Zinc-Catalyzed Enantioselective [3+2] Cycloaddition of Azomethine Ylides Using Planar Chiral [2.2]Paracyclophane-Imidazoline N,O-ligands.

We present a facile synthetic route toward a novel series of imidazolinyl-[2.2]paracyclophanol (UCD-Imphanol) ligands possessing central and planar chirality. Both sets of diastereomeric ligands were successfully purified by column chromatography. The preliminary application of this family of ligands showed excellent activities in the asymmetric Zn-catalyzed azomethine ylide cycloaddition. Enantioenriched pyrrolidines, in a substrate scope of 20 examples, were accessed in high levels of endo/exo ratios (up to >99/1) and enantioselectivities (up to >99 % ee) with excellent yields (up to 99 %) by using (S,S,SP )-UCD-Imphanol/(S,S,RP )-UCD-Imphanol, respectively.

Introduction

Planar chiral paracyclophanes have attracted widespread attention because of their unique structural and electronic features and have been extensively utilized in bio‐/material sciences and asymmetric catalysis. However, their asymmetric catalytic applications have been less explored compared to planar chiral metallocenes due to the significant challenges associated with the separation of enantiopure cyclophanes.[ , ] Based on the disubstitution pattern, chiefly, there are three types of paracyclophanes, namely pseudo‐geminal, pseudo‐ortho, and ortho‐derivatives that have been studied for asymmetric catalysis. The ortho‐disubstituted paracyclophanes offer the highest level of steric crowding and a tight catalytic pocket which could be useful for asymmetric catalytic applications. However, the potential synthetic attractiveness of ortho‐difunctionalized paracyclophane‐derived ligands is limited by their challenging synthesis and non‐facile structural modifications, including fine tuning of donor atom electronics.[ , , ] Nevertheless, considerable effort has been devoted to this class of bidentate ligands with N,O‐, N,S/Se‐, and P,N‐ligands 1–3 reported (Figure 1). Among these, N,O‐ligands have proven to be successful while other ligands showed moderate to poor levels of asymmetric induction in a range of transformations.[ , , , ] In 2001, Bräse introduced the first catalytic application of paracyclophane‐derived ketimine N,O‐ligands 1 for organozinc additions to aldehydes/imines, and were highly efficient for several transformations. The synthesis and application of oxazolinyl‐paracyclophane N,O‐ligand 3 in the diethylzinc addition to aldehydes was investigated by Bolm, but moderate enantioselectivities were obtained.

Figure 1

[2.2]Paracyclophane‐derived ortho‐disubstituted ligands.

Chiral oxazoline‐containing ligands have been extensively applied in asymmetric catalysis, whereas chiral imidazoline‐based ligands, which possess an additional nitrogen atom compared to oxazoline, affords the opportunity for fine‐tuning the ligand electronic and conformational properties by judicious choice of the substituent in the non‐ligating nitrogen atom. This frequently accounts for significantly improved levels of reactivity and asymmetric induction relative to the comparable oxazoline‐containing ligands.

As a part of our interest in designing chiral imidazoline based ligands, we proposed the synthesis of a series of sterically congested planar chiral [2.2]paracyclophane‐derived imidazoline ligands 4–6 (Figure 1). In this context, the imidazoline unit acts as an in‐built resolution unit, and the non‐ligating nitrogen would be useful for tuning the electronic properties of the ligand. This blueprint would allow us to generate an N,O‐ligand library possessing planar chirality and central chirality, along with the added advantages of convenient synthesis and resolution, modularity and electronic tunability. Here, we report the synthesis and resolution of this new class of planar chiral [2.2]paracyclophane‐derived imidazoline N,O‐ligands 4–6 and their application in the highly enantioselective ZnII‐catalyzed [3+2] azomethine ylide cycloadditions.

