Asymmetric intramolecular α-cyclopropanation of aldehydes using a donor/acceptor carbene mimetic

Enantioselective α-alkylation of carbonyl is considered as one of the most important processes for asymmetric synthesis. Common alkylation agents, that is, alkyl halides, are notorious substrates for both Lewis acids and organocatalysts. Recently, olefins emerged as a benign alkylating species via photo/radical mechanisms. However, examples of enantioselective alkylation of aldehydes/ketones are scarce and direct asymmetric dialkylation remains elusive. Here we report an intramolecular α-cyclopropanation reaction of olefinic aldehydes to form chiral cyclopropane aldehydes. We demonstrate that an α-iodo aldehyde can function as a donor/acceptor carbene equivalent, which engages in a formal [2+1] annulation with a tethered double bond. Privileged bicyclo[3.1.0]hexane-type scaffolds are prepared in good optical purity using a chiral amine. The synthetic utility of the products is demonstrated by versatile transformations of the bridgehead formyl functionality. We expect the concept of using α-iodo iminium as a donor/acceptor carbene surrogate will find wide applications in chemical reaction development.

Since the renaissance of enamine-iminium catalysis in 2000 (refs 39,40), enantioselective a-functionalization of carbonyls has taken the centre stage for synthetic innovation. Despite tremendous advance in this field, examples of direct amine catalysed a-alkylation of aldehydes are rare. Among them, two intramolecular a-alkylation strategies are notable: in 2004, Vignola and List reported the first enamine catalysed intramolecular alkylation using halo aldehydes 8 . Using this strategy, a cyclopropane aldehyde product was prepared. In 2013, MacMillan and co-workers accomplished intramolecular a-alkylation of aldehydes having a tethering double bond via enamine-singly occupied molecular orbital catalysis 41 . Alternatively, both MacMillan and Córdova reported methods to generate a-cyclopropane aldehydes from a,b-unsaturated aldehydes using an iminium-enamine cascade strategy [42][43][44][45] . Overviewing recent progress of organocatalytic a-alkylation reactions, efforts have been concentrated on replacing only one hydrogen atom at the a position of aldehydes with alkylating agents. In sharp contrast, double alkylation of the carbonyl a-methylene has been largely neglected 13 .
Herein, we report an orthogonal approach for direct a-cyclopropanation of aldehydes via intramolecular C ¼ C/CH 2 annulation, enabling rapid access to bicyclo[3.1.0]hexanes and their aza analogues (Fig. 1). This strategy uses an a-iodo iminium species as a key donor/acceptor carbene mimetic.

Results
Rationale design of a new donor/acceptor carbene mimetic. Recently, we reported an a-allenylation reaction of aldehyde using gold/amine synergistic catalysis, in which the a-methylene of carbonyl was converted to an sp 2 centre 46 . Along this line, we decided to explore the possibility of forming a cyclopropane through oxidative annulation between the aldehyde a-CH 2 and a tethering double bond. Conceptually, the net dehydrogenative nature of this transformation is counterintuitive. Normally, cyclopropanation of an olefin requires a donor/acceptor onecarbon counterpart such as a metal carbenoid, whereas the a-carbon of aldehydes is nucleophilic (donor only). We envisioned that an a-halo aldehyde intermediate might serve such a purpose. Upon mixing with a secondary amine, the corresponding a-halo iminium becomes a highly electrophilic alkylation species, whereas its complementary enamine form remains nucleophilic. This donor/acceptor reactivity might be synthetically equivalent to a carbene (Fig. 2).

