Patrick D.
Brown
and
Andrew L.
Lawrence
*
EaStCHEM School of Chemistry, University of Edinburgh, Joseph Black Building, David Brewster Road, Edinburgh, EH9 3FJ, UK. E-mail: a.lawrence@ed.ac.uk
First published on 29th August 2017
Covering: up to 2017
This review highlights why careful consideration of the biosynthetic origin (the how) and the biological function (the why) of a natural product can be so useful during the determination of its structure. Recent examples of structural reassignments inspired by biosynthetic and functional insights will be presented. This review will demonstrate the importance of viewing the origin, structure and function of a natural product as intertwined threads of a single story, best viewed as a whole rather than as discrete topics.
When faced with a natural product it is prudent to carefully consider how and why it exists. This will typically include some, or all, of the following steps:
(1) identification of likely biosynthetic building blocks.
(2) Proposal of feasible metabolic pathways.
(3) Direct comparison to biosynthetically related structures.
(4) Consideration of any known or proposed functional roles.
If nothing else, these relatively straightforward steps can serve as a reassuring check that any proposed structure, particularly if novel, makes logical sense. Obviously, this type of thought process does not provide hard evidence regarding a proposed structure, but it often leads to important insights and realizations. Then, if warranted, biomimetic synthetic studies or direct biosynthetic studies can be undertaken to probe the structure and origin of the natural product(s) further.
Fig. 2 Che's proposed structure for cytosporolide A alongside known natural products used to assist the structural assignment. |
The presence of this unusual and highly strained structural feature led George and Spence to further investigate the reported characterization data.9 Comparison of the key C(8) chemical shift in cytosporolide A to the C(8) positions of the caryophyllene-derived natural products guajadial (4) (84.3 ppm)10 and psidial A (5) (88.0 ppm)11 suggested that a similar 6-membered aryl ether ring could account for the observed 13C NMR data (Scheme 1a). George speculated that a biosynthetic pathway analogous to that proposed by Lee for guajadial (4) and psidial A (5) (see Scheme 1a)12 could produce an alternative structure for cytosporolide A (1b), which might adequately account for the reported characterization data. Thus, a [4 + 2] cycloaddition between fuscoatrol A (3) and an ortho-quinone methide (6), derived from the known fungal metabolite CJ-12,373,13 would give structure 1b (Scheme 1b).
Scheme 1 (a) Previously proposed biosynthesis of guajadial and psidial A. (b) George's biosynthetic proposal and proposed structural revision for cytosporolide A. (c) George's model biomimetic study. |
To test this biosynthetic hypothesis and structural revision, George and Spence conducted a biomimetic model study to produce the core framework of their proposed cytosporolide structure 1b.9 β-caryophyllene 8 was added to model ortho-quinone methide 7 and heating to 100 °C gave Diels–Alder adduct 9 in 53% yield as a single diastereomer (Scheme 1c). The structure of model compound 9 was confirmed by single crystal X-ray analysis and despite having opposite stereochemical configurations to the cytosporolides (e.g., 1b) at C(8), C(9) and C(16), the NMR and IR spectra of adduct 9 were found to be very similar to those reported for cytosporolides A–C. In particular, the key C(8) resonance in the model compound 9 occurred at 88.6 ppm, closely matching the cytosporolide A C(8) signal (88.0 ppm). George's proposed structural revision of cytosporolide A (1b) was subsequently confirmed by Takao and co-workers, in their 2015 total synthesis of cytosporolide A (1b),14 which followed a strategy analogous to George's biosynthetic hypothesis (Scheme 1b).9 Takao's total synthesis provides unequivocal validation of George's proposed structural reassignment and postulated biosynthesis.
