All-carbon [3 + 2] cycloaddition in natural product synthesis

Many natural products possess interesting medicinal properties that arise from their intriguing chemical structures. The highly-substituted carbocycle is one of the most common structural features in many structurally complicated natural products. However, the construction of highly-substituted, stereo-congested, five-membered carbocycles containing all-carbon quaternary center(s) is, at present, a distinct challenge in modern synthetic chemistry, which can be accessed through the all-carbon [3 + 2] cycloaddition. More importantly, the all-carbon [3 + 2] cycloaddition can forge vicinal all-carbon quaternary centers in a single step and has been demonstrated in the synthesis of complex natural products. In this review, we present the development of all-carbon [3 + 2] cycloadditions and illustrate their application in natural product synthesis reported in the last decade covering 2011–2020 (inclusive).


Introduction
The highly-substituted, stereo-congested, five-membered carbocycle containing contiguous stereocenters is one of the most common structural features in many structurally complicated, biologically important natural products [1][2][3][4][5][6][7] (Figure 1). Meanwhile, the construction of quaternary carbon stereocenter(s) is, at present, a distinct challenge in modern synthetic chemistry [8][9][10][11]. Therefore, the synthesis of highly-substituted five-membered carbocycles bearing congested arrays of stereocenters within the polycyclic framework of complex natural products usually require a sophisticated synthetic planning. This issue is not trivial because only a few strategies are available for the efficient synthesis of such an intriguing molecular architecture. More importantly, the all-carbon [3 + 2] cycloaddition can forge vicinal all-carbon quaternary centers [12] in a single-step operation and provides a direct access to various substituted fivemembered carbocycles. These characteristics make the all-carbon [3 + 2] cycloaddition an appealing method and/or strategy in the synthesis of complex natural products ( Figure 2). The 1,3-dipolar cycloaddition has been well-documented and widely used for the construction of five-membered heterocycles since the 1960s [13]. However, the development of the   [3 + 2] cycloaddition. (Quaternary carbon center(s) created by all-carbon [3 + 2] cyclization are highlighted in cyan; quaternary carbon center(s) created that are removed by subsequent transformations are highlighted in lilac; cyclopentane structures forged by the all-carbon [3 + 2] cyclization are labeled in red). (A) The intermolecular all-carbon [3 + 2] cyclization features as the key reaction. (B) The intramolecular all-carbon [3 + 2] cycloaddition features as the key reaction.
In this review, we present the development of the all-carbon [3 + 2] cycloaddition and discuss its application in natural product synthesis reported from 2011-2020. We begin with describing the brief history of the all-carbon [3 + 2] cycloaddition with selected natural product syntheses reported before 2011 [22][23][24][25][26]. Next, we discuss the synthetic methods including the proposed mechanism and/or catalytic cycle and focus on illustrative examples of natural product syntheses. Moreover, several natural product syntheses featuring all-carbon [3 + 2] annulation are elaborated. Lastly, we discuss future directions and opportunities for the all-carbon [3 + 2] cycloaddition.
With the successful preparation of angular fused triquinane 41 by trimethylenemethane diyl [3 + 2] cycloaddition [29], enabled the synthesis of (−)-crinipellin A (15) [30] and waihoensene (16) [31] by Lee and co-workers in 2014 and 2017, respectively (Scheme 2B and Scheme 2C). The synthesis of (−)-crinipellin A (15) began with the treatment of hydrazone 42 with sodium hydride under reflux to produce the tetraquinane 46 in 87% yield [30] (Scheme 2B). The authors suggested that the diazo compound 43 formed undergoes an intramolecular cycloaddition to give 44. Freshly prepared 44 was converted to diyl 45 followed by another cycloaddition to give the tetraquinane 46. A four-step synthesis from the tetraquinane 46 gave diketone 47. Treatment of sulfoximine 48 with n-butyllithium generated the corresponding anion, which selectively attacked the C-8 ketone moiety of 47 to give alcohol 49 in β-configuration in 80% yield [32]. Chemoselective and stereoselective reduction of the C-9 ketone of 49 was accomplished by treatment with NaBH(OAc) 3 [33] and produced 50 after a two-step synthesis. Removal of the sulfoximine group in 50 upon refluxing in toluene and subsequent epoxidation afforded 51 [32], which was converted to (−)-crinipelline A (15) in two steps.
The synthesis of waihoensene (16) commenced with the conversion of aldehyde 52a to the corresponding hydrazone 52b, which was treated with sodium hydride under reflux to give 56 in 83% yield over two steps [31] (Scheme 2C). This transformation was rationalized as follows: freshly prepared 52b was converted to diazo 53, which was subjected to [3 + 2] cycloaddition to give adduct 54. Formation of diyl 55 from 54 and subsequent [3 + 2] cycloaddition produced the tetracyclic compound 56. Dihydroxylation of freshly prepared 56 with OsO 4 and then selective tosylation afforded 57 in 39% yield over two steps. Exposure of 57 to DBU upon heating gave the elimination product 58, which was subjected to an oxidative rearrangement with PDC to give enone 59 in 68% yield. Copper-mediated conjugated addition of methyllithium to enone 59 in the presence of boron trifluoride ether [34,35] produced desired ketone 60 in 75% yield. The resultant ketone 60 was converted to waihoensene (16) in two steps.

