Vicinal stereocenters via asymmetric allylic alkylation and Cope rearrangement: a straightforward route to functionally and stereochemically rich heterocycles

An asymmetric allylic alkylation/Cope rearrangement (AAA/[3,3]) capable of stereoselectively constructing vicinal stereocenters has been developed. Strategically integrated 4-methylation on the 3,3-dicyano-1,5-diene controls stereoselectivity and drives Cope rearrangement equilibrium in the forward direction. The AAA/[3,3] sequence rapidly converts abundant achiral and racemic starting materials into valuable (hetero)cycloalkane building blocks bearing significant functional and stereochemical complexity, highlighting the value of (hetero)cyclohexylidenemalononitriles as launching points for complex heterocycle synthesis. On this line, the resulting alkylidenemalononitrile moiety can be readily converted into amides via Hayashi–Lear amidation to ultimately yield amido-piperidines, tropanes, and related scaffolds with 3–5 stereocenters and drug-like functionality.

Herein we report an enantio-and diastereoselective AAA/ Cope rearrangement sequence between alkylidenemalononitriles and 1,3-disubstituted allylic electrophiles to yield functionally and stereochemically complex building blocks (Fig. 1C). Strategically integrated 4-methylation on the 3,3-dicyano-1,5diene both controls stereoselectivity and drives Cope rearrangement equilibrium in the forward direction thus favoring the construction of the vicinal stereocenters at various temperatures (rt -110°C). The products contain two different alkene functional groups: the alkylidenemalononitrile can be mildly converted to amides diastereoselectively by NaBH 4 reduction and Hayashi oxidative amidation/esterication. 22,23 To maximize the impact of our work, we focused on the convergent coupling of functionalized starting materials, thus showcasing the tolerance of the sequence to structural modication and the potential value of the sequence for accessing functionally and stereochemically rich pharmaceutical leads. Therefore, this work yields complementary access to functionally-and structurally-complex piperidines, cyclohexanes, and related saturated heterocycles. 24 Recent state of the art approaches to complex piperidines include C-H functionalization, 25-28 from pyridine derivatives by hydrogenation or nucleophilic dearomatization, 29-36 and other methods. [37][38][39][40] A standout route to related cyclohexanes was described by Baran whereby Diels-Alder cycloaddition, desymmetrization, and decarboxylative coupling chemistry are impactfully harnessed. 41

