SPA-PNN ligand for the kinetic resolution of carbocyclic and heterocyclic spiranes by asymmetric hydrogenation

Abstract

The chiral carbocyclic and heterocyclic spirane scaffolds are widely present in natural products and pharmaceutical molecules. Efficient, concise, and enantioenriched synthesis of these structures is a significant objective in both academic and industrial fields. Here, we design and synthesize a series of chiral SPA-PNN ligands, which are used to kinetically resolve racemic spirocyclic 1,3-diketones and their analogs via Ir-catalyzed asymmetric direct hydrogenation to afford the corresponding chiral spirocyclic 1,3-diketones and chiral hydroxyspiroketones. This asymmetric transformation features broad substrate scope (four different types of spirocycle), low catalyst loading (S/C = 5000), and an excellent kinetic resolution efficiency (selectivity factor of up to 207). Notably, the resulting chiral products are successfully used in the total and formal synthesis of four natural products, and in the preparation of key synthetic intermediates for structurally attractive ligands. Furthermore, control experiments and DFT calculations enable the origin of the observed enantioselectivity to be elucidated.

Introduction

Catalytic asymmetric direct hydrogenation (ADH) has been widely used to produce chiral drugs, agricultural chemicals, perfumes, and chemical materials^1,2,3. Indeed, diverse substrates with various achiral functionalities have been transformed into value-added products bearing stereogenic centers owing to ease of operation, the use of readily available hydrogen sources, high atom economies, environmental friendliness, and richly available selective catalytic systems; consequently, ADH has been extensively investigated in both academic and industry^4,5,6. Notably, catalytic asymmetric hydrogenation of 1,3-diketones has received considerable attention from synthetic communities, which produce chiral 1,3-diols and β-hydroxyketones (depending on the selective reaction conditions), two important skeletons in different research fields^7,8.

As a substrate subtype, acyclic and monocyclic prochiral 2,2-disubstituted-1,3-diketones can produce synthetically challenging products bearing quaternary carbon centers via desymmetrization^9,10; to the best of our knowledge, the asymmetric direct hydrogenation of this category of substrates has only been accomplished by Zhang and coworkers using the Ir/f-ampha complex they developed¹¹, with other transformations involving enzyme-catalyzed reductions^{12,13,14,15,16,17,18}, asymmetric transfer hydrogenation (ATH)^19,20,21, and Corey–Bakshi–Shibata (CBS) reduction²². Notably, more challenging spirocyclic diketones that possess quaternary centers and two carbonyl groups in different rings have only been reduced by enzyme catalysis and the CBS reaction by Nakazaki²³ and Chan²⁴, respectively, who only used spiro[4,4]nonane-1,6-dione as the substrate. And the substrates with differently sized rings within the spirocyclic framework have not been studied using catalytic asymmetric approaches, which is possibly ascribable to the following issues (Fig. 1b): (1) numerous mono-ol and diol enantiomers and diastereoisomers (up to eight products each) could be produced during the hydrogenation process, which adds to the complexity associated with catalytically controlling the desired reaction; (2) the rigidity of the spirocenter in the substrate significantly influences how the carbonyl groups interact with the chiral catalyst; and (3) spirocyclic 1,3-diketones readily decompose in protic solvents in the presence of a base²⁵. In addition, the resulting chiral mono-reduced product (i.e., the β-hydroxyketone) can unexpectedly occurred the retro-aldol reaction.

Fig. 1: Asymmetric hydrogenation of 1,3-diketones.

a Previous work: desymmetric enantioselective reduction of monocyclic 1,3-diketones via asymmetric hydrogenation (AH). b Challenging kinetic resolution of spirocyclic 1,3-diketones via AH. c This work: kinetic resolution of spirocyclic 1,3-diketones and heterocyclic spiranes via AH. d Representative spirocyclic natural products. e Selected chiral ligands and materials containing the spirocyclic structural motif.

Despite facing the abovementioned difficulties, the successful development of such a challenging transformation not only fills the gap in the asymmetric hydrogenation chemistry of spirocyclic 1,3-diketones and their analogs but also produces two important privileged spirocyclic skeletons: spirocyclic 1,3-diketones and β-hydroxyketones, both of which can be smoothly transformed into the corresponding 1,3-diol, a third synthetically important spirocyclic skeleton. These three skeletons are widely found in bioactive natural products^{26,27,28,29,30,31}, chiral ligands^32,33,34, pharmaceutical drugs^35,36, and functional materials^37,38,39,40 (Fig. 1d, e). Therefore, exploring efficient catalytic systems for directly asymmetrically hydrogenating spirocyclic 1,3-diketones and their analogs is an important project.

