5−7a 5b
position 1 2 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19 20
6c
5.63, 5.66, 2.00, 1.97, 1.71, 1.37, 2.14, 1.47, 1.30, 1.08, 1.37, 1.18, 1.14, 1.49, 1.44, 1.90, 1.68, 3.71,
dd (10.3, 1.9) br d (10.3) m m m m m (9.8, 5.6, 5.6, 1.9) m m m m m m m ddd (12.4, 11.8, 4.8) ddd (12.4, 8.9, 4.8) m d (5.4)
5.62, 5.64, 1.99, 1.97, 1.70, 1.37, 2.14, 1.50, 1.34, 1.17, 2.03, 1.94, 5.14, 2.14,
0.98, 1.04, 1.31, 0.81, 1.47,
s s s d (6.9) t (1.9)
1.29, 1.25, 1.61, 0.80, 1.46,
d (10.4) d (10.4) m m m m m m m m m m br t (7.2) md
7c 5.64, 5.65, 2.01, 1.99, 1.71, 1.38, 2.15, 1.52, 1.39, 1.21, 2.08, 2.00, 5.22, 3.01,
br d (10.3) br d (10.3) m m m m m m m m m m br t (7.0) sd
2.29, 2.13, 0.90, 0.90, 1.61, 0.81, 1.47,
br d (6.9)d m d (6.7) d (6.7) s d (6.9) t (1.9)
1.61, md 2.69, t (6.3) s s s d (6.9) t (1.8)
Figure 2. Key COSY and HMBC correlations for pustulosaisonitrile-1 (5) and NOESY correlations for the 1,3,3-trimethyl-7oxabicyclo[2.2.1]heptane ring.
2; when the H-1 and H-2 signals were simultaneously decoupled, the H-6 signal simplified to a doublet of triplets with one large coupling (J = 9.8 Hz) and two smaller couplings each of 5.6 Hz. The 9.8 Hz value was suggestive of axial−axial coupling, thereby placing the side chain in the conformationally preferred equatorial orientation. The assignment of individual H-4 and H-5 protons was aided by the W coupling detected between H-1 and H-5a and between H-2 and H-4b in the gCOSY spectrum, which suggested that H-4b and H-5a were equatorially disposed. A NOESY spectrum then revealed nOes from H-6 to H-1, H-4a, and H-5a, but the overlapping 1H NMR signals for Me-20, H-7, H-5b, and H-9a led to ambiguity in the interpretation of the NOESY data. Fortunately, NMR data for the minor diterpenes pustulosaisonitrile-2 (6) and pustulosaisonitrile-3 (7) aided the preliminary stereochemical investigation. Isonitrile 6, with a molecular formula of C21H33NO identical to that of 5, clearly contained the same isocyano-substituted cyclohexene ring as 5, but differed in the side chain substituent. In addition to a wellresolved secondary methyl (δH 0.80 (d)) assigned to Me-19, a vinyl methyl (δH 1.61), a gem-dimethyl group (Me-16 and Me17 at δH 1.25 and 1.29, respectively), and an alkene (δH 5.14 (br t); δC 134.3 (s) and 124.8 (d)), there was an epoxide group (δH 2.69 (t); δC 64.0 (d) and 58.3 (s)). HMBC correlations from the gem-dimethyl signals at δH 1.25 and 1.29 to the epoxide carbon signals positioned the epoxide moiety, while HMBC correlations from the vinyl Me-18 to C-10 and C-11, from the alkene H-10 to C-8, C-9, C-11, and C-12, and from H-9a/b to C-7 located the Δ10,11 double bond. W couplings detected in a gCOSY spectrum between H-1 and H-5a, and between H-2 and H-4a, established H-5a and H-4a as equatorial. The NOESY correlations from H-6 to H-1, H-4b, and H-5a, and from Me-20 to H-4a and 5b, placed H-6 and Me-20 on opposite faces of the cyclohexene ring, supporting a 3R*,6S*, rather than a 3R*,6R* configuration (Figure 2c). The NOESY correlations from H-10 to (H-12)2 and from Me-18 to (H-9)2 established a (10E) configuration. Isonitrile 7, also with a
a c
Chemical shifts (ppm) referenced to CHCl3 (δH 7.26). bAt 900 MHz. At 700 MHz. dSignal integrated to 2H.
Table 2. 13C NMR (CDCl3) Data of Compounds 5−7a position
5b
6c
7c
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 NC
133.7, CH 130.0, CH 56.5, C 36.4, CH2 20.6, CH2 39.3, CH 37.6, CH 34.4, CH2 26.0, CH2 56.5, CH 86.9, C 39.2, CH2 25.9, CH2 86.2, CH 45.5, C 23.7, CH3 26.4, CH3 19.1, CH3 16.1, CH3 29.1, CH3 152.7, C
133.4, CH 129.5, CH 56.3, C 36.3, CH2 20.3 CH2 39.1, CH 36.1, CH 34.0 CH2 25.7 CH2 124.8, CH 134.3, C 36.4 CH2 27.4 CH2 64.0, CH 58.3, C 24.9, CH3 18.7, CH3 16.0, CH3 15.8, CH3 29.0, CH3 152.7, C
133.6, CH 130.0, CH 56.6, C 36.6, CH2 20.6, CH2 39.5, CH 36.4, CH 34.1, CH2 26.3, CH2 129.7, CH 120.8, C 54.7, CH2 206.6, C 51.1, CH2 24.8, CH 22.9, CH3 22.9, CH3 16.8, CH3 16.1, CH3 29.2, CH3 150.7, C
a
Chemical shifts (ppm) referenced to CHCl3 (δC 77.16). bFrom HMBC data at 900 MHz. cFrom HMBC data at 700 MHz.