Results and Discussion

This study commenced with the ligand synthesis which was achieved in eight simple steps from commercially available [2.2]paracyclophane 7, where the formyl group was installed by Ti‐mediated formylation to furnish racemic 8 in 84 % yield (Scheme 1). According to Bolm's procedure, the required ortho‐acetoxylation of paracyclophane 10 was achieved by Pd‐catalyzed C−H activation by a two‐step process. Hydrolysis of acetyl/oxime ether 10 through the action of p‐TsOH⋅H2O/HCHO led to the aldehyde 11 in 74 % yield. Later, the hydroxy group of 11 was protected by the reaction of MeI in the presence of K2CO3, and without further purification the intermediate was subjected to cyclization with (1S,2S)‐(−)‐1,2‐diphenylethylenediamine 12 mediated by NBS in CH2Cl2 to provide the inseparable diastereomers of the corresponding 2‐substituted imidazoline 13 in 73 % yield. Imidazoline 13 was subsequently treated with various benzyl bromide derivatives to afford the diastereomers of benzylated imidazolines 14 and 15 in 73 and 46 % (combined yields) respectively, which were readily separable. In contrast, pentafluorobenzylation afforded inseparable diastereomers 16 in 82 % yield. The absolute configuration of 14  a was unequivocally established as (S,S,S P) by single crystal X‐ray diffraction analysis. Finally, methyl ether protection in 14–16 was removed by the treatment of BBr3 leading to the smooth generation of ligands 4–6 in moderate to good yields, with diastereomers 6  a and 6  b now separable.

Scheme 1

Synthesis and resolution of imidazolinyl‐[2.2]paracyclophanols (UCD‐Imphanols).

With a newly developed ligand library in hand, we sought to explore their reactivity in asymmetric catalysis. Asymmetric azomethine ylide cycloaddition is a widely used methodology for the construction of enantioenriched pyrrolidines, useful in medicinal and biological applications. Several ligand classes along with suitable metal combinations have been investigated extensively,[ , ] and AgI‐ and CuI‐based catalytic systems have been most prominent,[ , ] whereas other metals such as CaII, FeII, NiII  and ZnII  ‐based catalytic systems have been explored relatively rarely (Figure S1, Supporting Information).

In 2002, Jørgensen established the first Zn‐( Bu‐BOX)‐catalyzed enantioselective azomethine ylide cycloaddition, which provided products in up to 94 % ee (Scheme 2). Later, Dogan and Garner reported up to 95 % ee for the synthesis of enantioenriched pyrrolidines by azomethine ylide cycloaddition using ferrocene‐derived N,O‐ligands/ZnII (Scheme 2). Since Dogan and Garner's report, the flexibility of N,O‐ligands along with a zinc platform would be more suitable to develop a general catalyst for asymmetric azomethine ylide cycloadditions. Further, the diverse coordination of ZnII supported by bidentate N,O‐ligands has been exploited to achieve high enantioselectivities in several asymmetric transformations. Moreover, the development of Zn‐based catalytic system would be desirable and valuable in terms of cost‐efficiency. For these reasons, we chose to apply our novel series of planar chiral UCD‐Imphanols 4–6 to this transformation.

Scheme 2

ZnII‐catalyzed asymmetric [3+2] azomethine ylide cycloaddition.

Our initial efforts on the application of ligands 4–6 in Zn‐catalysis focused on the asymmetric [3+2] cycloaddition of azomethine ylide with maleimides (Table 1). The reaction proceeded smoothly at room temperature and the results clearly indicated that yield, endo/exo ratios and enantioselectivity were affected by the electronic properties of the imidazoline N‐substituents (entries 1–6), and the planar chiral element is the dominant factor in controlling the asymmetric induction with reversal of planar chirality between ligands (S,S,S P)‐4  a–6  a and (S,S,R P)‐4  b–6  b, reversing the stereochemical outcome of the reaction. To further understand the importance of chirality at the C‐5 imidazoline ring and the role of the planar chiral element in these cycloadditions, we synthesized and tested the related ligands (S,S P)‐20 and (S,S)‐21 (see Supporting Information). Substitution at the C‐5 imidazoline ring is crucial for both reactivity and selectivity (entry 7), and the non‐planar chiral N,O‐ligand (S,S)‐21 showed poor reactivity and enantioselectivity (entry 8), which suggests that the planar paracyclophane unit is a requirement to achieve the high reactivity, and is necessary for enhanced asymmetric induction. We selected the ligands (S,S,S P)‐6  a and (S,S,R P)‐4  b for further optimization studies (see Supporting Information), which found dichloromethane to be the best solvent, and 1,4‐diazabicyclo[2.2.2]octane (DABCO) and N,N‐diisopropylethylamine (DIPEA) the optimal base for generating the respective enantiomeric cycloadducts with the lower temperature of 0 °C providing superior ee's (entry 9 and 10).

