Since Akiyama and Terada independently reported chiral Brønsted acid-catalyzed asymmetric Mannich reactions in 2004, respectively, BINOL-derived chiral phosphoric acids (CPAs) have become a mainstay of organocatalysis, and such catalysts are now commercialized.
The advantage of BINOL-type chiral phosphoric acids (CPAs) lies in their rigid, well-defined chiral environment, therefore, the design of Brønsted acid catalysts has focused on the development of more sterically hindered chiral phosphoric acid catalysts. Although this method helps to achieve high enantioselectivity, the large steric hindrance complicates the synthesis work and limits the possibility of catalyst diversification and further optimization. Furthermore, studies of key transition state structures have found that noncovalent interactions (NCIs) between catalysts and substrates often play an important role in stereoselectivity. These interactions lead to the dependence of the catalyst on the structure of the substrate. To address this challenge, researchers hope to develop an alternative class of chiral phosphoric acid catalysts that induce stereoselectivity through non-covalent interactions rather than relying on rigid chiral environments. Therefore, researchers have recently proposed a case of chiral phosphoric acid backbone through data science. Since the transfer hydrogenation of 8-aminoquinolines was reported to be not very selective (up to 86:14 er), the researchers chose this reaction as a benchmark for their study.
First, the researchers constructed a mock library of commercially available carboxylic acids, acid chlorides, sulfonyl chlorides, and symmetrical secondary amines, generating >300,000 potential fragment combinations. By establishing quantitative structure-activity relationship (QSAR) structure descriptors for all combinations, about 1100 candidate structures were finally obtained, and their optimized conformations were described by computational chemistry. The researchers emphasize the absolute advantages of computer algorithms in terms of cost control and rapid exploration of available chemical framework structures.
Subsequently, the researchers investigated the corresponding stereoselectivity of the screened catalysts based on the transfer hydrogenation reaction of 8-aminoquinoline, and obtained results consistent with the predictions of the initial training sample set of principal component analysis (PCA). This suggests that this strategy provides a possible optimization route during the catalyst design process. The researchers found that all catalysts containing the sulfonamide moiety achieved good selectivity (>85:15 er), indicating that the sulfonamide moiety contributes to the stereoselectivity of the reaction. Focusing on catalysts that do not exhibit enantioselectivity, the researchers believe that there may be a key structural feature that determines whether a catalyst is stereoselective or not.
The researchers next used data science to dig deeper into the reasons for this selectivity difference, and a more comprehensive set of catalyst descriptors was collected. The researchers used multiple linear regression (MLR) models to probe the stereoselectivity of transfer hydrogenation reactions to gain insight into important features of transfer selectivity and provide a basis for further catalyst optimization.
After the preliminary establishment of the computational model, the researchers expect the high enantioselectivity of the transfer hydrogenation reaction to be promoted by sulfonamide catalysts, thereby predicting the feasibility of the validation set of sulfonamide catalysts. The experimental results found the optimal catalyst (95:5 er) for the transfer hydrogenation reaction of 8-aminoquinoline, which also reflected the accuracy of the calculation model prediction.
Finally, the researchers explored the feasibility of this type of chiral phosphoric acid catalyst in catalyzing other types of reactions. The researchers chose the previously reported block-selective cyclodehydration reaction and oxetane desymmetry reaction. Experimental results show that the proposed classification parameter (ChelpG OR1) is also applicable to these two reactions, further showing the utility of OR1 partial charge and the feasibility of evaluating candidate catalysts.
Related Products & Services
Reference
- Data science enables the development of a new class of chiral phosphoric acid catalysts
Jordan P. Liles, Caroline Rouget-Virbel, Julie L.H. Wahlman, René Rahimoff, Jennifer M. Crawford, Abby Medlin, Veronica S. O'Connor, Junqi Li, Vladislav A. Roytman, F. Dean Toste, Matthew S. Sigman