Results

Design and synthesis of the chiral catalysis

Chiral PNN-type compounds have been employed as privileged ligands^41,42,43 for the catalytic asymmetric hydrogenations of ketones, especially the ligands reported by Zhou et al.⁴⁴, owing to their notable turnover numbers and excellent enantioselectivities. Inspired by these achievements of PNN ligands and as part of our ongoing research interests in the design, synthesis, and application of SPA (spirocyclic amide) skeleton ligands^{45,46,47,48,49,50}, we postulated that the structural rigidity of the SPA framework would play a pivotal role in directing the spatial orientation of the N–H moiety during the coordination of the PNN-type ligand to a transition metal center. This geometric constraint could serve as a key determinant for enabling the stereochemically demanding asymmetric transformations outlined above^51,52. Herein, we report that a SPA skeleton ligand kinetically resolves spirocyclic 1,3-diketones via asymmetric direct hydrogenation; this catalytic system was further expanded to include spirolactams and N-heterocyclic spirocycles.

Starting from the known chiral compound 2⁴⁵, three-step chemical transformations including phosphinylation, imine condensation, and reduction afforded the chiral ligands with various substituents on pyridine rings and/or different aryl rings at phosphorus atoms (Fig. 2).

Fig. 2: Synthesis of the chiral SPA-PNN ligands.

All yields are isolated yields. The absolute configuration of L4 was determined by single-crystal X-ray diffractometry.

Evaluation of reaction conditions for the kinetic resolution of spirocyclic 1,3-diketons

With the chiral ligands in hand, we selected racemic spirocyclic 1,3-diketone rac-5a as a model substrate for investigating the regio-, diastereo-, and enantioselective of the ADH reaction (Table 1). The impact of solvent and base was investigated using a combination of [Ir(COD)Cl]₂ (0.25 mol%) and ligand L1 (0.55 mol%). Fortunately, rac-5a was smoothly converted into mono-reduced product (R,R)−6a (67% conversion, 62% ee for (R,R)−6a) when K₂HPO₄ and ethanol were used as the base and solvent, respectively, with enantioenriched (S)−5a (> 99% ee) recovered (entry 1); the structures of compounds (S)−5a and (R,R)−6a were determined by comparison with literature data⁵³. Notably, the cyclohexanone was predominantly reduced, with byproducts stemming from the cyclopentanone and/or over-reduction almost completely suppressed at a controlled conversion of approximately 50%. Various ligands were subsequently screened, which revealed that L2 and L3 afforded similar results (entries 3 and 4) and L4 gave the best outcome (entry 5, selectivity factor (s-factor) = 85)⁵⁴. The other ligands either affected the reaction rate or led to dissatisfactory results (entries 6–9). We also examined other reaction parameters, which included lowering the temperature, altering the ligands and base equivalents, changing the reaction concentration, and examining various H₂ pressures; however, none of these modifications resulted in any significant enhancement (see Supplementary Table S1 for the details). Moreover, inferior resolution efficiency was obtained with other ligands (see Supplementary Table S3 and S4 for details).

Table 1 Condition optimizations for the kinetic resolution of spirocyclic 1,3-diketons^a

Substrate scope

After determining the optimal reaction conditions, we next investigated the scope of the alkyl spirocyclic 1,3-diketones (Fig. 3). Spiro[4.5]decane-1,6-dione, either gem-dimethylated at the ortho-position to the cyclopentanone carbonyl group or at the para-position of the cyclohexanone carbonyl group, was subjected to the optimal reaction conditions to afford the mono-reduced product (R,R)-6b or (S,R)-6c with high selectivity (s-factor = 127 for rac−5b; s-factor = 77 for rac-5c). Furthermore, the more complex tricyclic substrate rac-5d reacted with notable efficiency (s-factor = 207). The absolute configuration of (R,R)-6d was determined by single-crystal X-ray diffraction of its derivative, (R,R)-6de (CCDC: 2431263).

Fig. 3: Substrate scope of alkyl spirocyclic 1,3-diketons.

