Synthesis of cyclopropanated [2.2.1] heterobicycloalkenes: An improved procedure

ABSTRACT A safer and improved method to our previous report on palladium-catalyzed cyclopropanation of heterobicyclic alkenes has been developed. By using tetrahydrofuran as the solvent and a more dilute aqueous NaOH solution for the generation of diazomethane from Diazald, cyclopropanation could be achieved smoothly with minimal adjustment over the course of reaction. 7-Oxabicyclic substrates with bulky C1 or C2 groups, as well as 2,3-diazabicyclic substrates with various N-substituents, effectively underwent cyclopropanation. Using this methodology, yields to previously reported products were markedly increased, and 10 new cyclopropanated [2.2.1] heterobicyclic products were prepared. In addition, this work accounts for the first reported cyclopropanation of 2,3-diazabicyclic alkenes, which all gave excellent yields of >90%. GRAPHICAL ABSTRACT


Introduction
Cyclopropanes are present in a variety of natural products of plant, fungal, and bacterial origin, [1] and recently their stereoselective synthesis has received much attention in the context of biological activity studies. [2,3] The high ring strain and reactivity of cyclopropanes makes them key intermediates for further transformations and appealing building blocks to more complex structures. [4,5] Numerous methodologies exist for the preparation of cyclopropanated carbobicycloalkenes, [6][7][8][9][10][11] while the cyclopropanation of heterobicycloalkenes has proved more challenging, with only a limited number of publications to date. [12][13][14] Inspired by the work of Miller and Ji, [13] our group has recently succeeded in the first Pd(OAc) 2 -catalyzed cyclopropanation of 7-oxabenzonorbornadienes, demonstrating steadfast exo stereocontrol of cyclopropanation and broad applicability to variously substituted aromatic and C1-substituted oxabenzonorbornadiene derivatives (Scheme 1). [15] In addition, we have reported on several transformations which these novel systems can undergo. [16,17] In our previous report on cyclopropanation, however, we were faced with some procedural challenges: namely, the need to supply additional solvent to the reaction vessel during reaction; incomplete consumption of starting material, potentially due to low solubility of the substrate in Et 2 O; and minor complications with the addition of a viscous 50% w/v NaOH solution over time. The current work documents a safer and improved procedure to our former cyclopropanation methodology, where the two principle modifications described in this paper are (1) change in reaction solvent and (2) dilution of the aqueous NaOH solution used. In addition, the present work expands the scope of the reaction and introduces the first examples of the palladium-catalyzed cyclopropanation of other [2.2.1] heterobicycles, including 2,3-diazabicyclo-5-heptenes, as well as a hetarene-containing 7oxabicyclic alkene. The precursor bicycloalkenes that were cyclopropanated in this study were prepared following literature procedures of [4 þ 2] cycloadditions between either furans or cyclopentadiene with the appropriate dienophile. [18][19][20][21][22][23]