Table 1

Optimization of reaction conditions.[a]

Entry	Ligand	Base	T [°C]	Yield [%][b]	endo/exo [c]	ee [%][d]
1	(S,S,S P)‐4  a	Et3N	rt	61	94/6	92.3
2	(S,S,S P)‐5  a	Et3N	rt	68	95/5	86.7
3	(S,S,S P)‐6  a	Et3N	rt	75	97/3	91.6
4	(S,S,R P)‐4  b	Et3N	rt	67	94/6	88.0[e]
5	(S,S,R P)‐5  b	Et3N	rt	78	95/5	85.9[e]
6	(S,S,R P)‐6  b	Et3N	rt	59	92/8	63.7[e]
7	(S,S P)‐20	Et3N	rt	45	92/8	83.5
8	(S,S)‐21	Et3N	rt	34	87/13	5.6
9	(S,S,S P)‐6  a	DABCO	0	86	99/1	99.3
10	(S,S,R P)‐4  b	DIPEA	0	79	>99/1	95.4[e]

[a] Reaction conditions: 17  a (0.225 mmol) 18  a (0.15 mmol), Zn(OTf)2 (10 mol %), L (11.5 mol %), base (10 mol %) in 1.0 mL of CH2Cl2, 8–28 h. [b] Isolated yield. [c] The endo/exo ratio was determined by 1H NMR spectroscopy of crude reaction mixture. [d] The ee was determined by chiral SFC analysis. [e] Opposite enantiomer (1R,3S,3aR,6aS)‐19′a.

Please add color to the paracyclophane rings (as originally submitted) to Schemes 1 and 5, Table 1 and the Graphical Abstract.

Please add the X‐ray crystal structure (as originally submitted) of (S,S,S P)‐14a to Scheme 1.

With the optimal reaction conditions in hand, the scope of the reaction was investigated, Scheme 3, and both (S,S,S P)‐6  a and (S,S,R P)‐4  b, behaving as pseudo‐enantiomeric ligands, performed well to furnish the respective cycloadducts in good yields and excellent levels of endo/exo ratios and enantioselectivities. A variety of imino esters 17 derived from aldehydes with different steric and electronic natures are compatible in this cycloaddition affording cycloadducts 19/19′ in good yields (condition A, 61–98 % yield and condition B, 62–99 % yield) with excellent levels of endo/exo (up to >99/1) and enantioselectivities (condition A, 98.0–>99.9 % ee) and condition B, 82.9–99.8 % ee). It is noteworthy that sterically hindered 2‐bromo, 2‐methyl and 1‐naphthyl substituted imino esters 17  h–17  j are also well tolerated to afford the products 19  h–19  j in high yields (condition A, 91–98 % yield) and excellent endo/exo ratios and enantioselectivities (condition A, 98.4–>99.9 % ee). Ligand 4  b also displayed good reactivity (condition B, 80–98 % yield) but the levels of enantioselectivity were slightly compromised 19′h–19′j (condition B, 82.9–92.4 % ee). In addition, heteroaromatic 2‐thienyl derived imino ester 17  k and ‐methyl substituted imino ester 19  l are also viable substrates, delivering the cycloadducts 19  k/19  k′ and 19  l/19  l′ with excellent levels of enantiocontrol. The electronic and steric nature of N‐substitution of maleimides 18  p–18  s has a notable influence on the reaction yields without erosion of endo/exo ratio and enantioselectivities (19  p–19  s and 19′p–19′s).

Scheme 3

Substrate scope.[a–d] [a] Reaction conditions: 17 (0.225 mmol) 18 (0.15 mmol), Zn(OTf)2 (10 mol %), (S,S,S P)‐6  a/(S,S,R P)‐4  b (11.5 mol %), DABCO/DIPEA (10 mol %) in 1.0 mL of CH2Cl2, 28 h. [b] Isolated yield. [c] The endo/exo ratio was determined by 1H NMR spectroscopy of crude reaction mixture. [d] The ee was determined by chiral SFC analysis. [e] 1 mmol scale.