Results and discussion
To begin our work, we examined a model asymmetric allylic alkylation (AAA)/Cope rearrangement yielding vicinally stereogenic product 4a via 4-methyl-3,3-dicyano-1,5-dienes 3a enantioselectively from alkylidenemalononitrile 1a and racemic allylic electrophiles rac-2a (Table 1). While this class of 1,5diene has not yet been accessed enantioselectively, our previous work established the fundamental Cope rearrangement reactivity of these substrates: 4-methyl-3,3-dicyano-1,5-dienes have favourable energetic proles for rearrangement to vicinally stereogenic products whereas analogous des-methyl Cope substrates do not. 18,21 Based on our optimization studies (Table  1), it was found that the coupling of 1 equivalent of alkylidenemalononitrile 1a and 2.5 equivalents of racemic allylic carbonate rac-2a catalyzed by a Pd/(S,S)-DACH-phenyl Trost ligand complex facilitated full conversion to the desired product in good enantiomeric ratio ( Table 1, entry 1). Using equimolar amounts of the starting materials resulted in full conversion but insignicant 53 : 47 er (Table 1, entry 2). These ndings suggest the reaction proceeds by a Pd-catalyzed kinetic resolution [42][43][44][45] rather than a dynamic kinetic asymmetric transformation (DYKAT). [46][47][48][49] Further increasing the equivalents of the electrophilic component did not improve the enantiomeric ratio ( Table 1, entry 3), neither did switching to the Pd/naphthyl-Trost complex ( Table 1, entry 4). However, we did see some impact with respect to the leaving group, as the allyl acetate ( Table 1, entries 6-7) resulted in a slightly increased enantiomeric ratio compared to the methyl carbonate. Additionally, the residual allylcarbonate (2a) was found to be highly enantioenriched (92 : 8 er) ( Table 1). Considering 2.5 equivalents of rac-2a were initially used, the 92 : 8 er was somewhat unexpected. This potentially suggests that in situ generated reagents (e.g., carbonate anion, hydrogen methyl carbonate, or methanol) could be acting as a nucleophile or base to further consume the reactive enantiomer of 2a. Additional experimentation is required to better understand this enantioselective transformation.
A decrease in er was observed for several (hetero)aromatic examples (e.g., 4a, 4e, 4h, 4i), yet this could be improved by exchanging the allyl carbonate for the corresponding allyl acetate electrophiles (e.g., 4a, 4i, and 4j). The incorporation of a nitrile functional group (4k) was not particularly selective under conditions of asymmetric allylic alkylation via kinetic resolution. However, allylic alkylation of crotyl cyanohydrin can be achieved in high er via stereospecic allylic alkylation. 19,50 The N-tosyl protecting group common to products 4a-4k can be exchanged for a more easily removable N-Boc group (4l) without notable changes to yield or er. Similarly, we examined various (hetero)cyclohexylidenemalononitriles 1c-1e. On this line, substituted tetrahydropyran (1c to 4m-4n), thiopyran (1d to 4o-4p), and ketal-protected cyclohexanone (1e to 4q-4r) analogs were accessed in high er. We also targeted diuorocyclohexyl products (Scheme 2A). Interestingly, these substrates had poor thermodynamic proles for Cope rearrangement, and the respective Cope isomers 3 and 4 were inseparable by column chromatography. To yield the targeted building blocks, we employed our previously reported "reductive Cope rearrangement" to drive forward Scheme 1 Scope of the Pd-catalyzed kinetic resolution/Cope rearrangement sequence. the [3,3] process. 19,51,52 As such, promoting Cope rearrangement in the presence of Hantzsch ester resulted in full conversion and good yields and stereoselectivity for products 5a and 5b. Importantly, the enantioselective synthesis of products 4 is scalable; as shown in Scheme 2B, we efficiently prepared gram quantities of a representative building block 4b. The catalyst loading could also be reduced tenfold, though the efficiency decreased slightly (69% yield, 95 : 5 er). Overall, these results suggest that large quantities of these functionally and stereochemically dense building blocks can be efficiently generated for diverse downstream modication.
In addition to meso-bicyclic alkylidenemalononitriles 1g-1i (e.g., tropane derivatives; Scheme 3), we also examined meso-4substituted cyclohexylidenemalononitrile 1j (Scheme 4). This substrate displayed some practical challenges but could still be converted to valuable, enantioenriched polysubstituted and functionalized cyclohexanes. Like the chemistry described in Scheme 3, the AAA-step here yielded two diastereomeric 1,5dienes, epi-3ff and 3ff. In contrast, the [3,3] kinetic proles for both diastereomers were similar resulting in a major product and minor product diastereomeric mixture via [3,3]. For the case examined, the Cope starting material 3ff and Cope product 4ff diastereomers were inseparable via silica gel chromatography. Thus, to access characterizable products, we opted to chemically separate the isomers via NaBH 4 reduction and oxidative amidation with morpholine. This was effective and yielded product 6ff as a single diastereomer (88 : 12 er) in 14% overall yield from 1j. This sequence also showcases a key challenge in need of addressing in future studies: there are currently no methods to desymmetrize meso-alkylidenemalononitriles. Such a development would address the stereoselectivity challenges present in Schemes 3 and 4.
This efficient enantioselective coupling of alkylidenemalononitriles and racemic allylic electrophiles yields diverse vicinally stereogenic building blocks with orthogonal alkene functional handles. We envisaged that the alkylidenemalononitrile moiety could be converted to amides/esters via NaBH 4 reduction and Hayashi oxidative amidation or esterication, 22,23 and the neutral alkene could be utilized in olen metathesis 55 and ozonolysis 56 as summarized in Scheme 4. On this line, the alkylidenemalononitrile can be consistently converted to carboxylate derivatives (Scheme 5A). This transformation gives access to amide and ester products derived from 2°amines (6a), 1°amines (6b), methoxymethyl amine (yielding a Weinreb amide) (6c), and methanol (6d). High compatibility of this protocol was conrmed through the presence and persistence of various functional groups on products 6e-6j.
We examined the metathesis reactivity of 6j, which resulted in ring-opening/cross-metathesis 57,58 to 2,3,4,6-tetrasubstituted N-Boc-piperidine 9 (Scheme 5B). Under no circumstance was the formation of a product analogous to 8 observed. We speculate that the N-Boc group is either sterically blocking the initiation of ring-closing metathesis (RCM) or chelating the necessary ruthylidene to an unreactive conformer preventing RCM. Compound 7 was prepared via a telescoped asymmetric allylic alkylation/Cope rearrangement/NaBH 4 reduction sequence. This scaffold, by a tandem ring-rearrangement metathesis (RRM) 59-61 /oxidative amidation, yielded highly substituted pyran-fused cyclopentene 8. Altogether, a rigidbicyclic framework 8 decorated with various functional groups (alkenes, chloroarene, morpholine-amide) can be prepared efficiently from the alkylidenemalononitrile and the racemic allylic electrophile. The alkylidenemalononitrile moiety of tropanes 4v, 4w and 4z was surprisingly resistant to NaBH 4 reduction. However, it was found that treatment of these scaffolds with ozone followed by NaBH 4 yields products 10a-10c. It is proposed that the neutral alkene undergoes chemoselective ozonolysis, and upon NaBH 4 reduction, the alcohol is generated which directs and allows the otherwise sterically blocked alkylidenemalononitrile reduction. Finally, it was found that substrates 10a-10c undergo intramolecular oxidative esterication to 11a-11c. Notably, while Hayashi has reported intermolecular conversion of malononitriles into amides and esters, this work establishes the rst report of an intramolecular variant of this effective chemistry (Scheme 6).