Initial attempts on intramolecular a-cyclopropanation.
Enantioselective a-halogenation of aldehydes have been reported by several research groups [47][48][49][50][51][52][53][54][55][56] . Among those reported examples, chlorination of citronellal 55,56 seemed suitable as the starting point of our investigation. Unfortunately, attempts to affect the cyclopropanation reaction using a-chloro citronellal were fruitless. In the presence of a secondary amine, no cyclopropane aldehyde was detected. Further activation of the carbon-chloride bond using various Lewis acids resulted in substrate decomposition in most cases (Fig. 3). We speculated that chlorides are not electrophilic enough to react with an olefin, which is a rather poor close shell nucleophile. Next, we focused our attention on an iodination/annulation cascade. To further facilitate ring closure, we decided to install gem-diesters in the middle of the chain to assist conformational folding for intramolecular reaction (the Thorpe-Ingold effect).
We were excited to find out that treating aldehyde 1a with 1 equiv. of amine 3a (ref. 57) and N-iodosuccinimide (NIS) yielded the corresponding bicyclo[3.1.0]hexane aldehyde 2a in moderate yield and enantiomeric ratio (er; Fig. 3). No reaction occurred in the absence of NIS. Interestingly, a catalytic amount of amine 3a failed to give any product. Careful examination of the crude reaction mixture revealed that prolinol ether 3a quickly decomposed to benzophenone and pyrrolidin-2one. Control experiment showed that mixing 3a with NIS resulted in full conversion to benzophenone within 5 min at room temperature. As a result, a-iodination was undermined. The redox lability of amine 3a is likely a result of high electron density on the nitrogen atom, which might be attenuated using a weak Brønsted acid. Upon screening various acids, we found that p-nitrobenzoic acid (pNBA) stabilizes 3a under the reaction condition. When simple isopentaldehyde was subjected to the a-iodination reaction, the product was obtained in 35% yield using 20 mol% 3a, compared with a 90% conversion with 1 equiv. of 3a. Upon adding pNBA (1 equiv.) as additive, 90% yield was obtained using 20 mol% 3a and no decomposition of the amine was observed.
Optimization of the NIS-mediated a-cyclopropanation. Even though we were able to address the amine decomposition problem, the overall cyclopropanation cascade performed poorly using catalytic amine. Nevertheless, the use of pNBA continued to improve the efficiency of reactions employing 1 equiv. 3a. We next examined solvent effect. The reaction proceeds in a number of non-polar solvents ( Table 1, entries 1-6). Good enantioselectivity was observed in toluene and ether. Various iodination reagents were also investigated. 1,3-Diiodo-5,5-dimethylhydantoin afforded a very similar result as NIS ( dropped substantially. We speculated that both the electronic and the steric nature of the aryl group play an important role: electron-neutral aryls lead to higher yields; bulky 3,5-disubstitution improves er. Following this rationale, we prepared the corresponding 3,5-(t-Bu) 2 amine 3l. Improved yield was obtained at no cost of the selectivity (Table 1, entry 22). With this optimized chiral amine in hand, we quickly re-examined solvent effect. Further improvement of both yield and er was achieved using iPr 2 O as the reaction media (Table 1, entry 24, 75% yield, 93:7 er).
Substrate scope. The substrate scope was studied next ( Table 2). The cyclopropanation occurs smoothly using either carbon or nitrogen as the linker between carbonyl and olefin. Only cis-bicyclic isomers were isolated because of ring strain control. However, gem-disubstitution is required for the all-carbon linkers, without which the reaction becomes very slow. Trialkyl substituted double bond affords highest yield and er. Exo-cyclic double bonds were well tolerated, leading to bicyclo[3.1.0]hexanes with an additional ring spiro to the cyclopropane ( Table 2, products 2f-2h). The selectivity of these substrates increases with the size of the ring. When a cyclic ketal was used as the linker, both high yield and excellent er were obtained ( Table 2, product 2i). Interestingly, a 3-fuse-5-spiro-4 ring skeleton was obtained using an azetidine-containing substrate 1j. The alkyl substituents at the double bond terminal is essential. Removing these groups led to low reactivity and poor er ( Table 2, products 2k and 2l). Electrophile  Figure 2 | Design of a-halo iminium as a donor/acceptor carbene mimetic. We propose the a-carbon of a-halo aldehyde might serve as a donor/acceptor by the action of a secondary amine. In iminium form, the halide is a good electrophile. In enamine form, the carbon is a good nucleophile.