Palladium-catalyzed carboxylative trimethylenemethane cycloaddition
In 1986, Trost and co-workers disclosed the palladium-catalyzed intermolecular carboxylative TMM [3 + 2] cycloaddition [36] (Scheme 3). Exposure of coumarin 61 to the silyl-substituted TMM precursor 62 in the presence of a catalytic amount of Pd(PPh 3 ) 4 afforded adduct 63 in 81% yield as a single diastereomer (Scheme 3A). Trost and co-workers proposed that the catalytic mechanism involves an oxidative addition of palladium(0) into 62 affording the η 3 -Pd TMM complex A [37] (Scheme 3B). Methyl trimethylsilyl carbonate (64) is formed as side product, which is in equilibrium with carbon dioxide and methyl trimethylsilyl ether. The electron-rich end of complex A attacks the carbon dioxide to give carboxylate B. Migration of the TMS group on carboxylate B generates the 1,3-dipole on C in the form of TMS carboxylate. An intermolecular [3 + 2] cycloaddition of C and alkene D (see Scheme 3B, inset) gives the cycloaddition adduct E, which is converted to the corresponding carboxylic acid (not shown) upon reaction work-up. This elegant reaction was applied in the synthesis of marcfor-tine B (8), reported by Trost and co-workers in 2007 [38] and 2013 [39]. The synthesis of marcfortine B (8) began with palladium-catalyzed intermolecular carboxylatve TMM [3 + 2] cycloaddition [36] of enone 65 and TMM donor 62 to forge the highly-substituted spirocyclic cyclopentane 66a [38] (Scheme 4A). Methylation of the resultant cyclopentane 66a gave methyl ester 66b in 93% yield over two steps. A six-step synthesis from ester 66b gave α,β-unsaturated amide 67, which was treated with KHMDS to facilitate an intramolecular Michael addition to give lactam 68 in quantitative yield. The conversion of freshly prepared lactam 68 to xanthante ester 69 was achieved in three steps. Exposure of xanthante ester 69 to AIBN and a catalytic amount of tributylstannane [40] led to a radical cyclization, in which the resultant alkyl radical formed was trapped by AIBN to give a proposed nitrogen-centered radical 70. An 1,4-hydrogen abstraction of the nitrogen-centered radical on 70 produced carbon-centered radical 71, which underwent fragmentation to afford alkene 72 in 61% yield. Marcfortine B (8) was synthesized from alkene 72 in seven steps.
The enantioselective synthesis of marcfortine C (9) commenced with a catalytic asymmetric cyano-substituted TMM cycloaddition of oxindole 73 and TMM donor 75 with Pd(dba) 2 /74 as catalyst to give a cycloaddition adduct (not shown) [39] (Scheme 4B). Subsequent treatement with t-BuOLi resulted in the isomerization of the exo-olefin followed by exposure to n-butyllithium and Davis' oxaziridine 76 to give 77 in 60% yield with 89% ee. A three-step synthesis from 77 gave α,βunsaturated amide 78, which underwent successive intramolecular Michael addition and hydrolytic nitrile reduction to give 79 in 46% yield in two steps. Extensive studies of the nitrile reduction eventually identified that Et 3 Al and DIBAL-H could effectively reduce the nitrile group to the corresponding aldehyde and treatment with NaBH 4 afforded alcohol 79. Alcohol 79 was converted into the corresponding xanthate ester 80. This ester 80 was exposed to an excessive amount of AIBN and N,Obis(trimethylsilyl)acetamide in the presence of a catalytic amount of tributylstannane producing bicyclo[2.2.2]diazaoctane 81 in 54% yield. The authors mentioned that the employment of the previously reported conditions for the radical cyclization in the synthesis of marcfortine B (8) led to the decomposition of the starting material. It was suggested that the MOM group of 80 may contribute to undesired side reactions. Synthesis of marcfortine C (9) was accomplished from 81 in two steps.