Conclusions
In conclusion, we have developed a simple strategy to convert abundant alkylidenemalononitriles and allylic electrophiles into stereochemically and functionally rich (hetero)cycloalkanes. This work relied on the development of an enantioand diastereoselective AAA/ [3,3] sequence, ultimately yielding a broad range of trans-3,4-disubstituted piperidines, trans-3,4disubstituted tropanes, and a 1,2,4-trisubstituted cyclohexane. As shown in Scheme 5, heterocycles and carbocycles with these patterns are common to marketed drugs and drug leads/ medicinal chemistry campaigns. The possibilities for this work are many fold and can be grouped into two categories: (1) it is anticipated that the synthetic chemistry ndings, both the successes and challenges uncovered, will serve as key insight and inspiration for future method development related AAA/ [3,3] transformations. Representative areas of future study should include dynamic kinetic asymmetric transformation based approaches to reactive 1,5-dienes, stereoselective deprotonation of meso-alkylidenemalononitriles, and access to wholly unique 1,5-dienes for enantioselective sigmatropic rearrangement. (2) It is anticipated that this chemistry as described can be utilized to access important molecules for therapeutic discovery. For example, designed analogs of the marketed drugs and lead molecules in Scheme 5 are within reach by this Scheme 6 This method can uniquely populate piperidine, tropane, and cyclohexane pharmaceutical space.
method. Further, because the described chemistry is both convergent and divergent from abundant starting materials, we believe this chemistry has potential to generate novel and complex sp3-rich libraries for hit-to-lead screening efforts. Continued fundamental and applied studies of AAA/ [3,3] will be reported in due course.

Data availability
The datasets supporting this article have been uploaded as part of the ESI. †

Author contributions
AN and MDM were the primary researchers responsible for data collection and management. They also cowrote the manuscript. ANN and ES were secondary contributors to the research project. AJG conceptualized the project, oversaw the research, supervised the co-authors, and wrote the manuscript.

Conflicts of interest
There are no conicts to declare.