Donor/acceptor
When a tetrasubstituted olefin was used, a fully substituted cyclopropane product with three contiguous all-carbon quaternary centres was obtained as the single cis-diastereomer in 71% yield and 91:9 er (Table 2, product 2m). We also attempted intermolecular a-cyclopropanation reactions using simple aldehydes and various electron-rich olefins. Unfortunately, no desired cyclopropane product was obtained under the standard reaction condition. The a-iodo aldehyde intermediates were the major products. This result suggests a large entropy penalty is required for the key carbon-carbon bond formation.
Mechanistic experiments. To further understand the mechanism of this unconventional cascade cyclopropanation. We carried out real-time 1 H NMR experiments using deuterated solvent. The starting material was consumed within 10 min upon addition of NIS. The corresponding a-iodo aldehyde 4a was obtained in quantitative yield, which was slowly converted to product 2a.
In fact, a-iodo aldehyde 4a could be isolated by quick flash column chromatography. Subjecting 4a to amine alone led to the cyclopropane product in good yield without the need of NIS. Therefore, a-iodo aldehyde is very likely the intermediate for the overall cyclopropanation. It is noteworthy that the selectivity of the first step is poor. Er of 4a was 61:39 upon isolation after 5 min. However, when 4a was isolated after 40 min, it became racemic (Fig. 4a). This rapid decrease of optical purity for 4a indicated a facile racemization via enamine formation. Therefore, the stereoselectivity of the overall reaction is solely controlled in the annulation step. Er of the a-iodo intermediate ought to be inconsequential for the ultimate selectivity of the bicyclic product. Indeed, when racemic 4a was subjected to catalyst 3l in iPr 2 O, 2a was isolated in 82% yield and 93:7 er (Fig. 4b).
The annulation step was also studied. We prepared E/Z isomer pairs of disubstituted olefin 1n. The trans-product 2n was obtained as the major isomer regardless of the E/Z ratio of the starting material. Furthermore, a mixture (1.2:1) of diastereomers was obtained when the pure E-form of 1o was used (Fig. 5). Hence, the olefin/iodide annulation is a stepwise process, during which bond rotation occurs. Based on this observation, a free singlet carbene intermediate is highly unlikely. The cyclopropanation could occur through several pathways: (i) halogen extraction by the electron-rich olefin to form an iodiranium species, followed by enamine promoted cyclization and elimination of hydrogen iodide; (ii) double electrophilic alkylation pathway (first, Friedel-Crafts alkylation between olefin and iodide; second, alkylation of enamine by the resulting carbocation, Fig. 2); (iii) radical mechanism by homo-cleavage of the C-I bond. We obtained the X-ray structure of a-iodo aldehyde 4c. In solid phase, the double bond is oriented 180°away from the iodine atom, an ideal alignment for SN 2 displacement. Although we could not rule out the possibility of a radical cyclization, attempts to intercept the tertiary radical were unsuccessful. The   The a-cyclopropanation occurred smoothly for substrate 1a using 1 equiv. NIS and chiral amine 3a. A catalytic amount of 3a did not promote this reaction. (c) Amine 3a was found to react with NIS rapidly at room temperature (rt). Benzophenone was found to be the major decomposition product. (d) Owing to serious decomposition of 3a by NIS, a-iodination of isobutyraldehyde requires stoichiometric amine 3a. However, we found that para-nitrobenzoic acid (pNBA) stabilizes this amine, allowing for catalytic use. er, enantiomeric ratio; TMS, trimethylsilyl. TFA, trifluoroacetic acid; NCS, N-Chlorosuccinimide; OTMS, trimethylsilyloxyl. reaction proceeded with equal efficiency in dark and in the absence of any oxidant, suggesting a radical intermediate was unlikely. At this moment, we believe the reaction mechanism is in line with our original design (Fig. 2).
Rationale for the stereochemical control. The electrophilic alkylation is the stereoselectivity determining step, during which the a-iodo iminium intermediate folds into a chair-like conformation. The double bond approaches the a-carbon away from the bulky substituent of the amine and the olefin adopts a pseudo axial orientation in order to avoid steric crowding into the adjacent iminium moiety. This transition state model predicts the absolute stereochemistry of product to be (1R, 5S), which was confirmed by X-ray crystallography of derivative 8c (Fig. 6, vide infra).
Synthetic manipulations of the products. The unique bridgehead aldehyde functionality enabled us to perform various chemical manipulations to access a number of chiral scaffolds (Fig. 7). NaBH 4 reduction of 2b gave the corresponding alcohol 5b in 95% yield. Reductive amination using morpholine afforded amine 6b in 91% yield 58 . Horner-Wadsworth-Emmons reaction of 2c installed an a,b-unsaturated ester at the bridgehead 59 . Seyferth-Gilbert homolygation using the modified Bestmann's condition generated an alkyne functionality at the [3.1.0] bridgehead 60 . Epoxidation using trimethyl sulfonium iodide  afforded 9c as a mixture of diastereomers 61 . The formyl group of 2c was smoothly converted to methyl using Huang's modification of Wolff-Kishner reduction 62 . Deformylation using RhCl 3 resulted concurrent cyclopropane opening to give 3-vinylpyrrolidine 11c in 61% yield. The aldehyde group was retained in an acid promoted cyclopropane fragmentation to afford 3-formyl-4-vinylpyrrolidine 12c. Interestingly, skipped diene 13b was formed via Grob fragmentation 63 when we carried out sulfonylation on alcohol 5b. It is noteworthy that most of these derivatives would be very difficult to prepare otherwise.
Preliminary result for amine catalysis. It was intriguing that this cyclopropanation required 100 mol% of the chiral amine to achieve high conversion. Our earlier results showed that the a-iodo aldehyde could be generated in quantitative yield using 20 mol% amine. The lack of amine turnover in the annulation step was likely due to the generation of HI that interfered with the enamine/iminium equilibrium. Various quenching agents were used in an attempt to speed up catalyst regeneration. However, no improvement was observed using bases and silver salts. On the other hand, moderate turnover can be achieved using pNBA as buffer and substrate 1a in excess (Fig. 8). In-depth investigation on further improving catalyst turnover is underway and will be reported in due course.

