Propargylic Alcohol as a Key Substrate Motif for Achieving Enantioselective Gold-Catalyzed Enyne Cycloisomerization

Abstract

Despite successes in achieving asymmetric induction in enyne cycloisomerization, few systems are applicable to each of the typical 6-endo, 5-exo, and 5-endo cyclization/cycloisomerization modes. By appending a synthetically valuable hydroxymethyl group at the alkyne end, an H-bond between the HO group of the propargyl alcohol moiety and a chiral ligand basic group offers a novel asymmetric induction strategy in gold-catalyzed enyne cycloisomerization reactions. Both 1,5-enynes and 1,6-enynes are suitable substrates, and 5-exo, 5-endo, and 6-exo cyclizations lead to outstanding enantioselectivities. As a valuable reactive moiety, the hydroxymethyl group is converted into a versatile aldehyde moiety in the exo cyclization mode or engages in stereoselective cyclizations in a ligand-dictated chemo-divergent process. This strategy may further advance asymmetric gold catalysis.

Graphical Abstract

Over the past years, our laboratory¹ has focused on developing bifunctional biaryl-2-ylphosphine ligands to enable efficient/novel gold catalysis² involving metal-ligand cooperation.³ In this type of cooperative catalysis (shown as A in Scheme 1A), the ligand actively participates in the bond-forming/breaking events in the reactions, resulting in lowered reaction barriers and hence realizing novel reactivities⁴ and/or substantially accelerated reactions.⁵ To expand the utility of these ligands in homogeneous gold catalysis, we surmised that the (Lewis) basic functional group of these ligands could influence reaction outcome and, in particular, induce product chirality in a non-cooperative manner, i.e., a ligand is not directly involved in the bond-forming/breaking process in reactions. To this end, as shown as B in Scheme 1A, we envisioned that an H-bond between a chiral ligand and substrate could serve as an additional anchor⁶ in organizing reaction transition states but does not undergo bonding changes during the reaction, thus enabling asymmetric induction. This approach would permit asymmetric induction in various gold catalysis where metal-ligand cooperative catalysis is not possible. Of note is that biarylphosphine ligands employing ionic interactions⁷ have been successfully employed in achieving asymmetric gold catalysis.

Scheme 1.

Enantioselective gold catalysis enabled by non-cooperative H-bonding interactions

Gold-catalyzed enyne cycloisomerizations⁸ have been extensively studied, resulting in the discovery of a plethora of gold reactivities.^8b They have also served as valuable platforms for developing new chiral ligands⁹ for asymmetric gold catalysis.^2e,10 Despite past successes in achieving chiral induction, to our knowledge, no system is applicable to each of the typical 6-endo, 5-exo, and 5-endo cyclization modes. We surmised, as shown in Scheme 1B, by the substitution of a synthetically valuable hydroxymethyl group at the alkyne end, the 1,5-/1,6-enyne substrate 1 would form an H-bond with a chiral ligand basic group in addition to the metal-substrate binding, as depicted in B’, thereby achieving two-point substrate-catalyst interactions and hence permitting asymmetric access to chiral cyclopropyl gold carbene intermediates C and D via an endo and an exo cyclization pathways, respectively. In both intermediates, the H-bond between ligand and substrate may remain. Herein, we disclose the implementation of this strategy, which permits high levels of enantioselectivities in all three types of enyne cycloisomerizations. In addition, a ligand-dictated reaction divergence with excellent enantioselectivities is realized.

