|Home | About | Journals | Submit | Contact Us | Français|
Seleniranium ions at low temperatures (-90 to -78 °C) will initiate effective Friedel-Carfts cyclization if a suitably placed arene is allowed to react even when the arene is unactivated. These intermediates generated from N-aryl-N-tosylamides, undergo a novel, surprisingly efficient, detosylative cyclization to form 5- or 6-membered nitrogen heterocycles. A debenzylation route is preferred if both benzyl and tosyl groups are present in the substrate.
Electrophilic cyclizations of alkenyl carboxylic acids, alcohols, amines, amides and functionalized dienes initiated by seleniranium ions have been broadly applied for the syntheses of diverse heterocyclic and carbocyclic compounds.1 Intermolecular additions of carbon nucleophiles including electron-rich aromatic derivatives to seleniranium ions are also known and these reactions have been developed to a level that it is now possible to carry out highly diastereoselective additions, using enantiopure selenium reagents.2 Even though mechanistically related intramolecular additions of carbon nucleophiles to putative seleniranium ion intermediates were among the earliest examples of selenium-induced cyclizations,3 subsequent developments in this area have been sporadic. In one such rare example, Déziel in 19984 reported that 4-arylbutenes (1) with electron-rich aryl groups undergo competitive carbocyclization when the reaction is conducted in a mixture of CH2Cl2 and methanol (Scheme 1). The major side product, the β-methoxyselenide, 4, is readily converted into the cyclic product 3 by protic or Lewis acids. Formation of the methoxylated product 4 cannot be avoided since in the absence of methanol the reaction gave poor yields. Yet another limitation of this potentially powerful Friedel-Crafts cyclization is that non-activated arenes (e. g., phenyl) do not participate in this reaction.3j,4,5a
While searching for a general route to 1-methylenetetralin and analogous heterocyclic compounds in connection with our asymmetric hydrovinylation approach to 2,3-pyrrolidinoindoles (Figure 1), we decided to revisit this area. We expected the alkene-forming elimination reaction to be more facile via a selenide as compared to an iodide5 or sulfide6 arising from alternate cyclizations, especially for the synthesis of the more sensitive N and O containing heterocycles. Several novel observations that were made during the course of these investigations form the basis of this paper.
Our studies (Eq 1, 2; Table 1) started with cyclizations of two prototypical substrates, but-3-enyl 4-methoxyphenyl ether (5a) and N-phenyl-N-(but-3-enyl)-4-toluenesulfonamide (7a). A number of studies in the literature have suggested that counterions and additives in the reaction medium significantly affect the reactivity and selectivity of the reactions of seleniranium ions.7 For example, in a study that is highly pertinent to the present work, it has been reported that PhSeCl in the presence of a silver salt does not effect the cyclization of 4-aryl-1-butenes,5a even though such reactions are known to proceed in moderate yields using combination of N-(phenylseleno)succinimide and Lewis acids3j as long as the arene is activated and electron-rich.
We decided to explore the effect of alternate counterions and reaction conditions on the cyclizations. Initial scouting experiments to identify the optimal reaction conditions revealed that the best yields are obtained when a combination of PhSeBr and a silver salt in anhydrous CH2Cl2 is employed. Thus the but-3-enyl aryl ether 5a gave the expected product 6a in 85% isolated yield when the reaction was initiated at −90 °C, and then stirred for 8 h at −80 °C (entry 1, Table 1). Generation of the seleniranium ion and its subsequent reaction at low temperature are critical for success of this reaction. Entry 2 shows a reaction that was allowed to proceed only for a shorter period, resulting in a much lower yield. Entry 8 is a related example using the substrate 7a. Presumably, alternate modes of reactivity of the seleniranium ions have higher energies of activation and they are not competent at low temperatues where the slow cyclization ensues. Under the optimized reaction conditions 3.1 equivalents of the silver salt is used. Highly reproducible results were obtained when the crude reaction mixture was stirred with methanol for 60 minutes at the end of the reaction, presumably resulting in the decomposition of any silver complexes and/or salts that might be formed. We have noticed that skipping the methanol quench results in significant loss of material during isolation and purification by column chromatography (see entries 6 vs 7). As shown in entries 3 and 4, other silver salts such as AgBF4 and AgOTf, also work, giving slightly lower yields of the product(s). Reactions carried out using PhSeCl and Ag salts gave lower yields (entry 9). Silver ions are known to participate in reactions of alkenes including cyclization reactions.8 To rule out the unlikely involvement of Ag(I)-alkene intermediates in these reactions, the cyclization of 7a was carried out using putative seleniranium ion that was generated from PhSeBr and sodium tetrakis-[3,5-di-(trifluromethyl)phenyl] borate (Na BARF).9 The product distribution is not markedly different from the AgSbF6-mediated reaction (entry 10), ruling out any special role for Ag(I) in these reactions. Note that the but-3-enyl-N-sulfonamide 7a also gives varying amounts of an unexpected side product 9a, in which the sulfonyl group has been removed concomitant with the generation of a pyrrolidine nucleus (Eq 2, entries 5-10).
