Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Jun 1.
Published in final edited form as: Nat Chem. 2024 Feb 19;16(6):1003–1014. doi: 10.1038/s41557-024-01455-7

A catalytic process enables efficient and programmable access to precisely altered indole alkaloid scaffolds

Youming Huang 1,2,*, Xinghan Li 1,2,*, Binh Khanh Mai 3, Emily J Tonogai 4, Amanda J Smith 4, Paul J Hergenrother 4,5,†, Peng Liu 3,†, Amir H Hoveyda 1,2,†
PMCID: PMC11328697  NIHMSID: NIHMS2016024  PMID: 38374457

Abstract

A compound’s contour impacts its ability to elicit biological response, rendering access to distinctly shaped molecules desirable. Modification of a natural product’s framework is a common tactic, but possible only if the compound is abundant and contains suitably modifiable functional groups. Here, we introduce a strategy for programmable and concise synthesis of precisely altered scaffolds of scarce bridged polycyclic alkaloids. Our approach is founded on a scalable catalytic diastereo- and enantioselective multicomponent process that delivers tertiary homoallylic alcohols bearing chemo- and regioselectively differentiable alkenyl moieties. We used one product to launch concise and progressively divergent syntheses of a naturally occurring indole alkaloid and its precisely expanded, contracted, and/or distorted framework analogues (average number of steps/scaffold = 7). In vitro testing shows that a skeleton expanded by one methylene in two regions is cytotoxic against four types of cancer cell lines. Mechanistic and density functional theory (DFT) studies account for several unanticipated selectivity trends.

Graphical Abstract

graphic file with name nihms-2016024-f0001.jpg

Lead discovery is expedited when distinct compound arrays can be evaluated, placing the burden on the art of small molecule synthesis (molecular weight ≤1000 g/mol)1. Based on the premise that an even mildly bioactive natural product (NP) can be a pre-validated starting point2, methods3 have been introduced for functional group and/or stereochemical alterations of such entities. These strategies are applicable, however, if a NP is available in sufficient quantities and contains a suitably reactive functional group4. In contrast, schemes designed for making precisely altered scaffolds accessible are uncommon5, and those that are available typically generate analogues that contain multiple modifications (stereochemical, functional, and constitutional). Below , we present a strategy to address this shortcoming regarding a key class of scarce and bioactive natural products.

We undertook these studies to find ways for efficiently synthesising skeletally diverse bridged polycyclic indole alkaloids (Fig. 1a). Related collections have been prepared, but the resulting analogues were either differentiated by multiple changes, most stereochemical or functional group variations6, or also marked by peripheral and constitutional (bond connectivity) adjustments7. We had several reasons for focusing on this compound class. (1) The bridged bicyclic amine scaffold can be found in myriad NPs8. While some display different types of bioactivity, for example pericine being a weakly antimalarial (+)-16-hydroxy-16,22- dihydroapparicine9, and analgesic/cytotoxic agent10, little is known about others (for example, curan11 and isobrafuedin12). (2) Many indole alkaloids are available only in minute amounts, rendering their laboratory synthesis imperative. (3) The bridged polycyclic amine framework is often devoid of a functional group that can be used for scaffold alteration, especially if the modifications are expected to be without changing, adding, or removing stereochemistry (absolute or relative). One possibility might entail oxidative cleavage of the trisubstituted alkene followed by a Baeyer-Villiger reaction to generate a ring-expanded lactone. However, as we would later confirm, amines preclude the use of oxidative procedures. Further, our intent was to generate skeletally modified structures without concomitant incorporation of any additional polar groups. We favoured skeletal variants arising through precise insertion and/or deletion of one or more methylene units. The latter factors (2 and 3) render late-stage editing unrealistic. (4) There is structural diversity among indole alkaloid NPs. Several, represented by geissoschizoline (Fig. 1a), a candidate for treating Alzheimer’s disease13, feature a transannular linkage within their polycyclic structure. Most contain an E-trisubstituted alkene while some have a Z isomer (for example, ervaticine, Fig. 1a14). (5) There are unexplored gaps in the NP frameworks. Expansion and/or contraction of different scaffold regions by one or more methylene units leads to distorted analogues, one being a fragment of vinblastine (Fig. 1a), a cancer chemotherapy agent. We surmised that there might be as-of-yet unknown frameworks that posess similarly important attributes and could be an asset in fragment-based drug discovery15.

Figure 1 |. Bridged polycyclic indole alkaloids, the foundational multicomponent process and its applicability.

Figure 1 |

Open in a new tab

a, Alkaloids are large in number and exhibit different types of bioactivity. Consequently, analogues such as those of 16-hydroxy-16,22-dihydroapparicine arising from one- or two-methylene expansion, contraction or both (distortion), would be of interest. This is evident by a comparison with fragments found in related compounds with established therapeutic capabilities, such as vinblastine, showing that these scaffolds are linked by a combination of methylene unit expansion and/or contraction at different regions of their skeletons. As these natural products are available only in minute amounts and lack an easily modifiable site, late-stage framework editing strategies are not applicable. b, A multicomponent catalytic diastereo- and enantioselective process would generate a highly functionalized product, or a primary hub (1° hub), containing readily alterable and differentiable monosubstituted olefin and a stereochemically defined trisubstituted alkenyl boronate. Progressively divergent conversion to 2° hubs and a larger number of 3° hubs can then lead to precisely modified expanded, contracted or distorted frameworks (green dot denotes an added methylene and gray dot, a deleted methylene). The existing diastereo- and enantioselective protocols, catalyzed by a copper complex and involving ketones, generate multifunctional homoallylic alcohols, but versatility can still be greater if the allylic substituent were to be more easily alterable and the alkenyl moiety were to have stereochemical identity. NP, natural product.

We chose to focus on 16-hydroxy-16,22-dihydroapparicine (Fig. 1a; NP). Considering the relative simplicity of this NP, our decision was founded on the principle that complexity should be about more than the intricacy of a single (target) or les sthan a handful of natural products. It should also be defined by the number and diversity of the collection, and the efficiency of and the precision and selectivity with which it is assembled. The only reported enantioselective synthesis of NP consists of a longest linear sequence (LLS) of 23 steps16 and can likely offer access to serviceable quantities of the molecule (5% overall yield). Nevertheless, the sequence lacks a suitable diversification point and therefore does not lend itself to efficient preparation of precisely altered scaffold synthesis. For each skeletal variant a new scheme would have to be re-negotiated.

Our idea was that a multifunctional homoallylic tertiary alcohol might be a workable point of origin or primary hub (1° hub, Fig. 1b). This densely packed fragment could be obtained through a catalytic diastereo- and enantioselective process involving an indole-substituted ketone (1a), bis(pinacolato)diboron (B2(pin)2), and a vinylallene (rac-2a). With a vinyl group at the allylic position and a stereochemically defined trisubstituted alkenyl–B(pin) moiety, both modifiable and differentiable, together with indole’s mildly nucleophilic C3, it would be possible to convert 1° hub to several 2° hubs. The latter compound would in turn be transformed to an expanded suite of specifically outfitted 3° hubs, precursors to precisely remodelled scaffolds. Through a network of overlapping pathways, minimizing step-count and obviating the need for complete route repetition, NP and various precisely altered skeletal analogues would be secured (Expanded-I–II, Double-Expanded, Contracted, and Distorted-I–II).

