Iridium‐Catalyzed Enantioselective Intermolecular Indole C2‐Allylation

Abstract The enantioselective intermolecular C2‐allylation of 3‐substituted indoles is reported for the first time. This directing group‐free approach relies on a chiral Ir‐(P, olefin) complex and Mg(ClO4)2 Lewis acid catalyst system to promote allylic substitution, providing the C2‐allylated products in typically high yields (40–99 %) and enantioselectivities (83–99 % ee) with excellent regiocontrol. Experimental studies and DFT calculations suggest that the reaction proceeds via direct C2‐allylation, rather than C3‐allylation followed by in situ migration. Steric congestion at the indole‐C3 position and improved π–π stacking interactions have been identified as major contributors to the C2‐selectivity.


Introduction
Indole is ac ore structural element in many natural and synthetic organic compounds that possess awide diversity of important biological activities. [1] Allylation of the indole core is af undamental transformation, integral to numerous synthetic processes. [2] Consistent with the innate reactivity of indoles,i ntermolecular asymmetric allylation of the C3position of indole is aw ell-established process. [3] Thea nalogous N1-allylation is relatively less well-explored, although there exist various innovative methods for this recently developed by the groups of Hartwig, [4] You, [5] Krische [6] and others. [7] To the best of our knowledge,there are no methods capable of directly allylating indole at the C2-position in an enantioselective,i ntermolecular fashion.
Most intermolecular indole C2-allylation strategies involve either directed lithiation [8] (by lithium-halogen exchange or deprotonation) of the C2-position (Scheme 1, a), or C À Ha ctivation orchestrated by ad irecting group (typically attached at the N1-position) (Scheme 1, b). [9,10] Despite the numerous merits of these approaches,t heir dependencyo n directing groups is an obvious drawback, requiring additional installation and removal steps.F urthermore,t hese strategies typically result in the formation of achiral, linear allylated products,and the rare examples that form branched products generally require more forcing reaction conditions,are often low yielding,and are all racemic. [9c-g] Intramolecular reactions can overcome the regioselectivity issues associated with intermolecular indole reactivity. [11] Several asymmetric indole C2-allylation procedures have been developed following this approach. [12] However,t hese strategies almost always lead to the formation of annulated C2-allylated indoles.Asearch of the literature revealed that the sole exception to this generalization comes from Tambar et al. [13] who developed an innovative intramolecular route to branched, highly enantioenriched C2-allylated 3-amino indoles employing an enantioselective aza-Claisen rearrangement (Scheme 1, c).
In this study,w edemonstrate the successful implementation of as trategy to access enantioenriched C2-allylated 3substituted [14,15] indoles 3 via the intermolecular allylic substitution of branched allylic alcohols 1 and readily available indoles 2,c atalyzed by ac hiral Ir-(P,o lefin) complex [16] and aLewis acid additive (Scheme 1d). At the start of this study, gaining effective control over the regioselectivity of the allylic substitution step with respect to the indole (especially C2versus C3-substitution) was expected to be ak ey challenge. This motivated our decision to explore the use of Lewis acidic additives,a sw ep ostulated that under these reaction conditions,allylation of indole 2 at either its C2-or C3-position [17] would result in the convergent formation of the desired C2allylated product 3;e ither,i ndole 2 could react with the pallyl iridium complex via the less sterically hindered C2position, to form allylated product 3 directly,oralternatively, it could react via the more electron-rich C3-position but then undergo an in situ stereospecific migration. [18] In this latter scenario,the Lewis acid (that is essential for the activation of allylic alcohol 1)w ould also help to promote the required stereospecific migration. Mechanistic and computational studies (see below) suggest that direct C2-allylic substitution is the dominant pathway in the cases tested, but crucially, because both mechanistic pathways converge to the same product 3,t his means that effective C2-allylation can be achieved even in cases in which C3-allylic substitution competes.