Phosphine-catalyzed [3 + 2] cycloaddition
In 1995, Lu and co-workers reported a phosphine-catalyzed [3 + 2] cycloaddition, employing electron-deficient olefins and either 2,3-butadienoates or 2-butynoates to give a cyclopentene as product [17] (Scheme 5A). The reaction between ethyl 2,3butadienoate (82) and diethyl fumarate (83) in the presence of 10 mol % of triphenylphosphine afforded trans-84 in 67% yield. Under the same conditions, the use of diethyl maleate in place of diethyl fumarate (83) will give cis-84 in 46% yield (not shown). Lu and co-workers proposed that the catalytic mechanism involves a reaction between phosphine catalyst A and allene 82 to give B and/or C (Scheme 5B). Catalytic [3 + 2] cycloaddition of B and/or C and alkene D gives the cyclic intermediates E and F in an equilibrium state through a 1,2-proton transfer. The loss of phosphine catalyst from E or F affords the cycloaddition product G and the catalyst is regenerated. It is noteworthy that ethyl 2-butynoate (85) can be used as substrate  [38,39]. (B) Enantioselective synthesis of marcfortine C (9) features a palladium-catalyzed asymmetric cyano-substituted TMM [3 + 2] cycloaddition [39].

Phosphine-catalyzed enantioselective [3 + 2] annulation
In 2019, Lu and co-workers disclosed a novel chiral-phosphinecatalyzed enantioselective [3 + 2] annulation of allenes and isoindigos to give an enantioenriched annulation adduct bearing vicinal quaternary stereocenters [46] (Scheme 7A). Both symmetric and unsymmetric isoindigos can undergo enantioselective [3 + 2] annulation with an allene and produced a chiral adduct with high yield and high ee value. When unsymmetric isoindigo 100 was used as substrate, enantioselective [3 + 2] annulation with allene 101 in the presence of amino acidderived bifunctional phosphine 102 produced adduct 103 in 90% yield with 92% ee and 4:1 regioisomeric ratio (rr). The authors suggested that the observed regioselectivity could be rationalized by the proposed catalytic mechanism (Scheme 7B). The phosphine (i.e., PR 3, A) attacks the allene 101 to generate zwitterion intermediate B, which is subjected to a less hindered attack by the isoindigo 100. The oxindole bearing a chlorine atom on isoindigo 100 makes C-3 more electron deficient than C-3', which results in the regioselective formation of intermediate C. Cyclization of intermediate C gives D and subsequent proton transfer produces isomer E. It undergoes elimination to afford the annulation product 103 and the phosphine catalyst A is regenerated.
The synthesis of (−)-lingzhiol (17) was reported by Yang and co-workers in 2014 [49] (Scheme 9A). The synthesis began with the conversion of ketone 112 into alcohol 113 in four steps, which involved a hypervalent iodine-mediated ring expansion [60]. A two-step synthesis from 113 gave epoxide 114. Epoxide 114 was converted to the corresponding β-ketoester and subsequent treatment with Waser's reagent 116 [61] afforded alkyne 117 in 62% yield over two steps. Enyne 118, which was prepared in two steps from 117, was subjected to rhodium-catalyzed intramolecular [3 + 2] cycloaddition in the presence of carbon monoxide to give tricycle 119 bearing the desired vicinal quaternary carbon stereocenters in 86% yield. Reduction of aldehyde 119 and subsequent transesterification produced a lactone (not shown). It was exposed to SeO 2 to install the allylic hydroxy group to give 120 in 65% yield. Upon catalytic hydrogenation of 120, alcohol 121 was formed. This alcohol 120 was subjected to a bromination [62]/oxidation sequence followed by demethylation to produce (−)-lingzhiol (17).
The synthesis of sinensilactam A (20) commenced with a threestep synthesis from ketoeseter 131 to give enone 132 [59] (Scheme 9C). Selective reduction of the ketone moiety of 132 was accomplished under Luche's conditions [65] in the presence of calcium chloride [63] to produce the desired alcohol over three steps. The conversion of 138 to sinensilactam A (20) was achieved in two steps.