Several points merit brief note. We were aware that the approach might challenge the state of the art in chemical synthesis, and that method development, other than the foundational catalytic process, might be needed. Vis-à-vis diversification, our strategy represents a blend of two existing approaches: diversity-oriented synthesis (DOS)17,18,19 and diverted total synthesis (DTS)20. One type of DOS is centered on a core platform. In DTS, one or more intermediates, generated en route to the target molecule, might serve as branching points for implementing functional and/or stereochemical variations that may be accompanied by framework alterations caused by multiple connectivity changes21,22. Analogous to DTS but unlike DOS, in our approach a NP would serve as the parent skeleton. Neither DOS nor DTS has been used for programmatic implemention of precise scaffold alterations.

Results and Discussion

A suitable catalytic multicomponent process

There has been ample progress in the development of catalytic multicomponent processes involving ketones23. Notable advances include transformations that are promoted by a Cu–B(pin) or a Cu–H complex and involve monosubstituted allenes24,25, enynes26, and dienes27,28,29 as organocopper precursors. However, none of the reactions afford products that are amenable to efficient alkaloid synthesis.

We were far from confident that we would be able to identify conditions for the proposed multicomponent process to proceed efficiently or diastereo- and/or enantioselectively. 1,3-Disubstituted allenes are chiral, raining the question of whether they must be used in the enantiomerically enriched form. The only extant case pertains to 1,6-conjugate additions to α,β,γ,δ-unsaturated diesters, promoted by a N-heterocyclic carbene–copper complex30 , where a vinylallene’s enantiomeric purity turned out to be inconsequential. Mechanistic studies revealed that the Cu–allyl species derived from regioselective addition of the Cu–B(pin) complex undergoes π-allyl isomerization, engendering rapid and complete loss of stereochemical identity. Besides, high selectivity was found to be linked to a substrate’s 1,3-dicarbonyl moiety, which, according to density functional theory (DFT) studies, coordinates with the available alkali metal salt to enhance transition state organization. We also knew that, based on precedent24, and especially without having the advantages of the latter factors, with ketones high regio- and stereochemical control would be difficult to achieve. Making matters still dicier, the aforementioned method was found to be confined to aryl ketones24; additions to the less reactive aliphatic variants suffered from adventitious enolate formation.

We first evaluated the reaction of acetophenone with rac-2b (synthesized in two steps and 64% yield, Fig. 2a), and B2(pin)2 (commercially available) in the presence of the catalyst earlier found to be effective for the aforementioned 1,6-conjugate additions30. This was the copper complex generated from N-heterocyclic carbene (NHC) ligand derived from imid-1. The transformation proceeded to completion, affording 3a in 85:15 Z:E selectivity and >98:2 diastereomeric ratio (d.r.), but enantioselectivity was nominal (47:53 enantiomeric ratio, e.r.). Related complexes, such as that originating from imid-2, were slightly more discriminating (68:32 compared to 47:53 e.r.), and those generated from P,N-type ligands, such as aminophos, were only more Z-selective. With bisphos-1, optimal for the reactions between ketones and monosubstituted allenes24, 3a was generated in 83:17 Z:E and >98:2 d.r., albeit racemically. Reactions with bisphos-2–5, which contain a triarylphosphine moiety, afforded 3a with similar Z selectivity and d.r. Notably, the bisphos-5–Cu-catalyzed process afforded 3a in 92:8 e.r. (for determination of stereochemical identity, see Fig. 3a). With bisphos-6, the transformation was equally efficient and enantioselective, but the Z:E ratio was lower (see Fig. 3b and the related discussion for the importance of this finding).

Figure 2 |. Catalyst screening and the scope of the multicomponent process.

Figure 2 |

Open in a new tab

a, Screening of different types of chiral ligands began with the copper complex of N-heterocyclic carbene (NHC) derived from imid-1, which had formerly been found to be the most effective in 1,6-conjugate additions involving the same type of allenes, B2(pin)2 and acyclic dienoates; this reaction, while efficient and diastereoselective, was barely enantioselective. Similar results were obtained with other NHC ligands, aminophosphine with a P and N ligation site, as well as biaryl bisphosphines. Enantioselectivities improved with ferrocene-based bisphosphines, with bisphos-5 (in box) delivering the highest e.r. Ethyl-bridged bisphos-6 was as enantioselective as bisphos-5 gave a lower ratio of E:Z isomer (readily separable). b, The catalytic multicomponent process has considerable scope, in a single step generating densely functionalized fragments that contain chemoselectively alterable vinyl (in brown) and trisubstituted alkenyl–B(pin) (in red). A range of ketone substituents (in blue) is tolerated, including aryl moieties that are electron-dificient (3e) or electron-rich (3g-h), heteroaryl groups (4a-b), alkenyl units (5a-d), and alkyl moieties (6a-d). In all instances, the trisubstituted alkenyl isomers are readily separable; the yield shown are for the pure Z isomers. Reactions were performed under N2 atmosphere. Conversion to product (>98% in all cases), Z:E ratio, and d.r. were measured by analysis of 400 MHz 1H NMR spectra of unpurified mixtures with diphenylmethane serving as the internal standard (±2%). Yield of purified product, average over at least three runs (±5%) Enantioselectivities were determined by HPLC analysis (±1%). aWith bisphos-3; 59:41 e.r. with bisphos-5. bWith ent-bisphos-5; 39% yield, 80:20 d.r. with bisphos-5. See the supplementary information, section 4, for details. pin, pinacolato; d.r. diastereomeric ratio; e.r. enantiomeric ratio; TBS, tert-butyldimethylsilyl.

Figure 3 |. Diastereo- and enantioselective synthesis of the primary hub and some surprising observations along the way.

Figure 3 |

Open in a new tab

a, The absolute stereochemistry of products of the type in Fig. 2, which bear three key exchangeable/modifiable moieties (aryl/heteroaryl in blue; alkenyl boronate in red; vinyl unit in brown), was established by X-ray crystallography. b, Surprisingly, the chiral catalyst derived from bisphos-5, optimal for most other ketones (see Fig. 2b), was found to be minimally enantioselective when indolyl ketone 1b was a substrate. Further screening revealed that the most effective catalyst to generate the envisioned 1° hub is the copper complex derived from bisphos-2, namely, the complex that was found in the original screening studies to afford 3a in just 80:20 e.r. c, Equally surprising, as initially predicted by DFT studies and later substantiated by X-ray crystallography, the reactions with indolyl ketone 1b proceeded with the opposite sense of enantioselectivity (for mechanistic studies, see Fig. 6). d, Products derived from other vinylallenes and ketones may be used to generate functionally, as well as skeletally altered analogs. DIAD, diisopropylazadicarboxylate; pin, pinacolato; Bn, benzyl; Ar, aryl group; DME, dimethoxyethane; TBS, tert-butyldimethylsilyl; d.r., diastereomeric ratio; e.r., enantiomeric ratio. See the Supplementary Information for details, sections 4.2 and 6.

