Palladium-catalyzed heteroallylation of unactivated alkenes – synthesis of citalopram †

A palladium-catalyzed difunctionalization of unactivated alkenes with tethered nucleophiles is reported. The versatile reaction occurs with simple allylic halides and can be carried out under air. The methodology provides rapid access to a wide array of desirable heterocyclic targets, as illustrated by a concise synthesis of the widely prescribed antidepressantcitalopram.


Introduction
Palladium-catalyzed heterocyclization onto alkenes holds particular importance for synthetic chemists because of the complexity formed, the generally tolerant reaction conditions, and the prevalence of the heterocyclic products in bioactive targets. 1 The s-alkyl Pd(II) intermediate formed by heterocyclization can participate in a range of subsequent transformations including b-hydride elimination (Wacker-type reaction), alkoxycarbonylation, Heck reaction, or protodemetallation.1j In addition to these pathways, when the heterocyclization is preceded by oxidative addition, a reductive elimination can complete the catalytic cycle. 2 Recently, strategies have emerged that involve oxidation of the s-alkyl Pd(II) intermediate formed by heterocyclization (particularly with hypervalent iodine reagents) 3 as a means of subsequently forming new C-halogen, 4 C-O, 5 C-N, 6 and C-C 6f,7 bonds.
When a Pd(II)-induced heterocyclization onto an unactivated alkene is followed by C-C bond formation, the new C-C bond is typically to an sp 2 -hybridized carbon (e.g. 2 / 3, Scheme 1). 8,9xtending this methodology to form sp 3 -sp 3 C-C bonds would constitute a signicant advance, in part because increasing the sp 3 -C content of preclinical drug candidates has been identied as a potential strategy to improve clinical success rates. 10Pdcatalyzed allylation is a powerful method for C-C bond formation that has seen widespread use in the construction of an impressive array of complex targets; 11 however, this method has never been coupled to the Pd(II)-induced heterocyclization onto unactivated alkenes.In this communication, we report the catalyzed reaction of unactivated alkenes with tethered oxygen or nitrogen nucleophiles and allylic halides to generate a new fully substituted carbon center and a new sp 3 -sp 3 C-C bond.Furthermore, we demonstrate the utility of this method with a synthesis of the widely prescribed selective serotonin reuptake inhibitor (SSRI) citalopram.

Optimization
The combination of alkenyl phenol 5 with the commercially available catalyst Pd(hfacac) 2 was selected for initial screening of the planned heteroallylation based on precedent in other oxypalladation reactions (Table 1).7b Gratifyingly, benzofuran 6 was observed when the reaction with allyl bromide was conducted in THF, though conversion was relatively low aer 72 h (entry 1).Solvent screening revealed that ca.80% conversion could be achieved in PhMe aer 16 h (entry 2).Attempting to use alternative Pd catalysts (entries 3-6), or allylic electrophiles (entries 7, 8) resulted in substantially lower conversions.By replacing allyl bromide with allyl chloride, 9 full conversion was observed in 6 h (entry 9).Decreasing catalyst loading to 5 mol% resulted in increased reaction time, but full conversion was still observed aer 16 h (entry 10). 12It should be noted that the optimized conditions can be conducted under ambient atmosphere using solvent directly as procured. 13From a mechanistic point of view, it is noteworthy that only trace amounts of dihydrobenzofuran 6 formed when using a Pd(0) catalyst (entry 11, vide infra).