Miscellaneous
In 2012, Wang and co-workers reported a Lewis acid-catalyzed intramolecular [3 + 2] cross-cycloaddition (IMCC) of cyclopropane 1,1-diesters with non-activated alkene to generate bridged [n.2.1] carbocyclic skeletons, which is applied to the synthesis of phyllocladanol (21) [68] (Scheme 12A). The IMCC precursor 147 was prepared from aldehyde 146 in nine steps. The IMCC precursor 147 underwent an intramolecular crosscycloaddition catalyzed by tin tetrachloride to give tetracycle 149 in 81% yield. The authors suggested that the intramolecular [3 + 2] cross-cycloaddition of the less-substituted external carbon atom in the C=C double bond results in the formation of the more stable internal carbenium (i.e., 148) and promotes IMCC to give the bridged [3.2.1] octane 149. The transformation of 149 to phyllocladanol (21) was accomplished in four steps.
In 2016, Winne and co-workers reported that (5,6-dihydro-1,4dithiin-2-yl)methanol (151) can be served as a allyl-cation equivalent for the [3 + 2] cycloaddition and was applied in the synthesis of (±)-cuparene (13) [69] (Scheme 12B). An intermolecular [3 + 2] cycloaddition of tetrasubstituted alkene 150 and the dhdt-2-methanol reagent 151 under the effect of trifluoroacetic acid produced adduct 154 in 52% yield. The authors identified that the cyclic nature of the dhdt-2-methanol reagent 151 is essential for the cycloaddition to take place. The use of noncyclic analogues did not give the cycloaddition product. It is suggested that the restricted rotational freedom of 151 and the related enforced conjugation of the sulfur lone pair may block certain undesired cation reactions. Cycloaddition product 154 was subjected to the hydrodesulfurization with Raney nickel as catalyst and subsequent catalytic hydrogenation produced (±)-cuparene (13) in 90% yield.

All-carbon [3 + 2] annulation in natural product synthesis
The all-carbon [3 + 2] cycloaddition demonstrated the ability to assemble intricate polycyclic structures in the synthesis of complex natural products. Besides the all-carbon [3 + 2] cycloaddition reactions and the corresponding applications described above, the all-carbon [3 + 2] annulation, which undergoes other possible mechanistic pathways other than cycloaddition, proved Scheme 13: The recent advances of [3 + 2] annulation in natural product synthesis. (A) The preparation of meloscine (158) features a cascade radical annulation of divinylcyclopropane [70]. (B) Thiyl-radical-mediated [3 + 2] annulation reaction realizes the synthesis of (−)-pavidolide B (166) [71,72]. (C) A Danheiser's [3 + 2] annulation en route to conidiogenone B (171) [73] (inset, the suggested mechanism based on Danheiser's proposal disclosed in 1981 [74].) its usefulness in forging highly-substituted five-membered carbocycles. These reactions have been applied successfully in the synthesis of complex natural products. In 2011, Curran and co-workers reported the synthesis of meloscine (158) featuring a tandem radical cyclization of a divinylcyclopropane [70] (Scheme 13A). Slow addition of tributylstannane and AIBN to a refluxing solution of cyclopropane 155 afforded 156 in 38% yield. It was subjected to cleavage of the Boc group followed by N-allylation to give 157 in 73% yield over two steps. A ringclosing metathesis of freshly prepared 157 was effected by the second generation Hoveyda-Grubbs (HG II) catalyst and subsequent base-promoted epimerization produced meloscine (158) in 83% yield.
In 2017, Yang and co-workers disclosed the synthesis of (−)-pavidolide B (166) by using a thiyl-radical-mediated [3 + 2] annulation reaction to create four contiguous stereocenters on tricycle 162 in one step [71,72] (Scheme 13B). Exposure of ester 159 to PhSH [75], p-toluidine and a catalytic amount of Ir(dF(CF 3 )ppy) 2 (dtbbpy)PF 6 under the irradiation of blue LED light [76,77] afforded tricycle 162 in 50% yield. The authors suggested that this process involves an intramolecular 5-exoconjugated addition of a radical on 160 to the enone and produces 161. The newly formed 161 was subjected to 5-exo radical addition to the allyl sulfane and subsequent loss of a thiyl radical produces 162. A successive hydrolysis/decarboxylation upon heating and cleavage of acetal on 162 afforded aldehyde 163 in 90% yield. Coupling of aldehyde 163 and isoprene (164) with Ni(acac) 2 and diethylzinc [78] and then Dess-Martin oxidation gave a diene (not shown, 94% yield over two steps), which was subjected to ring-closing metathesis to give enone 165 in 85% yield. Isomerization of the freshly prepared 165 to more stable α,β-unsaturated enone with RhCl 3 [79] afforded pavidolide B (166) in 95% yield.
The synthesis of (−)-conidiogenone B (171) featured a Danheiser's [3 + 2] annulation [74,80] and was reported by Zhai and co-workers in 2020 [73] (Scheme 13C). Treatment of tricycle 167 with allene 168 in the presence of TiCl 4 gave the desired 169 carrying two vicinal quaternary carbons. A one-pot desilylation of the newly formed 169 with a trifluoride-acetic acid complex produced the tetraquinane 170a in 89% yield with a 4:1 dr. The conversion of the freshly prepared ketone 170a to 170b was achieved in three steps. Ozonolysis of the C=C double bond of 170b gave a keto aldehyde (not shown), which was subjected to an acid-mediated aldol reaction to give conidiogenone B (171) in 53% yield. The undesired isomer with β,γ-C=C double bond (not shown) was formed in 34% yield and can be isomerized to the more stable α,β-unsaturated enone to afford conidiogenone B (171) in 32% yield upon treatment with RhCl 3 in microwave.
The reaction mechanism of Danheiser's [3 + 2] annulation is shown according to the Danheiser's proposal [74] (Scheme 13C, inset). Initial complexation of the α,β-unsaturated ketone 167 and titanium tetrachloride produces an alkyoxy allylic carbocation (not shown). This carbocation is subjected to a regiospecific electrophilic substitution of allene 168 to generate a vinyl cation 172, which is stabilized by an adjacent carbon-silicon bond. The 1,2-shift of the silyl group in 172 produces an isomeric vinyl cation, which is intercepted by the titanium enolate and results in the new C-C bond formation to give the five-membered carbocycle 169.