By using bisphos-5, we synthesized an assortment of products (Fig. 2b), including those derived from aryl (3a–j), heteroaryl (4a–b), alkenyl (5a–d), and alkyl ketones (6a–d), which were formed in 90:10–97:3 e.r.; Z:E selectivities varied from 76:24 to 89:11 with one exception. Curiously, reaction of cyclohexyl-substituted vinylallene afforded 3c only as the Z isomer whereas the Z:E ratio for reaction of the same allene with n-hexyl-substituted ketone to afford 6c was in the expected range (82:18 Z:E; see Fig. 6 and the corresponding discussion for further analysis). The Z and E alkenyl isomers were separable by silica gel chromatography, and the trisubstituted alkenyl boronic acids were isolated as pure Z isomers (>98%, 54–84% yield).

Figure 6 |. DFT studies offer insight regarding several unexpected selectivity trends.

Figure 6 |

Open in a new tab

a, The higher Z:E ratio for cyclohexyl-substituted allene may be due to London dispersion attraction between C–H bonds and an aryl group. b, For 3g, the 59:41 e.r. with bisphos-5 compared to the 97:3 e.r. with bisphos-3 may be due to conformational differences leading to greater steric strain in ts-3g-bisphos-5-maj. With bisphos-3, the larger aryl phophine enhances steric pressure in ts-3g-bisphos-3-min. c, The unexpected low e.r. for 1° hub formation with bisphos-5 and enantioselectivity reversal (vs. 3d) with bisphos-2 was predicted by DFT studies, and appears to be owing to π-π interactions involving the N-benzyl group. The allyl–Cu complex and the ketone associate differently in ts-1° hub-bisphos-5-maj (vs. ts-3d-bisphos-5-min). The difference between ts-1° hub-bisphos-5 complexes is reduced due to π-π interaction in ts-1° hub-bisphos-5-maj. In contrast, steric strain from the proximity of the allyl–Cu and arylphosphine in ts-1° hub-bisphos-2-min and a face-to-face attraction in ts-1° hub-bisphos-2-maj cause enough energy difference for S,R-1° hub to be formed in 95:5 e.r. DFT studies were performed with M06/SDD-6–311+G(d,p)/SMD(THF)//B3LYP-D3/SDD-6–31G(d). See the supplementary information, section 13, for details. Bn, benzyl.

Reactions of relatively sizeable aryl and heteroaryl group ketones afforded homoallylic alcohols in >98:2 d.r. Whereas d.r. was lower with alkenyl (5a-c) and aliphatic ketones (6a-c), complete diastereoselectivity was observed for a bulkier enone (5d). With ent-bisphos-5 we obtained 6d in 54% yield (pure Z isomer) and 95:5 d.r.; this represented the matched catalyst-substrate pairing, since with bisphos-5 the transformation was lower yielding and less stereoselective (80% vs. >98% conv., 39% yield (pure Z), 80:20 d.r.).

Two additional points merit note: (1) Contrary to most instances when bisphos-5 was optimal, the reaction affording ortho-methoxyphenyl-substituted 3g was scarcely enantioselective (59:41 e.r.; 81% yield (pure Z), >98:2 d.r.). Further screening led us to the finding that, remarkably, bisphos-3 is superior in this case: 3g was formed with far higher enantioselectivity (97:3 e.r.; 64% yield (pure Z), >98:2 d.r.; see Fig. 6 and the related discussion for further analysis). Reaction with ortho-fluorophenyl ketone afforded 3i and 3j in 92:8 e.r. when bisphos-5 was used. (2) Transformations involving the smaller methyl-substituted vinylallene (compared to 2b), relevant to the planned indole alkaloids syntheses, were efficient and selective. Conversion of the C–B bond to a C–H bond was efficient (see the supplementary information, section 5.1, for details); for example, the disubstituted alkene derived from 3d was isolated in 82% yield (pure Z). Functionalization at the alkenyl–B(OH)2 site, generating C–aryl, C–alkenyl, or C–alkyl bonds, proceeded stereoretentively (see the supplementary information, section 5.2, for details). One example is the conversion of 3d to allylic alcohol 7a (Fig. 3a)31, the precursor to 7b, the X-ray structure of which confirmed the stereochemical identity of the products of the catalytic process (see the supplementary information, section 4.2, for details).

Preparation of 1° hubs

To synthesize the envisioned 1° hub (see Fig. 1b), we examined the reaction of 2-indolyl ketone 1a (purchasable), B2(pin)2 (purchasable), and vinylallene rac-2a (prepared in two steps, 50% yield) under the foregoing conditions (bisphos-5). There was no detectable conversion after 30 hours, leading us to suspect that the indole NH might cause catalyst inhibition. We therefore used the N-benzyl-protected derivative 1b (Fig. 3b; prepared in one step and 77% yield), which reacted efficiently and with complete diastereoselectivity. To our surprise and disconcertment, however, the 1° hub was formed in 53:47 e.r. Re-examination of alternative ligands (see Fig. 2a) led us to identify a quick solution in the form of using bisphos-2. That is, the catalyst that promoted addition to acetophenone in only 80:20 e.r. (see Fig. 2a), was, in this particular case, the most effective. These findings underscore the view, articulated nearly two decades ago32, that the most effective catalyst in a particular case can be different from what is marked as “optimal” based on methodological studies, pointing to the importance of the availability of a collection of catalyst candidates – the same diversity-based logic being applied to drug lead discovery.

We thus isolated 1° hub in 80% yield, >98:2 Z:E ratio, >98:2 d.r., and 95:5 e.r., along with the easily separable E-isomer (1° hub(E)) in 18% yield, >98:2 E:Z ratio, >98:2 d.r., and 88:12 e.r. Curiously, the latter two products were generated with the opposite sense of enantioselectivity compared to the other ketone substrates (see Fig. 2 and Fig. 3a). This was also unanticipated, to the extent that we did not suspect a stereochemical reversal until warned by the results of DFT studies (see Fig. 6 and the associated discussion). The stereochemical identity of 1° hub was verified by the X-ray structure of lactone 8 (Fig. 3c), confirming the validity of the DFT prediction (see the supplementary material, section 6.2, for further details). A more efficient way of obtaining 1° hub(E), applicable to preparation of Z-NP and the related skeletally modified analogs, would be by using the less Z-selective bisphos-6. Under these latter conditions, 1° hub(E) was formed in 47% yield and 94:6 e.r.

The reactions are scalable: 3.0 grams of 1b was converted to 3.6 grams of 1° hub and 0.81 gram of 1° hub(E). Transformations of 1b with other allenes were similarly efficient and selective, as exemplified by silyl ether 4c (Fig. 3d), but additions to ketones with less differentiable substituents were less enantioselective (4d). Such compounds may be viewed as alternative primary hubs that can be transformed to modified frameworks that contain peripheral/stereochemical alterations, or be made suitable for covalent ligation to another entity.