Results and Discussion
Our investigation started with the identification of suitable reaction conditions (see Supporting Information for full details), including finding aL ewis acid capable of performing up to three key roles within the allylationmigration cascade:1 )toa ctivate the allylic alcohol towards formation of the Ir-p-allyl complex, 2) to facilitate the enantioselective C2-or C3-allylation whilst avoiding competing N1-allylation, and 3) to facilitate the stereospecific migration of the allyl group from the C3-to the C2-position of indole if needed.
Remarkably,i nexpensive Mg(ClO 4 ) 2 (which has never previously been used in iridium-catalyzed allylic substitution) was identified as the best Lewis acid for the reaction between phenyl allylic alcohol 1a and 3-methyl indole 2a,enabling the formation of C2-allylated indole 3a in 99 %yield and 98 % ee ( Table 1, entry 1) when used in combination with [Ir(cod)Cl] 2 and the (S)-Carreira ligand (L1;s ee Scheme 2f or its structure). [19] When exploring the generality of this reaction we recognized that increasing the bulk at the C3-position of the indole reaction partner (e.g. 2b)resulted in alower yield due to the formation of ak etone side-product 4 (entry 2), however, this problem was easily rectified by aminor increase in the indole equivalents (1.1!1.3 equiv,e ntry 3). We also found that in some cases,asmall increase in the reaction temperature to 40 8 8Cw as needed to ensure complete conversion into the allylated products (entries 4a nd 5), and for consistency, these conditions (entry 6) were taken forward into the substrate scoping phase of the work, [20] which is summarized in Scheme 2.
Substituents were tolerated at all positions around the phenyl ring of the allylic alcohol partner (Scheme 2) including electron-deficient (di-nitro 3j)a nd electron-rich (trimethoxy 3k)aromatics.Enantioselectivity was universally high and the branched isomer product was formed exclusively in all of these examples.I tw as also possible to use heteroaryl allylic alcohols (3m-3o), although there was some erosion of the linear:branched regioselectivity when using indole allylic alcohol (3o).
Awide variety of substituents were well tolerated at the C3-position of indole including alkyl, aryl, benzyl and allyl as well as halide functionality [21] at the C3-, C5-and C6-positions of indole,a ll providing their corresponding C2-allylated products in excellent yields and enantioselectivities.T he lower yield obtained for 3-phenyl-substituted allylated product 3p was due to the competitive formation of propiophenone side-product 4;the extent of side-product formation was reduced as steric bulk decreased at the C3-position (phenyl ! benzyl ! allyl) and could also be suppressed by using increased indole equivalents (see earlier optimization). Alkyl tethers incorporating afree alcohol (3w), aprotected alcohol (3x)a nd amine functionality (3y-3aa)a tt he C3-position of indole were all well tolerated. N-Methyl indole was asuitable substrate,furnishing C2-allylated product (3ab)in72% yield and 84 % ee. Indole 3ac was also formed from at rimethoxybenzene-based allylic alcohol in good yield and excellent enantioselectivity;i ts structural assignment is supported by X-ray crystallographic data, from which the absolute stereochemistry of all other substrates was assigned by analogy. [19] It should be noted that when carbonyl groups were directly attached to the C3-position of indole,C 2-allylation was unsuccessful, [22] which is not altogether surprising,g iven that the nucleophilicity of indole is significantly reduced upon substitution with such electron-withdrawing groups.
Predictably,aC3-substituent is required on the indole to achieve selective C2-allylation, and in the absence of this substituent, allylation takes places exclusively at the C3position;f or example,t he formation of products 5a-h from indole 4 (Scheme 3). Notably,these products were all formed in good to excellent yields and excellent enantioselectivities, [23,24] reflecting the mild nature of our reaction conditions relative to previous methods. [3a-c] Indeed, this represents the most enantioselective method for the synthesis of these fundamental scaffolds to date.