^aAll reactions were run on a 0.1 mmol scale; isolated yields; the ee value of (S)-5 were determined by UPC² or GC analysis on a chiral stationary phase; the ee value of 6 was determined by UPC² or HPLC analysis after acetylation with acryloyl chloride; dr values were determined via ¹H NMR analysis of the reaction mixtures; s = ln[(1 − conv)(1 – ee₁)]/ln[(1 − conv)(1 + ee₁)], conv = ee₁/(ee₁+ ee₂). ^bL2 was used instead of L4. ^cEt₃N (0.5 equiv.) was used instead of K₂HPO₄. ^dAfter oxidized.

Spiro[4.4]nonane-1,6-dione (5e) was somewhat poorly kinetically resolved under the optimal reaction conditions to afford mono-ol (R,R)−6e, which prompted us to rescreen the ligands (Table 1, also see Supplementary Table S2 for the details), revealing that the use of ligand L2 led to a moderate kinetic resolution. In addition, while enantioenriched (S)-5e (96% ee) was achieved efficiently, it came at the expense of recovery yield (30%). Similar substrates 5f and 5g afforded the expected mono-reduced products, with chiral materials recovered in moderate yields.

In contrast to the abovementioned substrates, spiro[5.5]undecane-1,7-dione (rac-5h) was over-reduced to afford a mixture of the corresponding cis,cis- and cis,trans-diols that are difficult to separate by common column chromatography, as previously reported by Gerlach and coworkers⁵⁵. Therefore, the reaction of rac-5h was quenched at 51% conversion, and (S)-5h was separated by column chromatography (47% yield, 99% ee). The mixture of mono-ol, cis,cis-diol, and cis,trans-diol was then oxidized to afford chiral (R)-5h (40% yield from rac-5h, 92% ee). Notably, 4,4-dimethylspiro[5.5]undecane-1,7-dione (rac-5i) gave satisfactory results, with predominant reduction of the gem-dimethylated six-membered ring yielding two chiral products.

Larger-ring substrates (i.e., rac-5j–5 m) were also compatible with the developed catalytic system. Among these, rac-5m exhibited the highest selectivity (s-factor = 206). The absolute configurations of (S,R)−6i and (R,S)-6m were determined by single-crystal X-ray diffractometry, respectively, which enabled the configurations of other related products to be deduced. Unfortunately, spiro[6.6]tridecane-1,8-dione (rac-5n) was incompatible with the developed system; presumably, the structure of 5n is too large to fit into the chiral pocket of the catalytic system.

We next focused on the monoaryl-substituted spirocyclic 1,3-diketones scope (Fig. 4). The reaction parameters were thoroughly screened, which revealed that L8 is the optimal ligand in this case (see Supplementary Table S5 for the details). Gratifyingly, in addition to the C4-methyl-substited indanone (rac-7b), other methyl positions (i.e., rac-7c–7e) and a variety of functional groups on the aryl ring, including fluoride, chloride, and methoxy (i.e., rac-7g–7i, respectively), were well tolerated in this slightly modified system. Notably, the absolute configuration of (R,R)-8i was determined by single-crystal X-ray diffractometry (CCDC:2387315, see part six in the SI). Among the other examined ring systems, only rac-7l provided a good outcome (s-factor = 128).

Fig. 4: Substrate scope of monoaryl 1,3-diketospiranes.

^aAll reactions were run on a 0.1 mmol scale; isolated yields; the ee value of (S)-7 and (R,R)-8 were determined by UPC² analysis on a chiral stationary phase; dr values were determined via ¹H NMR analysis of the reaction mixtures; s = ln[(1 − conv)(1 – ee₁)]/ln[(1 − conv)(1 + ee₁)], conv = ee₁/(ee₁+ ee₂). ^b20  °C.

Our ability to successfully produce the two substrate types inspired to further research into spirolactams. Based on our previously screened reaction conditions, we selected those depicted in Fig. 3 for this particular substrate type (Fig. 5). Pleasingly, azaspiro[4,5], [4,4], and [5,5] substrates were all kinetically resolved, with rac-9h affording the best result (s-factor = 116).

Fig. 5: Substrate scope of spirolactams.

^aAll reactions were run on a 0.1 mmol scale; isolated yields; the ee value of (S)-9 and (R,R)-10 were determined by UPC² analysis on a chiral stationary phase; dr values were determined via ¹H NMR analysis of the reaction mixtures; s = ln[(1 − conv)(1 – ee₁)]/ln[(1 − conv)(1 + ee₁)], conv = ee₁/(ee₁+ ee₂). ^b[Ir(COD)Cl]₂ (0.5 mol%)/L4 (1.1 mol%) was used instead of [Ir(COD)Cl]₂ (0.25 mol%)/L2 (0.55 mol%).