Conclusion
The all-carbon [3 + 2] cycloaddition, together with the [3 + 2] annulation, continue to be an attractive class of reactions for the synthesis of highly-substituted and stereo-congested five-membered carbocycles. Also, one or more quaternary carbons can be created in a single reaction making this class of reactions appealing to complex natural product syntheses. This review outlines the development of the all-carbon [3 + 2] cycloaddition and its application in natural product synthesis reported from 2011-2020 (inclusive). The intermolecular all-carbon [3 + 2] cycloaddition offers a facile approach to install functionalized five-carbon carbocycles, including fused-rings (e.g., longeracinphyllin A (10)) and/or spiro-ring (e.g., marcfortine B (8)), at later stage of the synthesis without the need of preinstallation of necessary functional groups as a reaction precursor, for instance, ring-closing metathesis, intramolecular aldol condensation, and others.
One major issue that still needs to be addressed is the selectivity of the all carbon [3 + 2] cycloadditions, which are usually under substrate-control. Remarkable innovation of the stereoselective palladium-catalyzed trimethylenemethane cycloaddition reported by Trost's group, which makes use of catalytic amounts of palladium and chiral phosphine ligand 74, was applied successfully in the enantioselective synthesis of marcfortine C (9, Scheme 4B). Another brilliant example is the development of a chiral-phosphine-catalyzed [3 + 2] annulation reported by Lu in 2019, in which the chiral phosphine catalyst confers high stereocontrol on the formation of a spiro adduct bearing two vicinal all-carbon quaternary stereocenters (Scheme 7). We believe that the enantioselective all-carbon [3 + 2] cycloaddition provides a new strategy for the preparation of sp 3 -carbon-enriched complex scaffolds [81,82] for biological studies and potential new drug development.
The all-carbon [3 + 2] cycloaddition is undoubtedly an efficient synthetic transformation that creates two C-C bonds in a single reaction. However, the prior protection of the reactive functional groups, such as the hydroxy and amino groups, are still necessary for most of the all-carbon [3 + 2] cycloaddition reactions. We predict that further development of the all-carbon [3 + 2] cyclization with the reactive functional groups' compatibilities and/or without the use of protecting groups [83,84] can improve the synthetic efficiency and make this class of reactions more attractive to the synthetic scientist for applications. Lastly, we anticipate that the all-carbon [3 + 2] cycloaddition will gain further attention from the synthetic community, including scientists from academia and pharmaceutical industry, for methodic innovation and the efficient synthesis of biologically important natural products.

Funding
Financial support from the Shenzhen Human Resources and Social Security Bureau (50820190066) to Z. Wang is gratefully acknowledged. J. Liu acknowledges the financial support from Shenzhen Science and Technology Innovation Committee (grant nos. JCYJ20190809181011411).