Mechanistic Studies
Next, aseries of control reactions were conducted to help elucidate the roles of each reaction component (Table 2). Product 3a was not formed in the absence of Lewis acid (entry 2), clearly demonstrating its necessity for reactivity. Furthermore,t he nature of the Lewis acid, both the cation and anion, plays akey role in C2-vs.N1-selectivity,aswell as the enantioselectivity of the reaction (see Lewis acid optimization screen in SI). In the absence of the iridium catalyst  (entry 3) or the ligand (entry 4) am inor amount of product was formed, most probably via Lewis acidic activation of the allylic alcohol, as no asymmetric induction was observed.
Next, to help understand whether the reaction proceeds via direct C2-allylation or initial C3-allylation followed by migration, as eries of selective migration experiments were designed. First, C3-allylated indole 5a was reacted with electron-poor, p-nitrophenyl allylic alcohol 1e (Scheme 4, a). Thei dea was that if the reaction proceeded via initial C3allylation, dearomatized intermediate 6 would first form, which would presumably be followed by migration of the more electron-rich allyl substituent, in view of its higher migratory aptitude,t hus furnishing C2-allylated product 7a. However,ifthe reaction progressed via direct C2-attack, C2allylated product 7b would instead be formed. Theo utcome of this reaction was clear;t he sole isolation of indole 7b [25] (confirmed using NOE experiments), provided strong support that, in this case,t he reaction proceeded via the direct C2-pathway.S electivity between electronically similar substituents was next investigated (Scheme 4, b), via the reaction of C3-allylated indole 5a with deuterated allylic alcohol 1r. Due to the very similar electronic nature of the two allyl groups,amixture of products was expected if the reaction proceeded via C3-C2 migration, but in fact, bis-allylated product 8a was selectively formed (confirmed by NOE experiments), indicating again that the reaction proceeded via the direct C2-pathway.T he same direct C2-allylation pathway was also observed in the reaction of linear C3allylated indole 9 with phenyl allylic alcohol 1a (Scheme 4, c).
With experimental evidence supporting ad irect C2allylation pathway obtained, we then turned to DFT to gain adeeper understanding of the reaction mechanism ( Figure 1). All possible transition states for intermolecular C2-or C3allylation between indole 2a and p-allyl complex 11 were calculated. Them ost stable transition states for direct C2attack (TS-2,0 .0 kcal mol À1 )a nd direct C3-attack (TS-3, 2.6 kcal mol À1 )are represented in Figure 1. Thelower energy of TS-2 suggests adirect C2-allylation pathway.
Theinnate electronic property of indole favors C3-attack, which is exhibited by the more negative NPA( natural  population analysis) charge at C3 of TS-3 (À0.11) compared with that at C2 of TS-2 (À0.03). This property is supported by the direct C3-allylation of indoles not possessing aC 3substituent (Scheme 3). However,t he preference for direct C2-attack, by virtue of the more stabilized TS-2 over TS-3, stems from the compromise of several competitive effects: (1) the C3-methyl substituent causes unfavorable steric congestion due to the close non-bonding hydrogen atom pairs in TS-3 [B(H2···H5) = 2.15 and B(H2···H3) = 2.29 ]a nd (2) TS-2 enables superior p-p stacking between the electron-rich indole ring and the phenyl group of the electrophilic cinnamyl moiety,b oth of which contribute to the lower barrier of the direct C2-allylation pathway (see the Supporting Information for more details). [26] Finally,toexamine how p-stacking,identified in the above DFT studies (illustrated in Figure 1a,c), influences the regioselectivity of the reaction experimentally,a lkyl allylic alcohol 1p,w hich is unable to undergo aryl p-stacking,w as reacted under the standard conditions.T he reaction afforded an ear equal mixture of the C2-product 12 a and N1-product 12 a' ' (42 %and 41 %yield, respectively,Scheme 5) suggesting that the aryl p-stacking plays akey role in controlling the C2regioselectivity when reacting aryl allylic alcohols.B oth products were formed with high enantioselectivity. [27] Thus, at present we are only able to achieve high levels of C2regioselectivity using aryl substituted allylic alcohols.N onetheless,wewere pleased to observe that allylic substitution is still possible under our usual reaction conditions with aliphatic allylic alcohol 1p,g iven that p-allyl complex formation is known to be more challenging for such alcohols compared with the more activated benzylic systems that are the main focus of this study. [28] On the basis of the experimental data described above,in addition to previously described theoretical evidence, [29] the following mechanism for the enantioselective direct C2allylation of indole is proposed (Scheme 6). Thec atalyst precursor [Ir(cod)Cl] 2

Conclusion
In summary,ahighly enantioselective,d irecting groupfree intermolecular C2-allylation procedure has been demonstrated for the first time,f urnishing aw ide range of C2allylated products in excellent yields with high regiocontrol. Ac ombination of experimental migration studies and DFT calculations suggest the reaction proceeds via direct C2attack rather than C3-allylation followed by in situ migration; this mode of reaction results in greater reaction predictability than might be expected via aC 3-C2 migration pathway,i n which isomeric products could form as aresult of unselective migration. This unprecedented C2-selectivity was achieved due to the combination of several factors:( 1) steric congestion at the C3-position of indole increases the relative reactivity of the C2-and N1-positions;( 2) p-p stacking interactions between the electron-rich indole ring and the aryl group of the p-allyl intermediate increases selectivity for C2allylation over N1-allylation;( 3) as uitable Lewis acid was identified able to activate the allylic alcohol and influence the N1 vs.C 2s electivity,w ithout compromising enantioselectivity.The nature of both the cationic and anionic components of the Lewis acid were shown to be crucial for high selectivity and enantioselectivity.D uring the investigation, indole substrates without aC3-substitutent were also explored using the optimized allylic substitution conditions affording C3-allylated indoles in excellent enantioselectivities,f urther showcas-

Research Articles
ing the mildness and broad suitability of the identified reaction conditions.

General Procedure
To an oven-dried Schlenk tube charged with am agnetic stirrer bar was added [Ir(cod)Cl] 2 (0.016 mmol, 0.04 equiv) and (S)-Carreiras Ligand L1 (0.064 mmol, 0.16 equiv). Ther eaction vessel was purged by alternating vacuum and argon three times before dry CH 2 Cl 2 (2 mL) was added. This mixture was stirred at RT for 15 min to form the active catalyst during which the solution turns from yellow to ad eep red colour. Allylic alcohol (0.400 mmol, 1.0 equiv) was then added followed by the addition of indole derivative (0.520 mmol, 1.3 equiv) and Mg(ClO 4 ) 2 (0.100 mmol, 0.25 equiv) under ab ack pressure of argon. Thereaction mixture was then heated to reflux and stirred for 15 h. Thereaction mixture was directly concentrated on to silica and purifiedb yc olumn chromatography affording the desired allylated product.