correlations between Me-18/H-12b, H-12b/H-13b, and H13b/H-14. These data established the equatorial orientation of the 2-methylpropyl side-chain linker (Figure 2b). In the cyclohexene ring of 5, the signal for H-6 was broad in appearance owing to vicinal coupling interactions to H-1, both H-5 protons, and H-7, together with long-range coupling to H13315
DOI: 10.1021/acs.joc.7b02421 J. Org. Chem. 2017, 82, 13313−13323
Article
The Journal of Organic Chemistry
The (3R,6S,7R)-8c and (3R,6R,7R)-8a stereoisomers of truncated structure 8 were selected for gas phase and solution TDDFT-ECD calculations to determine the absolute configuration at C-3 of the cyclohexene ring;25 in addition, these calculations were expected to provide further confirmation of the C-6 configuration. In 8a, the low-energy conformers prefer an equatorial C-3 Me, while for 8c, low-energy conformers with an axial C-3 methyl group were found dominant. Comparison of the ECD spectrum run in both gas phase and PCM for nhexane with the experimental data for 5 established the (3R) configuration, which in combination with the established relative configuration allowed determining the absolute configuration of 5 as (3R,6S,7R) (Figure 4). Although the
molecular formula of C21H33NO, had NMR data that matched those for the cyclohexene ring and side-chain alkene of 6. The 13 C spectrum lacked signals corresponding to an epoxide group and instead revealed a signal at δC 206.6 for an aliphatic ketone. The 1H NMR spectrum showed signals for an isopropyl group (δH 0.90, d, 6H; δH 2.13, m) adjacent to a methylene ((H-14)2 at δH 2.29 (d)) and one other methylene ((H-12)2 at δH 3.01 (s)). HMBC correlations from H-12, H-14, and H-15 to the ketone signal positioned it at C-13. All other HMBC and COSY correlations were as expected; a NOESY correlation from H-10 to (H-12)2 verified the (10E) configuration. In each of 5−7, the δH value for H-6 was either δH 2.14 or 2.15; these values closely matched the published values (∼δH 2.12) for a series of four 1,10-bisaboladien-3-ol stereoisomers of either (3R,6S) or (3S,6R) configuration. In contrast, 1,10-bisaboladien-3-ol stereoisomers with either a (3S,6S) or a (3R,6R) configuration all had chemical shift values ∼δ 2.04 for H-6.20 A further comparison of 13C data for C-6 and C-8 of 5−7 with those of the various 1,10-bisaboladien-3-ol diastereomers suggested, but did not establish conclusively, a (7R) configuration. Molecular modeling and theoretical DFT NMR calculations were next used to probe the relative configuration at C-3, C-6, and C-7 of 5−7. Recognizing that a conformational search would generate a very large number of conformers, the side chain of 5−7 was truncated with four diastereomers (Figure 3;
Figure 4. Experimental ECD spectrum of 5 and Boltzmann-averaged ECD spectra of (3R,6S,7R)-8c computed at various levels of theory for theB3LYP/6-31G(d) in vacuo conformers.
Figure 3. Candidate diastereomers for the cyclohexene portion of pustulosaisonitrile-1, -2, and -3.
sign of the single positive ECD transition was governed by the C-3 chiral center, the difference in intensities of the in vacuo computed ECD spectra for 8a and 8c also tentatively suggested (6S) configuration in accordance with the NMR results. Stereochemical Disambiguation via Catalyst-Controlled Stereoselective Synthesis. Our inability to resolve the relative configurations of the two isolated portions of pustulosaisonitrile-1 via spectroscopic means provided an opportunity to showcase the power of catalyst-controlled stereoselective chemical synthesis to solve this difficult problem. Due to the secure footing established above with respect to the relative configuration of the oxabicyclic ring system and the absolute configuration of the cyclohexene motif, we aimed to make the two diastereomers shown as 5a and 5b in Figure 5, with the assumption that one of them would display spectroscopic data that were in full accord with those obtained from the natural product. We did not expect the spectroscopic differences between these two diastereomers to be large, but with the combination of high-field NMR spectroscopy, along with HPLC analysis (if needed), we were confident that we could establish the identity of the natural product. The ECD studies on the natural product, described above, indicated that the cyclohexene fragment likely had the absolute configuration as shown in target diastereomers 5a and 5b (Figure 5a). Our catalyst-controlled, stereodivergent approach shown in Figure 5b dictated that, to access each target diastereomer, we would need to start with opposite enantiomers of farnesyl epoxide; otherwise, there was a reasonable likelihood of generating the unnatural enantiomer
8a 3R,6R,7R; 8b 3R,6R,7S; 8c 3R,6S,7R; 8d 3R,6S,7S) selected for initial modeling. A Monte Carlo conformational search (Macromodel, version 10.5, Schrödinger, LLC, New York, NY, 2014) for each of 8a−8d yielded a set of conformers that were further optimized by DFT calculations at the B3LYP/6311+G(2d,p) level with the IEFPCM model for chloroform and using Gaussian software.21 The 1H and 13C NMR chemical shifts were calculated for all four stereoisomers; although the resulting mean absolute error (MAE) was a good fit to the experimental data in all cases (1H MAE