We began by studying the asymmetric cycloisomerization of tosylamide 1,6-enyne 1a. As shown in Table 1, entry 1, by using the binaphthyl-2-ylphosphine ligand (S)-L1 featuring a distal N-pyrrolidinyl carbonyl group, the expected azabicyclo[4.1.0]heptene 2a was formed via a 6-endo cyclization mode in 82% yield and with 39% ee. Modifying the ligand amide directing group in the cases of L2 (entry 2) and L3 (entry 3) did not improve the reaction enantioselectivity. The phosphine oxide ligand (S)-L4 did not lead to higher ee, either (entry 4). Interestingly, when we examined the tertiary amino ligands (S)-L5 featuring a chiral 1-cyclohexyltetrahydroisoquinoline moiety, the product ee value was improved to 60% (entry 5). The trimethyl variant (S)-L6, a better ligand in intramolecular propargylation chemistry,^4c further increased the ee value to 78% (entry 6). It needs to be pointed out that a 1:1 atropisomeric mixture of (S)-L6AuCl was employed here. By increasing the catalyst loading to 10 mol%, the ee was somehow boosted to 82% (entry 7). Decreasing reaction temperature led to better reaction enantioselectivities (entries 8–9). When the reaction was run at −10 °C, 92% ee was obtained, albeit requiring 30 h to reach completion (entry 9). Further lowering temperatures to −20 °C essentially shut down the reaction (entry 10). Changing the Ts group of 1a to Boc or Ms did not improve the reaction (entries 11–12). The reaction catalyzed by JohnPhosAuCl/NaBARF (10 mol%, entry 13) was faster than that using (S)-L6 (entry 7). This phenomenon is consistent with the absence of active ligand participation in bond-forming/breaking events. The slower reaction with (S)-L6 reflects the presence of its basic tertiary amine moiety, which may reversibly bind to the cationic gold center and hence decrease the concentration of active catalysts. Finally, the importance of the HO group in asymmetric induction is confirmed by the reaction of 1a-Bn, which resulted in only 13% ee (entry 14).

Table 1.

Conditions optimization of asymmetric cycloisomerizations^a

entry	1	L (X)	Conditions	Yieldb	eec
1	1a	(S)-L1 (5)	rt, 12 h	82%	39%
2	1a	(S)-L2 (5)	rt, 12 h	72%	34%
3	1a	(S)-L3 (5)	rt, 12 h	80%	32%
4	1a	(S)-L4(5)	rt, 12 h	84%	34%
5	1a	(S)-L5 (5)	rt, 12 h	85%	60%
6	1a	(S)-L6 (5)d	rt, 12 h	85%	78%
7	1a	(S)-L6 (10)d	rt, 8 h	80%	82%
8	1a	(S)-L6 (10)d	10 °C, 24 h	82%	88%
9	1a	(S)-L6 (10)d	−10 °C, 30 h	85%	92%
10	1a	(S)-L6 (10)d	−20 °C, 30 h	<5%	-
11	1b	(S)-L6 (10)d	−10 °C, 12 h	81%	80%
12	1c	(S)-L6 (10)d	−10 °C, 12 h	78%	87%
13	1a	JohnPhos (10)	rt, 3 h	76%	-
14	1a-Bn e	(S)-L6 (10)d	rt, 48 h	76%	13%

^aReaction run in 2-dram vials under Ar.

^bIsolated yield reported.

^cThe ee values were determined by chiral HPLC analysis.

^dA 1:1 atropisomeric mixture of (S)-L6AuCl used.

^eObtained via 0-benzylation of 1a.

With the optimal conditions (Table 1, entry 9) in hand, we explored the scope of the endo cyclization of tosylamide 1,6-enynes by varying the alkene moiety. As shown in Table 2A, with substrates featuring a 1,1-disubstituted alkene, various alkene substituents including ethyl (2e), bulky ^tBu (2f), siloxymethyl (2g), and phenyl (2i) were readily accommodated, affording the azabicyclo[4.1.0]hept-4-ene products with ≥91% ee. On the other hand, decreased enantioselectivities were observed in the cases of a vinyl group (2d, 75% ee) and a vinyl substituent (2h, 87% ee). This asymmetric catalysis was successfully extended to the ethereal 1,6-enyne substrates, and 3-oxabicyclo[4.1.0]hept-4-enes 2j and 2k were formed in 90% ee. Moreover, this strategy also worked with 1,5-enyne cycloisomerization via a 5-endo cyclization mode, delivering the bicyclo[3.1.0]hexenes 2l with 90% ee. It is noteworthy that despite much research in enyne cycloisomerization, highly enantioselective 1,5-enyne cycloisomerization has rarely been documented. The only related cases are a kinetic resolution study¹¹ and one with moderate ee values.¹²

Table 2.

Reaction Scope of 6-endo and 5-endo asymmetric cycloisomerization^a

^aReactions run in 2-dram sealed vials under argon at indicated temperature, isolated yields reported, and the ee values determined by chiral HPLC analysis.

^b−10 °C for 72 h.

^c−20 °C for 24 h.

^dAtropisomeric pure (S_a,S)-L6AuCl used, see SI for its preparation.

^eToluene as solvent.

With substrates featuring an aryl-substituted trans-double bond, we observed ligand-dictated reaction divergence. As shown in Table 2B, with the enynyl sulfonamide substrate possessing a β-styryl moiety, (S)-L6 led to the formation of the bicyclic tetrahydrofuran 2m’ and the desired azabicyclo[4.10]hept-4-ene 2m in a 1:0.15 ratio. The formation of racemic 2m’ was previously reported using a phosphite-gold catalyst¹³ and can be rationalized by the enhanced accommodation of positive charge α to the phenyl group. In our chemistry, it was isolated in 85% yield and with 98% ee. The ee value of the minor 2m is 93%. On the other hand, using the phosphine oxide ligand (S)-L4, the reaction selectivity was reversed, and the ratio of 2m to 2m’ is >10:1. Moreover, 2m exhibited 96% ee. Much to our surprise, the minor 2m’ of 84% ee shows enantioselectivity opposite to that formed by using the tertiary amine ligand (S)-L6. The underlying reason for this contrasting enantioselectivity is not clear and will be probed in the future. The electronic influence of the reactivity divergence was probed by substituting the phenyl group para position with mildly electron-withdrawing Cl (2n/2n’) and mildly electrondonating Me (2o/2o’) groups. As expected, decreasing amounts of the tetrahydrofuran product 2n’ were observed with both ligands. In contrast, 2o’ was formed exclusively in the case of (S)-L6 and as the major product in the case of (S)-L4. The enantioselectivities of the products are all excellent, and the opposing enantioselectivities of 2o’ were again observed. The same phenomena, i.e., ligand-dictated reactivity divergence and excellent enantioselectivities, were observed in the case of replacing the NTs group of 2m/2m’ with an oxygen atom (2p/2p’) or having it removed (2q/2q’).

In addition, although a cyclohexyl group as the alkene moiety in the case of 2r led to a moderate 66% ee, a n-propyl-substituted trans-double bond in the case of 2s resulted in an excellent 95% ee when (S)-L4AuNTf₂ was used as the catalyst (Table 2C).

With a carbon linker between the π systems, 1,6-enynes are known to undergo an initial 5-exo cyclization to form the exocyclic gold carbene intermediate of type D. As shown in Table 3, with the hydroxymethyl-terminated substrates 3a - 3j, the carbene moiety of type D undergoes facile 1,2-C-H insertion to deliver the aldehyde products 4a - 4j upon tautomerization. These transformations constitute new entries in gold-catalyzed enyne cycloisomerization and are highly enantioselective, with product ee values ranging from 90% to 98%. For the examined reaction scope, both CO₂Et (4a) and CO₂Bn (4b) groups worked well as the gem-substituents (i.e., the Z group) at the C4 position, and scaling up the latter case (see SI) did not affect the excellent reaction outcome. With Z = CO₂Bn, the reaction tolerated various substituents at the alkene internal carbon, including Me (4b), sterically bulky ^tBu (4c), TBSOCH₂ (4d), and phenyl (4e). Substitutions on the phenyl ring of 4e (4f-4h), including sterically congesting o-methyl (4g), were tolerated. Terminal substitutions on the C-C double bond were also examined. In the case of a phenyl group, only the (Z)-substrate led to the desired reaction, affording the aldehyde product 4i in 75% yield and with 90% ee from a mixture of substrate geometric isomers (E/Z = 1:6). The reaction of (Z)-3j possessing a methyl substituent on the alkene led to the formation of 4j in 85% yield and with an identical ee. The Z groups in 4a were replaced by BnOCH₂ without affecting the excellent enantioselectivity (4k, 94% ee). Finally, the Z group is a removable phenylsulfonyl group in 4l, and the reaction remained highly enantioselective, albeit in a moderate yield.