Discussion
Upon our attempt to cyclopropanate compound 1d (Z ¼ tetramethylsilane [TMS], Scheme 2) we found that the substrate did not dissolve well in Et 2 O, which was the solvent previously employed for these reactions. Probing the literature, it appeared that the use of other solvents was not uncommon in reactions employing diazomethane. [24,25] After screening the solubilities of substrates in some of these solvents, it appeared that tetrahydrofuran (THF) in fact did a better job of solubilizing several of the alkenes. Use of THF suggested another potential benefit: Because it is higher boiling than Et 2 O, evaporation could be minimized and thus periodic addition of solvent by syringe, which could potentially scratch the glassware or expose the chemist to small amounts of gas generated, could be avoided. We also found that by using a more dilute 25% w/v solution and increasing the dropping rate, the addition was smoother and could be left to be monitored less frequently. Accordingly, the glassware dimensions were modified slightly, as well ( Figure 3).
These modifications led to improved yields in compounds 2a-c, which were presented in our former report, [15] as well as successful production of seven new cyclopropanated oxabicycloalkenes 2d-j, each giving good to excellent yields. The substrates bearing bulky C1-substituents including tert-butyl and trimethylsilyl underwent cyclopropanation, providing appreciable yields of 85% and 75% of cyclopropane 2c and 2d, respectively. Cyclopropane 2e with both bridgehead and arene substitution also gave a good yield (82%), as did cyclopropane 2f with a long-chain bridgehead ethyl ester substituent (81%). Unlike all other cyclopropanes, which showed a singlet for bridgehead proton H a , cyclopropane 2f showed a doublet at 5.04 ppm with J ¼ 2.3 Hz. 1 H NMR coupling constants of bridgehead proton H a to cyclopropyl proton H b allowed us to deduce the dihedral angle of H a -C-C-H b , and the stereochemistry for each cyclopropane (Fig. 1). For the exo isomer, J values of 0-2 Hz are expected, whereas for the endo isomer a dihedral angle of ∼42°and a coupling constant of ∼5 Hz is expected. [26][27][28] The coupling seen in the present study for 2f is sufficiently close to J ¼ 2 Hz that the cyclopropane was concluded to be exo with respect to the [2.2.1] bicyclic framework, and this was even more obvious with all other cyclopropanated Scheme 2. Cyclopropanation of C1-, C2-, and arene-substituted 7-oxabicycloalkenes 1a-j. a Isolated yields after column chromatography. b Isolated yields reported in former study (Et 2 O solvent).
oxabicyclics, which showed no observable splitting for the peak of H a . In addition, tetradeuterated cyclopropane 2 g was obtained in good yield of 85%, comparable to the undeuterated parent compound 2a.
Next, cyclopropanations on substituents at the C2-position were attempted, to examine the effect of substituent proximity to the cyclopropanation site. Although both C2-electron-donating and electron-withdrawing substrates 1h and 1i required a larger amount of diazomethane for cyclopropanation (6 equivalents for 1h and 8 equivalents for 1i), each alkene underwent cyclopropanation to afford the corresponding exo cyclopropanated product in desirable yields (94% and 72%, respectively). The yield observed for product 2i was in good agreement with a known cyclopropanation of a similar C2-substituted 7-oxanorbornene (lit. 70% for two steps). [14] Finally, substrate 1j with a fused pyridyl ring also underwent cyclopropanation successfully, suggesting that the scope of this work could be further broadened to other hetarene-containing bicyclic substrates.
By repeating Miller and Ji's cyclopropanation using the current modified conditions on 2-oxa-3-azabicyclic alkene 3, [13] we were able to reduce both the quantity of diazomethane (from 8 equivalents to 2.6 equivalents) and catalyst loading (from 5 mol% to 1 mol%). Furthermore, we showed that the reaction proceeded equally well in THF as in Et 2 O (Scheme 3), giving an isolated yield of 97% relative to the previously reported 96%. As there was no precedent of 2,3-diazabicyclic alkene cyclopropanation, we decided to examine these substrates, as well.
Not surprisingly, cyclopropanation of bicyclic hydrazines 5a-c bearing various N-substituents all gave excellent yields of cyclopropanes 6a-c ( Table 1). Because of the highly fluxional nature of the nitrogen-containing framework of 6a-c, peaks were broadened with lower intensity in 1 H and 13 C NMR experiments. Thus, although the presence of 4 J-coupling (W-coupling) between H a and H b could not be verified, irradiation of H c in a selective gradient nuclear Overhauser effect spectroscopy (NOESY) experiment on 6a unambiguously showed positive enhancement for the peak belonging to H aʹ (as well as for the peak of H cʹ ), proving exo stereochemistry of the cyclopropane ring (Fig. 2). [29] As with compound 4, it is conceivable that the products 6a-c obtained in this work could be ring opened to afford nucleoside derivatives, which are key intermediates for the development of antiviral and antitumor agents. [13]  The mechanism of palladium-catalyzed cyclopropanation of alkenes is believed to occur in a stepwise manner, [30] which is consistent with computational work at the B3LYP level of theory. [31] The Pd(OAc) 2 precatalyst is reduced to its active Pd(0) state (A, Scheme 4) which then coordinates to the reactant alkene in a bridged η 2 fashion. This initial palladium complex is in equilibrium with the diazomethane-bound intermediate 7, which releases dinitrogen in a rate-determining step (B). Interaction of a second equivalent of alkene with carbene complex 8 promotes rearrangement to palladacyclobutane 9, which undergoes reductive elimination (C) to afford the cyclopropane product. In reality, the coordination modes to palladium are thought to be more complex with reversible ligand exchange and largely depend on alkene structure and reaction temperature. Based on Straub's work, [31] it is quite possible that the strained heterobicyclic alkenes in this study react via dialkene carbene complexes (initially coordinated to 2 equivalents of alkene), which later intramolecularly rearrange to palladacyclobutanes 9.
In conclusion, the present revision of our cyclopropanation methodology was used to create 14 cyclopropanated [2.2.1] heterobicycloalkenes, ten of which were novel compounds and four showing equivalent or improved yields to previous preparations. All cyclopropanes were obtained in good to excellent yields (65-98%) and the exo stereoselectivity for each cyclopropane was confirmed through NMR experiments. The current approach offers a safer reaction setup and is applicable to a broad class of heterobicycles, including hetaryne-containing substrates and 2,3-diazenes. Subsequent transformations of the cyclopropanes and expansion of the reaction scope will continue in our laboratories.

Experimental
All reactions were carried out using a continuous flow apparatus under an inert atmosphere. Commercial reagents and catalysts were used without further purification. Column chromatography was performed on 230-to 400-mesh silica gel following standard flash column chromatography techniques, [32] and analytical thin-layer chromatography (TLC) was performed on precoated silica gel 250 µm 60 F254 aluminum plates, visualized by ultraviolet (UV) light and p-anisaldehyde stain. Infrared (IR) spectra were obtained as NaCl or KBr discs on a Nicolet 380-FTIR spectrophotometer. 1  at a rate of 1-2 mL/min, maintaining efficient stirring and bubbling. Formation of the light yellow CH 2 N 2 gas was observed with the dissolution of Diazald. Upon complete dissipation of any yellow color (4-8 h), the reaction was assessed by TLC for completion. Once the reaction was seen to be complete by TLC, both septa [E] were removed and the apparatus was left to vent any trace CH 2 N 2 (8-16 h). Reactor [C] was removed and its contents were poured over Celite. The filter cake was washed with several portions (4 � 10-20 mL) of Et 2 O, and then concentrated and purified by column chromatography (hexanes/ethyl acetate mixture). -7-N-butoxycarbonyltricyclo[3.2.1.0 2

Funding
This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC).