Highly enantioselective construction of tertiary thioethers and alcohols via phosphine-catalyzed asymmetric γ-addition reactions of 5H-thiazol-4-ones and 5H-oxazol-4-ones: scope and mechanistic understandings

A new method for facile access to enantioenriched tertiary thioethers/alcohols.


Introduction
Organosulfur compounds are important molecular architectures in synthetic organic chemistry 1 and chemical biology. 2 Therefore, it is not surprising that many methods have been devised in recent years to access optically active thiol derivatives. Common approaches include: conjugate addition of thiols, 3 employment of sulfur-containing pronucleophiles, 4 kinetic resolution of racemic thiols, 5 and electrophilic sulfenylation reactions. 6 In this context, the catalytic synthesis of chiral tertiary thiols is a challenging task and remains largely unexplored. 7 Analogously, tertiary alcohol-containing structures are of great importance in the biological sciences and the pharmaceutical industry, 8 and asymmetric synthesis of chiral tertiary alcohols is an intensively investigated area. 9 A few selected biologically important tertiary thiols and alcohols are illustrated in Fig. 1. 10 From the outset of this research, we aimed to devise a versatile catalytic approach that would allow us to access both tertiary thiols and alcohols in an enantioselective manner.
To develop a method for the asymmetric synthesis of tertiary thiol molecules, it seems ideal to employ a readily accessible prochiral organosulfur compound. 5H-Thiazol-4-one and its derivatives, found to be useful in medicinal chemistry, 11 are suitable donors; the acidic protons at the 5-position can be readily removed to facilitate their reactions with various electrophiles. Surprisingly, 5H-thiazol-4-ones have been rarely used in asymmetric synthesis, and there are only three examples to date describing their applications in asymmetric catalysis. Palomo and co-workers have reported the Brønsted base-catalyzed Michael additions of thiazolones to nitroalkenes 12a and a 0 -silyloxy enone, 12b for the synthesis of tertiary thiols. Very recently, Hartwig disclosed an Ir-catalyzed allylation of 5H-thiazol-4-ones to form enantioenriched tertiary thioethers. 12c We envisioned that careful selection of the electrophilic reaction partners and catalytic systems, in combination with the utilization of pronucleophilic 5H-thiazol-4-ones, would lead to the discovery of novel synthetic methods for the asymmetric construction of tertiary thiols. Moreover, given the ready availability of the analogous 5H-oxazol-4-ones, 13 we anticipated that the methodology developed for the thiol synthesis could be easily adapted to include a-oxygenated carboxylate surrogates, thus allowing facile preparation of chiral tertiary alcohols as well.
Our group has been actively investigating asymmetric phosphine catalysis 14 in the past few years. We designed a series of amino acid-based bifunctional phosphine catalysts, and demonstrated their applications in a wide range of asymmetric transformations, including: (aza)-MBH reactions, 15 [3 + 2], [4 + 2], and [4 + 1] annulations, 16 allylic alkylations, 17 and Michael additions. 18 Very recently, we disclosed the utilization of 2,3butadienoates in phosphine-catalyzed enantioselective g-addition reactions. 19,20 To further expand the range of phosphinemediated asymmetric reactions, we envisaged that 5H-thiazol-4ones and 5H-oxazol-4-ones could serve as valuable donors in phosphine-catalyzed g-addition reactions to allenoates (Scheme 1). The chiral heteroatom-containing adducts formed have masked functionalities, and can be manipulated easily to give tertiary thiols/thioethers and alcohols.
In this article, we disclose the rst utilization of 5H-thiazol-4ones and 5H-oxazol-4-ones in phosphine-catalyzed asymmetric g-addition reactions, and the products can be readily converted to optically enriched tertiary thioethers and alcohols. In addition, we have also carried out DFT calculations to gain insights into the reaction mechanism and understand the origin of the stereochemical outcome of the reaction.