Divergent syntheses of 2° and 3° hubs

We converted 1° hub to 2° hubs-1–2 (Fig. 4a), featuring a one- and a two-methylene extension at the alkenyl boronic acid site, respectively. This was accomplished by single31 and double33 Matteson homologation, respectively, followed by oxidation (80% and 61% yield, respectively). As the first tertiary hub, we opted for a structure that, other than the missing indolyl linkage, is identical to NP. Conversion of 2° hub-1 to 3° hub-1 was effected in five steps and 43% yield. After silyl ether formation, the resulting alcohol was treated with zirconocene chloride hydride (Cp2ZrHCl) followed by hydroxylamine-O-sulfonic acid (HAS). This led to the corresponding α-primary NH2-amine34, which was converted to sulfonamide 9 (50% yield for 3 steps). Unmasking of the primary alcohol set the stage for the first cyclization, which furnished 3° hub-1 (86% yield for 2 steps).

Figure 4 |. Progressively divergent conversion of the primary hub to secondary and tertiary hubs.

Figure 4 |

Open in a new tab

a, 1° hub, containing two key modifiable moieties (alkenyl boronate in red; vinyl unit in brown) was converted to two 2° hubs, one an allylic and the other a homoallylic alcohol. 2° hub-1 was then transformed to various tertiary hubs. 3° hub-1 pertains to the natural framework, 3° hub-2 is a contracted ring; 2° hub-2 was similarly converted to 3° hubs-3–4. 3° hub-5 was generated by a five-step route that included a catalytic ring-closing metathesis and hydrogenation step. b, Hydroxy-assisted regioselective hydroamination, affording the α-secondary NH2-amine product, may be used for contraction of the bridged bicyclic amine. Bn, benzyl; d.r., diastereomeric ratio; e.r., enantiomeric ratio; r.r. regioisomeric ratio; Ns, 2-nitrobenzenesulfonyl; DIAD, diisopropylazadicarboxylate; Cp, cyclopentadiene; cod, cyclooctadiene; dppe, 1,2-bis(diphenylphosphino)ethane; Mes, mesityl or 2,4,6-trimethylphenyl; Ind(Bn), benzyl-protected indole. See the Supplementary Information for details, sections 7–8.

The ensuing single-methylene contraction (see Contracted, Fig. 1b) proved to be more complicated. The common approach for one-carbon excision, oxidative cleavage of the monosubstituted alkene, could not be implemented. Chemoselectivity issues aside, the presence of an indole, an amine, and/or a C–B bond was problematic. We reasoned that a solution might entail direct formation of a C–N bond at the internal carbon of the monosubstituted olefin, and subsequent use of the secondary amine for cyclization. Ideally, this would be accomplished by a hydroamination that would proceed with complementary regioselectivity, but such a method, particularly with a monosubstituted aliphatic olefin substrate35,36, is yet to be developed. We instead entertained a different possibility. In the foregoing synthesis of 3° hub-1, hydroamination of a monosubstituted alkene (see 2° hub-1 → 3° hub-1) was carried out by sequential treatment with Cp2ZrHCl and HAS. If the same were to be performed with an unmasked primary alcohol, we reasoned, an alkoxyzirconocene hydride might be generated first (2° hub-1 → i → ii, Fig. 4b). Ensuing intramolecular zirconocene–hydride addition via a more favourable six-membered zirconacycle (ii → iii)37,38, and trapping of the C–Zr bond with HAS would deliver an α-secondary NH2-amine (iv). In the event, subjection of 2° hub-1 to Cp2ZrHCl followed by HAS afforded 10 in 77% yield, as a single regioisomer and in 95:5 d.r. (after silica gel chromatography). Sulfonamide formation and cyclization furnished 3° hub-2 (67% yield from 2° hub-1, 3 steps).

Our next goal was to convert 2° hub-2 to different tertiary hubs (Fig. 4a), scaffolds wherein a bridge would be expanded by a single methylene group. The first, 3° hub-3, was prepared in five steps and 42% yield as before (see 2° hub-1 → 3° hub-1). Hydroxy-assited hydroamination of 2° hub-2 afforded α-secondary NH2-amine 11 with somewhat lower regio- and diastereoselectivity (77:23 regioisomeric ratio (r.r.) and 88:12 d.r.), probably on account of a more distal coordinating unit (compared to allylic alcohol in 2° hub-1). We isolated 11 in 61% yield, >98:2 r.r. and 95:5 e.r. (after chromatography) and transformed it to 3° hub-4 by the aforementioned sulfonamide formation/cyclization (52% yield from 2° hub 2, 3 steps).

The concluding tertiary hub (3° hub-5, Fig. 4a) contained two additional methylenes, one within each of the carbon bridges. We obtained this analogue by masking the alcohols and then selectively deprotecting the less hindered silyl ether, followed by conversion to triene 12 (73% yield, 3 steps). Ring-closing metathesis and site-selective hydrogenation of the cyclic alkene furnished 3° hub-5 (44% yield from 2° hub-2, 5 steps). With five tertiary hubs in hand, we were ready to convert them to NP and a collection of precisely modified scaffolds.

Syntheses of NP and altered scaffolds

Synthesis of the natural framework (Fig. 5a) began by removal of the nosyl and benzyl groups in 3° hub-1. The ensuing Pictet-Spengler reaction delivered ent-NP (the non-natural enantiomer; 66% yield). We accessed NP by using ent-bisphos-2 (also commercially available), and ent-Z-NP was prepared similarly from 3° hub-1(Z) (originating from 1°hub(E)). The same can be applied to every tertiary hub’s corresponding Z isomer.

Figure 5 |. Synthesis of the natural product and precisely altered scaffolds and in vitro testing.

Figure 5 |

Open in a new tab

(Green dot denotes an added methylene and gray dot, a deleted methylene; alterations derived from the indolyl moiety in blue, the alkenyl boronate in red, and from the vinyl unit in brown). a, The NP and Z-NP were obtained in two steps from 3° hub-1, as was α-chloro-amide 13, precursor to four expanded scaffolds. b, It took seven steps to transform 3° hub-2 to a contracted (ent-e2) and two distorted scaffolds (ent-16 and ent-e3). c–e, 3° hubs 3–5 were converted to expanded (ent-e4), distorted (ent-e5), and a doubly expanded (ent-e6) framework. f, Cytotoxicity (72-hour IC50 values, μM) against four cancer cell lines (Alamar Blue assay, n = 3 biological replicates, error is standard error of the mean (SEM)) underscores the importance of having access to precisely altered skeletal analogs. Oxaliplatin was used as a quantitative dead control. Z-e1 caused some cell death in two of the cell lines while e6 exhibited activity for all four lines. Conditions: A: 1. Na/NH3, THF, 1 h, −78 °C. 2. PPTs, HCHO, THF, 24 h, 22 °C. A’: Same as A, except step 2 in DMF. B: 1. TFA, CH2Cl2, 2 h, 22 °C. 2. Na/NH3, t-BuOH, 1 h, −78 °C. 3. ClCH2COCl, Et3N, 1 h, 22 °C. C. hn (254 nm), Na2CO3, EtOH/H2O, 15 min. D: LiAlH4, THF, 5 h, 60 °C. E. 1. (n-Bu)4NF, THF, 0.5 h, 22 °C. 2. Na/NH3, THF, 1 h, −78 °C. 3. PPTs, HCHO, MeCN, 24 h, 22 °C. See the supplementary information, sections 9 and 12, for details. ent, non-natural isomer; Bn, benzyl; e, edited scaffold; TFA, trifluoroacetic acid; Ns, 2-nitrobenzenesulfonyl; PPTs, pyridinium p-toluenesulfonate. MDA-MB-231: human breast cancer cell line; A549: adenocarcinomic human alveolar basal epithelial cell line; HCT-116: human colon cancer cell line; L1210: mouse lymphocytic leukemia cell line. IC50, half-maximal inhibitory concentration.