Late‐stage functionalizations of biologically active frameworks are often appealing to identify key medicinally active molecules. To demonstrate the use of our strategy in late‐stage functionalization, steroid derived imino ester 17  t and N‐methyl maleimide 18  a were subjected to the standard reaction conditions. Zinc complexes of ligands (S,S,S P)‐6  a and (S,S,R P)‐4  b were successful catalysts, affording the corresponding cycloadducts 19  t as a mixture of diastereomers (3 : 2 : 2 : 1) in 86 % yield (not shown in Scheme 3, see Supporting Information) and 19′t exclusively as a single diastereomer (>99 % de) in 94 % yield, respectively, with the synthesis of 19′t representing a rare example of an asymmetric approach to this class of compound. These results indicate that both catalyst and substrate control are involved in the asymmetric induction process.

Our optimized Zn‐catalyzed [3+2] cycloaddition can be easily scaled up to 1 mmol scale, furnishing 19  i in 85 % yield and excellent levels of diastereo‐ and enantioselectivities (endo/exo >99/1, 99.2 % ee). Similarly, the product 19′d was isolated in 92 % yield with endo/exo >99/1 and 99.7 % ee.

Functionalities present in pyrrolidines 19 offer an opportunity for further functionalization towards synthetically useful products, without erosion of stereochemistry. As examples, product 19  h was N‐allylated using allyl bromide 22 to generate 23 in 78 % yield, followed by a Pd‐catalyzed Heck reaction of 23 to give the fused pyrroloisoquinoline 24 in 47 % yield (Scheme 4). Further, we performed the Ag‐catalyzed multicomponent cycloaddition of N‐phenylmaleimide 25, cinnamaldehyde 26, and product 19′d, affording the complex fused tetracyclic pyrrolizidine 27 in 76 % yield with good diastereocontrol (endo/exo 92/8). Both pyrroloisoquinoline and pyrrolizidine cores are common in biologically active natural products.

Scheme 4

Synthetic transformations.

The experimentally observed high endo and enantioselectivities can be explained by our proposed transition‐state model (Scheme 5). We suggest that the transition state contains the azomethine ylide coordinating to the ZnII‐imidazolinyl‐paracyclophanol in a tetrahedral complex. With ligand (S,S,S P)‐6  a, the sterically crowded paracyclophane and C‐4 phenyl group effectively shield the top‐face approach to the Re‐face of the ylide, while the Si‐face is less crowded thus allowing the dipolarophile approach in an endo fashion to deliver the products in exceptionally high levels of enantioselectivity. In the case of ligand (S,S,R P)‐4  b, the sterically crowded paracyclophane and the C‐5 phenyl group effectively block the Si‐face of ylide, and the dipolarophile approaches in an endo fashion to the ylide Re‐face. Even though the Re‐face is slightly shielded by the C‐4 phenyl ring, the dipolarophile approaches from the bottom‐face, which leads to a slight compromise of enantioselectivity compared to ligand (S,S,S P)‐6  a. Both observations suggest that the sterics of paracyclophane are crucial for the observed high levels of asymmetric induction.

Scheme 5

Proposed transition‐state model for ligands 6  a and 4  b.

Conclusion

In summary, we have designed and synthesized a series of new planar chiral imidazolinyl‐paracyclophanol N,O‐ligand (UCD‐Imphanol) from commercially available [2.2]paracyclophane in eight steps. Notably, both sets of diastereomeric ligands were successfully resolved and their configuration was assigned by XRD studies. This novel class of ligand demonstrated excellent efficiency in the enantioselective ZnII‐catalyzed [3+2] azomethine ylide cycloaddition, furnishing the corresponding cycloadducts in excellent diastereo‐ and enantioselectivities, the best reported to date. Our studies also confirmed that planar chirality is the dominant role controlling asymmetric induction. The excellent level of ee's (up to >99 % ee), along with easy access to both enantiomerically pure cycloadducts using less expensive Zn‐catalysis, showcases our strategy as optimal compared to other Cu‐ or Ag‐based catalytic systems. Further applications of this family of ligands in asymmetric catalysis are currently ongoing in our laboratory and will be reported in due course.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.