Results and discussion
2.1 Phosphine-catalyzed enantioselective g-addition of 5Hthiazol-4-ones In the past few years, amino acid-based bifunctional phosphines have been shown to be very powerful in phosphine catalysis. In this study, readily available L-valine and L-threonine were chosen as the starting chiral skeletons for the preparation of the phosphine catalysts. By installing different hydrogen bond donating groups and introducing various O-silyl protecting groups, we prepared a wide range of amino acid-derived bifunctional phosphines (Scheme 2), which were used for subsequent studies.
We began our investigations by choosing 5-methyl-2-phenylthiazol-4(5H)-one 5a and allenoate 6c as substrates to evaluate the catalytic effects of the phosphine catalysts for the projected g-addition (Table 1). To our delight, all the bifunctional phosphines could effectively promote the reaction.
Among all the L-valine-derived phosphines, sulfonamidephosphine 2a was found to be the most efficient (entries 1-5). L-Threonine-derived phosphine sulfonamide catalysts (2b & 2c) were then employed, and the enantioselectivity of the reaction could be improved to 89% ee (entries 6 and 7). The dipeptide phosphines were found to be less effective (entries 8-11).
Subsequently, we further optimized the reaction conditions by varying the ester moiety in the allenoate structure (Table 2, entries 1-8). Among all the allenoates examined, the dibenzosuberyl ester proved to be the best, and the ee value of the reaction could be further improved to 91% (entry 6). Furthermore, solvent screening revealed that diethyl ether was the optimal solvent, and the desired product was obtained in 97% yield and with 95% ee under the optimized conditions (Table 2, Scheme 1 Construction of tertiary thiols/alcohols via phosphinecatalyzed g-additions of 5H-thia(oxa)zol-4-ones.
Scheme 2 Phosphine catalysts investigated. a Reactions were performed with 5a (0.1 mmol), 6c (0.12 mmol) and the catalyst (0.01 mmol) in toluene (1.0 mL) at room temperature. b Isolated yield. c Determined by HPLC analysis on a chiral stationary phase. entries 9-12). The use of different molecular sieves as additives did not result in a further improvement in enantioselectivity (entries [13][14][15]. In addition, lowering the reaction temperature resulted in a signicant decrease in reactivity coupled with reduced enantioselectivity (entry 16).
Having established the optimal reaction conditions, the substrate scope for the g-addition of thiazolones to allenoates was then evaluated (Table 3). A wide range of 5-alkyl substituted 5H-thiazol-4-ones could be employed, the reaction was insensitive to the length of the alkyl chain, and both linear and branched alkyl groups were well tolerated (entries 1-10). In addition, benzyl and 2-(naphthalen-2-yl)-substituted thiazolones also proved to be suitable substrates (entries 11 and 12).

Enantioselective g-addition of 5H-oxazol-4-ones
With the established protocol for the asymmetric g-addition of 5H-thiazol-4-ones in hand, we next targeted access to the analogous a-oxygenated carboxylate surrogates by employing 5Hoxazol-4-ones. This task could be challenging as reports containing both sulfur-and oxygen-substituted substrates are rare. 12c We hypothesized that the high tunability of our amino acid-based phosphine systems may provide a practical solution to this problem. The same set of phosphine catalysts were screened for the g-addition of 5H-oxazol-4-ones to allenoate 6c, and the results are summarized in Table 4. The best catalyst for the previous addition of 5H-thiazol-4-ones, 2c, only afforded moderate enantioselectivity (entry 1). Switching to the dipeptide phosphines resulted in highly effective catalytic systems. While O-TBDPS-L-Thr-L-tert-Leu-based 3 led to the desired adduct with a slightly improved ee value, O-silyl-D-Thr-L-tert-Leu-derived phosphines offered excellent catalytic effects (entries 9-11). Finally, phosphine 4c was found to be the best catalyst, affording 9a in 95% yield and 76% ee (entry 11).
To further improve the enantioselectivity, we next optimized the ester moiety of the allenoate (Table 5). Among the different allenoate esters, the dibenzosuberyl ester was most ideal, affording the desired adduct in 96% yield and 86% ee (entry 6). Solvent screening identied diethyl ether as the most suitable solvent for the reaction. When the reaction was performed in the presence of 3Å molecular sieves in diethyl ether, the g-addition product was obtained in 97% yield with 92% ee (entry 14).
Under the optimal reaction conditions, the reaction was applicable to a wide variety of 5-alkyl substituted 5H-oxazol-4ones. As shown in Table 6, the length of the alkyl chain can be varied, and both linear and branched alkyl groups can be employed, and high yields and excellent ee values were attainable in all cases (entries 1-10). When the 5-benzyl substrate was used, the enantioselectivity of the reaction dropped slightly (entry 11), which may be due to the unfavourable aromatic interactions induced by the Bn group. The absolute  conguration of the g-addition products was assigned by comparing the optical rotation of a derivative of 9b with the value reported in the literature. 21