After having successfully kinetically resolved the three types of spirocyclic 1,3-dicarbonyl compound, our interest shifted to heterocyclic spiranes owing to the prevalence of N-heterocyclic spirocycles in natural products and bioactive molecules (Fig. 1d). Ligand L6 was selected following condition screening (see Supplementary Table S6 for the details), which resulted in the formation of reduced product (S,R)-12e (47% yield with 81% ee, 11:1 dr) and recovered (R)-11e (47% yield with 80% ee). Next, we studied the scope of this protocol and found that satisfactory results were obtained for all tested substrates except for rac-11o (Fig. 6). The absolute configuration of (S,R)-12h (CCDC: 2387313, see part six in the SI) was also determined by single-crystal X-ray diffractometry.

Fig. 6: Substrate scope of N-heterocyclic spirocycles.

^aAll reactions were run on a 0.1 mmol scale; isolated yields; the ee value of (R)-11 and 12 were determined by UPC² analysis on a chiral stationary phase; dr values were determined via ¹H NMR analysis of the reaction mixtures; s = ln[(1 − conv)(1 – ee₁)]/ln[(1 − conv)(1 + ee₁)], conv = ee₁/(ee₁+ ee₂).

Synthetic applications

To investigate the practicality of the developed synthetic protocol, we kinetically resolved rac-5e on a 5.0 mmol scale using 0.02 mol% of the metal catalyst under the optimal reaction conditions for 24 h, which led to a similar outcome to that obtained on the small scale (Fig. 7a). The (S)-5e (92% ee) could be converted to ligand 1i⁵⁶ and ligand SpinPHOX⁵⁷. Moreover, the key intermediate (R)-5ee was prepared in two chemical steps from (S)-5e; (R)-5ee is used in the total synthesis of ginkgolide C by Barriault and coworkers³¹, or in the preparation of ligands SpinPO⁵⁸. Similarly, N-heterocyclic rac-11e was smoothly hydrogenated in the presence of 0.2 mol% of the catalyst to afford (R)-11e (42% yield with 91% ee), which could be converted into (−)-cephalotaxine according to a literature procedure (Fig. 7b)^59,60. Furthermore, rac-9h (2.0 mmol) was also satisfactorily kinetically resolved to afford the mono-reduced product (R,R)-10h (47% yield with 93% ee) and recovered (S)-9h (49% yield with 93% ee) (Fig. 7c). The former was reduced with BH₃•THF to (+)-nitramine in 56 % yield, whereas the latter was conveniently converted into the (−)-nitrabirine in two chemical steps using the Kagho’s procedure⁶¹.

Fig. 7: Synthetic applications.

a Kinetic resolution of rac-5e on a 5.0 mmol scale. b Kinetic resolution of rac-11e on a 5.0 mmol scale. c Kinetic resolution of rac−9h on a 2.0 mmol scale. Reaction condition: isolated yields; all ee values were determined by UPC² analysis on a chiral stationary phase. (I) PhNTf₂, LiHMDS, THF, – 78 °C to r.t., 10 h; (II) CO, Pd (OAc)₂, PPh₃, Et₃N, CH₃OH, DMF, r.t.; 8 h; (III) 1) Me₃OBF₄, CH₂Cl₂, 40  °C, 19 h; 2) 2,2-diethoxyethan-1-amine, CH₂Cl₂, r.t., 3 d; 3) HCl aq. (1 mol/L), 100  °C, 6 h, 75% (from (S)-9h); (IV) DIBAL-H, CH₂Cl₂, − 78  °C, 3 h; 40%.

Control experiments and DFT calculations

To verify the key role of the free N-H functional group^51,52 in ligand L2, we prepared a similar ligand L12 whose structure was designed to block the side of the N atom. A dramatically slower model reaction was observed when L12 was used instead of L2 under the optimal reaction conditions, with recovered racemic 7a along with racemic product 8a (Fig. 8a). In addition, ligand L13 devoid of the pyridine ring was used instead of L1 under the optimal reaction conditions, which led to only trace amounts of 8a after 3.5 h (Fig. 8b). These results demonstrate that both the free N-H and the N atom in the pyridinic ring are necessary, and that Ir metal should be triply coordinated to two N atoms and one P atom of the ligand. Taken together with previous works into the Ir-PNN-catalyzed hydrogenations of ketones and Noyori-type molecular catalysts, we postulated that the catalytic reaction pathway proceeds via an outer-sphere mechanism (Fig. 8c)^62,63,64,65.