Substrate scope
With these optimized conditions in place, we sought to examine the scope of this heteroallylation reaction to make substituted dihydrobenzofurans (Table 2).The unsubstituted case provided the allylation product in 70% isolated yield (entry 1).Substitution ortho to the reacting oxygen and inclusion of electronwithdrawing substituents were tolerated without decrease in yield (entries 2-4). 14Of particular importance is the successful heteroallylation in the presence of an aryl bromide (entry 5), which can serve as a handle for the subsequent introduction of a vast array of functionality.When attempting to utilize a monosubstituted alkene, only trace oxyallylation product was detected, with the major product arising from b-hydride elimination from the putative s-alkyl Pd(II) intermediate (entry 6, vide infra). 15ext, we chose to further probe the scope of the heteroallylation reaction by examining formation of heterocycles other than dihydrobenzofuran (Table 3).Use of the previously optimized reaction conditions with a benzylic alcohol led to a relatively low overall yield (entry 1). 16Fortunately, recourse to allyl bromide resulted in a cleaner reaction prole, with the desired isobenzofuran being obtained in 77% isolated yield (entry 2).Steric hindrance about the unactivated alkene did not prove to be an impediment to the hetroallylation reaction (entry 4).The dihydroisobenzopyran ring system could also be assembled using this methodology (entry 5). 17Use of a substrate containing an unprotected hydroxyl group that was not involved in the heterocyclization resulted in low conversion with allyl bromide; however, using allyl chloride for this substrate resulted in a synthetically viable yield (entries 7, 8).Secondary and tertiary alcohols, as well as benzoic acids also performed well in the reaction (entries 9-14).It should be highlighted that substitution on the unactivated alkene can include sp 3 -C (including a sterically encumbering t-Bu group), or an aromatic ring. 18Once again, when attempting the oxyallylation with a monosubstituted alkene, cyclization followed by b-hydride elimination occurred in relatively low conversion, with none of the allylated lactone being observed (entry 15). 15e then attempted to extend the method to an aminoallylation by cyclization of tosyl amides onto unactivated alkenes (Scheme 2).These efforts were initially plagued by low conversion on gem-disubstituted alkene precursors even at prolonged reaction times with 10 mol% catalyst loading.However, we were successfully able to access the isoquinolone (8b) and pyrrolopyrazinone (8c) ring systems from monosubstituted alkene precursors without signicant b-hydride elimination. 15Modication of the base to KH 2 PO 4 was required for these aminocyclizations in order to suppress N-allylation.

Mechanistic hypothesis
Although the precise mechanism for this heteroallylation reaction is yet to be fully elucidated, we have carried out preliminary studies designed to test the possible intermediacy of a p-allyl species.An experiment using dideuteroallyl bromide labelled at the allylic position resulted in exclusive formation of a product with two vinyl deuterons (Scheme 3). 15  Furthermore, while not denitive, the lack of reactivity using Pd 2 3 , coupled with the fact that precipitation of Pd(0) was not observed during the course of the reactions and the tolerance of the process for aryl halides (which might have undergone oxidative addition to an in situ-generated Pd(0) species) suggests that Pd(0) intermediates are not involved. 13These factors lead us to propose the following catalytic cycle (Scheme 4): Pd(II)-induced heterocyclization (9 / 10) might be followed by carbopalladation of the allyl halide to generate Pd(II)-alkyl complex 11. 1,19Subsequent b-halide elimination (which has been shown to occur more rapidly than b-hydride elimination in several systems 9,20 ) would then afford heteroallylation product 12, while releasing the Pd(II) catalyst. 21This mechanistic interpretation is consistent with results obtained in related cyclizations of allenes.9e,g,h

Synthesis of citalopram
In order to demonstrate the utility of the heteroallylation methodology for the preparation of bioactive heterocycles, and   further probe substrate scope, a synthesis of the SSRI alopram was performed (Scheme 22,23 The reaction of p-uorophenyl Grignard reagent 14 with commercially available cyanophthalide 13 is known from the patent literature.23c,f Subsequent Wittig reaction provided the requisite alkenyl alcohol 15.The pivotal Pd-catalyzed oxyallylation proceeded in good yield to furnish isobenzofuran 16.Conversion of terminal alkene 16 to citalopram was readily achieved onepot dihydroxylation-oxidative cleavage, 24 followed by reductive amination.

Conclusions
In summary, we have developed a versatile catalytic heteroallylation reaction of unactivated alkenes to form heterocycles that are prevalent in a plethora of targets including natural products and bioactive compounds.The process forms a fully substituted carbon center and a new sp 3 -sp 3 C-C bond in a single step, and occurs using a commercially available catalyst under operationally convenient conditions (e.g.under air).The orthogonality of the newly described heteroallylation to standard Pd(0)-catalyzed processes (as illustrated by the tolerance of the reaction for aryl halides, and lack of reactivity with a Pd(0) catalyst) clearly shows the potential for this method to be used in accessing complex targets.Phenols, alcohols, carboxylic acids and tosyl amides are all competent nucleophiles for this transformation, which we have also shown can be employed in a synthesis of the bioactive heterocycle citalopram.Ongoing investigations in our group seek to further expand the scope of the reaction, including ligand-based control of enantioselectivity, and explore the mechanism underpinning the process.

Table 1
Optimization of the Pd-catalyzed heteroallylation