Incargranine B formulates as a dimer, containing two shikimate-derived C6C2 units, two ornithine-derived C4N units and two glucose units (Scheme 2a). Consideration of the biosynthetic origins of this previously unknown indolo[1.7]napthyridine framework 10a, however, led us to question the validity of this structural proposal. It should be stressed that our analysis of the available spectroscopic data did not lead us to question Zhang's proposal. Ostensibly, the proposed structure 10a did fit with the spectroscopic data reported for incargranine B,15 and it was only our biosynthetic considerations that revealed a problem may exist. Specifically, any potential biosynthesis of indolo[1.7]napthyridine 10a, involving ornithine-derived building blocks, would necessitate the cleavage and formation of an unusually high number of bonds. This biosynthetic puzzle led us to speculate that a more biosynthetically plausible dipyrroloquinoline structure 10b could account for the characterization data reported for incargranine B (Fig. 3).16
We proposed phenylethanoid-diamine 11 represented a reasonable biosynthetic precursor to incargranine B (Scheme 2b). Oxidative deamination of diamine 11 could give an aldehyde 12, which, following intramolecular condensation, would give an N-aryl enamine 13. Enamine 13 could then dimerize with its corresponding iminium ion 14, via a domino Mannich/SEAr (electrophilic aromatic substitution) reaction sequence (i.e., a Povarov reaction), to give our proposed alternative structure for incargranine B (10b).17
To test both our proposed structural revision and new biosynthetic speculation, we undertook a short biomimetic synthesis of incargranine B (Scheme 2c).16 The acetal protected cyclisation precursor 15 (prepared in 3 steps) was exposed to 2 M aqueous HCl to deprotect the aldehyde and induce the condensation/Mannich/SEAr reaction sequence, giving a mixture of two dimeric products 16 and 17, in 50% isolated yield. Analysis of the 1H and 13C NMR spectra for the minor product, 17, revealed key similarities with the NMR data reported for incargranine B. Thus, dimer 17 was doubly glycosylated before global deprotection afforded the target structure 10b, as the expected 1:1 mixture of diastereomers (a result of glycosidation of a racemic aglycone). Data for this mixture of diastereomers, including optical rotation, matched extremely well with that reported for the natural product. Thus it is likely that natural incargranine B also exists as a mixture of diastereomers. Our proposed structural revision was, therefore, shown to be correct and the chemical feasibility of the biosynthetic hypothesis had been demonstrated.
This all cis arrangement, which is highly unusual within the wider family of resveratrol-derived indane natural products (cf. natural products 21–24, see Fig. 5),20–23 prompted Brill and Snyder to further investigate the structures of caraphenol B and C.19 They speculated that alternative structures with trans,trans stereochemistry (19b and 20b) could adequately account for all the spectroscopic data reported for caraphenol B and C (Fig. 4).19
To test this hypothesis, Snyder and Brill set about preparing samples of both Hu's (19a, 20a) and their own (19b, 20b) proposed structures for caraphenols B and C. Hu's structure for caraphenol B (19a) was prepared in five steps from permethylated ampelopsin D (25) (Scheme 3a). X-ray analysis of a derivative of synthetic 19a confirmed the anticipated all cis stereochemistry. Analysis of the NMR data for synthetic 19a showed significant discrepancies to that reported for caraphenol B.19
Scheme 3 (a) Snyder's synthesis of Hu's proposed caraphenol B structure. (b) Snyder's total synthesis and structural revision of caraphenol B. |
Snyder's proposed trans,trans structure (19b) was accessed in seven steps from permethylated paucifloral F (26) (Scheme 3b), via formation of the trans,cis compound 27, which was epimerized to the desired trans,trans diastereomer prior to global deprotection to give caraphenol B (19b). Similar synthetic sequences allowed access to the original and newly proposed structures for caraphenol C (20a and 20b). In both cases the newly proposed structures (19b, 20b) matched the reported data for the natural products in all respects, allowing Snyder and Brill to confirm their proposed structural revisions.19 This revision of natural product stereochemistry highlights how conserved structural features within a family of biosynthetically related natural products (i.e., configuration of the core motif) can help identify potentially anomalous structures.
Calvert and Sperry proposed a biosynthetic pathway towards yuremamine (28a) featuring an initial diastereoselective coupling of the known natural product, leucorobinetinidin (29)26 with N,N-dimethyltryptamine to give the flavonoid indole 28b.25 Photochemically induced cleavage of the benzylic C–O bond could then give the para-quinone methide 30 which could be trapped via [1,6]-addition of the indole nitrogen to give yuremamine (28a) (Scheme 4).