Fig. 8: Control experiments and possible reaction model.

a Examination of the effect of the N-H moiety on the outcome of catalysis. b Examination of the effect of the pyridine moiety on the outcome of catalysis. c Possible reaction model.

To further understand the origin of the stereochemistry observed for this reaction, we modeled the stereochemistry determining hydride/proton-transfer step involving rac-5a and IrH₃ (see Supplementary Figs. S5–S12 for the details) using Density Functional Theory (DFT) calculations^66,67,68, and found that the transition state energy of TS1-RR is lowest (Fig. 9). The principal interacting orbital (PIO) analysis⁶⁹ (see Supplementary Fig. S15 for the details) revealed that the interaction between Ir-H and the carbon atom of the carbonyl group in the substrate promotes the reaction. Through distortion-interaction analysis (see Supplementary Table S7 for the details)⁷⁰, we observed that although the total distortion energies of TS1-RR are similar to TS1-SS, the more intense interaction energies in TS1-RR result in its ΔE^≠ being 3.5 kcal/mol lower than that of TS1-SS. Furthermore, the ΔE^≠ of TS1-RR is also lower than that of TS1-RS and TS1-SR by 4.7 kcal/mol and 3.1 kcal/mol, respectively. Comparing the results of transition state TS1-RR, TS1-r1, TS1-r2, TS1-r3, and TS1-r4, we found that the dominant factors influencing the regioselectivity were the total distortion energies of both the Ir(L)H₃ and the substrate rac-5a (ΔE_{dis-Ir(L)H₃} and ΔE_dis-sub). This distortion-interaction analysis is consistent with the experimental observation that only the two enantiomers are obtained, with no diastereoisomers being formed. In addition, DFT calculations revealed that the energy of the cyclohexanone reduction product (TM1, TM2) is lower than that of the cyclopentanone reduction product (TM3, TM4) (see Supplementary Fig. S16b for the details). Therefore, the reduction of the cyclohexanone in rac-5a is thermodynamically and kinetically more favorable than that of the cyclopentanone.

Fig. 9: DFT calculations and distortion-interaction analysis for the stereo-determined step.

Calculations were performed at the B3LYP-D3/Def2-TZVP/SMD (Ethanol)//B3LYP-D3/Def2-SVP/SMD (Ethanol) level of theory. Graphical structures were visualized with CYLview (Version 1.0b).

Discussion

We designed and synthesized a series of SPA-PNN ligands and used them to kinetically resolve a range of spirocyclic dicarbonyl compounds, including alkyl-substituted substrates, aryl-substituted substrates, spirolactams, and N-heterocyclic spirocycles. Spirocyclic diketones, β-hydroxyketones, and their analogues were obtained in good yields and with high optical purities. The resulting products could be converted to useful chiral ligands and bioactive natural products. The origin of the stereochemistry observed for the reaction was elucidated through a combination of control experiments and DFT calculations, which demonstrated that the N-H functional group and pyridine ring in the ligand are two vital factors for achieving high stereoselectivity. Other SPA-PNN-catalyzed asymmetric transformations are currently being investigated by our group.

Methods

Preparation of precatalysts

To a 20 mL sample bottle was added [Ir(COD)Cl]₂ (16.8 mg, 0.025 mmol), L ((S,S)-L1 − L8, L12, and L13) (0.055 mmol, 1.1 equiv.), and anhydrous EtOH (10 mL) in an argon-filled glovebox. The mixture was stirred at room temperature for 2 h to give a yellow solution. The solution should be stored in a refrigerator at a temperature of − 20 °C.

General procedure for kinetic resolution of carbocyclic and heterocyclic spiranes

To a 5 mL ampoule in an autoclave (50 mL) were added spirocyclic substrates (0.1 mmol, 1.0 equiv.), K₂HPO₄ (3.4 mg, 20 mol%), EtOH (2.0 mL) and a solution of precatalysts of L in EtOH (0.005 mmol/mL, 0.1 mL, 0.0005 mmol, 0.5 mol%) under an atmospheric environment. The autoclave was purged with hydrogen by pressurizing to 5 atm and releasing the pressure. This procedure was repeated three times and then pressurized to 50 atm of H₂. The reaction mixture was stirred at 20 °C. After the reaction finished, releasing the hydrogen pressure, the reaction mixture was concentrated under reduced pressure. The residue was purified by flash column chromatography on silical gel with petroleum ether/ethyl acetate as an eluent to afford the chiral alcohols and the recovered chiral spirocyclic 1,3-diketons.