Conversion of 3° hub-1 to skeletons that have a methylene unit inserted between their nitrogen atom and the indole C3 (see Expanded-I, Fig. 1b) presented new challenges. The most effective solution was Witkop cyclization39, a photochemical transformation which, to the best of our knowledge, had not been applied to a substrate bearing a somewhat sensitive tertiary benzylic alcohol. As it turned out, efficient ring closure was feasible only without the tertiary hydroxy group (fast decomposition, otherwise). Photolysis of α-chloro-amide 13 (Fig. 5a), obtained in three steps from 3° hub-1 (74% yield, 96:4 d.r.), afforded ent-E-14 and ent-Z-14 (Expanded-1–2). Analogues ent-E-14 and ent-Z-14 (readily separable) contain an amide carbonyl, which is a possible point of H-bonding interaction. Amide reduction furnished the expanded, pericine-like (see Fig. 1a), bridged bicyclic amines ent-E-e1 and ent-Z-e1 (Expanded-3–4, label in italics signifies analog containing more than one alteration). While loss of the hydroxy group renders this modification less precise, other bioactive members of the alkaloid family, such as pericine (see Fig. 1a), apparicine40, and anti-neuroinflammatory geissoschizoline (see Fig. 1a) lack this unit as well. Moreover, curan (see Fig. 1a) carries a similar methyl-substituted stereogenic benzylic carbon. By and large, after a total of eight reactions, 3° hub-1 was transformed not only to NP, but also to four skeletally expanded analogs.

We transformed 3° hub-2 to ent-e2 (Contracted; 73% yield; Fig. 5b) by using the same reactions that converted 3° hub-1 to NP. Akin to synthesis of ent-E-e1 and ent-Z-e1, 3° hub-2 was converted to α-chloro-amide 15 and then ent-16 (Distorted-1, 30% yield from 3° hub-2, 4 steps), which was reduced to ent-e3 (Distorted-2, 21% yield from 3° hub-2, 5 steps). Nosyl and benzyl group removal and Pictet-Spengler cyclization turned 3° hub-3 into ent-e4 (Expanded-5; Fig. 5c) and 3° hub-4 into ent-e5 (Distorted-3; Fig. 5d) in 61% and 94% yield, respectively. The same two-step protocol, preceded by tertiary alcohol unmasking, delivered ent-e6 (Double-Expanded) in 60% yield from 3° hub-5 after three steps (Fig. 5e).

At the end, without needing to re-negotiate an entire route, we were able to secure seven distinct scaffolds after a total of 49 steps, with an average for steps/scaffold efficiency18 of seven and steps/analogue economy of four.

In vitro testing

In vitro testing was performed to determine whether the skeletal alterations lead to compounds that possess distinct bioactivity (Fig. 5f). Phenotypic screening41 was performed to gauge the cytotoxicity of the natural product (NP) and analogues (that is, enantiomers of compounds shown in Fig. 5a–e; see the supplementary information, section 12, for details) against an available panel of four cancer cell lines (with oxaliplatin as the control compound42). (There was no particular rationale for selecting these four cell lines, nor should this imply that further testing involving other cell lines is not warranted.) While the original naturally occurring compound (NP) was found to be weakly anti-malarial9, we performed this study for several reasons. Although an indole alkaloid natural product had been found to be cytotoxic43, we wanted to illustrate that application of the present strategy can lead to potential leads in other disease areas.

NP and most altered frameworks did not exert any influence. In contrast, two expanded frameworks promoted cell death: Z-e1 displayed appreciable cytotoxicity against two of the cell lines and, more notably, e6 exhibited cytotoxicity against all four cell lines. As shown by the dose-response curves, the double-expanded scaffold of e6 represents a notable shift in bioactivity. Limited testing of the corresponding unnatural enantiomers, which included ent-Z-e1, showed minimal cytotoxicity (>>50 μM; see the supplementary information, section 12.4), underscoring the importance of the catalytic enantioselective process.

The resulting structure-activity relationship (SAR) is optimisable. For example, comparison of the data for e6 together with those for NP and e4, bearing only one of the two one-methylene expansions in e6 implies that e6’s effectiveness might arise from the combined influence of expansion in two different bicyclic amine regions, or might be on account of its particular one-methylene enlargement. Conversion of 1° hub to the analog bearing only the latter one-methylene expansion and biological testing would be one way of gaining further clarity.

Regarding the puzzling selectivity trends

We encountered several perplexing selectivity profiles during the methodological phase of our investigations. Better understanding of the origins of these trends would enhance the applicability of the catalytic multicomponent process. DFT studies were thus performed.

The first experimental observation was the higher Z:E ratio for the transformation that involved cyclohexyl-substituted vinylallene (rac-2c) and generated homoallylic alcohol 3c through reaction with acetophenone (>98:2 Z:E; see Fig. 2b). In contrast, the Z:E ratio for reaction of rac-2c with 2-heptanone to afford 6c was at the typically observaed range (82:18 Z:E). DFT studies reveal that the transition state leading to the Z-alkene isomer of 3c is more favoured, partly owing to London dispersive attraction44,45 (ts-3c-bisphos-5-maj, Fig. 6a) between the axially-oriented C–H bonds of the cyclohexyl moiety and the p-cloud of acetophenone’s phenyl group. Due to diminished polarizability and enhanced conformational mobility, the same interaction would be less influential with an n-alkyl-ketone (for example, formation of 6c). Another reason for the increase in Z:E selectivity is the greater steric pressure between B(pin) and cyclohexyl groups in the less preferred ts-3c-bisphos-5-min compared to when an n-alkyl-allene is used (for example, see 3a-b). While the phenethyl moiety can be engaged in dispersive attraction in the lower energy transition state, its smaller size renders the alternative pathway more competitive (for further analysis, see the supplementary information, section 13.5).