Cyclization of but-3-eneyl phenyl ether 5b proceeds to give a nearly quantitative yield of the 6-exo-cyclization product 6b (entry 11). As noted before, significantly higher yields are obtained when more than stoichiometric amounts of AgSbF6 are used in the reaction (entries 11 vs 12).
In the most significant departure from previous studies, under these conditions, even unactivated arenes participate in the cyclization event. Thus 4-phenylbut-l-ene 10a and a dimethyl analog 10b undergo the cyclization under the standard conditions via a 6-endo-trig cyclization mode (Eq 3). BOC-protected N-allylaniline 12 also follows a similar course giving a single product (13) in 74% yield. Loss of the BOC protecting group under these highly electrophilic conditions is no surprise.
Existence of a discrete long-lived phenylseleniranium ion under these reaction conditions7a can be inferred from the reaction of 4-allylanisole 14 (Eq 5), a substrate that is resistant to cyclization because of stereoelectronic considerations. Upon treatment with PhSeBr and AgSbF6 under the conditions similar to the cyclization (2 h at -90 °C. then 8 h at −80 °C), followed by addition of methanol, 14 yields the expected methoxyselenated compounds 15 and 16 in 41% and 27% yields respectively. Similar products from allylbenzene have been observed under more standard selenoetherification conditions.10
N-(but-1-enyl)-N-methylaniline 17 failed to undergo the cyclization even though cyclization of non-aromatic homoallyl amines have been reported to give either an azitidine or a pyrrolidine upon treatment with PhSeX at room temperature.11 Allyl ethers 18a and 18b gave products arising from allylic-O cleavage. Allylic amine derivatives 19a and 19b gave low yields of the expected cyclization products 20a and 20b (29% and 39% respectively).
Since cyclizations of secondary amine substrates carrying Tos-1h,12 and BOC-13 protecting groups on nitrogen proceed without incident, we were surprised to find the partial loss of the toluenesulfonyl group in the cyclization of 7a to 9a (Eq 2). Even though detosylation14 and dissociative migration of tosylates15 have been observed under strongly electrophilic conditions, to the best of our knowledge, this detosylative cyclization represent a novel transformation. We wondered whether placing an electron-withdrawing groups on the aryl ring would divert the reaction away from the Friedel-Crafts cyclization. Indeed in substrates carrying a nitro or a trifluromethyl group in the arene, the Friedel-Crafts pathway is completely shut down and all cyclizations now proceed through the detosylative pathway (Eq 2, 7b –> 9b). Other examples of this transformation are shown in Table 2. The efficiency of the detosylative cyclization depends on the length of the alkene tether and nature of other substituents on nitrogen. While the N-but-3-enyl derivatives (e. g., 7b and 24) are generally sluggish (~ 8 h at -80 °C) in 5-endo-trig cyclizations, the N-pent-4-enyl derivatives (21 and 26) undergo more facile reactions giving both the 6-endo and 5-exo cycliztion products (entries 2 and 4) in respectable yields. With a pent-4-enyl substituent, a deactivated aromatic nucleus is not needed for an effective the detosylative cyclization (entry 5). The expected Friedel-Crafts cyclization (compare to 7a, Eq 2) would yield a 7-membered ring by exo cyclization, and presumably this reaction is not competitive with the piperidine formation. As a result the heterocycle 30 is formed in an astonishing 96% yield! If the substrate carries N-benzyl and N-tosyl substituents as in 31 and 33, exclusive loss of the benzyl group is observed in the formation of the nitrogen heterocycle.16
In summary, here we disclose new reactions involving seleniranium ions for intramolecular cyclizations. New protocols for the Friedel-Crafts cyclizations of non-activated arene-alkenes and novel detosylative cyclizations of tertiary N-p-toluenesulfonamides are described.
Financial assistance for this research by NSF (CHE-0610349) and NIH (General Medical Sciences, R01 GM075107) is gratefully acknowledged.