Scope of substrates and synthesis of tertiary thioethers and alcohols
Alkynes are common starting materials in organic synthesis, and the reaction here could be extended to alkyne substrates. Alkynoate (6 0 ) could be employed, instead of allenoates, in the gaddition reactions of both 5H-thiazol-4-ones and 5H-oxazol-4ones. Although the reactions were slower, the chemical yields and enantioselectivities were the same [eqn (1) and (2)]. (1) For the thia(oxa)zolone substrates, the inclusion of 5-arylsubstituted thiazol-4-ones and oxazol-4-ones was unsuccessful. Thia(oxa)zolones are known to exist in tautomeric forms as  a Reactions were performed with 8a (0.1 mmol), 6 (0.12 mmol) and 4c (0.01 mmol) in the solvent specied (1.0 mL) at room temperature overnight. b Isolated yield. c Determined by HPLC analysis on a chiral stationary phase. d 3Å molecular sieves were added. e 4Å molecular sieves were added. f 5Å molecular sieves were added. g The reaction was stirred at 0 C for 20 h. thia(oxa)zoles, 22 and the presence of a 5-aryl group makes the enol forms far more dominating. Indeed, the g-addition products were not observed when 5-aryl substituted substrates were used. Instead, O-attack of the tautomeric thiazole/oxazole to allenoates took place, and the corresponding achiral adducts were obtained in high yields (Scheme 3).
The g-addition products obtained possess a tertiary stereogenic center linked to a buried heteroatom, and they are valuable precursors for the convenient synthesis of enantiomerically enriched tertiary thiols/alcohols. As illustrated in Scheme 4, adduct 9b was converted to allyl oxazolone 15a, which was then treated with base to effect a ring opening, leading to the formation of a masked tertiary alcohol 15b in excellent yield. Reduction of the double bond and cleavage of the ester afforded a tertiary a-hydroxy acid derivative 17, which has both an ethyl and a propyl group present in the tertiary alcohol structure. Similarly, thiazolone 7a was transformed to allyl-substituted 18, which was readily converted to an enantiomerically enriched tertiary thioether 19 via a base-catalyzed ring opening. 23

Mechanistic insights and DFT calculations
Despite the popularity of phosphine-catalyzed organic reactions, mechanistic investigations remain very limited; a few theoretical studies appearing in the literature were disclosed by the groups of Yu, 24a-c Kwon, 24d,e and others. 24f-h We hypothesized that the hydrogen-bonding interactions between the Brønsted acid moiety of the phosphine catalyst and the donor molecules were essential for inducing enantioselectivity in our early reports on bifunctional phosphine catalyzed Michael and gaddition reactions, 18,20 and we believe such interactions are also crucial in our current reaction systems. The mechanism of the g-addition of 5H-thiazol-4-one to allenoate is shown in Scheme 5, which follows the general mechanism described in the literature for g-addition reactions. 19,20 The reaction is initiated by the nucleophilic attack of the phosphorus atom on allenoate to form intermediate B, which is weakly basic. Deprotonation of the donor 5H-thiazol-4-one by B then affords the corresponding enolate, which subsequently attacks the g-carbon of the allenoate to give intermediate E. Proton transfer takes place to afford 18, 20 and this is followed by the elimination of the phosphine catalyst to furnish the nal addition product. We propose that a hydrogen bonding interaction between the sulfonamide N-H and the thiazolone enolate dictates its addition to the C-C double bond, which is the key step for asymmetric induction.
In an effort to provide theoretical support for the proposed reaction mechanism and to rationalize the origin of enantioselectivity, detailed density functional theory (DFT) calculations were conducted. DFT methods, as implemented in the Gaussian 09 (ref. 25) program, have been employed to study the model reaction involving reactants 5H-thiazol-4-ones 5a, allenoate 6c and catalyst 2c. All the stationary points were optimized at the B3LYP 26 /6-31G (d) level 27 of theory. The vibrational frequencies were computed at the same level of theory to determine whether the optimized structure was at an energy minimum or a transition state and to evaluate the corrections of enthalpy and Gibbs free energy. Solvent effects were computed by the IEFPCM 28 solvation model at the M11 (ref. 29)/6-311+G (d) 30 and B3LYP-D3 (ref. 31)/6-311+G (d) levels of theory using the gas phase optimized structures. The conclusions are similar for both methods, and the B3LYP-D3 calculated Gibbs free energies in toluene are discussed in the text.
The calculated Gibbs free energy proles for the phosphinecatalyzed g-addition of 5a to 6c are summarized in Fig. 2 (blue line). As proposed, the reaction is initiated by the nucleophilic attack of the phosphine catalyst 2c on allenoate 6c via a transition state Ts1 with a barrier of 19.8 kcal mol À1 . This process is facilitated by the NH/O hydrogen bond, which brings the phosphine and the allene groups into proximity. A zwitterionic intermediate B is rst formed, endothermically and reversibly, Scheme 3 Reactions of 2,5-diphenyl-thiazol-4-ol 10 and 2,5diphenyl-oxazol-4-ol 12 with allenoate 6c. Scheme 4 Elaboration of the g-addition adducts into enantioenriched tertiary alcohols/thioethers.
with an overall barrier of 9.4 kcal mol À1 . Subsequent proton transfer between intermediate B and reactant 5a takes place via transition state Ts2 with a barrier of 14.4 kcal mol À1 (an overall barrier of 23.8 kcal mol À1 ). The nucleophilic attack can then occur via two possible pathways: the Re-face attack occurs through transition state Ts4-Re with a barrier of 7.8 kcal mol À1 to give intermediate E with R-conguration ((R)-E); and the alternative Si-face attack proceeds via transition state Ts4-Si with a barrier of 9.7 kcal mol À1 , 1.9 kcal mol À1 higher than that of Ts4-Re, leading to an intermediate with S-conguration ((S)-E). The corresponding addition product 7a can be generated by proton transfer and elimination of the catalyst, the pathway similar to those reported by Yu and co-workers. 24a-c These observations suggest that the enantioselectivity is determined by the nucleophilic attack step and a value of 92% ee predicted by the B3LYP-D3 method based on the energy difference of Ts4-Re and Ts4-Si is in good agreement with the experimental result, where the R-product was formed preferentially (89% ee, entry 7 in Table 1). When sulfur is replaced by oxygen (5-methyl-2-phenyloxazol-4(5H)-one (8a)), the B3LYP-D3 calculations predict a value of 47% ee for the R-isomer (based on the energy difference of 0.6 kcal mol À1 between transition states Ts5-Re and Ts5-Si), consistent with the experimental observation (65% ee, entry 1 in Table 4).
Upon evaluating transition states Ts4-Re and Ts4-Si (Fig. 3), it was found that the bond lengths for the forming C1-C2 bond are similar, in addition to the distances of the hydrogen bonds between H3 and O2 (about 1.8Å). However, the short H2/C3 distance of 3.04Å in Ts4-Si suggests a repulsion between the phenyl group of the reacting thiazolone and one of the phenyl groups in the phosphine catalyst, resulting in a higher transition state barrier. To better illustrate the steric repulsions in the nucleophilic addition step, a 2D contour map along the z-axis (dened as the forming C-C bond) of the van der Waals 32 surface of Ts4-na is plotted (Fig. 4), representing the nucleophile moiety of transition state Ts4-Re without the thiazolone substrate. When the thiazolone group is deprotonated and bound to the catalyst via the N-H/O hydrogen bond during the formation of the C-C bond along the z-axis, the steric hindrance for the Re-face attack (labelled as R) is smaller than that for the Si-face attack (labelled as S). As such, Re-face attack becomes more favorable.