Table 3.

Reaction Scope of 5-exo asymmetric cycloisomerization^a

^aReactions run in 2-dram sealed vials under argon, isolated yields reported, and ee determined by HPLC.

^bReaction run at rt for 24 h.

^cDetermined by measuring the dr of the Mosher ester of its reduced alcohol.

To establish the absolute configurations of the products, we oxidized 2a to the corresponding aldehyde 5a, the specific optical rotation of which is similar to that of reported (1R, 6S) −5a (Scheme 2).¹⁴ As such, the stereochemistries of the rest of the bicyclic cyclopropanes in Table 2 were assigned analogously. The absolute configurations of 2n’ were established via XRD studies (CCDC 2371026) and are consistent with those assigned to 2n. For the bicyclo[3.1.0]hexane products 4a-4j, the absolute configurations of 4b are inferred to be (1S, 5R) by converting it to the isopropyl ketone 5b and comparing its specific optical rotation with a literature report.¹⁵

Scheme 2:

Determination of absolute configurations

We performed DFT calculations to understand the stereoinduction and the role of the H-bonding interaction in the reactions of 1,6-enynes. The details are documented in the SI. In the reaction of 1a, two transition states in the initial 6-endo cyclization step, leading eventually to the opposite enantiomers of 2a, were located. Each of them experiences the highest energy barrier in their respective reaction pathways and is turnover-limiting. Throughout this elementary step, strong H-bonding interactions between the substrate HO group and the ligand tertiary amine nitrogen are maintained in both scenarios. As shown in Figure 1, endo-TS1-in leading to the major enantiomer is favored over endo-TS1-out by 0.47 kcal/mol and experiences an earlier transition state as the distance of the forming bond between C(sp) and methylene C(sp²) – 2.270 Å–is notably longer than that of the latter (2.075Å). On the other hand, the P-Au-C1 angle of 166.6° in endo-TS1-out is smaller than that in endo-TS1-in (173.7°) and substantially deviates from the typical linear 180°. In the exo cyclization of the methyl ester counterpart of 3a, the initial cyclization is again turnover-limiting. The transition state exo-TS1-in, leading to the observed major enantiomer, is favored over exo-TS1-out by 2.52 kcal/mol. These transition states are structurally similar in the bond distances of the H-bond and the forming C-C bonds, but the NCI plots of these two structures (see SI) reveal that enhanced van der Waals stabilization in exo-TS1-in arises from closer contact between the ligand tetrahydroisoquinoline aromatic ring and the substrate relative to exo-TS1-out.

Figure 1:

DFT-optimized transition states of 1,6-enyne cycloisomerization at the wB97XD/def2-TZVPP/SDD(Au) level. The distances are in angstroms, and the free energies are adjusted to the reaction temperature and concentration using GoodVibes, incorporating Truhlar’s frequency scaling for improved thermochemical accuracy. For details, please see SI.

In summary, we disclose that a synthetically valuable hydroxymethyl group in the form of a propargylic alcohol is a valuable handle for achieving asymmetric gold catalysis. The HO group of this structural motif engages in an H-bonding interaction with a basic moiety of a chiral bifunctional ligand and provides an additional anchor point for gold-substrate interaction and subsequent catalysis. This two-point catalyst-substrate interaction does not invoke metal-ligand cooperation in bond-forming/breaking events and offers a new approach to achieving inherently challenging asymmetric gold catalysis. In this work, we demonstrated the effectiveness of this strategy by realizing highly enantioselective 6-endo and 5-exo cycloisomerization of 1,6-enynes. In addition, rare asymmetric 5-endo enyne cyclizations are also achieved with excellent enantioselectivities. With substrates featuring an aryl-substituted trans-alkene, divergent reactivities along with excellent enantioselectivities were achieved with different ligands, further highlighting the utility of the strategy. DFT calculations support the critical role of the H-bonding interaction between substrate and ligand and hence substantiate the conceptual design.

Supplementary Material

Detailed experimental procedures, mechanistic studies, X-ray structure, DFT optimized structures, chiral HPLC chromatographs, and NMR spectra.