Based on this biosynthetic hypothesis, Calvert and Sperry undertook a short biomimetic synthesis, initially of the flavonoid indole 28b, which they hoped to subsequently rearrange to give the proposed structure of yuremamine (28a) (Scheme 5). They successfully accessed the key protected leucorobinetinidin analogue 31 in three steps from acetophenone 32 and aldehyde 33 in 32% yield. They were then able to diastereoselectively couple diol 31 with N,N-dimethyltryptamine in the presence of TMSOTf, followed by a global deprotection to give the postulated key biosynthetic intermediate 28b in a 56% yield over the two steps (Scheme 5).25 Intriguingly, the NMR data for this proposed biosynthetic intermediate 28b was found to be very similar to that reported for the natural product. On conversion to the TFA salt, which is how natural yuremamine was characterized, the spectroscopic data for this postulated intermediate 28b was found to match perfectly with that reported for the natural product.25 The possibility that the material had spontaneously rearranged to give 28a was ruled out by detailed analysis of 2D NMR data and the observation of an indole NH signal in the 1H-15N HSQC spectrum. Thus, Calvert and Sperry were able to conclusively reassign the natural product structure to that of their proposed biosynthetic intermediate 28b. This was further verified by Iwasawa and co-workers who synthesized the originally proposed structure of yuremamine 28a, plus all other diastereomers, and confirmed they all differ from the natural product.27 Calvert and Sperry's serendipitous discovery provides a striking example of the benefits of pursuing biomimetic strategies in total synthesis.
Scheme 6 Other known cyclic prodigiosin natural products, which are biosynthetically derived from undecylprodigiosin, and Thomson's proposed ion-binding model. |
An ongoing interest in the synthesis and biosynthesis of cyclic prodigiosins, led Thomson and co-workers to also re-examine the proposed structure of butylcycloheptylprodigiosin (34a).32 Thomson noted that several carbocyclic prodigiosin derivatives are biosynthesized from a common precursor, undecylprodigiosin (35), via oxidative cyclization processes exclusively to C(4) of the C-ring pyrrole, or to C(5) of the A-ring pyrrole (e.g., metacycloprodigiosin (36) and methylcyclodecylprodigiosin (37), see Scheme 6).33 This highly conserved cyclisation pattern within the cyclic prodigiosin family led Thomson to speculate that cyclisation to these positions must provide an evolutionarily important function. They suggested that these cyclisation patterns provide a conformational bias between the A, B and C pyrrole rings, which allows for effective ion binding (Scheme 6). The unusual cyclisation to C(4) of the C-ring pyrrole in butylcycloheptylprodigiosin (34a) would not provide the same conformational bias, removing this presumed evolutionary function.32
Thomson and co-workers, re-synthesized the proposed butylcycloheptylprodigiosin structure 34a and compared this synthetic material to previously reported data for synthetic butylcycloheptylprodigiosin (34a) and the natural product isolated by Floss and co-workers.32 Although the NMR data for their synthetic material matched well with previously reported synthetic material, they found discrepancies to the data reported for the naturally occurring compound, which they were not able to resolve using NMR spectroscopy. As a result, they turned to a comparison of the EI mass spectra. From this they found that synthetic butylcycloheptylprodigiosin (34a) and synthetic streptorubin B (34b)34 showed characteristic differences in their EI mass spectra, with the data attributed to natural butylcycloheptylprodigiosin matching the mass spectrum of synthetic streptorubin B (34b) extremely well. On this basis, they concluded that butylcycloheptylprodigiosin (34a) was unlikely to represent a genuine natural product.32 Thomson's confirmation that butylcycloheptylprodigiosin (34a) is not a natural product provides a clear example of how consideration of the biological function of a compound can further highlight possible discrepancies in structural assignments.35 As the quantity and quality of available functional information increases we predict this will become an increasingly useful consideration for structure determinations.