Experimental conrmation
The importance of hydrogen bonding interactions for asymmetric induction has been clearly demonstrated in the above computational studies. Sulfonamide 2c and its close structural analogs 2c 0 and 2c 00 were synthesized and applied to the g-addition of 5H-thiazol-4-one 5a to allenoate 6f (Table 7). Blockage of the sulfonamide N-H led to a dramatic decrease in reactivity and enantioselectivity (entry 2). When the sterically hindered O-silyl group was replaced by a free OH, not only the enantioselectivity, but also the reactivity of the reaction decreased signicantly (entry 3), suggesting  that the bulky silyl group may be crucial for locking the transition state geometry and differentiating the Re-and Siface attacks.

Conclusions
In summary, we have developed the rst phosphine-catalyzed highly enantioselective g-addition of 5H-thiazol-4-ones and 5Hoxazol-4-ones to 2,3-butadienoates. In the presence of amino acid-derived bifunctional phosphine catalysts, chiral thiazolones and oxazolones with a heteroatom (S or O)-containing tertiary chiral center were obtained in high yields and with excellent enantioselectivities. The optically enriched adducts are synthetically valuable, enabling the facile synthesis of optically enriched tertiary alcohols and thioethers. The method described in this report represents a method for rapid access to enantioenriched tertiary alcohol and thiol derivatives bearing an allylic chain, and may nd wide applications in synthetic organic chemistry. DFT calculations for understanding the mechanism revealed that the observed enantioselectivity results from a combination of three factors: (1) the hydrogen-bonding interaction between the amino moiety of the phosphine catalyst and the "C]O" unit of the thiazolone to activate the Michael donor, (2) the N-H/O interaction and the bulky O-silyl group to lock the conformation, and (3) the phenyl group of the thiazolone to differentiate the stereochemistry. It is noteworthy that this is the rst complete theoretical study for phosphine-catalyzed g-addition reactions. The theoretical results presented here are expected to offer new insight into the mechanisms of other phosphine-catalyzed asymmetric reactions, particularly those triggered by amino acid-derived phosphine catalysts. a Reactions were performed with 5a (0.1 mmol), 6f (0.12 mmol) and the catalyst (0.01 mmol) in Et 2 O (1.0 mL) at room temperature. b Isolated yield. c Determined by HPLC analysis on a chiral stationary phase.