Another unforeseen outcome was that the reaction with ortho-methoxyacetophenone afforded 3g more enantioselectively with bisphos-3 than the typically effective bisphos-5 (97:3 compared to 59:41, respectively). This discrepancy seems to originate from the larger angle between the aryl and methyl substituents of a ketone (to minimize allylic strain) in the case of the ortho-methoxy variant (120.5° (ts-3g-bisphos-5-maj, Fig. 6b) and 118° (ts-3d-bisphos-5-maj) for ortho-methoxyacetophenone and acetophenone, respectively). Consequently, with bisphos-5–Cu, the ortho-methoxyphenyl moiety is positioned closer to one of the tert-butyl groups of the bisphosphine ligand (2.11 compared to 2.15 Å in ts-3g-bisphos-5-maj and ts-3d-bisphos-5-maj, respectively). The ensuing steric pressure lowers the activation energy difference (ΔΔG‡) from 1.7 kcal mol−1 for the formation of 3d to 0.6 kcal mol−1, causing 3g to be formed with lower e.r. (59:41 compared to 94:6 for 3d). With bisphos-3 and its more sizeable triarylphosphine unit, while the preferred addition mode, ts-3g-bisphos-3-maj, is impacted nominally (compared to ts-3g-bisphos-5-maj), there is an increase in steric repulsion in the transition state for the minor enantiomer (ts-3g-bisphos-3-min). This arises from propinquity of the allylic moiety and the catalyst’s tert-butyl group, engendered by the larger ligand cone angle in ts-3g-bisphos-3-min (248.6°) compared to ts-3g-bisphos-5-min (244.9°). The gap between the two pathways thus widens to 1.6 kcal mol−1, and 3g is generated in 97:3 e.r. It follows that, the ketone with a smaller ortho-fluoro-substituted or bicyclic aryl moiety is converted in the presence of bisphos-5 to homoallylic alcohol 3i and 3j in 92:8 e.r., as anticipated.

Particularly enigmatic were the data regarding the catalytic multicomponent reaction that afforded 1° hub. With the typically effective bisphos-5, the homoallylic alcohol was formed as a near-racemate (53:47 e.r.), whereas with bisphos-2 the multifunctional platform was generated in 95:5 e.r. Equally surprising, and as presaged by DFT studies, the major enantiomer was not the one that we had predicted to be favoured based on the methodological studies (for example, see 3d, Fig. 2b). There seem to be several reasons for these. One is that, with most ketones, the methyl substituent is situated within the same quadrant as the ligand’s arylphosphine and the substrate’s aryl or heteroaryl and the tert-butyl phosphine moieties reside within a different quadrant (see ts-3d-bisphos-5-maj, Fig. 6b). The situation changes with benzyl-protected indolyl ketone 1b: now the indolyl fragment occupies the same quadrant as the ligand’s arylphosphine (see ts-1° hub-bisphos-5-maj and ts-1° hub-bisphos-2-maj; Fig. 6c). The reason for this is likely a stabilizing π-π interaction between the N-benzyl group and a phenyl substituent of the catalyst’s bisarylphosphine moiety (see ts-1° hub-bisphos-5-maj (edge-to-face) and ts-1° hub-bisphos-2-maj (face-to-face)). The preferred transition state with bisphos-5–Cu, namely, ts-1° hub-bisphos-5-maj, although stabilized by a π-π association, suffers from steric pressure on account of the proximity (1.91 Å) of the allyl moiety and one of the catalyst’s phenylphosphines. The net outcome is negligible energy difference between the pathways (ΔΔG‡ = 0.3 kcal mol−1; 47:53 e.r.). In ts-1° hub-bisphos-2-maj, on the other hand, the latter steric repulsion is ameliorated by more separation between that copper–allyl and diarylphosphine groups, a consequence of the ligand’s smaller cone angle (229.8° compared to 235.0° for ts-1° hub-bisphos-5-maj). The energy gap between the competing bisphos-2-derived transition states is therefore wider and enantioselectivity higher (ΔΔG‡ = 1.3 kcal mol−1; 5:95 e.r.). In brief, in the case of N-benzyl-protected indolyl ketone 1b, a blend of π-π interactions and ligand steric effects cause C–C bond formation to occur preferentially at the ketone enantiotopic face that would otherwise lead to the minor enantiomer.

The DFT-derived rationale is consistent with the lower e.r. for a chiral bisphosphine ligand wherein a phosphine is bound to two 3,5-bis-trifluoromethylphenyl moieties (instead of a 4-methoxy-3,5-dimethylphenyl groups in bisphos-2). In this instance, under otherwise identical conditions (Fig. 3b) S,R-1° hub was generated in 77:23 e.r. (54% yield (pure Z), 76:24 Z:E, >98:2 d.r.; see the supplementary information, section 13.8, for details), as the π-π interaction is probably weaker with a more electron-deficient arylphosphine. Further, the reaction involving N-SEM-protected (SEM, 2-trimethylsilylethoxymethyl) indolyl ketone and bisphos-2 was less enantioselective (see the supplementary information, section 13.9, for details). Phosphines that have the necessary electronic attributes and an N-benzyl group are thus key to the degree and sense of enantioselectivity.

Conclusions

We introduce a catalytic regio-, diastereo-, and enantioselective multicomponent process that played a central role in the realization of a concise total synthesis of an indole alkaloid and various precisely altered scaffolds. The catalytically generated 1° hub was first transformed through a progressively divergent series of 21 reactions, not more than five steps in each stage, to a pair of 2° hubs and their subsequent conversion to five 3° hubs, all in high diastereo- and enantiomeric purity. The collection of 3° hubs was then utilized to access NP (11-step LLS, 9.1% overall yield compared to 23-step LLS and 4.7% overall yield, formerly16). Ten skeletal analogues were generated after 26 operations. After all was said and done, after a total of 49 steps, seven unique scaffolds and twelve different analogs were in hand, translating to an average of seven steps/scaffold and four steps/analogue.

The foregoing 2° and 3° hubs represent a sample of alterable platforms. The alkenyl boronate could, for instance, be subjected to iterative Matteson homologation46 (instead of mono- or double homologation performed here). Alternatively, after a single-methylene truncation (hydroxy-assisted hydroamination), the alkenyl boronate might be used for chain extension prior to cyclization (for example by catalytic cross-metathesis or hydroformylation), affording differently expanded scaffolds. Nor do the transformations utilized for the five 3° hubs embody an exhaustive list, as the multicomponent process is not confined to indolyl ketones, and Witkop cyclizations can be performed with 3° hubs-3–5. Many such foci may be converted to skeletally altered analogs of other members of this extensive alkaloid family. The catalytically generated densely functionalized products may be utilized to access other NP types, such as isodactyloxene A47, charborol B48, fumagillin49 and structures represented by TNP-47050.

In vitro studies bring to light the implications of a made-to-order approach to ground-up precise skeletal diversification, illustrating that a specifically altered analogue – with one more or one less methylene unit or both – can be distinct from the initial natural product. The data provided here, involving a small number of cancer cell lines, suggest that the same skeletal analogs can show unique activity; the origins of such differences may then be further pursued. Consistent with these expectations, preliminary docking studies reveal that different skeltal analogues represent a superior fit for different receptor sites.