The sacoglossan molluscs have long been a source of great interest to chemists, providing a range of intriguing pyrone natural products. Previous biosynthetic and biomimetic studies by Faulkner, Ireland, Clardy and Solheim had demonstrated that many γ-pyrone natural products from Tridachiella and Tridachia molluscs are derived from a common biosynthetic pathway featuring a series of photochemically mediated rearrangements.38 Ireland and Scheuer further suggested that these strongly UV absorbing γ-pyrone natural products may serve an important biological function by acting as a sunscreen, protecting the molluscs from excessive UV exposure.39
Building on this biosynthetic speculation, Moses and co-workers proposed that this family of γ-pyrone natural products could be derived from a common polyene biosynthetic intermediate 41 (Scheme 7).40 Moses proposed that tridachiahydropyrone could be derived biosynthetically from polyene 41 through a 6π electrocyclisation. Depending on whether this electrocyclisation occurred in nature under thermal or photochemical conditions, a tridachiahydropyrone structure with a trans (disrotatory cyclization, structure 38a) or a cis (conrotatory cyclization, structure 38b) relationship between the C(9) and C(4) substituents would result. A previous total synthesis of the proposed structure of tridachiahydropyrone (38a) by Perkins and co-workers had indicated that the trans structure was incorrect.41 Therefore, in light of the postulated biological function of these compounds, Moses proposed the cis structure (38b), originating from a photochemical 6π electrocyclization, was likely to represent the correct structure of tridachiahydropyrone.40a Moses also suggested that the closely related natural products tridachiahydropyrones B (39a) and C (40a) were likely to be derived from tridachiahydropyrone (38b) via a [4 + 2] cycloaddition of singlet oxygen.40b
To test this biosynthetic hypothesis and proposed structural revision, Moses and co-workers undertook a biomimetic synthesis of the tridachiadhydropyrones (38b, 39a, 40a).40 The key polyene 41 was prepared in 75% yield by a Suzuki coupling of the pyrone bromide 42 and diene boronic ester 43 (Scheme 8). Polyene 41 was then subjected to both thermal and photochemical conditions to test their biosynthetic hypothesis. Heating polyene 41 to 150 °C produced no reaction; however, on exposure to sunlight for three days the desired tridachiahydropyrone 38b was formed in 29% yield, with all data matching that reported for the natural product (Scheme 8). NOESY and single crystal X-ray analysis of synthetic tridachiahydropyrone (38b) confirmed the anticipated cis stereochemistry, allowing Moses to confidently revise the structure of tridachiahydropyrone (38b) as well as lending support to the biosynthetic hypothesis.40a
With samples of synthetic tridachiahydropyrone (38b) now available, Moses and co-workers were also able to test their biosynthetic hypothesis for tridachiahydropyrones B (39a) and C (40a). Exposure of tridachiahydropyrone (38b) to singlet oxygen gave quantitative conversion to a 4:5 mixture of two products, which matched the reported data for tridachiahydropyrones B and C. However, extensive NOE experiments on this mixture, supported by computational work on the possible product structures, indicated that only a single adduct is formed in the cycloaddition with singlet oxygen, with retention of the (E)-geometry at the C(10)–C(11) alkene (39b). The two products observed were, in fact, rotamers resulting from restricted rotation around the C(9)–C(10) bond. This observation was further confirmed through the use of variable temperature NMR experiments, allowing the reassignment of tridachiahydropyrones B and C as a single natural product 39b.40b
The success of this synthesis provides an excellent example of the broad range of benefits biosynthetic considerations can provide. Moses proposed a biosynthetic hypothesis, which linked three natural products together, and by taking into account the possible sunscreen function of these compounds suggested a structural revision of tridachiahydropyrone (38b). By investigating this hypothesis synthetically they were able to validate this proposed structural revision and serendipitously reassign the structures of tridachiahydropyrones B (39a) and C (40a) as two rotamers of oxytridachiahydropyrone (39b).40
This journal is © The Royal Society of Chemistry 2017 |