Discussion
In summary, we developed the first example of asymmetric intramolecular a-cyclopropanation of aldehydes using a chiral amine promoter. We discovered that the a-iodo iminium functionality could serve as a donor/acceptor carbene mimetic to engage in an intramolecular annulation reaction with an olefin. This strategy enables direct access to privileged chiral bicyclo[3.1.0]hexane scaffolds. The formation of the threemember ring is believed to occur through a stepwise double electrophilic alkylation cascade. This reaction represents the first asymmetric double a-alkylation of aldehydes. The synthetic utility of the ring-fused cyclopropane products is demonstrated by diversified chemical derivatization. We expect the use of a-iodoaldehyde as a versatile donor/acceptor warhead will ARTICLE become a general approach for asymmetric di-functionalization of carbonyl compounds.

Methods
General methods and materials. Solvents for reactions were distilled according to general practice before use. All reagents were purchased and used without further purification unless specified otherwise. Amine catalyst was prepared based on the literature report 64,65 . Solvents for chromatography were technical grade and distilled before use. Flash chromatography was performed using 200-300 mesh silica gel with the indicated solvent system according to the standard techniques. Analytical thin-layer chromatography was performed using Huanghai silica gel plates with HSGF 254. Qingdao Haiyang Chemical HG/T2354-92 silica gel was used for silica gel flash chromatography. Visualization of the developed chromatogram was performed by ultraviolet absorbance (254 nm) or appropriate stains. 1  Synthesis of substrates. Synthetic method A for substrates 1a-1b, 1e. Dibenzyl malonate was converted to dibenzyl 2-allyl-2-(3-methylbut-2-en-1-yl)malonate by double alkylation according to a similar reported procedure 66 . To a solution of this diene (5 mmol, 1 equiv.) in tetrahydrofuran (THF) (50 ml) at 0°C, 9-borabicyclo[3.3.1]nonane (1.5 equiv., 0.5 M in THF) was slowly added by syringe over 10 min. The reaction was allowed to warm to room temperature and stirred for 3 h. Then 30 ml of water and sodium perborate tetrahydrate (4.5 g) was added slowly (exothermic). The white suspension was stirred overnight. The mixture was filtered to remove the white solid. The filtrate was poured into a separatory funnel and the layers were separated. The aqueous phase was back-extracted three times with 50 ml portions of ether. The combined organic layers were washed with 50 ml of brine. The resulting solution was dried over sodium sulfate, filtered and concentrated to yield a clear oil, which was further purified by flash column chromatography to provide dibenzyl 2-(3-hydroxypropyl)-2-(3-methylbut-2-en-1yl)malonate. To a flask containing the solution of this primary alcohol in dichloromethane was added pyridinium chlorochromate (1.5 equiv.) and the reaction stirred at room temperature until all the starting material was consumed. Filter to remove the inorganic impurity and the product was isolated by flash column chromatography (ethyl acetate and petrol ether as eluent).