Viewed through a wider lens, the findings described here underline the importance of, and the need for, synthesis methods and strategies the applicability of which is not confined to synthesis of a small set of natural products. The possibility of accessing precisely altered and unnatural frameworks should be taken into account as well.

Supplementary Material

supplementary information

Acknowledgements

The methodological aspects of this work were funded by the NIH (grant R35 GM-130395 to A. H. H.). Support for applications to synthesis of NP and the corresponding analogues was provided by the CNRS, the ANR (project PRACTACAL to A. H. H.), the Jean-Marie Lehn Foundation for Chemistry Research (University of Strasbourg, to A. H. H.), and the Circle Gutenberg Foundation (2020 Chair for A. H. H.). Additional support was provided by the University of Illinois (for P. J. H.), and the NIH (grant R35 GM-128779 to P. L.). We thank the Shanghai Institute of Organic Chemistry and Solvay, S. A. for a postdoctoral and a predoctoral fellowship to Y. H. and X. L., respectively. In vitro testing was performed at the University of Illinois, Urbana-Champain. DFT and PMI studies were carried out at the Center for Research Computing at the University of Pittsburgh, Bridges 2 supercomputer at the San Diego Supercomputer Center through allocation TG-CHE140139 from the Advanced Cyberinfrastructure Ecosystem: Services & Support (ACCESS) program, funded by NSF grants. We are grateful to F. Romiti, M. Formica, S. Ng, S. Xu, and A. Nikbakht for helpful discussions.

Footnotes

Competing interests

The authors declare no competing financial or non-financial interests.

Data availability

All data in support of the findings of this study are available within the Article and its Supplementary Information. Crystallographic data for the structures reported in this Article have been deposited at the Cambridge Crystallographic Data Centre, under deposition numbers CCDC 2237819 (7b) and 1874716 (8).

References

  • 1.Campos KR et al. The importance of synthetic chemistry in the pharmaceutical industry. Science 363, eaat0805 (2019). [DOI] [PubMed] [Google Scholar]
  • 2.Grigalunas M, Burhop A, Christoforow A & Waldmann H Pseudo-natural products and natural product-inspired methods in chemical biology and drug discovery. Curr. Opin. Chem. Biol 56, 111–118 (2020). [DOI] [PubMed] [Google Scholar]
  • 3.Guillemard L, Kaplaneris N, Ackermann L & Johansson MJ Late-stage C–H functionalization offers new opportunities in drug discovery. Nat. Rev. Chem 5, 522–545 (2021). [DOI] [PubMed] [Google Scholar]
  • 4.Huigens RW III et al. A ring-distortion strategy to construct stereochemically complex and structurally diverse compounds from natural products. Nat. Chem 5, 195–202 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Jurczyk J et al. Single-atom logic for heterocycle editing. Nat. Syn 1, 352–364 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.McLeod MC et al. Probing chemical space with alkaloid-inspired libraries. Nat. Chem 6, 133–140 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Fokas D, Yu L & Baldino CM Strategies for the synthesis of indole alkaloid-based screening libraries for drug discovery. Mol. Divers 9, 81–89 (2005). [DOI] [PubMed] [Google Scholar]
  • 8.Shearer J, Castro JL, Lawson ADG, McCoss M & Taylor RD Rings in clinical trials and drugs: present and future. J. Med. Chem 65, 8699–8712 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Noguchi Y et al. Synthesis and stereochemical determination of an antiparasitic pseudo-aminal type monoterpene indole alkaloid. J. Nat. Med 70, 302–317 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Arens H, Borbe HO, Ulbrich B & Stöckigt J Detection of pericine, a new CNS-active indole alkaloid from Picralima nitida cell suspension culture by opiate receptor binding studies. J. Med. Plant. Res 46, 210–214 (1982). [DOI] [PubMed] [Google Scholar]
  • 11.Bonjoch J, Solé D, GarcÍa-Rubio S & Bosch J A general synthetic entry to strychnos alkaloids of the curan type via a common 3a-(2-nitrophenyl)hexahydroindol-4-one intermediate. Total syntheses of (±)19,20-dihydroakuammicine, (±)-norfluorocurarine, (±)-echitamidine, and (±)-20-epilochneridine. J. Am. Chem. Soc 119, 7230–7240 (1997). [Google Scholar]
  • 12.Michel S, Tillequin F & Koch M Brafouédine et isobrafouédine: nouveux alcoïdes indoliques mineurs de strychnos dinklagei. J. Nat. Prod 49, 452–455 (1986). [Google Scholar]
  • 13.Lima JA et al. Geissoschizoline, a promising alkaloid for Alzheimer’s disease: inhibition of human cholinesterases, anti-inflammatory effects and molecular docking. Bioorg. Chem 104, 104215 (2020). [DOI] [PubMed] [Google Scholar]
  • 14.Arita T et al. Discovery of conolidine derivative DS39201083 as a potent novel analgesic without mu opioid agonist activity. Bioorg. Med. Chem. Lett 29, 1938–1942 (2019). [DOI] [PubMed] [Google Scholar]
  • 15.Jhoti H, Williams G, Rees DC & Murray CW The ‘rule of three’ for fragment-based drug discovery: where are we now? Nature Rev. Drug Discov 12, 644 (2013). [DOI] [PubMed] [Google Scholar]
  • 16.Hirose T et al. Structure determination and total synthesis of (+)-16-hydroxy-16, 22-dihydroapparicine. Chem. Eur. J 19, 10741–10750 (2013). [DOI] [PubMed] [Google Scholar]
  • 17.Schreiber SL Target-oriented and diversity-oriented organic synthesis in drug discovery. Science 287, 1964–1969 (2000). [DOI] [PubMed] [Google Scholar]
  • 18.Galloway WRJD, Isidro-Llobet A & Spring DR Diversity-oriented synthesis as a tool for the discovery of novel biologically active small molecules. Nat. Commun 1, 80 (2010). [DOI] [PubMed] [Google Scholar]
  • 19.Kim J, Kim H & Park SB Privileged structures: efficient chemical “navigators” toward unexplored biologically relevant chemical spaces. J. Am. Chem. Soc 136, 14629–14638 (2014). [DOI] [PubMed] [Google Scholar]
  • 20.Wilson RM & Danishefsky SJ Small molecule natural products in the discovery of therapeutic agents: the synthesis connection. J. Org. Chem 71, 8329–8351 (2006). [DOI] [PubMed] [Google Scholar]
  • 21.Garcia-Castro M, Zimmermann S, Sankar MG & Kumar K Scaffold diversity synthesis and its application in probe and drug discovery. Angew. Chem. Int. Ed 55, 7586–7605 (2016). [DOI] [PubMed] [Google Scholar]
  • 22.Lusi RF, Perea MA & Sarpong R C–C bond cleavage of α-pinene derivatives prepared from carvone as a general strategy for complex molecule synthesis. Acc. Chem. Res 55, 746–758 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Liu Y-L & Lin X-T Recent advances in catalytic asymmetric synthesis of tertiary alcohols via nucleophilic addition to ketones. Adv. Synth. Catal 361, 876–918 (2019). [Google Scholar]
  • 24.Meng F, Jang H, Jung B & Hoveyda AH Cu-catalyzed chemoselective preparation of 2-(pinacolato)boron-substituted allylcopper complexes and their in situ site-, diastereo- and enantioselective additions to aldehydes and ketones. Angew. Chem. Int. Ed 52, 5046–5051 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Tsai EY, Liu RY, Yang Y & Buchwald SL A regio- and enantioselective CuH-catalyzed ketone allylation with terminal allenes. J. Am. Chem. Soc 140, 2007–2011 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Yang Y, Perry IB, Lu G, Liu P & Buchwald SL Copper-catalyzed asymmetric addition of olefin-derived nucleophiles to ketones. Science 353, 144–150 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Li C et al. CuH-catalyzed enantioselective ketone allylation with 1,3-dienes: scope, mechanism, and applications. J. Am. Chem. Soc 141, 5062–5070 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Feng J-J, Xu Y & Oestreich M Ligand-controlled diastereodivergent, enantio- and regioselective copper-catalyzed hydroxyalkylboration of 1,3-dienes with ketones. Chem. Sci 10, 9679–9683 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Fu B et al. Copper-catalyzed asymmetric reductive allylation of ketones with 1,3-dienes. Org. Lett 21, 3576–3580 (2019). [DOI] [PubMed] [Google Scholar]
  • 30.Huang Y, Torker S, Li X, del Pozo J & Hoveyda AH Racemic vinylallenes in catalytic enantioselective multicomponent processes: rapid generation of complexity through 1,6-conjugate additions. Angew. Chem. Int. Ed 58, 2685–2691 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Matteson DS & Sadhu KM Boronic ester homologation with 99% chiral selectivity and its use in syntheses of the insect pheromones (3S,4S)-4-methyl-3-heptanol and exo-brevicomin. J. Am. Chem. Soc 105, 2077–2078 (1983). [Google Scholar]
  • 32.Hoveyda AH, Hird AW & Kacprzynski MA Small peptides as ligands for catalytic asymmetric alkylations of olefins. Rational design of catalysts or of searches that lead to them? Chem. Commun 1779–1785 (2004). [DOI] [PubMed] [Google Scholar]
  • 33.Blakemore PR & Burge MS Iterative stereospecific reagent-controlled homologation of pinacol boronates by enantioenriched α-chloroalkyllithium reagents. J. Am. Chem. Soc 129, 3068–3069 (2007). [DOI] [PubMed] [Google Scholar]
  • 34.Strom AE & Hartwig JF One-pot anti-Markovnikov hydroamination of unactivated alkenes by hydrozirconation and amination. J. Org. Chem 78, 8909–8914 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Bernoud E et al. Recent advances in metal free-and late transition metal-catalysed hydroamination of unactivated alkenes. Catal. Sci. Technol 5, 2017–2037 (2015). [Google Scholar]
  • 36.Huang L et al. Late transition metal-catalyzed hydroamination and hydroamidation. Chem. Rev 115, 2596–2697 (2015). [DOI] [PubMed] [Google Scholar]
  • 37.Takaya H, Yamakawa M, Mashima K Synthesis and characterization of 2-[di(cyclopentadienyl)zircona]-1-oxacyclopentanes. X-Ray crystal structure of [(η-C5H5)2ZrOCH2CH2CHMe]2. J. Chem. Soc., Chem. Commun 1283–1284 (1983). [Google Scholar]
  • 38.Liu X & Ready JM Directed hydrozirconation of homopropargylic alcohols. Tetrahedron 64, 6955–6960 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Gritsch PJ, Leitner C, Pfaffenbach M & Gaich T The Witkop cyclization: a photoinduced C–H activation of the indole system. Angew. Chem. Int. Ed 53, 1208–1217 (2014). [DOI] [PubMed] [Google Scholar]
  • 40.Joule JA et al. Alkaloid studies. Part XLVIII. The structure of apparicine, a novel aspidosperma alkaloid. J. Chem. Soc 87, 4773–4780 (1965). [PubMed] [Google Scholar]
  • 41.Moffat JG, Vincent F, Lee JA, Eder J & Prunotto M Opportunities and challenges in phenotypic drug discovery: an industry perspective. Nat. Rev. Drug Discov 16, 531–543 (2017). [DOI] [PubMed] [Google Scholar]
  • 42.Ghanbarian M, Afgar A, Yadegarazari R, Najafi R & Teimoori-Tolabi L Through oxaliplatin resistance induction in colorectal cancer cells, increasing ABCB1 level accompanies decreasing level of miR-302c-5p, miR-3664–5p and miR-129–5p. Biomed. Pharmacol 108, 1070–1080 (2018). [DOI] [PubMed] [Google Scholar]
  • 43.Kobayashi J et al. Subincanadines A–C, novel quaternary indole alkaloids from Aspidosperma subincanum. J. Org. Chem 67, 6449–6455 (2002). [DOI] [PubMed] [Google Scholar]
  • 44.Fujii A et al. Experimental and theoretical determination of the accurate CH/p interaction energies in benzene–alkane clusters: correlation between interaction energy and polarizability. Phys. Chem. Chem. Phys 13, 14131–14141 (2011). [DOI] [PubMed] [Google Scholar]
  • 45.Wagner JP & Schreiner PR London dispersion in molecular chemistry – reconsidering steric effects. Angew. Chem. Int. Ed 54, 12274–12296 (2015). [DOI] [PubMed] [Google Scholar]
  • 46.Burns, et al. Assembly-line synthesis of organic molecules with tailored shapes. Nature 513, 183–188 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Ji N-Y et al. Diterpenes, sesquiterpenes, and a C15-acetogenin from the marine red alga Laurencia mariannenins. J. Nat. Prod 70 1901–1905 (2007). [DOI] [PubMed] [Google Scholar]
  • 48.Zhang W-H, Williams ID & Che C-T Chabrolols A, B., and C, three new norditerpenes from the soft coral Nephthea charbroli. Tetrahedron Lett. 42, 4681–4685 (2001). [Google Scholar]
  • 49.Sin N, et al. The anti-angiogenic agent fumagillin covalently binds and inhibits the methionine aminopeptidase, MetAP-2. Proc. Natl. Acad. Sci 94, 6099–6103 (1997). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Lu J, Chong CR, Hu X & Liu JO Fumarranol, a rearranged fumagillin analogue that inhibits angiogenesis in vivo. J. Med. Chem 49, 5645–5648 (2006). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplementary information
NIHMS2016024-supplement-supplementary_information.pdf (789.2KB, pdf)

Data Availability Statement

All data in support of the findings of this study are available within the Article and its Supplementary Information. Crystallographic data for the structures reported in this Article have been deposited at the Cambridge Crystallographic Data Centre, under deposition numbers CCDC 2237819 (7b) and 1874716 (8).

RESOURCES